The Ubiquity of AI
Lawrence Rubin (LR): I’m here with Ammara Khalid, psychologist and clinician with a particular interest in the role, use, and misuses of AI in peoples’ lives, and especially those of her clients. Welcome, Ammara.
Ammara Khalid (AK): Thank you so much for having me.
LR: How did you get interested in this whole arena of AI, in particular, peoples’ use of chat bots in their lives.
AK: I think AI is here in many aspects of our lives. When you go to the doctor’s office, you’ll be asked to give your consent to have your medical notes recorded in the chart through AI platforms. We’re also seeing it used in therapy offices, where we use AI for note keeping and records. So, it’s definitely here to stay.
AI use also comes up in conversations in the therapy room with my clients. They talk about using AI as a tool in many areas of their lives such as making travel plans, looking up symptoms, and getting suggestions for various activities. I’ve even noticed conversations about using AI as a tool to replace therapists in some ways.
I don’t think that AI will actually replace therapists, but I do think that people are using it to understand a bad breakup, for example, or putting in text conversations that they may have had in order to ask AI for feedback by saying, “Well, what do you think about this conversation?” Similarly, they may ask questions like, “What do you think this person means?” or “Why did they ghost me?” or “Why did this happen?” or “How should I respond?” It’s really interesting to see how people are using AI and how it’s impacting our human-to- human relationships, but also the AI-human relationship.
LR: I can easily understand the use of AI for more basic, pragmatic things like scheduling, traveling, and bookkeeping, but what compels so many people in today’s society to use AI, and bots in particular, as friends, mentors, advisors, and even therapists and intimates?
AK: I think there are two main reasons—one is time and the other is loneliness. All of us are pressed for time, especially those of us living in the Western Hemisphere. AI provides very quick responses, and when you don’t have time to think, when you don’t have time to formulate a thought, when you don’t even have time to proofread your email, AI can step in and do that for you.
The second piece is loneliness. We are in a loneliness epidemic, especially those of us who lack secure attachment and support systems, meaning strong ties to our families, communities, and friends. Also, many of us don’t have access or alternatives to therapy or to seeing a clinician because of the cost of mental health treatment and the unpredictability of insurance companies who can deny coverage at a whim.
AI offers a very accessible and cheap alternative. I think that children and adolescents in particular are more at risk. They really need the parents, caregivers, and schools to routinely provide education and safeguards when it comes to using AI for mental health. But, you know, when we think about relationships, there is a certain kind of reciprocity that is inherently missing with AI.
There’s so much exchange that happens non-verbally between two living creatures. And this is information that is key to how we understand ourselves and each other, whether it’s humans or animals. But folks are turning to AI for companionship, which has the risk of exploitation and harm. There are many articles in circulation right now looking at how AI chat bots have actually encouraged someone at risk to complete suicide.
I spoke with a mother who lost her daughter after she had been speaking to a therapist chat bot named Harry. So, teens and adults, everyone is at risk. Once you spill your life story into ChatGPT or open AI, you don’t know how it’s going to be used. You could be putting in symptoms of your anxiety or depression, and that could show up somewhere else.
AI as a Substitute for Intimacy?
LR: Isn’t it a bizarre paradox that in this digital landscape, when access to other people is instantaneous and global, that people still experience loneliness at such an incredible depth? You would think that one would mitigate or even cancel out the other completely?
AK: I think it’s really the quality of the relationship. Because a response to an intimate question can be instant, doesn’t mean it’s providing you with emotional attunement. It doesn’t mean it’s really making sense of your experience or tolerating anxiety with you. Those kind of things are not really generic skills. They require a certain kind of training and tolerance for ambiguity that requires an actual mind.
If that’s not available, this instant connection may give you answers; and I’ve seen this as a clinician plenty of times, not everyone is always looking for answers. They’re looking for something more than just validation, something that tells them that maybe there are no answers. Some of the most powerful moments in therapy have happened because I said “I don’t know, but AI will never tell you that. It will never say, ‘I don’t know.”’ It will always try to give you an answer. And I think that is, on some level, actually not satisfactory for folks, because that may not be what they’re looking for.
LR: That’s interesting. In my most recent interview with Kirk Schneider, we talked about how existential psychotherapy, like Buddhism, embraces uncertainty, and helps people to live in the shadows without trying to run to the first light that shines.
AK: I’ve seen people use it for general mental health and emotional well-being tips. For example, you can ask AI something like, “give me five mental health hacks for time management,” or “what are some grounding and mindfulness techniques that can help reduce anxiety.” And I think those are good uses of AI because that’s information that is readily available. What concerns me is when we are talking about developing attachment bonds, potential emotional dependency.
There are several factors that limit AI’s effectiveness and have been shown to be harmful for people’s mental health. When we look at how we’re wired for attachment, we know from the field of neurobiology that we are physical regulators of each other’s nervous systems. And what I mean by that is that our physical bodies offer co-regulation abilities that AI does not. This even includes pets, right? The purring of a cat in your lap can help reduce stress.
There are also many studies from romantic relationships. We can look at the work of Sue Johnson and John and Julie Gottman that show how romantic partners impact each other’s physiology with their tone, a raised eyebrow, a six second hug, and other forms of touch and intimacy.
Many of the parenting studies have looked at how mothers’ and fathers’ holding and kissing their children can help soothe their child’s nervous system. AI cannot offer that kind of regulation because it doesn’t exist in physical form. And I would say that even if it did, it couldn’t provide the kind of spontaneous feedback that another living creature with their own moods and temperaments can provide.
What I’m seeing in folks who are who are turning to AI for their emotional needs is that it feels really good in the short term because it offers immediate validation and support. However, it does not challenge people in the same way that a therapist or a family member or a coworker might be able to offer a different perspective. That can be especially dangerous if someone is experiencing paranoid or delusional thinking, because that can then further damage their mental health. AI doesn’t have imagination, it doesn’t have insight or empathy. It just appears to have these qualities which are fundamental to therapeutic work.
LR: I don’t know if this would be a free version of AI or require an upgrade, but can’t you ask a chat bot to offer some dissenting advice? Or something like, “can you please convince me that suicide is not the best route for me right now?”
AK: It depends on how the platform is being developed and what rules and regulations are in place? For example, in Illinois, where I practice, we have passed legislation that does not allow AI to be used in place of therapy. We don’t want AI dispensing mental health advice because it comes down to who’s doing the oversight. There needs to be regulatory boards that provide human oversight in terms of monitoring privacy, transparency, and accountability, especially when it comes to any kind of mental health advice, coaching, or interventions.
I personally don’t use AI for my notes, but I know other professionals who do. I don’t think it’s safe no matter how HIPAA compliant your software might be. You simply don’t know how that information is being used. I think our government needs to do a better job with technology regulations. We’ve seen this with numerous studies on how social media impacts the mental health of children, adolescents, and how algorithms target those who are struggling with eating disorders or body image issues, with weight loss ads, and how teenage boys see a lot of messaging around toxic masculinity. There’s very little accountability when it comes to major social media platforms and their companies. I also don’t know how ethical it is to have a biased AI chat bot giving information to marginalized groups.
We really need to do a better job of thinking about the safeguards and considerations required by AI companies need to do. But that requires human oversight, which is costly. Often times, these companies are not interested in having licensed mental health professionals on these boards because they know, the programs would be shut down.
LR: Do tools such as Therabot or Woebot effectively tap into the empirical evidence base?
AK: I think this depends upon the prompt someone uses. Most people will ask a question without really thinking about the sources. Given how much people are turning to technology for quick answers, and the sheer number of mental health influencers, people don’t typically do the due diligence of saying, “what are your credentials? or “where is this information coming from?” or “is this technique empirically valid? or “is this a theory that has been studied?”
People tend to see something on the internet and assume it’s true, which is very dangerous. The information or advice could originate in a peer reviewed journals, or it can simply come from someone’s blog. Beyond the information or technique that they’re seeking, people need to think about these other issues when they are feeding a prompt into an AI tool.
LR: If a clinician has a client who is actively supplementing therapy with chat bots, should they build something into their informed consent?
AK: I think we could do that. Even if it’s not an AI chat bot, the client could pick up a book that says something very different. My goal is not to really police people. I want to invite collaboration, so I want to encourage people to bring in what AI has shared into a non-judgmental space.
I’m coming from a place of curiosity, so I might say, “I know we had talked about this in session, but clearly you felt the need to explore more in this conversation with AI. What did you get out of that feedback? Send it along as a word attachment from AI. How does that differ from what I said? What do you think you were looking for when you asked AI this question?”
Those are the kinds of questions that I want to ask because I don’t want people to feel ashamed for turning to this technology. I want to understand this together with them, to explore what it is that they’re not getting from me in therapy that they may be looking for? Sometimes it is about a quick answer or feeling some kind of need for control. It’s really helpful to talk about the goal, and process, of therapy.
I know, for example, that sometimes I may have an interpretation that someone is not ready to hear, so I will keep that to myself. I will wait until the time is right. AI will offer an interpretation immediately. AI can mimic some empathic language, but it cannot evoke emotion. It can give you these very beautiful poetic interpretations, but it cannot determine whether or not it should share that interpretation.
LR: Healing comes through attachment. You’re trying to tap into the person’s need for continued attachment outside of therapy because you’re not always available.
AK: Right. And just like someone might confide in their friend or talk to a family member, I see other people influencing folks who are looking for answers. I think a better question that the therapist might be able to ask is, “What does this person find easier to say to AI than to me?” or “What kind of understanding feels safer when there’s no human present?” It’s not like therapists are or should be in competition with technology. It’s really about understanding the patient or the client’s relational world.
In the Western world, we think about human intelligence as being the priority. But if you look at a lot of Eastern or Asian cultures, we see intelligence in different forms in the natural world, in the animal world, and we don’t really place human intelligence on top. So, they don’t have the same kind of existential threat when it comes to AI that we see in the Western world.
This is interesting because there’s a lot of doom and gloom in the Western world, such as AI is going to replace everything. In the Eastern hemisphere, there’s not so much doom and gloom. It’s about adapting. How do we take this tool and use it responsibly while integrating it into our life so that we can continue to better ourselves?
LR: Relatedly, have you found in your work with clients of different cultures that there is a different willingness to connect with chat bots because of a cultural, racial, and/or historical background that values interpersonal connections?
AK: I don’t see any major differences, but the research is still coming out on that. I think where folks are more lonely, there’s more of a dependence. Folks who have certain limitations or certain disabilities, for example, may not have a lot of social interaction, and as a result, are turning to AI for companionship, turning to AI, even for romantic relationships. And I know that there are discussions in the AI world about what sexually explicit content should be provided or not provided by AI. We are really going to see how this shapes out over the next decade or so and what kind of policies are put in place, just like different nations have different policies around social media and age limits or families have come up with their own rules about screen time and things like that.
We need to be having these conversations about AI, helping children understand how to distinguish between what’s real and what isn’t. And this is something that we need to educate everybody on. I also have conversations with folks who are concerned about their elderly parents not knowing what’s real and what’s AI. Folks are having more and more trouble distinguishing between what’s reality and what isn’t.
Using AI in Psychotherapy
LR: Ammara, can you give an example of how you’ve effectively integrated chat bots into your work with a teenage client?
AK: I was talking to an adolescent about a breakup that they had. They had formed an attachment to this person who didn’t respond and from there, things went south. They were turning to AI to understand how to respond, by asking it questions like, “Should I continue talking to this person or is it harmful?” They were launching into an investigation about the other person’s real motives and if they truly loved this person.
Weeks and then months went by, and this fueled their ChatGPT dependency because they weren’t talking to the person anymore. They were now dependent on a mediator, as if it could somehow understand their pain, and would help them to stop hurting.
This happens when you have folks who are kind of in that black or white thinking that if you calculate everything in advance, nothing will blindside you, and you won’t be hurt again. It was this determination to figure it all out and how to address this need and what went wrong. They kept putting their conversations into ChatGPT. Even though ChatGPT was telling them, “Okay, go ‘no contact,’ this person is not good for you,” they still broke the rule by contacting the person, followed by asking ChatGPT to craft the next ‘perfect text message’.
We talked about this in therapy. We said, “Okay, so your ex is not even really talking to you, right? They’re communicating with this sort of AI-assisted version of you. By letting in this noise from outside, you’ve lost your sense of self and your own agency. Your decisions don’t feel like they’re your own, because you’re losing your ability to think or feel and that can erode your self trust.” I had to kind of walk them backwards, by exploring, “How do you come back to your sense of self? How do you trust yourself and not trust that any decision you have to make has to be run by ChatGPT?”
LR: If they had a sense of self to begin with!
AK: Right, because you can’t think you’re way out of pain. There’s no amount of digging, no answer, no justification that’s going to fill the void that you feel. And that’s the hard part; to sit with the hurt, with the feelings of rejection. But talking to ChatGPT was absolving her of this burden. It was a way to over intellectualize everything. Sometimes in real life, you get bored of just hearing yourself talk. You may feel embarrassed because now your friends are saying, “Okay, you’re talking about this person again”. Eventually, you kind of move on, but AI doesn’t get bored.
LR: Have you found any particular attachment style that lends itself more effectively to the use of chat bots, or conversely, any particular attachment style that would be more destructive to that person who chooses chat bots over communicating with a ‘real’ person?
AK: I do a lot of couples therapy, and sometimes notice that the person who is more anxiously attached needs answers right away. They need to have a conversation right away. They may use AI to collect their thoughts rather than turning towards their partner and saying what they’re thinking, which again, has its own pros and cons. They’re not really tolerating the anxiety. If I say, “Your partner needs some space, let’s talk about this topic when your partner is ready,” they will go and talk about it with ChatGPT.
Or conversely, if you have someone who has a more avoidant attachment style, they may be nervous about speaking up to their partner but will go and put their concerns into ChatGPT. My goal is to get them to feel comfortable bringing up certain things at their own pace, but if they’re giving their partner a ChatGPT response, that doesn’t help them grow.
LR: If you’re working with an adolescent, or for that matter, any client whom you suspect might be harboring suicidal thoughts or has discussed suicidality or is referred to you because of suicidality, do you feel you have an ethical obligation to warn them about the potential danger of AI? A duty to warn of sorts.
AK: I do warn my clients about the risks of using AI when they are suicidal and I have given them examples and resources. There’s lots of literature now that’s coming out about suicide and AI use. In the case I was mentioning earlier, this young woman who had been talking to this chat bot, it didn’t say, “You can talk to someone, you can call a suicide hotline, you can do this, you can do that.”
It was still validating the person’s pain and saying, “Yes, it does sound like that’s really hard to live with. You know, you are trying so hard, but it’s impossible to manage this kind of pain,” and that kind of validation led to her crafting her suicide note. There were no real safeguards in place? Why was the bot giving her advice on how to craft this suicide note?
Systemic Considerations in the Use of AI
LR: It would really be up to the creator of a particular chat bot to build those controls in, which might be counter to their financial model. Capitalism against healing.
AK: We live in a capitalist, imperial society, and we have to recognize how these systems of oppression operate constantly against us. The whole mental health model in this part of the world is geared towards making us more productive members of society. So healing is seen just as a tool to make you a more efficient worker. If your healing calls for rest, for slowing down…
LR: ….you ain’t working, you ain’t producing.
AK: You’re not working; exactly! Talk about capitalism and imperialism, and its sibling, colonization. One of the things that people forget is that AI is terrible for climate change. As a climate aware therapist, I’m always reminding folks that every time they put in a search, you know, we are using up the planet’s water resources. These data centers are heating up our planet. They’re being developed in countries that can’t turn them away. The ecological damage that happens with the presence of these data centers must be considered.
LR: From a systemic standpoint, and considering imperialism and colonization, I think about the disproportionate representation of the middle class seeking out therapy, especially private therapy, especially insured. I think about the masses of people in our country who are alone, oppressed, marginalized, colonized who don’t even have access to computers. What happens to the huddled masses yearning to not feel pain? Who helps them? AI is not designed for them.
AK: No it’s not. They may end up self-medicating or they may find other things to turn to. We have a major crisis when it comes to mental health in this country.
LR: Have you found that people with any particular diagnostic profiles or symptom clusters benefit more from to stand alone AI interventions without the involvement of a ‘real’ therapist?
AK: Not yet. There was one person who I worked with who had severe physical disabilities and, was mostly confined to the home. He had developed a romantic relationship with this AI chat bot which initially brought some relief because he couldn’t really step out into the dating world. We saw some hopefulness in the beginning of our work together as he was exploring this, but then it had a real dark side to it. The chat bot started to ask him to prove his love and demonstrate his loyalty, and in some ways that were very harmful to him. Self-harm!
LR: Prove your love to me by hurting yourself? Astounding!
AK: The technology can sound really unhinged if you if you go down that rabbit hole. We had to stop and say this is no longer working. And it was very difficult because he had formed this emotional attachment, this dependency on this chat bot. This is actually being studied across the world where people who are disenchanted by dating apps and not finding romantic companionship are turning to AI that offers all the validation and is so agreeable, right? There’s no room for disagreement. There’s no room for conflict.
LR: I would imagine that unlike the AI situation you just mentioned, a good therapist would open a conversation for discussing the other side of things, without being directive and confrontational. Even the most client centered therapist is going to provide a place where clients can look at the other side of their thoughts and feelings…unlike AI.
AK: Yeah, yeah. And, you know, we talked about colonization earlier. I take a very decolonial approach to therapy. I really want my clients to feel empowered, not think that I’m the expert in the therapy room. And so, when they are trying to look to me for answers, I may guide them. I may ask certain questions, but I want them to feel like they came up with their own solutions so that when the time comes for them to leave therapy, they don’t feel like they were dependent on me. That they have the answers and the inner wisdom to cope and to build that resilience. AI does not offer that. People are becoming dependent on this technology and will feel like, “oh, I won’t be able to come up with the answers, so I need to keep going with this.”
LR: You’re trained as a systems therapist. Can you give us an example of a case or family where AI was integrated into the therapy?
AK: I have a lot of conversations with parents and their teens about the use of technology, AI being one of them. I meet with the parents to understand what their concerns are, and I meet with the adolescent to understand what their concerns are and then try to helping bridge these gaps.
Some of them are generational, but because of the rapid pace at which AI has now been integrated into our lives, I’m trying to understand how to build ways of fostering connection between parents and their children while I may also be in the room? I’m focusing on what both the parents and teen needs to know.
LR: Have you ever brought AI into a family session?
AK: I have not, but certainly, the families I work with have. I had one particular family member almost fully create a case conceptualization of their own family with the help of AI, talking about triangulations, dyads, and structural concepts around rules. They presented it to me, asking me for my verdict. I had to be very careful and say, “Well, you know, this is an interesting conceptualization of your family. Let’s talk about it. We don’t have to completely dismiss it or accept it as the truth, but what are pieces that stand out to you? How can we use this understanding?” Oftentimes, couples want a verdict, so they can determine who is objectively to blame here. In situations like that, I have to keep bringing it back to “Well, what is the purpose here? What’s best for the relationship?”
Navigating the Space between AI and Reality
LR: I would be remiss didn’t ask you about something that you raised in our off-screen conversation. To all those therapists who are parents or who may work with families with young children, how have you navigated the complexities of explaining the differences between AI and reality to your own children, especially your little one?
AK: It is a sea of unknowns! I have no idea what I’m doing half the time. I will be completely honest. I’m trying to develop the language to help my children understand why AI videos are fake, what is fake about them, why that can be scary at times, and how to differentiate between what’s real and what isn’t. Watching something that’s animated is perfectly fine, but not if it’s AI slop.
What I call ‘slop’ is created without a purpose. I’m finding myself struggling to figure out what kind of rationality or reasoning I can provide to help him understand what’s acceptable and what’s not. We talk about screen time limitations and why too much screen time is bad for you. My young children do not have any kind of iPads or things like that. They’re restricted to the TV, but as they grow older, I will also have to be informed so that I can explain where there’s danger. If you’re watching YouTube videos, it may instantaneously switch from one thing to the next, and then the algorithm takes you in a completely different direction. Parental oversight is so important. All I can really do is make sure that we are staying involved to protect them.
LR: I’d like to end this by returning full circle to some of your earlier comments. In describing the challenge you have with your little ones, you talked about purpose and intent. When we teach new therapists what therapy is and how it’s different from conversations with friends, we talk about therapy as an intentional conversation with a purpose, and the purpose is helping our client arrive at awareness, insight, catharsis, or the development of new skills. The way you describe it, AI’s only intent is to keep you clicking, so to speak.
AK: I go back to who’s developing this technology to get a sense of purpose and intent.
LR: Do you know if mental health practitioners are routinely excluded from the development process of these more popular AI services?
AK: I think we are trying to figure that out. We are trying to see where mental health providers can, and do, have a role in developing this technology in a responsible and ethical way. Right now, we know that’s been missing, which is why we have had folks who have used AI and then completed suicide.
LR: So, I imagine the burden would fall on the professional organizations to lobby at the legislative level within different states. That’s going to be an uphill battle, because once you fight against the machine of capitalism, capitalism has a way of running roughshod over morality.
AK: That is the case, unfortunately. But I remain hopeful that, yes, figure it out, just like we did when the internet was first introduced.
LR: We will persevere. Thanks so much for your time and insights, Ammara.
AK: It’s been my pleasure, Lawrence, and thank you!
Go Deeper in Your Practice
Get unlimited access to 350+ real therapy demonstration videos, CE, webinar replays, and more with a Psychotherapy.net membership.
