Excuse me, I need better AI!: employing collaborative diffusion to make game AI child's play

The idea of end-user game authoring environments is quickly gaining momentum in education. Environments such as AgentSheets have been used by thousands of children to learn about programming and design by creating their own computer games. With only hours of training these children initially create their own versions of classical games such as Frogger, Sokoban, and Space Invaders and later begin to design and implement their own game ideas. After creating numerous simple games including cursor controlled or randomly moving characters children are asking for more sophisticated AI. They would like to be able to build characters that can track each other through complex mazes and even to collaborate with each other. We have developed a simple to use multi-agent collaboration and competition framework called Collaborative Diffusion. At first we used collaborative diffusion for our own computational science applications and later also for educational applications at the university level. We found that by adding proper AI debugging tools we were able to convey the idea of collaborative diffusion even to middle school children. In this article we introduce the collaborative diffusion framework and present applications including a collaborative soccer game simulation.