Interactive entertainment is becoming an area actively studied
by researchers, artists, and product developers. An ideal system
would be highly interactive and yet have the emotional impact
of such traditional media as the novel, film, and television.
We believe this requires the computing system to make decisions
during interaction that are normally made off-line by artists
working in non-interactive media. Thus, artistic knowledge and
capabilities must migrate into the machine to achieve the ideal
of rich interactivity with the power of traditional media.
The Oz Project, an effort in CMU's School of Computer Science
assisted by the Drama Department, English Department, and others,
is exploring one variation of this theme. We want to provide
users with the experience of living in dramatic simulated worlds
that include engaging, believable characters. This talk summarizes
the project, emphasizing our efforts to build broadly capable,
but shallow, agents and to create a computational theory of interactive
drama. Along with a theory of presentation style, we see these
as the central topics in this research area.
One product of our work, an interactive world containing several
emotional, animated creatures was shown at the AAAI-92 AI-based
Arts Exhibition. I will show a video of this system (the "Woggles"),
and it might be available for interactive use after the talk.