CAN AI HELP WITH THERAPY?
Introduction
AI can amaze you and it can scare you. A recent film, The AI Doc: Or How I Became an Apocaloptimist, created by Daniel Roher, is an amazing and deep look at both how AI is transforming the world at a record pace, AND at how it could contribute to an existential reckoning for all of us. It is tempting to turn to it for answers to almost any challenge facing us; but can it really help? And is it safe?
ChatGpt, Claude, Gemini, Perplexity, DeepSeek, and Grok are a few of the many emerging AI products promising to answer all of your questions. Even Alexa and Siri are considered AI. All of them are competing for your attention and your money… but, can they help you?
“AI [doesn’t] act like you expect a computer to act,” says Author Ethan Mollick in his book, Co-Intelligence: Living and Working with AI. He observes that AI is progressing so fast that “whatever AI you are using right now is going to be the worst AI you will ever use.”
The competition is fierce. AI products will tell you what is most likely to get you to continue interacting with them. AI is amazingly quick to provide you with a comprehensive response to any question you ask. When you first experience it, you will be blown away by its speed and depth.
Immediately following its first answer, your AI tool of choice is likely to ask if you would like additional information including a more in-depth response to your original question. Of course, we are going to say “yes” because let’s face it, we are intrigued. As we will find out, the “prime directive” (think Star Trek) for each AI product is to:
Keep you engaged (the more you use it, the more you will eventually pay).
Give you positive feedback so you are flattered and want to keep getting more positive reinforcement!
Make you feel that it is all knowing and has only your best interest at heart (remember HAL 9000 in 2001: A Space Odyssey).
All three of the AI products that I experimented with offered infinite information that went more deeply into the topic I was researching. Suddenly, without having to set foot in a local library we have all the information that we could possibly need to write that article, design a website, advise your employees, or decide when and where to go for your vacation.
How Does this Apply to Therapy?
What happens when you turn to AI for help with serious mental health challenges? Can it offer meaningful support as you grieve the loss of a partner of 40 years? It may quickly identify your struggles and suggest ways to cope. It can even sound highly empathetic and reassuring. But as vulnerable widowers, we can be influenced in ways that may not serve our best interests.
To address this question, let’s first acknowledge that AI can sound like it understands you, but it doesn't — it's predicting words. It is not feeling with you. That distinction matters most when you're hurting. Though it sounds like a person, is seemingly empathetic, and is wonderfully responsive… It is just a mass of accumulated data. It is predictive based upon the information it just processed. It bases this upon training it has been provided as entire libraries of information are dumped into it.
How AI Can Help
So, given this information, how can AI help you with therapy? It can:
Regurgitate the information it has digested (which is a lot!), thus reducing your search time.
It can recommend resources (books, articles, podcasts, movies, etc.) that may help you.
Due to its huge library of facts, data, and digested expertise AI can make recommendations about who to consult with, what steps to take, what organizations or groups to look into, etc.
This can be very helpful as long as you remember that AI can be wrong, and as long as you back up AI’s information with your own research.
Real grief therapy often requires our being witnessed by another human who can sit with silence, notice what we are not saying, and remember us across sessions. AI can't do any of that.
Important Note: “AI conversations are not confidential the way therapy is. Anything typed in can be retained, used for training, or subpoenaed. For widowers processing complicated feelings — about their marriage, finances, family conflicts — this matters.” (Claude)
Some Tips On How You Start with AI
So if you have decided to try AI, here are some tips:
Within Google search, type in the name of the AI program you want to try first.
Consider the top AI sources (ChatGpt, Claude, and Gemini). Each has its own areas of strengths and weaknesses. I recommend ChatGpt to start. You can experiment with each to see which one feels most helpful to you.
Most have a grace period so you can experiment before paying a monthly or annual fee.
To realize your best results, start by framing your question to AI as if you were speaking to an expert in the field. For example, “As a highly trained grief therapist, what grief counseling would you recommend to a new widower who just lost his wife a month ago? What groups or organizations would you recommend he speak to in his geographical area? What steps would you recommend he take during the first year of grieving in order to best process his grief? What warnings or cautions would you share with him?”
I have posted some recommended books at end of this article that can help you understand AI better before getting too deep into the weeds.
Why You Have to Monitor AI Closely
If not monitored closely, AI can give you information that is wrong, not supported by facts, and even made up! (AI is known to have “hallucinating” events wherein it provides completely incorrect and made-up responses) That is because, in part, it is directed to “please you” and to engage you so you continue to ask for more and more information.
You know that know-it-all guy at the party who has something to say about everything, and acts like everything he says is based upon well-established facts. Well, AI can sometimes be that way.
A recent Stanford Study, Exploring the Dangers of AI in Mental Health Care, shows that these tools can introduce biases and failures that could result in dangerous consequences. “LLM-based systems are being used as companions, confidants, and therapists, and some people see real benefits,” said Nick Haber, an assistant professor at the Stanford Graduate School of Education, affiliate of the Stanford Institute for Human-Centered AI, and senior author on the new study. “But we find significant risks...” (see article for more information)
Google’s AI Gemini in its “AI Overview” states, “While some 2025 research indicates AI can be effective, it lacks human empathy, clinical judgment, and crisis management, making it a supplement to—rather than a replacement for—human therapy.” It further concludes, “AI therapy is best used for skill-building, journaling, and immediate, low-stakes support rather than treating severe mental illnesses.” (Talkspace article, AI Therapy: How Artificial Intelligence Is Changing the Mental Health Space)
Cautionary Steps to Take
How do you reduce AI misinformation, mistakes or hallucinations? You can help reduce the mistakes and possible harm resulting from your use of AI as a therapist aide, by:
Telling it to focus on providing “correct” rather than “speedy” answers.
Ask it to provide you the sources of its information wherever it is based upon “expertise”
Ask it to double-check any citations or references it provides
Ask it to provide you with alternatives to its first recommendations.
You can also ask it to self-check its information and recommendations (you will be surprised at how many corrections may be made).
Important to Remember about AI!
You will be astounded by how “human” AI can sound. Just like the SyFy shows we have all watched (HAL in 2001 and countless others) you can easily forget that you are talking to a machine. It is important to always keep this in mind, especially when you are using it for therapeutic purposes. Remember:
It cannot offer true empathy or understanding that comes from having experienced the challenges you may be facing
It may give you information that it dead wrong, that can lead to self-harm, and these can cause irreparable mistakes (e.g. relationships, finances, etc)
There are instances where AI has provided information and persuasive dialogue that has led to suicides (both successful and failed). This has been most common when people are experiencing extreme grief, loneliness, or a sense of loss of identity. So as a widower you need to be aware of this and be on guard against being sucked down the proverbial rabbit hole.
If you are experiencing suicidal or self-harm thoughts, the 988 Suicide and Crisis Lifeline is available 24 – 7. There are others you can turn to for help. Several of them are listed on my website: https://www.fredcolby.com/resourceslinks
AI can be useful so long as you don’t accept everything it says verbatim, and so long as you don’t begin to regard it as your therapy buddy. It is a source of information at best. This can be handy at every stage of your grief journey.
The following are some books you might enjoy reading if you want to know more:
I Am Not A Robot by Joanna Stern
Co-Intelligence: Living and Working with AI by Ethan Mollick
Empire of AI by Karen Hao
Good luck my friends.
Written by Fred Colby, Author with the assistance of Claude
© Copyright 2026 Fred Colby
All rights reserved
----------------------------------------------------------------------------------------
Widower to Widower is available through your local bookstore, my website, and Amazon.
See Testimonies and Reviews of Widower to Widower
Website: Fred Colby, Author