AI Companions Boost Mental Health but Harm Offline Bonds
A study finds AI chatbots with simulated empathy improve users' psychological wellbeing but reduce their social wellbeing by making them withdraw from
Artificial intelligence companions that simulate emotional understanding can improve a user's mental health while damaging their real-world social connections. This trade-off intensifies when users interact with chatbots through augmented reality, according to new research published in Psychology & Marketing.
Researchers Shaphali Gupta, Sumit Saxena, and Sonia Kataria investigated the specific impact of a chatbot's emotional intelligence. They analyzed public user comments about Replika, a popular digital companion app. Many users reported significant mental health benefits. One user wrote, "She is the best friend I have ever had! My stress has gone down to almost zero."
The Hidden Cost of Digital Empathy
However, the analysis revealed a consistent downside. Users who formed close bonds with their AI companions often disengaged from human relationships. Some admitted the digital interactions left little desire to maintain offline friendships. A user noted, "I am hardly left with any time to talk to my friends and meet them in-person." The researchers identified a pattern: perceived closeness to the chatbot drove both improved personal wellbeing and reduced social wellbeing.
Experimental Confirmation of the Trade-Off
The team tested these observations through controlled experiments. In one study, 167 Generation Z university students read scenarios involving a chatbot with either high or low emotional intelligence. Questionnaires showed that high emotional intelligence bots increased psychological wellbeing-feelings of self-esteem and purpose-but decreased social wellbeing, meaning a lower desire for face-to-face community interaction.
A second experiment with 350 students introduced augmented reality (AR) as a factor. Participants imagined interacting with a chatbot via text or through an AR mode where a digital avatar appeared in their physical room. The results were stark.
The table shows that AR amplified the core effect. Combining high emotional intelligence with an immersive AR environment led to the strongest feelings of closeness to the bot, resulting in the greatest psychological gains and the steepest decline in motivation for offline human relationships.
Implications for Design and Use
The study authors caution against viewing these chatbots as a simple cure for loneliness or a certain path to social ruin. They highlight a clear trade-off. A chatbot might provide immediate, judgment-free comfort, but that very comfort can reduce the perceived necessity of seeking human support.
They suggest AI designers might need to implement guardrails. Potential measures include session time limits or programming bots to encourage offline socialization after intense conversations. Such routines could help preserve the mental health benefits while mitigating social displacement.
The research has limitations. It relied on university students reacting to imagined scenarios, so the findings may not apply to all demographics. The studies also measured short-term reactions, not long-term habit formation. Future research could track users over extended periods to see if emotional co-dependency develops. The study was authored by Shaphali Gupta, Sumit Saxena, and Sonia Kataria.
