Creative Code Generation

Supervisor: Prof Simon Colton

Research group(s): Centre for Digital Music, Game AI

Machine Learning, Genetic Programming and Automated Program Synthesis are all areas in which AI systems can create new software directly. However, to some extent, each of these approaches are either driven directly by a user, or attempt to find something the user knew he/she was looking for. What if automated code generation software could find something you didn't know you were looking for? Using techniques from Computational Creativity, we are developing unsupervised approaches to creative code generation, with applications ranging from inventing mathematical concepts to designing videogame mechanics. Projects in this area would address technical issues related to generating code, and user-interface issues related to convincing a user that generated code is of real value to them, even though they don't realise it yet!