My gut instinct is that there is a world of difference between modelling a chess game and modelling an athlete. A chess game is ridiculously tightly constrained - there's only, what 32 components, and each is constrained very tightly in range and options... and it sounds like AI has only managed to model that in the last couple of years and with ridiculous amounts of computing power? And then you take an athlete whose body, life and environment is subject to literally *thousands* of inputs - many of them not input into any model and hence inaccessible to the AI..... Add to that the fact that one chess board is identical to another... but humans react *differently* on an individual basis to even the same training stimulus or input. Honestly? AI sounds like hype in the training space for the foreseeable future.
I agree that it's a very complex problem, but AI in sports is branching out in so many interesting directions. We're probably going to see it in smaller, niche areas first before it fully transforms athlete training. For example, race strategies like EF's Platapus have been in use for a long time. I reckon more teams will get on board with similar tech, kind of like what AlphaZero does. But when it comes to really advanced, prescriptive AI, like what AlphaFold represents, that's a whole different ball game and probably still a long way off. It's definitely an area to watch, but we've got to map out the body's systems before we can map out individuals. So we've got some ground to cover first.
Thanks @jenHry-ng3pw. I don't work in the field but as a coach this stuff is coming for me so this is me learning in public to figure out what is going to happen next.
