Step 4: tool results

get_claim_instances

{
  "claim_id": "50758adb-ea92-4663-adb8-70d8985d4ebd"
}

{
  "claim_id": "50758adb-ea92-4663-adb8-70d8985d4ebd",
  "total": 1,
  "limit": 20,
  "offset": 0,
  "has_more": false,
  "instances": [
    {
      "id": "49458575-3ec3-4e00-8a1c-193aebfb4f43",
      "source_id": "0a99fd17-2b29-45c8-9ab6-6afe1fdbc286",
      "verbatim_text": "Reliability curves replace a single p50 horizon with success probability across task duration, repeated attempts and time budgets.",
      "context": "Section 7 portfolio; Figure 6 lists 'One-point score as intrinsic ability' as an audit failure.",
      "stance": "affirms",
      "confidence": 0.55,
      "speaker": null,
      "publication": null,
      "source_date": null,
      "link": null,
      "source_title": "Frontier AI Forecasting Has a Measurement Problem: An Audit of Progress Evidence",
      "source_type": "preprint",
      "source_url": "https://arxiv.org/pdf/2608.14903",
      "source_published_date": "2026-08-14",
      "source_authors": [
        "Costa, Fabricio F"
      ],
      "source_publisher": "arXiv.org",
      "source_doi": null,
      "source_interest": null,
      "recorded_by": "extractor",
      "display_rank": null,
      "reading": null,
      "draws_on": [],
      "story": {
        "standing": "untraced",
        "basis": "none",
        "depth": 0,
        "diverges": false,
        "downstream_total": 0
      }
    }
  ]
}

search_similar_claims

{
  "limit": 8,
  "query": "METR 50% time horizon reliability 80% horizon forecasting AI agents"
}

{
  "query": "METR 50% time horizon reliability 80% horizon forecasting AI agents",
  "results": [
    {
      "id": "50758adb-ea92-4663-adb8-70d8985d4ebd",
      "kind": "claim",
      "text": "A single 50% task-completion horizon inadequately summarizes AI-agent capability for forecasting.",
      "claim_type": "evaluative",
      "state": "active",
      "similarity_score": 0.7001111147851123,
      "importance": 0.45,
      "assessment_status": null,
      "assessment_confidence": null
    },
    {
      "id": "b5ba920f-ea97-4b56-a7eb-33af9e2ecf8e",
      "kind": "claim",
      "text": "AI agents’ 50%-reliability software-task horizon will continue doubling approximately every seven months.",
      "claim_type": "empirical_derived",
      "state": "active",
      "similarity_score": 0.6831158080439599,
      "importance": 0.6,
      "assessment_status": "doubtful",
      "assessment_confidence": 0.65
    },
    {
      "id": "1d21c694-7f35-4518-ba59-770d0cfef986",
      "kind": "claim",
      "text": "METR Time Horizon 1.1's 50% time horizons above roughly 16 human-hours are unreliable measurements.",
      "claim_type": "empirical_verifiable",
      "state": "active",
      "similarity_score": 0.6721365301656933,
      "importance": 0.3,
      "assessment_status": null,
      "assessment_confidence": null
    },
    {
      "id": "a78dc4eb-d2da-49a8-b15b-b525779ed539",
      "kind": "claim",
      "text": "Frontier AI 50% time horizons on METR's task suite have grown approximately exponentially with release date since 2024.",
      "claim_type": "empirical_derived",
      "state": "active",
      "similarity_score": 0.661861907770914,
      "importance": 0.45,
      "assessment_status": "supported",
      "assessment_confidence": 0.65
    },
    {
      "id": "fab14deb-b0f5-45e4-bddc-a8c07d3a1719",
      "kind": "quantity",
      "text": "Doubling time of the 50%-task-completion time horizon of frontier AI models on METR's software task suite, 2019–2025, in days",
      "claim_type": "empirical_derived",
      "state": "active",
      "similarity_score": 0.6320326964685901,
      "importance": 0.65,
      "assessment_status": null,
      "assessment_confidence": null
    },
    {
      "id": "d8be49b1-0f3c-43e5-8426-cc8bc25d9165",
      "kind": "quantity",
      "text": "Doubling time, in days, of the 50%-task-completion horizon of frontier AI models released from 2024 onward on METR's software task suite",
      "claim_type": "empirical_derived",
      "state": "active",
      "similarity_score": 0.6270020005152808,
      "importance": 0.55,
      "assessment_status": null,
      "assessment_confidence": null
    },
    {
      "id": "434abbef-6a37-49af-acc9-7a534454b6b5",
      "kind": "claim",
      "text": "Benchmarks beyond METR's software suite show AI time horizons or capabilities growing faster since 2024.",
      "claim_type": "empirical_derived",
      "state": "active",
      "similarity_score": 0.6119926707042148,
      "importance": 0.3,
      "assessment_status": null,
      "assessment_confidence": null
    },
    {
      "id": "9493359e-c047-4b55-a8bb-be6049818df6",
      "kind": "claim",
      "text": "Frontier AI agents' performance on long-horizon AI R&D tasks varies substantially across repeated runs.",
      "claim_type": "empirical_verifiable",
      "state": "active",
      "similarity_score": 0.597532706255211,
      "importance": 0.55,
      "assessment_status": null,
      "assessment_confidence": null
    }
  ],
  "count": 8
}