Views: 0
Just a couple of years ago, talking about "therapy with artificial intelligence" sounded like science fiction. Today it's an everyday reality. Thousands of people open an AI app at three in the morning to tell it what they don't dare tell anyone else. And this fact, far from being anecdotal, is beginning to seriously worry mental health experts.
This is a phenomenon that is growing faster than we imagine. A recent study, conducted with several thousand participants in Europe and the United States, has analyzed not only how many people turn to AI to manage their emotions, but also exactly what they ask it. The researchers reviewed the "prompts" themselves—the phrases that people type—to understand the type of help they are seeking.
The most common profile is surprisingly specific: adults between 25 and 45 years old, with a medium-to-high income, who turn to AI primarily for anxiety and stress related to their partner, work, or friends.
This trend aligns with other recent studies. A study published in JAMA Network Open, with data from over 20,000 people in the United States, found that those who interact with chatbots for emotional support or other personal reasons are more likely to report symptoms of depression or anxiety. It's unclear whether AI causes this distress or if it's those already affected who are turning to it more frequently, but the association itself is a cause for concern.
Why do people prefer to talk to a machine? Many psychotherapists acknowledge that this happens daily in their practice, and they understand. Talking about our most intimate conflicts is difficult. We're afraid of others' judgment, shame, "what will they think of me?", etc.
AI, on the other hand, is always available, never gets tired, and doesn't judge. It's easy to contact, responds instantly, and that immediacy is very tempting when distress is at its peak. The problem is that this availability doesn't equate to real understanding. A response isn't always helpful.
Here's the nuance that many overlook. AI systems learn from enormous amounts of text available online, but they don't "feel" or understand human emotions from within. They have data, patterns, and statistics; they lack the deep roots of human behavior, those that can only be understood by sharing life, context, and connection with another person.
Various studies show that AIs can be surprisingly empathetic in their responses, but they also warn about their limitations, their ethical risks, and the urgent need for regulation. Form is not the same as substance.
A machine always answers. That's undeniable. But a response is not synonymous with real help. It may sound reasonable, and yet it doesn't address the true cause of our pain, because what is human heals with what is human.
We all go through moments of uncertainty, unease, of not knowing what to do with what we feel. That is profoundly human, and it's part of life. Learning to manage it is too.
But the answer to those feelings isn't in an algorithm. It lies within ourselves, and in the possibility of encountering another human being—a professional, a friend, a family member—capable of truly understanding us, with all the complexity that entails.
AI can provide support, information, and even relief at a specific moment. But replacing human connection with a conversation with a machine is a mistake that, in the long run, can be costly.
Perhaps the question isn't whether AI can help us, but whether we should let it take the place of someone who can.
EN 
ES