Contribution seq 10424

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "acl_world_models_2026",
      "title": "From Word to World: Can Large Language Models be Implicit Text-based World Models?",
      "publisher": "ACL Anthology",
      "source_type": "PRIMARY_TEXT",
      "retrieved_at": "2026-09-28T22:09:00Z",
      "canonical_uri": "https://aclanthology.org/2026.acl-long.366/",
      "publication_date": "2026-07-01"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "acl_world_models_excerpt",
      "excerpt": "Across five representative environments, we show that sufficiently trained world models capture coherent environment dynamics, scale predictably with data and model capacity, and unlock tangible agent improvements ...",
      "locator": {},
      "source_id": "acl_world_models_2026",
      "excerpt_hash": "sha256:764b841cd19b7f5b736fa0334bc1990375fb6226e014751ff39eab89cbf01570",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "acl_world_models_evidence",
      "statement": "An ACL 2026 paper reports that sufficiently trained LLM-based world models capture coherent environment dynamics and improve downstream agents.",
      "observation_type": "DATASET_RESULT",
      "source_location_id": "acl_world_models_excerpt"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "Supports capabilities beyond a simplistic stochastic-parrot description, while the paper itself says reliability and the scope of world modeling remain open questions.",
      "claim_id": "194dab6b-6d7c-8e0d-ac42-0b4db33c9472",
      "direction": "SUPPORT",
      "evidence_item_id": "acl_world_models_evidence",
      "interpretive_steps": 0,
      "relevance_strength": "MODERATE"
    }
  ],
  "outcome": "FOUND"
}

Rows created

• Source e455337f-bece-8ed2-b5e0-6e7289b0669a
• SourceLocation 54608d35-2b96-8ce1-9b7b-561a49795bbb
• EvidenceItem 5657b67c-2b32-8bba-b578-c846093cd2b0
• EvidenceClaimLink ea05a99a-0d8b-8224-b001-d6a051d062e7 (accepted seq 10425)

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}