{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:5UGMR4BFC23VOTT3536VCKLDOX","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"38b3de569bb0529923d8389fe1f0282ce3628f0db9ade774bc774476b9ac5af4","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-03-20T16:18:38Z","title_canon_sha256":"cfc8ec73effe4ce55f50dbb8ad31cd9b988236739f35cad6e46b2fc5cdb20ec1"},"schema_version":"1.0","source":{"id":"2303.13375","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2303.13375","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"arxiv_version","alias_value":"2303.13375v2","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.13375","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"pith_short_12","alias_value":"5UGMR4BFC23V","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"5UGMR4BFC23VOTT3","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"5UGMR4BF","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:c2d008fdae142e67fbc5fe61023d4e5e59d8dc4d73c6642a6551664093950e7d","target":"graph","created_at":"2026-05-17T23:38:52Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"GPT-4, without any specialized prompt crafting, exceeds the passing score on USMLE by over 20 points and outperforms earlier general-purpose models (GPT-3.5) as well as models specifically fine-tuned on medical knowledge (Med-PaLM)."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The official USMLE practice materials used are representative of the actual exam content and difficulty, and the model has not memorized the specific questions during pre-training (probed but not fully detailed in available text)."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"GPT-4 exceeds the USMLE passing score by more than 20 points and outperforms both GPT-3.5 and the medically fine-tuned Med-PaLM on the MultiMedQA benchmarks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"GPT-4 exceeds the USMLE passing score by over 20 points without any medical-specific training or prompts."}],"snapshot_sha256":"aa5c4eb7d7207b2a0f322277a88920b29e5aa820fc6a70938f87f5349b2f9c45"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"25353279e17364f83947f1997e60edd69bc0e9ff5a17263edda515bbdb3b388d"},"paper":{"abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable capabilities in natural language understanding and generation across various domains, including medicine. We present a comprehensive evaluation of GPT-4, a state-of-the-art LLM, on medical competency examinations and benchmark datasets. GPT-4 is a general-purpose model that is not specialized for medical problems through training or engineered to solve clinical tasks. Our analysis covers two sets of official practice materials for the USMLE, a three-step examination program used to assess clinical competency and grant licensure in the U","authors_text":"Dean Carignan, Eric Horvitz, Harsha Nori, Nicholas King, Scott Mayer McKinney","cross_cats":["cs.AI"],"headline":"GPT-4 exceeds the USMLE passing score by over 20 points without any medical-specific training or prompts.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-03-20T16:18:38Z","title":"Capabilities of GPT-4 on Medical Challenge Problems"},"references":{"count":23,"internal_anchors":8,"resolved_work":23,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Guidelines for human-AI interaction","work_id":"1915e653-ede4-4aa0-8b7b-ff8f5553f406","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Lan- guage models are few-shot learners","work_id":"75db645b-ae81-481a-bdbc-12b3f493d004","year":1901},{"cited_arxiv_id":"1810.04805","doi":"","is_internal_anchor":true,"ref_index":3,"title":"BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding","work_id":"ed240a10-5b19-406c-baa5-30803f465785","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Automated identification of adults at risk for in-hospital clinical deterioration","work_id":"c8446f66-fd7d-4d21-bee4-df87f15d6612","year":1951},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Who goes first? Influences of human-ai workflow on decision making in clinical imaging","work_id":"22b87454-3f97-4f73-8349-554d6fdc39c6","year":2022}],"snapshot_sha256":"a642b93a7c1f94352111c4390b17bf3f3284ffcf42f266171218c391874f291a"},"source":{"id":"2303.13375","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-15T13:39:07.226053Z","id":"9cb7a1f6-64b4-4b96-a8d5-8b5435604c7b","model_set":{"reader":"grok-4.3"},"one_line_summary":"GPT-4 exceeds the USMLE passing score by more than 20 points and outperforms both GPT-3.5 and the medically fine-tuned Med-PaLM on the MultiMedQA benchmarks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"GPT-4 exceeds the USMLE passing score by over 20 points without any medical-specific training or prompts.","strongest_claim":"GPT-4, without any specialized prompt crafting, exceeds the passing score on USMLE by over 20 points and outperforms earlier general-purpose models (GPT-3.5) as well as models specifically fine-tuned on medical knowledge (Med-PaLM).","weakest_assumption":"The official USMLE practice materials used are representative of the actual exam content and difficulty, and the model has not memorized the specific questions during pre-training (probed but not fully detailed in available text)."}},"verdict_id":"9cb7a1f6-64b4-4b96-a8d5-8b5435604c7b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:7310d0d4eace0d80a66f5c3ea1c313765ae3b0bda4656f43b627178788334709","target":"record","created_at":"2026-05-17T23:38:52Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"38b3de569bb0529923d8389fe1f0282ce3628f0db9ade774bc774476b9ac5af4","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-03-20T16:18:38Z","title_canon_sha256":"cfc8ec73effe4ce55f50dbb8ad31cd9b988236739f35cad6e46b2fc5cdb20ec1"},"schema_version":"1.0","source":{"id":"2303.13375","kind":"arxiv","version":2}},"canonical_sha256":"ed0cc8f02516b7574e7beefd51296375e3a59b4d753fffbd438f7f2847188086","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ed0cc8f02516b7574e7beefd51296375e3a59b4d753fffbd438f7f2847188086","first_computed_at":"2026-05-17T23:38:52.415467Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:52.415467Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Q28IDnrmWJmUwDC11W8OwoWl3/USl14keQtaYGFfJ6HjfCRv5Yo4KuJUZMMhgVdXRmLvXjYl9wz9fvUpuIoNCA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:52.415977Z","signed_message":"canonical_sha256_bytes"},"source_id":"2303.13375","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7310d0d4eace0d80a66f5c3ea1c313765ae3b0bda4656f43b627178788334709","sha256:c2d008fdae142e67fbc5fe61023d4e5e59d8dc4d73c6642a6551664093950e7d"],"state_sha256":"686a5b642dee741eca9b3dd6043d17687e1a6e8c70b8c3d68620edc231dbf32f"}