From the archive · June 26, 2026
The point past which models fail
Is there a horizon in the future we cannot see beyond?
01 · Word
Technological singularity
noun
A hypothesized point where machines become better than people at improving machines, and progress runs away so fast that nothing beyond it can be predicted
Examples
Forecasters disagree wildly about whether a singularity is decades away or impossible.
The word borrows from physics, where a singularity is a point at which the known equations stop making sense.
Origin
The mathematical sense, a point where a function becomes undefined, is old. The technological sense was popularized by the mathematician and science-fiction writer Vernor Vinge in a 1993 essay, drawing on earlier ideas about self-improving machine intelligence.
02 · Idea
The intelligence explosion
In 1965 the statistician I. J. Good described a feedback loop: build a machine even slightly better than humans at designing machines, and it can design a better machine, which designs a better one still. If each step is faster and smarter than the last, intelligence could climb explosively. Good called such a device “ultraintelligent” and remarked that it would be the last invention humanity ever needed to make: the point where the graph of progress could go nearly vertical. A “hard takeoff” is the version where this happens in a rush rather than gradually.
Good's 1965 paper “Speculations Concerning the First Ultraintelligent Machine” laid out the mechanism; Vinge's 1993 essay named the resulting horizon the singularity and argued it could arrive within decades. The idea now anchors much of the debate around advanced artificial intelligence, from optimistic forecasts to serious safety research.
Good's line: the first ultraintelligent machine is the last invention humans would ever need to make.
Once a system can improve the thing that improves it, the curve stops behaving like the past.
Limits and context
The explosion is a hypothesis, not a forecast. Self-improvement could hit hard limits (physical, computational, or economic), and progress could plateau instead of running away. Predicted dates have repeatedly slipped, and critics argue intelligence may not be the kind of thing that scales without bound. Confident timelines say more about the forecaster than the future.
Whether or not a full singularity ever arrives, the loop's early rungs are in place: AI systems already help design chips, write code, and train other models. That is why the once-fringe idea now shapes mainstream research and policy.
03 · Moment
The essay at the NASA symposium
The VISION-21 symposium, sponsored by NASA and the Ohio Aerospace Institute, 1993
In 1993, the mathematician, computer scientist, and Hugo-winning novelist Vernor Vinge presented an essay, “The Coming Technological Singularity: How to Survive in the Post-Human Era,” at a NASA-sponsored symposium called VISION-21. He argued plainly that within a few decades technology would create intelligence greater than human, that this would end the human era, and that the change would be so abrupt that our models of the future break down at that point: a singularity.
- Venue
- NASA VISION-21, 1993
- Author
- Vernor Vinge
- Claim
- End of the human era
The caveat
Vinge offered rough timeframes, and predictions of imminent superhuman AI have repeatedly proven premature; his essay is an argument and a warning, not a schedule. Whether a singularity is inevitable, possible, or incoherent remains genuinely disputed among serious researchers.
Vinge did not invent the underlying idea. He drew on Good's intelligence explosion and even earlier remarks attributed to John von Neumann, but he gave it its enduring name and framing. The essay moved the concept from science fiction toward serious discussion, and over the following decades it grew from a niche futurist notion into an organizing idea for AI research, forecasting, and safety, championed and contested by technologists, philosophers, and companies alike.
The idea that runaway progress marks a point of no return had been voiced before, and a 1950s remark by von Neumann is often quoted. Vinge's essay is what fixed the word “singularity” to it.