Sweet Learning Computer

To play a simple game called Hexapawn, an ‘artificially intelligent’ computer is created made entirely from sweets. The game is like a mini version of chess; the rules are explained fully, and a playing board drawn. The ‘machine’ then ‘learns’ how to improve its playing of the game by trial and error and by ‘remembering’ which strategies worked well. Repeated testing of the algorithm ought then to lead to a pattern of winning by the sweet computer – this pattern can be studied and, as an extension task, a decision tree can be constructed.