RFC: Banning "AI"-backed (LLM/GPT/whatever) contributions to Gentoo

Comments

• lvxferre@mander.xyz ⁨1⁩ ⁨year⁩ ago

  Good move.

  Putting aside ethical, moral, and legal concerns: LLMs are useful as long as you cross-check every bloody line of their output; that applies to text and, based on what I read about, code too. They aren’t good for direct info, but rather to inquire about information sources.

  But guess what. Muppets are abundant in this world. And muppets love assuming shite like “teh output is prolly fine lol lmao”. But they still want to believe that they’re “contriburing”, and then they defend their stupidity with shite like “BuT I HaVe gOOd InTeNtIonS!!!”.

  A few “replies” to the HN comments:

  The idea that intelligences - whether they be human, artificial or alien - should be forbidden from learning from code

  Assuming intelligence + taking the metaphor (“learning”) as the thing.

  It looks like a knee-jerk reaction by some “AI” hater rather than a well thought out request.
  Copyright: You have the same problem when copy-pasting code, and people do that, you can’t really single out AI.
  Quality: AI-generated code is often lower quality, but so is code written by bad coders

  Argumentum ad hominem plus two instances of whataboutism…

  ChatGPT3.5 already spots bugs, e.g. when I swap the order of conditions in fizzbuzz. An error that a human could make. We’ve been at the point where AI can help spot bugs for a while already.

  Cherry picking.

  Someone committing poor quality LLM generated code and deeming it appropriate for review could create equally bad, if not worse, handwritten code.

  I believe that, most of the time, that person would not be “contributing” (i.e. being a dead weight) on first place, if not with a system that output plausible-looking bullshit on their face.

  I had LLM patiently show me use after free bugs in non-existent Asterisk C code it just made up. :D

  Obviously that’s your fault for not having the code it found the bugs in. Why are you attacking progress?

  Highlighting a flaw in a system is apparently “attacking progress”. Yeah, nah.

  Those concerns seem legit? Surprised at the negativity here.

  HN thinks LLMs are the early days of the singularity and not a spam generator. I assume this is an echo chamber effect.

  I believe that echo chamber effect plays a huge role, but also the high concentration of tech bros there.

  Using LLMs as a glorified documentation search is the best use case I’ve found.

  …they said what I said but in prettier words. I like it.

  I suppose I can try to mark my projects in a such a way as to inform Gentoo that it’s against their policy to package them.

  “How to sound like a muppet because you did not RTFA” - the comment.

  From the link provided in the OP of the HN thread: “Just to be clear, I’m talking about our “original” content. We can’t do much about upstream projects using it.”

  Why don’t evaluate contributions based on how well the code/documentation is written? What does it matter who wrote it, if it’s good? Assuming no spamming by bots.

  Because eventually you need to use some heuristics to sort the rubbish out. And if people using a certain system are more often than not sending you rubbish, it’s simply better to tell them “we don’t want code from that system, sod off” than to waste your time reviewing their rubbish, then telling them why their “wonderful” contribution was not accepted.

  source