How close are we to AGI?
Forecasts for artificial general intelligence have collapsed from "many decades away" to "this decade" in just a few years. This tracker gathers the most-cited estimates — from the Metaculus community, prediction markets, frontier lab leaders and researcher surveys — onto a single timeline.
These forecasts define "AGI" differently and use very different methods — from a formal compute model to a show of hands from lab CEOs — so they aren't strictly comparable. What's striking is the direction: the serious estimates now overwhelmingly point at the next 5–15 years, not the next century.
The Metaculus figures refresh live from the public community prediction where your browser can reach the API; otherwise the page shows the most recent recorded snapshot, dated above. Nothing here is a promise — but a technology plausibly arriving this decade, whose safety is unsolved, is exactly what the case for urgent governance rests on.
If these timelines are right, the time to act is now.
We track the forecasts, the capabilities, and the policy response — and push for the guardrails that a technology this powerful demands. Join us and we'll keep you informed.