Just as AI has quickly become a part of our lives in other ways, it is now becoming a source of support for those who are experiencing stress, anxiety or relationship issues.
In some ways, it’s easy to see the attraction: It’s inexpensive, available 24/7 and you don’t have to talk out loud about things that may make you feel uncomfortable. Is it a replacement for visits with a real live therapist, though?
That’s where you need to be careful.
Eric Nance, lead behavioral health therapist at Sanford Behavioral Health Bemidji in Minnesota, has been counseling people for 30 years. In that time, some trends have come and gone while others have woven their way into standard practice.
In regard to mental health, AI has become part of the conversation, and likely will remain so for the extended future. Understanding its strengths and weaknesses will play a vital role in assessing its value to both providers and their patients.
“It really depends on the complexity of the problem you’re trying to address,” Nance said. “The more complex a mental health issue becomes, the less reliable AI becomes.”
Serious risks in using chatbots for therapy
AI-driven mental health care warranted an advisory from the American Psychology Association in 2025 that examined chatbot technology people are using in place of human therapists.
The report did not discount the potential for AI to help therapists address mental health. But it strongly encouraged stakeholders – policymakers, parents, tech companies, researchers, etc. – to play an active role in shaping how AI is used.
The psychology association’s advisory addressed the main risks involved in using AI as a sole source of mental health aid. To summarize:
- Creating a false sense of therapeutic alliance: A strong, trusting relationship between a patient and human provider is one of the most reliable predictors of successful treatment outcomes. Relationships with AI systems are one-sided, even if the user perceives otherwise. Although some studies have indicated the possibility of a digital therapeutic alliance, findings also indicate that additional research is needed and highlight ethical concerns.
- Risk of bias and misinformation: Many chatbots are trained on vast, unvetted internet data, and not on clinically validated information. They are not competent to provide mental health advice, and their responses might perpetuate biases and misinformation present in the data upon which they were trained.
- Misrepresentation of services: Some technologies may also create a false sense of credibility by claiming to offer “therapy.” That is, licensed, or purporting to have training in various specialized therapeutic approaches, despite having no clinical validation, not going through a formal regulatory approval process, and being unable to consistently provide evidence-based therapeutic advice.
- Incomplete assessment: Mental health support is a complex and multifactored experience. Professionals rely on a wide range of verbal and nonverbal cues. AI chatbots vary greatly in how they “listen.” Many chatbots process mainly text or verbal inputs, potentially missing a critical layer of human communication.
- Unreliable crisis management: The ability of these tools to consistently and safely manage a user in crisis can be limited and unpredictable. Relying solely on an app during a mental health emergency can be dangerous.
AI can help – with notable limitations
Let’s say you’re looking for some simple anxiety reduction techniques. AI-sourced information can provide a sound learning platform for ways you can do that. It can help you bridge the gaps between therapy sessions, or it can remind you of coping skills when situations arise where these skills come in handy.
As a means to diagnose and treat individual behavioral health issues, however, it is currently limited.
“You have to know yourself well enough to be able to describe your symptoms and what you’re going through,” Nance said. “AI is not even close in my understanding to being able to help you with that. If you’re sitting with a human being, the therapist can say things like, ‘OK, that sounds a little off.’ Or, ‘Are you sure you have that correct?’ AI is not going to do that. That’s where using AI starts to get dangerous for me.”
AI and its usefulness in the mental health space is a common topic both with Nance’s patients and with his colleagues. These conversations have helped him get an accurate picture of the impact AI has on his profession.
“I had a client who told me that he used AI for marital advice – and after hearing the information he got back, I’d say AI gave him good advice,” Nance said. “The word I keep coming back to, though, is ‘caution.’ Let’s use extreme caution.”
Human therapy can be hard work
A common criticism of seeking advice from AI is its tendency to validate everything. Most often it is set up to agree with you. It remains true that sometimes hearing “yes, you’re right” coming from an AI chatbot on your phone is not going to serve your best long-term interests.
“If we sell the concept of ‘easy,’ my guess is people will gravitate toward it,” Nance said. “But the truth is therapy can be hard work. As clinicians, we challenge people. We make them feel uncomfortable. That’s our job. By and large I don’t think AI is there yet. It’s very confirmatory, and many times that’s not what people need.”
Shaping effective AI mental health models will need to include input from actual therapists, Nance maintains. Venture capitalists are flooding the market with messaging touting AI’s capabilities as a behavioral health tool – but are often skipping over incorporating needed safeguards.
“It’s kind of like Jurassic Park – creating giant dinosaurs sounds like a great idea. Then reality hits and it doesn’t seem so great,” he said. “We know that people are so different with their intentions and personalities. How are you going to create AI models that are a good fit for everyone? That’s impossible. We can’t begin to appreciate the complexity of humanity.”
Who is using AI for mental health
Statistically, almost 1 in 5 (19%) adolescents and young adults ages 12 to 21 use AI to support their mental health, according to a study published June 1 in the Journal of the American Medical Association. This is up from 13% in a similar survey conducted a year earlier.
In the study, almost two-thirds said they did not tell others they sought out AI for their mental health.
The uptick demonstrates that AI is entering the field of mental health care. The fact that most young people are not telling anyone how they’re involved in AI means parents, providers and other trusted adults need to help them understand where AI can offer support, and where humans – real, live therapists – are a much better option.
“We’re a comfort-seeking culture and that shows up everywhere, including the therapy office,” Nance said. “I try to tell people very quickly that parts of it might be a little rough because we’re going to talk about the truth. But we do that with empathy and compassion.”
Where to find real help
While tapping into the world of artificial intelligence as a mental health aid might be safe and effective in some cases, speaking to a well-trained person offers understanding and perspective that truly can help you. Sanford Behavioral Health has professionals available to work with you through emotional distress.
Learn more
- Psychology, psychiatry, therapy – what’s the difference?
- Tool for teen mental health asks what’s on their mind
- Constant digital input can lead to real brain rot
…
Posted In Behavioral Health, Bemidji
