> probably the first real-world deployment of a self-improving system that beat humans.

Isn't that Checkers right from the dawn of reinforcement learning?