@TBFUnreality A very nice video! I'm personally convinced by arguments such as the Chinese Room thought experiment - there is no linear algebra large enough to constitute "understanding."
@ryxcommar@eatpraydiehard Two different architectures have access to two different submanifolds of C(R^n) (or whatever your favorite function space is). It's not obvious at all that their minima on a specific problem are close enough to be indistinguishable.
@nrui_tweet@cj_sheu If I read this, I would assume a non-native speaker. It's not a bad thing and the meaning is clear, but the phrasing isn't really natural.
@Swilua It's a very nice sounding argument, but gets almost everything wrong to the point of being complete nonsense (much like an llm!) Math is full of wrong but illuminative arguments as well as correct but useless ones. But you need a math education beyond high school to see that.
@Alan_Taylor_314@TonyTheLion2500 Not unique, but the set of functions that are Lebesgue integrable form a complete vector space. My professor in grad school described that as the great achievement of Lebesgue theory