What will be left for us to work on?

My keynote at ICML 2026

Jul 13 • Arvind Narayanan

281

25

49

Up the Stack: How AI’s Escape From the Commodity Trap Risks Enterprise Lock-in

Critics and boosters are both looking in the wrong place

Jul 9 • Arvind Narayanan

162

21

27

June 2026

Why AI hasn’t replaced software engineers, and won’t

Coding agents as normal technology

Jun 11 • Arvind Narayanan and Sayash Kapoor

566

83

106

May 2026

Did Google’s AI agents really build an operating system for $916?

The importance of independent evaluation

May 22 • Sayash Kapoor and Arvind Narayanan

140

24

10

Do AI Risks Require Extraordinary Government Intervention?

Let’s not skip the hard work of AI governance

May 21 • Sayash Kapoor and Arvind Narayanan

80

8

18

April 2026

Open-world evaluations for measuring frontier AI capabilities

Introducing CRUX, a new project for evaluating AI on long, messy tasks

Apr 16 • Sayash Kapoor and Arvind Narayanan

90

7

14

February 2026

New Paper: Towards a science of AI agent reliability

Quantifying the capability-reliability gap

Feb 24 • Sayash Kapoor and Arvind Narayanan

321

35

70

AI Won’t Automatically Make Legal Services Cheaper

Applying the AI as Normal Technology framework to legal services

Feb 12 • Justin Curl, Sayash Kapoor, and Arvind Narayanan

75

13

15

January 2026

Fact checking Moravec's paradox

This famous aphorism is neither true nor useful

Jan 29 • Arvind Narayanan

171

10

24

September 2025

A guide to understanding AI as normal technology

And a big change for this newsletter

Sep 9, 2025 • Arvind Narayanan and Sayash Kapoor

213

30

37

July 2025

Could AI slow science?

Confronting the production-progress paradox

Jul 16, 2025 • Sayash Kapoor and Arvind Narayanan

318

37

69

May 2025

AGI is not a milestone

There is no capability threshold that will lead to sudden impacts

May 1, 2025 • Sayash Kapoor and Arvind Narayanan

193

38

40