Evidence in AI audits
The records an auditor relies on when auditing AI.
Evidence includes model documentation, data lineage, validation reports, change records, monitoring dashboards, logs of inputs and outputs, human override records and re-performance results. Its reliability depends on logging and version control.
On the exam
Re-performing or independently testing model outputs gives stronger evidence than developers’ own reports.
Related terms
- Model cards and datasheets for datasetsStandard documentation describing a model or a dataset.
- Audit trailA record that lets you reconstruct who did what, and when.
- Model registry and versioningRecording each model version with its data, code, settings and approvals.
AAIA Test yourself on this with timed AAIA practice questions in the AAIA Exam Simulation Hub, or read the free AAIA exam guide.
Test yourself with today’s question of the day.
