AI Is Not Therapy and Why Over-Reliance on Algorithms Can Mislead Your Mental Health

Increasingly people are turning to AI to help with tasks to expedite productivity and free up time.For tasks such as scheduling, editing, organization, and related, AI has proven to be a great tool. I use it to assist with capturing broad concepts from data and/or written content that I have produced. Sometimes it is extremely helpful – and other times not so much, as it misses the point or will have erroneous conclusions.

A growing trend is that people are turning to AI for suggestions about their personal lives. This has led to an uptick of clients bringing in an AI-summary about their lives. Some is well summarized and other “conclusions” are generic or erroneous, in comparison to completing a clinical interview with the same client.

Human interaction is critical for identifying subtle nuances that drive behavior at an unconscious level. AI is not capable (yet) of reading these humanly generated cues from what I call “back brain” that will run the show of our behavior/choices until we make those inner-workings conscious.

Since the advent of AI, I have remained a slow-adopter for a good reason. AI language model programs are derived from what it is “fed” and the applied algorithms. The human condition (feeling-thinking-doing) is not algorithmic, although there are some patterns that do emerge. Rather, our individuality is formed from our earliest experiences, including during fetal development. Every experience is un/consciously registered and affects deep brain structures, as well as influences our DNA (through the epigenome). And this is what allows each of us to be a unique human being – and why as many people who experience the same event can have a diverse response to it.

Turning to AI for “therapy” loses the process of discernment and has great potential to shift the trajectory of thoughts into an algorithmic echo-chamber. Already, dangerous outcomes have resulted in suicides, progressive hopelessness/depression, and other deleterious situations from a mental health perspective.

“AI can be a wonderful tool for organizing our lives, but it cannot understand the human story. Your emotions, your history, your patterns of thought and behavior are shaped long before an algorithm ever meets you. What we uncover together in a clinical setting comes from subtle cues, unconscious processes, and lived experience—things no machine can read. When people turn to AI for personal insight, they often get information that feels certain but is not truly connected to who they are. My goal is to help patients return to thoughtful discernment, where clarity comes from genuine human understanding, not an echo of prediction. AI may summarize data, but it cannot know you.” — Dr. Wiet

Very recently, I have encouraged specific clients to withhold from “consulting” AI for personal assessments for the next month. One gladly accepted the challenge, stating the relief from not doing so. Upon further inquiry, “relief” was from the secondary anxiety they experienced from the responses – questioning if what was “spit out” was real or not. Another client noted hesitancy with this challenge; upon further examination, that hesitancy was due to and addiction-related pattern of behavior – not unlike most of us with our respective devices carried in hand, pocket or (now) eyewear.

My point is for anyone using AI for personal assessment to be aware that AI-generated feedback is NOT therapy. Rather, it is a sophisticated algorithm response and is NOT individualized for a unique human experience.

Key Takeaways

• AI is useful—but limited. While AI helps with productivity tasks like scheduling, editing, and organizing information, it often misses nuance and can draw incorrect conclusions.

• AI “life summaries” are not clinical assessments. Many clients now bring AI-generated summaries into sessions; some are accurate, but many oversimplify or misinterpret important psychological details.

• Human interaction is irreplaceable. Subtle emotional cues, unconscious patterns, and “back-brain” drivers of behavior cannot be detected or interpreted by AI.

• Human experience is not algorithmic. Our individuality is shaped by early life experiences, fetal development, and epigenetic influences—complex factors AI cannot personalize or fully understand.

• AI “therapy” can be harmful. Reliance on AI for personal guidance may create an echo-chamber, increasing anxiety, confusion, hopelessness, or worsening mental-health symptoms.

• Clients may feel relief when stepping back. Some clients who paused AI “self-diagnosis” noticed reduced anxiety, showing that AI can unintentionally fuel distress.

• AI reliance can mimic addictive patterns. Hesitancy to stop using AI for personal insights may reflect compulsive or habit-driven behavior similar to dependence on digital devices.

• AI is not therapy. AI outputs are algorithmic, generalized, and not tailored to a person’s unique psychological makeup or lived experience.

• Awareness is essential. Individuals should be cautious when using AI for personal assessments and understand its limitations.

• Upcoming topics will explore AI’s psychological impact. Future posts:

  1. From the Couch: What People Are Saying About AI

  2. AI at the Door and the Fear that Follows (How to Stay in Peace)

Upcoming Posts:

1. From the Couch: What People Are Saying about AI

2. AI at the Door and Fear that Follows (How to Stay in Peace)