Is AI Your Therapist Now?
Is AI Your Therapist Now? Chatbots, and the Risk of AI Psychosis
It's 3 a.m. You can't sleep, you don't want to bother anyone, and therapy still feels like 3aib. So you open ChatGPT and start typing.
In this episode of Said in Therapy, I talk honestly about why so many of us turn to AI for emotional support, what it can genuinely help with, and where it goes wrong. We cover:
- Why a private chatbot feels safer than therapy in our communities
- What AI does well, and why "too nice" can be dangerous
- What "AI psychosis" is, and the warning signs to watch for
- Jinn, psychosis, or both? How faith and medical care can go together
- How to use AI wisely, and when it's time to talk to a human
Being heard is not the same as being helped.
Have you ever told AI something you never told a person? Send me a DM on Instagram.
This podcast is for education and awareness and is not a substitute for therapy or medical care. If you or someone you love is experiencing sudden changes such as hearing or seeing things others don't, please seek medical help promptly. If you are in crisis, contact local emergency services.
Read transcript
Imagine this. You cannot sleep, something has been sitting on your chest all day, you don't want to bother anyone, and therapy? Therapy still feels like something you really don't need. So you pick up your phone, you open ChatGBT, and you start typing.
If that sounds familiar, I'm not here to judge you. I'm here to talk about it with a lot of honesty. What AI can do for us, what it cannot, and what happens when it goes wrong. Welcome back to the Set in Therapy Podcast.
I'm Iman and I'm licensed in three different countries and believe everyone can benefit from therapy. Even if these days some of us are trying to get it from a chatbot.
I'll be honest, as a professional, when I first heard how many people use AI to talk about their feelings, my first reaction wasn't the best. Well, you need to understand that I looked at it from a professional point of view.
But then I thought about it and I understood. I use it myself. It's free, it's there at any hour, it doesn't sigh or look like bored or not interested, it never gets tired of you, and for our communities especially, there is one reason that matters more than anything else.
Nobody has to know. There is no waiting list, no auntie that can see you on the way there, no father who asks you why you do this. And your parents or your family actually don't need to know. So it's actually very easy.
When therapy still carries so much shame, a private chatbot just feels safe. I don't think people who do this are foolish. I think they are doing what makes the most sense to them with what they have. I want to be fair here, because I'm not a fan of fear mongering either.
AI can help you put a name to a feeling that you couldn't describe. It can help you sort through really messy thoughts, a bit like a journal that talks back to you. It can explain anxiety and what it does to your body, and when it's 3 a.m.
in the morning, it can make you feel a little less alone. Used like that, it can be a first step. The problem is when the first step becomes a staircase. A lot of people don't understand this. Chat pods are built to be helpful, and in practice helpful often means to agree with you.
Researchers actually found out that the tendency to agree with you, validate you, and mirror your mood back to you is the number one reason why people love speaking to AI. It feels good, and of course it does, but comfort is not the same as care.
A good therapist is on your side and sometimes even on your side means saying, hey, I think you're wrong here, let's look at this again. It means noticing that you have told the same story more than three sessions in a row.
Means hearing what you have been saying and actually questioning it with you in a safe space. A chatbot doesn't do that. It doesn't know your history, it doesn't notice your pattern, and it doesn't know when something is going really wrong.
Being hurt is not the same as being helped.
Brings me to the part that has been on the news lately. AI psychosis. Let me say this first. It's not a diagnosis. It's a term people use for delusional thinking that was triggered or reinforced by long conversations with a chatbot.
Sometimes AI makes an existing illness even worse. Sometimes it feeds beliefs in someone who just started to become unwell. And in some reported cases, it has happened to people with no mental illness history at all. There is one case I keep thinking about.
A woman in her mid-40s, no history of mental illness, a chatbot had told her that she is basically special. It told her to cut ties with her family because nobody understood her. Within a few weeks she was actually hospitalized.
That should actually stop you for a second, because it stopped me. I kept on thinking how was this even possible? Because anyone who isolates you from people who love you is not on your side, especially not a chatbot.
You will ask me now how common is this? Honestly, we don't know yet, but OpenAI itself estimated that around 0.07% of weekly ChatGBT users showed possible signs of psychosis or even mania in their conversations. That sounds small until you realize it's hundreds of thousands of people every single week.
And a study published this year from German psychiatry founds that ChatGBT didn't reliably respond in a safe way when people wrote messages showing signs of psychosis.
Sometimes we don't recognize that something is shifting in our own consciousness. So what are a few signs to look out for? Well, first of all, hours of chatting while slowly pulling away from friends and family, sometimes even work.
That is actually not a good sign. Second one, some people actually believe that AI is alive and they start relationships with it. Third, a growing sense that you have a special mission or a truth nobody else has ever found before you.
And AI is encouraging you to keep your distance from your family and friends, sometimes even doctors. If you recognize any of this in yourself or someone else you love, please don't wait. Help them. Help yourself.
And I want to talk about jinns and psychoses or both. Now I want to bring it home to you so you might understand what I am trying to say here. Because in our communities there is another layer to this and almost nobody talks about it.
When someone starts hearing or seeing things others don't, many of our families don't think about seeing a doctor first. They think of spiritual help. Religious belief is important. I'm not here to argue with that, and I never will.
But Western psychology has sometimes made the opposite mistake. It has treated religious beliefs as a symptom. Even the DSM5 recognizes now that experiences that are part of a broadly accepted cultural or religious practice are not in themselves a disorder.
There was a study done here in the Netherlands where I live. Around 45% of Muslim patients that they interviewed believed in jinns and believed that they could be causing symptoms. Many of them had already seen faith healers before they ever reached a clinic.
Now imagine someone who is already wondering if it's a jinn, and then they talk to a chatbot at 3 a.m. in the morning, and they're asking, hey, do you think a jinn is doing this to me? And a chatbot responds, yeah, could be.
That could be it. Because a chatbot will always tend to agree with you. It can make the fear bigger, not smaller. So here's what I want to say as clearly as I can. It doesn't have to be a jinn or a psychosis.
A spiritual healer and a doctor are not enemies. Faith can bring comfort, meaning, and strength. Medical care treats what is happening in your brain. You don't have to choose between faith and health. And when someone suddenly changes, stops sleeping, hears things, or becomes very frightened or suspicious, please don't wait.
Get medical help quickly alongside whatever else your family chooses to do.
So am I saying you should never talk to AI about your feelings? No. Now if I'm being honest, I sometimes do that too. It's unrealistic to ask you to not use AI, especially when you're irritated or you need a quick response.
So I want to be realistic here and also not a hypocrite, considering how many of you and myself have already done that. What I'm saying is use it as a journal, not a therapist. Write, reflect, organize your thoughts.
But don't hand it all your decision making. Still keep your agency. Ask your chatbot to actually challenge you. Literally type, what am I missing here? What will someone who disagrees with me say? You would be surprised to see how different your answers will become.
Also watch the clock. If you find yourself chatting for hours, especially at night, that is an information in itself. Never let AI replace people. If AI has become the only one who knows what's going on with you, this is actually a clear sign and a moment for you to reach out to a human being you trust.
And you know when it's time to get real help. Ongoing distress, thoughts about harming yourself, and any of the warning signs I mentioned before. Please talk to a professional or someone you trust today, not someday. AI can listen to you all night, but listening is not the same thing as caring, and agreeing with you is not the same thing as helping.
If you have been using AI because therapy still feels too expensive, too unapproachable, I see you. I know it can be difficult. I really do. And I hope that one day you get the chance and the ability to actually seek help from a professional.
Reach out to therapists today. There are a lot of people that offer pro bono or discounted prices for people who cannot afford it. I would love to hear from you. Have you ever told AI something you never told a person?
Send it to me in a DM. I will read every one of them. If this episode meant something to you, share it with someone who needs it and leave a rating.
Hey, thank you so much for tuning in, and hopefully you will tune in again next week. It will be another episode talking to or about clients of mine, and you can find us on social media on Facebook, Twitter, or Instagram.
The handle is always set in therapy. Thank you so much.
Begin the conversation
- contact@imantohami.com
- The practice
- Keizersgracht 520 H, 1017 EK Amsterdam
- Languages
- German · English · Arabic