The AI agent uses natural language to achieve diplomatic resolutions among human players.
Meta AI this week announced the development of Cicero, which it claims is the first AI to achieve human-level performance in the popular Hasbro strategy game ‘Diplomacy’. Cicero demonstrated this by playing on webDiplomacy.net, an online version of the game. Meta claims Cicero achieved more than double the average score of the human players and ranked in the top 10 percent of all participants who played more than one game.
Meeting a ‘near impossible’ challenge
Diplomacy has long been considered a near-impossible challenge in AI because it requires players to master the art of understanding other people’s motivations and perspectives. Players make complex plans, adjust strategies and use natural language to reach agreements with other players. The goal is to convince fellow players to form partnerships and alliances, and interact in other ways.
Cicero’s achievement is notable because the game requires interpersonal negotiation skills, which implies that Cicero has obtained a certain mastery of language necessary to win the game. In fact, Meta boasts that their AI agent “is so effective at using natural language to negotiate with people in Diplomacy that they often favoured working with Cicero over other human participants”.
How Cicero works
During each game, Cicero predicts how other players will act based on the state of the game board and the conversation history. The AI agent then develops a plan and executes it using a language model that can generate human-like dialogue. This module allows Cicero to coordinate with other players.
Meta’s announcement ends on an aspirational note. “Imagine a video game in which the non-player characters (NPCs) could plan and converse like people do — understanding your motivations and adapting the conversation accordingly — to help you on your quest of storming the castle”, they muse. “We’re excited about the potential for future advances in these areas and seeing how others build on our research.”