How to assess claims about AI extinction risk
Commentators have put widely varying probabilities on the possibility that machines could wipe out humanity, ranging from zero to more than 95 percent. New Scientist columnist Jacob Aron argues that such figures should not be accepted without examining how they were produced and what assumptions underpin them.
Commentators have estimated the probability of machines wiping out humanity at figures ranging from 0 to more than 95 percent.
New Scientist columnist Jacob Aron explains why those numbers should not be trusted without examining the assumptions and methods behind them.
THE QUESTIONS THIS EVENT LEAVES BEHIND
What assumptions produce such widely different estimates of AI extinction risk?
What evidence would justify assigning a numerical probability to this outcome?
Who benefits when extreme risk estimates shape public debate?
How should uncertainty be communicated when the available evidence does not support precise figures?
YOUR QUESTION
Does this story leave you with another question?
Send us the question the report did not answer. It may become the next question IAQ investigates.
GLOBAL CURIOSITY MAP · AUGUST 2026