This paper presents an innovative framework that integrates Large Language Models
(LLMs) with an external Thinker module to enhance the reasoning capabilities of LLM-based …

Are current language models capable of deception and lie detection? We study this question
by introducing a text-based game called $\textit {Hoodwinked} $, inspired by $\textit {Mafia} …

In this paper, we study the class of games known as hidden-role games in which players get
privately assigned a team and are faced with the challenge of recognizing and cooperating …

Abstract The Game Theory & Multi-Agent team at DeepMind studies several aspects of multi-
agent learning ranging from computing approximations to fundamental concepts in game …

As humans, we use linguistic elements referencing time, such as “before” or “tomorrow”, to
easily share past experiences and future predictions. While temporal aspects of the …

In this paper we present a Reinforcement Learning environment that leverages agent
cooperation and communication, aimed at detection, learning and ultimately penalizing …

Ad-hoc teamwork models are a relevant tool for addressing Multi-Agent Systems (MAS)
problems with unknown teammates [5, 20]. For example, we often encounter situations in …

This paper introduces Werewolf Arena, a novel framework for evaluating large language
models (LLMs) through the lens of the classic social deduction game, Werewolf. In Werewolf …

Computational game theory has led to significant breakthroughs in AI dating back to the start
of AI as a discipline. These include the strongest AI agents for both recreational and practical …

Strategic interactions between agents can be formalized by game theory. In recent years
modern learning techniques, coupled with the ubiquity of data and proliferation of compute …