AI in Testing
Ways a model can take on part of testing and QA work — what it does, when it's worth it, where it falls short, and the judgement that stays with you. This is the other side of Evals: there the AI is the thing under test, here it's a helper inside the testing.
Any model you put into your testing is a component that can be wrong, so test it the way you’d test any other — see Evals. Where an entry says what to watch for, it names one risk: the way that approach fails silently. Every approach has more.