The Importance of Probabilistic Strategies in Poker

When I think about the algorithms that we use in AI, the fact that you're forced to be probabilistic seems a big challenge. It seems like you can't just run standard reinforcement learning to solve poker. You need to do something else. So there's two ways to get around this problem. One is these regret minimization techniques or like self-flate techniques that are specifically designed to converge to a mixed randomized strategy which is unexplainable. The other way is playing rock paper scissors with some random arbitrary strategy and then looking at what would do optimally against that strategy.