The word singularity is used loosely and often misleadingly. In popular culture it has come to mean something like “the moment AI takes over” — a vague catastrophe or utopia arriving from the future. In serious scientific and philosophical discussion it means something considerably more specific, more interesting, and more uncertain.
A singularity, in its precise mathematical sense, is a point at which a function reaches infinity or becomes undefined — a point beyond which the mathematics breaks down and predictions become impossible. Black holes contain gravitational singularities. The Big Bang emerged from a cosmological singularity. These are points where our physical theories, which work extraordinarily well everywhere else, stop being able to tell us what happens next.
The technological singularity is a proposed analogy: a point in the development of artificial intelligence at which the pace of technological change becomes so rapid, and AI capabilities so far beyond human comprehension, that prediction of what comes next becomes impossible. Beyond the singularity, by definition, we do not know what we do not know.
Ray Kurzweil — inventor, futurist, and now a principal researcher at Google — has predicted the technological singularity will occur around 2045. More recently, projections based on the accelerating pace of AI development have suggested that something like a singularity threshold — a point at which AI systems surpass human capabilities across virtually all cognitive domains — may arrive closer to 2037. This article examines what that claim means, what evidence supports or challenges it, and what the stakes are if it proves even approximately correct.
The History of an Idea
The concept of a technological singularity has a longer history than its recent popular prominence suggests. The mathematician John von Neumann, one of the founders of modern computing, noted in conversation with his colleague Stanislaw Ulam in the early 1950s that technological change seemed to be accelerating toward “some essential singularity in the history of the race beyond which human affairs, as we know them, could not continue.” This was decades before the personal computer, let alone the internet or modern AI.
Alan Turing, in his landmark 1950 paper “Computing Machinery and Intelligence” — the paper that introduced the Turing Test — speculated about machines that could learn and improve, noting that once such machines existed, they would quickly surpass human intelligence. Turing was right that the question was important. He did not live to see how difficult it would prove to build such machines.
The science fiction writer Vernor Vinge popularised the term “technological singularity” in a 1993 essay, defining it specifically as the creation of superhuman intelligence — AI that exceeds human cognitive capabilities — which would trigger a runaway process of self-improvement leading to outcomes beyond human prediction or control.
Ray Kurzweil, in his 2005 book The Singularity Is Near, grounded the prediction in what he called the Law of Accelerating Returns — his observation that information technology improves exponentially over time, with each generation of technology enabling the next generation to be built faster. Kurzweil predicted this exponential progression would lead to machines matching and then exceeding human intelligence by 2029, with the full singularity — the merger of human and machine intelligence — arriving around 2045. In 2024, he updated his prediction, suggesting 2029 for human-level AI was still on track.
The 2037 Projection: What It Is Based On
The 2037 timeframe has gained traction from several converging lines of evidence, none individually conclusive but collectively suggestive.
The first is the observed pace of large language model development. From GPT-2 in 2019 to GPT-4 in 2023 to the frontier models of 2025 and 2026, AI capabilities have advanced dramatically in a very short time. Systems that could not hold a coherent conversation in 2018 can now write code, pass professional examinations, conduct scientific reasoning, and engage in sophisticated multi-step planning. Extrapolating this trajectory — with appropriate caution about the difficulty of extrapolation — suggests that systems approaching or matching human-level performance across broad cognitive domains could arrive before 2040.
The second line of evidence comes from AI researchers themselves. Surveys of AI researchers — including the influential AI Impacts survey — have found that the median estimate among researchers for human-level machine intelligence has moved significantly earlier over the past decade. In 2022, the median estimate was 2059. By 2024, it had moved to the mid-2030s for several researchers and organisations tracking the question.
The third is compute scaling. The amount of computing power used to train frontier AI models has been doubling approximately every six months for the past decade — a rate far faster than Moore’s Law for general computing. As long as this scaling continues to produce capability improvements — which is an empirical question that could change — the trajectory points toward increasingly powerful systems arriving on a compressed timeline.
Importantly, none of this implies that 2037 is certain, or even likely. The history of AI is littered with confident predictions of imminent human-level intelligence that were not borne out. The 2037 figure represents the central estimate of a distribution with enormous uncertainty on both sides — human-level AI could arrive much sooner, or it could take much longer if current scaling approaches hit fundamental limits.
What the Singularity Would Actually Mean
Discussions of the singularity often conflate several distinct scenarios that deserve to be kept separate, because they have very different implications and different levels of plausibility.
Human-level AI — artificial general intelligence (AGI) capable of performing any cognitive task a human can perform — is the threshold most commonly discussed. Its arrival would be transformative but not necessarily catastrophic. Humans have, after all, been building tools that extend our cognitive capabilities for centuries. AGI would be a more powerful tool than any previous one, with significant economic and social implications, but it would not by itself constitute a singularity in the sense of making the future unpredictable.
Recursive self-improvement is the scenario that most deserves the name singularity. If an AI system capable of improving its own capabilities is created, and if those improvements make the system better at improving itself, the result could be an extremely rapid and potentially uncontrollable increase in AI capabilities — an intelligence explosion. Whether this scenario is physically possible, and whether it would unfold on timescales of years, months, or seconds, is deeply uncertain. Most AI researchers consider it a genuine long-term risk but not an imminent one.
Artificial superintelligence — AI that exceeds human cognitive capabilities in every domain — is the scenario most associated with existential risk. The concern, articulated by philosophers including Nick Bostrom and researchers at organisations including Anthropic and OpenAI, is that a superintelligent system pursuing goals misaligned with human values could cause catastrophic harm — not through science fiction malevolence but through the ordinary execution of objectives that we specified imperfectly. Solving this alignment problem before superintelligence arrives is the central concern of AI safety research.
The Cosmological Singularity: A Different but Related Concept
The technological singularity is not the only kind of singularity relevant to our understanding of reality. In cosmology, singularities arise in two places: the Big Bang and black holes.
The Big Bang singularity is the point at the beginning of time at which the density and temperature of the universe were, mathematically, infinite. General relativity predicts this singularity but cannot describe what happens within it — the theory breaks down at the Planck scale. Understanding the Big Bang requires a theory of quantum gravity that we do not yet possess.
Black hole singularities are the points at the centres of black holes where, according to general relativity, matter is compressed to infinite density. The singularity theorems of Roger Penrose and Stephen Hawking proved that singularities are not exotic mathematical curiosities but are predicted to occur whenever matter is compressed beyond a certain density under general relativistic gravity.
Both types of singularity point to the same deep problem: our best theories of physics predict their own breakdown at extreme conditions. Resolving these singularities requires physics beyond what we currently know — and the search for that physics is one of the most active frontiers in theoretical science. For a look at how these ideas connect to the quantum vacuum and the origin of the universe, see our article on from nothing to everything: how the universe emerged from the quantum vacuum.
What We Should Actually Do About the Technological Singularity
Whether the singularity arrives in 2037, 2045, or 2075, the trajectory of AI development raises questions that need to be addressed now — before the most powerful systems exist, while there is still time to shape how they are built and governed.
AI safety research — the attempt to ensure that AI systems pursue goals aligned with human values, and that their behaviour remains predictable and controllable as capabilities increase — is arguably the most important technical research programme of our time. It is also significantly underfunded relative to AI capabilities research, a disparity that several major research organisations and governments are beginning to address.
Governance frameworks for AI development — international agreements, regulatory structures, and accountability mechanisms — are being developed in real time, with significant uncertainty about whether they will be adequate to the challenge. The EU’s AI Act, executive orders on AI safety from the US government, and multilateral discussions at forums including the AI Safety Summit represent early steps.
For a broader look at how quantum computing — another transformative technology with potentially civilisation-scale implications — is developing alongside AI, see our article on quantum computing in 2026. And for an exploration of the deepest questions about consciousness that the singularity ultimately raises — what would it mean for a machine to be conscious, and how would we know? — see our article on the hard problem of consciousness.
Frequently Asked Questions
What is the technological singularity?
The technological singularity is a proposed point in the future at which artificial intelligence surpasses human cognitive capabilities, triggering a period of accelerating technological change so rapid that it becomes impossible to predict what comes after. The concept was popularised by Vernor Vinge and Ray Kurzweil, drawing on earlier ideas from mathematicians John von Neumann and Alan Turing.
Why is 2037 mentioned as a significant date?
2037 represents a central estimate derived from extrapolating the current pace of AI development, compute scaling trends, and shifting estimates among AI researchers for when human-level machine intelligence might be achieved. It is an uncertain projection, not a prediction — the actual timeline could be earlier or considerably later depending on whether current scaling approaches continue to yield capability improvements.
Is the singularity the same as artificial general intelligence?
Not exactly. Artificial general intelligence — AI capable of performing any cognitive task a human can — is one threshold on the path to a singularity. The singularity proper refers to a point beyond which prediction becomes impossible, typically associated with recursive self-improvement or superintelligence rather than merely human-level capability.
What is the difference between a technological and a cosmological singularity?
A cosmological singularity is a mathematical point in spacetime — such as the Big Bang or the centre of a black hole — where physical quantities become infinite and current theories break down. A technological singularity is a proposed point in technological development where the pace of change becomes too rapid for human prediction. Both share the concept of a point beyond which existing frameworks cannot extend.
Should we be afraid of the singularity?
Fear is less useful than preparation. The genuine risks associated with advanced AI — misalignment, misuse, concentration of power — are serious and warrant serious research and governance work. They are also not inevitable. The trajectory of AI development is shaped by choices being made now, and those choices matter enormously.
What is AI alignment?
AI alignment is the research problem of ensuring that AI systems pursue goals consistent with human values and remain under meaningful human control as their capabilities increase. It is considered one of the most important and difficult problems in AI development, and is the primary focus of organisations including Anthropic, the Machine Intelligence Research Institute, and the AI safety teams at major AI laboratories.
Further Reading
- Wikipedia — Technological Singularity
- Wikipedia — AI Alignment
- Stanford Encyclopedia of Philosophy — Artificial Intelligence
- The Singularity Is Near by Ray Kurzweil
- Superintelligence by Nick Bostrom
Sources
- Wikipedia — Technological Singularity
- Wikipedia — Artificial General Intelligence
- Wikipedia — AI Alignment
- Wikipedia — Ray Kurzweil
- Web News For Us — Quantum Computing in 2026
- Web News For Us — The Hard Problem of Consciousness
- Web News For Us — From Nothing to Everything
About the Author
Baryon is the founder and editor of Web News For Us. Driven by a deep fascination with the biggest unanswered questions in science — from quantum physics and cosmology to the nature of consciousness and the genetic code written into every living cell — he has spent years studying modern physics, biology, and the history of scientific thought. He covers Science & AI, Space, Genetics & Research, and the timeless wisdom of history’s greatest thinkers and mystics.
If you have ever looked at the night sky and felt that pull to understand what is out there — or the wonder of an entire universe coiled inside your genes — you are in the right place.
Discover more from Web News For Us
Subscribe to get the latest posts sent to your email.
