Many users praised Arvind Narayanan's ICML keynote for its thoughtful framing of AI as normal technology that cuts through hype and offers important adaptation insights, while a few dismissed its novelty.
Based on 13 visible X reactions from 19 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@random_walker Very informative talk! Really appreciate the ambition to elegantly decompose what key terms actually mean (AGI, reliability, automation) -> to expose where unfounded causal leaps are made. Complements Messy Jobs' use of economic abstraction -> explain AIs impact on tasks.
@random_walker There is nothing new in AI except it is busting tribal knowledge...out in the open... Lawyers used to charge $500 to draft a complain... AI can do it for you. Super-intelligent AI will actually need super data from actually super intelligent people... and we can't make Gauss.
@random_walker @sayashk Thanks for sharing I’d call this one of the best perspectives on AI for anyone in AI or wanting to cut through the haystack of hype.
@random_walker A very thoughtful presentation. Thank you for sharing it. https://x.com/DrTechlash/status/2076886833636806835/photo/1
His ICML keynote slides reject sudden, single-milestone job automation.
@random_walker Very informative talk! Really appreciate the ambition to elegantly decompose what key terms actually mean (AGI, reliability, automation) -> to expose where unfounded causal leaps are made. Complements Messy Jobs' use of economic abstraction -> explain AIs impact on tasks.
@random_walker There is nothing new in AI except it is busting tribal knowledge...out in the open... Lawyers used to charge $500 to draft a complain... AI can do it for you. Super-intelligent AI will actually need super data from actually super intelligent people... and we can't make Gauss.
@random_walker @sayashk Thanks for sharing I’d call this one of the best perspectives on AI for anyone in AI or wanting to cut through the haystack of hype.
@random_walker A very thoughtful presentation. Thank you for sharing it. https://x.com/DrTechlash/status/2076886833636806835/photo/1
@random_walker I like the talk, decided to follow you after it
@IEthics call me a luddite but if i smell AI output i turn it off
I had the honor of giving a keynote at the International Conference on Machine Learning in Seoul last week titled “What will be left for us to work on?” I addressed the widespread anxiety about how we should adapt as AI capabilities increase. I was thrilled by the talk’s reception, so I have made my slides available, annotated with a lightly edited transcript: https://www.cs.princeton.edu/~arvindn/talks/icml-2026-annotated-slides/ I made three arguments. First, the "AI as Normal Technology" framework is a correct and useful as a way to think about AI’s impacts, unless and until there is some future discontinuity such as through recursive self-improvement. Second, even though we should take recursive self-improvement seriously, there is no milestone that companies might achieve in the lab that will suddenly put us all out of work. Third and finally, jobs of the future will be radically different, and a lot of adaptation will be needed. I shared my thinking about what this might look like and ended with a vision of human/AI “co-superintelligence”.
I also recently started talking about how our field needs a new north star as we reach AGI. I also concluded that humans are the last bottleneck and we will need to find ways to better interact with and steer the legions of agents we will all soon command. I also started to wonder how we as humans will navigate our own competing objectives with AI. Will we use it to negotiate on our behalf? Can we trust it to do what is in our best interests? Does it find better compromises? It might be time for AI researchers to learn about how to study people!
Thank you for all the kind words! In case the webpage doesn't render correctly on your mobile device, there's a static version here: https://www.normaltech.ai/p/what-will-be-left-for-us-to-work
Many users praised Arvind Narayanan's ICML keynote for its thoughtful framing of AI as normal technology that cuts through hype and offers important adaptation insights, while a few dismissed its novelty.
Based on 13 visible X reactions from 19 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@IEthics call me a luddite but if i smell AI output i turn it off
I had the honor of giving a keynote at the International Conference on Machine Learning in Seoul last week titled “What will be left for us to work on?” I addressed the widespread anxiety about how we should adapt as AI capabilities increase. I was thrilled by the talk’s reception, so I have made my slides available, annotated with a lightly edited transcript: https://www.cs.princeton.edu/~arvindn/talks/icml-2026-annotated-slides/ I made three arguments. First, the "AI as Normal Technology" framework is a correct and useful as a way to think about AI’s impacts, unless and until there is some future discontinuity such as through recursive self-improvement. Second, even though we should take recursive self-improvement seriously, there is no milestone that companies might achieve in the lab that will suddenly put us all out of work. Third and finally, jobs of the future will be radically different, and a lot of adaptation will be needed. I shared my thinking about what this might look like and ended with a vision of human/AI “co-superintelligence”.
I also recently started talking about how our field needs a new north star as we reach AGI. I also concluded that humans are the last bottleneck and we will need to find ways to better interact with and steer the legions of agents we will all soon command. I also started to wonder how we as humans will navigate our own competing objectives with AI. Will we use it to negotiate on our behalf? Can we trust it to do what is in our best interests? Does it find better compromises? It might be time for AI researchers to learn about how to study people!