Do LLMs need to be good at math with the same approach?
To draw an a analogy, we've got our human brain specialized.
Why not implement a part of the AI brain that's not neural nets, but instead circuitry specialized to math?
Maybe a dumb question since I'm a layperson!