NPCAgency: Procedural Generation of NPCs

The NPCAgency project aims to automatically create deep, believable and interactive agents that can be easily generated and imported into game projects. These agents can be used as non-player characters (NPC) whose traits, personalities, personal histories and action conform to a “universe” selected by the user at creation time. The idea is that using NPCAgency would significantly reduce the game designer’s authorial burden, at least for NPC creation. Dozens, even hundreds, of characters, can be created and imported into a game engine using this webbased tool. We demonstrate the concept by encoding the “Game of Thrones” (GOT) / “A Song of Ice and Fire” universe. Users can parameterize the generative process and export the characters in the form of extensions to be included in Inform 7 based interactive fiction projects. The authors conduct a 37person user study whereby subjects familiar with Inform 7 created 50 unique GOT NPCs for their Inform 7 games. The majority of the users indicate they could use the tool on their own projects. Our results demonstrate this technique could work on a larger scale, and in principle, with graphical game engines.