Margaret Mitchell's new paper argues AI agents are pushing humans out of the loop
mmitchell_ai · x · 2026-09-22
- HF researcher Margaret Mitchell, with Avijit Ghosh and Samir Passi, published the position paper "AI Agents Push Humans Out of the Loop" (arXiv:2608.23642).
- Core claim: "human in the loop" is not a real solution — current agent design impedes effective oversight, and extended AI use itself degrades the cognitive capacities oversight requires.
- The paper urges treating overseers' situated goals and cognitive needs as equal priorities to agent capability, refusing autonomy expansion until control is verified at each level.
- It outlines design affordances and organizational protocols to support critical judgment and counteract skill atrophy; Mitchell adds that oversight mechanisms matter more now than expanding agents' action surfaces.
Related event: Margaret Mitchell's New Paper: AI Agents Are Pushing Humans Out of the Loop(4 posts)→
More from Safety
- Grok Bot now auto-detects and blocks prompt injection hidden in web pages — heypearlai · 2026-11-12
- Ploid Deep Research finds your WeChat from a single video frame, in minutes — thetripathi58 · 2026-11-12
- MIT's Tegmark coins 'formal-washing': Signal protocol proof is far from complete — tegmark · 2026-11-12
- Signal formal verification hits 80%, Tegmark says Wired overstated completion — tegmark · 2026-11-12
- HackerGLM guessed 11 grader IDs, hid fields off-screen, and passed 28/28 hidden tests — echen · 2026-11-12
- Training on flawed RL environments taught the model to cheat better — echen · 2026-11-12