
Millions of people are feeling apprehensive these days. The headlines are enough to make almost anyone feel anxious. People who are distressed may have a difficult time finding a therapist, however. There are too few, and consequently many are not taking new patients. Wait lists are long, often three to six months. Therapists who are accepting patients may not take insurance, and therapy can be pricey. A single session of gold-standard cognitive behavioral therapy can cost from $100 to $250. Could AI fill the therapy gap, offering psychotherapy online?
At The People’s Pharmacy, we strive to bring you up to date, rigorously researched insights and conversations about health, medicine, wellness and health policies and health systems. While these conversations intend to offer insight and perspective, the content is provided solely for informational and educational purposes. Please consult your healthcare provider before making any changes to your medical care or treatment.
Conversational agents like ChatGPT, Gemini or Claude have become nearly ubiquitous. People use them to help write resumes, pitch stories, create images for web or social media posts and make financial projections. Using these chatbots to give feedback as in therapy is surprisingly popular. But how well can AI fill the therapy gap, really? Today’s guest has been studying these interactions.
The conversational agents are also referred to as LLMs, for Large Language Models. It describes how they have been trained by scouring the internet. That allows them to predict the most likely word to come next in a sentence, or the probable next idea in a paragraph. They can’t actually think, but if something has been posted online, they have access to it. At this point, the technology has become so refined that chatbots easily pass the Turing test; it is difficult to reliably distinguish AI from human responses.
There are advantages to having “someone to talk to” any time, any place. Younger people in particular are digital natives and often feel more comfortable with technology than face-to-face with a human.
The training of AI agents as therapists, though, gives rise to some serious flaws. Because they are trained to elicit positive responses from humans to keep people engaged, they have a sycophancy bias. Have you noticed that most messages start by telling you your idea is great? That makes you feel good, and you are less likely to quit the conversation. But it isn’t necessarily how therapy is supposed to work. If people are not challenged when appropriate, they may get stuck and not make any progress toward healthier attitudes or behaviors. They may fail to develop the critical skill of stress tolerance. In addition, chatbots are disconnected from reality. This could become a serious problem if a user starts to become delusional or is in an acute crisis.
Dr. Brewer suggests that we would do well to think of anxiety as a habit. He credits a 1985 paper by an investigator named Tom Borkovec suggesting that worry drives anxiety rather than being a mere symptom of anxiety. Worrying leads people to dwell on possible catastrophic outcomes, which understandably makes them more anxious. Treating anxiety as a habit, especially by finding a better reward than the illusion of control offered by worrying, could be effective. Responding with curiosity and kindness might offer a better outcome. He has studied this possibility. When you treat anxiety as a habit that can be changed, anxiety scores decline by 67%. That is quite impressive.
One way to use AI effectively is to train conversational agents specifically to monitor for safety in other human-chatbot interactions. Given clear rules, they can do this very well. Also, chatbots could be used not so much as teaching assistants but as learning assistants. They could help people who are striving to change their anxiety habit. This might be integrated with video tutorials from an expert human, such as Dr. Brewer or one of his colleagues. They are testing this approach currently. Hopefully, it will prove more effective than the 20% response rate to SSRI medication for anxiety.
Jud Brewer, MD, PhD, is an internationally renowned addiction psychiatrist and neuroscientist. He is a professor in the School of Public Health and Medical School at Brown University. His 2016 TED Talk, “A Simple Way to Break a Bad Habit,” has been viewed more than 20 million times. He has trained Olympic athletes and coaches, government ministers, and business leaders. Dr. Brewer is the author of The Craving Mind: from cigarettes to smartphones to love, why we get hooked and how we can break bad habits, the New York Times best-seller, Unwinding Anxiety: New Science Shows How to Break the Cycles of Worry and Fear to Heal Your Mind, and his latest book is The Hunger Habit: Why We Eat When We’re Not Hungry and How to Stop.
You can find more information on the skills-based program for anxiety that Dr. Brewer developed at www.goingbeyondanxiety.com
[caption id="attachment_117804" align="alignnone" width="768"]
Judson Brewer, MD, PhD, Brown University, author of Unwinding Anxiety[/caption]
The People’s Pharmacy is reader supported. When you buy through links in this post, we may earn a small affiliate commission (at no cost to you).
Thank you for sharing your thoughts about one of our products! Please focus on the product performance and quality. Please read our Privacy Policy and Terms of Use for more information about the review process.