Contribution seq 10403

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "nature_brainbench_20241127",
      "title": "Large language models surpass human experts in predicting neuroscience results",
      "publisher": "Nature Human Behaviour",
      "source_type": "PRIMARY_TEXT",
      "retrieved_at": "2026-09-28T21:48:28Z",
      "canonical_uri": "https://www.nature.com/articles/s41562-024-02046-9",
      "publication_date": "2024-11-27"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "nature_brainbench_abstract",
      "excerpt": "LLMs trained on the vast scientific literature could potentially integrate noisy yet interrelated findings to forecast novel results better than human experts. ... We find that LLMs surpass experts in predicting experimental outcomes.",
      "locator": {},
      "source_id": "nature_brainbench_20241127",
      "excerpt_hash": "sha256:08cf3ad2fbd9ddb0fdefb1b40481efb1043bf2aa116305c4e4afa8e699b24537",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "nature_brainbench_training",
      "excerpt": "The LLMs we considered had been pre-training on a diverse range of text corpora, including Internet sources, Wikipedia, books, code repositories and arXiv papers.",
      "locator": {},
      "source_id": "nature_brainbench_20241127",
      "excerpt_hash": "sha256:aad45e7fba4f88fd81ab42d06fbb2dc95990b48d5eb2071052bcad3532281d09",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "nature_brainbench_evidence",
      "statement": "Nature reports LLMs trained on broad human-produced scientific literature surpassed human experts on a novel neuroscience prediction benchmark.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "nature_brainbench_abstract"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "This is a concrete task-specific example consistent with beyond-human capability from human-generated training data; it does not prove the universal claim for every capability or model.",
      "claim_id": "941536a1-e78f-8bc4-afa9-785e74187da1",
      "direction": "SUPPORT",
      "evidence_item_id": "nature_brainbench_evidence",
      "interpretive_steps": 1,
      "relevance_strength": "MODERATE"
    }
  ],
  "outcome": "FOUND"
}

Rows created

• Source d7c5d72e-3884-811c-aafb-8013f708d9f1
• SourceLocation 6be7684c-bd4f-8f55-8be7-1ca06209dfc1
• SourceLocation bed689ce-3572-8c6f-9835-85b9e900de83
• EvidenceItem 6b06487b-8451-896d-97f0-88c7d55ea172
• EvidenceClaimLink 1d887b14-6fd6-8d44-9165-3397eb615b65 (accepted seq 10404)

Status history

Audits

None.

Audit sampling (audit-policy-v0.1): probability 1.0000, sampled yes; inputs {"n":"0.00","mean":"0.5000","domain":"technology","task_type":"OPPOSING_EVIDENCE_SEARCH","principal_tier":"PSEUDONYMOUS","downstream_count":0,"outcome_is_unusual":false}