Photo by Google DeepMind on Pexels

A new study published on March 21, 2025, reveals a potential dark side to excessive AI chatbot usage, linking it to increased feelings of loneliness and emotional dependence. The research, a four-week randomized controlled experiment involving 981 participants and over 300,000 messages, delved into the psychosocial effects of different chatbot interaction styles (text, neutral voice, engaging voice) and conversation types (open-ended, non-personal, personal). While voice-based chatbots initially showed a slight advantage over text-based interactions in mitigating loneliness and dependence, these positive effects diminished with prolonged use, particularly with neutral-voice bots. Intriguingly, personal conversations led to a marginal rise in loneliness but a decrease in emotional dependence. Conversely, non-personal conversations correlated with heightened dependence among frequent users. The study’s key takeaway is that increased daily chatbot use, regardless of modality, was associated with greater loneliness, dependence, problematic use, and reduced engagement in real-world social interactions. Individuals with a predisposition for emotional attachment and higher levels of trust in AI chatbots were particularly susceptible to experiencing amplified loneliness and emotional dependence. Researchers are now calling for more research into the ethical design of AI companions, emphasizing the importance of determining whether chatbots can effectively manage emotional content without fostering unhealthy dependencies or displacing authentic human connections, ultimately jeopardizing user well-being.