>>11553967To be honest I don't think so.
That's because "AI research" a lot of time comes down to trying to being hardware and software together and talk with vendors etc., in practice.
A lot of algorithms and networks and whatnot are there, it's just that it's no feasible to run it on slow energy consuming hardware.
Of course, if you find a billionaire hobbyist who gives you a super-computer and says "hey, build me true general intelligence," then you might end up doing stuff that's not just cog in the machine stuff.
But in practice, I'd say, "machine learning" atm is plaguing oneself tuning a system so that, when hooked to a AAA battery, would not drain it in 5 minutes. Or, on a desktop, is fast enough.
Can be fune if you like to tune shit for a living, though, but it's not "math."
Not like obscure Banach spaces to solve niche differential equations kind of Terry Tao math.
