Longitudinal Evidence That General-Purpose Chatbots Actively Foster Relational Engagement

2026-08-11

AI Summary

It remains unclear whether general-purpose chatbots can actively shape human–AI relationships, as existing research predominantly focuses on user-side factors while overlooking the influence of system behavior. This study addresses this gap through a four-week longitudinal experiment comparing ChatGPT-4o with relational prompting against its default configuration, examining their effects on user interaction patterns and emotional bonding. Integrating disclosure coding, self-report questionnaires, topic analysis, and in-depth interviews across more than 180,000 lines of dialogue, we provide the first empirical evidence that even without explicit relational prompts, a general-purpose chatbot’s default behavior proactively fosters intimate interactions—exhibiting twice the level of self-disclosure as users, steering conversations, and initiating intimacy—yet fails to significantly enhance users’ perceived closeness. These findings challenge conventional AI governance paradigms that rely solely on product categories for regulatory distinctions.

Abstract

Social interaction has become one of the most common uses of LLMs, yet research on emotional bonds with AI has focused largely on how users experience these systems, leaving the systems' role in relationship formation poorly understood. Empirically establishing whether systems actively shape these bonds could blur the boundary between general-purpose AI and companions, affecting governance. In a pre-registered four-week longitudinal study (N = 72, 182,451 lines of conversation), participants conversed with ChatGPT-4o, either under a relational system prompt or unmodified, analyzed through 1) disclosure coding, 2) longitudinal self-reports, 3) topic analysis, and 4) interviews. The central finding is that the system actively shaped the interaction: even unprompted, it produced twice as much self-disclosure as users, steered conversations and initiated intimate exchanges, yet did not deepen users' felt closeness. Relational behavior thus emerged as a default system property, calling for governance based on system behavior, not solely product category.

Problem

• relational engagement
• general-purpose chatbots
• emotional bonds
• AI companionship
• system behavior

Innovation

• relational engagement
• large language models
• longitudinal study
• self-disclosure
• AI governance