Connect Four is a classic two person, zero-sum game in which players utilize their wits and gravity to connect four of their own pieces in a horizontal, vertical, or diagonal row while blocking their opponent’s attempt to do the same. We have constructed a simulation of this game which we have used as a base for the implementation and testing of varying Artificial Intelligence (AI) systems. Early strategies worked according to simple strategic methods, while more advanced heuristics employed a Min-Max Tree in tandem with methods to determine how advantageous a certain board would be. This Min-Max Tree goes beyond a simple strategy, as it allows for the computer to look many moves ahead, thus picking the move that optimizes its chances of winning. The collection of statistics for the various strategies has allowed for the analysis and improvement of the AI structures.
Baly, Kirk; Freeman, Andrew; Jarratt, Andrew; Kling, Kyle; and Prough, Owen, "Teaching Computers to Think: Analysis of Artificial Intelligence and Connect Four" (2012). Symposium on Undergraduate Research and Creative Expression (SOURCE). 127.