What is superintelligence?

What the evidence says

In a 1998 paper, Nick Bostrom of Oxford University defined superintelligence as “an intellect that is much smarter than the best human brains in practically every field”.1 His 2014 book Superintelligence: Paths, Dangers, Strategies put it as “any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest.”2 The key word is “all”: a chess program that beats every human at chess is not superintelligent, because it can do nothing else.

Superintelligence is usually described as the step after artificial general intelligence (AGI), meaning AI that matches people across most tasks. The philosopher David Chalmers, for example, argues that AGI is “a very likely path to artificial superintelligence.”2 Surveyed researchers expect even that to take years: in a 2023 survey of 2,778 published AI researchers, the median answer put a 50% chance on machines outperforming humans in every possible task by 2047.8

Some AI companies now name superintelligence as their aim. In January 2025, OpenAI chief executive Sam Altman wrote that “we are beginning to turn our aim beyond that, to superintelligence in the true sense of the word.”3 The company Safe Superintelligence Inc. describes building safe superintelligence as “the most important technical problem of our time.”4

That ambition is why the word comes up in debates about AI risk. A statement hosted by the Future of Life Institute calls for “a prohibition on the development of superintelligence” until there is “broad scientific consensus that it will be done safely and controllably” and “strong public buy-in”; its page showed more than 77,000 signatures (October 2026).7 Others reject the framing: Arvind Narayanan and Sayash Kapoor call superintelligence “incoherent as usually conceptualized” and argue AI should be treated as a tool people can keep under control.9 For how likely experts think the worst outcomes are, see our p(doom) guide.

On September 29, 2026, President Trump signed Executive Order 14434, “Inaugurating the Era of Super Intelligence.” It tells federal agencies to use “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” in official correspondence, public communications, websites, reports and policy documents. Regulations, contracts, grants and other documents already issued do not have to change.5

Despite the name, the order’s legal definition is not Bostrom’s. Section 3 defines “Super Intelligence” as the technologies already covered by the legal definition of “artificial intelligence” in 15 U.S.C. 9401(3),5 which describes a “machine-based system” that can “make predictions, recommendations or decisions influencing real or virtual environments.”6 The order also asks the President’s science adviser to propose a new federal definition of the term within 60 days, so this may change.5