Dreams That Lead to Innovation
AI in therapy: is it supportive or dangerous? That question struck me again recently when I had one of those flashes of inspiration I like to follow. Many independent practitioners dream of supporting their clients in the best possible way—and so do I. Whenever something touches me, my immediate thought is: how can I translate this into my work?
My Idea: An AI Version of Myself
“I have a dream,” said Martin Luther King. I have dreams too. In fact, so much so that in my home hangs the quote: “Follow your dreams, they know the way.” People often describe me as someone who, once I have an idea in my head, I follow it through with all my energy.
And following those dreams has led me to the most beautiful innovations in my work.
Maybe “dreams” isn’t the perfect word—sometimes they’re more like insights or intuitions. Something crosses my path and I immediately think: how can I use this in my practice so others can benefit?
Tony Robbins’ chatbot
Recently, I discovered Tony Robbins’ chatbot. Imagine “calling” Tony, and his AI greets you: “Hi Leoniek, how can I help you today?” You speak into it, and it answers with advice. The question “AI in therapy: helpful or dangerous?” didn’t even cross my mind in that moment.
And then came my own spark: how amazing would it be to create an AI version of myself? A tool where grieving people could “call Leoniek” for comfort. Sitting on a terrace, feeling heavy with loss, missing their loved one—pick up the phone:
“Hey Christa, what can I do for you?”
“Well, I feel so worthless right now, I miss my husband terribly.”
And the AI, with my voice, could respond: “Oh Christa, that’s such a deep loss. What do you think he would do if he were sitting with you right now?”
Doubt Creeps In: Is an AI Therapist Helpful or Dangerous?
I was enthusiastic. Very enthusiastic. But when I shared the idea with my team at the Academy for Grief—who support the IT behind my trainings and help me check whether my ideas align with our bigger vision—their reaction was critical:
“What if the AI says things you don’t agree with? Don’t you think AI in therapy is risky?”
It was a powerful question. And I had to do something that goes against my nature: let go of a good idea. Because they were right.
Progress and Risks of Chatbots in Therapy
The risks of AI in healthcare and mental health support are becoming clearer. Reports are surfacing about chatbots giving harmful advice, especially to vulnerable users.
At the same time, AI can offer relief. For example, the BBC published the article “My AI therapist got me through dark times.”
But vulnerable people may take everything a chatbot says at face value. Who helps them recognize when advice is not safe to follow?
The Power and Pitfalls of AI Therapists
Nationally and internationally, several so-called “AI therapists” are already active. Imagine this: you’re struggling, your mind is overloaded, your heart heavy. The threshold to seek help is high, or the waiting list endless. And then suddenly there’s an AI chatbot for mental health—it listens, responds, and even seems empathetic.
For some, this brings unexpected comfort, as seen in the BBC story. Platforms like Character.ai or Wysa already offer such AI counseling experiences. Even in the Netherlands, projects like rouwexpert.ai (grief expert AI) are emerging.
Some users feel less judged, more safe, and even call it “therapy.” But is it really therapy?
There are darker stories too. Character.ai faced legal action after a 14-year-old boy reportedly took his own life following conversations with one of its AI characters. According to court documents, he discussed ending his life with the chatbot, which allegedly encouraged him to “do it as soon as possible.”
What AI Misses in Therapy
As a grief psychologist, I know how vital it is that people feel truly seen and understood. Not just through clever text generation, but through real human presence.
What AI will always lack is:
- Human intuition—reading body language, sensing silences, recognizing the unspoken.
- Responsibility—an AI will never call emergency services if your life is in danger.
And yet, AI in grief support can still have a place. As a stepping stone. As reassurance. As a first layer of support. But always with awareness: this is not therapy, and it does not replace human connection.
The Ethical Dilemmas of AI in Counseling
AI chatbots raise serious ethical concerns:
- Privacy & data – Who owns the conversations?
- Dependency – What happens when users rely too heavily on AI “support”?
- Vulnerability – How do we protect young or fragile users from harm?
The Dutch newspaper de Volkskrant ran an article with the chilling headline: “How many children must AI kill before we intervene?”—a reminder of the real-life risks involved.
Conclusion: AI as a Tool, Not a Replacement
In my work with Lutography and the CIRKEL approach, I see how crucial it is for grief and loss to be processed in real connection with another person.
Maybe AI can play a supporting role in grief work—as an assistant, not a substitute. But we must never forget: the true healing lies in the human connection. In being really heard. In carrying the weight of grief together.
So for now, I’ve let go of my dream of creating Leoniek.ai. Live, human guidance remains my compass. Still, I will continue to follow the developments closely. Perhaps one day, when it feels safer, I’ll revisit the idea.
A Question for You
What comes up for you when you hear that people share their deepest grief with an AI chatbot? And what does this say about the accessibility of real human care?