Back
Metaculus AGI Question
webCredibility Rating
3/5
Good(3)Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.
Rating inherited from publication venue: Metaculus
This Metaculus question serves as a live aggregation of crowd forecasts on AGI timelines, useful for understanding how the informed forecasting community assesses the pace of AI development over time.
Metadata
Importance: 45/100otherreference
Summary
A Metaculus community forecasting question tracking crowd-sourced predictions for when artificial general intelligence will be achieved. The question aggregates probabilistic estimates from forecasters worldwide, providing a continuously updated community median and distribution of expected AGI arrival dates.
Key Points
- •Aggregates probabilistic forecasts from the Metaculus community on when AGI will be developed
- •Provides a continuously updated median prediction and confidence intervals based on community input
- •Reflects evolving consensus as new AI developments influence forecaster expectations
- •Useful as a reference point for how informed forecasters assess AGI timelines
- •Community predictions have historically shifted earlier as AI capabilities have advanced rapidly
Cited by 2 pages
| Page | Type | Quality |
|---|---|---|
| Self-Improvement and Recursive Enhancement | Capability | 69.0 |
| AI Acceleration Tradeoff Model | Analysis | 50.0 |
Cached Content Preview
HTTP 200Fetched Mar 15, 202629 KB
[**650** comments](https://www.metaculus.com/questions/5121/date-of-artificial-general-intelligence/#comments)
**1.8k** forecasters
# When will the first general AI system be devised, tested, and publicly announced?
Current estimate
Sep 2032
20202079
Share
Predict
Top Key Factors
View all (10)
↑ reliable >50-step agent chains with published evals
Impact
later
Strength
10 votes
AI companies dont get funding (due to AI bubble bursting)
Impact
Earlier
Strength
16 votes
AI demonstrates the ability to apply knowledge from training to novel or unseen scenarios without crashing or failing
Impact
later
Strength
1 vote
↓ grid/permit delays & export controls on HBM/nodes
Impact
Earlier
Strength
7 votes
China starts a war with Taiwan
Impact
Earlier
Strength
84 votes
Strong regulation of AI companies by US government
Impact
Earlier
Strength
52 votes
Genuinely good US government policy on AI safety (something like SB 1047)
Impact
later
Strength
7 votes
Someone solves the alignment problem (I'd hope timelines are longer without this & shorter with it)
Impact
later
Strength
8 votes
CommentsTimelineKey FactorsQuestion Info
Timeline
1d1w2mall
30 Jul 2025Aug 2031Sep 2037Sep 2043Oct 204930 Jul 2025Aug 2031Sep 2037Sep 2043Oct 2049
Jan 15Jan 17Jan 19Jan 21Jan 23Jan 25Jan 27Jan 29Jan 31Feb 02Feb 04Feb 06Feb 08Feb 10Feb 12Feb 14Feb 16Feb 18Feb 20Feb 22Feb 24Feb 26Feb 28Mar 02Mar 04Mar 06Mar 08Mar 10Mar 12Mar 14Sep 2032
Resolution Criteria
We will thus define "an AI system" as a single unified software system that can satisfy the following criteria, all completable by at least some humans.
- Able to reliably pass a 2-hour, adversarial Turing test during which the participants can send text, images, and audio files (as is done in ordinary text messaging applications) during the course of their conversation. An 'adversarial' Turing test is one in which the human judges are instructed to ask interesting and difficult questions, designed to advantage human participants, and to successfully unmask the computer as an impostor. A single demonstration of an AI passing such a Turing test, or one that is sufficiently similar, will be sufficient for this condition, so long as the test is well-designed to the estimation of Metaculus Admins.
- Has general robotic capabilities, of the type able to autonomously, when equipped with appropriate actuators and when given human-readable instructions, satisfactorily assemble a (or the equivalent of a) [circa-2021 Ferrari 312 T4 1:8 scale automobile model](https://web.archive.org/web/20250109024142/https://www.deagostini.com/uk/assembly-guides/). A single demonstration of this ability, or a sufficiently similar demonstration, will be considered sufficient.
- High competency at a diverse fields of expertise, as measured by achieving at least 90% mean accuracy across all tasks in the Q\\&A dataset developed by [Dan Hendrycks et al.](https://arxiv.org/abs/2009.03300).
- Able to get top-1
... (truncated, 29 KB total)Resource ID:
0aa1710a67875e8e | Stable ID: ZWE3NWEwNz