We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s critiques. Larger models are better at self-critiquing, with scale improving critique-writing more than summary-writing. This shows promise for using AI systems to assist human supervision of AI systems on difficult tasks.
AI-written critiques help humans notice flaws
We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s critiques.
AI & ML
Editorial note: This article represents original analysis and commentary by the TechDailyPulse editorial team.