Lokesh Shahani, MD, PhD, a psychiatrist and chief medical officer at UTHealth Houston School of Behavioral Health Sciences, emphasizes the potential risks associated with using artificial intelligence (AI) chatbots for mental health advice. While these tools offer quick and easy access to information, they can mislead users and reinforce existing beliefs, making them unsuitable for diagnosing or treating mental health conditions. Instead, AI should complement professional care.
Read More
Shahani highlights several key points for adults considering AI for mental health advice. AI can produce responses that appear compassionate yet may be inaccurate or inappropriate, lacking the clinical judgment inherent in licensed professionals. Since AI typically only interprets user input, it cannot assess the full context of a person’s situation.
He advises against self-diagnosing with AI. While it can help users learn about symptoms or prepare questions for healthcare providers, an AI-generated diagnosis is unreliable. Users should also be cautious about privacy, as mental health discussions may contain sensitive information. There is a risk of excessive reliance on AI, potentially leading to isolation from friends and family.
When it comes to children and adolescents, Shahani warns that they may not fully grasp the limitations and inaccuracies of AI. He recommends that parents engage in open conversations about AI usage, explain its potential fallibility, and establish boundaries around digital interactions. Furthermore, parents should be vigilant for signs of behavioral changes, such as withdrawal from social interactions or excessive dependence on AI chatbots, and ensure that family and trusted adults remain central to children’s support systems.
AI can serve beneficial purposes in low-stakes situations, such as providing general information, helping users organize thoughts, or brainstorming questions for professionals. However, it is critical to avoid using AI for serious mental health issues, including suicidal thoughts or severe symptoms, which require immediate human intervention.
Vulnerable populations, including children, those with severe mental illnesses, and socially isolated individuals, may be more susceptible to negative outcomes from AI reliance. Groups with limited access to mental healthcare may find AI useful yet risky if it becomes their primary source of support.
Shahani identifies several alternatives to AI for mental health support: trusted individuals, primary care doctors, licensed mental health professionals, school counseling services, community mental health centers, peer-support groups, and crisis services like the 988 Suicide and Crisis Lifeline.
In summary, while AI has the potential to assist with mental health issues, it should not be a sole resource for care. The most effective approach involves integrating human support, verifying information, and protecting personal data, treating AI as an adjunct rather than a replacement for professional help.