Some context is in https://www.uzpg.me/general/2023/12/30/review
I will not bet in this market.
I have been studying some Math/CS/Phys at MIT and am also in the AI community doing research, went to NeurIPS recently. I guess I feel like a lot of the things I learn in my courses are not really essential for a lot of the ML work, and some people seem to be focusing more already. I'm doing it because it's pretty but the world is moving quickly, maybe too quickly 👀
The more math/physics/chemistry/etc. I learn, the more connections I see to other fields, including AI. For example:
Entropy is a physics and a statistics thing (used in training LLMs).
Energy and the Boltzmann equation are relevant for chemistry and [RL](https://www.sciencedirect.com/science/article/pii/S0925231208002130).
Singular learning theory is a thing, though I haven't taken the time to learn it.