Artificial intelligence for games: Searching for the winning strategy

With the games industry surpassing Hollywood in terms of monetary value, it’s clear that games are big business. Dr Edward Powley, along with researchers at the University of Bradford, Imperial College London, and the University of Essex, is tackling the problem of unknown factors in games.

A game of imperfect information is the term for games that have hidden information; for example, when playing poker, you can’t predict the outcome of decisions, as you can’t see your opponent’s cards. (The opposite of this are games of perfect information, such as chess. The game is laid out in front of you; assuming that you can accurately guess your opponent’s moves, you can accurately guess the outcome of every possible move on this turn.)

Artificial intelligence has been used to create computers capable of estimating, with a high degree of accuracy, the outcome of every decision in a given turn, resulting in computers that are World Champion players of a given game. However, Dr Powley’s research focuses on creating artificial intelligence that is a superior player of any game of imperfect information. The research primarily applies to board and card games, with applications for computer games.

The impacts of a computer that can accurately estimate the outcome of moves in a game of imperfect information are wide-reaching. Using the same processes, it could also be used in economics, where several companies are in competition. Another example of applications of this research is in security – a computer could predict the path of a criminal, helping police to catch the culprit.

Dr Powley said:

“As a best-case scenario, this kind of artificial intelligence could be available within the next few years.”

Dr Powley’s event as part of the British Science Festival will be a drop-in session, open to all ages. Participants will be able to play against a computer, whilst being able to see how the computer makes decisions with each move.


Image courtesy of