Reliability
Regression test
A saved set of cases re-run on every change, so an improvement in one place cannot silently break another.
AI systems are unusually prone to whack-a-mole: fixing one behaviour breaks a different one, and nobody notices until a customer complains.
A regression suite is the antidote. Every bug found in production becomes a permanent test case. Over time the suite encodes everything you have learned about how the system fails.
The practical bar is low: it does not need to be exhaustive, it needs to run automatically on every change and fail the build when something breaks.
Related