Promoting Mental Health — 10

Asystematic review of fifteen randomized trials of AI chatbots for mental health has shown a significant reduction of depression and psychological distress that was more pronounced than with multimodal, generative AI. There was no improvement noted for overall psychological well-being. Another review of ten studies, seven of which were randomized trials, conducted in women, concluded similarly that chatbots were effective in reducing anxiety, distress, and depression. In a related review of nineteen trials for chatbot impact on lifestyle, physical activity, healthy diet, and sleep health were promoted.

Results of an important study conducted in England with the Limbic AI chatbot in nearly 130,000 participants countered the belief that this technology would worsen inequalities. Self-referrals for mental health through Limbic were substantially increased among minority ethnic groups (29 percent increase) and nonbinary individuals (179 percent increase), “providing strong evidence that digital tools may help overcome the pervasive inequality in mental healthcare.”

While these findings with conversation chatbots and dCBT are encouraging, the concern about privacy cannot be emphasized enough. It is worth reviewing the privacy policy, security measures, and whether the companies sell information to third parties. This is also an issue that requires more regulatory oversight. Adopting policies that prevent breaches, sale, or any compromise of an individual’s data or engagement is essential.

The need for oversight extends to safety, with notable precedents, such as a chatbot that was designed to help with eating disorders but instead provided harmful advice to make matters worse, necessitating its removal by the National Eating Disorders Association.

What about the next generation of chatbots that use generative AI? Large language models have been assessed for their ability to track people’s mental states, and they compared favorably to human performance for a comprehensive battery of tests. Generative AI, integrated with therapeutic sessions, has specific potential to improve behavioral health by recognizing, expressing, and managing emotions. However, the important risks of harm from generative AI wellness apps, which do not go through regulatory oversight and can be used in unintended ways, have been emphasized. It’s still too early to determine the net benefit of generative AI for promoting mental health, but there’s considerable optimism, especially for reduction of anxiety. Not as a substitute for a therapist but as an adjunctive means of support.

LAST RESORTS

So the problem remains. With relatively fewer trained professionals to provide mental health care, how will we help older people feel less anxious, stressed, and depressed? Medications should be considered as the absolute last resort. Again, many lifestyle factors work as well or better than medications. That’s for the common mental health issues, not serious ones like bipolar or major depression disorder, schizophrenia, or post-traumatic stress disorder, which are beyond the scope of our principal focus of expanding health span.