Christian Guckelsberger

Christian Guckelsberger

Intrinsic Motivation in Computational Creativity with Applications to Games

This research aims at designing artificial systems which are perceived as genuinely creative, by means of an agent-centric approach and intrinsic motivation. Video games are considered a “killer application” for computational creativity because of their many interacting facets, and are used as an exciting application area. Intrinsically motivated agents perform an activity for its inherent satisfaction rather than for some externally specified goal. The main focus of this project is currently on investigating how intrinsic motivation can be employed to trigger new and surprising, yet sensible behaviour in companion and enemy NPCs. Compared to established methods such as traditional reinforcement learning, there is no need to specify the game mechanics or goals at design-time. This project consequently focusses on the challenges which new technologies such as procedural content generation impose on NPCs, with the view of creating a generic AI which shows companion- or enemy-like behaviour in an arbitrary, previously unknown game. The preliminary results are promising, highlighting the technology’s potential to decrease efforts in manually scripting NPCs, to extend the spectrum of game variants accessible to automatic game evolution, and to make NPC behaviour more surprising and thus interesting to the player.

Christian studied Computer Science, History of Art and Business in Germany and the UK, and is now based in London to work towards a PhD in Computational Creativity, a subfield of Artificial Intelligence. He challenges the question how computers could ever become genuinely creative, by drawing on a highly interdisciplinary approach based on Philosophy, Cognitive Science and Computing. Having worked in the R&D department of SAP SE for three years, Christian has substantial industrial experience. He enjoys working in an international environment with open-minded, passionate people. His main game-related interests are in procedural content generation and general game AI.

Home institution: Goldsmiths

Supervisors: Professor Simon Colton, Dr Jeremy Gow

Ready to apply?

Once you have identified your potential supervisor, we would encourage you to contact them to discuss your research proposal.

Learn More