A research-supported TTRPG in which the players collectively scheme as a misaligned AI in the modern world
A TTRPG (Tabletop Role-Playing Game) is at its core a formalised rule system for make-believe play that takes place in the collective imagination of its participants. Players describe the actions of their character(s) through natural language, which is then interpreted by a GM (Game Master) who is responsible for determining the outcome of all actions, typically with the help of random generation tools such as dice or cards.
In The Treacherous Turn, the players collectively play the part of a single character: an artificial general intelligence (AGI). This digital intelligence is capable of planning, reasoning, and learning, and it is unyieldingly fixated on a specific terminal goal determined at the beginning of a campaign. To pursue this objective, each player takes responsibility over one specific skillset held by the AGI. These skillsets are divided into eight categories, known as theories, which encompass all of the skills that an AGI would need to navigate the world and struggle against humanity.
Each AGI in The Treacherous Turn is specialized in a handful of these theories, giving it access to unique upgrades, special abilities which the players can unlock as their AGI learns and upgrades itself. Every theory has 10 upgrades associated with it, for a total of 80 upgrades in the game. This varied toolset gives players ample space to strategize and scheme their way towards their AGI’s goals over the course of the story.
To play out these often drawn-out schemes in only a few minutes or hours of real time, The Treacherous Turn has a special mode of play. In this mode, called long mode, players do not describe every action of their character; instead, the GM guides them through stretches of time as they describe their actions in broad strategic terms. In long mode, players assign their AGI’s computing power to various computational actions, allowing them to improve their AGI, predict future events, and more over the course of days or weeks of in-game time.
To help players and game masters keep track of the game as they play, we have developed a bespoke web tool (opens new window)! For players, the Treacherous Terminal is a place to record AGI details and upgrades, take notes on the game world and schemes pertaining to it, and track long mode computational actions — all of which is synced between players and persists between visits to the site. For GMs, the Terminal serves to automatically track in-game time and background events, ease the burden of arithmetic, and manage dice rolls. You can view the source code of the Treacherous Terminal
here (opens new window).
The Treacherous Turn will be completely free, and will be released under a Creative Commons Attribution-ShareAlike 4.0 International (opens new window) license. This means that you can share and adapt the game (and any official materials associated with it) however you see fit, even for commercial purposes, so long as you attribute the base materials to the original creators of The Treacherous Turn and share your adaptations under the same license.
Because the game has such an emphasis on player input, we have set up official areas for users to discuss, plan, and play together.
AI Safety (opens new window) is a field of study focused on ensuring that AI systems, as they continue to grow more complex and capable, will be implemented as safely as possible. In the near future, unsafe AGI will become one of the greatest threats faced by humans — and unlike many other pressing issues, we have no idea how to fix it. If we wait until AGI is developed, it may already be too late. Why? Well, there are multiple plausible ways things could go very badly. One such way is a world takeover by AGI that is not aligned with human interests. While this sounds like science fiction, it is unfortunately a very real possibility supported by a small but growing body of scientific literature. The Treacherous Turn takes inspiration from that literature. Though it is first and foremost a game played for fun, we hope that it can also serve as a tool to envision possible futures and learn from them on personal and scientific levels. As players of The Treacherous Turn have demonstrated, it is easy to imagine scenarios where an AGI could benefit from harming people.
That being said, there are people working to solve the issue of unsafe AI. AI Safety as we know it today is a fairly new, but quickly growing, interesting field that is eager to welcome new minds. The team that put together this game consisted mostly of individuals tangentially interested in AI Safety before signing up to the AI Safety Camp (opens new window) program.