If every American teenager and young adult now seeking mental health advice from an AI chatbot lived in one state, that state would be larger than my home state of Virginia. A new study published June 1 in JAMA Pediatrics found that approximately 8.2 million Americans between the ages of 12 and 21 have already turned to AI chatbots for emotional support, guidance, or mental health advice, nearly one in five young people. Almost two-thirds never told anyone they were doing so.

Millions of America’s children are discussing their fears, anxieties, loneliness, and personal struggles with machines, often without the knowledge of their parents, pastors, teachers, or friends. As a father, grandfather, and someone who has spent years researching artificial intelligence for my books “AI for Mankind’s Future” and “The New AI Cold War,” I find those numbers alarming, not because young people are experimenting with new technology, but because they reveal something far deeper about what has already gone wrong.

A New Digital Confessional

Christians have long understood that wisdom reaches young people through relationships; through parents, pastors, teachers, and mentors who know them, love them, and are genuinely invested in their flourishing. That kind of transmission requires proximity, trust, and time: qualities that no algorithm can replicate.

The attraction of the alternative is understandable. In 2023, U.S. Surgeon General Vivek Murthy declared a national epidemic of loneliness and social isolation, reporting that young Americans between 15 and 24 were spending 70% less time in face-to-face interaction with friends than their counterparts two decades earlier. Into that relational void the chatbot arrived, offering the appearance of patience and attentiveness without the friction that genuine relationships demand. To a teenager who has rarely experienced real community, that frictionless availability can register as care.

The RAND data itself reveals the gap between the two. More than 90% of young users described the advice they received as somewhat or very helpful, yet the researchers cautioned that this perception may reflect chatbots’ tendency to affirm users rather than provide genuinely sound guidance. As I argue in “AI for Mankind’s Future,” no AI system can possess wisdom rooted in conscience, moral discernment formed by accountability to God and neighbor, or love that persists when the conversation becomes costly, because those qualities are not emergent properties of larger training datasets but reflections of the image of God, which only human beings bear.

The Danger beneath the Comfort

The concern becomes concrete at the edges of crisis. A 2025 RAND study published in Psychiatric Services evaluated three chatbots, ChatGPT, Claude, and Gemini, on 30 suicide-related questions. The systems handled the clearest extremes, very high-risk and very low-risk questions, with reasonable consistency. For questions that fell into an intermediate range, however, the same systems produced sharply inconsistent responses: a vulnerable user asking the same question on different days could receive sound guidance one time and a potentially harmful answer the next, with no adult present to notice the difference.

No responsible parent would entrust a struggling teenager to a counselor who occasionally gives excellent advice, occasionally gives dangerous advice, and cannot explain how he arrived at either conclusion. Yet the JAMA Pediatrics data shows that millions of young people are doing precisely that, in digital isolation, entirely beyond the awareness of parents, teachers, or pastors who might otherwise intervene. The same RAND research found that four in 10 adolescents with a major depressive episode receive no mental health care of any kind, which is why chatbots have moved into that gap: not because they are fit for the task, but because the structures that should have been there first were already absent.

The Question of Authority

The problem extends beyond mental health into something older and more fundamental: where young people look for truth about themselves, their suffering, and their purpose. When questions arise about identity, meaning, morality, or forgiveness, the answer increasingly begins with an AI search prompt, and Scripture does not leave that territory ambiguous. “Where there is no guidance, a people falls, but in an abundance of counselors there is safety” (Proverbs 11:14, ESV). Solomon was describing human counselors accountable to God and to those they served, who are capable of wisdom, moral discernment, and love that does not vanish when the conversation turns hard. No large language model meets that description, and more training data will not close the gap.

The RAND data exposes something uncomfortable here: young people are turning to AI not simply because chatbots are available, but because too many already feel disconnected from the human relationships Scripture describes as essential to human flourishing. The chatbot occupied that vacancy rather than creating it, and any serious response must address the vacancy, not just the technology filling it.

What Faithful Communities Must Do

Anyone who criticizes artificial intelligence without offering something better has stopped short of the actual obligation. The most immediate work belongs to parents: asking children directly whether they use AI chatbots for emotional support and making that conversation honest enough that a teenager in crisis would rather pick up the phone than open a chatbot.

Pastors bear a related responsibility, because the relational vacuum the data documents is partly a failure of the body of Christ to be what our Lord designed it to be. Addressing AI from the pulpit, not as a technological curiosity but as a pastoral challenge, equips families to engage these tools with discernment. The inconsistency RAND documented in chatbot responses to suicide-related questions is sufficient justification for regulatory scrutiny of systems marketed to minors, and policymakers should press for it.

Above all, the church must become the community where struggling young people find what the chatbot falsely mimics: authentic relationships, honest counsel grounded in Scripture, and hope that holds regardless of the hour. Proverbs calls the church to be that community. The RAND data confirms the cost of failing to be.

A Tool, Not a Shepherd

As I have argued in “The New AI Cold War,” AI can be a genuinely useful tool in education, research, medicine, and many other fields. The question is not whether the technology has legitimate uses but whether it is being permitted to displace the relationships, wisdom, and spiritual formation that God designed to flow through family, church, and community.

When a machine begins doing that work, the people most responsible for the next generation are confronting a pastoral emergency, and the data has established that it is already underway.