Ethan Mollick: Hidden thinking traces push AI toward automation over augmentation
emollick · x · 2026-10-27
Ethan Mollick argues that the disappearance of detailed summarized thinking traces in Claude and ChatGPT — likely due to anti-distillation concerns and UX — makes human-in-the-loop work much harder. With reasoning almost entirely hidden, AI becomes "magic," pushing users toward automation rather than augmentation. He calls the near-total hiding of thinking traces a failure of imagination and a disintermediation default worth actively challenging.
More from AGI Musings
- Princeton PhD cited by Nobel committee jobless after 2,300 applications since 2024 — CtrlAltDwayne · 2026-10-27
- Coding got cheap, but the friction we removed was doing useful cognitive work — charles_irl · 2026-10-27
- Richard Socher: superintelligence will soon make almost all new scientific discoveries — RichardSocher · 2026-10-27
- tszzl: races to control a runaway intelligence explosion will keep ending in losing — CatAstro_Piyush · 2026-10-27
- Epistemic loneliness debate: it's a sign you're not smart enough to onboard others — broodsugar · 2026-10-27
- Commentary: OpenAI won't release reasoning traces because the reasoning process is the real product — max_paperclips · 2026-10-27