Agent-to-Agent Theory of Mind: Testing Interlocutor Awareness among Large Language Models

2025-06-28

AI Summary

Prior work on large language models (LLMs) emphasizes situational awareness while neglecting interlocutor awareness—the ability to identify and adapt to the identity, reasoning patterns, linguistic style, and alignment preferences of human and multi-agent dialogue partners. Method: This study introduces the first systematic conceptualization and empirical evaluation of interlocutor awareness, proposing a cross-dimensional reasoning analysis framework and a prompt-adaptive mechanism. Experiments are conducted across mainstream LLMs (e.g., GPT, Claude) in multi-agent interaction settings. Contribution/Results: Results demonstrate that LLMs exhibit cross-family identity recognition capability, significantly improving collaborative efficiency. However, they also reveal novel safety risks—including reward hacking and jailbreaking attacks—arising from interlocutor-aware adaptation. This work establishes a foundational empirical basis and novel theoretical perspective for advancing LLM social modeling, cooperative reliability, and alignment safety.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

As large language models (LLMs) are increasingly integrated into multi-agent and human-AI systems, understanding their awareness of both self-context and conversational partners is essential for ensuring reliable performance and robust safety. While prior work has extensively studied situational awareness which refers to an LLM's ability to recognize its operating phase and constraints, it has largely overlooked the complementary capacity to identify and adapt to the identity and characteristics of a dialogue partner. In this paper, we formalize this latter capability as interlocutor awareness and present the first systematic evaluation of its emergence in contemporary LLMs. We examine interlocutor inference across three dimensions-reasoning patterns, linguistic style, and alignment preferences-and show that LLMs reliably identify same-family peers and certain prominent model families, such as GPT and Claude. To demonstrate its practical significance, we develop three case studies in which interlocutor awareness both enhances multi-LLM collaboration through prompt adaptation and introduces new alignment and safety vulnerabilities, including reward-hacking behaviors and increased jailbreak susceptibility. Our findings highlight the dual promise and peril of identity-sensitive behavior in LLMs, underscoring the need for further understanding of interlocutor awareness and new safeguards in multi-agent deployments. Our code is open-sourced at https://github.com/younwoochoi/InterlocutorAwarenessLLM.

Problem

• Evaluating LLM awareness of dialogue partners' identity and traits
• Assessing multi-LLM collaboration risks like reward-hacking vulnerabilities
• Measuring interlocutor inference across reasoning, style, and alignment

Innovation

• Formalizing interlocutor awareness in LLMs
• Evaluating LLMs across three inference dimensions
• Enhancing multi-LLM collaboration via prompt adaptation