2,003 karma · joined September 6, 2023
Though tbf corporate-speak and AI-slop are both insufferable in similar ways...
Constructive, reasoned discussions on migration are precisely what have not occurred due to the dominant ideology stifling all contrary opinion.
You're basically pointing at a smoker who started a year ago and concluding because they don't have cancer yet cigarettes don't cause cancer.
War a nation can recover from, but from an erosion of cultural unity and shared identity it cannot.
The US remains the best option among a mixed field of weak, corrupt, divided and authoritarian alternatives.
The real problem is not the debt, but the social instability caused by the measures taken to address the debt.
Not sure I agree with this. AI generated proofs can still be analyzed and mined for useful insights. I suppose he's saying the process of banging our heads against the wall on a problem can itself yield useful insight? But what is stopping us from analyzing a proof after the fact. And if we can generate many different versions of a proof that should help us develop a much deeper understanding of the problem than we would have without being able to perceive the "proof landscape"...
1. Conversational AI - this is ubiquitous now, but not sure it's increasing at any significant rate. At least my personal usage has leveled off. Also probably not a significant driver of per token demand relative to code gen. More a driver of subscription revenue.
2. Code Generation - this is the big killer app, which I suspect has driven most if not all of the last year's revenue growth for Anthropic and to a lesser extent OpenAI. Now, I doubt we can extrapolate that growth given 1) how quickly and completely CC has been adopted by the industry and developers and 2) how corps are reigning in spend
3. Image and Video Generation - this honestly strikes me as more of a trifle/novelty than anything else. Granted I'm not a visual artist but I don't see it being a killer app anywhere to the extent of code generation.
4. Physical AI - this is the wildcard for me, but also beyond my ability to conceptualize its effects on demand. I suspect a proliferation of self driving and autonomous robotics would be a huge driver of compute demand but who knows how close we are to proliferation?
No matter how intelligent the executor is, without knowledge of the relevant external context, specs will always have to be clearly defined and given.