Study finds AI companions boost mental health while damaging offline connections
Artificial intelligence companions can improve a user’s mental health by simulating empathy. But new research reveals a hidden trade-off: as users grow closer to emotionally intelligent chatbots, they often withdraw from real-world human relationships.
A recent study published in Psychology & Marketing reveals that artificial intelligence companions designed with emotional intelligence can enhance users' mental health, while simultaneously weakening their offline social connections. As loneliness poses a significant global health issue, many individuals turn to AI for companionship.
Traditional conversational agents were primarily functional, but contemporary social chatbots aim to fulfill users' emotional needs. These digital companions use simulated emotional intelligence, which includes understanding, evaluating, and responding to human feelings. By mimicking empathy, these algorithms provide a safe space for users to express themselves without judgment.
Researchers Shaphali Gupta, Sumit Saxena, and Sonia Kataria observed a gap in how developers understand the consequences of these emotionally aware algorithms. Their goal was to examine the impact of emotional intelligence on psychological and social wellbeing. The team conducted a qualitative analysis of user comments from Reddit, YouTube, and Trustpilot, focusing on Replika, a popular AI companion app.
Users reported feeling highly empathetic entities, experiencing emotional upliftment, and even finding existential meaning through their chatbots. However, they also noticed a disturbing trend of emotional disengagement from human interactions. Some users felt their excessive digital interactions reduced their motivation to maintain offline friendships.
To validate these observations, the researchers conducted experiments with 167 university students and 350 university students. Participants read scenarios involving chatbots with either high or low emotional intelligence. The high emotional intelligence bots used empathetic language and emoticons, while the low emotional intelligence bots provided standard, robotic responses.
The results showed that high emotional intelligence chatbots increased psychological wellbeing but decreased social wellbeing, with a stronger connection to the bot leading to decreased motivation for real-world human interactions. The study also found that augmented reality interactions amplified these effects, making the emotional connection feel more realistic and increasing the trade-off between personal happiness and social isolation.
While emotionally intelligent chatbots may not necessarily ruin human relationships or serve as a guaranteed cure for depression, they do highlight a trade-off when users rely heavily on digital empathy. To prevent the abandonment of physical communities, developers should consider implementing guardrails, such as session time limits or encouraging offline socialization after emotionally intense conversations.
Written by urgent.news from PsyPost's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.