UnlikelyLass on Nostr: I am not trying to paint all LLM users as plagiarists, either. I just think the ...

I am not trying to paint all LLM users as plagiarists, either. I just think the nature of the tool means there is a mathematical certainty that, much like an infinite number of monkeys, you will get some amount of duplication of input training data. It's not the goal, but I think it's a mathematical certainty precisely because the mechanisms in play are statistical in nature. Statistically, it's going to happen, even if it's rare. I don't think it's possible to ensure it can't happen.