AI Is Not Replacing Therapy. It Is Replacing Confusion
Over the past year, a growing body of research has documented something many parents and educators are only beginning to notice: adolescents and young adults are increasingly using artificial intelligence tools to process emotional distress, confusion, and relational conflict.
The Rand Corporation recently published this:
A survey by Common Sense Media published in July found that 72 percent (PDF) of American teenagers said they had used A.I. chatbots as companions. Nearly one-eighth had sought “emotional or mental health support” from them, a share that, if scaled to the U.S. population, would equal 5.2 million adolescents. In another recent study by Stanford researchers, almost a quarter of student users of Replika, an A.I. chatbot designed for companionship, reported turning to it for mental health support.
At first glance, this trend is often described as “AI replacing therapy.” That framing misses what is actually happening.
A more accurate description is this: many teens are using AI as a clarity engine, not a therapist.
Rather than seeking diagnosis or treatment, young people are increasingly using AI to examine confusing communication. They upload email threads, direct messages, text message threads, or social media comment exchanges and ask a simple question: What is actually happening here? The appeal is not authority, but legibility. AI allows an entire interaction to be viewed at once rather than reconstructed from memory or selective retelling. Patterns of language, repetition, escalation, and tone become easier to see when placed in a single analytic frame.
This distinction matters. In traditional therapeutic or mentoring contexts, much of the work necessarily depends on recollection and interpretation. AI introduces a supplementary tool that allows the content itself to be examined directly. Used in this limited way, AI does not tell someone what to think or feel, nor does it replace judgment or relationship. It offers a clearer starting point for understanding what was communicated and how it may have been received.
For years, I’ve drawn on Karen Horney’s work to help students think more clearly about why people say and do what they do, especially in emotionally charged or ambiguous situations. Horney argued that anxiety often arises not from internal pathology, but from relational confusion and unresolved tension. When messages are contradictory, unclear, or freighted with unspoken expectations, anxiety follows. Clarity, by contrast, reduces unnecessary psychological pressure.
Seen in this light, teen use of AI for mental health begins to make sense. Adolescents today are navigating an environment saturated with rapid communication, moralized language, and constant evaluation. When content is intense or confusing, having a tool that can slow the moment down and make patterns visible without judgment is not escapism. It is orientation.
This is especially relevant for teens and young adults active on social media, who often receive conflicting messages about identity, responsibility, emotional expression, and authority. Before any productive conversation can happen, they need to understand what they are responding to in the first place. AI is increasingly filling that preliminary role.
What follows is a review of several recent peer-reviewed studies on AI and mental health, interpreted through a Horneyan framework. Together, they suggest that this is not a passing technological curiosity, but a response to deeper conditions of psychological and relational confusion.
Karen Horney would not ask whether AI is good or bad. She would ask what kind of self it is helping to form. When AI is used to clarify experience so that young people can engage the world more honestly, responsibly, and relationally, it can support growth toward what Horney called the real self. When it replaces responsibility, relationship, or moral agency, it reinforces neurosis.
The difference matters. Parents and educators play a crucial role in helping young people understand it.
