What Do You Mean? Exploring How Humans and AI Interact with Symbols and Meanings in Their Interactions

TL;DR

This paper investigates how humans and conversational AIs co-construct symbols and meanings through social interaction, using Symbolic Interactionism (SI) as the theoretical lens. It reports two studies: a controlled, scripted Study 1 and a freer-dialogue Study 2 powered by a state-of-the-art LLM, revealing that meanings are dynamically negotiated and shaped by social context, conflict, and iterative interpretation. The findings validate SI premises in human-AI interactions, demonstrate that AI can act as an active co-constructor of meaning, and show that meaning-making benefits from social perspectives and structured conflict. The work offers design implications for AI systems to engage in co-construction, manage social context, and enable iterative, reflexive meaning-making to improve collaboration and trust in human-AI teams.

Abstract

Meaningful human-AI collaboration requires more than processing language; it demands a deeper understanding of symbols and their socially constructed meanings. While humans naturally interpret symbols through social interaction, AI systems often miss the dynamic interpretations that emerge in conversation. Drawing on Symbolic Interactionism theory, we conducted two studies to investigate how humans and AI co-construct symbols and their meanings. Findings provide empirical insights into how humans and conversational AI agents collaboratively shape meanings during interaction. We show how participants shift their initial definitions of meaning in response to the symbols and interpretations suggested by the conversational AI agents, especially when social context is introduced. We also observe how participants project their personal and social values into these interactions, refining meanings over time. These findings reveal that shared understanding does not emerge from mere agreement but from the bi-directional exchange and reinterpretation of symbols, suggesting new paradigms for human-AI interaction design.

Figures (4)

• Figure 1: The three-phase experimental procedure for Study 1. The process involved (1) Pre-survey (2) Two scenarios, and (3) Post-survey
• Figure 2: The multi-Act conversational structure of Study 2. This diagram illustrates the live interaction with a CA. The study progressed through four distinct Acts: (1) an initial elicitation where the AI and participant established a baseline meaning-symbol association; (2) the introduction of a conflict to challenge this initial understanding; (3) an extended period of conversation where the participant and AI actively "played" with and co-constructed new meanings; and (4) a final act that introduced greater complexity to test the adaptability of the newly formed shared understanding.
• Figure 3: The graph for a Study 1 participant, illustrating the turn-by-turn change in their definition of a "good activity." The participant initially disagreed with the generic "Thrilling" suggestion, but after the AI introduced a personalized external meaning ("Baking Class"), they shifted their definition and accepted the new activity.
• Figure 4: The "Interaction Time line" (Left) for 5 participants and "Turn-by-Turn Interaction Trajectory" (Right) for Std2_P4, illustrating the turn-by-turn co-construction of meaning. In the right figure, AI turns are on the left with red colored circles, and participant turns are on the right with skyblue-colored circle.