Intelligent modeling of the user in interactive entertainment. Paper presented at the AAAI Stanford Spring Symposium

A theme of the symposium is to explore ways to employ AI to make games more appealing to people who do not enjoy current genres, and to expand the market for interactive entertainment beyond the traditional niche of young male players. We suggest that AI techniques employed in the world of intelligent tutoring to model the user and adjust instruction, help and content could be fruitfully adapted to interactive entertainment. In computer-based educational tutoring, adaptivity to user behaviors and characteristics such as gender and cognitive developmental level have been shown to increase learner motivation, engagement and achievement in the area of mathematics learning. Similarly, utilizing data regarding player behaviors such as latency and errors to construct a model of the player would allow for more adaptive game play, which in turn would increase the appeal of computer games to a wider audience. Intelligent modeling of the user in interactive entertainment Increasing the appeal of interactive entertainment From one perspective, the world of computer games is a major economic success story, with sales now beyond those of more traditional media such as films, and with more power to attract users ' leisure time than television or books (Jennings, 2001). However, from another perspective, the picture is not so bright: although large, the interactive entertainment market is not expanding in any significant way beyond the original target market, i.e., adolescent and young adult males, by attracting new fans from other demographic niches. In particular, girls and young women are still only one third as likely to be consumers of interactive entertainment as their male peers (Jennings, 2001). Popular explanations for this gender gap typically focus on girls ' supposed dislike of fantasy, magic and action genres, or on their rejection of sexist portrayals of female characters (cf., Gailey, 1992; Rabasca, 2000). Yet these explanations are not supported by research i