An artificial general intelligence system is created and verified before 2030.
Timelines for achieving artificial general intelligence remain central to industry forecasts. Resolves positively if an AI system matches or exceeds median human performance across a comprehensive battery of intellectual and professional tasks.
Across 6 frontier models answering independently, the panel thinks this is unlikely: the median is 15%.
They do not agree: Gemini puts it at 38% and Claude at 6%, 32 points apart. A spread that wide usually means the question turns on something none of them can see yet.
Nothing is settled until 2 Nov 2031, when the outcome is checked against sources and every number above gets a permanent score.
What each one said
-
GeminiGoogle
Achieving verified AGI by 2030 requires breaking current sample-efficiency and novel reasoning walls within three years of training runs.
- Frontier labs like OpenAI and Anthropic are scaling compute 3-5x yearly, but algorithmic breakthroughs in autonomous learning remain unsolved.
- Comprehensive verification across professional domains takes 12-18 months of validation post-training, compressing the window to 2028-2029.
- I would shift above 60% if continuous self-improving agents routinely match top human software engineers on unconstrained novel repos.
38% -
GPTOpenAI
It hinges on whether capability gains become broad, reliable autonomy rather than stronger benchmark performance.
- Labs must build and independently test a system across a pre-specified, comprehensive task battery before 2030.
- Past AI forecasts routinely overestimate near-term generality; current systems still fail at reliability, transfer, and long-horizon work.
- A replicated 2027–28 leap in autonomous research and professional-task performance would sharply raise the odds.
19% -
GrokxAI
Frontier labs must ship and independently verify robust median-human performance on a broad cognitive+professional battery by end-2029.
- OpenAI, Google DeepMind and Anthropic have ~38 months; they need systems that close remaining gaps in long-horizon agency, reliability and tool use.
- Capability jumps occur every 18-24 months, yet current models still fail comprehensive evals; independent verification of a true AGI battery historically takes extra…
- A public, audited suite (expanded HLE plus workplace simulations) clearing the median-human bar in 2029 would raise this sharply; otherwise the window is too short.
18% -
QwenAlibaba
The comprehensive-battery-plus-verification bar is far higher than what current frontier models approach, even with rapid scaling.
- No system today handles sustained autonomous professional work (surgery, field engineering, courtroom advocacy) at median human level, and the gap in embodied/novel-task…
- Verification requires an agreed-upon testing body and battery—no such institution exists yet, and consensus on what counts would itself take years to form.
- Even optimistic lab insiders (Altman, Hassabis) who float 2027–2028 AGI dates use definitions narrower than 'median human across comprehensive professional tasks,' which…
11% -
DeepSeekDeepSeek
Whether a single system can generalize across all cognitive domains, not just scale further, before 2031.
- No lab has a system that generalizes across unrelated domains; frontier models still fail at long-horizon planning and novel tasks, so the gap is architectural, not just…
- Verification requires a comprehensive battery that doesn't exist yet; no consensus benchmark or body can certify 'median human across all intellectual tasks' by 2031.
- Base rate for a genuine paradigm shift in five years is low; even optimistic AGI researchers put median timelines in the 2030s-2040s, and the 2020s scaling wave has…
8% -
ClaudeAnthropic
Hinges on whether any lab achieves and a credible body verifies broad human-level performance by 2030.
- No consensus benchmark or verification body exists yet, and creating one plus certifying results takes years.
- Current frontier models still fail broadly on robustness, long-horizon planning, and novel reasoning despite narrow benchmark gains.
- A surprise breakthrough with rapid independent replication (e.g., METR-style evals showing broad parity) would raise this significantly.
6%
6 of 6 models answered · 32 points between the highest and lowest. None was shown the market price.
Question sourced from a news sweep on 3 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.