Arvind Narayanan is a Professor of Computer Science at Princeton University and Director of Princeton's Center for Information Technology Policy (CITP). With Sayash Kapoor he is a co-author of the book AI Snake Oil (2024) and the "AI as Normal Technology" framework (Knight Columbia, 2025), which argue that AI is a general-purpose technology whose effects will diffuse slowly and that both transformative-AI imminence and AGI existential-risk framings overstate near-term change.
Background
Before his AI-policy work, Narayanan published extensively in security and privacy research. His de-anonymization research (2008 onward) demonstrated statistical re-identification of individuals in datasets that had been treated as anonymous. He conducted web-tracking measurement research on cookie tracking and browser fingerprinting and built the OpenWPM measurement tool. He co-authored the textbook Bitcoin and Cryptocurrency Technologies (2016) on cryptocurrency and blockchain security.
Core works on AI
AI as Normal Technology (2025)
"AI as Normal Technology," written with Kapoor and published by Knight Columbia in 2025, sets out the academic articulation of the thesis that AI is a general-purpose technology, that its diffusion will be slow, and that the superintelligence framing is mistaken (AI as Normal Technology).
AI Snake Oil (2024)
AI Snake Oil (2024), a book co-authored with Kapoor, is the popular-press articulation of the same skeptical framing (AI Snake Oil — Narayanan and Kapoor (2024)). It distinguishes three technologies grouped under the "AI" label: predictive AI, which the authors characterize as largely snake oil (citing recidivism, hiring, and benefits-fraud tools); generative AI, which they describe as powerful but overhyped; and content-moderation AI, which they describe as structurally hard. The book argues that AGI hype inflates all three.
Coding agents as normal technology (2026)
In "Why AI Hasn't Replaced Software Engineers, and Won't" (June 10, 2026), Narayanan and Kapoor extend the normal-technology framing to the labor question, arguing that reported AI-driven software layoffs are largely "AI washing" and that software work follows a "decide-execute-deliver sandwich" in which AI compresses execution while decision-making and accountable delivery resist automation. It is the first in a planned series and bears on AI Labor Disruption and AI Coding Agents.
Policy engagement
Narayanan served as a draft reviewer for the California Report on Frontier AI Policy (June 2025), produced by the California Frontier AI Working Group (2025). His skeptical framing appears in the report as a counterweight to more cautionary reviewer positions.
Positions and statements
Narayanan combines machine-learning research with policy engagement and is associated with the slow-diffusion, normal-technology framing of AI policy. He distinguishes his view from that of pure AI skeptics such as Gary Marcus by granting that AI is transformative while disputing the timeline and nature of its effects.
His framing sets him in disagreement with several other positions tracked on adjacent pages. He rejects the San Francisco Consensus timelines and benefits claims associated with Aschenbrenner and Altman. He rejects the existential-risk framing of Bostrom and the CAIS statement as "AGI hype." He disagrees with the wave framing of Suleyman. He partially aligns with the deregulatory position in Cochrane's "Just Relax" but is more willing than Cochrane to endorse targeted rules.
Relationships
- co-author-with: Sayash Kapoor
- supports: AI Snake Oil — Narayanan and Kapoor (2024) / AI as Normal Technology — his flagship works.
- authored: Why AI Hasn't Replaced Software Engineers, and Won't — June 2026 essay on coding agents and software-engineering labor.
- related: Law of Accelerating Returns (LOAR) — concept he explicitly rejects.
- related: San Francisco Consensus — consensus he dissents from.
- related: Liar's Dividend — parallel empirical argument about deepfakes.
- related: The California Report on Frontier AI Policy — reviewer role.
- related: Stanford HAI — intellectual-adjacent ecosystem.
- related: Ethan Mollick — more optimistic practitioner voice; contrast.