@grono@mastodon.com.pl > A useful AI-assisted review looks like: “Here are three edge cases this diff may break; I checked the call sites; this test is missing; this naming obscures the invariant.” If a 'hallucination'(*) is a false positive, what does a false negative look like in this case? How would you even find it? ( * - the term OpenAI's PR department uses to anthropomorphize their broken-by-design technology. ) @jacqueline@chaos.social