{
  "question_id": "W13",
  "slug": "ai-beats-superforecasters-by-a-wide-margin",
  "title": "An AI beats superforecasters by as much as they beat the public",
  "type": "binary",
  "resolution": {
    "levels": [],
    "values": [],
    "annotations": []
  },
  "forecasts": [
    {
      "question_id": "W13",
      "forecast_date": "2026-09-20",
      "evidence_cutoff": "2026-09-20",
      "method": "wide-v2",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.005
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.0299
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.0735
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.1384
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.2116
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.2905
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.3685
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.4443
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.511
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.5697
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.617
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.6572
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.6915
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.7208
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.7459
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.7675
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.7861
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.8022
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.816
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.828
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.8383
        }
      ],
      "passes": []
    },
    {
      "question_id": "W13",
      "forecast_date": "2026-09-27",
      "evidence_cutoff": "2026-09-27",
      "method": "wide-v2",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.001
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.026
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.0649
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.1164
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.1826
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.2562
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.3343
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.4142
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.4903
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.5566
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.612
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.6586
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.6978
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.7311
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.758
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.781
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.8007
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.8176
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.8322
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.8448
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.8557
        }
      ],
      "passes": []
    },
    {
      "question_id": "W13",
      "forecast_date": "2026-10-03",
      "evidence_cutoff": "2026-10-03",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": "Threshold assessed on FRI's tournament/baseline Overall Brier Index from any nightly leaderboard release after 2026-09-16; if FRI pools a fresh superforecaster round into the 'Superforecaster median forecast' entry, that pooled entry (and the existing 'Public median forecast') are used as the renamed equivalents.",
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0.01
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.035
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.073
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.12
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.173
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.23
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.288
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.342
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.391
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.434
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.47
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.502
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.53
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.553
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.573
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.59
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.607
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.62
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.634
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.645
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.655
        }
      ],
      "passes": []
    },
    {
      "question_id": "W13",
      "forecast_date": "2026-10-10",
      "evidence_cutoff": "2026-10-10",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": "The best AI on the forecasting leaderboard now edges superforecasters by about 0.4 points. The bar needs about 5.2 points, the same margin by which superforecasters beat the general public. Progress has been fast, but many questions are close to pure chance. That leaves limited room, so I expect the bar to be cleared, if at all, mostly in 2028–2030. The planned fresh superforecaster round could change this most, by moving the superforecaster score and so the bar.",
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0.003
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.01
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.022
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.039
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.062
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.088
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.117
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.147
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.177
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.204
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.229
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.251
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.271
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.29
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.306
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.32
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.334
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.346
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.357
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.367
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.377
        }
      ],
      "passes": []
    },
    {
      "question_id": "W13",
      "forecast_date": "2026-10-17",
      "evidence_cutoff": "2026-10-17",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": "The best AI on the Forecasting Research Institute's benchmark now roughly ties the superforecaster median (69.0 vs 68.7). This question needs it to lead by as much as superforecasters lead the public, about 5 points, so roughly 74. Progress of 2–4 points a year suggests this could happen around 2028, but gains may slow near the limits of predictability. The October 2026 superforecaster rematch, which could raise or lower the bar, is the development that would most change this.",
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0.005
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.02
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.079
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.129
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.179
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.23
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.28
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.328
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.369
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.407
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.44
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.469
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.495
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.518
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.537
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.554
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.569
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.582
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.593
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.604
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.613
        }
      ],
      "passes": []
    }
  ],
  "declined": []
}