Had there been any mention of video generation via Sora, Pulse, etc. and how it’s linked to improving AI robotics? I listened to a talk where this was mentioned briefly but I can’t seem to find anything written about it
not a lot! the video generation stuff is more on the “what are the things we need to concerned about” side - which happened in lunches and dinner conversations
Jaggedness has been a feature of the IT revolution since the start and contrast sharply with the relatively uniform technological improvement of the industrial economy until the late 20th century. The big-picture illustration is the contrast between near-stasis in transport (40 year old cars, ships and planes, if they've been looked after, are barely distinguishable from those made today) and the giga-fold improvements of Moore's Law.
At a smaller level, it's fractal. Someone finds a way to replace analog with digital in some application (watches were one of the first examples). Even if the digital version is initially clunky and expensive, Moore's Law (taken as doubling every 1.5 years) guarantees that it will be 1000 times more powerful in 15 years time. But some things resist digitization or have exponentially scaling costs, so the analog version stays competitiive until some workaround is found.
In this context, the emergence of the term "hyperscalar" suggests to me that AI, at least as represented by LLMs is in the latter category.
Great writeup. The lack of genuine interest in China among AI insiders totally jives with my experience too having just spent a couple weeks in the bay area including an evening at Lighthaven. My running theory is that China fundamentally undermines the crucial crux that animates so many people there: that of inevitability. They don't understand why China doesn't see AI the same way they do, and they're not that interested in finding out.
Thank you, Vincent! Yes, there’s definitely a rationalistic calculation when it comes to the “urgency and priority” of how to rank the China question. Compared to the AGI timeline, China is not ranked high enough.
Ted Chiang and Gordon-Levitt in the same room, that alone justifies the conference! The juxtaposition of three writers' views on AI is fascinating. Perhaps these seemingly incompatible perspectives point to a bifurcated future: some may use AI as a true 'bicycle for the mind' to amplify creativity and personal growth, while the rest slides into passive consumption of AI-generated dopamine hits? The optimist in me hopes Ben Thompson is right: by lowering the cost of creation, AI may meaningfully expand the 1% creators and bends the internet’s 90/9/1 curve toward creators?
I had no idea who Joseph Gordon-Levitt was and this is classic The Curve and I love it.
i love the fact that you love the fact that you don't know who Joseph Gordon-Levitt was! classic nerdy AI researcher.
have you saw Inception??
yep i googled it!
Had there been any mention of video generation via Sora, Pulse, etc. and how it’s linked to improving AI robotics? I listened to a talk where this was mentioned briefly but I can’t seem to find anything written about it
not a lot! the video generation stuff is more on the “what are the things we need to concerned about” side - which happened in lunches and dinner conversations
Jaggedness has been a feature of the IT revolution since the start and contrast sharply with the relatively uniform technological improvement of the industrial economy until the late 20th century. The big-picture illustration is the contrast between near-stasis in transport (40 year old cars, ships and planes, if they've been looked after, are barely distinguishable from those made today) and the giga-fold improvements of Moore's Law.
At a smaller level, it's fractal. Someone finds a way to replace analog with digital in some application (watches were one of the first examples). Even if the digital version is initially clunky and expensive, Moore's Law (taken as doubling every 1.5 years) guarantees that it will be 1000 times more powerful in 15 years time. But some things resist digitization or have exponentially scaling costs, so the analog version stays competitiive until some workaround is found.
In this context, the emergence of the term "hyperscalar" suggests to me that AI, at least as represented by LLMs is in the latter category.
hi John! that’s great to know. i thought it’s a new term!
The term is new, I think, but the phenomenon is not
Fr fr deedmd dc dddd
Great writeup. The lack of genuine interest in China among AI insiders totally jives with my experience too having just spent a couple weeks in the bay area including an evening at Lighthaven. My running theory is that China fundamentally undermines the crucial crux that animates so many people there: that of inevitability. They don't understand why China doesn't see AI the same way they do, and they're not that interested in finding out.
Thank you, Vincent! Yes, there’s definitely a rationalistic calculation when it comes to the “urgency and priority” of how to rank the China question. Compared to the AGI timeline, China is not ranked high enough.
Ted Chiang and Gordon-Levitt in the same room, that alone justifies the conference! The juxtaposition of three writers' views on AI is fascinating. Perhaps these seemingly incompatible perspectives point to a bifurcated future: some may use AI as a true 'bicycle for the mind' to amplify creativity and personal growth, while the rest slides into passive consumption of AI-generated dopamine hits? The optimist in me hopes Ben Thompson is right: by lowering the cost of creation, AI may meaningfully expand the 1% creators and bends the internet’s 90/9/1 curve toward creators?