How does AI risk compare?

Asteroids, nuclear war, pandemics, climate change, how does the risk from artificial superintelligence stack up against the other ways this century could go catastrophically wrong? These are philosopher Toby Ord's estimates from The Precipice. Set the AI number to your own view and watch the ranking change.

Your estimate for AI catastrophe this century 10%

Toby Ord, who wrote the book these figures come from, put AI at about 1 in 10. A survey of machine-learning researchers landed nearer 5%. Some AI safety researchers go much higher. Move the slider to set your own number.

Show risk as

← less likelylog scale · per next 100 yearsmore likely →
1 in 6Total, this century

Add it all up and Ord's overall estimate for existential catastrophe this century is about one in six, the odds of a single round of Russian roulette. Anthropogenic risks, led by AI, account for nearly all of it; the natural risks our species survived for 200,000 years are now a rounding error.

Where these numbers come from

The estimates are from Toby Ord's The Precipice: Existential Risk and the Future of Humanity (2020), Table 6.1, his best guesses for the probability that each threat causes an existential catastrophe within the next 100 years. Ord is a philosopher at Oxford's Future of Humanity Institute and a co-founder of the effective altruism movement.

They are deliberately rough, order-of-magnitude judgments, not measurements. Ord says they could be off by a factor of a few. Their purpose is to show the shape of the risk: the greatest threats now are ones we create ourselves, and AI sits at the top.

The AI figure is the only one you can change here, because it is the most contested and the one this foundation exists to address. The others are held at Ord's values.

All tools →