"I'm concerned about LLM code in #curl and would like to suggest a code ban"

[bagder@mastodon.social]

"I'm concerned about LLM code in #curl and would like to suggest a code ban"

https://github.com/curl/curl/discussions/20972

[antoinechambertloir@mathstodon.xyz]

@bagder ethical rules cannot be enforced a priori, they can be breached, but they exist to state what a scientific field expects from its participants.
It would make sense, on the basis of the initial post, that a collaborative programming community states as an ethical rule that no use of code generating software whatsoever is made in the submitted code.
(People will argue about word completion.)

[doragasu@mastodon.sdf.org]

@bagder Very good state of the art report about LLM plagiarism. Yeah, that's a difficult problem.

[charon@hachyderm.io]

@bagder idk kinda seems like trusting users not to plagiarize would be similar to trusting users not to use a plagiarizing machine, if you wanted to ban llm usage.

[bagder@mastodon.social]

@charon but what is "llm usage" ?

[djh@chaos.social]

@bagder @charon Is it ok to use language models to improve man pages, documentation, the web presence and so on? It's not even only about code.

[yukijoou@fedi.kemonomimi.gay]

@bagder@mastodon.social i feel like "pretty please don't use LLMs" is not too bad a policy imo:

1. it shows the project values human contributors, which should help attracting human contributors
2. it implies that nobody will be there to help you untangle LLM spagetti and you can't hide behind your AI's output
3. it makes it more likely that knowing LLM-user will pick another project to contribute to
4. IANAL, but it may help the project legaly defend itself if plagiarised code is submitted and merged, it shows an attempt at avoiding plagiarised code

though, maybe being more explicit about the goals rather than "pretty please don't use LLMs" may be better!