Regression
Definition
Something that worked yesterday and doesn't today, because of a change you made somewhere else. In AI, one extra instruction can reroute everything around it — so re-testing only the case you fixed is how you ship regressions and find out from a customer.
A regression in AI is a previously-working behavior that breaks because of an unrelated change. In deterministic software, a fix is local — you change one function, other functions are safe. In AI, prompts, examples, and context are entangled; changing one instruction can silently rewrite the model's behavior on cases the fix wasn't targeting.
Why it matters
Regressions are the specific reason you need a full golden dataset, not just the case you're fixing. Every iteration should be scored against the whole set, and every drop investigated before merge. Without that discipline, every fix ships a new hidden failure somewhere else.
Related terms