For me, I've always believed and still believe that Ai can only be used to improve self, technologies, games and what have you, rather than destroy them, so I agree with you at most, newer generations have the ability of mastering most games that seems difficult to master in time past, like many today still do not know how to play poker, with the help of Ai, anybody can master the game easily, this is the benefit i see with the use of Ai in our games.
But then, setting up Ai to play the game in actuality, I don't think that is feasible, and even if casinos permit it, and they are able to win, I don't think their winning will be consistent in the long term due to the dynamics of the game.
Of course, it is possible to use AI as a teacher in order to learn how to play poker, but in my opinion it is not at all interesting. In addition, simple training programs for training have long been invented and implemented, which in no way fit the definition of AI. It is much more interesting to play with partners just with your friends or acquaintances. It's more interesting and more fun.
As for the actions of AI as a human assistant, the question arises until when AI will provide information that really improves the quality of human life and what will it do in case of conflicts of interests of different people? And here, when solving such problems, destructive tendencies in the actions of AI may well begin to appear. In approximately the same way, this situation can be reproduced in the question of AI participation in gambling, including when playing poker. As a result, we will get in the form of an AI player just some average poker player, and this becomes generally completely uninteresting and even unnecessary.
So AI in poker, I think it's a dead end application of AI itself.