We've seen with ML and Chess (an go to some extent) that brute forcing the problem space is clever than using heuristics.

I think this is only true because, so far, our heuristics are not that clever.