If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All
by Eliezer Yudkowsky and Nate Soares
Little, Brown and Company, 272 pp., $30.00
by Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland, and Romeo Dean
AI Futures Project, 71 pp., free online
Proponents of extinction-level AI risk perennially suffer from the curse of Cassandra, the existential stakes of their advocacy farcically contrasted with their failure to communicate these immense concerns to the highest level of policymakers or to the general public. While concerns about the calamitous potential of Artificial General Intelligence have become more visible in recent years, coinciding with the sudden ubiquity of much simpler AI tools in everyday life, the more apocalyptic predictions still register as far-fetched and alarmist to the reading public, to the extent they register at all. Failing to cohere neatly to either side of the omnipresent information age culture war, AI doomers are exiled to the political wilderness.
Eliezer Yudkowsky and Nate Soares have endeavored to escape this condition with the release of a new manifesto, the bluntly titled If Anyone Builds It, Everyone Dies, which may be the yet most ambitious and prominent attempt to bring AI alarmism to a mass audience.
