Doctor: You’re dying sir, but we have a drug that has cured everyone who’s taken it.
Terence Tao: Has some human figured out why?
Doctor: Not yet.
Terence Tao: LOL! Get out of here.
Terrible take by Terence Tao. I’d take not one but ten cocktails all at once if that gave me not a cure, but even a small chance of one.
I think I understand the point some mathematicians are making. What I don't understand is why they can't continue doing what they were doing before while AI is still solving problems.
Ilusionado con la nueva subida de la vivienda gracias a las medidas que aprobará el gobierno debido al desahucio de la señora de 87 años.
It's confusing to see the incredible capabilities AI demonstrates from time to time and, at the same time, the incredibly stupid takes it sometimes produces.
So, I’m going to be clear: if distillation worries Anthropic and OpenAI and is described as equivalent to copyright, then both companies have zero, literally zero moat. I would say these valuations are pretty hard, if not impossible, to maintain if there isn't a moat.
So the solution seems to be to leverage the application layer and build a moat there, but then what’s the incentive to develop new models? Another strategy is what we are seeing: banning Chinese models and creating a set of rules to avoid distillation.
The limit of AI in programming is not marked by how long it takes to generate code, but by how long it takes an organization to understand and trust it.
