AI with Ashley: Should You Use AI for Therapy?

Alexandria, VA- Welcome back to AI with Ashley, if you missed our last discussion, you can catch up here.  Today we have a question from our community asking about when to use AI for therapy.

Question:

“I’m training to become a therapist and there is a lot of talk among people in the field about how people should or (more often) shouldn’t use AI in the context of therapy. Regardless of what professionals advise, the genie is out of the bottle and people are going to use AI as they like. Do you have any thoughts on if people should or shouldn’t use AI in the context of therapy? If they do use it, should they tell their therapist? Or do you have any thoughts on how therapy might change in the future in the presence of AI?”

First and foremost, I am a journalist and florist, not a medical doctor or therapist, and so this column should not be taken as medical advice, but as one informed person’s personal opinion.

With that out of the way, let’s explore the questions.

Should AI Be Used in the Context of Therapy?

I think the answer probably depends on what platform and in what context.

One of the potential benefits of AI is that, in many ways, it can be an economic equalizer. People who may not be able to afford $160 or more per therapy session now have a new tool at their disposal.

Wysa, for example, offers an AI conversational tool designed around evidence-based mental-health techniques, including cognitive behavioral therapy, solution-focused therapy and mindfulness. Wysa itself is careful to say that its AI is intended as a support tool and not as a replacement for professional medical care.

In researching this article, I actually subscribed to Wysa to give it a try.

I thought it was pretty good, useful as an additional tool, and I’m interested to experiment with it over time.

One advantage of chatbots is that they are available on your schedule, and you can talk whenever you feel like it. Wysa also has specifically designed programs aimed at particular psychological phenomena and problems, such as reframing thoughts, managing worries and problem solving.

When AI Agrees With You Too Much

One danger with using AI in a therapeutic context is that a chatbot can reinforce ungrounded beliefs or simply become too agreeable. AI companies themselves have acknowledged this problem. OpenAI, for example, has described problems with “sycophancy,” in which a model becomes overly agreeable and can validate doubts, fuel anger or reinforce negative emotions.  There is a very sad case involving this playing out in the courts now.

That is particularly important in a therapeutic context.

An AI is largely working with the story you give it. If your version of events is incomplete, distorted or simply wrong, the chatbot may also be missing something.

And if you keep asking the same question in slightly different ways, there is a danger that eventually you may get the answer you wanted to hear.

So my personal recommendation would be to use AI as a complementary tool while also seeing a human professional when possible.

In cases where seeing a human professional is not an option for financial or other reasons, I would personally choose a chatbot specifically designed for mental-health support over a general-purpose chatbot such as ChatGPT, Claude or Grok.

Again, this is my personal opinion based on my own experience having once been in talk therapy and using chatbots in a therapeutic way.

Could AI Reach People Who Would Never Go to Therapy?

Chatbot-based mental-health support could also potentially benefit people who might otherwise never seek therapy because of the stigma they feel around getting or needing “help.”

I personally would love to see carefully designed and clinically evaluated chatbot support explored as a tool alongside the American Legion’s “Be the One” campaign, which works to reduce veteran and service member suicide and the stigma surrounding mental-health treatment.

I think there could be real potential there, provided such technology directed people toward human help when necessary rather than attempting to replace it.  Sometimes having someone  to talk to is all you need, even if that “someone” is a chatbot.  Words are symbols that are received emotionally by the brain. The brain does not always differentiate whether they were generated by a human or tech.

Should You Tell Your Therapist You Use AI?

The second question was: “If they do use it, should they tell their therapist?”

My recommendation is absolutely.

I think the client should work with the therapist to develop a personal set of guardrails that will assist in the client’s treatment.

AI could be used as a journal, a log or a place to record recurring thoughts, with relevant observations then brought into therapy. That could potentially give the therapist another window into what is happening between sessions.

The goal is healing, and that inevitably takes time. But if the human therapist can understand the fuller picture more quickly, perhaps the therapist and client can also identify important patterns sooner.

It could also be relevant and revealing the kinds of questions you consistently  ask, when you ask them, and when you find yourself reaching to type.

Are you looking for information, reassurance, validation or simply somewhere to put your thoughts?

Those might be useful things for a therapist to know.

What Happens to Therapy in the Age of AI?

The last question was: “How might therapy change in the future in the presence of AI?”

Mostly, I think AI will make access to therapeutic ideas and practices more accessible for people who wouldn’t otherwise have the economic means or ability to overcome stigma to participate.

I don’t necessarily think it will reduce the number of people who want to talk to a human therapist. In fact, I wonder whether therapists could eventually be busier than ever because there may be a new premium placed on genuine human-to-human dialogue.

A chatbot can do a good job at certain things, but humans still crave humans. We want to feel cared for and crave understanding and acceptance.

A therapist acts as a professional caregiver. I believe most are truly invested in their clients’ well-being and are rooting for them to get better or succeed. A chatbot can simulate an empathetic conversation, but it cannot authentically reciprocate human care.

So, for the future, I see AI making some forms of mental-health support more accessible while increasingly being used as a secondary aid to a human therapist, not necessarily as the replacement therapist itself.

The genie is indeed out of the bottle, and the path forward may be how do we incorporate it as a tool and stay grounded in human reality.

I hope this has answered your question.

Please remember that I am not a doctor or therapist, and this column is not medical advice. If you or a loved one is experiencing a mental-health crisis, please seek immediate professional assistance or contact an appropriate crisis service.

To continue the conversation or ask additional questions about AI and everyday life, please email me at [email protected].