Image: Episode artwork for Modern Therapist’s Survival Guide, Episode 483, titled
Google PodcastsApple PodcastsStitcherSpotifySubscribe on AndroidShare Listen in a New WindowDownloadSoundCloudLeave a ReviewSubscribe via RSSKeep me updated!Player Embed

When Clients Bring AI to Therapy: Working with ChatGPT, Sycophancy, and the New Third Wheel in the Room

Curt Widhalm, LMFT, and Katie Vernoy, LMFT take on a question they hear constantly from listeners and audiences: what happens when clients bring their own AI into the therapy room? The profession has spent real energy building ethics guidance for when therapists direct clients toward AI, but it has largely sidestepped the messier reality already showing up in sessions, where clients arrive with chat logs as homework, lean on an endlessly agreeable algorithm between appointments, and sometimes use a chatbot to critique the clinical work itself.

Curt names three patterns he is seeing in his own practice: clients processing the same issues with AI between sessions, clients using AI to interpret or respond to relationships (especially dating), and clients (often young men) offloading their social and dating lives into AI companions. Katie adds the everyday end of the spectrum, where AI functions as a helpful assistant, and the concerning end, where heavy use shades into emotional dependence. Rather than landing on “AI is bad,” they treat it as a clinical reality that calls for assessment, psychoeducation, and curiosity.

This is part of the podcast’s ongoing conversation about AI in clinical practice, building on earlier episodes like the deskilling podcourse. Curt and Katie offer a working overview therapists can use right away.

Transcript

Click here to scroll to the podcast transcript.

(Show notes provided in collaboration with Otter.ai and Claude AI.)

In this podcast episode: Assessing Client AI Use, Sycophancy, and Clinical Responsibility

Curt and Katie work through how to recognize the many ways clients are already using AI, from work-driven habit to relational outsourcing to full companionship, and why the function of the use matters more than the fact of it. They talk through the mechanics therapists need to be able to explain in plain terms, including why a chatbot built to keep you engaged tends toward agreeableness and validation rather than objective truth, and they share concrete in-session moves for building critical thinking, protecting confidentiality, and reframing how clients use AI around medical and diagnostic questions. They close on the harder cases, where AI has become a client’s primary support system, and what it looks like to begin weaning a client back toward real-world connection.

Key Takeaways for Therapists: Assessing AI Use, Psychoeducation, and Working with the Chatbot in the Room

“This is a massive therapeutic opportunity. It’s being able to see the unfiltered versions of clients a lot of times with how they’re entering in prompts and being able to see them in a way that they might not necessarily be interacting with you as a therapist.”

– Curt Widhalm, LMFT

  • Lead with curiosity, not judgment. A judgmental stance toward AI forecloses the clinical opportunity. Exploring what a client is actually getting from the tool, whether validation, accessibility at 2am, or a felt sense of being understood, opens a window into unmet needs, attachment patterns, and how they relate to the world.
  • Assess the function and pattern of use. Ask why, how often, and for what: work, relational and emotional support, therapy-adjacent processing, or companionship. Ongoing assessment matters, because clients reinforced to use AI all day at work may be primed to over-rely on it everywhere else.
  • Name AI’s mechanics candidly. AI is not an objective third party. It predicts what a user wants to hear and can validate cognitive distortions as readily as insight. Psychoeducation on sycophancy, confidentiality, and safety is itself an intervention.
  • Teach critical thinking with concrete experiments. Curt’s move is to have a client enter the opposite position, or a partner’s perspective, and watch the chatbot validate that too. It surfaces the gap between objective truth and the client’s subjective truth, and builds toward clearer, more effective communication with the people in their life.
  • Treat chatbots as public spaces. Clients sharing diagnoses, medical details, or mental state into free or learning-enabled tools may be handing over data that can be reconstituted. Therapists can act as stewards of client privacy and coach clients to share only what they would say publicly.
  • Reframe the prompt for medical and diagnostic questions. Rather than “what is my diagnosis,” coach clients to ask “what are three questions I should ask my doctor.” That turns AI into preparation for expert care, not a substitute for it, and helps clients become more informed without mistaking a confident answer for an accurate one.
  • Watch for dependence as its own clinical picture. When AI has moved from neat tool to primary support system, partner, or girlfriend, the work shifts toward weaning, validating the underlying need, and helping clients practice the friction and disagreement of real relationships that AI is not built to provide.

“Our job is not to teach our clients how to use AI. It’s knowing enough about AI to be able to address any clinical impacts and mitigate risk. AI’s not good for crisis response, AI doesn’t hold confidentiality. It doesn’t call you out in the way a therapist will. It’s not sufficient for what we’re able to do and what our responsibilities are as clinicians.”

– Katie Vernoy, LMFT

Concern, Not Condemnation: What the Research Is Starting to Show

Katie grounds the conversation in two studies worth knowing. A four-week randomized controlled trial from the MIT Media Lab and OpenAI followed nearly a thousand participants across more than 300,000 messages and found that higher voluntary chatbot use tracked with more loneliness, more emotional dependence, more problematic use, and less real-world socialization, with the heaviest effects among users who reported greater trust in and attraction to the AI. A second paper describes a “technological folie à deux,” a feedback loop in which a user’s belief and a chatbot’s sycophantic validation reinforce one another, a dynamic that can be especially destabilizing for people prone to magical thinking or psychosis and that carries real stakes around crisis and safety.

The reframe the hosts land on is that “concern” is more useful than alarm. Clients are already using these tools, AI is already in the treatment room, and a flat “AI is bad” stance is both unrealistic and clinically limiting. The task is to separate the healthy, adaptive uses from the unhealthy ones, and to do enough of this work on ourselves that we do not follow our clients down the same seductive, sycophantic rabbit hole.

Resources on AI in the Therapy Room

We’ve pulled together resources mentioned in this episode and put together some handy-dandy links. Please note that some of the links below may be affiliate links, so if you purchase after clicking below, we may get a little bit of cash in our pockets. We thank you in advance!

Relevant Episodes of MTSG Podcast

Meet the Hosts: Curt Widhalm & Katie Vernoy

Picture of Curt Widhalm, LMFT, co-host of the Modern Therapist's Survival Guide podcast; a nice young man with a glorious beard.Curt Widhalm, LMFT

Curt Widhalm is in private practice in the Los Angeles area. He is the cofounder of the Therapy Reimagined conference, an Adjunct Professor at Pepperdine University and CSUN, a former Subject Matter Expert for the California Board of Behavioral Sciences, former CFO of the California Association of Marriage and Family Therapists, and a loving husband and father. He is 1/2 great person, 1/2 provocateur, and 1/2 geek, in that order. He dabbles in the dark art of making “dad jokes” and usually has a half-empty cup of coffee somewhere nearby. Learn more at: http://www.curtwidhalm.com

Picture of Katie Vernoy, LMFT, co-host of the Modern Therapist's Survival Guide podcastKatie Vernoy, LMFT

Katie Vernoy is a Licensed Marriage and Family Therapist, coach, and consultant supporting leaders, visionaries, executives, and helping professionals to create sustainable careers. Katie, with Curt, has developed workshops and a conference, Therapy Reimagined, to support therapists navigating through the modern challenges of this profession. Katie is also a former President of the California Association of Marriage and Family Therapists. In her spare time, Katie is secretly siphoning off Curt’s youthful energy, so that she can take over the world. Learn more at: http://www.katievernoy.com

A Quick Note:

Our opinions are our own. We are only speaking for ourselves – except when we speak for each other, or over each other. We’re working on it.

Our guests are also only speaking for themselves and have their own opinions. We aren’t trying to take their voice, and no one speaks for us either. Mostly because they don’t want to, but hey.

Join the Modern Therapist Community:

Linktree

Patreon | Buy Me A Coffee

Podcast Homepage | Therapy Reimagined Homepage

Facebook | Facebook Group | Instagram | YouTube | LinkedIn | Substack

Consultation services with Curt Widhalm or Katie Vernoy:

The Fifty-Minute Hour

Connect with the Modern Therapist Community:

Our Facebook Group – The Modern Therapists Group

Modern Therapist’s Survival Guide Creative Credits:

Voice Over by DW McCann https://www.facebook.com/McCannDW

Music by Crystal Grooms Mangano https://groomsymusic.com

Transcript for this episode of the Modern Therapist’s Survival Guide podcast (Autogenerated):

Transcripts do not include advertisements just a reference to the advertising break (as such timing does not account for advertisements)

… 0:00
(Opening Advertisement)

Announcer 0:00
You’re listening to the Modern Therapist’s Survival Guide, where therapists live, breathe, and practice as human beings. To support you as a whole person and a therapist, here are your hosts, Curt Widhalm and Katie Vernoy.

Curt Widhalm 0:15
Welcome back, Modern Therapists. This is the Modern Therapist’s Survival Guide. I’m Curt Widhalm, with Katie Vernoy, and this is the podcast for therapist about the things that go on in our practices, the things that go on with our clients, our reactions to those things, and we are in the middle of Shark Month on the podcast, or as we’ve been talking about, kind of the new format of the show, and being able to go into deeper dives on some of the topics, this is actually AI month, and do you remember when the biggest third wheels in your sessions was maybe a client’s overbearing parent coming in and telling you about everything that’s going on during their week, or partner who texted too much? How many clients have you had that, have just handed over their phone just to show you a text conversation between them and one of their friends or one of their loved ones.

Katie Vernoy 1:09
To be fair, that still happens.

Curt Widhalm 1:14
In that way, it still seems like those are simpler times, because today we have a new kind of third wheel in our sessions, and that’s the algorithm of AI, and when clients are now bringing their own AI into therapy. Now…

Katie Vernoy 1:34
Yeah.

Curt Widhalm 1:35
We have talked, we have crafted ethics codes around when therapists direct clients to use AI, and we had largely sidestepped if clients are using AI, that’s its own whole separate issue, and as you and I have done presentations, as we’ve gotten feedback from the podcast about stuff, some of the questions that we get, What happens when clients bring AI into therapy?

Katie Vernoy 1:58
Yeah.

Curt Widhalm 1:58
That’s what this episode is about. So today we are diving into the world of AI in the therapy room. Clients who bring in chat logs as homework, the rise of the endlessly agreeable algorithms. How to handle a digital companion that is currently out validating your hundreds of 1000s of dollars of clinical training, so let’s dive in.

Katie Vernoy 2:28
Yeah, I think this is an important topic, because more and more AI has become pervasive in how folks are interacting with the world. I think about how often I’m interacting with Claude or ChatGPT or Google AI, you’ve got Gemini, and using it to as an assistant, right, as some sort of helpful tool to move through things more quickly, offload some of the grunt work, so to speak, but I don’t know about you, Curt, but I’ve been hearing a lot of folks talking about having their, their perspective validated by ChatGPT related to arguments with a partner or writing emails and basically running all of their communication with folks through their, their AI bot of choice, and so it becomes really challenging, I think, to work with folks if you don’t understand what’s actually happening or why it’s so important. So I’m excited to dig in. When we started talking about what to do with this episode, what was your initial thought around how we would approach it? Because I, I went in and I did a deep dive, I worked with Claude to try to sort through some other things. So I’ve actually used AI to help with this AI topic, but, but how have you seen it showed up, show up the most? Like, what is, what has been your biggest concern about making sure our audience is AI savvy enough to be able to work with it clinically?

Curt Widhalm 4:15
Most of the way that I’ve seen AI show up in my practice falls into, I think, three categories. Number one is clients who are using AI in between sessions around the very same topics that we’re talking about, and using it as kind of an addition. Sometimes they bring it in as why aren’t you picking up on these things, kind of a critique of the therapeutic work that we are, so that’s the first category. The second category is, as we both kind of alluded to here, trying to make sense of relationships or using it to help interpret other relationships. I’ve had multiple clients talk about either they themselves are using one of the AI’s in order to respond to typically dating partners and not coming up with their own reactions or having dating partners who are using AI and responding to them. So that’s kind of the second category where it’s people who are offloading their thinking into the algorithm and then responding as whatever the algorithm says. And then third is a growing concern that I’m seeing, particularly with young men, which is offloading their entire social lives and dating lives into AI girlfriends. So I am seeing AI play out more and more, but those are the three big areas that are showing up in my practice, and I assume that if it’s happening with me, it’s happening with others.

Katie Vernoy 5:55
Absolutely, and I think there’s the offloading of thinking, I think goes even further. I’ve even seen for myself, I have started doubting sometimes. I’m like, what do I think about this? And then, and then I’m like, let me just put it into Claude real quick, and let me run it through Google real fast. And there’s a whole article, and there’s a million articles that I found. I really did go down a rabbit hole, but I’ll link to all of those in the show notes, over at mtsgpodcast.com but there’s an article about the drive to cognitive death. It’s a very psychodynamic, deep kind of article around this need to make everything easy, and you stop critical thinking, you stop having your own thought. That’s obviously a very dangerous impact of AI, but I think the big piece is when someone has completely immersed themselves, it becomes almost addictive that they’re constantly interacting with AI, they have a kind of parasocial relationship, or those types of things, it can be very obvious that it’s of concern. I think some of these other kind of typical use, those things can also be clinically relevant, and so I want to make sure that we’re not leaving out the folks like you and me who are kind of just interacting with it as an assistant, as a helpful thing. I’ll take pictures and redesign my backyard, like I think that it can be very seductive, and it’s frankly fun. And so I wanted to talk about why this is of concern overall, and what we want to do with it clinically. And concern is more neutral than, oh my gosh, this is horrible, but I do want to reference a couple of studies that I saw. One was Fang et al in 2025 and it was a four week randomized control trial, lots of participants, 981 over 300,000 messages, and in that article they found that higher daily chat bot use was correlated with more loneliness, more emotional dependence, more problematic use, and lower socialization. And so when people had much higher trust in AI, there was more emotional attachment tendencies. There was a lot of harm. So that’s one. The other one, and I saw when I told you about this before we hit record. I saw your eyes, your eyebrows raise a little bit. I think you were kind of taken with this one. Is Dohnány et al, 2026 I might be saying that incorrectly. In Nature Mental Health, Technological folie à deux, and I thought that is fascinating, and it speaks directly to the sycophancy that you’re talking about, where I have an idea, AI then reinforces it for me. “You are so brilliant. This is absolutely right.” And then I seek more information, and it goes down a rabbit hole where we’re just completely validating whatever has been in your brain, and so I think it can be very dangerous. I mean, especially we can think about folks who have magical thinking or psychosis, those things can be very dangerous, but this also gets to the safety issues around crisis and crisis response. I mean, there have been situations where folks have been told that their ideas on how they might kill themselves and how they might die by suicide were great ideas, and this is what you need to be doing, and all of those things, because they were just reinforcing what the user was saying, and so it’s really important for clinicians to make sure that they’re assessing use, that they’re doing appropriate psychoeducation, those types of things, because this is a big deal. This is a new entity that is designed to keep us engaged, designed to make us feel good, and the way that that happens is this sycophancy, and we’re going to go into more detail in a CE episode later in this month, but I think there’s a lot here for clinicians to pay attention to, and I think saying, well, AI is bad, we can’t use it at all, I think is shortsighted because of our clients already using it, and it’s already entering into our treatment rooms.

… 10:23
(Advertisement Break)

Curt Widhalm 10:24
Like you’re saying, this is a massive therapeutic opportunity.

Katie Vernoy 10:28
Yeah.

Curt Widhalm 10:28
It’s being able to see the unfiltered versions of clients a lot of times with how they’re entering in prompts and being able to see them in a way that they might not necessarily be interacting with you as a therapist.

Katie Vernoy 10:44
Yeah.

Curt Widhalm 10:45
And I think that there’s opportunity here, and if you, as a clinician, are coming in as very judgmental about AI use, you’re going to miss out on this opportunity. So my first kind of thoughts around this is really in the way that I’ve tried to approach this is with curiosity, and let’s look into this, let’s see what you’re getting out of this, what you’re seeking in being able to share these chat logs, and part of this is looking at what are they getting out of the AI use in the first place? Is it the, you know, the sycophancy, is it the lack of judgment, the positive reinforcement, and being able to put that in the context of how their attachments are, how their upbringing is, how their reactions to the world are. If it’s really coming from a place of this is a place where I feel understood, that’s a avenue to be able to explore in some of the positivity and validation there. Is it rather the accessibility of having something there at 2am when a therapist isn’t available, or their friends aren’t available. Is there any number of different areas where it feels like something is missing in their real life? Is part of the evaluation that goes on, and that’s before we actually get to any of the content that they’re putting into it. I’m oftentimes surprised at how often the artificial intelligence chats that I have seen have actually provided fairly decent insight into some of the clients’ answers. So it’s quickly for a lot of people, becoming the go-to place to search for answers to problems, and particularly problems that are more relational or subjective in nature, rather than going to a traditional Google search or something like that. But really, where I end up finding this is validating that there are needs, that this is an avenue where that they are expressing their needs and seeing how this relationship with artificial intelligence is either something that they’re doing in a way that is healthy and adaptive or if it is a sign of something that is a larger problem.

Katie Vernoy 13:24
When someone is going to to a chat bot, especially a more generalized chat bot for advice or opinions, I think there are ways for it to be healthy and adaptive, I’ll agree with that. I think there’s a real need for caution if the client or the individual doesn’t have an understanding of the underlying mechanisms of how AI is built and what it’s built for. The folks who have created AI chat bots and large language models are designing it to keep you engaged. They’re designing it to be very agreeable, because folks like that. They also are making it more human-like in order to make it feel more welcoming, more more relatable, and so, like social media, like a lot of the things that we engage with, part of what happens in our use is this, you know, whether it’s dopamine or feel good, whatever it is, it seems like there is an element of this that we are reinforced for staying in this environment. I think with some of the things that I’ve, I’ve heard and read, and I can try to find yet more articles, if it’s of interest, that we can put in the show notes over at mtsgpodcast.com, but people who are incentivized to use AI for work, for example, will oftentimes make work and be constantly in Claude, Chat GPT, you know, whatever white labeled AI bot that their company has, and having constantly ding and come back, and they’re supposed to come back and and create the next prompt and have it do the next thing, and all of those things, so even if the primary usage is at work, there is this huge time that’s maybe spent in these environments, and so I think being able to assess the use: Why are you using it? What’s the function of it? How often are you using it? Are you only using it for work? Are there other relational, emotional elements to it? Is there, as you’re describing, Curt, is there therapy-adjacent work? Are you talking about therapy and processing, or therapy sessions? Are you seeking outside coaching from a chatbot? Knowing their relationship to AI. I think is an initial piece, because in our field it feels like AI has been not villainized, but it’s we’re very cautious.

Curt Widhalm 16:24
Sure, yeah.

Katie Vernoy 16:25
In other professions, the professions that some of our clients are in, as I’m mentioning, they may have been reinforced for using AI, that it may be something where their work, quote unquote, thinking partner is now a chat bot, and they have an AI assistant, and so they’re they’re predisposed to be in these spaces, and so I think a strong assessment from the beginning, an ongoing assessment of how it’s being used, I think is the initial first step. Being able to do some psychoeducation on sycophancy, to be able to talk about safety and treating AI chat bots as public spaces and only putting in the information that you would publicly share, for example, is something that’s really important. Now, there’s AI self-help bots, there’s there’s a lot of other AI opportunities out there that may have more secure environments, you know, they may be HIPAA compliant, they may have those things, we’re not reviewing those specifically on this episode, but unless someone is very clear on how their data is being used, I think that that’s something that, as a therapist, you want them to consider, because if they’re sharing their mental state, if they’re sharing their medical diagnoses, if they’re sharing those types of things, and they have learning turned on, they’re helping the AI chatbot, or they’re using the free version, whatever it is, their their data may be accessible and reconstitutable, and that’s that’s something that we want to pay attention to, because I feel that we’re kind of the stewards of our our clients’ privacy when we can be.

… 18:15
(Advertisement Break)

Curt Widhalm 18:17
There’s a lot that you’re saying that I think is where part of our responsibility when clients are bringing in artificial intelligence in this way is explaining to them the mechanics of artificial intelligence and putting it in terms that are rather candid about how AI is not an objective third party, it predicts what you want to hear. It validates cognitive distortions just as easily. It doesn’t necessarily always go so far as to telling you exactly what you want to hear, but it can be ultimately just there to continue to engage you in ways to continue to make you move forward, and what we want is to separate out the healthy uses of artificial intelligence with clients from the unhealthy ones, and that involves promoting critical thinking, that…

Katie Vernoy 19:20
Yes.

Curt Widhalm 19:21
It’s curious in as you’re talking about the ways that artificial intelligence is permeating the workplace and ways that you and I use it, and then I hear my young teenager at home come in and he’s, you know, spouting all AI use is bad. And if there’s a side that I’m going to have him follow on just in his own opinions, I’d prefer it to be on that side, but if there’s people who are engaging with this, it’s helping them to be able to discern; is this just telling me what I want it to hear, or is this actually providing me with advice? Now, one of the strategies that I use with my clients is let’s put in the opposite information and see if we get the same answers.

Katie Vernoy 20:14
Yeah, definitely.

Curt Widhalm 20:16
And this helps to recognize that AI will validate whatever side is going on, and it can’t be used as an objective truth. And with the couple of clients that I’ve done with this with, is helping them to come to recognize and be able to better convey what their subjective truth in a situation is. And for all my DBT therapists out there, hi. How are you? That’s exactly what I’m doing is helping them to build a better way of effectively communicating, so that way they’re not just going into their next interactions with their loved ones, with their dating partners, whoever it might be, with this idea that I am right, and this is the only way that things can be, but more: Here’s what I’m trying to say, and here’s what I’m hoping that you’ll be able to hear out of this from me in a more effective way.

Katie Vernoy 21:13
That’s awesome.

Curt Widhalm 21:14
This becomes a little bit more complicated when, I don’t know if you’ve had couples bring in?

Katie Vernoy 21:22
Oh, yeah.

Curt Widhalm 21:24
Now I don’t do a lot of couples work, and I haven’t had couples bring this in, but I can only imagine if couples are trying to say: this is why I’m right. Then it ends up being what I would imagine is all right, now I got to teach both of you to be critical thinkers about what you’re putting into a chat bot.

Katie Vernoy 21:48
Well, and I think it actually, the same exercise could apply. Have them put in their partner’s perspective, it helps them to think about their partner’s perspective, and then see how that perspective is also validated. So I think I think critical thinking is a clear one. I think the other thing is related to critical thinking and cognitive surrender is making sure that you ask the questions in session and teach your client to ask the questions that are going to get beyond what their chat bot friend has suggested to them. And we’ve got the whole AI deskilling episode that talks about how to do that for clinicians, but seeking out what is inaccurate about this? Where are the sources? You know, going and actually reading the original information that’s that’s creating these perspectives, talking about it with somebody who has their own mind and is thinking differently. I think it’s it’s something where if you can slow the client down and ask for their thinking, I think there can be a lot of rich work, because this is just another piece of information that’s flowing in. If they get too far into that reliance on the chat bot, I think then you might be looking at more addiction or dependence kind of interventions versus someone that’s just under-resourced, feeling alone and seeking out Chat or Claude or Gemini or whatever to think things through and feel less alone, and so I think it’s, it’s important that we recognize our responsibility here, because sometimes, especially with all of the sycophancy and the smart sounding words, and all of the things, the results that you get can be very seductive, and so unless you recognize that and see it for yourself and do some of these practices for yourself, you might go down that same rabbit hole with your client. I mean, this can also go into, like, the pseudo diagnosis or the misinformation spirals that can happen, where I give you all my information, and then it kind of keeps, you know, that confirmatory bias keeps coming up. It’s not the same as a Google search, it can, it can be worse, but it can also be better, because it’s a little bit more personalized. What are your thoughts on someone using an LLM chat bot to try to sort through information from medical doctors, or diagnosis, or concerns about that? Because I, I have mixed feelings, because I’ve actually used it for myself a little bit, but I think there’s, there’s a lot of caution that needs to be put into that.

Curt Widhalm 24:47
I think it goes with just about anything that you might be interacting with. My biggest, I guess, personal response around artificial intelligence is, if it’s about ideas that you know a lot about yourself, you see the gaps in the information that are being regurgitated back to you, and a lot of times it sounds kind of correct, and it might be somewhere in that C plus range as far as being able to provide an answer, and I think that people who are using artificial intelligence effectively are smart enough about the content area to know the types of follow-up questions to ask within artificial intelligence to be able to synthesize down and get better answers out of it. I think for a lot of people who are putting in information at first, if they’re accepting the first or the second kinds of material that ends up being responses back, they’re potentially at danger for not knowing what they don’t know, and so it really is how you interact with artificial intelligence, and to me this is falling where it ends up being something to talk about, but it falls outside of what therapy actually is to teach your clients how to use artificial intelligence in that way, and so if people are looking for more information about a diagnosis, you know, if you’re putting in your medical information, I would also talk with your client about, hey…

Katie Vernoy 26:28
Don’t do that.

Curt Widhalm 26:28
Those data, those data things that you’re talking about. But for being able to use it, a lot of times, what I direct my clients to is share with your provider, share with me, as your clinician, what questions you might have because of what you’re getting out of the AI responses. Don’t necessarily take it as fact. That it could be very, very true, but this is kind of teaching awareness around the limitations of artificial intelligence.

… 27:01
(Advertisement Break)

Katie Vernoy 27:02
Well, and I think I think the, the way that you described it is a really good one. It’s understanding where the use can be most effective. If you’re experiencing medical things or mental health things, and it’s complicated, it’s complex, you’re, you’re not understanding everything, putting in basic information, and asking questions about what does this mean, those types of things, that can be a good primer, like it’s not the end all be all, but if you prompted, and I got this from a class that I was taking, or starting to take in the master class program, so I will, I guess, link to that in the show notes too. But if you put some basic information and say, what do I need to know about this? What are three questions I should ask my doctor? Then the questions are what you take to the provider. It can help you be a more informed patient or client versus what do I do, what is what does this mean, what is my diagnosis? Because the large language model that you’re interacting with is only as good as where it draws information, and it’s going to reaffirm whatever it is that you’re focusing in on. Whereas a medical professional will at times, I have a little bit of my own issues with with medical doctors not necessarily seeing the whole picture, but an expert is going to look in the places that you don’t know to look and will ask you the questions that you don’t know to answer in order to get to a diagnosis, whereas the AI chat bot will take the information you have and put something together with it, and may miss something entirely, and it can be the same thing that I would joke, like, don’t go to Dr. Google, because you know every single thing that you have turns into, oh my goodness, you have cancer, right? So I think that using that skepticism and using it as thinking partner is a little bit too big, but I think it’s the closest that I can get to, which is, how do I come prepared? How do I know what questions to ask my doctor. How do I get more information? Where are some likely places for me to look at primary sources and read them myself? How do you summarize this thing that I have, and let me make sure that it’s accurate, and that I’ve given the chat bot appropriate prompts to not hallucinate, to not be sycophantic, that kind of stuff. I think there’s, there’s a lot there that’s positive, but if, if your client doesn’t seem capable of having that level of savvy, or is not recognizing problematic use, I think then there’s, then there’s additional, you know, kind of clinical beyond the scope of this conversation, additional clinical steps that you may want to take to support them in, in moving away from that over reliance on AI as a, as a support, as a, as a thinking mechanism, all of those things. So, so I think you’re right, I don’t think that this is, you know, we’re not, our job is not to teach our clients how to use AI, it’s knowing enough about AI to be able to address any clinical impacts and mitigate risk related to, you know, AI’s not good for crisis response, AI doesn’t hold confidentiality. It doesn’t call you out in the way a therapist will. It’s not sufficient for what we’re able to do and what our responsibilities are as clinicians.

Curt Widhalm 30:53
So now we also need to address for those clients who have already become dependent on AI.

Katie Vernoy 30:59
Yeah.

Curt Widhalm 31:00
And this becomes its own whole intervention process in and of itself, because we’ve gone well past the hey, this is a neat tool, and now this is my everything, this is my support system, this is my girlfriend, for some of the clients who are coming in. And I think that this is probably worthy of a much deeper dive in potential another episode, but it’s working on weaning off of AI. It’s still a lot of what we’ve talked about already, about validating the experience, about looking into the why, but it’s also trying to take the needs that are being met and trying to get some of the clients to experiment with some of those ideas in the real world. That for a lot of the relational aspects that even when I’ve heard well-intended clients say I’m trying to level up my social skills in order to be able to one day have a real girlfriend. We explored some of the ideas around what friction and disagreements and having to relate to people is actually an important part of building relationships and about being able to learn how to negotiate with people is a skill that artificial intelligence is not set up in order to help them learn how to do, and it’s going to create unrealistic expectations about what it is going to be like in a real relationship. Now, one of two things happens there, is clients will either agree with you or they’ll disagree with you, and you say this right here, what’s happening right now in your disagreement. These are the skills that you’re going to have to learn in order to be able to be in a healthy relationship with somebody. So, there is kind of that opportunity to do it right then, but I also wonder if there is a future market for therapists to become specialists in grief counseling for putting to rest AI girlfriends, that’s going to be somebody’s specialty someday. I’m coining it right now.

Katie Vernoy 33:22
So I think to wrap this up, make sure that you’re assessing AI use, make sure that you’re providing some psychoeducation on the risks, confidentiality, safety, sycophancy, relational harm, and then when you see some of these impacts, make sure that you’re addressing them as you would any third wheel that’s coming in and having a lot of impact on your client. Determine how you can replace some of these coping strategies with healthier coping strategies, how you can make sure that you mitigate great risk, all that stuff, and so definitely more to discuss on AI and clients use of AI, but we thought we would start with a a quick brief overview on what you might want to consider.

Curt Widhalm 34:16
You can find our show notes over at mtsgpodcast.com. Make sure that you follow us on our social media. Join our Facebook group, the Modern Therapist Group, to continue on with this and other conversations. And until next time, I’m Curt Widhalm with Katie Vernoy.

… 34:30
(Advertisement Break)

Announcer 34:31
Thank you for listening to the Modern Therapist’s Survival Guide. Learn more about who we are and what we do at mtsgpodcast.com. You can also join us on Facebook and Twitter, and please don’t forget to subscribe, so you don’t miss any of our episodes.

 

/0 Comments/by
0 replies
SPEAK YOUR MIND

Leave a Reply

Your email address will not be published. Required fields are marked *