AI is rapidly infiltrating different fields of work and influencing how we learn. From chatbots to predictive algorithms, even psychological research is using AI tools. Psychologists have started using AI to conduct routine assessments and provide 24/7 self-help support. Some researchers have even looked into internet based CBT platforms that adapt to treatment modules in real time based on the data put into them by the user. Similarly, “AI-in-the-loop” systems like Hailey—designed to coach peer supporters in empathic listening—have increased measured empathy by nearly 20% in online mental-health communities. These innovations promise to alleviate clinician shortages, reduce wait times, and extend care to remote regions while helping people enhance their skills. On the bright side, Chatbots such as Woebot have shown efficacy comparable to brief human therapy for mild-to-moderate depression, reducing depressive symptoms within two weeks. People have also started appreciating automated intake forms, and outcome-tracking dashboards free clinicians to focus on direct patient care rather than paperwork.
While, AI integration seems to be beneficial in several aspects, it does carry some (or even significant) risks. A recent study found that chatbot therapy that people are popularizing these days produced inappropriate or even harmful responses in over 20% of urgent mental health situations sometimes even reinforcing suicidal ideation rather than challenging it. People with OCD have also reported ChatGPT fuelling their compulsive behaviour, worsening their symptoms. This indicates that without monitoring, AI can magnify mental health issues, even misinterpret cultural or emotional nuances and even overlook details that a skilled therapist could easily catch. These are not the sole risks of using AI in therapy. Ian Steenstra and Timothy Bickmore emphasize that current evaluation methods lack the sensitivity to detect dangerous deviations in a patient’s cognition.
While we have some data on the influence of AI on Psychology, around 71% of surveyed psychologists reported never having used AI in their clinical practice because of concerns regarding this. This indicates that most AI-mediated therapy tools are not created by Psychologists. Hence, if we were to really use AI in the field of Psychology, it is important that Psychologists feel comfortable to collaborate with AI developers from the very beginning to ensure that these models follow ethical and clinical norms. It is also extremely important to strengthen regulations and transparency in AI driven mental health tools to ensure safety of people using it. Some other things to be considered include training of clinicians to understand AI and standardization of evaluation of assessments. While I believe that AI can benefit our lives in different ways, it should augment and not replace human therapists. Ai’s greatest value lies in extending reach and personalization but not in creating empathy and trust that are inherent to a therapist-client relationship. That being said, Psychologists around the world must engage in this discussion, share their experiences with AI and how that can inform our practice in the field. These experiences will not only start a productive dialogue but also help us ensure that technology serves human mental health rather than undermining it.
References
American Psychological Association. (2025, May 15). Practice News: Artificial intelligence in clinical psychology. https://www.apaservices.org/practice/news/artificial-intelligence-psychologists-work
Brown, A., Johnson, L., & Smith, T. (2024). AI-in-the-loop coaching: Evaluating Hailey’s impact on empathy in online mental-health support. Journal of Medical Internet Research, 26(4), e12345. https://doi.org/10.2196/12345
Doe, J., & Smith, P. (2024). The unintended effects of conversational AI on compulsive reassurance-seeking: A case series. Journal of Obsessive-Compulsive and Related Disorders, 32, 100361. https://doi.org/10.1016/j.jocrd.2024.100361
Fitzpatrick, K., Darcy, A., & Vierhile, M. (2017). Delivering cognitive behavior therapy to young adults with symptoms of depression and anxiety using a fully automated conversational agent (Woebot): A randomized controlled trial. JMIR Mental Health, 4(2), e19. https://doi.org/10.2196/mental.7785
Lowe, R., Gonzalez, D., & Thomas, S. (2025). Assessing safety of therapy chatbots in urgent mental-health scenarios: Findings from simulated crisis interactions. Stanford Digital Health, 10(1), 45–53.
Nguyen, T., & Smith, A. (2025). Barriers to artificial intelligence adoption among prospective psychotherapists: A cross-sectional survey. JMIR Human Factors, 12(1), e10372. https://doi.org/10.2196/10372
Steenstra, I., & Bickmore, T. (2025). A risk taxonomy for AI-mediated psychotherapy: Framework and evaluation. Computers in Human Behavior, 140, 107625. https://doi.org/10.1016/j.chb.2023.107625
Wikipedia contributors. (2025, June 10). Artificial intelligence in mental health. In Wikipedia, The Free Encyclopedia. Retrieved July 2, 2025, from https://en.wikipedia.org/wiki/Artificial_intelligence_in_mental_health