Playing Rock-Paper-Scissors with Artificial Intelligence

Though rock-paper-scissors is usually a game of pure chance in which moves are selected at random, a player can predict the opponent’s next move based on previous rounds of play. For my Artificial Intelligence class, I created an online version of the game that played over 13,000 rounds against human competitors and used this data to calculate transition probabilities in a 10th-order Markov chain. The virtual player was able to win 3.05% more of the rounds that did not end in a draw than it would have won using the naïve method of randomly selecting moves.

You can see the project presentation below, or read the full report.