Writer: Kyle Heim Your chatbot therapist will see you now. While they might still be years away from providing traditional therapy, artificially intelligent chatbots could play a pivotal role in mental health care in the future, according to Ryan Crane, executive director of the National Alliance on Mental Illness (NAMI) in Iowa. Crane, along with Joshua Larson, generative AI developer at Hilos and co-founder of Numinous Games; Jon Lensing, CEO of OpenLoop; and Ciara Lewis, director of Des Moines University’s Student Counseling Center, explained some of the opportunities and challenges of AI’s expansion in mental health care during dsm magazine’s Lifting the Veil panel discussion in June 2024. Crane said it will be important to train the software, which uses large language models, so it doesn’t share clinically or medically inaccurate information or react inappropriately to a patient’s deeply personal disclosure. “This is not six or 12 months out; this is far out,” Crane said of AI therapy. In the meantime, he added, people are already experimenting with various AI chat programs. “Even though it sounds kind of dystopian and kind of weird and kind of uncomfortable for us, currently, I do see a future where some really highly refined AI-type chatbots are providing some kind of therapy,” Crane said. “And the reason that I like this or the reason that doesn’t freak me out and make me go Orwellian, is because of access. A lot of young people are way more comfortable online than we are.” TRAINING AND RISK ASSESSMENT One concern about AI tools is their ability to assess whether people might harm themselves or others. “Some of the bots that exist currently don’t even actually prompt a user to seek help, even if the person directly names a word like ‘suicide’ or ‘self-harm,’” Lewis said. “I think there’s a huge area of risk there and additional training that’s going to be so important for this technology to catch up to.” Lewis said she has not found a tool that is doing any kind of assessment of a patient’s risk of committing a homicide, which she said is incredibly important. “As a licensed professional, I would have a duty to warn if something like that should occur, alerting law enforcement, of course, but then also warning the intended potential victim,” Lewis said. “So where does that duty to warn fall with something like AI? I think there are some really important, safety-related and risk assessment-related areas where that training needs to grow a bit more.” CREATIVE OUTLETS Larson said in recent years, he struggled with depression for the first time in his life. “I’m naturally a content person, so it was very unusual,” he said. AI helped him reconnect with his creative side. “I got really into AI image generation, and that kind of helped me almost as an art therapy practice to just be creative again because I’m a very creative person,” he said. “So I was using those tools to help reignite my creative drive, in addition to working with a therapist and exercising more and other tools.” People use AI to generate images for a variety of creative projects, including social media graphics, personalized marketing materials and website visuals. Joshua Larson, generative AI developer at Hilos and co-founder of Numinous Games, uses AI to make images like this one to spark creativity. PATIENT-FOCUSED TECHNOLOGY New technologies in health care, including telehealth services, have increased access for patients and have improved convenience. “I think as health care providers, we’ve always said, [the] patient comes first. But in traditional health care, oftentimes, it’s like, ‘Yes, [the] patient is what we’re trying to work toward, but you’re going to come to our site, you’re going to go to this place, you’re going to be on our schedule, our time,’” Lensing said. Now, patients have more options for when, where and how they receive care. “It’s user choice, it’s patient choice at the end of the day, and I think technology’s only further empowering that. It’s putting the patient’s care back in their control,” Lensing said. “I’m really excited about the next five to 10 years in the health care space. I think the pandemic [ushered] in some miraculous innovation, and I’m excited to watch it continue to unfold.” UNDERSTANDING HOW AI WORKS While attending a futurist conference in California in June, Larson met someone who is developing software to better understand how AI works. “Up until now,” he said, “a lot of people think of it as a black box, and that’s concerning, especially with something as high stakes as mental health.” Larson said that understanding exactly how AI works will allow for the creation of safer guardrails and, as a result, a safer experience for users. POTENTIAL RISKS If organizations and companies start to rely too heavily on technology-based applications, it could eventually limit overall access to mental health services, Lewis said. She believes overdependence on AI mental health delivery could shrink the mental health workforce. “Could that actually reduce in-person access even more for the people that really do benefit from that one-on-one, in-person connection?” she said. FROM AI TO ZOOM Larson said the evolution of technology within mental health care in the past five years started with therapy sessions on Zoom when the COVID-19 pandemic disrupted everyday life. Soon enough, health care providers were allowed to send prescriptions directly to a patient’s local pharmacy. Now some AI companions check in on patients on a daily basis, whereas traditional therapists might be available for just once a week. “These AI companions are checking in every few hours,” Larson said. “‘Hey, how are you feeling right now? Did this work out for you? Did you try this new tactic that you discussed in your last session?’ That’s pulling down stuff from the care plans of the therapist and then kind of bringing that back into the daily living of these patients.”