The p(doom) database · asked and declined
Andy Barkett’s p(doom)
?
What they said
Andy Barkett was asked for a p(doom) on the record (Oct 8, 2026) and did not give a number.
I think assigning a specific probability to something like that is misleading. There are certainly real risks associated with AI, including potentially existential ones.
Asked "Should we worry that there is a 10% chance that AI will kill us all?" (Berkeley Social Sciences interview).
Why this is here
The database records refusals as well as numbers. A refusal is not a number, so it never counts toward any median. If Andy Barkett states a number in public, it gets a row. What p(doom) means, and the case against it.