AGI Timeline Forecasts: What the Labs and Surveys Actually Say
This page is a structured working draft — real analysis, not yet expanded with the full expert sourcing given to the flagship pages. Safe to build on; treat specifics as provisional until sourced.
The spread is the story
Ask when AGI arrives and you’ll get answers from “within two years” to “not this century, if ever.” Both camps include credentialed researchers. That spread is itself informative: it means no one has a validated model of AGI timing, only extrapolations from trends that could bend, break, or continue.
What lab leaders are saying
Executives at the labs closest to the frontier have moved their public estimates earlier over 2024–2025, not later:
What forecasting surveys say
Structured surveys of AI researchers (distinct from single lab-leader quotes) have consistently shown median estimates for “high-level machine intelligence” shifting earlier with each survey wave since 2022, while the variance across individual respondents remains enormous — some assign it decades out, others a handful of years.
Why skeptics push back
Skeptics point to a consistent pattern: benchmark performance improves quickly while real-world reliability, long-horizon planning, and robustness to novel situations lag well behind. Critics also note that lab leaders have a commercial incentive to sound imminent. Both the optimistic and skeptical camps are tracking real signals; they weight them differently.
How to use a timeline forecast responsibly
Treat any single timeline as a scenario weight, not a fact. The protocols in this manual are built to be useful whether AGI-level capability arrives in 2027 or 2040 — tiered preparedness (see Survival & Preparedness) is explicitly designed not to depend on getting the date right.
Sources
“meaningful evidence of AGI being in play in 2025, not years away”