michaelothomas
Joined 2 years ago
Karma
145
Submissions
9
Member since
Jun 27
Joined 2 years ago
JavaScript plays fast and loose with types. Which means that when JIT compilers generate code for it, they have to make all sorts of conservative decisions. This makes generated code more complex and, in turn, slower.
Does that make sense?
We had a pretty awesome team of people working across the entire stack optimizing absolutely everything. The blog post only talks about the three optimizations that made the biggest difference.
Two years ago we were told AGI, massive job losses, and trillions in profits were around the corner; justifying enormous investments in Anthropic and OpenAI. Today I'm supposed to be impressed that they solved an obscure math problem by kinda/maybe copying someone else's work?
These models are not living up to the gargantuan hype fed to investors. So OpenAI and Anthropic are stuck: How do they have successful IPOs to generate exit capital for their investors when their IPOs are taking place right when frontier advancements are kinda petering out? You find a way, any way, to keep the hype train going and hope for the best.
Judging from the front pages of the world's newspapers, people are lapping this "AI will kill us all" nonsense up ... so it sure seems like it's going to work. Kudos.