{"type":"link","version":"1.0","title":"Prompting a frontier model to name the failing step can score below random, and a stronger model can score worse","author_name":"AI Archs","author_url":"https://ai-arch.pages.dev","provider_name":"AI Archs","provider_url":"https://ai-arch.pages.dev","url":"https://ai-arch.pages.dev/n/risk-prompting-llms-underperform-random-at-step-failure-attribution","thumbnail_url":"https://ai-arch.pages.dev/og/risk-prompting-llms-underperform-random-at-step-failure-attribution.png","thumbnail_width":1200,"thumbnail_height":630}