A fresh study suggests that responses from artificial intelligence (AI) are perceived as more compassionate than those from mental health professionals. The findings challenge the assumption that only humans with lived experience can provide genuine emotional support.

Published on Jan. 10 in Communications Psychology, the research involved four experiments with 550 participants. Volunteers shared personal experiences and rated responses—both AI-generated and from trained professionals—based on compassion, attentiveness, and preference. Surprisingly, AI responses were rated 16% more empathetic and were chosen 68% of the time, even when participants knew whether a human or AI had written them.

Lead author Dariya Ovsyannikova from the University of Toronto attributed AI’s success to its ability to analyze intricate details and maintain an unbiased approach. Unlike human therapists, who can suffer from burnout and fatigue, AI remains consistent in providing emotionally supportive replies.

Eleanor Watson, an AI ethics expert, noted that AI’s vast data processing capability offers a unique advantage. It can present insights and techniques beyond a therapist’s individual training. Additionally, AI’s accessibility makes mental health support more reachable, especially in regions where professional help is scarce.

However, risks remain. AI-generated empathy may be exaggerated, creating unrealistic expectations. Privacy concerns also loom, as sensitive mental health data could be misused. Experts stress the need for strong ethical guidelines to prevent exploitation.