{"version":1,"type":"story","url":"https://digestai.news/story/researchers-test-recursive-knowledge-building-for-science-claim-verifi","json":"https://digestai.news/story/researchers-test-recursive-knowledge-building-for-science-claim-verifi.json","markdown":"https://digestai.news/story/researchers-test-recursive-knowledge-building-for-science-claim-verifi.md","slug":"researchers-test-recursive-knowledge-building-for-science-claim-verifi","headline":"Researchers test recursive knowledge building for science claim verification","summary":"A new study on arXiv explores how recursive knowledge construction affects agents’ ability to verify scientific claims. The authors tested two design choices: what evidence the agent stores and how it decides when to stop. They created a dataset called ProClaim-eval with 718 claims across molecular biology, nuclear physics, and astronomy, each labeled as supported or refuted by experts.\n\nThe team built a system called ProClaim that maintains an external evidence state, separate from the model’s context. Their findings show that structured evidence storage improves performance over in-context accumulation. A trained judge that reviews this evidence outperformed fixed-round stopping. When tested against other verification systems and general-purpose agents, ProClaim ranked in the top three, though a separately trained judge matched its accuracy on three of five subsets. The study suggests recursive, structured evidence handling could be key for scientific agents.","keyPoints":["ProClaim-eval dataset includes 718 expert-labeled claims from biology, physics, and astronomy","Recursive evidence storage outperforms in-context accumulation for claim verification","Trained judge matches ProClaim’s accuracy on three of five subsets, using Claude Sonnet 4.6"],"whyItMatters":"If recursive evidence handling improves scientific claim verification, it could set a new standard for AI agents in research and fact-checking, reducing errors in automated reasoning.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":["Claude Sonnet 4.6","Claude Haiku 4.5"],"people":[]},"firstPublishedAt":"2026-11-02T05:00:00Z","updatedAt":"2026-11-02T05:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.AI","title":"Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?","url":"https://arxiv.org/abs/2610.40906","publishedAt":"2026-11-02T05:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":{"title":"AI Reliability in Scientific Verification","url":"https://digestai.news/thread/german-study-finds-ai-reviewers-favor-confident-scientific-papers-over-cautious","storyCount":2},"cite":{"text":"Digest AI, \"Researchers test recursive knowledge building for science claim verification\", 2 November 2026, https://digestai.news/story/researchers-test-recursive-knowledge-building-for-science-claim-verifi","publisher":"Digest AI","title":"Researchers test recursive knowledge building for science claim verification","datePublished":"2026-11-02T05:00:00Z","url":"https://digestai.news/story/researchers-test-recursive-knowledge-building-for-science-claim-verifi"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}