“AI as Normal Technology” is an essay by the Princeton computer scientist Arvind Narayanan and the researcher Sayash Kapoor, published in April 2025 by the Knight First Amendment Institute at Columbia University. It offers the leading intellectual alternative to the view that AI is on a fast path to superintelligence. The authors argue that AI is better understood as a normal technology - transformative, like electricity or the internet, but still a tool that society can and should remain in control of, rather than a separate species or an autonomous agent racing ahead of human oversight.
Their case rests on the observation that technological impact unfolds through slow, interlocking processes: inventing methods, building applications, and adopting them across organizations, each on its own timescale. They argue that benchmark results, like an AI passing the bar exam, systematically overstate real-world capability, because the tasks easiest to measure are the least representative of messy professional work. On this view, transformation will be gradual enough for institutions to adapt, and a future with advanced AI will involve more human jobs focused on overseeing, specifying, and controlling AI, not mass redundancy.
The essay is notable for how it reorders risk. Rather than prioritizing speculative catastrophe such as the paperclip-maximizer scenario, the authors elevate systemic, accumulating harms - inequality, surveillance, bias entrenchment, and erosion of democratic institutions - as the risks most strongly supported by the history of past technologies. They recommend resilience and sector-specific regulation over sweeping restrictions premised on uncontrollable AI, and they treat misalignment as a genuine but speculative research problem rather than an imminent certainty.
Why a general reader should care: this essay frames the main counter-position in today’s AI debate. How seriously a company or government takes superintelligence risk versus normal-technology risk shapes very different decisions about regulation, investment, and adoption.