How Will AI Integrate Into Psychiatry Long-Term?

Hans Eriksson, MD, PhD, shared his perspectives on the evolving role of artificial intelligence in psychiatric practice and clinical research.

Eriksson expressed measured optimism about AI's trajectory, noting that commercially available tools already performed tasks that would have been unimaginable just 5 years prior. He cautioned that the field should not underestimate the potential of this technology. Drawing an analogy to psychotherapy, Eriksson framed the trained therapist as someone employing a kind of "hidden algorithm"—raising the question of whether AI could replicate or standardize that function with built-in safety barriers.¹

He anticipated that AI would enter psychiatric drug development before clinical practice, particularly through standardization of rating scale interviews. Variability in assessments using instruments such as the Hamilton Depression Rating Scale, the Montgomery-Asberg Depression Rating Scale, and the Positive and Negative Syndrome Scale, represented a meaningful threat to clinical trial validity. Eriksson posited that standardizing these interviews via AI could yield significant methodological gains and represent "the first step of AI coming into a role that is really therapist-like or clinician-like."

Eriksson also raised substantial safety concerns, warning that unregulated AI therapy engines could place vulnerable psychiatric patients at risk. He noted that such patients "could be quite suggestible, perhaps even to doing harmful and dangerous things to themselves or to others."² He advocated for strict regulatory oversight analogous to US Food and Drug Administration frameworks governing drug development, expressing uncertainty about how AI use in clinical practice would ultimately be governed. Eriksson did acknowledge a potential benefit: some patients might prefer disclosing sensitive information to an AI system, perceiving it as a more anonymous and less judgmental interaction than speaking with a human clinician.

Dr Eriksson is chief medical officer of HMNC Brain Health in Munich.

References

1. Spytska L. The use of artificial intelligence in psychotherapy: development of intelligent therapeutic systems. BMC Psychol. 2025;13(1):175.

2. Exploring the dangers of AI in mental health care. Stanford University Human-Centered Artificial Intelligence. June 11, 2025. Accessed March 9, 2026. https://hai.stanford.edu/news/exploring-the-dangers-of-ai-in-mental-health-care