Silicon Valley built a machine that reconstructs human cognitive artifacts, and mistook the reconstruction for the birth of an alien mind.
Here’s what AI actually is, where its intelligence comes from, and why it has a ceiling:
https://t.co/w9MlorKMKl
@jaykreps@sarthakgh In the next few years we’re going to have fully utilized data centers, commoditized models, compressed profit margins, compressed valuations, and a big whiff on ROI. It’s the 90’s all over again
@ilmagicoLorenzo@AricFloyd@MIRIBerkeley You misunderstand me. I don’t think ASI is possible with our current techniques and architectures. These predictions of a coming superior intelligence are wrong
@RokoMijic@Altimor The Riemann Hypothesis is perfectly predictable under the cognitive channel thesis. You are navigating reality with a broken model. Time to update
@NinaDSchick I expect us to continue making progress in domains where problems can be decomposed, intermediate steps are objectively verifiable, and correctness is clear. Do not underestimate how powerful this is. But I strongly doubt we are on some sort of exponential curve to SI
@mattyglesias What exactly has been proven at this point? He committed adultery? What does that have to do with running a company? If the “spy” is in fact a spy I will change my mind. But so far I haven’t seen any proof of that
@schorndorfer@mattyglesias It’s very well understood in tech that nothing happens by accident or as a result of a committee. One person needs to own the problem and deliver results. And very few people, if any, are capable of leading that company effectively besides Jensen
@TaxEveryone@cremieuxrecueil@grok how much money has CA spent on the high speed rail, and how many miles have been built? Would you label this project "inefficient" or "efficient"
@luckeyfaraday@misraetel “Hey ChatGPT, start a business with at least 1M in annual profit”
I’ll bet you all the money in the world AI never solves that problem