intelligence explosion
In plain wordsThe idea that once AI gets good enough at improving AI, each smarter version could build a smarter one faster still, so abilities would snowball far beyond human level in a short time.
A hypothesized runaway increase in machine intelligence driven by a feedback loop in which AI systems improve the systems that succeed them. If AI can do AI research at least as well as human researchers, each gain in capability speeds up the next, and progress could accelerate until it meets physical or economic limits, possibly within months. A 2025 analysis by Tom Davidson, Rose Hadshar and William MacAskill of the research group Forethought distinguishes three types by what must be automated: a software intelligence explosion, in which AI improves algorithms and training methods on existing chips; an AI-technology intelligence explosion, in which AI also improves chip technology; and a full-stack intelligence explosion, in which AI and robots also expand chip production. Whether any loop explodes or fizzles depends on whether research returns diminish faster than capability grows, a quantitative question about algorithmic progress, scaling lawsscaling laws, n.: Measured patterns showing that an AI model's error shrinks smoothly and predictably as the model is made bigger and given more data and computing power, so labs can forecast a giant model's performance from small, cheap experiments. and computecompute, n.: The raw computing power used to train and run AI, usually measured by how many specialized chips are used and for how long. With data and algorithms, it is a main driver of AI progress. bottlenecks on which forecasters still disagree.
If returns to software research hold up, automating our researchers could set off an intelligence explosion within a year.
Origin
Coined by I. J. Good in 1965, who argued that an ultraintelligent machine could design better machines, so there would 'unquestionably be an intelligence explosion'. Eliezer Yudkowsky and the Singularity Institute (now MIRI) revived the phrase in the 2000s, David Chalmers analyzed it philosophically in 2010, and from 2021 economists and forecasters such as Tom Davidson began modeling it quantitatively.
Usage note
Often equated with the technological singularitytechnological singularity, n.: A moment some people expect in the future when technology, usually AI smarter than us, starts changing the world so fast and so deeply that we cannot see past it., but the explosion is a mechanism and the singularity a claimed consequence. Skeptics argue that research is bottlenecked by experiments, compute and physical capital rather than by thinking speed, so automating researchers would speed progress far less than the feedback picture implies.
Related
- recursive self-improvementrecursive self-improvement, n.: An AI improving its own design or training, so that each improved version is better at making the next improvement. It is the engine usually imagined behind an intelligence explosion.
- automated AI researchautomated AI research, n.: Using AI systems to do the work of AI researchers, such as running experiments and writing code for new models, so that AI helps build its own successors.
- takeofftakeoff, n.: The stretch of time between AI reaching roughly human level and AI becoming vastly superhuman. People argue over whether it would take decades, years, months or just days.
- FOOMFOOM, n.: Slang for a sudden, explosive jump in an AI's abilities, the AI going 'foom' and racing from about human level to vastly superhuman in a very short time.
- technological singularitytechnological singularity, n.: A moment some people expect in the future when technology, usually AI smarter than us, starts changing the world so fast and so deeply that we cannot see past it.
- ultraintelligent machineultraintelligent machine, n.: I. J. Good's 1965 name for a machine smarter than any person at every kind of thinking, which could therefore build machines smarter than itself.
- compute overhangcompute overhang, n.: A situation in which far more computing power is available than AI currently uses, so that a new method or a decision to scale up could suddenly produce a big jump in ability.
- superintelligencesuperintelligence, n.: A mind, usually imagined as an AI, that would leave even the cleverest people far behind in nearly every kind of thinking, not just in one game or skill. None is known to exist.
Sources
- Good, I. J. (1965). Speculations Concerning the First Ultraintelligent Machine. Advances in Computers 6.
- Chalmers, D. J. (2010). The Singularity: A Philosophical Analysis. Journal of Consciousness Studies 17(9–10).
- Davidson, T., Hadshar, R. & MacAskill, W. (2025). Three Types of Intelligence Explosion. Forethought, 17 March 2025. forethought.org/research/three-types-of-intelligence-explosion
Mentioned in
superintelligencesuperintelligence, n.: A mind, usually imagined as an AI, that would leave even the cleverest people far behind in nearly every kind of thinking, not just in one game or skill. None is known to exist., technological singularitytechnological singularity, n.: A moment some people expect in the future when technology, usually AI smarter than us, starts changing the world so fast and so deeply that we cannot see past it., AI 2027AI 2027, prop. n.: A detailed 2025 forecast, written as a month-by-month story, imagining AI becoming superhuman by the end of 2027, with two alternative endings depending on whether humanity slows down., AI as normal technologyAI as normal technology, phr.: The view that AI, however important, will change society gradually, like electricity or the internet, through slow adoption by people and institutions, rather than as a sudden, uncontrollable superintelligence., artificial superintelligenceartificial superintelligence, n.: Superintelligence in machine form: AI that would outthink the best humans at nearly every intellectual task. Often shortened to ASI and pictured as the stage after human-level AI., automated AI researchautomated AI research, n.: Using AI systems to do the work of AI researchers, such as running experiments and writing code for new models, so that AI helps build its own successors., David ChalmersDavid Chalmers, prop. n.: An Australian philosopher who named the 'hard problem' of consciousness — why physical processes feel like anything at all — and who asks whether AI systems could be conscious., Eliezer YudkowskyEliezer Yudkowsky, prop. n.: An American writer and AI theorist who, from around 2000, argued that superhuman AI would likely destroy humanity unless its goals were made safe, and who founded an online community devoted to rational thinking., fast takeofffast takeoff, n.: A scenario in which AI goes from about human level to far beyond it very quickly, in days, weeks or a few months, leaving people little time to react., I. J. GoodI. J. Good, prop. n.: A British statistician who broke codes with Alan Turing and in 1965 argued that a machine smarter than us could design still smarter machines, setting off an 'intelligence explosion'., Machine Intelligence Research InstituteMachine Intelligence Research Institute, prop. n.: A small nonprofit in Berkeley, California, among the first to warn that smarter-than-human AI could destroy humanity, which now campaigns for a worldwide halt to building it., recursive self-improvementrecursive self-improvement, n.: An AI improving its own design or training, so that each improved version is better at making the next improvement. It is the engine usually imagined behind an intelligence explosion., self-improvementself-improvement, n.: When an AI gets better by learning from material it produces itself — its own answers, critiques or practice problems — rather than only from new human-made examples., takeofftakeoff, n.: The stretch of time between AI reaching roughly human level and AI becoming vastly superhuman. People argue over whether it would take decades, years, months or just days., Vernor VingeVernor Vinge, prop. n.: An American science-fiction author and mathematics professor who popularized the 'technological singularity': a coming point when smarter-than-human minds change the world so fast that no one can foresee what follows., AI Futures ProjectAI Futures Project, prop. n.: A small nonprofit research group that writes detailed forecasts and scenarios of how advanced AI could develop, best known for the 2025 scenario AI 2027., compute overhangcompute overhang, n.: A situation in which far more computing power is available than AI currently uses, so that a new method or a decision to scale up could suddenly produce a big jump in ability., Daniel KokotajloDaniel Kokotajlo, prop. n.: An American AI forecaster and former OpenAI researcher who led 'AI 2027', a detailed scenario of a race to superintelligence, and campaigns for transparency at AI companies., explosive growthexplosive growth, n.: A hypothetical jump in the world economy's growth rate to 30% a year or more, about ten times today's pace, which some economists think AI that replaces human workers could cause., FOOMFOOM, n.: Slang for a sudden, explosive jump in an AI's abilities, the AI going 'foom' and racing from about human level to vastly superhuman in a very short time., ForethoughtForethought, prop. n.: A small Oxford research nonprofit that asks what problems, besides keeping AI safe, the rapid arrival of superintelligent AI would create, from power grabs to the rights of digital minds., Gödel machineGödel machine, n.: A purely theoretical design, proposed by Jürgen Schmidhuber, for a program allowed to change anything about itself, including how it learns, but only once it has proved that the change is an improvement., internal deploymentinternal deployment, n.: When an AI company uses its own newest, unreleased AI systems inside the company — for example to write code or do research — before or instead of offering them to the public., law of accelerating returnslaw of accelerating returns, n.: Ray Kurzweil's claim that technology improves at an ever-faster rate, because each advance helps build the next, so progress speeds up exponentially rather than steadily.