[vga1]

How would you differentiate a 3000 line LLM commit made by the best models and good AI processes from a 3000 line commit made by the best human developer?

edit Okay, I set the bar too high here with "best human developer" and vague "good AI processes". My bad. Yes, LLM is not quite there yet.

[vurudlxtyt]

A personal relationship and trust, as seems to be the case here?

[wiseowise]

By using my brain.

[dnikolovv]

Don't be ridiculous! We don't do that anymore.

[saagarjha]

Read it?

[IshKebab]

It's still fairly obvious just by skimming the code. The best AI models are still quite far from the best human developers in ability and especially in code quality.

[matwood]

When the best AI models are the same or better than the best[1] human developers, what then?

We're already at the point talking about best vs. best.

[IshKebab]

If that happens and we have a way of reliably knowing if some code is produced to that high quality, then I think we probably can accept that AI coding is the only sensible option.

We definitely are not close to that point though and it's unclear if/when we will get there.

[vga1]

It seems to me that people might be arguing from conflicting hidden premises here. "AI Coding" is a spectrum that could mean something as simple as letting the LLM proofread your changes and then act on those with your own human brain, or it could mean just telling the agent what you want and let it rip and tear until it is done.

If I do the latter and submit a PR to something like Zig, I'll be certainly caught doing it and rightfully chastised. If I do the former, my PR will be better without anybody besides myself having any way of knowing how it got better. Probably I do something in between when I contribute to open-source these days.

Blanket banning all of these seems like a bad idea to me. It actively gates people like myself from contributing, because I respect these people and projects that much. It feels like I would be doing something they find disgusting if my work has touched an LLM and I obviously don't want to do that to people I respect. But it's fine, there are plenty of things to do in the world even when some doors are closed.

I do not presume to have any say on Zig project's well argued decisions[0] -- I'm not really even their user let alone someone important like a contributor. Their point of preferring human contact is superb, frankly. Probably a different kind of problem in an open-source project staffed with a lot of remote working people, where human contact is scarce.

https://kristoff.it/blog/contributor-poker-and-ai/

[em-bee]

Blanket banning all of these seems like a bad idea to me. It actively gates people like myself from contributing

in my projects i will reject any contribution that i do not understand. even if the contribution is handwritten by an expert developer. that developer will have to earn my trust like anyone else, like you would have too.

LLM contributions are non-deterministic, which means they can never be trusted.

therefore, if you use LLM to contribute, you can not earn my trust. if you believe that you can not create a meaningful contribution without the use of LLM then you are realizing that you are not skilled enough to understand the code that you contribute. because if you could understand it, then you could write it yourself. i want your personal contributions, not those of your LLM. i want contributions that the submitter actually understands. i want you to earn my trust by showing me that you understand what you are doing. i want you to grow your understanding of my project. none of this happens when you use LLMs.

if you are unable to make a contribution without the help of an LLM then you are not ready to contribute. try looking for smaller issues that you can work on instead until you learned enough to make larger contributions.

[SuperV1234]

> i will reject any contribution that i do not understand

Fair.

> that developer will have to earn my trust like anyone else

What does it take to "earn your trust"?

> LLM contributions are non-deterministic, which means they can never be trusted.

Provably incorrect. LLM contributions can be reviewed, tested, and understood like any other contribution. There's nothing "special" about LLM contributions.

Contributions authored by human brains are also non-deterministic, perhaps if the author was feeling in a slightly different way they'd have formatted the code a bit differently.

> therefore, if you use LLM to contribute, you can not earn my trust.

The premise is wrong.

> if you believe that you can not create a meaningful contribution without the use of LLM then you are realizing that you are not skilled enough to understand the code that you contribute

What if I believe I can do so without an LLM, but that it could be even better with an LLM?

What if I'm great at understanding code, but terrible at writing it?

Again, this is a premise that you just decided to take as truth, without proof.

> because if you could understand it, then you could write it yourself.

False. I can understand a novel algorithm by reading and studying it, but perhaps I could have not come up with it myself.

> i want you to earn my trust by showing me that you understand what you are doing

I can easily do that even if my contribution involves LLM assistance.

> i want you to grow your understanding of my project

Ditto.

> none of this happens when you use LLMs

False. Why do you think so?

> if you are unable to make a contribution without the help of an LLM then you are not ready to contribute.

Again, this is your opinion and you have no way of proving it. I can prove the opposite.

[vga1]

>because if you could understand it, then you could write it yourself.

I accept most things you said there as valid opinions, but this is where the logic goes wrong.

I use LLMs to give me more from the only resource (now that my basic and mid-level needs are largely met) that ultimately matters: time. That means that I need to waste far less time in front of the computer, typing code, and use far more time doing more useful things, like hobbies, art, being with my children.

But as I said before, every project is obviously allowed to make their own rules, and contributors should obey those rules. There are plenty of projects that take both AI deniers and plenty of projects who prefer AI aficiandos.

At least for now. My belief is that one those groups will fade away like horseback riding did, but we'll see. Perhaps you have heard the famous stages quoted by many different people in different forms: first an idea is ridiculed, then it's attacked, then it's accepted. Some open-source communities have clearly entered the attacking phase in the last year so.

[irishcoffee]

How can AI possibly be better than “the best” when the corpus of training data now includes its own slop in addition to all the code by new devs/lazy devs/bad devs scattered all over the internet? Law of averages applies here.

[vga1]

Because LLM models are obviously much more than the sum of their parts.

[irishcoffee]

Oh, which parts are those? Do tell!

[bmacho]

Don't use "the corpus", but use thinking, source code of the libraries and existing software, documentation, tools, best practices.

Billion times faster than a human, no tiring, no miscalculation, no brain-fart, no cheating.

[maccard]

The post that inspired this post [0] says:

> So while one could in theory be a valid contributor that makes use of LLMs, from the perspective of contributor poker it’s simply irrational for us to bet on LLM users while there’s a huge pool of other contributors that don’t present this risk factor.

> The people who remarked on how it’s impossible to know if a contribution comes from an LLM or not have completely missed the point of this policy and are clearly unaware of contributor poker.

The point isn't about the 3000 line PR, it's about do we think the submitter is going to stick around.

[0] https://kristoff.it/blog/contributor-poker-and-ai/

[PunchyHamster]

It seems to be trivially easy for everyone but people heavily invested into LLM to spot LLM slop