Investigating and automating the creative act of software engineering

We take the position that the creative act of computer programming has been under-investigated in Computational Creativity research. It is time for a concerted study of software engineering from the perspective of creative software. Such software should produce code and algorithms as artefacts of interest in their own right, rather than as a means to an end. We briefly survey and critique existing automated programming approaches, propose some novel methods for this, and investigate potential application areas for automated software engineering. Central to our position is the notion that creative software generators should perform in an unsupervised manner in order to problematise the world rather than (or in addition to) solving given problems. This will necessarily utilise some current methodologies and philosophies from Computational Creativity research, and we explore the ways in which these could guide future software synthesis approaches.