Here's some "empirical" proof of this.

Most people aren't aware that computers became better than humans at backgammon almost twenty years before the dominated chess and go.

And the reason is simple - AI wasn't needed, just probability theory. Researchers were able to directly program in probability models - and I'm talking model types invented in the seventeenth century that a bright high school student would understand perfectly - and defeat world class players.

It came down to the fact that evaluating a position in backgammon comes down to computing a few thousand small probabilities and summing them - something a computer can do without sweating, but something humans can't do very well at all.

Long-chain reasoning, on the other hand, is something we do so well that a single unaided human Go player is still strong competition for a computer costing hundreds of millions of dollars with decades of programming.