Singularity

The technological singularity is a hypothesized point beyond which technological change becomes so fast and so deep that present-day models of the future stop working. Most accounts attribute it to the creation of superhuman intelligence.

Where the term comes from

In “John von Neumann 1903–1957”, a memorial in the Bulletin of the American Mathematical Society in 1958, the mathematician Stanislaw Ulam recalled a conversation with von Neumann about the ever-accelerating progress of technology. It seemed to approach, in Ulam’s words, “some essential singularity in the history of the race”, beyond which human affairs as known could not continue. In 1965 I. J. Good supplied a mechanism: a machine cleverer than any human could design still better machines, producing an intelligence explosion (see FOOM).

The modern sense comes from Vernor Vinge, a San Diego State University mathematician and science-fiction writer. After a short piece in Omni in January 1983, he set out the idea in “The Coming Technological Singularity: How to Survive in the Post-Human Era”. He presented it at the VISION-21 symposium, sponsored by NASA’s Lewis Research Center and the Ohio Aerospace Institute, on March 30–31, 1993. Vinge predicted that within thirty years humanity would have the means to create superhuman intelligence, and that the human era would end shortly after. He named four routes: AI, computer networks that become superhumanly intelligent, intimate human–computer interfaces, and biological enhancement of human intellect. He would be surprised, he wrote, if it happened before 2005 or after 2030. He called it a singularity because at that point our models must be discarded. He noted that von Neumann had seemed to mean ordinary progress, not superhuman intellect.

Kurzweil’s version

Ray Kurzweil made the term popular. In a 2001 essay, “The Law of Accelerating Returns”, he argued that technological progress is exponential, so that the 21st century would bring not a hundred years of progress at today’s rate but something like twenty thousand. The Singularity Is Near: When Humans Transcend Biology (Viking, September 2005) described the singularity as a period of change so rapid and profound that human life would be irreversibly transformed, with biology and technology merging. Its sequel, The Singularity Is Nearer: When We Merge with AI (June 2024), kept the timetable: human-level AI by 2029 and the singularity by 2045, when, Kurzweil told The Observer, brain–computer interfaces would expand intelligence “a millionfold”.

Singularity and FOOM

The terms overlap but answer different questions. FOOM is a claim about a mechanism: a single AI improving itself so fast that it outruns everyone else. The singularity is a claim about a horizon: beyond some point, prediction fails. A FOOM would produce a singularity, but a singularity need not come from a FOOM. Vinge allowed for routes through networks, interfaces and biology; Kurzweil describes a smooth exponential across many technologies that ends in merger rather than in one machine overtaking humanity. The 2008 debate between Eliezer Yudkowsky and Robin Hanson was in effect a dispute over which kind to expect.

Criticism

• No evidence. In a 2008 IEEE Spectrum survey of technologists, the psychologist Steven Pinker wrote: “There is not the slightest reason to believe in a coming singularity.”
• The complexity brake. In MIT Technology Review in October 2011, Paul Allen and Mark Greaves argued that accelerating returns are an extrapolation, not a law, and that the software of intelligence depends on an understanding of cognition that advances irregularly. Kurzweil replied that aggregate trends in information technology are predictable even when individual efforts are not.
• Secular eschatology. Critics have compared the idea to religious end-time expectations; IEEE Spectrum’s 2008 special report on the singularity included an essay titled “Waiting for the Rapture”.

Why it matters now

Vinge’s window closes in 2030 and Kurzweil’s first milestone falls in 2029, so both will soon be tested. The International AI Safety Report 2026 does not use the word. Its four scenarios for 2030 run from stalled progress to AI systems comparable to human remote workers, and it finds little expert consensus on which is most likely. Its concern is close to Vinge’s all the same: if AI speeds up AI research, it warns, decades of progress could happen in years.

Related

FOOM · Scaling Laws · Sharp Left Turn · p(doom) · A Horizon, Not a Date (essay) · What is AIciety?

Sources: Stanislaw Ulam, “John von Neumann 1903–1957”, Bull. Amer. Math. Soc. 64 (1958) · Vernor Vinge: The Coming Technological Singularity (1993) · Ray Kurzweil: The Law of Accelerating Returns · Press release for The Singularity Is Near (2005) · Zoë Corbyn, interview with Ray Kurzweil, The Observer, 29 June 2024 (syndicated copy) · IEEE Spectrum: Tech Luminaries Address Singularity (2008) · IEEE Spectrum: Special Report: The Singularity · Paul Allen and Mark Greaves: The Singularity Isn’t Near (2011) · Ray Kurzweil: Don’t Underestimate the Singularity (2011) · International AI Safety Report 2026 · Wikipedia: Technological singularity (orientation) · Wikipedia: The Singularity Is Nearer (orientation)