AI as normal technology
In plain wordsThe 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.
A worldview and policy framework, set out by Arvind Narayanan and Sayash Kapoor, that treats AI as a transformative but normal general-purpose technologygeneral-purpose technology, n.: A technology, like electricity or the computer, that spreads through the whole economy, keeps improving and sparks many other inventions. Many economists think AI is one. whose economic and social effects unfold over decades, not as a new species or an imminent 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.. Its main claims are that the speed of impact is limited by diffusion and adoption, which depend on organizational change, regulation and safety requirements, not only by capability gains; that methods, applications and adoption must be distinguished; that humans will retain control through oversighthuman oversight, n.: Keeping people able to watch, understand, question and, if necessary, stop or overrule an AI system, rather than letting it act unchecked., auditing and institutions, with human work shifting toward specifying and supervising AI; and that policy should favor resilience, reducing uncertainty and addressing known harms over drastic nonproliferation measures premised on speculative catastrophe. It is offered as an alternative to both utopian and dystopian superintelligence narratives.
Treating AI as normal technology, the agency focused on procurement rules and liability rather than compute caps.
Origin
Introduced by Princeton computer scientists Arvind Narayanan and Sayash Kapoor in the essay 'AI as Normal Technology', published by the Knight First Amendment Institute at Columbia University in April 2025, following their book AI Snake Oil (2024).
Usage note
Critics, especially those expecting an intelligence explosionintelligence explosion, n.: The 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. or scenarios such as 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., argue that it underestimates how quickly AI could automate AI research or acquire dangerous capabilities; the authors present it as both prediction and prescription, to be tested against events.
Related
- general-purpose technologygeneral-purpose technology, n.: A technology, like electricity or the computer, that spreads through the whole economy, keeps improving and sparks many other inventions. Many economists think AI is one.
- AI hypeAI hype, n.: Exaggerated claims about what AI can do or will soon do, made to attract money, attention or customers, which can mislead the public and policymakers.
- 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.
- existential riskexistential risk, n.: A danger that could end humanity, or shut it out of a good future forever. Some experts count advanced AI among the largest such risks; others think this worry is overblown.
- productivity paradoxproductivity paradox, n.: The puzzle that a powerful new technology, like computers in the 1980s or AI today, can be everywhere without showing up as faster growth in how much workers produce.
- intelligence explosionintelligence explosion, n.: The 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.
- intelligence explosionintelligence explosion, n.: The 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.
Sources
- Narayanan, A. & Kapoor, S. (2025). AI as Normal Technology: An alternative to the vision of AI as a potential superintelligence. Knight First Amendment Institute at Columbia University.
Mentioned in
AI hypeAI hype, n.: Exaggerated claims about what AI can do or will soon do, made to attract money, attention or customers, which can mislead the public and policymakers., general-purpose technologygeneral-purpose technology, n.: A technology, like electricity or the computer, that spreads through the whole economy, keeps improving and sparks many other inventions. Many economists think AI is one., productivity paradoxproductivity paradox, n.: The puzzle that a powerful new technology, like computers in the 1980s or AI today, can be everywhere without showing up as faster growth in how much workers produce., Arvind NarayananArvind Narayanan, prop. n.: An Indian-American computer scientist at Princeton who studies AI's real-world effects and argues that AI is a 'normal technology' whose impact will unfold gradually.