The likelyhood of an LLM producing an exact match to someone else's code is no higher than if a human happens to do it. It's not really something that happens in real life.
LLMs just "learn from more sources" than any human has the chance to - but LLMs _don't_ memorize/store the training data any more than we do. That's what usually gets misunderstood.
Chad McCullough (npub1ded…kves) elementary (npub16nm…cc46)
