Post #1261877
2026-04-16 15:56 UTC
@larsmb I'm not arguing against labelling and disclosing LLM use. But I think it's a bit odd to have a strict "no LLM assistance" policy.
Agentic assistants are good at *some* things, *some* of which are a terrible chore for humans. Refactoring comes to mind. Why is it a good idea to insist on a human doing refactoring alone, without LLM assistance?
You could argue that refactoring might not even establish a copyright claim, if everything behaves exactly the same way afterwards.
Replies (1)
-
@larsmb@mastodon.online 2026-04-16 17:53
@xahteiwi You don't need to argue with me; I'm aware. I'm in this weird middle ground where I'm disliked by both the AI fans because I want to talk about the problems/constraints, and hated by the others because I use/engage with it at all. I know they're genuinely useful for a few things. But if someone _doesn't_ want it, I think that's their choice that should be respected. The copyright question for the generated code is ... complicated, but currently seems to tend towards "nope" in court.