Princeton professor and CITP director, AI policy and evaluation researcher
Arvind Narayanan
Profile
Arvind Narayanan is a computer science professor at Princeton and director of the university’s Center for Information Technology Policy. If you want someone who actually reads the papers, understands the math, and will still tell you the emperor has no clothes, this is your guy. His work is the single best counterweight to AI marketing currently in circulation — not because he dismisses the technology, but because he takes it seriously enough to demand evidence.
Before AI became his main beat, Narayanan was already a force in technical privacy and security research. His 2008 paper with Vitaly Shmatikov on de-anonymizing the Netflix Prize dataset is a classic — showing that “anonymized” data often isn’t, using nothing more than public IMDb ratings as a side channel. He later co-authored Bitcoin and Cryptocurrency Technologies, the Princeton textbook that many developers still cite as the cleanest technical introduction to the field.
Today his focus is AI hype and accountability, mostly with his PhD student and co-author Sayash Kapoor. Together they wrote AI Snake Oil — the book and the Substack — and in 2025 published AI as Normal Technology, a long essay arguing that AI is a general-purpose technology like electricity or the internet, not a superintelligent alien. The piece is worth reading even if you disagree with it: it’s the most rigorous articulation of the “moderate optimist” position out there, and it directly engages with both the doomers and the accelerationists on their own terms.
For developers, Narayanan’s value is simple. When you see a breathless claim — “AI predicts crime,” “AI screens resumes better than humans,” “AI will collapse the economy in 18 months” — his writing is where you go to figure out what’s actually being measured, what the benchmark is hiding, and whether anyone has reproduced the result. He and Kapoor made the TIME100 AI list in 2023 for exactly this reason: evidence-based criticism is scarce, and theirs is the best.
Books
Key Articles & Papers
AI as Normal Technology AI Snake Oil (Substack newsletter) Robust De-anonymization of Large Sparse Datasets Leakage and the Reproducibility Crisis in ML-based Science GPT-4 and professional benchmarks: the wrong answer to the wrong question A guide to understanding AI as normal technology Evaluating LLMs is a minefieldVideos
YouTube
Spotify Podcasts
