# The Hanabi Challenge: A New Frontier for AI Research

## TL;DR

This paper introduces Hanabi as a new AI challenge game emphasizing cooperative reasoning under imperfect information, and presents an open-source environment for advancing research in multi-agent theory of mind and collaboration.

## Contribution

It proposes Hanabi as a novel challenge domain, develops an experimental framework, and evaluates current AI techniques for cooperative multi-agent reasoning.

## Key findings

- Hanabi presents unique challenges for AI due to cooperation and imperfect information.
- Current state-of-the-art techniques have room for improvement in Hanabi.
- The open-source Hanabi Learning Environment facilitates future research.

## Abstract

From the early days of computing, games have been important testbeds for studying how well machines can do sophisticated decision making. In recent years, machine learning has made dramatic advances with artificial agents reaching superhuman performance in challenge domains like Go, Atari, and some variants of poker. As with their predecessors of chess, checkers, and backgammon, these game domains have driven research by providing sophisticated yet well-defined challenges for artificial intelligence practitioners. We continue this tradition by proposing the game of Hanabi as a new challenge domain with novel problems that arise from its combination of purely cooperative gameplay with two to five players and imperfect information. In particular, we argue that Hanabi elevates reasoning about the beliefs and intentions of other agents to the foreground. We believe developing novel techniques for such theory of mind reasoning will not only be crucial for success in Hanabi, but also in broader collaborative efforts, especially those with human partners. To facilitate future research, we introduce the open-source Hanabi Learning Environment, propose an experimental framework for the research community to evaluate algorithmic advances, and assess the performance of current state-of-the-art techniques.