Loved this--it inspired an AI use case for me today. My team sent me a bunch of differently formatted text docs with time spend in meetings. Instead of doing what I would normally do, normalizing in Excel, I just dump the mess into Claude and started asking it questions. It worked beautifully.
Another thought which occurs to me - inspired by this essay and also the idea that "big data is when it's cheaper to store it than to figure out what to do with it" - a bit like Quantum Mechanics, there's something fundamentally inhuman about the way AI works. You basically can't mental model it with intuitive understanding or folk physics - you need to learn a new set of conceptual tools.
Hayek is not a very popular thinking these days. However, I think it is Hayek who had a theory about law [I mean literally our legal systems] and politics that a very serious problem facing long-surviving nation states is: that we pass layer upon layer of laws and make things worse and worse. Because when we pass a law that doesn't do what we wished it to do, we don't repeal it and try again, we add layers of 'fixes' and 'complexifications' onto it. Do we want million page contracts in the future that dueling law AI's argue over? Or do we want 5 page human contracts.
Since grit and 10K hours have made their re-appearance from 2011-2014, maybe shu-ha-ri or wu-wei could enter into this model too? Proliferation of mastery in GPT-n+1? Mastering detailed domains beyond our patience and working memory limits? More Move 37s to come
Loved this--it inspired an AI use case for me today. My team sent me a bunch of differently formatted text docs with time spend in meetings. Instead of doing what I would normally do, normalizing in Excel, I just dump the mess into Claude and started asking it questions. It worked beautifully.
Another thought which occurs to me - inspired by this essay and also the idea that "big data is when it's cheaper to store it than to figure out what to do with it" - a bit like Quantum Mechanics, there's something fundamentally inhuman about the way AI works. You basically can't mental model it with intuitive understanding or folk physics - you need to learn a new set of conceptual tools.
This is great. Oozification is really a restatement of "simple algorithms and more data >> complex algorithms and less data".
Hayek is not a very popular thinking these days. However, I think it is Hayek who had a theory about law [I mean literally our legal systems] and politics that a very serious problem facing long-surviving nation states is: that we pass layer upon layer of laws and make things worse and worse. Because when we pass a law that doesn't do what we wished it to do, we don't repeal it and try again, we add layers of 'fixes' and 'complexifications' onto it. Do we want million page contracts in the future that dueling law AI's argue over? Or do we want 5 page human contracts.
I think this is the first time I’ve been thrilled by what AI can do. Speed of patience, indeed!
Since grit and 10K hours have made their re-appearance from 2011-2014, maybe shu-ha-ri or wu-wei could enter into this model too? Proliferation of mastery in GPT-n+1? Mastering detailed domains beyond our patience and working memory limits? More Move 37s to come