The role of a Dungeon Master, or DM, in the game Dungeons & Dragons is to perform multiple tasks simultaneously. The DM must digest information about the game setting and monsters, synthesize scenes to present to other players, and respond to the players' interactions with the scene. Doing all of these tasks while maintaining consistency within the narrative and story world is no small feat of human cognition, making the task tiring and unapproachable to new players. Large language models (LLMs) like GPT-3 and ChatGPT have shown remarkable abilities to generate coherent natural language text. In this paper, we conduct a formative evaluation with DMs to establish the use cases of LLMs in D&D and tabletop gaming generally. We introduce CALYPSO, a system of LLM-powered interfaces that support DMs with information and inspiration specific to their own scenario. CALYPSO distills game context into bite-sized prose and helps brainstorm ideas without distracting the DM from the game. When given access to CALYPSO, DMs reported that it generated high-fidelity text suitable for direct presentation to players, and low-fidelity ideas that the DM could develop further while maintaining their creative agency. We see CALYPSO as exemplifying a paradigm of AI-augmented tools that provide synchronous creative assistance within established game worlds, and tabletop gaming more broadly.

Our overeager party--an elvish druid; a dwarven wizard; a halfling rogue; and a human paladin--has arrived at a dusty, cluttered library. Hearing of our quest for the fabled Orb of Zarekath, the head librarian--Thimblewick, a gnome--recounts how it was once "a powerful artifact" that has long since disappeared in the nearby ruined city. But the rogue is less curious about Orb-lore and more interested in snooping and stealing from the nearby shelves. Sneaking into the shadows, he's caught by a librarian. "Oh, sorry," the rogue says with a disarming smile.

We propose a novel task, G4C, to study teacher-student natural language interactions in a goal-driven and grounded environment. Dungeons and Dragons (D&D), a role-playing game, provides an ideal setting to investigate such interactions. Here, the Dungeon Master (DM), i.e., the teacher, guides the actions of several players -- students, each with their own personas and abilities -- to achieve shared goals grounded in a fantasy world. Our approach is to decompose and model these interactions into (1) the DM's intent to guide players toward a given goal; (2) the DM's guidance utterance to the players expressing this intent; and (3) a theory-of-mind (ToM) model that anticipates the players' reaction to the guidance one turn into the future. We develop a novel reinforcement learning (RL) method for training a DM that generates guidance for players by rewarding utterances where the intent matches the ToM-anticipated player actions. Human and automated evaluations show that a DM trained to explicitly model intents and incorporate ToM of the players using RL generates better-quality guidance that is 3x more likely to fulfill the DM's intent than a vanilla natural language generation (NLG) approach.

Deep in some underground ruins below a drought-stricken village, four brave adventurers found themselves in grave danger. The dungeon belonged to the Order of the Pure, a potentially nefarious cult that may have had something to do with the droughts that have been wreaking havoc in the village of Havenshire. Despite the danger, the stakes were too high to turn back. After all, not only did the fate of the villagers depend on their success, but the town's mayor also promised a small fortune if they succeeded. Eventually, the heroes discovered a wooden chest hidden under a slab of rock in one of the inner chambers of the dungeon.

Four bedraggled adventurers stand together on the shore of a desolate island, shivering in the evening mist. They don't know each other, and their motives for being here are unclear. But as they make stilted conversation they see, emerging from the briny waters, figures dressed in the rags of sailor outfits, moaning and shuffling and horrible. The adventurers stand around, roll some dice and chat some more, as the undead seamen lurch ever closer. Looking on at this desperate scene, I think to myself, "What the hell? Why can't anyone make a decision? We've been here for half an hour! We've not even begun the proper adventure yet!" Dungeons and Dragons has always been there in the background of my life.

Dr Lara Martin wants to teach artificial intelligence how to tell a tale and tell it well. Lara is a Computing Innovation Fellow postdoctoral researcher at the University of Pennsylvania, where she teaches AI to generate stories and produce language that is natural and human-like. She reveals why we need to train machines how to be storytellers and what Dungeons & Dragons has to do with it all. People have been telling stories since before we could write; we're natural storytellers. So if machines were able to tell and understand stories as well, we'd be able to communicate with them more naturally.

I still remember walking past the tabletop game store in the mall when I was a kid. I used to think, "that looks really interesting, but everyone would think I'm a nerd if I started playing it." Admittedly, I am most definitely a nerd, and proud of it. But only recently have I begun diving into the world of tabletop games like Dungeons & Dragons (otherwise known as D&D). The poster (left), from one of the many Dungeons & Dragons-themed films of recent decades, gives some sense of the genre.

Fans of games like Dungeons & Dragons know that the fun comes, in part, from a creative Dungeon Master--an all-powerful narrator who follows a storyline but has free rein to improvise in response to players' actions and the fate of the dice. This kind of spontaneous yet coherent storytelling is extremely difficult for artificial intelligence, even as AI has mastered more constrained board games such as chess and Go. The best text-generating AI programs too often produce confused and disjointed prose. So some researchers view spontaneous storytelling as a good test of progress toward more intelligent machines. An attempt to build an artificial Dungeon Master offers hope that machines able to improvise a good storyline might be built.

In late November, a college senior at Brigham Young University named Nick Walton published a short fable called "My Musical Troupe of Orcs Uses Music to Advance Orc Rights." In the story, written in the second person, you are a goblin. "I am a goblin!" you say proudly. "And I'm glad to be one." "Well then, congratulations," says the orc captain. Over the course of a few hundred words, some big things happen: You ask if you can join the orc band.

In late November, a college senior at Brigham Young University named Nick Walton published a short fable called "My Musical Troupe of Orcs Uses Music to Advance Orc Rights." In the story, written in the second person, you are a goblin. "I am a goblin!" you say proudly. "And I'm glad to be one." "Well then, congratulations," says the orc captain. Over the course of a few hundred words, some big things happen: You ask if you can join the orc band.