You are correct. MCTS + UCB and other variants were state of the art leading up to AlphaGo. And even then, MCTS was also used in AlphaGo.
The main change in AlphaGo was using a deep learning network to encode a value network for fast rollouts and a policy network for move selection (rather than using the UCB rule). They later removed the value network and rollouts entirely, but even AlphaZero uses MCTS.
The main change in AlphaGo was using a deep learning network to encode a value network for fast rollouts and a policy network for move selection (rather than using the UCB rule). They later removed the value network and rollouts entirely, but even AlphaZero uses MCTS.