“Existential catastrophe via unaligned artificial intelligence: ~1 in 10 within the next 100 years — the single largest existential risk humanity faces, out of a total existential risk of about 1 in 6.”
Said Mar 2020Deadline Dec 31, 2120The Precipice: Existential Risk and the Future of Humanity (2020), risk table; figure restated by Ord on the 80,000 Hours Podcast #72, March 2020
Ord explains the 10% as roughly a 50% chance of creating something more intelligent than humanity this century combined with a substantial chance that it goes badly; he has since revisited these estimates in later talks.