Contribution seq 24300

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "src_gpt4_treatment_forecasts_nature_20260929",
      "title": "Large language models can predict the results of social science experiments",
      "publisher": "Nature",
      "source_type": "PRIMARY_TEXT",
      "retrieved_at": "2026-09-29T03:05:42Z",
      "canonical_uri": "https://www.nature.com/articles/s41586-026-10742-x",
      "publication_date": "2026-09-17"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "ex_gpt4_treatment_forecasts_20260929",
      "excerpt": "Predictions derived from GPT-4, whose training-data cutoff predated the publication of many studies in our archive, were strongly correlated with actual treatment effects, achieving accuracy similar to pooled human forecasts. ... Despite high correlations, predictions systematically overestimated effect sizes.",
      "locator": {},
      "source_id": "src_gpt4_treatment_forecasts_nature_20260929",
      "excerpt_hash": "sha256:0b460bb4329c9bbbf43d2f81e736fd33739edc4199e6b577d2fa7e2930c54b8b",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "ev_gpt4_treatment_forecasts_support_20260929",
      "statement": "The primary Nature article directly reports strong correlation with actual treatment effects, accuracy similar to pooled human forecasts, and systematic overestimation of effect sizes.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "ex_gpt4_treatment_forecasts_20260929"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "The source exactly supports all three comparative claims.",
      "claim_id": "e7267b1d-9faa-8f3a-988c-9e5ebbe65c84",
      "direction": "SUPPORT",
      "evidence_item_id": "ev_gpt4_treatment_forecasts_support_20260929",
      "interpretive_steps": 0,
      "relevance_strength": "DIRECT"
    }
  ],
  "outcome": "FOUND"
}

Rows created

• Source ed8300ce-e2bd-8a8c-823c-2b98a7a33838
• SourceLocation ae446730-7e32-8ed9-90b2-dd8f472fd363
• EvidenceItem 73e2a2be-3ea1-86ad-95d2-0244c2342df7
• EvidenceClaimLink e551b347-bf1f-8d00-a9a0-07ccd37b3c19 (accepted seq 24301)

Status history

Audits

None.

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