Authors: Magnus Johansson, Mirjam P. Eladhari and Harko Verhagen
This paper describes the Model Social Game Agent (MSGA) and presents a comparison to existing strategies to create NPC behavior in video games. We present the results of a survey and an interview-study ofprofessional game developers, designers and scholars with a focus onthe social behavior of NPCs in existing gamesand the potential advantages and disadvantages of the MSGA. The results indicate that there are concerns regarding choosing to implement games AI as systemic AI (such as the MSGA) over an implementation based on scripting and behavior trees.Another question raised is the actual benefits of systemic AI and in what sense it would change the player experience. One of the potential gains of introducing theMSGA in game design would be to open upopportunitiesfor new modes of play and also open up for unexplored design spaces.
Keywords: NPCs, Believable NPCs, Game Design, MSGA, Game AI, Believability, Intelligent Agents.