{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:V3OHOFGP4QOW5PRR54DM4Y74WR","short_pith_number":"pith:V3OHOFGP","schema_version":"1.0","canonical_sha256":"aedc7714cfe41d6ebe31ef06ce63fcb474c7d116d6e571c6308445a0f73d24bd","source":{"kind":"arxiv","id":"2402.06782","version":4},"attestation_state":"computed","paper":{"title":"Debating with More Persuasive LLMs Leads to More Truthful Answers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Akbir Khan, Ansh Radhakrishnan, Dan Valentine, Edward Grefenstette, Ethan Perez, John Hughes, Kshitij Sachan, Laura Ruis, Samuel R. Bowman, Tim Rockt\\\"aschel","submitted_at":"2024-02-09T21:05:01Z","abstract_excerpt":"Common methods for aligning large language models (LLMs) with desired behaviour heavily rely on human-labelled data. However, as models grow increasingly sophisticated, they will surpass human expertise, and the role of human evaluation will evolve into non-experts overseeing experts. In anticipation of this, we ask: can weaker models assess the correctness of stronger models? We investigate this question in an analogous setting, where stronger models (experts) possess the necessary information to answer questions and weaker models (non-experts) lack this information. The method we evaluate is"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2402.06782","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-02-09T21:05:01Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"11448bd75042b782594be7292b98ec2d1225554db9c6b3df1f1459c6fe3b2cc3","abstract_canon_sha256":"4b83a323fb110cc01394c9c18fe949ab731413a54801a10cba5cf91c85b9aa85"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:48:41.149468Z","signature_b64":"FBdmzaWqwegbOOvGpbpWsHS+zV3BeP+R6f3zi0T+LRhGyVXkw3Z4x2VVhs+ElPrLhfMG7PFt+rXmvBd9PZNTBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aedc7714cfe41d6ebe31ef06ce63fcb474c7d116d6e571c6308445a0f73d24bd","last_reissued_at":"2026-07-05T08:48:41.149047Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:48:41.149047Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Debating with More Persuasive LLMs Leads to More Truthful Answers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Akbir Khan, Ansh Radhakrishnan, Dan Valentine, Edward Grefenstette, Ethan Perez, John Hughes, Kshitij Sachan, Laura Ruis, Samuel R. Bowman, Tim Rockt\\\"aschel","submitted_at":"2024-02-09T21:05:01Z","abstract_excerpt":"Common methods for aligning large language models (LLMs) with desired behaviour heavily rely on human-labelled data. However, as models grow increasingly sophisticated, they will surpass human expertise, and the role of human evaluation will evolve into non-experts overseeing experts. In anticipation of this, we ask: can weaker models assess the correctness of stronger models? We investigate this question in an analogous setting, where stronger models (experts) possess the necessary information to answer questions and weaker models (non-experts) lack this information. The method we evaluate is"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.06782","kind":"arxiv","version":4},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2402.06782/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2402.06782","created_at":"2026-07-05T08:48:41.149101+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.06782v4","created_at":"2026-07-05T08:48:41.149101+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.06782","created_at":"2026-07-05T08:48:41.149101+00:00"},{"alias_kind":"pith_short_12","alias_value":"V3OHOFGP4QOW","created_at":"2026-07-05T08:48:41.149101+00:00"},{"alias_kind":"pith_short_16","alias_value":"V3OHOFGP4QOW5PRR","created_at":"2026-07-05T08:48:41.149101+00:00"},{"alias_kind":"pith_short_8","alias_value":"V3OHOFGP","created_at":"2026-07-05T08:48:41.149101+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":20,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08535","citing_title":"When the Judge Changes, So Does the Measurement: Auditing LLM-as-Judge Reliability","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2606.24627","citing_title":"The Warrant Gap: Claim-Conditioned Re-scoring for Fact-Checking","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19826","citing_title":"Heterogeneous LLM Debate Under Adversarial Peers: Honest Gains, Replacement Costs, and Resilience","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17657","citing_title":"Using Cognitive Models to Improve Language Model Simulation of Human Persuasion Games","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08098","citing_title":"When Does Delegation Beat Majority? A Delegation-Based Aggregator for Multi-Sample LLM Inference","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04987","citing_title":"DeliChess: A Multi-party Dialogue Dataset for Deliberation in Chess Puzzle Solving","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05330","citing_title":"A Model of Multi-turn Human Persuadability Using Probabilistic Belief Tracing","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00424","citing_title":"Weak Critics Make Strong Learners: On-Policy Critique Distillation for Scalable Oversight","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2407.04183","citing_title":"Seeing Like an AI: How LLMs Apply (and Misapply) Wikipedia Neutrality Norms","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2502.16810","citing_title":"AI Realtor: Towards Grounded Persuasive Language Generation for Automated Copywriting","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2503.21460","citing_title":"Large Language Model Agent: A Survey on Methodology, Applications and Challenges","ref_index":77,"is_internal_anchor":false},{"citing_arxiv_id":"2504.10166","citing_title":"Fact-Checking with Contextual Narratives: Leveraging Retrieval-Augmented LLMs for Social Media Analysis","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17247","citing_title":"Towards Robust Argumentative Essay Understanding via TIDE: An Interactive Framework with Trial and Debate","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13334","citing_title":"LLM-Based Persuasion Enables Guardrail Override in Frontier LLMs","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08327","citing_title":"Interactive Critique-Revision Training for Reliable Structured LLM Generation","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09278","citing_title":"EquiMem: Calibrating Shared Memory in Multi-Agent Debate via Game-Theoretic Equilibrium","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05579","citing_title":"LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods","ref_index":111,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22109","citing_title":"Spontaneous Persuasion: An Audit of Model Persuasiveness in Everyday Conversations","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2502.18864","citing_title":"Towards an AI co-scientist","ref_index":275,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12195","citing_title":"Representing expertise accelerates learning from pedagogical interaction data","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR","json":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR.json","graph_json":"https://pith.science/api/pith-number/V3OHOFGP4QOW5PRR54DM4Y74WR/graph.json","events_json":"https://pith.science/api/pith-number/V3OHOFGP4QOW5PRR54DM4Y74WR/events.json","paper":"https://pith.science/paper/V3OHOFGP"},"agent_actions":{"view_html":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR","download_json":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR.json","view_paper":"https://pith.science/paper/V3OHOFGP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.06782&json=true","fetch_graph":"https://pith.science/api/pith-number/V3OHOFGP4QOW5PRR54DM4Y74WR/graph.json","fetch_events":"https://pith.science/api/pith-number/V3OHOFGP4QOW5PRR54DM4Y74WR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR/action/storage_attestation","attest_author":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR/action/author_attestation","sign_citation":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR/action/citation_signature","submit_replication":"https://pith.science/pith/V3OHOFGP4QOW5PRR54DM4Y74WR/action/replication_record"}},"created_at":"2026-07-05T08:48:41.149101+00:00","updated_at":"2026-07-05T08:48:41.149101+00:00"}