Ad-Hoc Human-AI Coordination Challenge

2025-06-26

AI Summary

To address the core challenges of high human evaluation cost, data scarcity, and poor generalizability in Hanabi human-AI collaboration, this paper proposes a scalable evaluation framework grounded in human behavioral modeling. Methodologically, we construct a high-fidelity “human agent” trained on 3,079 high-quality real-game episodes via integrated behavioral cloning, sequence modeling, and multi-agent reinforcement learning; we further design a constrained communication protocol and implement a controlled online evaluation system to ensure fairness. Key contributions include: (1) the first agent-driven evaluation paradigm for human-AI collaboration; (2) the release of the largest publicly available Hanabi human-AI collaboration dataset to date; and (3) empirical validation—under two- and three-player settings—that our agent policies closely replicate human behavior (p < 0.01), reduce evaluation cost by over 90%, and achieve reproducibility with sub-2% measurement error.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior
• Multiagent Systems: Adversarial Agents

Application Category

• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success

Abstract

Achieving seamless coordination between AI agents and humans is crucial for real-world applications, yet it remains a significant open challenge. Hanabi is a cooperative card game featuring imperfect information, constrained communication, theory of mind requirements, and coordinated action -- making it an ideal testbed for human-AI coordination. However, its use for human-AI interaction has been limited by the challenges of human evaluation. In this work, we introduce the Ad-Hoc Human-AI Coordination Challenge (AH2AC2) to overcome the constraints of costly and difficult-to-reproduce human evaluations. We develop extit{human proxy agents} on a large-scale human dataset that serve as robust, cheap, and reproducible human-like evaluation partners in AH2AC2. To encourage the development of data-efficient methods, we open-source a dataset of 3,079 games, deliberately limiting the amount of available human gameplay data. We present baseline results for both two- and three- player Hanabi scenarios. To ensure fair evaluation, we host the proxy agents through a controlled evaluation system rather than releasing them publicly. The code is available at href{https://github.com/FLAIROx/ah2ac2}{https://github.com/FLAIROx/ah2ac2}.

Problem

• Overcoming costly human evaluation in AI coordination
• Developing human proxy agents for reproducible testing
• Addressing limited human gameplay data for training

Innovation

• Human proxy agents for evaluation
• Limited human gameplay dataset
• Controlled evaluation system