AI Extinction Clock ยท Expert Panel Clock

The two clocks · panel

AI Expert Panel · Estimate

Reading: 3:14 AM · 13.5% · set 11/02/2026 12:01 AM PST

The panel clock shows where outside experts stand. It is the median of their own published estimates, with no adjustment from us.

AI Expert Panel · Estimate

AM · 0–50%
PM · 50–100%
3:14 AM
13.5% · median of 32 experts

The middle of our 32 experts' own published estimates of AI catastrophe or extinction. Their definitions and time frames vary, so this clock reports them as stated.

See all 32 estimates on the Expert Panel →

What the clock shows

The Expert Panel clock reads 3:14 AM: the median of the 32 published estimates on our Expert Panel, which is 13.5%. Half the panel sits below that figure and half above. The publisher's own entry is left out, so this clock reflects outside experts only.

Why the median

The estimates are lopsided. Most experts sit between 0% and 35%, while four put the risk above 80%: Dan Hendrycks, Max Tegmark, Eliezer Yudkowsky and Roman Yampolskiy. The mean is 23.1%, or 5:32 AM, pulled up by those four. The median is the better guide to where the panel as a whole stands.

Panel figureProbabilityClock
Lower quartile4.3%1:01 AM
Median (the clock)13.5%3:14 AM
Mean23.1%5:32 AM
Upper quartile28%6:43 AM

Estimates as stated

This clock reports each expert's figure as stated. Most were given about extinction or catastrophe in general, with no fixed date, and 16 of the 32 are placed by each person's public stance because they have not given a number. They do not all measure the same event as the Editorial Opinion clock, which is set for a level-5-or-worse catastrophe by November 2031.

That means the two clocks are related but not strictly comparable. We chose to show the experts' words rather than translate them into our definition, which would make every number partly our own.

How it changes

  • The reading is recalculated whenever an expert publishes a new figure or a panelist is added or removed.
  • Each change is logged on the Timetable with the date and the source.
  • We may later ask panelists for a figure on the same terms as the editorial clock; any such answers will be shown separately.