Narayanan and Kapoor publish 'AI as Normal Technology'

Princeton researchers Arvind Narayanan and Sayash Kapoor published “AI as Normal Technology,” an essay proposing to treat AI the way earlier general-purpose technologies such as electricity or the internet had come to be treated — as consequential but not categorically different from other tools, rather than as an emerging non-human intelligence whose trajectory could only be described in terms of existential risk or utopian transformation. They framed the claim as simultaneously descriptive, predictive and prescriptive: a characterisation of what AI currently was, a forecast of its likely path, and an argument for how policy should respond.

The essay’s central empirical claim was that societal transformation from a technology is paced by diffusion and adoption, not by the underlying rate of technical invention. Narayanan and Kapoor distinguished between methods (research advances), applications (products built on them) and deployment (actual use at scale), arguing that even in domains where AI methods had improved rapidly, high-consequence applications often still ran on “decades-old statistical techniques,” and that the gap between what a model could do in a demonstration and what institutions actually adopted would remain wide for years. They argued this made “superintelligence,” understood as a rapid, discontinuous jump to autonomous human-surpassing capability, an unhelpful frame for thinking about near-term risk.

On policy, the essay argued for regulating deployment rather than development, building on existing safety frameworks from aviation and cybersecurity rather than inventing AI-specific regulatory machinery, and prioritising societal resilience — the capacity to detect and correct harms after they occur — over attempts to prevent capable models from being built or proliferating in the first place. It became one of the most widely cited statements of a position distinct from both the leading AI-safety organisations’ emphasis on existential risk and industry’s more triumphalist framing of imminent transformation, offering a third position grounded in the economic history of technology diffusion.