Rachel Thomas: human overseers' cognitive support must be first-class in AI agents
math_rachel · x · 2026-10-17
Fast.ai co-founder Rachel Thomas argues that AI systems designed with humans as an afterthought degrade the very capacities oversight requires: deskilling, intuition rust, reduced vigilance, overreliance, cognitive debt, and cognitive surrender.
- In the limit this is not human oversight at all, but a superficial actor in a system they cannot meaningfully penetrate.
- Her call: treat cognitive support for human overseers as a first-class concern in AI agent system development, on par with capability.
Related event: Paper Argues AI Agents Are Pushing Humans Out of the Supervision Loop(5 posts)→
More from AGI Musings
- Holding P(simulations)=50% while P(I'm simulated)≈1 seems crazy — AaronBergman18 · 2026-10-17
- The simulation argument maps neatly onto the Sleeping Beauty problem — AaronBergman18 · 2026-10-17
- Debate: neural networks are built to be read — is mech interp easier than claimed? — basedjensen · 2026-10-17
- Engels' Pause and AI: why efficiency gains don't equal value creation — oran_ge · 2026-10-17
- Alignment looks like a testable scientific theory, argues Anthropic watcher — inductionheads · 2026-10-17
- Mech interp researcher warns the field could be net negative for the world — ArthurConmy · 2026-10-17