Do this
- Check a few AI results in full each month, even the routine ones.
- Track how often your team changes what the AI suggested.
- Look closer if that number is near zero.
Sources
- IMDA Singapore, Model AI Governance Framework for Agentic AI, Version 1.5 (2026), section 2.2.2, a low human override rate “may signal rubber-stamping” (pp. 29–30)
- NIST Generative AI Profile (NIST AI 600-1), section 2.7, Human-AI Configuration (over-reliance and automation bias)
