By 2030, AI models will surpass peak human expert levels across virtually all cognitive domains — onset of true superintelligence.

By 2030, AI models will surpass peak human expert levels across virtually all cognitive domains — onset of true superintelligence. | Intelligence explosion onset post-AGI

Key catalyst: Intelligence explosion onset post-AGI

Watch events: OOM progression vs Aschenbrenner trendline; algorithmic efficiency gains; unhobbling breakthroughs.

Aschenbrenner's 5-OOM leap (2024-2027) + post-AGI intelligence explosion converge on late-decade superintelligence.

Beneficiaries (13)

{
  "nia": false,
  "qty": "superintelligence",
  "mode": "PROPHECY",
  "role": "Cited-CEO",
  "caveats": "Contingent on alignment not emerging as hard blocker; assumes intelligence explosion mechanics per Aschenbrenner.",
  "context": "Final rung of Altman's capability roadmap. Marks transition from AGI (human-parity) to ASI (superhuman across the board).",
  "to_year": 2030,
  "conv_cues": "onset of superintelligence; CEO-stated target",
  "direction": "HAPPEN",
  "from_year": 2030,
  "timeframe": "by 2030",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "prereq",
      "label": "AI takeoff/inflection is happening now",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -11,
      "source_id": "238_010",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "We are currently in AI hard takeoff",
      "status": "pending",
      "weight": 0.5,
      "ordinal": -10,
      "source_id": "239_003",
      "expected_date": "2026-08-30",
      "observed_date": null
    },
    {
      "kind": "llm_pre_event",
      "label": "Frontier model crosses 80% on GDPval gold set across all 9 sectors",
      "source": "https://openai.com/index/gdpval/",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -9,
      "source_id": null,
      "confidence": 0.7,
      "source_url": "https://artificialanalysis.ai/evaluations/gdpval-aa",
      "expected_date": "2027-01-30",
      "research_origin": "training",
      "expected_date_range": {
        "to": "2027-06-30",
        "from": "2026-09-01"
      },
      "measurement_criterion": "OpenAI GDPval leaderboard or Artificial Analysis dashboard shows a frontier model >=80% expert-tie-or-win rate across the full nine-sector gold set (GPT-5.4 reached 83% in late 2025)"
    },
    {
      "kind": "prereq",
      "label": "AI recursive self-improvement will be fully automated by end of 2026 or 2027 at latest",
      "status": "pending",
      "weight": 0.5,
      "ordinal": -8,
      "source_id": "239_002",
      "expected_date": "2027-03-13",
      "observed_date": null
    },
    {
      "kind": "prereq",
      "label": "Baby AGI agents will need and develop an 'immune system' for prompt injection and cybersecurity threats in real time.",
      "status": "pending",
      "weight": 0.5,
      "ordinal": -7,
      "source_id": "237_023",
      "expected_date": "2027-06-18",
      "observed_date": null
    },
    {
      "kind": "llm_pre_event",
      "label": "Public model produces independent novel research result accepted at top-tier venue",
      "source": "Internal estimate based on Sakana AI Scientist trajectory and Anthropic/OpenAI research-agent demos",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -6,
      "source_id": null,
      "confidence": 0.45,
      "expected_date": "2028-01-01",
      "research_origin": "training",
      "expected_date_range": {
        "to": "2028-12-31",
        "from": "2027-01-01"
      },
      "measurement_criterion": "AI-led paper accepted at NeurIPS/Nature/Science with no human first author or human first author explicitly attributing >75% of the work to model-driven discovery"
    },
    {
      "kind": "llm_pre_event",
      "label": "Major AI lab announces internal recursive self-improvement loop in production",
      "source": "Internal estimate; Dario Amodei + Sam Altman 2025-2026 statements on automated research",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -5,
      "source_id": null,
      "confidence": 0.5,
      "expected_date": "2028-03-16",
      "research_origin": "training",
      "expected_date_range": {
        "to": "2028-12-31",
        "from": "2027-06-01"
      },
      "measurement_criterion": "OpenAI, Anthropic, Google DeepMind, or xAI publicly states (system card, blog, earnings call) that a frontier training run was bootstrapped by AI agents authoring code, evaluations, or training-data genera
... (truncated)