Seriously though, there are definitely elements of both involved in AI, but I don't know that there's much going on in AI that involves math beyond what you'd get in a typical C.S. curriculum anyway. And if you're worried about that, you could always go with a "major in C.S. and take as many maths electives as possible" strategy.
As noted in the linked FB thing, the key things you want are: "Calc I, Calc II, Calc III, Linear Algebra, (and) Probability and Statistics".
There's nothing to stop you from getting all of those while majoring in C.S. And here's the thing - the "other" stuff you get in C.S. will empower you in terms of programming and - for now - programming is how you do "experiments" in AI/ML.
As for majoring in Mathematics... I don't think it can ever hurt to have too much math, and I wish I'd taken more maths classes in my own past. But I haven't seen much evidence to suggest that you'll use a lot of, say, complex analysis, or measure theory, or group theory in studying AI/ML. Note that that's not to say that I'm 100% (or even 2%) confident that you'd never need those things. Given that measure theory is strongly related to probability, for example, it wouldn't shock me if some knowledge in that area did indeed prove useful somehow. But by and large, it seems to be the case that the core math needed really is Calculus, Linear Algebra and Probability/Statistics.