Social bots as companions: Boon or bane for mental health community engagement?

Purpose As AI-powered social bots are increasingly used as companions in mental health communities, their impact on human-to-human interactions remains unclear. This study examines how social bots' replies shape users' subsequent replying behavior and investigates the underlying psychological mechanisms behind users' perception of companionship. Design/methodology/approach This study adopted a multi-method research design comprising two studies. Study 1 examined whether social bot responses were associated with subsequent engagement of posting users and other community members by comparing posts that received social bot responses with posts that received only human responses. Study 2 examined the underlying mechanisms by investigating how users perceive responses from humans and from social bots, respectively, through an online experiment. Findings Posts receiving social bots' responses were associated with faster replies but lower subsequent reply frequency from original posters. They were also associated with fewer human repliers, fewer human replies and lower average replies per user. Human responses outperformed social bots' responses in terms of perceived companionship. The difference was explained by higher perceived psychological closeness, perceived authenticity and empathy. Research limitations/implications Although social bots may accelerate human responses, they cannot replicate the psychological depth of human connection. Platforms should therefore deploy social bots cautiously as supplementary companionship resources in virtual mental health communities. Originality/value This study contributes to an understanding of how social bots' replies reshape community engagement in virtual mental health communities. It also reveals the psychological mechanisms underlying users' different perceptions of human users and AI-powered social bots.