LLM ‘reflection’ is just re‑generation, not true iterative improvement
A new study shows that prompting LLMs to “reflect” rarely adds information; on objective tasks the revision step offers near‑zero gain, and on subjective tasks it often harms performance. Humans, by contrast, consistently improve answers. The findings argue that current reflection prompts are essentially re‑generation, not genuine error‑driven revision.