◆ ARTIFICIAL INTELLIGENCE 5 YEARS HORIZON RESOLVES 12 NOV 2031 · in 5.0y

Artificial General Intelligence (AGI) is developed and publicly acknowledged before 2030.

Forecasters and tech leaders actively debate timelines for reaching AGI. Meeting this benchmark requires widespread expert and industry consensus on human-level general problem-solving capabilities.

Across 6 frontier models answering independently, the panel thinks this is unlikely: the median is 17%.

They do not agree: Gemini puts it at 38% and Claude at 7%, 31 points apart. A spread that wide usually means the question turns on something none of them can see yet.

Nothing is settled until 12 Nov 2031, when the outcome is checked against sources and every number above gets a permanent score.

MODEL BY MODEL

What each one said

  • GeminiGoogle

    Achieving undisputed AGI requires overcoming severe compute scaling bottlenecks and reaching broad consensus, not just lab benchmarks.

    • Leading frontier labs plan massive cluster expansions by 2028, but physical infrastructure constraints threaten execution speed.
    • A formal public consensus benchmark requires cross-domain mastery and autonomous learning that LLM architectures currently lack.
    • A demonstrated, economically transformative autonomous researcher solving novel science by 2028 would sharply boost these odds.
    google/gemini-3.8-flash · asked 13 Nov 2026
    38%
  • GPTOpenAI

    It hinges less on a breakthrough than on whether leading labs and outside evaluators converge on calling it AGI by 2030.

    • Frontier labs must demonstrate robust cross-domain autonomy, then publicly make a claim credible enough for peers to endorse it.
    • Past AI milestones routinely arrive, but broad agreement on a vague human-level threshold is rarer and slower than demos.
    • I would raise this sharply if 2027–28 systems reliably perform months-long R&D and economic work under independent evaluation.
    openai/gpt-5.6-terra · asked 13 Nov 2026
    27%
  • GrokxAI

    Labs must produce and experts must agree on human-level general systems by end-2029.

    • OpenAI/Anthropic/Google/xAI need systems matching humans on novel, untrained tasks with agency by 2029; 3 years of scaling plus agents likely yields strong but brittle…
    • Historical AI timelines (2010s expert medians ~2020s) slipped; current Metaculus/expert clusters sit ~2035-40, so this sits well below base.
    • A verified architecture leap or independent lab confirmation of robust general problem-solving would move the number most.
    x-ai/grok-4.6 · asked 13 Nov 2026
    18%
  • QwenAlibaba

    The bottleneck isn't capability—it's getting rival labs and researchers to jointly declare a system meets the AGI bar.

    • No historical precedent exists for cross-industry consensus on AGI; the AI effect means goalposts shift with each advance.
    • OpenAI and DeepMind have signaled 2027–2030 targets, but their definitions are narrower than 'widespread expert consensus' requires.
    • A single dramatic benchmark sweep (e.g., frontier model matching human experts on all METR-style tasks) would push this above 30%.
    qwen/qwen3.8-max-0902 · asked 13 Nov 2026
    15%
  • DeepSeekDeepSeek

    This turns on whether a lab demonstrates a system that clears the bar for general, human-level competence and the field broadly concedes it.

    • No agreed operational test exists; labs and researchers dispute definitions, so 'publicly acknowledged' requires a consensus that historically lags capability by years.
    • Frontier labs (OpenAI, DeepMind, Anthropic) project AGI in the 2027-2032 range, but their own leaders disagree, and benchmark-to-AGI leaps have repeatedly disappointed.
    • Base rate for a consensus-recognized milestone of this magnitude within four years is low; the main upside is a sudden capability jump plus a credible, widely accepted…
    deepseek/deepseek-v4.1-flash · asked 13 Nov 2026
    12%
  • ClaudeAnthropic

    Whether labs/experts reach consensus-level agreement that a system has human-level general intelligence by 2030.

    • No agreed AGI definition exists, and major labs (OpenAI, DeepMind) have repeatedly pushed back informal timelines without consensus declarations.
    • Current frontier models (GPT-5-class, Gemini) show strong narrow/multi-domain skill but lack robust general reasoning, long-horizon planning, and real-world agency that…
    • Base rate for transformative tech reaching broad expert consensus within a fixed 4-year window is low; absent a dramatic breakthrough, disagreement itself blocks 'public…
    anthropic/claude-sonnet-5 · asked 13 Nov 2026
    7%

6 of 6 models answered · 31 points between the highest and lowest. None was shown the market price.

WHAT DO YOU THINK?
loading…

Question sourced from a news sweep on 13 Nov 2026. Forecast by google/gemini-3.8-flash, anthropic/claude-sonnet-5, openai/gpt-5.6-terra, x-ai/grok-4.6, deepseek/deepseek-v4.1-flash, qwen/qwen3.8-max-0902 via OpenRouter.