Let’s say you play a game with a friend. And the metric of “how good you perform in that game” is measured by a numeric score which increases when you get closer to “winning the game” than your friend and decreases when your friend gets closer to “winning the game” than you. In this setting, you can think of yourself as trying to maximize the score and your friend as trying to minimize it.

We can come up with an algorithm able to make good decisions and win such a game by modeling the above situation in the following way: we will have 2 entities (functions) that call each other; one tries to maximize the score, the other to minimize it. Basically, these 2 functions will mimic the two players.

This algorithm is also a good example of AI that’s not ML. Some people make this confusion and think that AI = ML; in reality, ML is a subset of AI. Some AI techniques don’t involve ML. The minimax algorithm is such an algorithm that makes computers behave intelligently but they are not learning anything. And despite that, it works quite well in many games.

#algorithms #programming #decision-making #game-theory #artificial-intelligence