Automated Dungeon Mastering

Often touted as a "grand challenge" of interactive storytelling, this involves creating an AI agent capable of running a tabletop role-playing game (i.e. serving as a "dungeon master"). Simple right?

On the surface of it, the AI agent can automate a lot of the necessary minutiae that happens during a tabletop session. For example, it can calculate perception checks, attack rolls, movement for characters, and so on. But an AI dungeon master (AI-DM) must also determine several things that are less obvious: it needs to perform dynamic difficulty adjustment, it needs to improvise new story situations based on the actions of characters, and it needs to treat gameplay as a form of conversation. Working in this area will involve combining technologies around natural language processing (e.g. Amazon's Lex) with technologies around story generation (e.g. automated planning), and understanding how to navigate tradeoffs between authorial control and player agency.


  • Jennifer Nelson
  • Matthew Christensen

Project Details

Name: Automated Dungeon Mastering

Project Stage: Growing

Launched: 2017

Categories: Interactive Narrative,Artificial Intelligence

Tags: Role-playing Games,Automated Storytelling


University of Utah's Undergraduate Research Opportunities Program

External Site: None.

Our most recent projects

Automated Dungeon Mastering

Link to Project Page

Decompositional Planning

Link to Project Page

Determinants of Player Choice

Link to Project Page

The PSYCHIC Narrative UI

Link to Project Page

Procedural Puzzle Rooms

Link to Project Page

Elsewhere on the Web


Merrill Engineering Building, #3153
50 Central Campus Drive
Salt Lake City, UT, 84112, USA