AI chatbots may already be the largest providers of mental health services in the United States, raising big questions about safety, effectiveness, and oversight. In this episode of the Being Well Podcast, Dr. Rick and Forrest are joined by Dr. Nick Jacobson, to explore the risks and opportunities of AI therapy: Can a chatbot be good at therapy? Will it replace human therapists? What about AI psychosis? How should we think about privacy, bias, and regulation? Is this a silver bullet for mental health access, or are we just opening a new can of worms?
About our Guest: Nick is associate professor of biomedical data science, psychiatry, and computer science at Dartmouth, and directs the AI and Mental Health Laboratory there. He’s also the developer of Therabot, a generative AI therapy chatbot that predates ChatGPT, and he’s one of the first researchers to run a clinical trial on AI therapy.
Key Topics
- 02:35: Is AI going to replace human therapists?
- 05:00: Risks of using ChatGPT as your therapist, and general vs. therapy-specific AI
- 14:30: What should people be worried about?
- 19:14: Is AI good at therapy?
- 29:58: Bias, values, and “who’s watching the watchers”
- 39:17: Is there something unique about a human therapist?
- 52:21: Oversight and the self-driving car analogy
- 1:00:51: Personhood, consciousness, and risks of anthropomorphizing AI
- 1:11:00: Recap
Forrest is now writing on Substack, check out his work there.
