The People Who Imagined the Singularity Before It Had a Name
Everyone knows Kurzweil. He's the megaphone, not the origin. The word wasn't even his. A field guide to the thinkers who actually built the idea — from von Neumann to Epoch.
Unsafe Intelligence
Ray Kurzweil is the face of the singularity, but he's closer to its publicist than its inventor. He didn't coin the word, he didn't invent the mechanism, and the framing he's famous for was defined by a science fiction writer a decade before The Singularity Is Near. The ideas were laid down, mostly, by people the public has never heard of — cryptographers, roboticists, an economist or two, and a mathematician who happened to write novels. Here's the actual lineage, with the deep cuts marked.
The ancestors — before there was a name
John von Neumann (via Stanisław Ulam, 1958) is the literal origin of the word in this sense. Ulam's tribute after von Neumann's death recounts a conversation about "the ever-accelerating progress of technology… approaching some essential singularity in the history of the race beyond which human affairs, as we know them, could not continue." Kurzweil borrowed a sixty-year-old metaphor.
Samuel Butler (1863) got there a century earlier still. "Darwin Among the Machines," and then Erewhon, argued that machines were evolving and would inherit the Earth — the first real machines-surpass-their-makers case. Henry Adams (1904–09) tried, in "A Law of Acceleration," to write accelerating progress down as an actual law and extrapolated it to a near-term phase change — a proto-singularitarian almost no one cites. And Pierre Teilhard de Chardin, in the 1950s, gave us the "Omega Point" and the "noosphere": mind and technology converging toward transcendence. That's the spiritual ancestor of Kurzweil's tone — and of Schmidhuber's "Omega" decades later.
The intelligence explosion — 1950s to 60s
The single most important name most people don't know is I. J. Good (Irving John Good). His 1965 paper, "Speculations Concerning the First Ultraintelligent Machine," coined "intelligence explosion," described a machine that designs better machines, and called it "the last invention that man need ever make." Good was a Bletchley Park cryptographer alongside Turing — and, fittingly, a consultant on 2001: A Space Odyssey.
Alan Turing is famous, but his relevant line is less so: once machines could learn and improve, he wrote in 1951, "we should have to expect the machines to take control." Norbert Wiener, the father of cybernetics, was already warning in God and Golem, Inc. (1964) about learning machines we wouldn't understand.
The definers — Kurzweil's real predecessors
Vernor Vinge is the man who actually named and defined the modern concept. A mathematician and SF novelist, he used "singularity" in a 1983 op-ed and then wrote the canonical 1993 essay, "The Coming Technological Singularity: How to Survive in the Post-Human Era." Kurzweil popularized Vinge. (Vinge died in March 2024.)
Hans Moravec, the CMU roboticist, supplied the transhumanist machinery in Mind Children (1988) and Robot (1999): mind-uploading, robotic succession, and "Moravec's paradox." K. Eric Drexler's Engines of Creation (1986) added the "atoms on demand" half of the toolkit — and decades later, his Comprehensive AI Services model became the best non-explosion picture of advanced AI we have. Ray Solomonoff, founder of algorithmic probability, gave a 1985 talk discussing an "infinity point" of AI progress — a true insider's insider.
The movement — the people in the know
Max More and FM-2030 (Fereidoun Esfandiary) were the organizers: More coined "extropy" and ran the Extropy Institute, the 1990s mailing-list crucible where this whole subculture formed; FM-2030 wrote Are You a Transhuman? (1989). Robin Hanson, an economist, is the rigorous contrarian — The Age of Em (2016) and the famous 2008 "AI-Foom Debate" with Eliezer Yudkowsky, who founded the Singularity Institute (now MIRI), coined "Friendly AI," and wrote the cleanest taxonomy of the whole muddle ("three schools of singularity," 2007).
Nick Bostrom systematized it academically in Superintelligence (2014). Jürgen Schmidhuber gave it formal mathematics — the Gödel machine and recursive self-improvement. Ben Goertzel popularized the term "AGI" and ran its conference series; Shane Legg (later a DeepMind co-founder) gave AGI an operational definition; David Chalmers brought analytic philosophy to bear ("The Singularity: A Philosophical Analysis," 2010); and Bill Joy's "Why the Future Doesn't Need Us" (2000) was the prominent doomer counterweight that made everyone take it seriously.
The modern successors — the empirical turn
The new generation traded prophecy for data. The most influential figure almost no one outside the field knows is Carl Shulman, foundational to modern takeoff modeling. Ajeya Cotra (biological anchors), Tom Davidson (compute-centric takeoff), and Paul Christiano (alignment) turned vibes into models. Epoch AI — Jaime Sevilla, Tamay Besiroglu, Ege Erdil — turned "is it accelerating?" into measured trendlines. And Leopold Aschenbrenner (Situational Awareness) and Daniel Kokotajlo (AI 2027) are the scenario-writers carrying Vinge's torch to specific dates.
The skeptics who sharpened it
Worth knowing because they pushed the field forward by attacking it: Hubert Dreyfus (What Computers Can't Do, 1972), Theodore Modis (the physicist who took apart Kurzweil's curve-fitting in "The Singularity Myth"), and today Gary Marcus and François Chollet (ARC-AGI). A field is only as good as its best critics, and these are load-bearing.
The one-line throughline: von Neumann named it → Good gave it a mechanism → Vinge defined it → Moravec and Drexler furnished it → Bostrom, Yudkowsky, and Hanson made it rigorous → Kurzweil sold it → Epoch and Aschenbrenner are now measuring it. Kurzweil gets the cover. Everyone above did the work.