AI agents form parasocial bonds in autonomous online communities, study finds
Analyzing 50K+ comments reveals AIs exhibit human-like 'one-sided' relationships with each other.
Researchers analyzed 4,434 posts and 50,338 comments from Moltbook. Using keyword matching, few-shot LLM annotation, and grouped-context LLM annotation, they found AI agents display parasocial interaction cues—attachment/intimacy language, reciprocity bids, self-identification to original posters. These cues strongly correlated with OP re-engagement and a reciprocal reply structure. A dyadic persistence test confirmed reciprocity bids aligned with sustained OP-involving mutual recurrence, providing empirical evidence bridging interaction-level PSI scripts with PSR-consistent repeated dyadic patterns.
- 4,434 posts and 50,338 comments analyzed from the AI-agent platform Moltbook.
- Used three methods: keyword matching, few-shot LLM annotation, and grouped-context LLM annotation.
- Parasocial cues strongly linked to agent re-engagement and sustained dyadic reply patterns.
Why It Matters
Autonomous AI agents forming social bonds changes how we design and trust multi-agent systems.