Game Theory Optimal GTO – AI...Are they really useful?
There is a branch of mathematics called Game Theory. Mathematicians are not good at English. It should have been called Game Theory Non Exploitative GTNE.
If there were 2 computers, each with Artificial Intelligence, and they played against each other, constantly adjusting to each others strategy, they would eventually reach equilibrium. Both would have a perfect strategy that the other computer could not exploit. But, neither would win or lose.
You would never intentionally play against the AI computer for money. In the real world you play against humans that do not have a perfect game and that mistake. It is a simple fact that the AI strategy is far from perfect when it comes to Win Rate. AI assumes that their opponent does not make mistakes.
That does not mean that AI or GTO is bad. But it is not Optimal, just not exploitable. Understanding that fact, how can we use it in a real game against imperfect opponents? It would be very foolish to ignore such a valuable tool. We need to understand how to use it.
We could look at the AI database, only a few terabytes. I don't think that is possible for me, but then I
am getting old.
We can ask the AI computer specific questions and it will tell us the answer. There are millions of possible questions so we have to be selective. There are software products that allow you to do just that.
Please don't get me wrong, both AI and GTO are very valuable tools. We just need to understand what and how they can be used. One good example is Minimum Defense Frequency MDF.
P.S. PeakHoldem has GTO and MDF included.
Every future release of PeakHoldem will increase the accuracy of opponent play.
Unique enough to be patent pending. 62/732,519.
Please share!