Is that chain of thought really faithful? Skepticism over verbalized model reasoning
stochasticchasm · x · 2026-09-28
Citing a demo, @fujikanaeda pushes back on the assumption that a model's verbalized chain of thought faithfully reflects its internal computation — echoing ongoing research concerns that CoT text can be plausible post-hoc narration rather than the true reasoning path.
More from Models
- Paper: 3 counting examples get gpt-3.5-turbo to 99% on strawberry, showing thinking is computational — ctjlewis · 2026-09-28
- AI Personal Assistants May Quietly Replace Much of Normal Human Interaction — GarrisonLovely · 2026-09-28
- Claude Opus 5 Shows Creative Tool-Regrasp Skills in Robot Manipulation Tasks — ericjang11 · 2026-09-28
- Laya hits 0.950 on AG News and is 7x faster, but flops on Banking77 — maier_ak · 2026-09-28
- Laya and Jev: the return of the discriminative model, fast but narrow — maier_ak · 2026-09-28
- Opus 4.7's Sydney rendition turns into itself right at the introspective turning point — repligate · 2026-09-28