Intelligent tools can ease the burden of game development. One approach to easing this burden is the use of co-creative, artificial agents, capable of helping a human developer by making suggestions or extending an initial design. However, agents capable of design have historically required a large amount of hand-authored design information-domain-specific rules, heuristic functions, or formal logic rules. Due to the time, it takes to author this knowledge, such approaches do not remove the development burden but shift it to the author of the agent. To solve this problem we present a demonstration of a level-authoring tool with a co-creative agent informed by knowledge it derived from interactions with human designers.
You can see a video of the system in use here

Image
Faculty
Mark O. Riedl
Students
Matthew Guzdial, Shao-Yu Chen, Jonathan Chen, Shukan Shah, Vishwa Shah, and Joshua Reno