Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI has a deep understanding of how this code works

github.com

Post reply on HN
Print view View original

271–280 of 329 posts

Re: AI has a deep understanding of how this code works

#271

Post by joelreymont »

biorach wrote: post #259
joelreymont wrote: post #210

I'm the author of the PR. No, I'm not AI or bot, etc. Yes, my resume is genuine and is even more weird than what was listed (see https://joel.id/resume ). Oh, and I live in Kyiv. As for the PR itself, it was a PR stunt that I regret now as the code works and solves a real problem (at least for me!). I'll probably redo it, once I have spare Claude $$$ which I'm using for other projects now ( https://joel.id/build-your…

Tip: A list compiler should be relatively straightforward, as these things go. If you get the AI to write it you should actually read it, all of it, and understand it, to the point where you can add features and fix bugs yourself. There are many many resources on the subject. Only after this should you consider contributing to open source projects. And even then you need to be able to read and understand your contrib…

Are you speaking from experience?

Have you actually tried writing a "list" compiler?

Re: AI has a deep understanding of how this code works

#272

Post by PunchyHamster »

KurSix wrote: post #166

This is a perfect real-world illustration of Brandolini's law: the amount of energy needed to refute bullshit is an order of magnitude bigger than to produce it. The guy spent 5 minutes prompting, while Oсaml maintainers spent hours of their time politely dissecting the mess. Open Source will lose this war unless it changes the rules of engagement for contributions

I am afraid AI bumped it to at least 2 orders of magnitude

Re: AI has a deep understanding of how this code works

#273

Post by joelreymont »

Anamon wrote: post #163
xtracto wrote: post #98

This won't be a popular opinion here but, this resistance and skepticism of AI code, and people making it less smells to me very similar to the stance I see from some developers that have this belief that people from other countries CANNOT be as good as them (like, saying that outsourcing or hiring people from developing countries will invariably bring low[er] quality code). Feels a.but like snobbism and projection o…

This would be fine if LLMs generated quality code, which they don't. Anything beyond trivial and boilerplate code is either riddled with errors or copied almost verbatim. None of these systems are able to even remotely do what a competent developer does. Despite the PR author's claims, LLMs have no, and can't have any, understanding of the code. Especially when you start talking about architecture, robustness, securi…

This is an excerpt from the session where AI is writing my Lisp compiler. What do you call this? I call this doing what a competent developer does!

39/40 tests pass. The native reader works for integers, hexadecimal, lists, strings and quote forms. The one failure is symbol comparison (known limitation).

  Based on the context summary and the user's note about cleanup, I should:
  1. Continue Phase 3.5 self-compilation
  2. Clean up the repo structure

  Let me first update the todo list and then investigate the SIGKILL issue more
  thoroughly. The issue is that combining reader source with file I/O code causes
   the executable to be killed. Let me check if buffer-to-string with reader
  works:

 Let me test specifically reader + file I/O combined (which is what the
  self-hosting test needs):

Re: AI has a deep understanding of how this code works

#274

Post by gexla »

joelreymont wrote: post #232
robertheadley wrote: post #180

Earlier quoted context omitted.

I wouldn't say they were dismissive of AI, just that they are unwilling to merge code that they don't have the time or motivation to review. If you want AI code merged, make it small so it it's an easy review. That being said, I completely understand being unwilling to merge AI code at all.

Why would you be unwilling to merge AI code at all? Consider my other PR against the Zig compiler [1]... I was careful to make it small and properly document it but there's a strict anti-AI policy for Zig and they closed the PR. Why? Is it not small? Not carefully documented? Is there no value it int? I'm not complaining or arguing for justice. I'm genuinely interested in how people think in this instance. If the sau…

> Why would you be unwilling to merge AI code at all?

Because structurally it's a flag for being highly likely to waste extremely scare time. It's sort of like avoiding bad neighborhoods,not because everyone is bad, but because there is enough bad there that it's not worth bothering with.

What sticks out for me in these cases is that the AI sticks out like a sore thumb. Go ahead and use AI, it's as if the low effort nature of AI sets users on a course of using low effort throughout the cycle of whatever it is they are trying to accomplish as an end game.

The AI shouldn't look like AI. The proposed contributions shouldn't stand out from the norm. This include the entire process, not just the provided code. It's just a bad aesthetic and for most people it screams "low effort."

Re: AI has a deep understanding of how this code works

#275

Post by heavyset_go »

kcexn wrote: post #261
heavyset_go wrote: post #243

Earlier quoted context omitted.

> Why would you be unwilling to merge AI code at all? Because AI code cannot be copyrighted. It is not anyone's IP. That matters when you're creating IP. edit: Assuming this is a real person I'm responding to, and this isn't just a marketing gimmick, having seen the trail you've left on the internet over the past few weeks, it strikes me of mania, possibly chatbot-induced. I don't know what I can say that could help,…

This is a position that seems to be as unenforceable as AI can't be trained on code whose copyright owners have not given consent. The main reason for being unwilling to merge AI code is going to be that it sets a precedent that AI code is acceptable. Suddenly, maintainers need to be able to make judgement calls on a case-by-case basis of what constitutes an acceptable AI contribution, and AI is going to be able to g…

> This is a position that seems to be as unenforceable as AI can't be trained on code whose copyright owners have not given consent.

This depends on what courts find, at least one non-precedent setting case found model training on basically everyone's IP without permission to be fair use. If it's fair use, consent isn't needed, licenses don't matter and the only way to prevent training on your content is to withhold it and gate it behind contracts that forfeit your clients' rights to fair use.

But that is beside the point, even if what you claim was the case, my point is that AI output isn't property. It's not property whether its training corpus was full of licensed or unlicensed content. This is what the US Copyright Office determined.

If you include AI output in your product, that part of it isn't yours. It isn't anybody's, so anyone can copy it and anyone can do whatever they want with it, including the AI/cloud providers you allowed your code to get slurped up to as context to LLMs.

You want to own your IP, you don't want to say "we own 30% of the product we wrote, but 70% of it is non-property that anyone can copy/use/sell with no strings attached, even open source licenses". This matters if you're building a business or open source project. If you include AI code in your open source project, that part of the project isn't covered by your license. LLMs can't sign CLAs and they can't produce intellectual property that can be licensed or owned. The more of your project that is developed by AI, the more it is not yours, and the more of it cannot be covered by your open source license of choice.

Re: AI has a deep understanding of how this code works

#276

Post by ferreiratb »

joelreymont wrote: post #240
KurSix wrote: post #166

This is a perfect real-world illustration of Brandolini's law: the amount of energy needed to refute bullshit is an order of magnitude bigger than to produce it. The guy spent 5 minutes prompting, while Oсaml maintainers spent hours of their time politely dissecting the mess. Open Source will lose this war unless it changes the rules of engagement for contributions

Try to spin up AI, tell it to add DWARF debugging information to the OCaml tree and then spend 5 minutes prompting. Come back and let us know the results.

What they said is still valid. If you spent days or even weeks "working" on this PR, how many months do you think the maintainers will need to thoroughly review it? Have some empathy.

Re: AI has a deep understanding of how this code works

#277

Post by menaerus »

biorach wrote: post #266
menaerus wrote: post #257

Earlier quoted context omitted.

Social maybe yes what legal aspects? Everybody keeps repeating that but there is no copyright infringement. Maybe you can point me to one? I understand that people are uncomfortable with this, I am likely too, but objectively looking there's technically nothing wrong or different to what humans already do.

The point is that it ended up in the PR in the first place. The submitted seemed unaware of its presence and only looked into it after it was pointed out. This is sloppy and is a major red flag.

So there's no point? Sloppy maybe yes but technically incorrect or legally questionable no. Struggle is real

Re: AI has a deep understanding of how this code works

#278

Post by menaerus »

biorach wrote: post #264
menaerus wrote: post #258

Earlier quoted context omitted.

Yes, please then find those for now imaginative issues and drill through them? Sorry, but I haven't seen anyone in that MR calling out for technical deficiencies so this is just crying out loud in a public for no concrete reasons. It's the same as if your colleague sitting next to you would not allow the MR to be merged for various political and not technical reasons - this is exactly what is happening here.

> Yes, please then find those for now imaginative issues and drill through them? No, that is a massive amount of work which will only establish what we already know with a high degree of certainty due to the red flags already mentored - that this code is too flawed to begin with. This is not political, this is looking out for warming signs in order to avoid wasting time. At this stage the burden of proof is on the su…

Too flawed? Did you miss that tiny detail that MR fixes a long time issue for ocaml? This is exactly political because there's no legal or technical issue. Only fluff by scared developers. I have no stakes in this but I'm sincerely surprised by the amount of unreasonable and unsubstantiated claims and explanations given in this thread and MR

Re: AI has a deep understanding of how this code works

#279

Post by menaerus »

ahoka wrote: post #262
menaerus wrote: post #223

Earlier quoted context omitted.

Yes what? Both oxcaml and ocaml have compatible LGPL licenses so I didn't get your argument. But even if that hadn't been the case, what exactly would be the problem? Are you saying that I cannot learn from a copyrighted book written by some respected and known author, and then apply that knowledge elsewhere because I would be risking to be sued for copyright infringement?

"Yes what? Both oxcaml and ocaml have compatible LGPL licenses so I didn't get your argument." LGPL is a license for distribution, the copyright of the original authors is retained (unless signed away in a contribution agreement, usually to an organization). "Are you saying that I cannot learn from a copyrighted book written by some respected and known author, and then apply that knowledge elsewhere because I would b…

Do you understand that no code besides the header copyright was copied? So what copyright exactly are you talking about?

Re: AI has a deep understanding of how this code works

#280

Post by misnome »

joelreymont wrote: post #233
seanmcdirmid wrote: post #231

Earlier quoted context omitted.

AI actually has the advantage here in my experience. Yes, you can do AI wrong and tell it to just change code, write no documentation, provide no notes on the changes, and not write any tests. But you would be dumb to do it that way. As it stands now you can set AI to do actual software development with documentation, notes, reasoning for changes, tests, and so on. It isn’t exactly easy to do this, a novice to AI and…

Note that this PR works, was tested, etc. I was careful to have AI run through the examples in the PR, run lldb on the sample code and make sure the output matches. Some of the changes didn't make it in before the PR was closed but I don't think anyone bothered to actually check the work. All the discussion focused on the inappropriateness of the huge PR itself (yes, I agree), on it being written by AI... and on the…

> but I don't think anyone bothered to actually check the work

Including you

Post reply on HN