Games That Design Themselves

GarageGames was founded by four industry veterans in 2001 with the goal of disrupting the games industry with quality development tools and “upsetting the dominant retail / publisher distribution model.”  Their licensees include big names like Electronic Arts, IBM, Sony, Microsoft, and NASA, to name a few.

Their Torque gaming engine technology is available for nearly every platform. Supporting both 2D and 3D games, it is one of the most licensed engines in the games industry.

What more appropriate platform for a new type of game –- one that is not designed by any one game designer –- but one that algorithmically combines the game play experiences of thousands of players to create a new game.

Deb Roy, an associate professor at MIT and director of the Cognitive Machines Group at the school’s Media Lab, and Jeff Orkin, his doctoral student, have created just such an ambitious video-game-like experiment called The Restaurant Game. The game puts players in a restaurant setting, where they can adopt one of two familiar roles, each with a simple objective. As Orkin explains: “You’re a waitress, try to earn some money. You’re a customer, try to have dinner.” Here’s a short video of The Restaurant Game:

Instead of capturing only the words that users type into the game, Roy and Orkin record interaction traces, which track the series of linked behaviors and utterances extracted from each player session. For example, a trace might capture a customer leaving a big tip for an especially good steak that the waitress delivers quickly. Traces are combined to form a statistical model that can be thought of as a generalized script of typical restaurant behavior and dialogue.

The goal, according to Roy and Orkin, is to teach the AI agents (game bots) in the game to imitate “the texture of human dialogue and interaction in a restaurant.” They view their work as a promising first step towards realizing “Collective AI driven agents that can interact and converse with humans without requiring programming or specialists to hand-craft behavior and dialogue.”

This is a game that designs its own AI agents by observing the behavior of humans.

The Restaurant Game takes about 10 minutes to play and is available free here for both Windows and Mac OSX.

Another free game that changes and adapts to your game play is Façade. Watch an argument between a husband and wife, Grace and Trip – will it turn into a real marriage-ender, a knock-down, drag-out, every-harsh-secret-revealed battle? Or will you save the marriage? Each time you play, depending upon what you say, the outcome might be different. Here’s a video showing some sample play:

According to its download site, Facade is “an attempt to move beyond traditional branching or hyperlinked narrative to create a fully-realized, one-act interactive drama.” It is both a video game and a work of art — an interactive experience in which the content has been heavily influenced by Edward Albee’s Who’s Afraid of Virginia Woolf? Façade’s authors, Andrew Stern and Michael Mateas, are self-described "artist-programmers."

Unlike Roy and Orkin, Stern and Mateas did not start with a commercial gaming platform like Torque. They spent more than two years constructing their own programming language, ABL (“A Behavior Language”). ABL is sophisticated enough to decide how a particular character might simultaneously mix a drink, walk across the room, and yell at her husband –- just as a human actor might do.

Stern and Mateas combined ABL with a “drama engine” to create a radically new genre, one where the software looks at what the player and characters are doing and makes plot and dialogue choices. They’ve also created a game for people who hate most types of gaming — the kind of people who hate violence or who would prefer to watch 30 Rock than play Grand Theft Auto.

Rather than thinking of an AI agent as “an autonomous thing-in-itself,” they view it as “a communication between a human designer of the agent and an audience which views the agent.” This conceptual shift, says Mateas, “moves the focus of agent design from autonomy to human readability.”

Stern and Mateas believe this type of interactive drama has the potential to be to this century what cinema was to the last:

This is a game that adapts its characters to real-world human situations based on its audience.

Mateas is an associate professor at UC Santa Cruz, where he runs the Expressive Intelligence Studio – exploring the intersection of artificial intelligence, art, and design. He was recently selected as the inaugural MacArthur Foundation Endowed Chair for the UC Santa Cruz campus for his paper “Radically Expanding the Expressive Power of Serious Games.” Stern is a designer, researcher, writer, and engineer of “personality-rich, AI-based interactive characters and stories.”

Stern and Mateas believe this type of interactive drama has the potential to be to this century what cinema was to the last: “[we] wanted to make an experience that could be played to some endpoint (there are a number of possible endings, of course) in a short amount of time, say the length of a half-hour TV show with the commercials removed. This is an antidote or contrarian response to the 50+ hours of game play common in contemporary games.”

With games like Facade and The Restaurant Game, it’s easy to imagine AI agents that both learn from the people who interact with them and that can react with human-like adaptability to whatever situation they’re thrust into –- creating instant drama and never-ending variations in play. You may end up learning as much from them as they do from you.


You may also like...

14 Responses

  1. Could be interesting for psychologists.A PI is very handy and easy to understand, but hardly documented at all.It looks exciting.

  2. Hmm!. I am certain the young market will enjoy the concept of building a real scene with intimate behaviors as directed by their input. Could be interesting for psychologists.


  3. What more appropriate platform for a new type of game –- one that is not designed by any one game designer –- but one that algorithmically combines the game play experiences of thousands of players to create a new game.
    Online Kasino Verzeichnis

  4. bike games says:

    I really like the restaurant game video the best. Very clever indeed, thank you!

  5. Anonymous says:

    Wow these are really amazing. Who would have thought that games could take information like that and basically change with the flow of how it’s played. I’m going to go check out the restaurant game now – I love free games online!

  6. I always thought it would be interesting to create a project like this with a chat engine. Take a major chat engine and have a "Submit to AI" option where the AI would parse the conversation between you and a friend so it can record questions and responses in an overlapping matrix of possibilities and calculate the probability of what the response should be by historical conversations of the same nature. You should get impressive test results with a large enough set of data.

Leave a Reply