A deep dive into AI as a normal technology vs. a humanlike intelligence and how major public policy based on controlling superintelligence may make things worse
An alternative to the vision of AI as a potential superintelligence — We articulate a vision of artificial intelligence (AI) as normal technology. Bluesky: @taumuyi , @knightcolumbia.org , @tedunderwood.me , @manfrommolise , @jwherrman , @andytseng , @badlogic , @emollick , and @moskov.goodventures.org . Mastodon: @nhamiel@infosec.exchange X: @techczech , @random_walker , @random_walker , @kevinroose , and @sayashk LinkedIn: Hunter Dunbar and Eitan Loewenstein . Forums: Hacker News Bluesky: Tau-Mu Yi / @taumuyi : This is a really smart framing of #AI as normal #technology knightcolumbia.org/content/ai- a... “if AI turns out to be normal technology— the downsides of which will be likely to mirror those of previous technologies that are deployed in capitalistic societies, such as inequality.” [embedded post] @knightcolumbia.org : In a new essay from our “Artificial Intelligence and Democratic Freedoms” series, @randomwalker.bsky.social & @sayash.bsky.social make the case for thinking of #AI as normal technology, instead of superintelligence. Read here: knightcolumbia.org/content/ai- a... Ted Underwood / @tedunderwood.me : The past five years have taught us that “intelligence” is actually a bundle of separable capacities. Language can do amazing things without conscious subjectivity, long-term memory, planning, or intent. — That I think is why apocalyptic futures are being replaced by this relatively chill take: M. Alessandro / @manfrommolise : Truly thoughtful and essential analysis of the AI field from @randomwalker.bsky.social @sayash.bsky.social. States what many felt, but haven't articulated. Pairs well with Shazeda Ahmed's “epistemic culture of AI safety” and others' work on risk and anti-trust. — knightcolumbia.org/content/ai- a... John Herrman / @jwherrman : This is a really thorough and sober ai-as-automation paper- passes through some of the same territory knightcolumbia.org/content/ai- a... Andy Tseng / @andytseng : Interesting perspective in this “AI as Normal Technology” article. It's definitely valuable to consider #AI as a natural progression of technology, rather than a radical departure. It's good to have diverse views when we think about AI's future. Thanks for sharing @mariannhardey.com. [embedded post] Mario Zechner / @badlogic : Excellent read. The only thing I disagree with is that systemic barriers prohibit speedy deployment of ML and AI in high risk areas. — We have ample evidence, that governments and orgs fall over backwards to deploy ML/AI without risk analysis. — knightcolumbia.org/content/ai- a... Ethan Mollick / @emollick : There is a bunch I disagree with here, but it makes an argument for why AI adoption may be less rapid and less immediately world-changing (& dangerous) than people expect. — Well worth reading, as it is a useful base case where AI is just another technology. knightcolumbia.org/content/ai- a... Dustin Moskovitz / @moskov.goodventures.org : First, they say it's surprisingly minimal now. I read this as wow! a half point increase in labor productivity across all U.S. adults, *just* with chatbots and IDEs for devs. — So if you just scale that to more hours per above, a 2.5% gain, but there's more 2/8 — knightcolumbia.org/content/ai- a... … Mastodon: @nhamiel@infosec.exchange : We are bombarded daily with over-the-top predictions and what equates to AI performance art from leaders and wannabe influencers. This document is a grounded look at the future of AI, with a vision that is a lot more likely than ASI in 2027. https://knightcolumbia.org/... X: Dominik Lukes / @techczech : I think I mostly tend to frame AI (even LLMs) as “normal technology” but it's possible that it's not a special kind of “normal technology”. One that it is still subject to diffusion bottlenecks, but whose impact will be bigger than thought. [image] Arvind Narayanan / @random_walker : I was glad to discuss the “AI as Normal Technology” paper with Kevin Roose and Casey Newton. We had an in-depth and fun conversation where we tried to identify our points of agreement and disagreement. I very much appreciate it when journalists take the time to read researchers' Arvind Narayanan / @random_walker : 📢📢New paper: AI as Normal Technology. I've worked on this w/ @sayashk for the last 2 years. The full version will be our next book! Utopian & dystopian AGI visions are quite similar. A far more important axis of disagreement—will AI be superintelligence or normal technology? [image] Kevin Roose / @kevinroose : It's Hard Fork Friday! This week on the show: — Meta has its day in court — Princeton's @random_walker makes the case that AI is a “normal technology” — And a round of HatGPT — Plus, a big announcement about our LIVE SHOW! [image] Sayash Kapoor / @sayashk : How will AI impact the economy? Can we defend against misuse? What policies would mitigate the risks of AI? Thrilled to share that @random_walker and I are writing another book to tackle these questions! Today, we release a paper laying out our argument: AI as Normal Technology. [image] LinkedIn: Hunter Dunbar : For a while I was very into AI doom/boom, paperclips, super intelligence, etc. Recently, I am somewhat disillusioned because, despite all the hype … Eitan Loewenstein : Although I've been tempted to share everything I read on AI by Arvind Narayanan and Sayash Kapoor this essay is particularly interesting and worth passing along. … Forums: Hacker News : AI as Normal Technology
Context & Ripple Effects
The essay enters a debate already shaped by warnings against both reflexive dismissal of AI and overly humanlike interpretations of language models. Earlier coverage of the limits of LLM capabilities provides a relevant backdrop for its argument that AI should be assessed as a technology rather than presumed to be an emerging person-like intelligence.
It also shifts attention from hypothetical control problems toward distributional and institutional consequences, echoing coverage that AI deployment may concentrate wealth and weaken workers' bargaining power. Published in the Knight Institute's democratic-freedoms series, the intervention is principally about what policy questions receive priority.
First-order effects
- The essay gives policymakers, researchers, and civil-society groups a framework for evaluating AI through present-day deployment, power, and accountability rather than through superintelligence scenarios alone.
- Its central warning challenges policy agendas that allocate major attention or authority to controlling a presumed future superintelligence while current social and economic harms remain less directly addressed.
Second-order effects
- The framing increases pressure on AI-governance proposals to show how they address concrete use, institutional incentives, and affected workers or users—not only frontier-model catastrophe risks.
- Public debate may become less binary: concerns about AI's material harms can coexist with recognition that the technology is consequential, rather than being sidelined by either dismissal or speculative intelligence claims.
Third-order effects
- If this approach gains traction, AI governance could increasingly resemble oversight of a general-purpose industrial technology: focused on deployment conditions, market power, labor effects, and democratic accountability.
- The underlying fault line is likely to persist: institutions must decide how to balance uncertain high-consequence future risks against harms from systems already being integrated into economic and public life.
The trend: AI policy is moving toward a contest over whether governance should prioritize speculative frontier-risk control or the institutional consequences of AI's ordinary deployment.