We Shape AI, and Thereafter AI Shape Us: Humans Align with AI through Social Influences

Jingshu Li, Tianqi Song, Beichen Xue, Yi-Chieh Lee

Keywords: Opinions, Alignment from Social Influence, Human-AI Interaction, Challenges and Future Directions

TL;DR: A perspective on how AI can subtly shape human cognition, emotions, and behavior, mirroring social influence in human interactions.

Abstract:

The rapid advancement of AI technologies has facilitated the integration of AI into various social scenarios in daily life, making it crucial to understand how AI influences humans and fosters human-AI alignment. While previous research has extensively explored how AI can be consciously designed to affect human behavior, a critical yet underexplored area is how AI can subtly shape human cognition, emotions, and behavior, mirroring social influence in human interactions.
By examining social influence mechanisms and case studies from recent human-AI interaction studies, this paper identifies two key mechanisms through which AI influences humans: contagion and conformity. We further explore the challenges and opportunities of AI-driven social influence, urging future research to address risks such as cognitive exploitation and group manipulation, while also leveraging its potential benefits, including fostering positive emotional contagion and supporting healthy behavioral interventions.
We advocate for a governance framework that integrates technological, ethical, and societal considerations and call for multidisciplinary collaboration to explore new paradigms of social influence in human-AI interactions. Ultimately, this perspective study provides both theoretical insights and practical pathways toward a harmonious and symbiotic human-AI society.

Submission Type: Short Paper (4 Pages)

Archival Option: This is an archival submission

Presentation Venue Preference: CHI 2025

Submission Number: 83