ongoing

completed

Industrial Graveyard

Traditionally, AI focuses on activity in the world as problem-solving rationality. The goal for autonomous agents is often to behave optimally rationally in approaching some goal. For interactive computer characters, this focus is problematic, since characters do not need to be particularly smart or rational, instead needing to project emotion and personality in a way that is understandable to users. In the Industrial Graveyard, I explored how to create agents, not as rational problem-solvers, but as experienced by human users. Users observe the antics of a discarded lamp in a junkyard, while controlling the behavior of its unsympathetic overseer. The technology is based on narrative psychology, which argues that humans interpret activity by organizing it into narrative. I support human interpretation of character action by providing visible cues for narrative interpretation of agent behavior, most notably through transitions between behaviors that connect them by expressing the reason for the behavior change to the user.