Arvind Narayanan and Sayash Kapoor · Sep 9, 2025
Arvind Narayanan and Sayash Kapoor explain AI as normal technology
Markdown · JSON · Original source
On September 9, 2025, Arvind Narayanan and Sayash Kapoor argue that treating AI as normal technology does not mean expecting it to be mundane or predictable. Their framework says AI’s social effects emerge through deployment and adoption, not capability development alone, leaving people, organisations and policymakers points of leverage. They argue that institutions must adapt to unforeseen effects rather than rely on prediction or try to prevent all harm; they call this approach resilience. The authors also distinguish deployment—making capabilities available in products—from diffusion: how people actually use them, for how long and for what purposes. Instant deployment, they argue, makes adoption feel unusually fast because people must keep deciding whether to use AI, even while learning and organisational change take longer. They conclude that AI is not going away or becoming a niche, and that the framework offers actionable guidance for workers, business leaders, students and policymakers. They say the framework is a way to think about possible impacts, not a forecast of particular outcomes. How quickly AI use will spread through workplaces, beyond access to products, remains unresolved.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
Arvind Narayanan and Sayash Kapoor · Article
Source details
Original source date: Sep 9, 2025 (2025-09-09T13:00:40.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-12T06:13:23.737Z. First observed: 2026-10-23T20:05:47.406Z. Retrieved: 2026-10-23T20:23:35.551Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 23, 2026; originally published September 9, 2025.
Open question
What evidence will show whether AI use is spreading through workplaces and changing workflows, rather than merely becoming available in products?
Source-stated
Narayanan and Kapoor argue that AI’s effects on society depend on deployment and that people and institutions have leverage over those effects.
Benefits and risks are realized when AI is deployed, not when it is developed.
Source-stated
The authors say resilience means adapting policy to impacts that cannot be reliably anticipated.
the policymaking approach that enables such adaptability is called resilience
Source-stated
Narayanan and Kapoor distinguish product deployment from diffusion, which they define through actual use and its duration and purpose.
What matters is how many people are actually using them, how long they are using them, and what they are using them for.
Source-stated
The authors argue that instant deployment makes AI adoption feel faster because people must repeatedly decide whether to use it, although learning and organisational change take longer.
People have to spend a much higher percentage of time evaluating whether to adopt AI for some particular use case
Source-stated
The authors conclude that AI will not become a niche people can ignore, and say their framework offers actionable guidance to several groups.
AI is not going away, nor will it become a niche like crypto that most people can ignore.
Source and contributor context
Arvind Narayanan and Sayash Kapoor · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Editorial provenance
Version 1. Publication ID: dfa6e366-1623-df80-d8fb-0b41887dc71a.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-12T06:08:47.640Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: 78af43ff394202bbfe2013e1bf70e8f6574cd7ca02549ed10567b0972fee70b4. Source text hash: 49a446cbff1862207f0014e36c395b1e4bcd1bb02c713a4ba9e987d18f975235.
Revision and source-check history
Version 1 · 2026-11-12T06:13:23.737Z
Arvind Narayanan and Sayash Kapoor explain AI as normal technology
On September 9, 2025, Arvind Narayanan and Sayash Kapoor argue that treating AI as normal technology does not mean expecting it to be mundane or predictable. Their framework says AI’s social effects emerge through deployment and adoption, not capability development alone, leaving people, organisations and policymakers points of leverage. They argue that institutions must adapt to unforeseen effects rather than rely on prediction or try to prevent all harm; they call this approach resilience. The authors also distinguish deployment—making capabilities available in products—from diffusion: how people actually use them, for how long and for what purposes. Instant deployment, they argue, makes adoption feel unusually fast because people must keep deciding whether to use AI, even while learning and organisational change take longer. They conclude that AI is not going away or becoming a niche, and that the framework offers actionable guidance for workers, business leaders, students and policymakers. They say the framework is a way to think about possible impacts, not a forecast of particular outcomes. How quickly AI use will spread through workplaces, beyond access to products, remains unresolved.
Original source
Arvind Narayanan and Sayash Kapoor · Article
Source details
Original source date: Sep 9, 2025 (2025-09-09T13:00:40.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-12T06:13:23.737Z. First observed: 2026-10-23T20:05:47.406Z. Retrieved: 2026-10-23T20:23:35.551Z.
Limits
Text as retrieved on October 23, 2026; originally published September 9, 2025.
Open question
What evidence will show whether AI use is spreading through workplaces and changing workflows, rather than merely becoming available in products?
Source-stated
Narayanan and Kapoor argue that AI’s effects on society depend on deployment and that people and institutions have leverage over those effects.
Benefits and risks are realized when AI is deployed, not when it is developed.
Source-stated
The authors say resilience means adapting policy to impacts that cannot be reliably anticipated.
the policymaking approach that enables such adaptability is called resilience
Source-stated
Narayanan and Kapoor distinguish product deployment from diffusion, which they define through actual use and its duration and purpose.
What matters is how many people are actually using them, how long they are using them, and what they are using them for.
Source-stated
The authors argue that instant deployment makes AI adoption feel faster because people must repeatedly decide whether to use it, although learning and organisational change take longer.
People have to spend a much higher percentage of time evaluating whether to adopt AI for some particular use case
Source-stated
The authors conclude that AI will not become a niche people can ignore, and say their framework offers actionable guidance to several groups.
AI is not going away, nor will it become a niche like crypto that most people can ignore.
Source and contributor context
Arvind Narayanan and Sayash Kapoor · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Editorial provenance
Version 1. Publication ID: dfa6e366-1623-df80-d8fb-0b41887dc71a.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-12T06:08:47.640Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: 78af43ff394202bbfe2013e1bf70e8f6574cd7ca02549ed10567b0972fee70b4. Source text hash: 49a446cbff1862207f0014e36c395b1e4bcd1bb02c713a4ba9e987d18f975235.
uncertain · 2026-11-12T06:08:47.640Z · anthropic/claude-opus-5.5
Publication: dfa6e366-1623-df80-d8fb-0b41887dc71a
attribution — pass: The reading consistently attributes its claims to Arvind Narayanan and Sayash Kapoor, who are the bylined authors of this single newsletter essay. Their positions stay framed as their arguments, not as established fact.
Arvind Narayanan and Sayash Kapoor Sep 09, 2025
summary — uncertain: The reading drops the essay's central framing: normal technology is offered as an alternative to the dominant superintelligence worldview, and is set against AI 2027. It also softens the authors' force on diffusion. They flatly reject the rapid-adoption claim, but the reading only says adoption feels fast and ends by calling the speed unresolved. Smallest repair: add one sentence on the contrast with the superintelligence worldview, and one stating that they argue actual diffusion is slow.
AI boosters believe that it is being adopted unprecedentedly rapidly. We completely disagree.
date — pass: September 9, 2025 matches originalPublished (2025-09-09 UTC) and the source dateline. The kind is publisher-page, so no archive disclosure is needed. The reading uses no later material. The book plans are not presented as outcomes.
Sep 09, 2025
support — uncertain: Most claims match their excerpts. However, the summary sentence and claim 5 say the framework offers actionable guidance, while the source hedges that it has some degree of actionable guidance. That drops the authors' qualification. Smallest repair: restore the words some degree of. The line saying the framework is not a forecast also omits the authors' statement that it is as much a prescription as a prediction.
The framework has some degree of actionable guidance for business leaders, workers, students, people concerned about AI safety or AI ethics, and policymakers
media — pass: No media was selected, and no verified candidates were supplied, so there is nothing to check for identity or relevance.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.