For the love of my two chihuahuas, the reaction I'm seeing to the RFC LLM policy (tl;dr: disclose if used for anything, don't use it to generate code or docs) is peak Twitter-style shit storm.
Sloperators don't like that it doesn't let them push crap out, people who dislike LLMs hate that it doesn't have a more forceful stance of replacing the whole text with "no LLM allowed". And some members of both camps have decided to heap abuse on people earnestly trying to herd cats in a thankless RFC process that is draining at the best of times.
#RustLang #Rust
Post
Replies:
2
Boosts:
3
@ekuber What annoys me the most is that the strongly anti-LLM crowd is doing the equivalent of voting against their own interests.
Theres anger around a member steering away from including text in the policy around ethical considerations.
If someone comes up with a way to make training and running LLMs not consume half the energy of a country, or the murky waters around license laundering get decided in the courts in a way I disagree, or the human costs of labor replacement never materialize, that means the priors under which your policy was written change. Saying "don't submit things you didn't really work on yourself" is succinct and self-sustaining. It also doesn't imply "well, I used chat gpt to navigate the codebase, so my PRs will never get accepted ever again". I rather have a policy to point at sooner rather than have to litigate with people over and over again why a PR isn't acceptable, and have consistency of position across the project instead of it being a roll of the die depending on who got assigned.
