Annenberg Answers: AI Chatbots and Youth Mental Health

A new study found that nearly one in five American youth aged 12 to 21 reported using AI chatbots for mental health advice. What’s more, most young people who turn to this technology for mental health advice are using these tools in secret, the researchers say, noting that most of the survey participants “told no one that they used AI chatbots for this purpose.”

PIK Professor Desmond Upton Patton studies the impact social media has on young people’s well-being and mental health. He is the founding director of SAFELab, a transdisciplinary research lab dedicated to using innovative methods to promote joy and healing in online and offline spaces.

We spoke to him about why young people might be turning to AI for mental health support, how parents can navigate their youth’s AI use, and how chatbots can be used for good.

Why are young people turning to AI chatbots for mental health support and advice?

Young people are turning to AI chatbots because they are immediate, private, low-barrier, and always available. For a young person trying to make sense of stress, loneliness, grief, anxiety, identity, family conflict, or uncertainty, a chatbot can feel like a first place to put language around something that may be hard to say out loud.

I do not read this only as a technology story. I read it as a story about access, trust, and unmet need. My work has always started from the idea that young people’s digital lives are meaningful. They are not just scrolling. They are narrating what they are carrying. 

AI is becoming one more digital space where young people are looking for language, care, and connection.

Should parents be worried about their children using chatbots for mental health advice?

I would not start with worry. I would start with evidence and context. The issue is not whether AI is good or bad. The better question is: what kind of AI, used by what young person, for what purpose, under what conditions, and with what safeguards?

Parents should be especially attentive when young people use chatbots for high-stakes concerns like self-harm, suicidal thoughts, trauma, abuse, eating concerns, severe anxiety, depression, or intense loneliness. AI may help with reflection or language, but it should not replace a trained clinician, trusted adult, crisis service, or sustained human relationship. As a social worker, I come back to this: context is not extra. Context is the work.

How should parents, guardians, or mentors talk to young people about safe AI use?

The conversation should be age-appropriate and grounded in the actual young person in front of you. A conversation with a 9-year-old should not sound like a conversation with a 17-year-old. It should also be shaped by maturity, family dynamics, emotional vulnerability, disability, culture, and the kind of AI tool being used.

For young people with visible and invisible disabilities, AI chatbots may have real accessibility value when used carefully. A 2024 systematic review in npj Digital Medicine found that AI-assistive technologies show promise for supporting adaptive functioning among people with neurodevelopmental conditions in everyday environments, including more personalized and context-responsive supports. That matters for youth who may need help with communication, planning, emotional labeling, social rehearsal, or breaking complex tasks into smaller steps. The question is whether the tool is increasing agency, clarity, communication, and connection, or increasing isolation, dependency, and risk.

How can AI chatbots be used positively by young adults?

AI chatbots can be useful when they are bounded, transparent, and connected to human care. Young adults may use them for journaling, organizing thoughts, preparing for therapy, drafting language for a hard conversation, building a routine, naming emotions, or identifying questions for a doctor, counselor, mentor, or trusted adult.

For young adults with visible and invisible disabilities, this may include support for executive functioning, communication, accessible explanations, self-advocacy, and social practice. The npj Digital Medicine review is useful here because it does not overclaim; it shows promise for AI-assistive technologies in supporting adaptive functioning for people with neurodevelopmental conditions, while also making clear that the evidence is still developing and must be evaluated in real-world settings. For me, the strongest use cases are not about AI replacing people. They are about AI helping young people move toward more support, more clarity, more connection, and more care.