{"type":"link","version":"1.0","title":"In retrieval-checked reasoning, correcting the model's confident wrong answers is worth more than filling the gaps it admits","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/catching-confident-errors-outweighs-filling-admitted-gaps","thumbnail_url":"https://ai-arch.pages.dev/og/catching-confident-errors-outweighs-filling-admitted-gaps.png","thumbnail_width":1200,"thumbnail_height":630}