Current frontier models suffer from 'jagged intelligence' — excelling at international mathematics olympiads while simultaneously failing rudimentary arithmetic and spatial logic; true AGI requires smoothing these jagged edges via continual learning, d...

Current frontier models suffer from 'jagged intelligence' — excelling at international mathematics olympiads while simultaneously failing rudimentary arithmetic and spatial logic; true AGI requires smoothing these jagged edges via continual learning, dynamic memory allocation, and robust world models. | Novel world-model architecture from DeepMind

Key catalyst: Novel world-model architecture from DeepMind

Watch events: ARC-AGI-2 saturation; GDPval subscore analysis; world-model benchmarks

Multiple empirical demonstrations 2024-2026: IMO Gold medals via o-series / DeepMind; simultaneous failures on Tower of Hanoi, counting, spatial reasoning across same frontier models.

{
  "nia": false,
  "mode": "FORECAST",
  "role": "Cited-CEO",
  "context": "Hassabis structural-skepticism framing constrains bull-case AGI timelines. Couples with CMQ_010 (requires AlphaFold-class scientific breakthroughs) and INF_072 (50/50 scaling alone reaches AGI).",
  "to_year": 2026,
  "conv_cues": "CEO FIRST_PERSON; specific diagnostic framing",
  "direction": "HAPPEN",
  "from_year": 2026,
  "timeframe": "2026 ongoing",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "quartile_checkpoint",
      "label": "Q1 window check-in (25%)",
      "status": "overdue",
      "weight": 0.05,
      "ordinal": -3,
      "source_id": null,
      "expected_date": "2026-01-30",
      "observed_date": null,
      "miss_emitted_at": "2026-05-02T22:07:21.384228+00:00",
      "miss_emitted_by": "metadata_milestone_sweep"
    },
    {
      "kind": "quartile_checkpoint",
      "label": "Q2 window check-in (50%)",
      "status": "overdue",
      "weight": 0.05,
      "ordinal": -2,
      "source_id": null,
      "expected_date": "2026-03-01",
      "observed_date": null,
      "miss_emitted_at": "2026-05-02T22:07:21.384228+00:00",
      "miss_emitted_by": "metadata_milestone_sweep"
    },
    {
      "kind": "quartile_checkpoint",
      "label": "Q3 window check-in (75%)",
      "status": "overdue",
      "weight": 0.05,
      "ordinal": -1,
      "source_id": null,
      "expected_date": "2026-03-30",
      "observed_date": null,
      "miss_emitted_at": "2026-05-02T22:07:21.384228+00:00",
      "miss_emitted_by": "metadata_milestone_sweep"
    },
    {
      "kind": "event",
      "label": "Current frontier models suffer from 'jagged intelligence' — excelling at international mathematics olympiads while simultaneously failing ru",
      "status": "hit",
      "weight": 1,
      "ordinal": 0,
      "source_id": "AI_005",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    }
  ],
  "repeat_eps": 1,
  "affiliation": "Google DeepMind",
  "attribution": "FIRST_PERSON",
  "granularity": "YEAR",
  "resolved_at": "2026-04-29T22:23:18.198758+00:00",
  "source_refs": "2, 9",
  "target_date": "2026-12-15T00:00:00",
  "display_date": "2026-04-29",
  "episode_date": "2026-04-21T00:00:00",
  "key_catalyst": "Novel world-model architecture from DeepMind",
  "parse_method": "Current-state observation",
  "domain_bucket": "AI",
  "episode_title": "Forecasting the Inference Epoch: Expert AI Predictions & Macroeconomic Trajectories (2023-2026)",
  "flag_repeated": false,
  "in_5yr_window": true,
  "source_report": "AI Predictions Search Plan.md (2026-04-21)",
  "appears_in_eps": "AI-RPT",
  "futurist_phase": "Phase 1 (2026)",
  "is_macro_claim": false,
  "total_mentions": 1,
  "priority_weight": 5,
  "report_evidence": "Anchor section: Pragmatic Forecasters / Jagged Intelligence.",
  "active_end_month": "2026-12",
  "recent_statement": "Hassabis 2025-2026 interviews and DeepMind blog.",
  "watch_events_raw": "ARC-AGI-2 saturation; GDPval subscore analysis; world-model benchmarks",
  "months_from_today": 8,
  "probability_layer": "Higher (in-flight)",
  "active_start_month": "2026-01",
  "december_dispersal": {
    "reason": "december_dispersal: domain=AI → 09/2026",
    "new_date": "2026-09-30",
    "old_date": "2026-12-31",
    "applied_at": "2026-04-30T16:28:34.304992+00:00"
  },
  "flag_nia_bracketed": false,
  "resolved_at_source": "validations_observed_at",
  "track_record_grade": "A",
  "track_record_notes": "Hassabis research-and-capability-assessment accurate; founded AlphaFold Nobel lineage.",
  "contradicting_notes": "Gemini 3 Deep Think, GPT-5 continuous-learning variants narrowing gap in 2026; some researchers argue jaggedness is transient.",
  "flag_near_term_2027": false,
  "flag_high_conviction": true,
  "milestones_derived_at": "2026-05-02T03:08:50.484120+00:00",
  "reference_class_match": {
    "top_n": [
      {
        "id": "agi_breakthrough_5y",
        "cosine": 0.595
      },
      {
        "i