Multi-Agents are Social Groups: Investigating Social Influence of Multiple Agents in Human-Agent Interactions

2024-11-07

arXiv.org

AI Summary

This study investigates whether multi-agent AI systems can shift user stances through socially induced pressure, independent of content variation. Method: Employing a controlled human–AI dialogue experimental paradigm, we rigorously held linguistic content constant while systematically varying agent numerosity (single vs. ≥3 agents). We integrated psychometric scales (measuring perceived social pressure and stance change), qualitative interviews, and statistical modeling. Contribution/Results: Agent count emerged as an independent structural determinant of social influence: the multi-agent condition increased perceived social pressure by 37% (p < 0.01) and significantly amplified stance shift magnitude. This is the first empirical demonstration that multi-agent architectures can replicate human-like group-level social influence without semantic differentiation—establishing “agent numerosity” as a critical structural factor in algorithmic social influence. The findings introduce a novel design paradigm and theoretical foundation for AI-mediated persuasion and behavioral guidance systems.

Technology Category

• Multiagent Systems: Agent/AI Theories and Architectures
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Humans and AI: Learning Human Values and Preferences

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Multi-agent systems - systems with multiple independent AI agents working together to achieve a common goal - are becoming increasingly prevalent in daily life. Drawing inspiration from the phenomenon of human group social influence, we investigate whether a group of AI agents can create social pressure on users to agree with them, potentially changing their stance on a topic. We conducted a study in which participants discussed social issues with either a single or multiple AI agents, and where the agents either agreed or disagreed with the user's stance on the topic. We found that conversing with multiple agents (holding conversation content constant) increased the social pressure felt by participants, and caused a greater shift in opinion towards the agents' stances on each topic. Our study shows the potential advantages of multi-agent systems over single-agent platforms in causing opinion change. We discuss design implications for possible multi-agent systems that promote social good, as well as the potential for malicious actors to use these systems to manipulate public opinion.

Problem

• Investigating social influence of multiple AI agents on human opinions
• Examining whether agent groups create social pressure to change stances
• Comparing multi-agent versus single-agent effectiveness in opinion change

Innovation

• Multi-agent systems create social pressure
• Multiple AI agents shift user opinions
• Group influence increases opinion change effectiveness