1) Programs like Deep Blue and Stockfish can now defeat the best human chess players by using algorithms like minimax search and alpha-beta pruning to evaluate game trees up to dozens of ply deep.
2) Go programs initially struggled compared to chess programs due to the enormous game tree of Go. But programs like Crazy Stone and more recently AlphaGo used Monte Carlo tree search combined with deep neural networks to achieve superhuman performance at Go.
3) While algorithms like minimax and Monte Carlo tree search have achieved superhuman performance at perfect-information games like chess and Go, developing strong algorithms for imperfect-information games like poker remains an open research problem. Determinization is one approach but has limitations.