A bit ago, I found out that researchers had taught a machine to play Texas Hold'Em at a level that beat most champions. However, that AI had access to the information of what cards it was dealt.
So I got to wondering. If a poker-playing AI were made, where the AI did not know what its own cards were (that is, it never 'looked' at them), would it be possible for the AI to reach a similar level of success?
If the AI does not know its own cards, it only has less knowledge and is therefore at a disadvantage. The AI could only act on the basis of the other players' play, while they know their own cards and can use the information.
For example, if you look at a game with 2 players before any action was taken, from the AI's point of view the chance of winning would be 50% while the other player can calculate a much more precise probability based on his own cards and build on that.
A super-human level cannot be reached with this.
If the strategy is cleverly adapted, a beginner level can perhaps be reached.
In brief, yes. It has just to pick up a random private hand and play according to the opponent's decision. To sum up, it would play optimally giving this random hand and his opponent's decisions.
