Back

Google CEO says more than a quarter of the company's new code is created by AI

539 points2 daysbusinessinsider.com
S0y2 days ago
asdfman12316 hours ago

I work for Google, and I just got done with my work day. I was just writing I guess what you'd call "AI generated code."

But the code completion engine is basically just good at finishing the lines I'm writing. If I'm writing "function getAc..." it's smart enough to complete to "function getActionHandler()", and maybe suggest the correct arguments and a decent jsdoc comment.

So basically, it's a helpful productivity tool but it's not doing any engineering at all. It's probably about as good, maybe slightly worse, than Copilot. (I haven't used it recently though.)

NotAnOtter8 hours ago

I also work at google (until last Friday). Agree with what you said. My thoughts are

1. This quote is clearly meant to exaggerate reality, and they are likely including things like fully automated CL/PR's which have been around for a decade as "AI generated".

2. I stated before that if a team of 10 is equally as productive as a team of 8 utilizing things like copilot, it's fair to say "AI replaced 2 engineers", in my opinion. More importantly, Tech leaders would be making this claim if it were true. Copilot and it's clones have been around long enough know for the evidence to be in, and no one is stating "we've replaced X% of our workforce with AI" - therefor my claim is (by 'denying the consequent'), using copilot does not materially accelerate development.

ahmedfromtunis6 hours ago

> no one is stating "we've replaced X% of our workforce with AI"

Even if that's been happening, I don't think it would be politically savvy to admit it.

In today's social climate claiming to replace humans with AI would attract the wrong kind of attention from politicians (during an election year) and from the public in general.

This would be even more unwise to admit for a company like Google who's an "AI producer". They may leave such a language for closed meetings with potential customers during sales pitches though.

burningChrome8 minutes ago

>> Even if that's been happening, I don't think it would be politically savvy to admit it.

When I was working in RPA (robotic process automation) about 7 years ago, we were explicitly told not to say "You can reduce your team size by having use develop an automation that handles what they're doing!"

Even back then we were told to talk about how RPA (and by proxy AI) empowers your team to focus on the really important things. Automation just reduces the friction to getting things done. Instead of doing 4 hours of mindless data input or moving folders from one place to the other, automation gives you back those four hours so your team can do something sufficiently more important and focus on the bigger picture stuff.

Some teams loved the idea. Other leaders were skeptical and never adopted it. I spent the majority of those three years trying to selling them on this idea automation was good and very little time actually coding. Its interesting seeing the paradigm shift and seeing this stuff everywhere now.

whywhywhywhy5 hours ago

> and from the public in general

Don't think the public will be that concerned about people in Google's salary bracket losing their jobs.

+1
jl651 minutes ago
pjmlp40 minutes ago

No one cares about self checkout on supermarkets impact on their employees, until their employer does something similar.

+1
ytss5 hours ago
wbl19 minutes ago

The bank tellers didn't go away: they just became higher paid and higher skilled when cash management was no longer the job.

lenerdenator3 hours ago

> I don't think it would be politically savvy to admit it.

Would it be? Do they care?

Sam Altman's been talking about how GenAI could break capitalism (maybe not the exact quote, but something similar), and these companies have been pushing out GenAI products that could obviously and easily be used to fake photographic or video evidence of things that have occurred in the real world. Elon's obsessed with making an AI that's trained to be a 20-year-old male edgelord from the sewer pits of the internet.

Compared to those things, "we've replaced X% of our workforce with AI" is absolutely anodyne.

agentultra57 minutes ago

100%.

Altman encourages anyone that will listen to him that monopolies are the only path to success in business. He has a lot riding on making sure everyone is addicted to AI and that he’s the one selling the shovels.

Google isn’t far off.

Most capitalists have this fantasy that they can reduce their labour expenses with AI and continue stock buy-backs and ever-increasing executive payouts.

What sucks is that they rely on class divisions so that people don’t feel bad when the “overpaid” software developers get replaced. Problem is that software developers are also part of the proletariat and creating these artificial class divisions is breaking up the ability to organize.

It’s not AI replacing jobs, it’s capital holders. AI is just the smoke and mirrors.

rty324 hours ago

> I don't think it would be politically savvy to admit it.

Depends on who you ask.

If Trump wins and Elon Musk actually gets a new job, they would be bragging about replacing humans with AI all day long. And corporates are going to love it.

Not sure about what voters think though. But the fact that most of these companies are in California, New York etc means that it barely matters.

petre11 minutes ago

Yup, just like full self driving and ending the war in Ukraine on 24 hours.

ulfw19 minutes ago

Corporates will soon have to realise the hard reality that when masses of humans have been replaced there won't be masses of humans with salaries to buy said corporate's goods anymore.

+1
tjahg3 hours ago
onion2k8 hours ago

no one is stating "we've replaced X% of our workforce with AI"

That's only worth doing if you're trying to cut costs though. If the company has unmet ambitions there's no reason to shrink the headcount from 10 to 8 and have the same amount of output when you can keep 10 people and have the output of 12 by leveraging AI.

hyperpape7 hours ago

Almost all the big tech companies have had layoffs over the past several years. I think it’s safe to say cost cutting is very much part of their goal.

lupire5 hours ago

But the specific roles being laid off are arbitrary, and the overall goal headcount reduction is driven by macroeconomics factors (I'm being generous there), not based on new efficiencies.

Note the difft between "cost cutting" (do less, to lower cost) and "efficiency" (do same, but with less cost)

wcoenen24 minutes ago

> including things like fully automated CL/PR's which have been around for a decade

I haven't seen this yet so I'm intrigued. Is this a commercial product, or internal tooling?

pj_mukh3 hours ago

"we've replaced X% of our workforce with AI"

Most likely what is actually happening is that the X% of workforce you would lay off is being put to other projects and Google in general can take on X% more projects for the same labor $$. So there is no real reason to make that particular "replaced" statement.

ktnaWA5 hours ago

Thanks, that is how I read the announcement. The powers that be decided that there must be some quota to be fulfilled, and magically that quota was fulfilled.

AI engineers will not yet get a Nobel prize for putting everyone out of work.

nlehuen8 hours ago

I also work at Google and I agree with the general sentiment that AI completion is not doing engineering per se, simply because writing code is just a small part of engineering.

However in my experience the system is much more powerful than you described. Maybe this is because I'm mostly writing C++ for which there is a much bigger training corpus than JavaScript.

One thing the system is already pretty good at is writing entire short functions from a comment. The trick is not to write:

  function getAc...
But instead:

  // This function smargls the bleurgh
  // by flooming the trux.
  function getAc...
This way the completion goes much farther and the quality improves a lot. Essentially, use comments as the prompt to generate large chunks of code, instead of giving minimum context to the system, which limits it to single line completion.
Aachen6 hours ago

This type of not having to think about the implementation, especially in a language that we've by now well-established can't be written safely by humans (including by Google's own research into Android vulnerabilities if I'm not mistaken), at least with the current level of LLM, worries me the most

Time will tell whether it outputs worse, equal, or better quality than skilled humans, but I'd be very wary of anything it suggests beyond obvious boilerplate (like all the symbols needed in a for loop) or naming things (function name and comment autocompletes like the person above you described)

munksbeer2 hours ago

> worries me the most

It isn't something I worry about at all. If it doesn't work and starts creating bugs and horrible code, the best places will adjust to that and it won't be used or will be used more judiciously.

I'll still review code like I always do and prevent bad code from making it into our repo. I don't see why it's my problem to worry about. Why is it yours?

+1
Aachen1 hour ago
mbfg46 minutes ago

Except that no one trusts Barney down the hall that has stack overflow open 24/7. People naturally trust AI implicitly.

caeril2 hours ago

It's worrying, yes, but we've had stackoverflow copy-paste coding for over a decade now already, which has exactly the same effects.

This isn't a new concern. Thoughtless software development started a long time ago.

Aachen59 minutes ago

As a security consultant, I think I'm aware of security risks all the time, also when I'm developing code just as a hobby in spare time. I can't say that I've come across a lot of stackoverflow code that was unsafe. It happened (like unsafe SVG file upload handling advice) and I know of analyses that find it in spades, but I personally correct the few that I see (got enough stackoverflow rep to downvote, comment, or even edit without the user's approval though I'm not sure I've ever needed that) and the ones found in studies may be in less-popular answers that people don't come across as often because we should be seeing more of them otherwise, both personally and in the customer's code

So that's not to say there is nothing to be concerned about on stackoverflow, just that the risk seems manageable and understood. You also nearly always have to fit it to your own situation anyway. With the custom solutions from generative models, this is all not yet established and you're not having to customise (look at) it further if it made a plausible-looking suggestion

Perhaps this way of coding ends up introducing fewer bugs. Time will tell, but we all know how many wrong answers these things generate in text as well as what they were trained on, giving grounds for worry—while also gathering experience, of course. I'm not saying to not use it at all. It's a balance and something to be aware of

I also can't say that I find it to be thoughtless when I look for answers on stackoverflow. Perhaps as a beginning coder, you might copy bigger bits? Or without knowing what it does? That's not my current experience, though

miki1232117 hours ago

This is a good idea even outside of Google, with tools like copilot and such.

Often when I don't know exactly what function / sequence of functions I need to achieve a particular outcome, I put in a comment describing what I want to do, and Copilot does the rest. I then remove the comment once I make sure that the generated code actually works.

I find it a lot less flow-breaking than stackoverflow or even asking an LLM.

It doesn't work all of the time, and sometimes you do have to Google still, but for the cases it does work for, it's pretty nice.

Aachen6 hours ago

Why remove the comment that summarises the intent for humans? The compiler will ignore your comment anyway, so it's only there for the next human who comes along and will help them understand the code

+1
randomdata4 hours ago
+1
qwertox6 hours ago
+2
cryptonym6 hours ago
jappgar6 hours ago

I can guarantee you there is more publicly accessible javascript in the world than C++.

Copilot will autocomplete entire functions as well, sometimes without comments or even after just typing "f". It uses your previous edits as context and can assume what you're implementing pretty well.

infecto5 hours ago

I can guarantee you that the author was referencing code within Google. That is, their tooling is trained off internal code bases. I am imagining c++ dwarfs javascript.

lupire5 hours ago

Google does not write much publicly available JavaScript. They wrote their own special flavor. (Same for any hugel legacy operation)

+1
bilekas5 hours ago
tomhallett3 hours ago

Has anyone made a coding assistant which can do this based off audio which I’m saying out loud while I’m typing (interview/pairing style), so instead of typing the comment I can just say it?

hecanjog2 hours ago

I had some success using this for basic input, but never took it very far. It's meant to be customizable for that sort of thing though: https://talon.wiki/quickstart/getting_started/ (Edit: just the voice input part)

cryptonym6 hours ago

I find writing code to be almost relaxing plus that's really a tiny fraction of dev work. Not too excited about potential productivity gains based purely on authoring snippets. I find it much more interesting on boosting maintainability, robustness and other quality metrics (not focusing on quality of AI output, actual quality of the code base).

michaelbuckbee6 hours ago

I don't work at Google, but I do something similar with my code: write comments, generate the code, and then have the AI tooling create test cases.

AI coding assistants are generally really good at ramping up a base level of tests which you can then direct to add more specific scenario's to.

alickz6 hours ago

Comment Driven Programming might be interesting, as an offshoot of Documentation Driven Programming

gniv5 hours ago

That's pretty nice. Does it write modern C++, as I guess it's expected?

atoav12 hours ago

So this is basically the google CEO saying "a quarter of our terminal inputs is written by a glorified tab completion"?

asdfman12311 hours ago

Yes. Most AI hype is this bad. They have to justify the valuations.

remus9 hours ago

"tab completion good enough to write 25% of code" feels like a pretty good hit rate to me! Especially when you consider that a good chink of the other 75% is going to be the complex, detailed stuff where you probably want someone thinking about it fairly carefully.

+2
rantallion8 hours ago
red_admiral5 hours ago

"rm re[TAB]" to remove a file called something like "report-accounting-Q1_2024.docx" is really helpful, especially when it adds quotes as required, but not exciting enough to get me out of bed any earlier in the morning.

I feel it's a bit like the old "measuring developer productivity in LoC" metric.

As I hinted at in another comment, in Java if you had a "private String name;" then the following:

    /**
     * Returns the name.
     * @return The name.
     */
    public String getName() {
        return this.name;
    }
and the matching setter, are easy enough to generate automatically and you don't need a LLM for it. If AI can do that part of coding a bit better, sure it's helpful in a way, but I'm not worried about my job just yet (or rather, I'm more worried about the state of the economy and other factors).
+1
Maxion9 hours ago
keybored6 hours ago

25% is a great win if you are prone to RSI. And for quicker feedback. But in terms of the overarching programming goal? Churning out code is a small part of it.

Code is often a liability.

shombaboor3 hours ago

It would be funny if they had a metric for how much code is completed by CTRL+V

unglaublich10 hours ago

Yes, isn't that the essential idea of industrialization and automation?

OtherShrezzing8 hours ago

I think the critique here is that the AI currently deployed at Google hasn't meaningfully automated this user's life, because most IDEs already solved "very good autocomplete" more than a decade ago.

+1
tormeh4 hours ago
busterarm2 hours ago

glorified, EXPENSIVE tab completion.

walthamstow2 hours ago

I assume you're referring to the compute/energy used to run the completion?

busterarm2 hours ago

to train the model

mmmpetrichor11 hours ago

Yeah, but he wants people to hear "reduce headcount by 25% if you buy our shit!"

mewpmewp29 hours ago

How do you know that? You are creating this false sense of expectations and hype yourself.

I am going to argue contrary. If AI increases productivity 2x, it opens up as much new usecases that previously didn't seem worthy to do for its cost. So overall there will just be more work.

+1
JimDabell7 hours ago
pawelmurias6 hours ago

[flagged]

binkHN5 hours ago

I wouldn't call it genius tab completion. Unfortunately, more than half of the time that the "genius" produces the code, I'm wasting my time reviewing code that is incorrect.

tguinot6 hours ago

I'm sorry but I don't understand how people say LLMs are simply "tab completion".

They allow me to do much more than that thanks to all the knowledge they contain.

For instance, yesterday I wanted to write a tool that transfers any large file that is still being appended to to multiple remote hosts, with a fast throughput.

By asking Claude for help I obtained exactly what I want in under two hours.

I'm no C/C++ expert yet I have now a functional program using libtorrent and libfuse.

By using libfuse my program creates a continuously growing list of virtual files (chunks of the big file).

A torrent is created to transfer the chunks to remote hosts.

Each chunk is added to the torrent as it appears on the file system thanks to the BEP46 mutable torrent feature in libtorrent.

On each receving host, the program rebuilds the large file by appending new chunks as soon as they are downloaded through the torrent.

Now I can transfer a 25GB file (and growing) to 15 hosts as it is being written too.

Before LLM this would have taken me at least four days as I did not know those libraries.

LLMs aren't just parrots or tab completers, they actually contain a lot of useful knowledge and they're very good at explaining it clearly.

qwertox6 hours ago

> By asking Claude for help I obtained exactly what I want in under two hours.

Did you use it in your editor or via the chat interface in the browser? Because they are two different approaches, and the one in the editor is mostly a (pretty awesome) tab completion.

When I tell an LLM to "create a script which does ..." I won't be doing this in the editor, even if copilot does have the chat interface. I'll be doing this in the browser because there I have a proper chat topic to which I can get back later, or review it.

+1
tguinot5 hours ago
bitcharmer2 hours ago

> thanks to all the knowledge they contain

This is what's problematic with modern "AI". Most people inexperienced with it, like the parent commenter will uncritically assume these LLMs poses "knowledge". This I find the most dangerous and prevalent assumption. Most people are oblivious to the fact how bad LLMs are.

tguinot2 hours ago

I know excatly how bad the output they give is, because I ask for output that I can understand, debug and improve.

People misusing tools don't make tools useless or bad. Especially since LLMs designers never claimed the compressed information inside models is spotless or 100% accurate, or based on logical reasoning.

Any serious engineer with a modicum of knowledge about neural networks knows what can or can't be done with the output.

lupire5 hours ago

That's fine for your quick hack that is probably a reimplementation of an existing program you can't find.

But it's not a production quality implementation of new need.

pizzafeelsright1 hour ago

I am of the strong opinion most problems were solved 20-40 years ago and that most code written today is reimplementation using different languages.

I have shipped production code using LLMs in languages I did not study approved by seasoned SWE's is evidence that an acceleration is happening.

rty323 hours ago

True, but hey, even if it's not production code, it may be an ad-hoc thing that never gets push to production, it may be code reviewed by C++ experts and improved to production quality. At very least, someone saved four days with it, and could use the time for something, maybe something they are expert at. Isn't that still good?

tguinot5 hours ago

It's a knowledge base that can explain the knowledge it returns when you ask, how is that not useful in a professional environment for production code?

I mean if you assume all devs are script kiddies who simply copy paste what they find on google (or ChatGPT without asking for explanations) then yeah it's never gonna be useful in a prod setting.

Also you're very wrong to believe every technical need or combination of libraries has already been implemented in open source before.

znpy6 hours ago

That sounds like a great idea, are you going to open source that?

+1
tguinot5 hours ago
OnionBlender12 hours ago

Do people find these AI auto complete things helpful? I was trying the XCode one and it kept suggesting API calls that don't exist. I spent more time fixing its errors than I would have spent typing the correct API call.

_kidlike9 hours ago

I really really dislike the ones that get in your way. Like I start typing something and it injects random stuff (yes in the auto-complete colors). I have a similar feeling to when you hear your voice back in a phone: completely disabling your thought process.

In IntelliJ thankfully you can disable that part of the AI, and keep the part that you trigger it when you want something from it.

frereubu7 hours ago

> I have a similar feeling to when you hear your voice back in a phone: completely disabling your thought process.

This is a fantastic description of how it disturbs my coding practice which I hadn't been able to put into words. It's like someone is constantly interrupting you with small suggestions whether you want them or not.

gtirloni6 hours ago

This is it. I have a picture in my mind and then it puts 10 lines of code in front of me and my brain can't ignore. When I'm done reviewing that, it's already tainted my idea.

andyjohnson07 hours ago

Visual Studio "intellisense" has always been pretty good for me. Seemed to make good guesses about my intentions without doing anything wild. It seemed to use ad hoc rules and patterns, but it worked and then got out of the way.

Then it got worse a couple of years ago when they tried some early-stage AI approach. I turned it off. I expect that next time I update VS it'll have got substantially worse and it will have removed the option for me to disable it.

mu5312 hours ago

I find the simpler engines work better.

I want the end of the line completed with focus on context from the working code base, and I don't want an entire 5 line function completed with incomplete requirements.

It is really impressive when it implements a 5 line function correctly, but its like hitting the lottery

ncruces9 hours ago

I particularly like the part where it suggests changes to pasted code.

When I copy and paste code, very often it needs some small changes (like changing all xs to ys and at the same time widths to heights).

It's very good at this, and does the right thing the vast majority of the time.

It's also good with test code. Test code is supposed to be explicit, and not very abstracted (so someone only mildly familiar with a codebase that's looking at a failing test can at least figure the cause). This means it's full of boilerplate, and a smart code generator can help fill that in.

binkHN5 hours ago

Right now my opinion is that they're 60% unhelpful, so I largely agree with you. Sometimes I'll find the AI came up with a somewhat better way of doing something, but the vast majority of the time it does something wrong or does something that appears right, but it's actually wrong and I can only spot it with a somewhat decent code review.

4lb02 hours ago

I use Codeium in NeoVim and yes I find it very helpful. Of course, is not 100% error free, but even when it has errors most of the time it is easier for me to fix them than to write it from scratch.

mcintyre199410 hours ago

I like Cursor, it seems very good at keeping its autocomplete within my code base. If I use its chat feature and ask it to generate new code that doesn’t work super well. But it’ll almost always autocomplete the right function name as I’m typing, and then infer the correct parameters to pass in if they’re variables and if the function is in my codebase rather than a library. It’s also unsurprisingly really good at pattern recognition, so if you’re adding to an enum or something it’ll autocomplete that sensibly too.

I think it’d be more useful if it was clipboard aware though. Sometimes I’ll copy a type, then add a param of that type to a function, and it won’t have the clipboard context to suggest the param I’m trying to add.

qeternity8 hours ago

I really like Cursor but the more I use it the more frustrated I get when it ends up in a tight loop of wanting to do something that I do not want to do. There doesn’t seem to be a good way to say “do not do this thing or things like it for the next 5 minutes”.

M4v3R11 hours ago

It probably depends on the tool you use and on the programming language. I use Supermaven autocomplete when writing Typescript and it’s working great, it often feels like it’s reading my mind, suggesting what I would write next myself.

I_AM_A_SMURF11 hours ago

I use the one at G and it's definitely helpful. It's not revolutionary, but it makes writing code less of a headache when I kinda know what that method is called but not quite.

skybrian12 hours ago

I often delete large chunks of it unread if it doesn't do what I expected. It's much like copy and paste; deleting code doesn't take long.

card_zero11 hours ago

So your test is "seems to work"?

+1
skybrian10 hours ago
DecoySalamander3 hours ago

I was evaluating it for a month and caught myself regularly switching to an IDE with non-AI intellisense because I wanted code that actually works.

vbezhenar9 hours ago

I mostly use one-line completes and they are pretty good. Also I really like when Copilot generates boilerplate like

    if err != nil {
      return fmt.Errorf("Cannot open settings: %w", err);
    }
guappa9 hours ago

I suspect that if you work on trivial stuff that has been asked on stackoverflow countless of times they work very nicely.

OnionBlender43 minutes ago

This is what I've been noticing. For C++ and Swift, it makes pretty unhelpful suggestions. For Python, its suggestions are fine.

Swift is especially frustrating because it will hallucinate the method name and/or the argument names (since you often have to specify the argument names when calling a method).

simne6 hours ago

Looks like model is not trained well. From my exp, after make few projects (2 looks enough), oldest XCode managed to give good suggestions in much more than 50% cases.

0points9 hours ago

No, not at all.

"classic" intellisense is reliable, so why introduce random source in the process?

karmasimida9 hours ago

It is useful in our use case.

Realtime tab completion is good at some really mundane things within the current file.

You still need a chat model, like Claude 3.5 to do more explorational things.

sharpy11 hours ago

Often yes. There were times when I was writing unit tests that was me just naming the test case, with 99% of the test code auto generated based on the existing code, and the name.

mdavid6269 hours ago

No, not at all. It’s just the hype. It doesn’t replace engineering.

saagarjha9 hours ago

The one Xcode has is particularly bad, unfortunately.

myworkinisgood8 hours ago

Copilot is very good.

red_admiral6 hours ago

A few years ago we called that IntelliSense, right?

I remember many years ago as a Java developer, Netbeans could do such things as complete `psvm` to "public static void main() {...}", or if you had a field "private String name;" you could press some key combination and it would generate you the getter and setter, complete with javadoc which was mandatory at that place because apparently you need "Returns the name.\n @return The name." on a method called getName() in case you wondered what it was for.

rty323 hours ago

I think most people define "Intellisense" as "IDE suggestions based on static anaysis results". Sometimes it blends a bit of heuristics/usage statistics as added feature depending on the tool. They are mostly deterministic, based on actual AST of your code, and never hallucinates. They may not be helpful but can never be wrong.

On the other hand, LLMs are completely different -- based on machine learning and everything is random and about statistics. It depends on training data and context. It is more useful but make a ton of mistakes.

_heimdall4 hours ago

Yes, Copilot and other LLM coding tools are just a (much) better version of IntelliSense.

snowe20104 hours ago

Much worse imo.

_heimdall4 hours ago

That could be too. I don't use LLMs so I'm just giving it the benefit of the doubt based on other commentors here.

skydhash4 hours ago

Most jetbrains IDEs come with those snippets and if you’re using IDEA, the code will be 50%+ generated by the IDE.

davedx8 hours ago

I'm working on a CRM with a flexible data model, and ChatGPT has written most of the code. I don't use the IDE integrations because I find them too "low level" - I work with GPT more in a sort of "pair programming" session: I give it high level, focused tasks with bits of low level detail if necessary; I paste code back and forth; and I let it develop new features or do refactorings.

This workflow is not perfect but I am definitely building out all the core features way faster than if I wrote the code myself, and the code is in quite a good state. Quite often I do some bits of cleanup, refactorings, making sure typings are complete myself, then update ChatGPT with what the code now looks like.

I think what people miss is there are dozens of different ways to apply AI to your day-to-day as a software engineer. It also helps with thinking things through, architecture, describing best practices.

_heimdall4 hours ago

Its interesting to see these LLM tools turning developers into no-code customers. Where tools like visual site builders allowed those without coding experience to code a webpage, LLMs are letting those with coding experience to avoid the step of coding.

There's not even anything wrong with that, don't take my comment the wrong way. It is an interesting question of what happens at scale though. We could easily find ourselves in a spot where very few people know how to code and most producing code don't actually know how it works and couldn't find or fix a bug if they needed to. It also means LLMs would be stuck with today's code for a training set until it can invent its own coding paradigms and languages, at which point we're all left in the dust trusting it to work right.

sampo5 hours ago

> I paste code back and forth

There is this tool Aider. Takes your prompt, adds code files (sometimes not all of your code files but files it figures relevant) and prepares one long prompt, sends it to an LLM, receives the response, and makes a git commit based on the response. If you rather review git commits, it can save you the back-and-forth copy-pasting. https://aider.chat/

littlestymaar8 hours ago

I share your sentiment, I've written three apps where I've used language models extensively (a different one for each: ChatGPT, Mixtral and Llama-70B) and while I agree that they where immensely helpful in terms of velocity, there are a bunch of caveats:

- it only works well when you write code from scratch, context length is too short to be really helpful for working on existing codebase.

- the output code is pretty much always broken in some way, and you need to be accustomed to doing code reviews to use them effectively. If you trust the output and had to debug it later it would be a painfully slow process.

Also, I didn't really noticed a significant difference in code quality, even the best model (GPT-4) write code that doesn't work, and I find it much more efficient to use open models on Groq due to the really fast inference. Looking at ChatGPT slowly typing is really annoying (I didn't test o1 and I have no interest in doing so because of its very low throughput).

davedx8 hours ago

> context length is too short to be really helpful for working on existing codebase.

This is kind of true, my approach is I spend a fairly large amount of time copy-pasting code from relevant modules back and forth into ChatGPT so it has enough context to make the correct changes. Most changes I need to make don't need more than 2-3 modules though.

> the output code is pretty much always broken in some way, and you need to be accustomed to doing code reviews to use them effectively.

I think this really depends on what you're building. Making a CRM is a very well trodden path so I think that helps? But even when it came to asking ChatGPT to design and implement a flexible data model it did a very good job. Most of the code it's written has worked well. I'd say maybe 60-70% of the code it writes I don't have to touch at all.

The slow typing is definitely a hindrance! Sometimes when it's a big change I lose focus and alt-tab away, like I used to do when building large C++ codebases or waiting for big test suites to run. So that aspect saps productivity. Conversely though I don't want to use a faster model that might give me inferior results.

+1
chrisjj7 hours ago
littlestymaar7 hours ago

> approach is I spend a fairly large amount of time copy-pasting code from relevant modules back and forth into ChatGPT

It can work, but what a terrible developer experience.

> I'd say maybe 60-70% of the code it writes I don't have to touch at all

I used to to write web apps so the ratio was even higher I'd say (maybe 80/90% of the code didn't need any modification) but the app itself wouldn't work at all if I didn't make those 10% changes. And you really need to read 100% of the code because you won't know upfront where those 10% will be.

> The slow typing is definitely a hindrance! Sometimes when it's a big change I lose focus and alt-tab away, like I used to do when building large C++ codebases or waiting for big test suites to run.

Yeah exactly, it's xkcd 303 but with “IA processing the response” instead of “compiling”. Having instant response was a game changer for me in terms of focus hence productivity.

> I don't want to use a faster model that might give me inferior results

As I said earlier, I didn't really feel the difference in quality so the switch was without drawbacks.

creesch6 hours ago

> Also, I didn't really noticed a significant difference in code quality, even the best model (GPT-4) write code that doesn't work,

Interesting, personally I have noticed a difference. Mostly in how well the models pick up small details and context. Although I do have to agree that the open Llama models are generally fairly serviceable.

Recently I have tended to lean towards Claude Sonnet 3.5 as it seems slightly better. Although that does differ per language as well.

As far as them being slow, I haven't really noticed a difference. I use them mostly through the API with open webui and the answers come quick enough.

jszymborski25 minutes ago

Sounds like the JetBrains new local AI autocomplete. If it's anything like that, it's honestly my ideal application of generative deep learning.

alxjrvs13 hours ago

In my day to day, this still remains the main way I interact with AI coding tools.

I regularly describe it as "The best snippet tool I've ever used (because it plays horseshoes)".

tomcam12 hours ago

Horseshoes? As in “close enough”?

ttul11 hours ago

Or, as in, “Ouch, man! You hit my foot!”

+1
goykasi11 hours ago
simplyluke10 hours ago

This is exactly how I’ve used copilot for over a year now. It’s really helpful! Especially with repetitive code. Certainly worth what my employer pays for it.

The general public has a very different idea of that though and I frequently meet people very surprised the entire profession hasn’t been automated yet based on headlines like this.

arisAlexis10 hours ago

Because you are using it like that doesn't mean that it can't be used for the whole stack and on its own and the public including laymen such as the Nvidia CEO and Sam think that yes, we (I'm a dev) will be replaced. Plan accordingly my friend.

robertlagrant9 hours ago

> Because you are using it like that doesn't mean that it can't be used for the whole stack

Well no, but we have no evidence it can be used for the whole stack, whatever that means.

+3
arisAlexis8 hours ago
simplyluke2 hours ago

I was a hacker before the entire stack I work in was common or released, and I’ll be one when all our tools change again in the future. I have family who programmed with punch cards.

But I doubt the predictions from men whose net worth depends on the hype they foment.

arisAlexis12 minutes ago

It's not tools. It's intelligent agents capable of human output.

arisAlexis5 hours ago

The laymen was ironic of course..

awkward1 hour ago

Stuff that works well with AI seems to correlate pretty well with high churn changes. I've had good luck using AI to port large numbers of features from version A to version B, or getting code with a a lot of dependencies under mocked unit tests.

It's easy to see that adding up quickly to represent large percentages of the codebase by line, but it's not feature development or solving hard problems.

blindhippo1 hour ago

Same things I use it for as well - crap like "update this class to use JDK21" or "re-implement this client to use AWS SDKv2" or whatever.

And it works maybe... 80% of the way and I spend all my time fixing the remaining 20%. Anecdotally I don't "feel" like this really accelerates me or reduces the time it would take me to do the change if I just implemented the translation manually.

hgomersall9 hours ago

Before I go and rip out and replace my development workflow, is it notably better than auto complete suggestions from CoC in neovim (with say, rust-analyzer)? I'm generally pretty impressed how quickly it gives me the right function call or whatever, or it's the one of the top few.

Leherenn5 hours ago

It's more than choosing the right function call, it goes further than that. If your code has patterns, it recognises and suggests them.

For instance, one I find very useful is that we have this pattern of checking the result of a function call, logging the error and returning, or whatever. So now, every time you have `result = foo()`, it will auto suggest `if (!result) log_error...` with a generally very good error message.

Very basic, but damn convenient. The more patterns you use, the more helpful it becomes.

bambax2 hours ago

Thank you for this comment. So the code written in this manner isn't really "created by AI"; AI is just a nice additional feature of an editor.

I wonder if the enormous hype around AI is a good or bad thing; it's obviously both but will the good win out the bad, or will the disappointment eventually be so overwhelming as to extinguish any enthusiasm.

insane_dreamer58 minutes ago

also useful for writing unit tests, comments, descriptions, so if you count all of that as code, together with boilerplate stuff, then yeah, it could add up to 25%.

jb19913 hours ago

Long before this current AI hype cycle, we’ve had excellent code completion in editors for decades. So I guess by that definition, we’ve all been writing AI assisted code for a very long time.

fhd23 hours ago

I'd say so, and it's a bit misleading to leave that out. Code generation is almost as old as computing. So far, most of it happened to be deterministic.

karmasimida9 hours ago

I can totally see it.

It is actually a testament that, part of Google's code are ... kinda formulaic to some degree. Prior to the LLM take over, we already heard praise how Google's code search works wonder in helping its engineer writing code, LLM just brought that experience to next level.

nycdatasci13 hours ago

This is a great anecdote. SOTA models will not provide “engineering” per se, but they will easily double productivity of a product manager that is exploring new product ideas or technologies. They are much more than intelligent auto-complete. I have done more with side projects in the last year than I did in the preceding decade.

llm_trw12 hours ago

One of my friends put it best: I just did a months worth of experimentation in two hours.

Sateeshm12 hours ago

I find this hard to believe. Can someone give me an example of something that takes months that AI can correctly do in hours?

+1
jvanveen12 hours ago
ghostpepper13 hours ago

Does it make you 25% more productive?

vundercind12 hours ago

Between the fraction of my time I spend actually writing code, and how much of the typing time I’m using to think anyway, I dunno how much of an increase in my overall productivity could realistically be achieved by something that just helped me type the code in faster. Probably not 25% no matter how fast it made that part. 5% is maybe possible, for something that made that part like 2-3x faster, but much more than that and it’d run up against a wall and stop speeding things up.

imchillyb12 hours ago

I imagine that those who cherished the written word thought similar thoughts when the printing press was invented, when the typewriter was invented, and before excel took over bookkeeping.

My productivity isn't so much enhanced. It's only 1%... 2%... 5%... globally, for each employee.

Have you ever dabbled with, mucked around in, a command line? Autocomplete functions there save millions of man-hour-typing-units per year. Something to think about.

A single employee, in a single task, for a single location may not equal much gained productivity, but companies now think on much larger scales than a single office location.

+1
moron4hire11 hours ago
rustcleaner12 hours ago

If 25% of code was AI-written, wouldn't it be a 33[.333...]% increase in productivity?

vundercind1 hour ago

Who spends 100% of their time actually typing code?

It’s probably closer to 10% than 100%, especially at big companies.

One thing I would love to see is reports of benefits from various tools coming with one’s typing ability in WPM. I’d also like to see that on posts where people express a preference for “a quick call” or stopping by your desk rather than posting what they want in chat. I have some hypotheses I’d like to test out.

PeterStuer8 hours ago

It is not a direct correlation. I might write 80% of the lines of code in a week, then spend the next 6 months on the remaining 20%. If the AI was mostly helpfull in that first week, overall productivity gain would be very low.

card_zero10 hours ago

Not if there was also an 8.333̅% increase in slacking off.

Wait, no. That should be based on how much slacking off Google employees do ordinarily, an unknown quantity.

saagarjha9 hours ago

You can just check Memegen traffic to figure that one out.

aforty3 hours ago

We have this at our company too. I guess it’s useful but doesn’t really have a whole lot of time.

markstos3 hours ago

Which editor is Google's AI code completion integrated with? VS Code?

hoveringhen50 minutes ago

Yeah

afro8812 hours ago

So more or less on par with continue.dev using a local starcoder2:3b model

josh_carterPDX57 minutes ago

My first reaction to the title was, "That explains why things are broken." but this explanation makes so much sense. Thanks for clarifying.

But yeah, I wish the new version of Chrome worked better. ¯\_(ツ)_/¯

hackerknew13 hours ago

I wondered if this the real context. i.e. They are just referring to code-completion as AI-generated code. But, the article seems like it is referring to more than that?

ImaCake13 hours ago

This autocomplete seems about on par with github copilot. Do you also get options for prompting it on specific chunks of code and performing specific actions such as writing docs or editing existing code? All things that come standard with gh copilot now.

heresie-dabord6 hours ago

"Our overhyped Autocomplete Implementation (A.I.) is completing 25% of our lines of code so well that we need to fund nuclear reactors to power the server farms."

theodric2 hours ago

I wonder how this works with IP rights in the USA. Like, is `function getAc` eligible for copyright protection, but `tionHandler()` isn't? After all, [1]

[1] https://www.reuters.com/legal/ai-generated-art-cannot-receiv...

cryptica10 hours ago

This is my experience as well. LLMs are great to boost productivity, especially in the hands of senior engineers who have a deep understanding of what they're doing because they know what questions to ask, they know when it's safe to use AI-generated code and they know what issues to look for.

In the hands of a junior, AI can create a false sense of confidence and it acts as a technical debt and security flaw multiplier.

We should bring back the title "Software engineer" instead of "Software developer." Many people from other engineering professions look down on software engineers as "Not real engineers" but that's because they have the same perspective on coding as typical management types have. They think all code is equal, it's unavoidable spaghetti. They think software design and architecture doesn't matter.

The problems a software engineer faces when building a software system are the same kinds of problems that a mechanical or electrical engineer faces when building any engine or system. It's about weighing up trade-offs and making a large number of nuanced technical decisions to ultimately meet operational requirements in the most efficient, cost-effective way possible.

grecy2 hours ago

I'm confused, I've been doing similar tab completion for function names in eclipse since about 2003...

znpy6 hours ago

> If I'm writing "function getAc..." it's smart enough to complete to "function getActionHandler()", and maybe suggest the correct arguments and a decent jsdoc comment.

I really mean no offense, but your example doesn't sound much different from what old IDEs (say, Netbeans) used to do 15 years ago.

I could design a Swing ui and it would generate the code and if I wanted to override a method it would generate a decent boilerplate boilerplate (a getter, like in your example) along with usual comments and definitely correct parameters list (with correct types).

Is this "AI Code" thing something that appears new because at some point we abandoned IDEs with very strong intellisense (etc) ?

hoveringhen40 minutes ago

This video is a pretty good one on how it works in practice: https://storage.googleapis.com/gweb-research2023-media/media...

napierzaza13 hours ago

[dead]

Galatians4_1613 hours ago

Kerry said hi

ntulpule2 days ago

Hi, I lead the teams responsible for our internal developer tools, including AI features. We work very closely with Google DeepMind to adapt Gemini models for Google-scale coding and other Software Engineering usecases. Google has a unique, massive monorepo which poses a lot of fun challenges when it comes to deploying AI capabilities at scale.

1. We take a lot of care to make sure the AI recommendations are safe and have a high quality bar (regular monitoring, code provenance tracking, adversarial testing, and more).

2. We also do regular A/B tests and randomized control trials to ensure these features are improving SWE productivity and throughput.

3. We see similar efficiencies across all programming languages and frameworks used internally at Google and engineers across all tenure and experience cohorts show similar gain in productivity.

You can read more on our approach here:

https://research.google/blog/ai-in-software-engineering-at-g...

hitradostava1 day ago

I'm continually surprised by the amount of negativity that accompanies these sort of statements. The direction of travel is very clear - LLM based systems will be writing more and more code at all companies.

I don't think this is a bad thing - if this can be accompanied by an increase in software quality, which is possible. Right now its very hit and miss and everyone has examples of LLMs producing buggy or ridiculous code. But once the tooling improves to:

1. align produced code better to existing patterns and architecture 2. fix the feedback loop - with TDD, other LLM agents reviewing code, feeding in compile errors, letting other LLM agents interact with the produced code, etc.

Then we will definitely start seeing more and more code produced by LLMs. Don't look at the state of the art not, look at the direction of travel.

latexr1 day ago

> if this can be accompanied by an increase in software quality

That’s a huge “if”, and by your own admission not what’s happening now.

> other LLM agents reviewing code, feeding in compile errors, letting other LLM agents interact with the produced code, etc.

What a stupid future. Machines which make errors being “corrected” by machines which make errors in a death spiral. An unbelievable waste of figurative and literal energy.

> Then we will definitely start seeing more and more code produced by LLMs.

We’re already there. And there’s a lot of bad code being pumped out. Which will in turn be fed back to the LLMs.

> Don't look at the state of the art not, look at the direction of travel.

That’s what leads to the eternal “in five years” which eventually sinks everyone’s trust.

danielmarkbruce21 hours ago

> What a stupid future. Machines which make errors being “corrected” by machines which make errors in a death spiral. An unbelievable waste of figurative and literal energy.

Humans are machines which make errors. Somehow, we got to the moon. The suggestion that errors just mindlessly compound and that there is no way around it, is what's stupid.

+2
latexr18 hours ago
+1
kelnos17 hours ago
+1
openrisk7 hours ago
+3
reverius4220 hours ago
nuancebydefault20 hours ago

Exactly my thought. Humans can correct humans. Machines can correct, or at least point to failures in the product of, machines.

goatlover10 hours ago

Machines are intelligently designed for a purpose. Humans are born and grow up, have social lives, a moral status and are conscious, and are ultimately the product of a long line of mindless evolution that has no goals. Biology is not design. It's way messier.

spockz12 hours ago

My main gripe with this form of code generation is that is primarily used to generate “leaf” code. Code that will not be further adjusted or refactored into the right abstractions.

It is now very easy to sprinkle in regexes to validate user input , like email addresses, on every controller instead of using a central lib/utility for that.

In the hands of a skilled engineer it is a good tool. But for the rest it mainly serves to output more garbage at a higher rate.

cdchn11 hours ago

>It is now very easy to sprinkle in regexes to validate user input , like email addresses, on every controller instead of using a central lib/utility for that.

Some people are touting this as a major feature. "I don't have to pull in some dependency for a minor function - I can just have AI write that simple function for me." I, personally, don't see this as a net positive.

spockz11 hours ago

Yes, I have heard similar arguments before. It could be an argument for including the functionality in the standard lib for the language. There can be a long debate about dependencies, and then there is still the benefit of being able to vendor and prune them.

The way it is now just leads to bloat and cruft.

paradox24220 hours ago

I don't see how this is sustainable. We have essentially eaten the seed corn. These current LLMs have been trained by an enormous corpus of mostly human-generated technical knowledge from sources which we already know to be currently being polluted by AI-generated slop. We also have preliminary research into how poorly these models do when training on data generated by other LLMs. Sure, it can coast off of that initial training set for maybe 5 or more years, but where will the next giant set of unpolluted training data come from? I just don't see it, unless we get something better than LLMs which is closer to AGI or an entire industry is created to explicitly create curated training data to be fed to future models.

_DeadFred_20 hours ago

These tools also require the developer class to that they are intended to replace to continue to do what they currently do (create the knowledge source to train the AI on). It's not like the AIs are going to be creating the accessible knowledge bases to train AIs on, especially for new language extensions/libraries/etc. This is a one and f'd development. It will give a one time gain and then companies will be shocked when it falls apart and there's no developers trained up (because they all had to switch careers) to replace them. Unless Google's expectation is that all languages/development/libraries will just be static going forward.

+4
layer818 hours ago
0points9 hours ago

Yea, I'm thinking along the same lines.

The companies valuing the expensive talent currently working on Google will be the winner.

Google and others are betting big right now, but I feel the winner might be those who watches how it unfolds first.

brainwad20 hours ago

The LLM codegen at Google isn't unsupervised. It's integrated into the IDE as both autocomplete and prompt-based assistant, so you get a lot of feedback from a) what suggestions the human accepts and b) how they fix the suggestion when it's not perfect. So future iterations of the model won't be trained on LLM output, but on a mixture of human written code and human-corrected LLM output.

As a dev, I like it. It speeds up writing easy but tedious code. It's just a bit smarter version of the refactoring tools already common in IDEs...

+1
kelnos17 hours ago
loki-ai13 hours ago

maybe most of the code in the future will be very different from what we’re used to. For instance, AI image processing/computer vision algorithms are being adopted very quickly given the best ones are now mostly transformers networks.

lelanthran4 hours ago

> Don't look at the state of the art not, look at the direction of travel.

That's what people are doing. The direction of travel over the most recent few (6-12) months is mostly flat.

The direction of travel when first introduced was a very steep line going from bottom-left to top-right.

We are not there anymore.

philipwhiuk15 hours ago

> The direction of travel is very clear

And if we get 9 women we can produce a baby in a single month.

There's no guarantee such progression will continue. Indeed, there's much more evidence it is coming to a a halt.

Towaway699 hours ago

It might also be an example of 80/20 - we're just entering the 20% of features that take 80% of the time & effort.

It might be possible but will shareholders/investors foot the bill for the 80% that they still have to pay.

farseer10 hours ago

Its not even been 2 years, and you think things are coming to a halt?

+1
0points9 hours ago
simianparrot10 hours ago

I know for a fact they are because rate _and_ quality of improvement is diminishing exponentially. I keep a close eye on this field as part of my job.

dogleash1 hour ago

> I'm continually surprised by the amount of negativity that accompanies these sort of statements.

I'm excited about the possibilities and I still recoil at the refined marketer prose.

olalonde10 hours ago

> I'm continually surprised by the amount of negativity

Maybe I'm just old, but to me, LLMs feel like magic. A decade ago, anyone predicting their future capabilities would have been laughed at.

Towaway699 hours ago

Magic Makes Money - the more magical something seems, the more people are willing to pay for that something.

The discussion here seems to bare this out: CEO claims AI is magical, here the truth becomes that it’s just an auto-complete engine.

guappa9 hours ago

Nah, you just were not up to speed with the current research. Which is completely normal. Now marketing departments are on the job.

davedx8 hours ago

Transformers were proposed in 2017. A decade ago none of this was predictable.

mmmpetrichor11 hours ago

That's the hype isn't it. The direction of travel hasn't been proven to be more than a surface level yet.

0points9 hours ago

> LLM based systems will be writing more and more code at all companies.

At Google, today, for sure.

I do believe we still are not across the road on this one.

> if this can be accompanied by an increase in software quality, which is possible. Right now its very hit and miss

So, is it really a smart move of Google to enforce this today, before quality have increased? Or did this set off their path to losing market shares because their software quality will deteriorate further over the next couple years?

From the outside it just seems Google and others have no choice, they must walk this path or lose market valuation.

randomNumber719 hours ago

Because there seems to be a fundamental misunderstanding producing a lot of nonsense.

Of course LLMs are a fantastic tool to improve productivity, but current LLM's cannot produce anything novel. They can only reproduce what they have seen.

visarga12 hours ago

But they assist developers and collect novel coding experience from their projects all the time. Each application of LLM creates feedback to the AI code - the human might leave it as is, slightly change it, or refuse it.

baxtr13 hours ago

I think that at least partially the negativity is due to the tech bros hyping AI just like they hyped crypto.

fallingknife14 hours ago

I'm not really seeing this direction of travel. I hear a lot of claims, but they are always 3rd person. I don't know or work with any engineers who rely heavily on these tools for productivity. I don't even see any convincing videos on Youtube. Just show me on engineer sitting down with theses tools for a couple hours and writing a feature that would normally take a couple of days. I'll believe it when I see it.

Roark6610 hours ago

Well, I rely on it a lot, but not in the IDE, I copy/paste my code and prompts between the ide and LLM. By now I have a library of prompts in each project I can tweak that I can just reuse. It makes me 25% up to 50% faster. Does this mean every project t is done in 50/75% of the time? No, the actual completion time is maybe 10% faster, but i do get a lot more time to spend on thinking about the overall design instead of writing boilerplate and reading reference documents.

Why no youtube videos thought? Well, most dev you tubers are actual devs that cultivate an image of "I'm faster than LLM, I never re-read library references, I memorise them on first read" and do on. If they then show you a video how they forgot the syntax for this or that maven plugin config and how LLM fills it in 10s instead of a 5min Google search that makes them look less capable on their own. Why would they do that?

skydhash4 hours ago

Why don’t you read reference documents? The thing with bite-sized information is that is never gives you a coherent global view of the space. It’s like exploring a territory by crawling instead of using a map.

fuzztester12 hours ago

you said it, bro.

reverius422 days ago

To me the most interesting part of this is the claim that you can accurately and meaningfully measure software engineering productivity.

zac23or56 minutes ago

I knew a superstar developer who worked on reports in an SQL tool. In the company metrics, the developer scored 420 points per month, the second developer scored 60 points. “Please learn how to score more points from the leader”, the boss would say.

The superstar developer’s secret… he would send blank reports to clients (who would only realize it days later, and someone else would end up redoing the report), and he would score many more points without doing anything. I’ve seen this happen a lot in many different companies. As a friend of mine used to say, “it’s very rare, but it happens all the time.”

I have no doubt that AI can help developers, but I don’t trust the metrics of the CEO or people who work on AI, because they are too involved in the subject.

ozim1 day ago

You can - but not on the level of a single developer and you cannot use those measures to manage productivity of a specific dev.

For teams you can measure meaningful outcomes and improve team metrics.

You shouldn’t really compare teams but it also is possible if you know what teams are doing.

If you are some disconnected manager that thinks he can make decisions or improvements reducing things to single numbers - yeah that’s not possible.

deely31 day ago

> For teams you can measure meaningful outcomes and improve team metrics.

How? Which metrics?

+1
anthonyskipper21 hours ago
+2
neaanopri14 hours ago
+3
ozim1 day ago
UncleMeat1 day ago

At scale you can do this in a bunch of interesting ways. For example, you could measure "amount of time between opening a crash log and writing the first character of a new change" across 10,000s of engineers. Yes, each individual data point is highly messy. Alice might start coding as a means of investigation. Bob might like to think about the crash over dinner. Carol might get a really hard bug while David gets a really easy one. But at scale you can see how changes in the tools change this metric.

None of this works to evaluate individuals or even teams. But it can be effective at evaluating tools.

fwip19 hours ago

There's lots of stuff you can measure. It's not clear whether any of it is correlated with productivity.

To use your example, a user with an LLM might say "LLM please fix this" as a first line of action, drastically improving this metric, even if it ruins your overall productivity.

valval1 day ago

You can come up with measures for it and then watch them, that’s for sure.

lr19701 day ago

when metric becomes the target it ceases to be a good metric. when discovered how it works developers will type the first character immediately after opening the log.

edit: typo

joshuamorton19 hours ago

Only if the developer is being judged on the thing. If the tool is being judged on the thing, it's much less relevant.

That is, I, personally, am not measured on how much AI generated code I create, and while the number is non-zero, I can't tell you what it is because I don't care and don't have any incentive to care. And I'm someone who is personally fairly bearish on the value of LLM-based codegen/autocomplete.

LinuxBender1 day ago

Is AI ready to crawl through all open source and find / fix all the potential security bugs or all bugs for that matter? If so will that become a commercial service or a free service?

Will AI be able to detect bugs and back doors that require multiple pieces of code working together rather than being in a single piece of code? Humans have a hard time with this.

- Hypothetical Example: Authentication bugs in sshd that requires a flaw in systemd which then requires a flaw in udev or nss or PAM or some underlying library ... but looking at each individual library or daemon there are no bugs that a professional penetration testing organization such as the NCC group or Google's Project Zero would find. In other words, will AI soon be able to find more complex bugs in a year than Tavis has found in his career and will they start to compete with one another and start finding all the state sponsored complex bugs and then ultimately be able to create a map that suggests a common set of developers that may need to be notified? Will there be a table that logs where AI found things that professional human penetration testers could not?

0points8 hours ago

No, that would require AGI. Actual reasoning.

Adversaries are already detecting issues tho, using proven means such as code review and fuzzing.

Google project zero consists of a team of rock star hackers. I don't see LLM even replacing junior devs right now.

paradox24220 hours ago

Seems like there is more gain on the adversary side of this equation. Think nation-states like North Korea or China, and commercial entities like Pegasus Group.

saagarjha9 hours ago

FWIW: NSO is the group, Pegasus is their product

AnimalMuppet20 hours ago

Google's AI would have the advantage of the source code. The adversaries would not. (At least, not without hacking Google's code repository, which isn't impossible...)

nycdatasci12 hours ago

You mention safety as #1, but my impression is that Google has taken a uniquely primitive approach to safety with many of their models. Instead of influencing the weights of the core model, they check core model outputs with a tiny and much less competent “safety model”. This approach leads to things like a text-to-image model that refuses to output images when a user asks to generate “a picture of a child playing hopscotch in front of their school, shot with a Sony A1 at 200 mm, f2.8”. Gemini has similar issue: it will stop mid-sentence, erase its entire response and then claim that something is likely offensive and it can’t continue.

The whole paradigm should change. If you are indeed responsible for developer tools, I would hope that you’re activity leveraging Claude 3.5 Sonnet and o1-preview.

ActionHank2 hours ago

Would you say that the efficiency gain is less than, equal to, or greater than the cost?

It's always felt like having AI in the cloud for better autocomplete is a lot for a small gain.

bcherny15 hours ago

How are you measuring productivity? And is the effect you see in A/B tests statistically significant? Both of these were challenging to do at Meta, even with many thousands of engineers —- curious what worked for you.

assanineass18 hours ago

Was this comment cleared by comms

bogwog20 hours ago

Is any of the AI generated code being committed to Google's open source repos, or is it only being used for private/internal stuff?

hshshshshsh21 hours ago

Seems like everything is working out without any issues. Shouldn't you be a bit suspicious?

fhdsgbbcaA1 day ago

I’ve been thinking a lot lately about how an LLM trained in really high quality code would perform.

I’m far from impressed with the output of GPT/Claude, all they’ve done is weight against stack overflow - which is still low quality code relative to Google.

What is probability Google makes this a real product, or is it too likely to autocomplete trade secrets?

mysterydip1 day ago

I assume the amount of monitoring effort is less than the amount of effort that would be required to replicate the AI generated code by humans, but do you have numbers on what that ROI looks like? Is it more like 10% or 200%?

wslh20 hours ago

As someone working in cybersecurity and actively researching vulnerability scanning in codebases (including with LLMs), I’m struggling to understand what you mean by “safe.” If you’re referring to detecting security vulnerabilities, then you’re either working on a confidential project with unpublished methods, or your approach is likely on par with the current state of the art, which primarily addresses basic vulnerabilities.

Twirrim20 hours ago

> We work very closely with Google DeepMind to adapt Gemini models for Google-scale coding and other Software Engineering usecases.

Considering how terrible and frequently broken the code that the public facing Gemini produces, I'll have to be honest that that kind of scares me.

Gemini frequently fails at some fairly basic stuff, even in popular languages where it would have had a lot of source material to work from; where other public models (even free ones) sail through.

To give a fun, fairly recent example, here's a prime factorisation algorithm it produce for python:

  # Find the prime factorization of n
  prime_factors = []
  while n > 1:
    p = 2
    while n % p == 0:
      prime_factors.append(p)
      n //= p
    p += 1
  prime_factors.append(n)
Can you spot all the problems?
kgeist20 hours ago

They probably use AI for writing tests, small internal tools/scripts, building generic frontends and quick prototypes/demos/proofs of concept. That could easily be that 25% of the code. And modern LLMs are pretty okayish with that.

gerash20 hours ago

I believe most people use AI to help them quickly figure out how to use a library or an API without having to read all their (often out dated) documentation instead of helping them solve some mathematical challenge

taeric20 hours ago

If the documentation is out of date, such that it doesn't help, this doesn't bode well for the training data of the AI helping it get it right, either?

+4
macintux20 hours ago
delfinom16 hours ago

I've never had an AI not just make up API when it didn't exist, instead of saying "it doesn't exist". Lol

randomNumber719 hours ago

I think that too but google claims something else.

calf19 hours ago

We are sorely lacking a "Make Computer Science a Science" movement, the tech lead's blurb is par for the course, talking about "SWE productivity" with no reference to scientific inquiry and a foundational understanding of safety, correctness, verification, validation of these new LLM technologies.

almostgotcaught17 hours ago

Did you know that Google is a for-profit business and not a university? Did you know that most places where people work on software are the same?

+1
zifpanachr238 hours ago
+1
calf8 hours ago
justinpombrio19 hours ago

> Can you spot all the problems?

You were probably being rhetorical, but there are two problems:

- `p = 2` should be outside the loop

- `prime_factors.append(n)` appends `1` onto the end of the list for no reason

With those two changes I'm pretty sure it's correct.

kenjackson14 hours ago

You don't need to append 'p' in the inner while loop more than once. Maybe instead of an array for keeping the list of prime factors do it in a set.

zeroonetwothree12 hours ago

It’s valid to return the multiplicity of each prime, depending on the goal of this.

rmbyrro17 hours ago

`n` isn't defined

justinpombrio15 hours ago

The implicit context that the poster removed (as you can tell from the indentation) was a function definition:

    def factorize(n):
      ...
      return prime_factors
dangsux18 hours ago

[dead]

ijidak16 hours ago

I'm the first to say that AI will not replace human coders.

But I don't understand this attempt to tell companies/persons that are successfully using AI that no they really aren't.

In my opinion, if they feel they're using AI successfully, the goal should be to learn from that.

I don't understand this need to tell individuals who say they are successfully using AI that, "no you aren't."

It feels like a form of denial.

Like someone saying, "I refuse to accept that this could work for you, no matter what you say."

senko20 hours ago

We collectively deride leetcoding interviews yet ask AI to flawlessly solve leetcode questions.

I bet I'd make more errors on my first try at it.

AnimalMuppet20 hours ago

Writing a prime-number factorization function is hardly "leetcode".

+2
senko19 hours ago
+4
atomic12819 hours ago
gamesetmath1 day ago

[flagged]

pixxel1 day ago

[flagged]

dep_b3 hours ago

A quarter of all new code? Of course. Especially if you include all "smart autocomplete" code.

When dealing with a fermenting pile of technical debt? I expect very little. LLM's don't have application-wide context yet.

AI is definitely revolutionizing our field, but the same people that said that no-code tools and all of the other hype-of-the-decade technologies would make developers jobless are actually the people AI is making jobless.

Generate an opinion piece about how AI is going to make developers jobless, using AI? Less than a minute. And you don't need to maintain that article, once it's published, it's done.

While there's a tsunami of AI-generated almost-there projects coming that need to be moved to a shippable and sellable state. So I'm more afraid about the kind of work I'm going to get while still getting paid handsomely for my skills, than ever being jobless as the only guy that really understands the whole stack from top to bottom.

randomdata3 hours ago

At the end of the day an LLM is just a compiler anyway. The developer isn't going away even if 100% of the code is generated by LLMs, just as the developer didn't go away when we stopped spending our days flipping toggle switches.

dep_b3 hours ago

I'm actually surprised that _the others_ always think that the programmers somehow will make themselves obsolete first? If it gets cheaper to make software, more software will be made, until we reach the point again we're running short on people capable enough to keep it all running.

xen011 minutes ago

I really do wonder who these engineers are, that the current 'AI' tools are able to write so much of their code.

Maybe my situation is unusual; I haven't written all that much code at Google lately, but what I do write is pretty tied to specific details of the program and the AI auto completion is just not that useful. Sometimes it auto completes a method signature correctly, but it never gets the body right (or even particularly close).

And it routinely making up methods or fields on objects I want to use is anti productive.

yangcheng11 hours ago

Having worked at both FAANG companies and startups, I can offer a perspective on AI's coding impact in different environments. At startups, engineers work with new tech stacks, start projects from scratch, and need to ship something quickly. LLMs can wrtie way more code. I've seen ML engineers build React frontends without any previous frontend experience, flutter developers write 100-line SQL queries for data analysis, with LLM 10x productivity for this type of work. At FAANG companies, codebases contain years of business logic, edge cases, and 'not-bugs-but-features.' Engineers know their tech stacks well, and legacy constraints make LLMs less effective, and can generate wrong code that needs to be fixed

davnicwil7 hours ago

It might not quite be there yet, but one key advantage large codebases have that I think LLMs in time will be able to better exploit is the detection of existing patterns - presuming they're consistent - and application to new code doing similar things or to fix bugs in existing code that deviates from the pattern in some way that causes a bug.

It's a different thing to what you're talking about, but it's one way I'd expect to see LLMs contribute a lot to productivity on larger codebases specifically.

mdgrech233 hours ago

large application codebase - consistent - have you worked in the field? I feel like usually there are 3 or 4 patterns from different people/teams at different points in time that spearheaded a particular ideology about how things "should" be done.

devonbleak20 hours ago

It's Go. 25% of the code is just basic error checking and returning nil.

QuercusMax20 hours ago

In Java, 25% of the code is import statements and curly braces

layer818 hours ago

You generally don’t write those by hand though.

I’m pretty sure around 50% of the code I write is already auto-complete, without any AI.

amomchilov17 hours ago

Exactly, you write them with AI

+1
throwaway10638217 hours ago
jansan10 hours ago

Simply strech your definition of AI and voilá, you are writing it with AI.

rwmj5 hours ago

The most important thing is to put out a press release about how half your code is written by AI.

contravariant19 hours ago

In lisp about 50% of the code is just closing parentheses.

harry819 hours ago

Heh, but it can't be that, no reason to think llms can count brackets needing a close any more than they can count words.

int_19h16 hours ago

LLMs can count words (and letters) just fine if you train them to do so.

Consider the fact that GPT-4 can generate valid XML (meaning balanced tags, quotes etc) in base64-encoded form. Without CoT, just direct output.

+1
overhead407518 hours ago
xxs10 hours ago

Over 3 imports from the same package - use an asterisk.

NeoTar19 hours ago

Does auto-code generation count as AI?

remram16 hours ago

Another 60% is auto-generated protobuf/grpc code. Maybe protoc counts as "AI".

GeneralMayhem13 hours ago

Google does not check in protoc-generated code. It's all generated on demand by Blaze/Bazel.

remram2 hours ago

Oh thanks for the info.

On the other hand, that doesn't mean it doesn't count for the purpose of this press release/advertisement...

hiddencost12 hours ago

Go is a very small fraction of the code at Google.

fzysingularity20 hours ago

While I get the MBA-speak of lines-of-code that AI is now able to accomplish, it does make me think about their highly-curated internal codebase that makes them well placed to potentially get to 50% AI-generated code.

One common misconception is that all LLMs are the same. The models are trained the same, but trained on wildly different datasets. Google, and more specifically the Google codebase is arguably one of the most curated, and iterated on datasets in existence. This is a massive lever for Google to train their internal code-gen models, that realistically could easily replace any entry-level or junior developer.

- Code review is another dimension of the process of maintaining a codebase that we can expect huge improvements with LLMs. The highly-curated commentary on existing code / flawed diff / corrected diff that Google possesses give them an opportunity to build a whole set of new internal tools / infra that's extremely tailored to their own coding standard / culture.

bqmjjx0kac19 hours ago

> that realistically could easily replace any entry-level or junior developer.

This is a massive, unsubstantiated leap.

throwaway10638217 hours ago

I’d take pair programming with a junior over a GPT bot any day.

neaanopri14 hours ago

I'd take coding by own damn self over either a junior or a gpt bot

risyachka18 hours ago

The issue is it doesn't really replace junior dev. You become one - as you have to babysit it all the time, check every line of code, and beg it to make it work.

In many cases it is counterproductive

jtbetz227 hours ago

> Google codebase is arguably one of the most curated, and iterated on datasets in existence

I spent 12 years of my career in the Google codebase.

This assertion is technically correct in that google3 has been around for 20 years, and all code gets reviewed, but the implication that Google's codebase is a high-quality training set is not consistent with my experience.

unit14911 hours ago

Philosophically, these models are akin to scholars prior to differentiation during their course of study. Throttling data, depending on one's course of study, and this shifting of the period in history step-by-step. Either it's a tit-for-tat manner of exchange that the junior developer is engaged in, when overseeing every edit that an LLM has modified, or I'd assume that there are in-built methods of garbage collection, that another LLM evaluating a hash function partly identifying a block of tokenized internal code would be responsible for.

morkalork19 hours ago

Is the public gemini code gen LLM trained on their internal repo? I wonder if one could get it to cough up propriety code with the right prompt.

p1esk18 hours ago

I’m curious if Microsoft lets OpenAI train on GH private repos.

happyopossum11 hours ago

> Is the public gemini code gen LLM trained on their internal repo?

Nope

ttul11 hours ago

I wanted a new feature in our customer support console and the dev lead suggested I write a JIRA. I’m the CEO, so this is not my usual thing (and probably should not be). I told Claude what I wanted and pasted in a .js file from the existing project so that it would get a sense of the context. It cranked out a fully functional React component that actually looks quite nice too. Two new API calls were needed, but Claude helpfully told me that. So I pasted the code sample and a screenshot of the HTML output into the JIRA and then got Claude to write me the rest of the JIRA as well.

Everyone knows this was “made by AI” because there’s no way in hell I would ever have the time. These models might not be able to sit there and build an entire project from scratch yet, but if what you need is some help adding the next control panel page, Claude’s got your back on that.

simianparrot10 hours ago

You’re also the CEO so chances are the people looking at that ticket aren’t going to tell you the absolute mess the AI snippet actually is and how pointless it was to include it instead of a simple succinct sentence explaining the requirements.

If you’re not a developer chances are very high the code it produces will look passable but is actually worthless — or worse, it’s misleading and now a dev has to spend more time deciphering the task.

JonChesterfield8 hours ago

> Everyone knows this was “made by AI” because there’s no way in hell I would ever have the time.

Doubtful. A decent fraction of the people reading it will guess that you've wasted your time writing incoherent nonsense in the jira. Engineers don't usually have much insight into what the C suite are doing. It would be a prudent move to spend the couple of seconds to write "something like this AI sketch:" before the copy&paste.

gloflo9 hours ago

> Everyone knows this was “made by AI” because ...

They should know because you told them so.

Having to decipher weird code only to discover it was not written by a human is not nice.

zac23or52 minutes ago

> dev lead suggested I write a JIRA. I’m the CEO, so this is not my usual thing (and probably should not be)

Fascinating point of view.

Taylor_OD20 hours ago

If we are talking about the boilerplate code and autofill syntax code that copilot or any other "AI" will offer me when I start typing... Then sure. Sounds about right.

The other 75% is the stuff you actually have to think about.

This feels like saying linters impact x0% of code. This just feels like an extension of that.

creativenolo18 hours ago

It probably does. But an amazing number of commenters think they are prompting the copy & pasting, and hoping for the best.

Kalabasa14 hours ago

Yep, a lot of headline readers here.

It's just a very advanced autocomplete, completely integrated into the internal codebase and IDE. You can read this on the research blog (maybe if everyone just read the blog).

e.g.

I start typing `var notificationManager`

It would suggest `= (Notification Manager) context.getSystemService(NOTIFICATION_MANAGER);`

If you've done Android then you know how much boilerplate there is to suggest.

I press Ctrl+Enter or something to accept the suggestion.

Voila, more than 50% of that code was written by AI.

> blindly committing AI code

Even before AI, no one blindly accepts autocomplete.

A lot of headline-readers seem to imagine some sort of semi-autonomous or prompt based code generation that writes whole blocks of code to then be blindly accepted by engineers.

skydhash3 hours ago

That makes a while since I’ve done Android, but I’m sure that this variable should be a property and be set as part of the lifecycle. And while Android (and any big project) is full of boilerplate, each line is subtly different or it would have already been abstracted in some base class. And even then, the code completion is already so good in Android Studio that you would have to be a complete junior (in this case, you wouldn’t know that the AI suggestion is good) to complain that writing code is slow. Most time spent is designing code, fixing subtle bugs, and refactoring to clean up the code.

esjeon13 hours ago

> The other 75% is the stuff you actually have to think about.

I’m pretty sure the actual ratio is much lower than that. In other words, LLMs are currently not good enough to remove the majority of chores, even with the state of the art model trained on highly curated dataset.

d_burfoot3 hours ago

I'd be far more impressed if the CEO said "The AI deleted a quarter of our company's code".

ryoshu21 hours ago

Spoken like an MBA who counts lines of code.

pfannkuchen21 hours ago

It’s replaced the 25% previously copy pasted from stack overflow.

rkagerer19 hours ago

This may have been intended as a joke, but it's the only explanation that reconciles the quote for me.

brainwad20 hours ago

The split is roughly 25% AI, 25% typed, 50% pasted.

imaginebit2 days ago

I think he's trying to promote AI, somehow raises questions about thrir code quality among some

dietr1ch2 days ago

I think it just shows how much noise there is in coding. Code gets reviewed anyways (although review quality was going down rapidly the more PMs where added to the team)

Most of the code must be what could be snippets (opening files and handling errors with absl::, and moving data from proto to proto). One thing that doesn't help here, is that when writing for many engineers on different teams to read, spelling out simple code instead of depending on too many abstractions seems to be preferred by most teams.

I guess that LLMs do provide smarter snippets that I don't need to fill out in detail, and when it understands types and whether things compile it gets quite good and "smart" when it comes to write down boilerplate.

drunken_thor16 hours ago

A company that used to be the pinnacle of software development is now just generating code in order to sell their big data models. Horrifying. Devastating.

summerlight17 hours ago

In Google, there is a process called "Large Scale Change" which is primarily meant for trivial/safe but extremely tedious code changes that potentially span over the entire monorepo. Such as foundational API changes, trivial optimization, code style etc etc. This is a perfectly suitable for LLM driven code changes (in fact I'm seeing more and more of LLM generated LSC) and I guess a large fraction of mentioned "AI generated codes" can be actually attributable to this.

bubaumba13 hours ago

yeh, but the main problem is the quality. with algorithm bug can be fixed. with llm it's more complicated. in practice they do some mistakes consistently, and in some cases cannot recover even with assistance. (don't take me wrong, I'm very happy with the results most of the time)

afro8811 hours ago

You just fix the mistakes and keep moving. It's like autocomplete where you still need to fill in the blanks or select a different completion.

saagarjha9 hours ago

Spotting and fixing mistakes in a LSC is no small feat.

0xCAP21 hours ago

People overestimate faang. There are many talents working there, sure, but a lot of garbage gets pumped into their codebases as well.

fuzzfactor7 hours ago

>a lot of garbage gets pumped into their codebases

I would imagine it always has.

>Google CEO says more than a quarter of the company's new code is created by AI

It may very well be starting to become apparent anyway :\

motoxpro18 hours ago

People talk about how AI is bad at generating non-trivial code, but why are people using it to generate non-trivial code?

25% of coding is just the most basic boilerplate. I think of AI not as a thinking machine but as a 1000 WPM boilerplate typer.

If it is halucinating, you're trying to make it do stuff that is too complex.

ghosty14117 hours ago

But for this boiletplate creating a few snippets in your code generally works better. Especially if things change you dont have to retrain your model.

Thats my main problem: for trivial things it works but isnt much better than conventional tools, for hard things it just produces incorrect code such that writing it from scratch barely makes a difference

motoxpro17 hours ago

I think thats a great analogy.

What would it look like if I could have 3-500 snippets instead of 30. Those 300 are things that I do all over my codebase e.g. same basic where query but in the context of whatever function I am in, a click handler with the correct types for that purpose, etc.

There is no way I can have enough hotkeys or memorize that much, and I truly can't type faster than I can hit tab.

I don't need it to think for me. Most coding (front-end/back-end web) involves typing super basic stuff, not writing complex algorithms.

This is where the 10-20% speed-up comes in. On average I am just typing 20% faster by hitting tab.

globular-toast9 hours ago

Were people seriously writing this boilerplate by hand up until this point? I started using snippets and stuff more than 15 years ago!

nosbo2 days ago

I don't write code as I'm a sysadmin. Mostly just scripts. But is this like saying intellisense writes 25% of my code? Because I use autocomplete to shortcut stuff or to create a for loop to fill with things I want to do.

n_ary1 day ago

You just made it less attractive to the target corps who are to buy this product from Google. Saying, intellisense means corps already have license of various of these and some are even mostly free. Saying AI generate our 25% code sounds more attractive to corps, because it feels like something new and novel and you can imagine laying off 25% of the personnel and justify buying this product from Google.

When someone who uses a product says it, there is a 50% chance of it being true, but when someone far away from the user says it, it is 100% promotion of product and setup for trust building for a future sale.

coldpie20 hours ago

Looks like it's an impressive autocomplete feature, yeah. Check out the video about halfway down here: https://research.google/blog/ai-in-software-engineering-at-g... (linked from other comment https://news.ycombinator.com/item?id=41992028 )

Not what I thought when I heard "AI coding", but seems pretty neat.

stephenr13 hours ago

> I don't write code as I'm a sysadmin. Mostly just scripts.

.... so what do you put in your scripts if not code?

sanj2 hours ago

Caveat: I formerly worked at Google.

What missing is that code being written by AI may have less of an impact than dataset that are developed or refined by AI. Consider examples like a utility function's coefficients, or the weights of a model.

As these are aggressively tuned using ML feedback, they'll influence far more systems than raw code.

arethuza8 hours ago

I'm waiting for some Google developer to say "More than a quarter of the CEOs statements are now created by AI"... ;-)

freilanzer8 hours ago

I'd say most CEO statements are quite useless already, as they're mostly corporate newspeak.

randomNumber720 hours ago

I cannot imagine this to be true, cause imo current LLM's coding abilities are very limited. It definitely makes me more productive to use it as a tool, but I use it mainly for boilerplate and short examples (where I had to read some library documentation before).

Whenever the problem requires thinking, it horribly fails because it cannot reason (yet). So unless this is also true for google devs, I cannot see that 25% number.

jdefr892 hours ago

80% or more of the code you write day to day is just grunt work. Boring code that has, for the most part, already been written in some form such that it was copied from Google or StackOverflow. AI is basically a shortcut to using that stuff..

Wheatman10 hours ago

My guess is that they counted each line of code made by an engineer using AI coding tools.

Besides, even google employees write a lot of boilerplate, especially android IIRC, not to mention simple but essential code, so AI can prevent carpal tunnel for the junior devs working on that.

zifpanachr238 hours ago

Roughly only one quarter (assuming they are outputting similar amounts of code as non AI using engineers) of engineers actually using AI regularly for coding is a statistic that is actually believable to me based on my own experience. A lot of small teams have their "AI guy" who has drunk the kool aid, but it's not as widespread as HackerNews would make you think.

chrisjj6 hours ago

> My guess is that they counted each line of code made by an engineer using AI coding tools.

... and forgot to count the Delete presses.

alienchow9 hours ago

When setting up unit tests traditionally took more time and LOC than the logic itself, LLMs are particularly useful.

1. Paste in my actual code.

2. Prompt: Write unit tests, test tables. Include scenarios: A, B, C, D, E. Include all other scenarios I left out, isolate suggestions for review.

I used to spend the majority of the coding time writing unit tests and mocking test data, now it's more like 10%.

arkh8 hours ago

> Paste in my actual code.

> Prompt: Write unit tests

TDD in shambles. What you'd like is:

> Give your specs to some AI

> Get a test suite generated with all edge cases accounted for

> Code

alienchow8 hours ago

Matter of preference. I've found TDD to be inflexible for my working style. But your suggestion would indeed work for a staunch TDD practitioner.

avsteele15 hours ago

Everyone here is arguing about the average AI code quality and I'm here just not believing the claim.

Is Google out there monitoring the IDE activity of every engineer, logging the amount of code created, by what, lines, characters, and how it was generated? Dubious.

kunley8 hours ago

Very good point. How was the 25% measured?

Jyaif7 hours ago

> Is Google out there monitoring the IDE activity of every engineer, logging the amount of code created, by what, lines, characters, and how it was generated

A good chunk () of their code goes in a centralized repo, and is written via a centralized web IDE. So measuring everything you mentioned is not hard.

() Android, Chrome, and other similar projects are exceptions.

avsteele5 hours ago

How does this allow them to measure the % generated by AI tooling?

Jyaif1 hour ago

The IDE integrates the AI generator, like copilot.

Yes, they'll miss AI-generated code that is copy pasted, so they only have a lower bound of AI-generated code.

makerofthings9 hours ago

I keep trying to use these things but I always end up back in vim (in which I don't have any ai autocomplete set up.)

The AI is fine, but every time it makes a little mistake that I have to correct it really breaks my flow. I might type a lot more boilerplate without it but I get better flow and overall that saves me time with less mistakes.

submeta6 hours ago

Pandora‘s box has been opened.

Some say „this is mere tab completion“, some say „it won’t replace the senior engineer.“

I can remember how many fiercely argued 2 years ago that GenAI and Copilot are producing garbage. But here we are: These systems improve the workflow of creating / editing code enormously. You seniors might not be affected, but there are endless many scenarios where it replaces the junior who‘d write code to transform data, write scripts, write one-off scripts, or even write boilerplate, test code and what not.

And this is only after a short time. I cannot even imagine what we‘ll have ten years from now where we can propably have much larger context windows where the system can „unterstand“ the whole code base, not just parts.

I am sorry for low level engineering jobs, but I am super exited as well.

With GebAI I have been writing super complex Elisp code to automate workflow in Emacs, or VBA scripts in Excel, or Bash scripts I wouldn’t have otherwise been able to write, or JavaScript, or quickly write Python code to solve very tricky problems (and I am very high level in Python), or even React code for web apps for my personal use.

The future looks exiting to me.

gmm19904 hours ago

I don’t fully understand the workflow were you hand boiler plate code off to a junior wouldn’t the communication overhead be higher than writing it yourself. Certainly llms have valid uses but I see improving junior productivity more than senior productivity

thelittleone4 hours ago

I understand CEOs need to promote their companies, but it's notable that Google - arguably the world's leading information technology company - fell behind in AI development under Pichai's leadership. Now he's touting Google's internal capabilities, yet Gemini is being outperformed by relative newcomers like Anthropic and OpenAI.

His position seems secure despite these missteps, which highlights an interesting double standard: there appears to be far more tolerance for strategic failures at the CEO level compared to the rigorous performance standards expected of engineering staff.

jdefr892 hours ago

To be fair the paper that helped launch LLMs to a new level was from Google. “All You Need Is Attention”, Keras… They fell behind when it comes to marketing AI maybe…

ausbah2 days ago

i would be may more impressed if LLMs could do code compression. more code == more things that can break, and when llms can generate boatloads of it with a click you can imagine what might happen

Scene_Cast22 days ago

This actually sparked an idea for me. Could code complexity be measured as cumulative entropy as measured by running LLM token predictions on a codebase? Notably, verbose boilerplate would be pretty low entropy, and straightforward code should be decently low as well.

jeffparsons2 days ago

Not quite, I think. Some kinds of redundancy are good, and some are bad. Good redundancy tends to reduce mistakes rather than introduce them. E.g. there's lots of redundancy in natural languages, and it helps resolve ambiguity and fill in blanks or corruption if you didn't hear something properly. Similarly, a lot of "entropy" in code could be reduced by shortening names, deleting types, etc., but all those things were helping to clarify intent to other humans, thereby reducing mistakes. But some is copy+paste of rules that should be enforce in one place. Teaching a computer to understand the difference is... hard.

Although, if we were to ignore all this for a second, you could also make similar estimates with, e.g., gzip: the higher the compression ratio attained, the more "verbose"/"fluffy" the code is.

Fun tangent: there are a lot of researchers who believe that compression and intelligence are equivalent or at least very tightly linked.

8note2 days ago

Interpreting this comment, it would predict low complexity for code copied unnecessarily.

I'm not sure though. If it's copied a bunch of times, and it actually doesn't matter because each usecase of the copying is linearly independent, does it matter that it was copied?

Over time, you'd still see copies being changed by themselves show up as increased entropy

david-gpu21 hours ago

> Could code complexity be measured as cumulative entropy as measured by running LLM token predictions on a codebase? Notably, verbose boilerplate would be pretty low entropy, and straightforward code should be decently low as well.

WinRAR can do that for you quite effectively.

malfist1 day ago

Code complexity can already be measured deterministically with cyclomatic complexity. No need to use an AI fuzzy logic at this. Especially when they're bad at math.

contravariant19 hours ago

There's nothing fuzzy about letting an LLM determine the probability of a particular piece of text.

In fact it's the one thing they are explicitly designed to do, the rest is more or less a side-effect.

ks20482 days ago

I agree. It seems like counting lines of generated code is like counting bytes/instructions of compiled code - who cares? If “code” becomes prompts, then AI should lead to much smaller code than before.

I’m aware that the difference is that AI-generated code can be read and modified by humans. But that quantity is bad because humans have to understand it to read or modify it.

TZubiri2 days ago

What's that line about accounting for lines of code on the wrong side of the balance sheet?

latexr1 day ago

> If “code” becomes prompts, then AI should lead to much smaller code than before.

What’s the point of shorter code if you can’t trust it to do what it’s supposed to?

I’ll take 20 lines of code that do what they should consistently over 1 line that may or may not do the task depending on the direction of the wind.

AlexandrB2 days ago

Exactly this. Code is a liability, if you can do the same thing with less code you're often better off.

EasyMark2 days ago

Not if it’s already stable and has been running for years. Legacy doesn’t necessarily mean “need replacement because of technical debt”. I’ve seen lots of people want to replace code that has been running basically bug free for years because “there are better coding styles and practices now”

8note2 days ago

How would it know which edge cases are being useful and which ones aren't?

I understand more code as being more edge cases

wvenable20 hours ago

More code could just be useless code that no longer serves any purpose but still looks reasonable to the naked eye. An LLM can certainly figure out and suggest maybe some conditional is impossible given the rest of the code.

I can also suggest alternatives, like using existing library functions for things that might have been coded manually.

ekwav8 hours ago

Or just refactor to use early returns

asah2 days ago

meh - the LLM code I'm seeing isn't particularly more verbose. And as others have said, if you want tighter code, just add that to the prompt.

fun story: today I had an LLM write me a non-trivial perl one-liner. It tried to be verbose but I insisted and it gave me one tight line.

baalimago3 hours ago

To me, programming assistants have two usecases:

1. Generate unit tests for modules which are already written to be tested 2. Generate documentation for interfaces

Both of these require quite deep knowledge in what to write, then it simply documents and fills in the blanks using the context which already has been laid out.

sbochins20 hours ago

It’s probably code that was previously machine generated that they’re now calling “AI Generated”.

frank_nitti20 hours ago

That would make sense and be a good use case, essentially doing what OpenAPI generators do (or Yeoman generators of yore), but less deterministic I’d imagine. So optimistically I would guess it covers ground that isn’t already solved by mainstream tools.

For the example of generating an http app scaffolding from an openapi spec, it would probably account for at least 25% of the text in the generated source code. But I imagine this report would conveniently exclude the creation of the original source yaml driving the generator — I can’t imagine you’d save much typing (or mental overhead) trying to prompt a chatbot to design your api spec correctly before the codegen

prmoustache18 hours ago

Aren't we just talking about auto completion?

In that case those 25% are probably the very same 25% that were automatically generated by LTP based auto-completion.

gilfoyle2 hours ago

This is like saying more than a quarter of the code is from oss, examples and stackoverflow before LLMs.

redbell6 hours ago

Wait a second—didn't Google warn its employees against using AI-generated code? (https://news.ycombinator.com/item?id=36399021). What had changed?! Has Gemini now surpassed Bard in capabilities? Did they manage to resolve the copyright issues? Or maybe they've noticed a boost in productivity? I'm not sure, but let’s see if other big tech companies would follow this path.

KeplerBoy5 hours ago

Different audiences.

You tell investors that AI is freaking magic and going to usher in an age of savings and productivity gains.

You tell your developers that it's a neat autocomplete, they should use carefully.

nenadg5 hours ago

Internet random person (me) says more than 99% of Google's 25%+ code written by AI has already been written by humans.

hodder18 minutes ago

The market would be even more shocked to learn that another 30% is pasted in from Stack Overflow!

lysace21 hours ago

Github Copilot had an outage for me this morning. It was kind of shocking. I now believe this metric. :-)

I'll be looking into ways of running a local LLM for this purpose (code assistance in VS Code). I'm already really impressed with various quite large models running on my 32 GB Mac Studio M2 Max via Ollama. It feels like having a locally running chatgpt.

evoke490820 hours ago

Ollama, docker and "open webui".

It immediately works out of the box and that's it. I've been using local LLMs on my laptop for a while, it's pretty nice.

The only thing you really need to worry about is VRAM. Make sure your GPU has enough memory to run your model and that's pretty much it.

Also "open webui" is the worst project name I've ever seen.

kulahan20 hours ago

I'm very happy to hear this; maybe it's finally time to buy a ton of ram for my PC! A local, private LLM would be great. I'd try talking to it about stuff I don't feel comfortable being on OpenAI's servers.

lysace20 hours ago

Getting lots of ram will let you run large models on the CPU, but it will be so slow.

The Apple Silicon Macs have this shared memory between CPU and GPU that let's the (relatively underpowered GPU, compared to a decent Nvidia GPU) run these models at decent speeds, compared with a CPU, when using llama.cpp.

This should all get dramatically better/faster/cheaper within a few years, I suspect. Capitalism will figure this one out.

kulahan20 hours ago

Interesting, so this is a Mac-specific solution? That's pretty cool.

I assume, then, that the primary goal would be to drop in the beefiest GPU possible when on windows/linux?

evilduck14 hours ago

There's nothing Mac specific about running LLMs locally, they just happen to be a convenient way to get a ton of VRAM in a single small power efficient package.

In Windows and Linux, yes you'll want at least 12GB of VRAM to have much of any utility but the beefiest consumer GPUs are still topping out at 24GB which is still pretty limiting.

+1
lysace20 hours ago
mirkodrummer18 hours ago

Sometimes I wonder why we would want LLMs spit out human readable code. Wouldn’t be a better future where LLMs generate highly efficient machine code and eventually we read the “source map” for debugging? Wasn’t source code just for humans?

sparcpile15 hours ago

You just reinvented the compiler.

palata17 hours ago

Because you can't trust what the LLM generates, so you have to read it. Of course the question then is whether you can trust your developer or not.

mirkodrummer17 hours ago

I’d rather reply with LLMs aren’t just capable of that. They’re okay with Python and JS simply because there’s a lot of training data out in the open. My point was that it seems like we’re delegating the future to tools that could generate critical code using languages originally thought to be easy to learn.. it doesn’t make sense

mattxxx16 hours ago

I think they spit out human-readable code, because they've been tried on human authors.

But you make an interesting point: eventually AI will be making for other AI's + machines, and human verification can be an after thought.

standardUser18 hours ago

I use it all the time for work. Not much for actual code that goes into production, but a lot for "tell me what this does" or "given x, how do I do y". It speeds me up a ton. I'll also have it do code review when I'm uncertain about something, asking if there's any bugs or inefficiencies in a given chunk of code. I've actually found it to be more reliable about code than more general topics. Though I'm using it in a fairly specific way with code, versus asking for deep information about history for example, where is frequently gets facts very wrong.

haccount8 hours ago

No wonder Gemini is a garbage fire if had chatgpt write the code for it.

rcarmo1 day ago

There is a running gag among my friends using Google Chat (or whatever their corporate IM tool is now called) that this explains a lot of what they’re experiencing while using it…

tdeck20 hours ago

I didn't know anyone outside Google actually used that...

SavageBeast19 hours ago

Google needs to bolster their AI story and this is good click bait. I'm not buying it personally.

klocksib2 hours ago

it's quicker and easier than ever to generate a project to send to the Google Graveyard.

psunavy0339 minutes ago

And yet the 2024 State of DevOps report THAT GOOGLE PRODUCES has a butt-ton of caveats about the effectiveness of GenAI . . .

agilob8 hours ago

So we're using CoL as a metric now?

tgtweak14 hours ago

I feel like, given my experience lately with all the API models currently available, that this is only a fact if the models google is using internally are SIGNIFICANTLY better than what is available publicly even on closed models.

Claude 3.5-sonnet (latest) is barely able to stay coherent on 500 LOC files, and easily gets tripped up when there are several files in the same directory.

I have tried similarly with o1-preview and 4o, and gemini pro...

If google is using a 5M token context window LLM with 100k+ token-output trained on all the code that is not public... then I can believe this claim.

This just goes to show how critical of an issue this is that these models are behind closed doors.

nomel14 hours ago

> This just goes to show how critical of an issue this is that these models are behind closed doors.

How is competitive advantage, using in-house developed/funded tools, a critical issue? Every company has tools that only they have, that they pay significantly for to develop, and use extensively. It's can often be the primary thing that really differentiates companies who are all doing similar things.

teknopaul5 hours ago

I'd say the same. But 90% of my time not writing code. It is mostly time wasted with github and k8s build issues.

LudwigNagasena20 hours ago

How much of that generated code is `if err != nil { return err }`?

bryanrasmussen16 hours ago

Public says more than a quarter of Google's search results are absolute crap.

holtkam218 hours ago

Can we also see the stats for how much code used to come from StackOverflow? Probably 25%

me551ah6 hours ago

AI has boosted my productivity but only marginally. Earlier I used to copy paste stuff from stackoverflow and now AI generates that for me.

mgaunard8 hours ago

AI is pretty good at helping you manage a messy large codebase and making it even more messy and verbose.

Is that a good thing though? We should work and making code small and easy to manage without AI tools.

mjhay19 hours ago

100% of Sundar Pichai could be replaced by an AI.

agomez31417 hours ago

I thought great engineers reduce the amount of new code in a codebase?

elzbardico19 hours ago

Well. When I developed in Java, I think that Eclipse did similar figures circa 2005.

fennecbutt2 hours ago

That explains a lot.

zxilly9 hours ago

As a go developer, Copilot write 100% "if err != nnil for me

Kiro9 hours ago

I find it interesting that the people who dismiss the utility of AI are being so aggressive, sarcastic and hateful about it. Why all the anger? Where's the curiosity?

echoangle11 hours ago

Does protobuf count as AI now?

hsuduebc215 hours ago

I believe it is absolutely suitable for generating controllers in java spring or connecting to database and making a simple query which from my experience as an ordinary enterprise developer in Fintech is most of the job. Making these huge applicatins is a lot of repetitive work and integrations. Not a work that usually requires some advanced logic.

ajkjk49 minutes ago

Well yeah he sells AI and wants you to believe in it so the stock price stays good.

blibble18 hours ago

this is the 2024 version of "25% of our code is now produced by outsourced resources"

skatanski18 hours ago

I think at this moment, this sounds more like "quarter of the company's new code is created using stackoverflow and other forums. Many many people use all these tools to find information, as they did using stackoverflow a month ago, but now suddenly we can call it "created by AI". It'd be nice to have a distinction. I'm saying this, while being very excited about using LLMs as a developer.

twis20 hours ago

How much code was "written by" autocomplete before LLMs came along? From my experience, LLM integration is advanced autocomplete. 25% is believable, but misleading.

scottyah19 hours ago

My linux terminal tab-complete has written 50% of my code

nottorp7 hours ago

The protobuf boilerplate, right? :)

jmartin268318 hours ago

I’m gonna bet this is a lie.

freedomben17 hours ago

I don't think it's a lie, but I do think it's very misleading. With common languages probably 25% of code can be generated by an AI, but IME it's mostly just boilerplate or some pattern that largely just saves typing time, not programming/thinking time. In other words it's the 25% lowest hanging fruit, so thinking like "1/4 of programming is now done by AI" is misleading. It's probably more like 5 to 10 percent.

meindnoch8 hours ago

I saw code on master which was parsing HTML with regex. The author was proud that this code was mostly generated by AI.

:)

sheeshkebab4 hours ago

and it shows… Google codebases I see in the wild are the worst - jumbled mess of hard to read code.

mastazi17 hours ago

The auto-linter in my editor probably generates a similar percentage of the characters I commit.

tabbott18 hours ago

Without a clear explanation of methodology, this is meaningless. My guess is this statistic is generated using misleading techniques like classifying "code changes generated by existing bulk/automated refactoring tools" as "AI generated".

ThinkBeat20 hours ago

This is quite interesting to know.

I will be curious to see if it has any impact positive or negative over a couple of years.

Will the code be more secure since the AI does not make the mistakes humans do?

Or will the code, not well enough understood by the employees, exposes exploits that would not be there?

Will it change average up time?

kunley8 hours ago

what makes you think that current direction of AI development would lead to making less mistakes than humans do, as opposed to repeating same miskates plus hallucinating more?

davidclark19 hours ago

If I tab complete my function and variable symbols, does my lsp write 80%+ of my lines of code?

Starlevel00420 hours ago

No wonder search barely works anymore

syngrog662 hours ago

> "and we continue to be laser-focused on building great products."

NO! False. I can confirm they are not. I've known of several major obvious unfixed bugs/flaws in Google apps for years. and in the last year or so especially theres been an explosion in the number of head-scratching, jaw-dropping fails and UX anti-patterns in their code. GMail, Search, Maps and Android are now riddled with them.

on Sundar Pichai's watch he's been devolving Google to be yet another Microsoft type in terms of quality, care and taste.

1GZ08 hours ago

I wonder how much of that code is boilerplate vs. actual functionality.

Terr_18 hours ago

My concern is that "frequently needed and immediately useful results" is strongly correlated to "this code should already be abstracted away into a library by now."

Search Copy-Paste as a Service is hiding a deeper issue.

_spduchamp20 hours ago

I can ask AI to generate the same code multiple times, and get new variations on programming style each time, and get the occasional solution that is just not quite right but sort of works. Sounds like a recipe for a gloppy mushy mess of style salad.

okokwhatever2 hours ago

People still don't understand those who pay the bills are those who claim developers are less and less necessary. It doesn't matter how much we love our job and how much we care for quality, at the end those who pay take more care of reducing workforce for something potentially free or cheap. We are less needed, less cared and less seen as engineers. We are just a cost in a wrong column of Quickbooks. Get use to it.

nektro16 hours ago

Google used to be respected, a place so highly sought after that engineers who worked there were revered like wizards. oh how they've fallen :(

fredgrott5 hours ago

Kind of useless stat given how much code a typical dev refactors....

tylerchilds3 hours ago

as a consumer, i never could have guessed

hiptobecubic12 hours ago

I've had mixed results writing "normal" business logic in c++, but i gotta say, for SQL it's pretty incredible. Granted SQL has a lot of boilerplate and predictable structure, but it saves a ton of time honestly.

shane_kerns14 hours ago

It's no wonder that their search absolutely sucks now. Duckduckgo is so much better in comparison now.

cebert20 hours ago

Did AI have to go thru several rounds of Leetcode interviews?

hi_hi19 hours ago

> More than a quarter of new code created at Google is generated by AI, said CEO Sundar Pichai...

How do they know this? At face value, it sounds like alot, but it only says "new code generated". Nothing about code making it into source control or production, or even which parts of googles vast business units.

For all we know, this could be the result of some internal poll "Tell us if you've been using Goose recently" or some marketing analytics on the Goose "Generate" button.

It's puff piece to put Google back in the lime light, and everyone is lapping it up.

mjbale1162 days ago

If you manage to convince software engineers that you are doing them a favour by employing them then they will approach any workplace negotiations with a specific mindset which will make them grab the first number it gets thrown to them.

These statements are brilliant.

akira250120 hours ago

These statements rely on an unchallenged monopoly position. This is not sustainable. These statements will hasten the collapse.

fmardini8 hours ago

Proto-plumbing is very LLM amenable

odinkara2 hours ago

and it shows

anacrolix4 hours ago

Puts on Google

defactor14 hours ago

Try any AI tool to write basic factor code.hallucinates most of the time

foobarian17 hours ago

The real question is, what fraction of the company’s code is deleted by AI :-)

ChrisArchitect2 days ago

Related:

Alphabet ($GOOG) 2024 Q3 earnings release

https://news.ycombinator.com/item?id=41988811

chabes20 hours ago

When Google announced their big layoffs, I noted the timing in relation to some big AI announcements. People here told me I was crazy for suggesting that corporations could replace employees with AI this early. Now the CEO is confirming that more than a quarter of new code is created by AI. Can’t really deny that reality anymore folks.

hbn20 hours ago

I'd suggest the bigger factor in those layoffs is the money was made in earlier covid years where money was flowing and everyone was overhiring to show off record growth, then none of those employees had any justification for being kept around and were just a money sink so they fired them all.

Not to mention Elon publicly demonstrated losing 80% of staff when he took over twitter and - you can complain about his management all you want - as someone who's been using it the whole way through, from a technical POV their downtimes and software quality has not been any worse and they're shipping features faster. A lot of software companies are overstaffed, especially Google who has spent years paying people to make projects just to get a PO promoted, then letting the projects rot and die to be replaced by something else. That's a lot of useless work being done.

akira250120 hours ago

> Can’t really deny that reality anymore folks.

You have to establish that the CEO is actually aware of the reality and is interested in accurately conveying that to you. As far as I can tell there is absolutely no reason to believe any part of this.

paradox24220 hours ago

When leaders without the requisite technical knowledge are making decisions then the question of whether AI is capable of replacing human workers is orthogonal to the question of whether human workers will be replaced by AI.

robohoe20 hours ago

Who claims that he is speaking the truth and not some marketing jargon?

randomNumber719 hours ago

People who have replaced 25% of their brain with ai.

josephd794 hours ago

that explains everything.

skywhopper18 hours ago

All this means is that 25% of code at Google is trivial boilerplate that would be better factored out of their process rather than tasking inefficient LLM tools with. The more they are willing to leave the “grunt work” to an LLM, the less likely they are to ever eliminate it from the process.

soperj20 hours ago

The real question is how many lines of code was it responsible for removing.

AmazingTurtle3 hours ago

Yeah, go ahead and lay off another 25% of development staff and see how well AI coders perform.:))

marstall21 hours ago

maps with recent headlines about AI improving programmer productivity 20-30%.

which puts it in line with previous code-generation technologies i would imagine. I wonder which of these increased productivity the most?

- Assembly Language

- early Compilers

- databases

- graphics frameworks

- ui frameworks (windows)

- web apps

- code generators (rails scaffolding)

- genAI

akira250120 hours ago

Early Compilers. By a wide margin. They are the enabling factor for everything that comes below it. It's what allows you to share library interfaces and actually use them in a consistent manor and across multiple architectures. It entirely changed the shape of software development.

The gap between "high level assembly" and "compiled language" is about as large as it gets.

hggigg20 hours ago

I reckon he’s talking bollocks. Same as IBM was when it was about to disguise layoffs as AI uplift and actually just shovelled the existing workload on to other people.

niobe17 hours ago

This explains a LOT about Google's quality decline.

otabdeveloper420 hours ago

That explains a lot about Google's so-called "quality".

zxvkhkxvdvbdxz19 hours ago

I feel this made me loose the respect I still had for Google

wokkaflokka19 hours ago

No wonder their products are getting worse and worse...

rockskon16 hours ago

No shit a quarter of Google's new code is created by AI. How else do you explain why Google search has been so aggressively awful for the past 5~ years?

Seriously. The penchant for outright ignoring user search terms, relentlessly forcing irrelevant or just plain wrong information on users, and the obnoxious UI changes on YouTube! If I'm watching a video on full screen I have explicitly made it clear that I want YouTube to only show me video! STOP BRINGING UP THE FUCKING VIDEO DESCRIPTION TO TAKE UP HALF THE SCREEN IF I TRY TO BRIEFLY SWIPE TO VIEW THE TIME OR READ A MESSAGE.

I have such deep-seated contempt for AI and it's products for just how much worse it makes people's lives.

remram16 hours ago

Yeah that might explain some of the loss of quality. Google apps and sites used to be solid, now they are full of not-breaking-but-annoying bugs like race conditions (don't press buttons too fast), display glitches, awful recommendations, and other usability problems.

Then again, their devices are also coming out with known fatal design flaws, like not being able to make phone calls, or the screen going black permanently.

sigmonsays15 hours ago

imho code that is written by AI is code that is not worth having.

ThinkBeat19 hours ago

So um. With making this public statement, can we expect that 25% of "the bottom" coders at Google will soon be granted a lot more time and ability to spend time with their loves ones.

jdmoreira7 hours ago

I would prefer if he was more competent and made the stock price go up. I guess grifters are going to grift

Hamuko19 hours ago

How do Google's IP lawyers feel about a quarter of the company's code not being copyrightable?

sjs3823 hours ago

This was one of my first thoughts, too. In what ways can this contaminate their codebase? What if they use AI to add uncopyrightable code to GPL projects?

deterministic19 hours ago

Not impressed. I currently auto generate 90% or more of the code I need to implement business solutions. With no AI involved. Just high level declarations of intent auto translated to C++/Typescript/…

Timber-653910 hours ago

All this talk means nothing until Google gives AI permissions to push to prod.

marstall21 hours ago

first thought is that much of that 25% is test code for non-ai-gen code...

jagged-chisel5 hours ago

“Created by” or “with the assistance of”?

oglop1 day ago

No surprise. I give my career about 2 years before I’m useless.

k4rli20 hours ago

Seems just overhyped tech to push up stock prices. It was already claimed 2 years ago that half of the jobs would be taken by "AI" but barely any have and AI has barely improved since GPT3.5. Latest Anthropic is only slightly helpful for software development, mostly for unusual bug investigations and logs analysis, at least in my experience.

phi-go22 hours ago

They still need someone to write 75% of the code.

DidYaWipe10 hours ago

No wonder it sucks. Google's vaunted engineering has always been suspect, but their douchebaggery has been an accepted fact (even by them)>

marviel19 hours ago

> 80% at Reasonote

horns4lyfe16 hours ago

I’d bet at least a quarter of their code is class definitions, constructors, and all the other minutiae files required for modern software, so that makes sense. But people weren’t writing most of that before either, we’ve had autocomplete and code geb for a long time.

martin8211 hours ago

I guess that must be the reason for the shocking enshitification of Google

jeffbee19 hours ago

It's quite amusing to me because I am old enough to remember when Copilot emerged the HN mainthought was that it was the death sentence for big corps, the scrappy independent hacker was going to run circles around them. But here we see the predictable reality: an organization that is already in an elite league in terms of developer velocity gets more benefit from LLM code assistants than Joe Hacker. These technologies serve to entrench and empower those who are already enormously powerful.

arminiusreturns19 hours ago

I was a luddite about the generative LLMs at first, as a crusty sysadmin type. I came around and started experimenting. It's been a boon for me.

My conclusion is that we are at the first wave of a split between those who use LLMs to augment their abilities and knowledge, and those who delay. In cyberpunk terminally, it's aug-tech, not real AGI. (and the lesser ones code abilities and simpler the task, the more benefit, it's an accelerator)

est15 hours ago

Now maintain quarter of your old code base with AI, don't shut down services randomly.

nephy17 hours ago

Can we move on to the next grift yet?

yapyap15 hours ago

yikes

xyst18 hours ago

I remember Google used to market "lines of code" for their products. Chrome at one point had 6.7 LoC. Now the new marketing term is: "product was made with 1M lines of AI generated code (slop)!11!". Or "Chrome refactored with 10% AI" or some bs

skrebbel21 hours ago

To my experience, AIs can generate perfectly good code relatively easy things, the kind you might as well copy&paste from stackoverflow, and they'll very confidently generate subtly wrong code for anything that's non-trivial for an experienced programmer to write. How do people deal with this? I simply don't understand the value proposition. Does Google now have 25% subtly wrong code? Or do they have 25% trivial code? Or do all their engineers babysit the AI and bugfix the subtly wrong code? Or are all their engineers so junior that an AI is such a substantial help?

Like, isn't this announcement a terrible indictment of how inexperienced their engineers are, or how trivial the problems they solve are, or both?

toasteros20 hours ago

> the kind you might as well copy&paste from stackoverflow

This bothers me. I completely understand the conversational aspect - "what approach might work for this?", "how could we reduce the crud in this function?" - it worked a lot for me last year when I tried learning C.

But the vast majority of AI use that I see is...not that. It's just glorified, very expensive search. We are willing to burn far, far more fuel than necessary because we've decided we can't be bothered with traditional search.

A lot of enterprise software is poorly cobbled together using stackoverflow gathered code as it is. It's part of the reason why MS Teams makes your laptop run so hot. We've decided that power-inefficient software is the best approach. Now we want to amplify that effect by burning more fuel to get the same answers, but from an LLM.

It's frustrating. It should be snowing where I am now, but it's not. Because we want to frivolously chase false convenience and burn gallons and gallons of fuel to do it. LLM usage is a part of that.

jcgrillo17 hours ago

What I can't wrap my head around is that making good, efficient software doesn't (by and large) take significantly longer than making bloated, inefficient enterprise spaghetti. The problem is finding people to do it with who care enough to think rigorously about what they're going to do before they start doing it. There's this bizarre misconception popular among bigtech managers that there's some tunable tradeoff between quality and development speed. But it doesn't actually work that way at all. I can't even count anymore how many times I've had to explain how taking this or that locally optimal shortcut will make it take longer overall to complete the project.

In other words, it's a skill issue. LLMs can only make this worse. Hiring unskilled programmers and giving them a machine for generating garbage isn't the way. Instead, train them, and reject low quality work.

aleph_minus_one15 hours ago

> What I can't wrap my head around is that making good, efficient software doesn't (by and large) take significantly longer than making bloated, inefficient enterprise spaghetti. The problem is finding people to do it with who care enough to think rigorously about what they're going to do before they start doing it.

I don't think finding such programmers is really difficult. What is difficult is finding such people if you expect them to be docile to incompetent managers and other incompetent people involved in the project who, for example, got their position not by merit and competence, but by playing political games.

giantg217 hours ago

"What I can't wrap my head around is that making good, efficient software doesn't (by and large) take significantly longer than making bloated, inefficient enterprise spaghetti."

In my opinion the reason we get enterprise spaghetti is largely due to requirement issues and scope creep. It's nearly impossible to create a streamlined system without knowing what it should look like. And once the system gets to a certain size, it's impossible to get business buy-in to rearchitect or refactor to the degree that is necessary. Plus the full requirements are usually poorly documented and long forgotten by that time.

+2
jcgrillo16 hours ago
galdosdi14 hours ago

It's a market for lemons.

Without redoing their work or finding a way to have deep trust (which is possible, but uncommon at a bigcorp) it's hard enough to tell who is earnest and who is faking it (or buying their own baloney) when it comes to propositions like "investing in this piece of tech debt will pay off big time"

As a result, if managers tend to believe such plans, bad ideas drive out good and you end up investing in a tech debt proposal that just wastes time. Burned managers therefore cope by undervaluing any such proposals and preferring the crappy car that at least you know is crappy over the car that allegedly has a brand new 0 mile motor on it but you have no way of distinguishing from a car with a rolled back odometer. They take the locally optimal path because it's the best they can do.

It's taken me 15 years of working in the field and thinking about this to figure it out.

The only way out is an organization where everyone is trusted and competent and is worthy of trust, which again, hard to do at most random bigcorps.

This is my current theory anyway. It's sad, but I think it kind of makes sense.

c0balt16 hours ago

It's relatively easy to find a programmer(s) who can realize enterprise project X, it's hard to find a programmer(s) who cares about X. Throwing an increased requirement like speed at it makes this worse because it usually ends up burning out both ends of the equation.

jihadjihad16 hours ago

> The problem is finding people to do it with who care enough to think rigorously

> ...

> train them, and reject low quality work.

I agree very strongly with both of these points.

But I've observed a truth about each of them over the last decade-plus of building software.

1) very few people approach the field of software engineering with anything remotely resembling rigor, and

2) there is often little incentive to train juniors and reject subpar output (move fast and break things, etc.)

I don't know where this takes us as an industry? But I feel your comment on a deep level.

+1
jcgrillo15 hours ago
steve_adams_8616 hours ago

I agree as well. These are actually things that bother me a lot about the industry. I’d love to write software that should run problem-free in 2035, but the reality is almost no one cares.

I’ve had the good fortune of getting to write some firmware that will likely work well for a long time to come, but I find most things being written on computers are written with (or very close to) the minimum care possible in order to get the product out. Clean up is intended but rarely occurs.

I think we’d see real benefits from doing a better job, but like many things, we fail to invest early and crave immediate gratification.

+1
karolinepauls16 hours ago
wpietri14 hours ago

Agreed.

The way I explain this to managers is that software development is unlike most work. If I'm making widgets and I fuck up, that widget goes out the door never to be seen again. But in software, today's outputs are tomorrow's raw materials. You can trade quality for speed in the very short term at the cost of future productivity, so you're really trading speed for speed.

I should add, though, that one can do the rigorous thinking before or after the doing, and ideally one should do both. That was the key insight behind Martin Fowler's "Refactoring: Improving the Design of Existing Code". Think up front if you can, but the best designs are based on the most information, and there's a lot of information that is not available until later in a project. So you'll want to think as information comes in and adjust designs as you go.

That's something an LLM absolutely can't do, because it doesn't have access to that flow of information and it can't think about where the system should be going.

noisy_boy11 hours ago

> The problem is finding people to do it with who care enough to think rigorously about what they're going to do before they start doing it.

There is no incentive to do it. I worked that way, focused on quality and testing and none of my changes blew up in production. My manager opined that this approach is too slow and that it was ok to have minor breakages as long as they are fixed soon. When things break though, it's blame game all around. Loads of hypocrisy.

sethammons17 hours ago

"Slow is smooth and smooth is fast"

jcgrillo15 hours ago

It's true every single time.

A4ET8a8uTh016 hours ago

<< Instead, train them, and reject low quality work.

Ahh, well, in order to save money, training is done via an online class with multiple choice questions, or, if your company is like mine and really committed to making sure that you know they take your training seriously, they put portions of a generic book on 'tech Z' in pdf spread spread over a drm ridden web pages.

As for code, that is reviewed, commented and rejected by llms as well. It is used to be turtles. Now it truly is llms all the way down.

That said, in a sane world, this is what should be happening for a company that actually wants to get good results over time .

chongli20 hours ago

we've decided we can't be bothered with traditional search

Traditional search (at least on the web) is dying. The entire edifice is drowning under a rapidly rising tide of spam and scam sites. No one, including Google, knows what to do about it so we're punting on the whole project and hoping AI will swoop in like deus ex machina and save the day.

photonthug19 hours ago

Maybe it is naive but I think search would probably work again if they could roll back code to 10 or 15 years ago and just make search engines look for text in webpages.

Google wasn’t crushed by spam, they decided to stop doing text search and build search bubbles that are user specific, location-specific, decided to surface pages that mention search terms in metadata instead of in text users might read, etc. Oh yeah, and about a decade before LLMs were actually usable, they started to sabotage simple substring searches and kind of force this more conversational interface. That’s when simple search terms stopped working very well, and you had to instead ask yourself “hmm how would a very old person or a small child phrase this question for a magic oracle”

This is how we get stuff like: Did you mean “when did Shakespeare die near my location”? If anyone at google cared more about quality than printing money, that thirsty gambit would at least be at the bottom of the page instead of the top.

+1
hughesjj18 hours ago
+3
layer818 hours ago
+1
CapeTheory16 hours ago
masfuerte19 hours ago

Google results are not polluted with spam because Google doesn't know how to deal with it.

Google results are polluted with spam because it is more profitable for Google. This is a conscious decision they made five years ago.

+1
chongli19 hours ago
redwall_hp15 hours ago

If you own the largest ad network that spam sites use and own the traffic firehose, pointing the hose at the spam sites and ensuring people spend more time clicking multiple results that point to ad-filled sites will make you more money.

Google not only has multiple monopolies, but a cut and dry perverse incentive to produce lower quality results to make the whole session longer instead of short and effective.

skissane19 hours ago

I personally think a big problem with search is major search engines try to be all things to all people and hence suffer as a result.

For example: a beginner developer is possibly better served by some SEO-heavy tutorial blog post; an experienced developer would prefer results weighted towards the official docs, the project’s bug tracker and mailing list, etc. But since less technical and non-technical people vastly outnumber highly technical people, Google and Bing end up focusing on the needs of the former, at the cost of making search worse for the later.

One positive about AI: if an AI is doing the search, it likely wants the more advanced material not the more beginner-focused one. It can take more advanced material and simplify it for the benefit of less experienced users. It is (I suspect) less likely to make mistakes if you ask it to simplify the more advanced material than if you just gave it more beginner-oriented material instead. So if AI starts to replace humans as the main clients of search, that may reverse some of the pressure to “dumb it down”.

photonthug18 hours ago

> But since less technical and non-technical people vastly outnumber highly technical people, Google and Bing end up focusing on the needs of the former, at the cost of making search worse for the later.

I mostly agree with your interesting comment, and I think your analysis basically jives with my sibling comment.

But one thing I take issue with is the idea that this type of thing is a good faith effort, because it’s more like a convenient excuse. Explaining substring search or even include/exclude ops to children and grandparents is actually easy. Setting preferences for tutorials vs API docs would also be easy. But companies don’t really want user-directed behavior as much as they want to herd users to preferred content with algorithms, then convince the user it was their idea or at least the result of relatively static ranking processes.

The push towards more fuzzy semantic search and “related content” everywhere is not to cater to novice users but to blur the line between paid advertisement and organic user-directed discovery.

No need to give megacorp the benefit of the doubt on stuff like this, or make the underlying problems seem harder than they are. All platforms land in this place by convergent evolution wherein the driving forces are money and influence, not insurmountable technical difficulties or good intentions for usability.

+1
consp18 hours ago
skydhash19 hours ago

> Traditional search (at least on the web) is dying.

That's not my experience at all. While there are scammy sites, using the search engines as an index instead of an oracle still yields useful results. It only requires to learn the keywords which you can do by reading the relevant materials .

rubyfan18 hours ago

AI will make the problem of low quality, fake, fraudulent and arbitrage content way worse. I highly doubt it will improve searching for quality content at all.

layer818 hours ago

Without a usable web search index, AI will be in trouble eventually as well. There is no substitute for it.

ponector18 hours ago

>> No one, including Google, knows what to do about it

I'm sure they can. But they have no incentive. Try to Google an item, and it will show you a perfect match of sponsored ads and some other not-so-relevant non-sponsored results

lokar20 hours ago

It took the scam/spam sites a few years to catch up to Google search. Just wait a bit, equilibrium will return.

fmos12 hours ago

Kagi has fixed traditional search for me.

AtlasBarfed18 hours ago

There's no way the search AI will beat out the spamgen AI.

Tailoring/retraining the main search AI will be so much more expensive that retraining the spam special purpose AIs.

AnimalMuppet20 hours ago

But it can't save the day.

The problem with Google search is that it indexes all the web, and there's (as you say) a rising tide of scam and spam sites.

The problem with AI is that it scoops up all the web as training data, and there's a rising tide of scam and spam sites.

petre20 hours ago

AI will generate even more spam and scam sites more trivially.

ses198415 hours ago

What do you mean “will”, we are a few years past that point.

quickthrowman18 hours ago

Google could fix the problem if they wanted to, but it’s not in their interests to fix it since the spam sites generally buy ads from Google and/or display Google ads on their spam websites. Google wants to maximize their income, so..

cyanydeez19 hours ago

If only google was trying to solve search rather than shareholdet values.

akoboldfrying19 hours ago

>The entire edifice is drowning under a rapidly rising tide of spam and scam sites.

You make this claim with such confidence, but what is it based on?

There have always been hordes of spam and scam websites. Can you point to anything that actually indicates that the ratio is now getting worse?

+1
chongli19 hours ago
romwell20 hours ago

Narrator: it did not, in fact, save the day.

jihadjihad16 hours ago

Another frustration I have with these models is that it is yet another crutch and excuse for turning off your brain. I was tagged on a PR a couple days ago where a coworker had added a GIN index to a column in Postgres, courtesy of GPT-4o, of course.

He couldn't pronounce the name of the extension, apparently not noticing that trgm == trigram, or what that might even be. Copying the output from the LLM and pasting it into a PR didn't result in anything other than him checking off a box, moving a ticket in Jira, and then onto the next thing--not even a pretense of being curious about what any of it all meant. But look at those query times now!

It's been possible for a while to shut off your brain as a programmer and blindly copy-paste from StackOverflow etc., but the level of enablement that LLMs afford is staggering.

tru3_power16 hours ago

Out of curiosity- did it work though?

gonzobonzo14 hours ago

Doesn't this get to one of the fundamental issues though, that many of these frameworks and languages are poorly constructed in the first place? A lot of the times people turn to web searches, Stack Overflow, or AI is because they want to do X, and there's no quick, clear, and intuitive way to do X. I write cheat sheets for opaque parts of various frameworks myself. A lot of them aren't fundamentally difficult once you understand them, but they're constructed in an extremely convoluted way, and there's usually extremely poor documentation explaining how to actually use them.

In fact, I'd say I use AI more for documentation than I do for code itself, because AI generated documentation is often superior to official documentation.

In the end, these things shouldn't be necessary (or barely necessary) if we had well constructed languages, frameworks, libraries and documentation, but it appears like it's easier to build AI than to make things non-convoluted in the first place.

braiamp15 hours ago

> because we've decided we can't be bothered with traditional search

Traditional search was only Google, and Google figured out that they don't need to improve their tools to make it better, because everyone will continue to use it as a force of habit (google is a verb!). Traditional search is being abandoned because traditional search isn't good enough for the kinds of search we need (also, while google may claim their search is very useful, people rarely search stuff nowadays, instead prefer being passively fed content via recommendations algorithm (that also use AI!))

dleeftink15 hours ago

Algolia, Marginalia, Kagi, Scopus, ConnectedPapers, Lense[0] all stick to more or less traditional search and yield consistent high quality results. It shouldn't be one or the other, and I think the first one to combine both paradigms in a seamless fashion would be quite successfull (it has been tried, I know, but it's still a niche in many cases).

[0]: https://www.lens.org/lens/search/

hawski18 hours ago

A human can't be trusted to not make memory safety bugs. At the same time we can trust AI with logic bugs.

kelnos17 hours ago

Since LLMs are just based on human output, we should trust LLMs (at best) as much as we trust the average human coder. And in reality we should probably trust them less.

GaggiX16 hours ago

These models are simply much more powerful than a tradition search engine and stackoverflow, so many people use these models for a reason, a friend of mine that never tried ChatGPT until very recently managed to solve a problem he couldn't find a solution on stackoverflow using GPT-4o, next time he's probably going to ask the model directly.

toasteros10 minutes ago

I don't know what your friend's prompts were, but this probably speaks to the conversational aspect. I've found success in using LLMs to "search" for things I don't know how to search for - a 'tip of my tongue' type scenario.

"How do I do a for loop" though is a waste of time and energy and should be put into a search engine. There is no need to use the inefficient power needs of an LLM to answer that question. The search engine will have cached the results of that question, leading to a much faster discovery of the answer, and less power draw to do it, whereas an LLM needs to ponder your question EVERY. SINGLE. TIME. A huge waste.

Stop using LLMs for simple things.

worik18 hours ago

> But the vast majority of AI use that I see is...not that. It's just glorified, very expensive search.

Since the collapse of Internet search (rose tinted hindsight - was it ever any good?) I have been using a LLM as my syntax advisor. I pay for my own tokens, and I can say it is astonishingly cheap

It is also very good.

Dalewyn17 hours ago

>We are willing to burn far, far more fuel than necessary because we've decided we can't be bothered with traditional search.

That's because traditional search fucking sucks balls.

rpcope120 hours ago

I don't get it either. People will say all sorts of strange stuff about how it writes the code for them or whatever, but even using the new Claude 3.5 Sonnet or whatever variant of GPT4, the moment I ask it anything that isn't the most basic done-to-death boilerplate, it generates stuff that's wrong, and often subtly wrong. If you're not at least pretty knowledgeable about exactly what it's generating, you'll be stuck trying to troubleshoot bad code, and if you are it's often about as quick to just write it yourself. It's especially bad if you get away from Python, and try to make it do anything else. SQL especially, for whatever reason, I've seen all of the major players generate either stuff that's just junk or will cause problems (things that your run of the mill DBA will catch).

Honestly, I think it will become a better Intellisense but not much more. I'm a little excited because there's going to be so many people buying into this, generating so much bad code/bad architecture/etc. that will inevitably need someone to fix after the hype dies down and the rug is pulled, that I think there will continue to be employment opportunities.

solumunus19 hours ago

Supermaven is an incredible intellisense. Most code IS trivial and I barely write trivial code anymore. My imports appear instantly, with high accuracy. I have lots of embedded SQL queries and it’s able to guess the structure of my database very accurately. As I’m writing a query the suggested joins are accurate probably 80% of the time. I’m significantly more productive and having to type much less. If this is as good as it ever gets I’m quite happy. I rarely use AI for non trivial code, but non trivial code is what I want to work on…

ta_113819 hours ago

This is all about the tooling most companies choose when building software: Things with more than enough boilerplate most code is trivial. We can build tools that have far less triviality and more density, where the distance between the code we write and business logic is very narrow.. but then every line of code we write is hard, because it's meaningful, and that feels bad enough to many developers, so we end up with tools where we might not be more productive, but we might feel productive, even though most of that apparent productivity is trivially generated.

We also have the ceremonial layers of certain forms of corporate architecture, where nothing actually happens, but the steps must exist to match the holy box, box cylinder architecture. Ceremonial input massaging here, ceremonial data transformation over there, duplicated error checking... if it's easy for the LLM to do, maybe we shouldn't be doing it everywhere in the first place.

+1
thfuran19 hours ago
codr713 hours ago

I think you just nailed the paradox of Go's popularity among developers, managers are obvious.

monksy19 hours ago

I don't think that is the signal that I think most people are hoping for here.

When I hear that most code is trivial, I think of this as a language design or a framework related issue making things harder than they should be.

Throwing AI or generates at the problem just to claim that they fixed it is just frustrating.

throw23423423417 hours ago

> When I hear that most code is trivial, I think of this as a language design or a framework related issue making things harder than they should be.

This was one of my thoughts too. If the pain of using bad frameworks and clunky languages can be mitigated by AI, it seems like the popular but ugly/verbose languages will win out since there's almost no point to better designed languages/framework. I would rather a good language/framework/etc where it is just as easy to just write the code directly. Similar time in implementation to a LLM prompt, but more deterministic.

If people don't feel the pain of AI slop why move to greener pastures? It almost encourages things to not improve at the code level.

solumunus7 hours ago

I'm writing software independently, with an extremely barebones framework (just handles routing pretty much) and very lean architecture. Maybe I should re-phrase it, "a lot of characters in the code base are trivial". Imports, function declarations, variable declarations. Is this stuff code/logic? Barely, but it's completely unavoidable. It all takes time and it's now time I rarely have to spend.

Just as an example, I have "service" functions. They're incredibly simple, a higher order function where I can inject the DB handler, user permissions, config, etc. Every time I write one of these I have to import the ServiceDependencies type and declare which dependencies I need to write the service. I now spend close to zero time doing that and all my time focusing on the service logic. I don't see a downside to this.

Most of my business logic is done in raw SQL, which can be complex, but the autocomplete often helps there too. It's not helping me figure out the logic, it's simply cutting down on my typing. I don't know how anyone could be offered "do you want to have type significantly less characters on your keyboard to get the same thing done?" and say "no thanks". The AI is almost NEVER coding for me, it's just typing for me and it's awesome.

I don't care how lean your system is, there will at least be repetition in how you declare things. There will be imports, there will be dependencies. You can remove 90% of this repetitive work for almost no cost...

I've tried to use ChatGPT to "code for me", and I agree with you that it's not a good option if you're trying to do anything remotely complex and want to avoid bugs. I never do this. But integrated code suggestions (with Supermaven, NOT CoPilot) are incredibly beneficial and maybe you should just try it instead of trying to come up with theoretical arguments. I was also a non-believer once.

int_19h17 hours ago

Well, Google did design Go...

Kiro19 hours ago

Interesting that you believe your subjective experience outweighs the claims of all others who report successfully using LLMs for coding. Wouldn't a more charitable interpretation be that it doesn't fit the stuff you're doing?

kelnos17 hours ago

Why wouldn't someone's subjective experience outweigh someone else's subjective experience?

Regardless, I do wonder how accurate those successful reports are. Do people take LLM output, use it verbatim, not notice subtle bugs, and report that as success?

Kiro10 hours ago

There's a big difference between "I've seen X" and "I've not seen X". The latter does not invalidate the former, unless you believe the person is lying or being delusional.

tobyjsullivan20 hours ago

I'm not a Google employee but I've heard enough stories to know that a surprising amount of code changes at google are basically updating API interfaces.

The way google works, the person changing an interface is responsible for updating all dependent code. They create PRs which are then sent to code owners for approval. For lower-level dependencies, this can involve creating thousands of PRs across hundreds of projects.

Google has had tooling to help with these large-scale refactors for decades, generally taking the form of static analysis tools. However, these would be inherently limited in their capability. Manual PR authoring would still be required in many cases.

With this background, LLM code gen seems like a natural tool to augment Google's existing process.

I expect Google is currently executing a wave of newly-unblocked refactoring projects.

If anyone works/worked at google, feel free to correct me on this.

cj17 hours ago

Do they have tooling for generating scaffolding for various things (like unit/integration tests)?

If we’re guessing what code is easiest and largest proportion of codebase to write, my first guess would be test suites. Lots of lines of repetitive code patterns that repeat and AI is decent at dealing with

slibhb19 hours ago

Most programming is trivial. Lots of non-trivial programming tasks can be broken down into pure, trivial sections. Then, the non-trivial part becomes knowing how the entire system fits together.

I've been using LLMs for about a month now. It's a nice productivity gain. You do have to read generated code and understand it. Another useful strategy is pasting a buggy function and ask for revisions.

I think most programmers who claim that LLMs aren't useful are reacting emotionally. They don't want LLMs to be useful because, in their eyes, that would lower the status of programming. This is a silly insecurity: ultimately programmers are useful because they can think formally better than most people. For the forseeable future, there's going to be massive demand for that, and people who can do it will be high status.

tonyedgecombe19 hours ago

>I think most programmers who claim that LLMs aren't useful are reacting emotionally.

I don't think that's true. Most programmers I speak to have been keen to try it out and reap some benefits.

The almost universal experience has been that it works for trivial problems, starts injecting mistakes for harder problems and goes completely off the rails for anything really difficult.

theshackleford16 hours ago

> I don't think that's true. Most programmers I speak to have been keen to try it out and reap some benefits.

I’ve been seeing the complete opposite. So it’s out there.

gorjusborg19 hours ago

> Most programming is trivial

That's a bold statement, and incorrect, in my opinion.

At a junior level software development can be about churning out trivial code in a previously defined box. I don't think its fair to call that 'most programming'.

BobbyJo19 hours ago

Probably overloading of the term "programming" is the issue here. Most "software engineering" is non-programming work. Most programming is not actually typing code.

Most of the time, when I am typing code, the code I am producing is trivial, however.

uh_uh15 hours ago

Think of all the menial stuff you must perform regardless of experience level. E.g. you change the return type of a function and now you have to unpack the results slightly differently. Traditional automated tools fail at this. But if you show some examples to Cursor, it quickly catches on to the pattern and start autocompleting semi-automatically (semi because you still have to put the cursor to the right place but then you can tab, tab, tab…).

r14c19 hours ago

From my perspective, writing out the requirements for an AI to produce the code I want is just as easy as writing it myself. There are some types of boilerplate code that I can see being useful to produce with an LLM, but I don't write them often enough to warrant actually setting up the workflow.

Even with the debugging example, if I just read what I wrote I'll find the bug because I understand the language. For more complex bugs, I'd have to feed the LLM a large fraction of my codebase and at that point we're exceeding the level of understanding these things can have.

I would be pretty happy to see an AI that can do effective code reviews, but until that point I probably won't bother.

er4hn19 hours ago

It's reasonable to say that LLMs are not completely useless. There is also a very valid case to make that LLMs are not good at generating production ready code. I have found asking LLMs to make me Nix flakes to be a very nice way to make use of Nix without learning the Nix language.

As an example of not being production ready: I recently tried to use ChatGPT-4 to provide me with a script to manage my gmail labels. The APIs for these are all online, I didn't want to read them. ChatGPT-4 gave me a workable PoC that was extremely slow because it was using inefficient APIs. It then lied to me about better APIs existing and I realized that when reading the docs. The "vibes" outcome of this is that it can produce working slop code. For the curious I discuss this in more specific detail at: https://er4hn.info/blog/2024.10.26-gmail-labels/#using-ai-to...

Aeolun18 hours ago

I find a recurring theme in these kind of comments where people seem to blame their laziness on the tool. The problem is not that the tools are imperfect, it’s that you apparently use them in situations where you expect perfection.

Does a carpenter blame their hammer when it fails to drive in a screw?

er4hn18 hours ago

I'd argue that a closer analogy is I bought a laser based measuring device. I point it a distant point and it tells me the distance from the tip of the device to that point. Many people are excited that this tool will replace rulers and measuring tapes because of the ease of use.

However this laser measuring tool is accurate within a range. There's a lot of factors that affect it's accuracy like time of day, how you hold it, the material you point it at, etc. Sometimes these accuracy errors are minimal, sometimes they are pretty big. You end up getting a lot of measurements that seem "close enough". but you still need to ask if each one is correct. "Measure Twice, Cut Once" begins to require one measurement with the laser tool and once with the conventional tool when accuracy matters.

One could have a convoluted analogy where the carpenter has an electric hammer that for some reason has a rounded head that does cause some number of nails to not go in cleanly, but I like my analogy better :)

johnnyanmac14 hours ago

>Does a carpenter blame their hammer when it fails to drive in a screw?

That's the exact problem. I have plenty of screwdrivers but there's so much pressure from people not in carpentry telling me to use this shiny new army Swiss knife contraption. Will it work? Probably, if I'm just screwing in a few screws. Would I readily abandon my set of precision built, magnetic tip, etc. Screwdriver set for it? Definitely not.

I'm sure it's great for non-carpenters to have so many tools in so small a space. But I developed skills and tools already. My job isn't just to screw in a few screws a day and call it quits. People wanting to replace me for a quarter the cost for this Swiss army carpenter will quickly see a quality difference and realize why it's not a solution to everything.

Or in the software sense, maybe they are fine with unlevel shelves and hanging nails in carpet. It's certainly not work I'd find acceptable.

johnnyanmac15 hours ago

> I think most programmers who claim that LLMs aren't useful are reacting emotionally. They don't want LLMs to be useful because, in their eyes, that would lower the status of programming.

I think revealing the domain each programmer works in and asking in hose domains would reveal obvious trends. I imagine if you work in Web that you'll get workable enough AI gen code, but something like High Performance computing would get slop worse than copying and lasting the first result on Stackoverflow.

A model is only as good as its learning set, and not all types are code are readily able to be indexable.

adriand19 hours ago

> Lots of non-trivial programming tasks can be broken down into pure, trivial sections. Then, the non-trivial part becomes knowing how the entire system fits together.

I think that’s exactly right. I used to have to create the puzzle pieces and then fit them together. Now, a lot of the time something else makes the piece and I’m just doing the fitting together part. Whether there will come a day when we just need to describe the completed puzzle remains to be seen.

boringg16 hours ago

Trivial is fine but as you compound all the triviality the system starts to have a difficult time with putting it together. I don't expect it to nail it but then you have to unwind everything and figure out the issues so it isn't all gravy - fair bit of debug.

shinycode17 hours ago

It’s always harder to build a mental model of the code written by someone else. No matter what, if you trust an LLM on small things in the long run you’ll trust it for bigger things. And the most code the LLM writes, the harder it is to build this mental construct. In the end it’ll be « it worked on 90% of cases so we trust it ». And who will debug 300 millions of code written by a machine that no one read based on trust ?

jolt4218 hours ago

They are useful, but so far, I haven't seen LLMs being obviously more useful than stackoverflow. It might generate code closer to what I need than what I find already coded, but it also produces buggier code. Sometimes it will show me a function I wasn't aware of or approach I wouldn't have considered, but I have to balance that with all the other attempts that didn't produce something useful.

jerb16 hours ago

Yes. Productivity tools make programmer time more valuable, not less. This is basic economics. You’re now able to generate more value per hour than before.

(Or if you’re being paid to waste time, maybe consider coding in assembly?)

So don’t be afraid. Learn to use the tools. They’re not magic, so stop expecting that. It’s like anything else, good at some things and not others.

Reason07719 hours ago

A good farmer isn’t likely to complain about getting a new tractor. But it might put a few horses out of work.

derefr18 hours ago

I would add that a lot of the time when I'm programming, I'm an expert on the problem domain but not the solution domain — that is, I know exactly what the pseudocode to solve my problem should look like; but I'm not necessarily fluent in the particular language and libraries/APIs I happen to have to use, in the particular codebase I'm working on, to operationalize that pseudocode.

LLMs are great at translating already-rigorously-thought-out pseudocode requirements, into a specific (non-esoteric) programming language, with calls to (popular) libraries/APIs of that language. They might make little mistakes — but so can human developers. If you're good at catching little mistakes, then this can still be faster!

For a concrete example of what I mean:

I hardly ever code in JavaScript; I'm mostly a backend developer. But sometimes I want to quickly fix a problem with our frontend that's preventing end-to-end testing; or I want to add a proof-of-concept frontend half to a new backend feature, to demonstrate to the frontend devs by example the way the frontend should be using the new API endpoint.

Now, I can sit down with a JS syntax + browser-DOM API cheat-sheet, and probably, eventually write correct code that doesn't accidentally e.g. incorrectly reject reject zero or empty strings because they're "false-y", or incorrectly interpolate the literal string "null" into a template string, or incorrectly try to call Element.setAttribute with a boolean true instead of an empty string (or any of JS's other thousand warts.) And I can do that because I have written some JS, and have been bitten by those things, just enough times now to recognize those JS code smells when I see them when reviewing code.

But just because I can recognize bad JS code, doesn't mean that I can instantly conjure to mind whole blocks of JS code that do everything right and avoid all those pitfalls. I know "the right way" exists, and I've probably even used it before, and I would know it if I saw it... but it's not "on the tip of my tongue" like it would be for languages I'm more familiar with. I'd probably need to look it up, or check-and-test in a REPL, or look at some other code in the codebase to verify how it's done.

With an LLM, though, I can just tell it the pseudocode (or equivalent code in a language I know better), get an initial attempt at the JS version of it out, immediately see whether it passes the "sniff test"; and if it doesn't, iterate just by pointing out my concerns in plain English — which will either result in code updated to solve the problem, or an explanation of why my concern isn't relevant. (Which, in the latter case, is a learning opportunity — but one to follow up in non-LLM sources.)

The product of this iteration process is basically the same JS code I would have written myself — the same code I wanted to write myself, but didn't remember exactly "how it went." But I didn't have to spend any time dredging my memory for "how it went." The LLM handled that part.

I would liken this to the difference between asking someone who knows anatomy but only ever does sculpture, to draw (rather than sculpt) someone's face; vs sitting the sculptor in front of a professional illustrator (who also knows anatomy), and having the sculptor describe the person's face to the illustrator in anatomical terms, with the sketch being iteratively improved through conversation and observation. The illustrator won't perfectly understand the requirements of the sculptor immediately — but the illustrator is still a lot more fluent in the medium than the sculptor is; and both parties have all the required knowledge of the domain (anatomy) to communicate efficiently about the sculptor's vision. So it still goes faster!

coliveira15 hours ago

> people who can do it will be high status

They don't have high status even today, imagine in a world where they will be seen as just reviewers for AI code...

uh_uh15 hours ago

> They don't have high status even today

Try putting on a dating website that you work at Google vs you work in agriculture and tell us which yielded more dates.

+1
johnnyanmac14 hours ago
wvenable20 hours ago

> Or do they have 25% trivial code?

We all have probably 25% or more trivial code. AI is great for that. I have X (table structure, model, data, etc) and I want to make Y with it. A lot of code is pretty much mindless shuffling data around.

The other thing is good for is anything pretty standard. If I'm using a new technology and I just want to get started with whatever is the best practice, it's going to do that.

If I ever have to do PowerShell (I hate PowerShell), I can get AI to generate pretty much whatever I want and then I'm smart enough to fix any issues. But I really don't like starting from nothing in a tech I hate.

lambdasquirrel19 hours ago

I’ve already had one job interview where the applicant seemed broadly knowledgeable about everything we asked them during lead-in questions before actual debugging. Then when they had to actually dig deeper or demonstrate understanding while solving some problem, they fell short.

I’m pretty sure they weren’t the first and there’ve been others we didn’t know about. So now I don’t ask lead-in questions anymore. Surprisingly, it doesn’t seem to make much of a difference and I don’t need to get burned again.

randomNumber719 hours ago

Yes but then it would be more logical to say "AI makes our devs 25% more efficient". This is not what he said, but imo you are obviously right.

wvenable19 hours ago

Not necessarily. If 25% of the code is written by AI but that code isn't very interesting or difficult, it might not be making the devs 25% more efficient. It could even possibly be more but, either way, these are different metrics.

johannes123432119 hours ago

The benefit doesn't translate 1:1. The generated code has to be read and verified and might require small adaptions. (Partially that can be done by AI as well)

But for me it massively improved all the boilerplate generic work. A lot of those things which are just annoying work, but not interesting.

Then I can focus on the bigger things, on the important parts.

groestl20 hours ago

> do they have 25% trivial code?

From what I've seen on Google Cloud, both as a user and from leaked source code, 25% of their code is probably just packing and unpacking of protobufs.

hughesjj18 hours ago

I'd bet at least 25% of code attributes to me in gitfarm at Amazon was generated by octane and/or bones.

God I miss that, thanks for the other person on HN introducing me to projen. Yeoman wasnt cutting it.

These days I write a surprising amount of shell script and awk with LLMs. I review and adapt it, of course, but for short snippets of low context scripting it's been a huge time saver. I'm talking like 3-4, up to 20 lines of POSIX shell.

Idk. Some day I'll actually learn AWK, and while I've gotten decent with POSIX shell (and bash), it's definitely been more monkey see monkey do than me going over all the libraries and reference docs like I did for python and the cpp FAQ.

akira250120 hours ago

> isn't this announcement a terrible indictment

Of obviously flawed corporate structures. This CEO has no particular programming expertise and most of his companies profits do not seem to flow from this activity. I strongly doubt he has a grip on the actual facts here and is uncritically repeating what was told to him in a meeting.

He should, given his position, been the very _first_ person to ask the questions you've posed here.

nimithryn16 hours ago

An example:

I'm looking for a new job, so I've been grinding leetcode (oof). I'm an experienced engineer and have worked at multiple FAANGs, so I'm pretty good at leetcode.

Today I solved a leetcode problem 95% of the way to completion, but there was a subtle bug (maybe 10% of the test cases failing). I decided to see if Claude could help debug the code.

I put the problem and the code into Claude and asked it to debug. Over the course of the conversation, Claude managed to provide 5 or 6 totally plausible but also completely wrong "fixes". Luckily, I am experienced enough at leetcode, and leetcode problems are simple enough, that I could easily tell that Claude was mistaken. Note that I am also very experienced with prompt engineering, as I ran a startup that used prompt engineering very heavily. Maybe it's a skill issue (my company did fail, hence why I need a job), but somehow I doubt it.

Eventually, I found the bug on my own, without Claude's help. But leetcode are super simple, with known answers, and probably mostly in the training set! I can't imagine writing a big system and using an LLM heavily.

Similarly, the other day I was trying to learn about e-graphs (the data structure). I went to Claude for help. I noticed that the more I used Claude, the more confused I became. I found other sources, and as it turns out, Claude was subtly wrong about e-graphs, an uncommon but reasonably well-researched data structure! Once again, it's lucky I was able to recognize that something was up. If the problem wasn't limited in scope, I'd have been totally lost!

I use LLMs to help me code. I'm pro new technology. But when I see people bragging on Twitter about their fully automated coding solutions, or coding complex systems, or using LLMs for medical records or law or military or other highly critical domains, I seriously question their wisdom and experience.

bluerooibos18 hours ago

At what point are people going to stop shitting on the code that Copilot or other LLM tools generate?

> how trivial the problems they solve are

A single line of code IS trivial. Simple code is good code. If I write the first 3 lines of a complex method and I let Copilot complete the 4th, that's 25% of my code written by an LLM.

These tools have exploded in popularity for good reason. If they were no good, people wouldn't be using them.

I can only assume people making such comments don't actually code on a daily basis and use these tools daily. Either that or you haven't figured out the knack of how to make it work properly for you.

thegrim3316 hours ago

These tools have exploded in popularity for good reason. If they were no good, people wouldn't be using them.

You're saying anything that's ever been popular is popular for a good reason? You can't think of counter examples that disprove this?

You're saying anything that people decide to do is good, or else people wouldn't do it? People never act irrationally? People never blindly act on trends? People never sacrifice long-term results for short-term gain? You can't come up with any counter examples?

nijave15 hours ago

remembers Bitcoin et al

ghosty14118 hours ago

I havent seen anybody use them and be more productive.

With c++ my experience is that the results are completely worthless. It saves you from writing a few keywords but nothing that really helps in a big way.

Yes Copilot CAN work, for example writing some JS or filter functions, but in my job these trivial snippets are rather uncommon.

I‘d genuinely love to see some resources that show its usefulness that arent just PR bs.

fuzzy218 hours ago

I'll just answer here, but this isn't about this post in particular. It's about all of them. I've been struggling with a team of junior devs for the past months. How would I describe the experience? It's easy: just take any of these posts, replace "AI" with "junior dev", done.

Except of course AI at least can do spelling. (Or at least I haven't encountered a problem in that regard.)

I'm highly skeptical regarding LLM-assisted development. But I must admit: it works. If paired with an experienced senior developer. IMHO it must not be used otherwise.

palata18 hours ago

Isn't the whole point of hiring a junior dev that they will learn and become senior devs eventually?

johnnyanmac14 hours ago

Your mindset is sadly a decade put of touch. Companies long since shifted to churn mentality. They not only slashed retention perks, they actively expect people to move around every few years. So they don't bother stopping them or counter offering unless they are a truly exceptional person.

alfiedotwtf18 hours ago

> replace "AI" with "junior dev", done.

Damn, that’s a good way of putting it. But I’ll go one further:

replace "AI" with "junior dev who doesn’t like reading documentation or googling how things work so instead confidently types away while guessing the syntax and API so it kind of looks right”

hughesjj18 hours ago

I've been saying it's like an intern who has an incredible breadth of knowledge but very little depth, is excessively over confident in their own abilities given the error rates they commit, and is anxious to the point they'll straight up lie to you rather than admit a mistake.

Currently, they don't learn skills as fast as a motivated intern. A stellar intern can go from no idea to "makes relevant contributions to our product with significant independence and low error rate" (hi Matt if you ever see this) in 3 months. LLMs, to my understanding, take significantly more attention from super smart people working long hours and an army of mechanical Turks, but won't be able to independently implement a feature and will still have a higher error rate in the same 3 months.

It's still super impressive what LLMs can do, but that same intern is going to keep growing at that faster rate in skills and competency as they go from jr->mid->sr. Sure the intern won't have as large of a competency pool, and takes longer to respond to any given question, but the scope of what they can implement is so much greater.

skissane20 hours ago

> To my experience, AIs can generate perfectly good code relatively easy things, the kind you might as well copy&paste from stackoverflow, and they'll very confidently generate subtly wrong code for anything that's non-trivial for an experienced programmer to write. How do people deal with this?

Well, just in the last 24 hours, ChatGPT gave me solutions to some relatively complex problems that turned out to be significantly wrong.

Did that mean it was a complete waste of my time? I’m not sure. Its broken code gave me a starting point for tinkering and exploring and trying to understand why it wasn’t working (even if superficially it looked like it should). I’m not convinced I lost anything by trying its suggestions. And I learned some things in the process (e.g. asyncio doesn’t play well together with Flask-Sock)

aiforecastthway17 hours ago

I decided to go into programming instead of becoming an Engineer because most Engineering jobs seemed systematic and boring. (Software Engineers weren't really a thing at the time.)

For most of my career, Software Engineering was a misnomer. The field was too young, and the tools used changed too quickly, for an appreciable amount of the work to be systematic and boring enough to consider it an Engineering discipline.

I think we're now at the point where Software Engineering is... actually Engineering. Particularly in the case of large established companies that take software seriously, like Google (as opposed to e.g. a bank).

Call it "trivial" and "boring" all you want, but at some point a road is just a road, and a train track is just a train track, and if it's not "trivial and boring" then you've probably fucked up pretty badly.

javaunsafe201917 hours ago

Since when is engineering boring? Stranges ideas and claims u made.

I’m an engineer who writes code since 20 years and it’s far away from trivial . Maybe to do web dev for a simple Webshop is. Elsewhere software has often times special requirements. Be them technical or domain wise both make the process complex and not simple IMHO

aiforecastthway17 hours ago

Boring is the opposite of exciting/dynamic.

Not all engineering is boring. Also, boring is not bad.

A lot of my career has been spent working to make software boring. To the extent that I've helped contribute to the status quo, where we can build certain types of software in a relatively secure fashion and on relatively predictable timelines, I am proud to have made the world more boring!

(Also, complexity can be extraordinarily boring. Some of the most complex things are also the most boring. Nothing more boring than a set of business rules that has an irreducible complexity coming in at 5,211 lines of if-else blocks wrapped in two while loops! Give me a simple set of partial differential equations any day -- much more exciting to work with those! If you're the type of person who enjoys reading tax code, then we just have different definitions of boring; and if you're the type of person doesn't think tax code is complex, then I'm just a dummy compared to you :))

But e.g. in the early naughts doing structural engineering work for residential new build projects was certainly less engaging and exciting work than building websites.

Most engineering works aims for repeatable and predictable outcomes. That's a good thing, and it's not easy to achieve! But if Software has reached the point where the process of building certain types of software is "repeatable and predictable", and if Google needs a lot of that type of software, then if the main criticism of AI code assistants is "it's only good for repeatable and predictable", well, then the criticism isn't exactly the indictment that skeptics think it is.

There is nothing wrong with boring in the sense I'm using it. Boring can be tremendously intellectually demanding. Also, predictable and repeatable processes are incredibly important if you want quality work at scale. Engineering is a good thing. Maturing as a field is a good thing.

But if we're maturing from "wild west everything is a greenfield project" to "70% of things are pretty systematic and repeatable" then that says something about the criticism of AI coding assistants as being only good for the systematic and repeatable stuff, right?

Also: the AI coding assistant paradigm is coming for structural/mechanical/civil engineering next, and in a big way!

+1
sally_glance16 hours ago
JohnMakin19 hours ago

> To my experience, AIs can generate perfectly good code relatively easy things, the kind you might as well copy&paste from stackoverflow,

This, imho, is what is happening. In the olden days, when StackOverflow + Google used to magically find the exact problem from the exact domain you needed every time - even then you'd often need to sift through the answers (top voted one was increasingly not what you needed) to find what you needed, then modify it further to precisely fit whatever you were doing. This worked fine for me for a long time until search rendered itself worthless and the overall answer quality of StackOverflow has gone down (imo). So, we are here, essentially doing the exact same thing in a much more expensive way, as you said.

Regarding future employment opportunities - this rot is already happening and hires are coming from it, at least from what I'm seeing in my own domain.

eco20 hours ago

I'd be terribly scared to use it in a language that isn't statically typed with many, many compile time error checks.

Unless you're the type of programmer that is writing sabots all day (connecting round pegs into square holes between two data sources) you've got to be very critical of what these things are spitting out.

int_19h16 hours ago

I can't help but think that Go might be one of the better languages for AI to target - statically typed, verbose with a lot of repeated scaffolding, yet generally not that easy to shoot yourself in the foot. Which might explain why this is a thing at Google specifically.

randomNumber719 hours ago

It is way more scary to use it for C or C++ than Python imo.

cybrox19 hours ago

If you use it as advanced IntelliSense/auto-complete, it's not any worse than with typed languages.

If you just let it generate and run the code... yeah, probably, since you won't catch the issues at compile time.

grepLeigh18 hours ago

I have a whole "chop wood, carry water" speech born from leading corporate software teams. A lot of work at a company of sufficient size boils down to keeping up with software entropy while also chipping away at some initiative that rolls up to an OKR. It can be such a demotivating experience for the type of smart, passionate people that FANNGs like to hire.

There's even a buzzword for it: KTLO (keep the lights on). You don't want to be spending 100% of your time on KTLO work, but it's unrealistic to expect to do done of it. Most software engineers would gladly outsource this type of scutwork.

girvo13 hours ago

> KTLO (keep the lights on)

Some places also call this "RTB" for "run the business" type work. Nothing but respect for the engineers who enjoy that kind of approach, I work with several!

dmurray18 hours ago

> Or do they have 25% trivial code?

Surely yes.

I (not at Google) rarely use the LLM for anything more than two lines at a time, but it writes/autocompletes 25% of my code no problem.

I believe Google have character-level telemetry for measuring things like this, so they can easily count it in a way that can be called "writing 25% of the code".

Having plenty of "trivial code" isn't an indictment of the organisation. Every codebase has parts that are straightforward.

geodel14 hours ago

> Like, isn't this announcement a terrible indictment of how inexperienced their engineers are..

Well, Rob Pike said same thing about experience and that seemed to pissed lot of people endlessly.

However I don't think it as indictment It just seems very reasonable to me. In fact 25% seem to be on lower end. Amazon seems to have thousands of software engineers who are doing API calling API calling API.. kind of crap. Now their annual income might be more than my lifetime earnings. But to think that all these highly paid engineers are doing highly complex work that need high skills seems just a myth that is useful to boost ego of engineers and their employers alike.

hifromwork21 hours ago

25% trivial code sounds like a reasonable guess.

fzysingularity20 hours ago

This seems reasonable - but I'm interpreting this as most junior-level coding needs will end and be replaced with AI.

mrguyorama20 hours ago

And the non junior developers will then just magically appear from the aether!With 10 years experience in a four year old stack.

herval15 hours ago

In my experience, that was always the case with gpt3.5, most times the case with gpt4, some times the case with the latest sonnet. It’s getting better FAST, and the kind of code they can handle is increasing fast too

pjmorris16 hours ago

> Like, isn't this announcement a terrible indictment of how inexperienced their engineers are, or how trivial the problems they solve are, or both?

Or maybe there's a KPI around lines of code or commits.

ants_everywhere19 hours ago

Google's internal codebase is nicer and more structured than the average open source code base.

Their internal AI tools are presumably trained on their code, and it wouldn't surprise me if the AI is capable of much more internally than public coding AIs are.

sangnoir20 hours ago

> Does Google now have 25% subtly wrong code?

How do you quantify "new code" - is it by lines of code or number of PRs/changesets generated? I can easily see it being the latter - if an AI workflow suggests 1 naming-change/cleanup commit to your PR made of 3 other human-authored commits, has it authored 25% of code? Arguably, yes - but it's trivial code that ought to be reviewed by humans. Dependabot is responsible for a good chunk of PRs already.

Having a monorepo brings plenty of opportunities for automation when refactoring - whether its AI, AST manipulation or even good old grep. The trick is not to merge the code directly, but have humans in the loop to approve, or take-over and correct the code first.

asdfman12316 hours ago

No, AI is generating a quarter of all characters. It's an autocomplete engine. You press tab, it finishes the line. Doesn't do any heavy lifting at all.

Source: I work there, see my previous comment.

afavour19 hours ago

> Or do they have 25% trivial code?

If anything that's probably an underestimate. Not to downplay the complexity in much of what Google does but I'm sure they also do an absolute ton of tedious, boring CRUD operations that an AI could write.

djvuvtgcuehb18 hours ago

A better analogy is a self driving car where you need to keep your hands on the wheel in case something goes wrong.

For the most part, it drives itself.

Yes, the majority of my code is trivial. But I've also had ai iterate on some very non trivial work including writing the test suite.

It's basically autocomplete on steroids that predicts your next change in the file, not just the next change on the line.

The copy paste from stack overflow trope is a bit weird, I haven't done that in ten years and I don't think the code it produces is that low quality either. Copy paste from an open source repo on GitHub maybe?

fsckboy18 hours ago

> Does Google now have 25% subtly wrong code?

maybe the ai generates 100% of the company's new code, and then by the time the programmers have fixed it, only 25% is left of the AI's ship of Theseus

manquer16 hours ago

if their sales and stock depends on saying that new shinny thing is changing the world then they have to say so, and say how it is changing their world .

It is not Netflix or Airbnb or Stripe etc making this claim, google managers have a vested interest in this.

If this metric was meaningful either of two things should have happened - google should have fired 25 % developers or built 25 % more product .

Both of this would visible in their financial reporting and has not happened.

metrics like this claim depends on how you count, that is easily gamed and can be made to show any % between 0-99 you want. Of the top of head

- I could count all AI generated code used for training as new code

- consider compiler output to assembly as AI code by adding some meaningless AI step in it

- code generated with boilerplate perhaps even generated by llm now

- mix autocomplete with llm prompts so on

The number only needs to believable , 25 is believable now, it is not true but you would believe it >50 has psychological significance and bad PR on machines replacing humans jobs , less than 10 is bad for AI sales , 25 works all the commenters in this thread is testament to that

airstrike21 hours ago

By definition, "trivial" code should make up a significant portion of any code base, so perhaps the 25% is precisely the bit that is trivial and easily automated.

Smaug12320 hours ago

I don't think the word "definition" means what you think it means!

signa1117 hours ago

``` Like, isn't this announcement a terrible indictment of how inexperienced their engineers are, or how trivial the problems they solve are, or both? ```

there is a 3rd possibility as well: having spent a huge chunk of change on these techniques, why not overhype it (not outright lie about it) and hope to, somewhat recoup the cost from unsuspecting masses ?

sally_glance17 hours ago

I guess the obvious response would be - yes, they have _at least_ 25% trivial code (as any other enterprise), and yes, they should have lots of engineers 'babysitting' (aka generating training data). So in another year or two there will be no manpower at all needed for the trivial tasks.

cybrox19 hours ago

Depends if they include test code in this metric. I have found AI most valuable in generating test code. I usually want to keep tests as simple as possible, so I prefer some repetition over abstraction to make sure there's no issues with the test logic itself, AI makes this somewhat verbose process very easy and efficient.

Cthulhu_18 hours ago

You're quick to jump to the assertion that AI only generates SO style utility code to do X, but it can also be used to generate boring mapping code (e.g. to/from SQL datasets). I heard one ex Google dev say that most of his job wat fiddling with Protobuf definitions and payloads.

skeeter202020 hours ago

trivial code could very easily include the vast majority of most apps we're building these days. Most of it's just glue, and AI can probably stitch together a bunch of API calls and some UI as well as a human. It could also be a lot of non-product code, tooling, one-time things, etc.

aorloff19 hours ago

Its been a while since I was really fully in the trenches, but not that long.

How people deal with this is they start by writing the test case.

Once they have that, debugging that 25% comes relatively easily and after that its basically packaging up the PR

andyjohnson020 hours ago

I suspect that a lot of the hard, google-scale stuff has already been done and packaged as an internal service or library - and just gets re-used. So the AIs are probably churning out new settings dialogs and the like.

nwellinghoff20 hours ago

They probably have ai that scans existing human written code and auto generates patches and fixes to improve performance or security. The 25% is just a top level stat with no real meaning without context.

jjtheblunt20 hours ago

Maybe the trick is to hide vetted correct code, of whatever origin, behind function calls for documented functions, thereby iteratively simplifying the work a later-trained LLM would need to do?

notyourwork19 hours ago

To your point, I don't buy the truth of the statement. I work in big tech and am convinced that 25% of the code being written is not coming from AI.

rh2323o4jl23414 hours ago

> Does Google now have 25% subtly wrong code?

I think you underestimate the amount of boiler-plate code that a typical job at Google requires. I found it soul-crushingly boring (though their pay is insane).

ZiiS18 hours ago

Yes 25% of code is trivial; certainly for companies like Google that have always been a bit NIH.

tmoravec21 hours ago

Does the figure include unit tests?

ithkuil20 hours ago

Or perhaps that even for excellent engineers and complicated problems a quarter of the code one writes is stupid almost copy-pasteable boilerplate which is now an excellent target for the magic lArge text Interpolator

Kiro19 hours ago

You're doing circular reasoning based on your initial concern actually being a problem in practice. In my experience it's not, which makes all your other speculations inherently incorrect.

TacticalCoder16 hours ago

> and they'll very confidently generate subtly wrong code for anything that's non-trivial for an experienced programmer to write

Thankfully I don't find it subtle but plain wrong for anything but trivial stuff. I use it (and pay an AI subscription) for things where false positive won't ruin the day, like parameters validation.

But for anything advanced, it's pretty hopeless.

I've talked with lawyers: same thing. With doctors: same thing.

Which ain't no surprise see how these things do work.

> Like, isn't this announcement a terrible indictment of how inexperienced their engineers are, or how trivial the problems they solve are, or both?

Probably lots of highly repetitive boilerplate stuff everywhere. Which in itself is quite horrifying if you think about it.

dyauspitr18 hours ago

This subtly wrong thing happens maybe 10% of the time in my experience and asking it to generate unit tests or writing your own ahead of time almost completely eliminates it.

uoaei18 hours ago

I've suspected for a while now that the people who find value in AI-generated code don't actually have hard problems to solve. I wonder how else they might justify their salary.

vkou18 hours ago

How would you react to a tech firm that in 2018, proudly announced that 25% of their code was generated by IntelliJ/Resharper/Visual Studio's codegen and autocomplete and refactoring tools?

jajko20 hours ago

I can generate in eclipse pojo classes or their accessor methods. I can let maven build entire packages from say XSDs (I know I am talking old boring tech, just giving an example). I can copy&paste half the code (if not more) from stack overflow.

Now replace all this and much more with 'AI'. If they said AI helped them increase say ad effectivity by 3-5%, I'll start paying attention.

Nasrudith20 hours ago

I wouldn't call it an indictment necessarily, because so much is dependent upon circumstances. They can't all be "deep problems" in the real world. Projects tend to have two components, "deep" work which is difficult and requires high skill and cannot be made up with by using masses of inexperienced and "shallow" work where being skilled doesn't really help, or doesn't help too much compared to throwing more bodies at the problem. To use an example it is like advanced accounting vs just counting up sales receipts.

Even if their engineers were inexperienced that wouldn't be an indictment in itself so long as they had a sufficient necessary amount of shallow work. Using all experienced engineers to do shallow work is just inefficient, like having brain surgeons removing bunions. Automation is basically a way to transform deep work to a producer of "free" shallow work.

That said, the real impressive thing with code isn't in its creation but in its ability to losslessly delete code and maintain or improve functionality.

name_nick_sex_m18 hours ago

[flagged]

llm_trw15 hours ago

Or alternatively you don't know how to use AI to help you code and are in the 2020s equivalent of the 'Why do I need google when I have the yellow pages?' phase a lot of adults went through in the 2000s.

This is not a bad thing since you can improve, but constantly dismissing something that a lot of people are finding an amazing productivity boost should give you some pause.

johnnyanmac14 hours ago

It's like blockchain right now. I'm sure there is some killer feature that can justify its problem space.

But as of now the field is full of swamps. Of grifters, of people with a solution looking for a problem. Of outright scams of questionable legality being challenged as we speak.

I'll wait until the swamps work itself out before evaluating an LLM workflow.

llm_trw12 hours ago

Blockchain was always a solution looking for a problem.

LLMs are being used right now by a lot of people, myself included, to do tasks which we would have never bothered with before.

Again, if you don't know how to use them you can learn.

johnnyanmac11 hours ago

And the same was said with the last fad when Blockcbain was all investors wanted to hear about ("Big Data" I suppose). It's all a pattern.

It's a legal nightmare in my domain as of now, so I'll make sure the Sam Breaker-Friends are weeded out. If it's really all the hype it won't be going anywhere in 5 years.

nine_zeros2 days ago

Writing more code means more needs to be maintained and they are cleverly hiding that fact. Software is a lot more like complex plumbing than people want to admit:

More lines == more shit to maintain. Complex lines == the shit is unmanageable.

But wall street investors love simplistic narratives such as More X == More revenue. So here we are. Pretty clever marketing imo.

bamboozled6 hours ago

"Product has not improved, or maybe even become worse in that time"

evbogue2 days ago

I'd be turning off the autocomplete in my IDE if I was at Google. Seems to double as a keylogger.

throwaway29013 hours ago

"More than a quarter of our code is created by autocomplete!"

That's not that much...

AI_beffr15 hours ago

i like how people say that ai can only write "trivial" code well or without mistakes. but what about from the point of view of the AI? writing "trivial" code is probably almost exactly as much of a challenge as writing the most complex code a human could ever write. the scales are not the same. dont allow yourself to feel so safe..

pixelat3d1 day ago

[flagged]

jrockway2 days ago

When I was there, way more than 25% of the code was copying one proto into another proto, or so people complained. What sort of memes are people making now that this task has been automated?

hn_throwaway_992 days ago

I am very interested in how this 25% number is calculated, and if it's a lot of boilerplate that in the past would have been just been big copy-paste jobs like a lot of protobuffers work. Would be curious if any Googlers could comment.

Not that I'm really discounting the value of AI here. For example, I've found a ton of value and saved time getting AI to write CDKTF (basically, Terraform in Typescript) config scripts for me. I don't write Terraform that often, there are a ton of options I always forget, etc. So asking ChatGPT to write a Terraform config for, say, a new scheduled task for example saves me from a lot of manual lookup.

But at the same time, the AI isn't really writing the complicated logic pieces for me. I think that comes down to the fact that when I do need to write complicated logic, I'm a decent enough programmer that it's probably faster for me to write it out in a high-level programming language than write it in English first.

dietr1ch2 days ago

I miss old memegen, but it got ruined by HR :/

rcarmo1 day ago

I am reliably told that it is alive and well, even if it’s changed a bit.

anon124319 hours ago

Memegen is there but unrecognizable now. A dedicated moderator team deletes memes, locks comments, bans people for mentioning "killing a process" (threatening language!) and contacts their managers.

dietr1ch19 hours ago

Yup, I simply stopped using it, which means they won.

kev0092 days ago

I would hope a CEO, especially a technical one, would have enough sense to couple that statement to some useful business metric, because in isolation it might be announcement of public humiliation.

dmix2 days ago

The elitism of programmers who think the boilerplate code they write for 25% of the job, that's already been written before by 1000 other people before, is in fact a valuable use of company time to write by hand again.

IMO it's only really an issue if a competent human wasn't involved in the process, basically a person who could have written it if needed, then they do the work connecting it to the useful stuff, and have appropriate QA/testing in place...the latter often taking far more effort than the actual writing-the-code time itself, even when a human does it.

marcosdumay2 days ago

If 25% of your code is boilerplate, you have a serious architectural problem.

That said, I've seen even higher ratios. But never in any place that survived for long.

hn_throwaway_992 days ago

Depends on how you define "boilerplate". E.g. Terraform configs count for a significant number of the total lines in one of my repos. It's not really "boilerplate" in that it's not the exact same everywhere, but it is boilerplate in the since that setting up, say, a pretty standard Cloud SQL instance can take many, many lines of code just because there are so many config options.

marcosdumay1 day ago

Terraform is verbose.

It's only boilerplate if you write it again to set almost the same thing again. What, granted, if you are writing bare terraform config, it's probably both.

But on either case, if your terraform config is repetitive and a large part of the code on an entire thing (not a repo, repos are arbitraty divisions, maybe "product", but it's also a bad name). Than that thing is certainly close to useless.

TheNewsIsHere1 day ago

To add: it’s been my experience that it’s the company that thinks the boilerplate code is some special, secret, proprietary thing that no other business could possibly have produced.

Not the developer who has written the same effective stanza 10 times before.

wvenable20 hours ago

25% of new code might be boilerplate. All my apps in my organization start out roughly the same way with all the same stuff. You could argue on day one that 100% of the code is boilerplate and by the end of the project it is only a small percentage.

8note2 days ago

Is it though? It seems to me like a team ownership boundary question rather than an architecture question.

Architecturally, it sounds like different architecture components map somewhere close to 1:1 to teams, rather than teams hacking components to be closer coupled to each other because they have the same ownership.

I'd see too much boilerplate as being a organization/management org issue rather than a code architecture issue

cryptoz2 days ago

Android mobile development has gotten so …architectured that I would guess most apps have a much higher rate of “boilerplate” than you’d hope for.

Everything is getting forced into a scalable, general purpose way, that most apps have to add a ridiculous amount of boilerplate.

dmix2 days ago

You're probably thinking of just raw codebases, your company source code repo. Programmers do far, far more boilerplate stuff than raw code they commit with git. Debugging, data processing, system scripts, writing SQL queries, etc.

Combine that with generic functions, framework boilerplate, OS/browser stuff, or explicit x-y-z code then your 'boilerplate' (ie repetitive, easily reproducible) easily gets to 25% of code you're programmers write every month. If your job is >75% pure human cognition problem solving you're probably in a higher tier of jobs than the vast majority of programmers on the planet.

kev0092 days ago

Doing the same thing but faster might just mean you are masturbating more furiously. Show me the money, especially from a CEO.

mistrial92 days ago

you probably underestimate the endless miles of verbose code that are possible, by human or machine but especially by machine.

dyauspitr2 days ago

Or a statement of pride that the intelligence they created is capable of lofty tasks.

joeevans10002 days ago

I read these threads and the usual 'I have to fix the AI code for longer than it would have taken to write it from scratch' and can't help but feel folks are truly trying to downplay what is going to eat the software industry alive.

steve_adams_8616 hours ago

I’m not convinced it’s there yet. I think it’s actively eating part of the software industry, but I wonder where that’ll stop—at least for some time—and a new shape of the industry is settled upon.

There are still things I do in my IDE that I can’t seem to get AI to do. It’s not really close yet. I don’t doubt it could get there eventually, but I suppose I don’t believe it’s about to eat those parts of the industry.

I do anticipate a massive issue from lower skill software jobs vanishing. I don’t know what entry into the industry will look like. There will be a strange gap that’s filled by AI and some people who use it to do basic things but have no idea how it does it. They will be somewhat like data entry workers, knowing how to use a spreadsheet or word processor but having no idea how the program actually works let alone the underlying operating system. I fully expect that to happen, and I can’t properly imagine what the implications will be.

tylerchilds2 days ago

if the golden rule is that code is a liability, what does this headline imply?

eddd-ddde2 days ago

The code would be getting written anyways, its an invariant. The difference is less time wasted typing keys (albeit small amount of time) and more importantly (in my experience) it helps A LOT for discoverability.

With g3's immense amount of context, LLMs can vastly help you discover how other people are using existing libraries.

tylerchilds1 day ago

my experience dabbling with the ai and code is that it is terrible at coming up with new stuff unless it already exists

in regards to how others are using libraries, that’s where the technology will excel— re-writing code. once it has a stable AST to work with, the mathematical equation it is solving is a refactor.

until it has that AST that solves the business need, the game is just prompt spaghetti until it hits altitude to be able to refactor.

JimDabell1 day ago

Nothing at all. The headline talks about the proportion of code written by AI. Contrary to what a lot of comments here are assuming, it does not say that the volume of code written has increased.

Google could be writing the same amount of code with fewer developers (they have had multiple layoffs lately), or their developers could be focusing more of their time and attention on the code they do write.

contravariant19 hours ago

Well, either they just didn't spend as much time writing the code or they increased their liability by about 33%.

The truth is likely somewhere in between.

danielmarkbruce2 days ago

I'm sure google won't pay you money to take all their code off their hands.

AlexandrB2 days ago

But they would pay me money to audit it for security.

danielmarkbruce2 days ago

yup, you can get paid all kinds of money to fix/guard/check billion/trillion dollar assets..

yearolinuxdsktp14 hours ago

[dead]

an_d_rew2 days ago

Huh.

That may explain why google search has, in the past couple of months, become so unusable for me that I switched (happily) to kagi.

twarge2 days ago

Which uses Google results?

croes2 days ago

Related?

> New tool bypasses Google Chrome’s new cookie encryption system

https://news.ycombinator.com/item?id=41988648

floor_5 hours ago

So no one owns a quarter of the new code at google. It's going to be very funny when it hits 100%.

hipadev232 days ago

Google is now mass-producing techdebt at rates not seen since Martin Fowler’s first design pattern blogposts.

nelup2019 hours ago

We've now entered the age of exponential tech debt, it'll be a sight to behold

joeevans10002 days ago

Not really technical debt when you will be able to regenerate 20K lines of code in a minute then QA and deploy it automatically.

kibwen2 days ago

So a fresh, new ledger of technical debt every morning, impossible to ever pay off?

1attice2 days ago

Assuming, of course:

- You know which 20K lines need changing - You have perfect QA - Nothing ever goes wrong in deployment.

I think there's a tendency in our industry to only take the hypotenuse of curves at the steepest point

TheNewsIsHere1 day ago

That is a fantastic way to put it. I’d argue that you’ve described a bubble, which fits perfectly with the topic and where _most_ of it will eventually end up.

Tier3r2 days ago

Google is getting enshittified. It's already visible in many small ways. I was just using Google maps and in the route they called X (bus) Interchange as X International. I can only assume this happened because they are using AI to summarise routes now. Why in the world are they doing that? They have exact location names available.

FactKnower692 days ago

[flagged]

eob2 days ago

So GCS customers will trust their codegen product. (Engineers aren’t the buyer; corp suite is)

hn_throwaway_992 days ago

I don't understand why you think this at all. Care to explain?

dartharva2 days ago

Why? Especially when said AI helpers are a part of what the company itself is selling?

joeevans10002 days ago

These companies are competing to be the next codegen service provider.

foota2 days ago

Translation: They'd love to lay off all the engineers.

sfmz2 days ago

We should watch for dev layoffs as a sign/signal of the impact of generated code. I remember reading about an anime shop that fired 80% of its illustrators due to ai-images.

TheNewsIsHere1 day ago

By some intuitive measures, it’s surprising they have very many still writing their code. Google’s product quality isn’t what it once was. There is no amount of AI accelerators and energy they can burn through to fix that without humans.

lesuorac2 days ago

Well, the article has a paywall so it might go into this.

I'm not sure this stat is as important as people point it out to be. If I start of `for` and the AI auto-completes `for(int i=0; i<args.length; i++) {` then a lot more than 25% of the code is AI written but it's also not significant. I could've figured out how to write the for-loop and its also not a meaningful amount of time saved because most of the time is figuring out and testing which the AI doesn't do.

dyauspitr2 days ago

I don’t think the public cares wether their code is written by machines or real people as long as the product works.

Nullabillity2 days ago

Just today, Google Calendar asked me whether I wanted the "easy" or "depressed" colour scheme.

mattigames2 days ago

It's for when you have an upcoming funeral, the calendar it's just trying to dress appropriately.

Mistletoe2 days ago

Ironically, your comment brightened my day.

microtherion2 days ago

[flagged]

Tiktaalik2 days ago

[flagged]

calmbonsai2 days ago

[flagged]

pyuser5832 days ago

[flagged]

YPPH2 days ago

Actually 0%, assembly language is assembled to machine code, not compiled.

ndesaulniers2 days ago

Inline asm has to go through the compiler to get wired up by the register allocator.

bakugo2 days ago

[flagged]

mergisi1 day ago

[dead]

ultra_nick2 days ago

Why work at big businesses anymore? Let's just create more startups.

IAmGraydon2 days ago

Risk appetite.

game_the0ry19 hours ago

Not so sure nowadays. Given how often big tech lays off employees and the abundance of recently laid off tech talent, trying to start your own company sounds a lot more appealing than ever.

I consider myself risk-averse and even I am contemplating starting a small business in the event I get laid off.

shiroiushi13 hours ago

> trying to start your own company sounds a lot more appealing than ever.

It really isn't. Even if you get laid off from a large tech company, you probably didn't have to pay a cent to get the job there in the first place, and you started drawing a paycheck right away (after the initial delay due to the pay cycle). If you only work there for 6 months, you can save a really good amount of money if you have frugal habits.

Starting a company isn't nearly as easy, usually requires up-front investment, and there can be a long time before you generate any profit. Either you need some business idea that's going to generate profit (or at least enough revenue to give the founder(s) a paycheck), or a business loan or other funding, which means convincing someone to invest in your company somehow.

Starting your own company only sounds appealing if you ignore reality, or have the privilege of having plenty of cash saved up for such a venture.

wayoverthecloud17 hours ago

Interesting. I think the same thing but I wonder if the market is not ready for products created by the big guys, what can I offer? Have you thought in that line?

1oooqooq2 days ago

this only means employees sign up to use new toys and they are paying enough seats for all employees.

it's like companies paying all those todolist and tutorial apps left running on aws ec2 instances in 2007ish.

I'd be worried if i were a google investor. lol.

fragmede2 days ago

I'm not sure I get your point. Google created Gemini and whatever internal LLM their employees are using for code generation. Who are they paying, and for what seats? Not Microsoft or OpenAI or Anthropic...