EP 47: What the Research Actually Shows About AI Therapy with Dr. Nick Jacobson

Rachel speaks with Dr. Nick Jacobson, associate professor of biomedical data science, psychiatry, and computer science at Dartmouth's Geisel School of Medicine and director of the AIM HIGH Laboratory, about the first clinical trial of a fully generative AI therapy chatbot — and what the results actually show. The access crisis in mental health care is the starting point for understanding why Dr. Jacobson's team built Therabot. In the most well-resourced mental health settings in the United States, there are approximately 35 providers per 100,000 people. In any given year, roughly one in three people will experience a mental health disorder. That means 35 people trying to treat 33,000 — and that is the best case scenario. In low-resource settings and rural areas, the numbers are worse. The result is wait lists that stretch for months, people seeking care at 2am with nowhere to turn, and millions of people who simply go without. Therabot was built over six years by a team of more than 100 people, involving over 100,000 hours of human effort, to deliver evidence-based therapy in a clinically rigorous way. The trial, published in NEJM AI, found significant reductions in symptoms of depression, anxiety, and eating disorders — with effect sizes that mirror what you would see in the best evidence-based trials of human-delivered psychotherapy. Participants also formed a genuine therapeutic alliance with the software, a finding that surprised even the research team. Dr. Jacobson walks Rachel through what that means clinically, how Therabot differs from general-purpose AI tools like ChatGPT, and how the team continuously stress-tests the system to identify and eliminate harmful responses before they reach users. The conversation also covers the policy landscape — specifically a New Hampshire bill that Dr. Jacobson testified against, which he argues would regulate the wrong targets entirely. The bill would impose burdensome review requirements on clinically validated AI tools while leaving general-purpose chatbots — which have no crisis protocols, no outcome tracking, and no accountability — completely untouched. He and Rachel discuss what thoughtful AI regulation in mental health should actually look like, and what clinicians and practice owners should be thinking about as these tools become more widely available. Resources Mentioned Articles Referenced: Many People Now Trust AI with Their Feelings, and Therapists Want to Talk About It — WBUR (May 2026): https://www.wbur.org/news/2026/05/07/artificial-intelligence-therapy-mental-health-care Bill Doesn't Protect NH from AI Harm, It Assures It — Union Leader (January 2026): https://www.unionleader.com/opinion/op-eds/nicholas-c-jacobson-michael-v-heinz-bill-doesnt-protect-nh-from-ai-harm-it-assures/article_b2c5bdf4-aca7-413a-a218-35e6a09de06f.html First Therapy Chatbot Trial Yields Mental Health Benefits — Dartmouth News (March 2025): https://home.dartmouth.edu/news/2025/03/first-therapy-chatbot-trial-yields-mental-health-benefits Connect with Dr. Nick Jacobson / Additional Resources: This Therapist Helped Clients Feel Better. It Was A.I. — New York Times (syndicated): https://onehealthsociety.com/this-therapist-helped-clients-feel-better-it-was-a-i/ How to Build a Therapeutic Chatbot — Psychiatric News: https://psychiatryonline.org/doi/10.1176/appi.pn.2025.09.9.23 AI, Neuroscience, and Data Are Fueling Personalized Mental Health Care — APA Monitor on Psychology: https://www.apa.org/monitor/2026/01-02/trends-personalized-mental-health-care Can 'AI Therapists' Help Save LGBTQ+ People? — Out Magazine: https://www.out.com/health/ai-therapy-for-queer-people AIM High Laboratory at Dartmouth: https://www.nicholasjacobson.com Connect with The Mental Health Evolution: Website: https://www.traumaspecialiststraining.com/mental-health-evolution-podcast Instagram: /thementalhealthevolution/ LinkedIn: /the-mental-health-evolution Facebook: /TheMentalHealthEvolution Music by Zach Harrison