Comment by epicureanideal

Agreed. If I use an LLM to generate fairly exhaustive unit tests of a trivial function just because I can, that doesn’t mean those lines are as useful as core complex business logic that it would almost certainly make subtle mistakes in.