AI assurance work means checking that AI systems are safe, fair and accountable, before and after they are used on real people. It is one of the newest careers around, growing as the rules tighten, and because it is so new, careful outsiders can get in early. It suits sceptical, methodical people who prefer evidence to hype: you test claims, document findings and keep honest records. The hard parts are the pace of change and the amount of reading. The purpose is easy to believe in.
You spend the morning running checks on an AI system: does it treat different groups consistently, does it fail safely, are its records complete. You write up what you found, good and bad. In the afternoon you meet the team who built it and agree the fixes. You end by updating the audit trail.
AI can generate the test cases and the paperwork for its own inspection, which speeds up the routine work enormously. But a system cannot credibly vouch for itself. Deciding whether it is fair, safe and fit to use, and putting your name to that, stays human.
Entrants normally have a degree, GCSEs, a BTEC or A levels. Training is mostly given on the job.
Tick off steps, save the job and grow your own tree. Free, nothing captured.
QA / software tester · Data governance officer