Evan Dumas  

You’re listening to Group Practice Tech, a podcast by Person Centered Tech where we help mental health group practice owners ethically and effectively leverage tech to improve their practices. I’m your co host, Evan Dumas,

 

Liath Dalton  

and I’m Liath Dalton and we are Person Centered Tech.  This episode is brought to you by TherapyNotes. TherapyNotes is a robust online Practice Management and Electronic Health Record system to support you in growing your thriving practice. TherapyNotes is a complete practice management system with all the functionality you need to manage client records, meet with clients remotely, create rich documentation, schedule appointments, and bill insurance all right at your fingertips. To get two free months of TherapyNotes as a new TherapyNotes user go to therapynotes.com and use promo code PCT.

 

Evan Dumas  

Hello and welcome to Episode 530: When Clients Use AI for Therapy: Risks, Realities and Conversations We Need to Have.

 

Liath Dalton  

Conversations we need to have, indeed. And I mean that–we mean that in two senses, actually; that we need to be having this conversation among the professional community, amongst mental health therapists, and then therapists to be having this conversation with clients. So we’re going to get into all of that, because the reality currently is that clinicians have really been focused primarily on their own ethical and legal use of AI, or looking at evaluating risks and benefits. Is it useful to use in a clinical context, or is it more appropriate at this point in time to just be using it for operational and business support, et cetera? And the conversations that have been oriented to clients have been more focused honestly, until very recently, from what we’ve seen and heard on sort of fears around being replaced as therapists by AI, right? And evaluating it from that perspective of, is it really a threat–an existential threat or not. But what we need to be talking about as well and not overlooking, is how clients who are in therapy are using AI for therapy, emotional support, guidance, etc. Because the reality is that is happening. 

 

Evan Dumas  

Yeah, they’re already doing it. 

 

Liath Dalton  

Yeah, they are. They are doing it. And part of the reality as well is that I feel like every day we are seeing additional news stories about AI gone awry, right? AI psychosis, AI being supportive of delusions and reinforcing delusions or recommending tall bridges to someone expressing suicidal ideation, right? This is consequential stuff, like, really consequential. So what we want to talk about is just that that landscape of client AI use. Some of the evolving laws, because Illinois has just passed a ban on the use of AI and AI chat bots for therapeutic use. Nevada and Utah are trying to follow suit, so we’re starting to see more regulatory and legal oversight because these risks are so significant, and we’re seeing them kind of being realized and playing out. And then we’re going to talk through some practical guidance for practice leadership on how to address this with your clinical team and how to support them in having these important conversations with their clients. 

 

Evan Dumas  

Yeah.

 

Liath Dalton  

So we’ve been seeing that clients are using AI, not just for, you know, planning trip itineraries or helping with meal planning, and/or their work and some of those more innocuous uses. Folks are going to AI for a lot of emotional reassurance, advice on personal and relationship issues, using it as a virtual counselor. There was a recent NPR story where ChatGPT was tested out as a couples counselor, which was really interesting. I’m going to be putting the links to the specific news stories we reference into the show notes, so you can do a deeper dive, and they would also be really good to use as source material when talking to your clinical teams about client use of of AI, because it provides some specificity, and then that specificity, in turn, can shape how they might broach the subject with clients, or, you know, be an opener. They can be like, oh, I, I saw this news story, and it made me wonder about how my clients might be using AI, and if that’s feeling supportive and helpful, or or how we can work with that effectively together in the therapeutic relationship, etc. Evan, have you heard some interesting reports, in addition to this about how AI is being used, or any anecdotes from folks in your life?

 

Evan Dumas  

Totally. So AI has been around a long time. You know, recently it’s called AI. It used to be called chat bots, and so people would converse with chat bots, and it was really just automated systems to sort of guess what you’re saying and respond in kind. And that is just sort of what AI is now, it just has a lot more material to base its responses on. But some paint, some like, like articles that I’ve read and seen–there was a guy recently who, AI advised him to replace table salt with sodium bromide. But then he got what’s called bromide intoxication, which is like psychosis, and he had to be, like, hospitalized for psychiatric things. And like, you know, they took this as well meaning, well researched advice when it was really just some computer throwing words together, saying, yeah, it sounds about right. let’s send this. You know, no factual basis, no doctoral degree, nothing yet, still very much trusted. Like people who put AI on high–I’ve even heard that some people are, you know, starting a religion around AI saying, yes, this will save us when you know, it’s still made by us. It’s still made by people. It’s still very fallable.

 

Liath Dalton  

Oh, my goodness, that part I had not heard, though, that is fascinating and something I will definitely be reading up more about because, wow, that’s that is quite something. Well, again, we also know that folks are using it like some of the the data so far is that half of Gen Zers are using AI for dating and relationship advice, and I think you know so much of that is taking place online in the first place that you know, having AI review profiles of potential matches, and then analyze the text communications between the asker and, you know, potential partner, or whatever is occurring. So it’s just like it’s proliferating through kind of every aspect of life. And we know that there are a lot of risks that come along with that right. We already spoke to some of the more extreme ones in terms of it recommending suicide or facilitating harmful behavior, reinforcing delusions and hallucination. But something that I think is also really impactful, even if it’s not giving negative advice, and sometimes especially if it’s not, is that AI is programmed to be sycophantic, right? 

 

Evan Dumas  

Oh yeah.

 

Liath Dalton  

To keep you engaged with it, and to provide that validation. So it will typically align with what users are saying, which can reinforce unhealthy thought patterns. But even if it’s not reinforcing unhealthy thought patterns, creating a dependency on that validation, or, you know, setting things up to kind of have an addiction to that relationship, or sense of relationship where so much positive feedback is being given is another risk, as well as eroding self trust. That’s something I’ve heard clinicians name as a concern about how client usage of AI and reliance on AI could, you know, jeopardize the development of internal resilience, right? And then, of course, there’s the bigger piece, or not bigger piece, but a piece that everyone listening to this will already be thinking of, because we’ve addressed it in terms of clinicians, use of AI, which are the confidentiality gaps right? And these consumer facing AI platforms, generative AI platforms, like chatgpt, are not subject to HIPAA, do not have any of those privacy protections in place, so what folks are inputting into it is not confidential and private, and there are, of course, real risks to the information that is being input into, AI, than just being out there, right? So it’s a big, big deal. Of course, we have to also name that there are, just like with clinical use of AI, there are some potential, you know, opportunities or benefits in terms of how clients might be using AI. And these really intersect with a lot of the reasons why clients use AI. I mean, one thing that can be useful and is more innocuous than other use applications is using it as a tool for journaling, practicing dialogs or organizing thoughts, but that all needs to be really considered and intentional, right? And that’s part of why it is prudent to be having this conversation with clients and trying to bring whatever ways they are using AI into the therapeutic space, so that you as a clinician or the clinicians on your team can be kind of monitoring how it’s impacting them and helping them to use it in safe and constructive ways, and if it’s getting out of hand and having negative consequences or exacerbating existing challenges, to be able to recognize and address that as best as possible, right? Evan, I think it merits talking a little bit about why clients are–who are in therapy and like, getting support from their therapist, might use AI though, right?

 

Evan Dumas  

Oh, yeah, totally. It’s fun. It’s always available, like, two in the morning, you can ask AI the most weirdest, random questions, and you can have a dialog with it, and it kind of scratches that itch of communication, even if it is just a computer. But, you know, therapy is expensive. Therapy is every once in a while, and so AI is pretty much always available. Sometimes it’s, you know, just way more responsive. It’s way quicker. And, you know, some people definitely use it to just bounce ideas off of, like, journaling, practicing, organizing thoughts, stuff like that, just as a sort of like, hey, how does this sound? Or what’s an alternative take on this? So you’re not directly like, what should I do? But it is sort of just like a soundboard. Who knows what you’ll get out of it. 

 

Liath Dalton  

Right. 

 

Evan Dumas  

But it’s always there. They all have their own flavor. And, yeah, yeah, it can be fun.

 

Liath Dalton  

And I think something that a number of clinicians have brought up too in terms of reasons they think or have had clients report to them that they are using AI for sort of therapy related purposes or therapeutic conversations, is that lack of, that sense of lack of judgment and safe space, right, like disclosure of deepest darkest secrets, feels somehow safer when it’s going to a not real human right? Into a computer than to another human being, and you want your therapist to hold you in high regard. And so sometimes disclosing the darkest stuff or the hardest stuff you know really gets held back for a long, long time. So that’s something that we know is influencing and a contributing driving factor to why clients are using AI in this way. All right, so now that we’ve kind of set the stage, what are some things that clinicians can do to address this, to have these conversations, and help sort of manage the risks within the therapeutic relationship.

 

Evan Dumas  

Yeah, well, you could talk about it. You can normalize it, like if people are going to AI because they’re so ashamed of how they’re feeling, you can reduce that shame by normalizing the use of AI, you know, talk about best practices, safety planning, other things. But you can also just be like, hey, is this–is this even a thing? Talk about it just like, you know, be curious.

 

Liath Dalton  

Exactly. Curiosity is key in this, and so is normalizing the disclosure and not reacting to the disclosure as though it is a existential threat, right? Asking follow up questions to explore impact, like, what are, what’s their emotional response? How is it impacting trust in their own thinking and assessment of circumstances, right? 

 

Evan Dumas  

Yeah.

 

Liath Dalton  

And it’s also important to weave into that some context and boundaries, like clarifying that AI isn’t a substitute for for therapy, referencing things like limitations on privacy risks, for example, lack of empathy, like real empathy, Right? There’s the performative empathy of AI, but that is qualitatively different to human empathy. And then I think, in that vein of curiosity, really inviting collaboration, something like, let’s unpack what the AI suggested, and how does it or how doesn’t it align with your goals or values, etc. And it’s also important to be including the use of AI and what’s revealed in these discussions to just include that in the client record and documentation. That’s something that’s protective for the clinicians as well.

 

Evan Dumas  

Yeah, yeah. I think there’s also space to just celebrate clients doing the work outside of sessions. So, you know, AI is rife with potential issues, but clients themselves, being curious about themselves, asking questions, seeking guidance, you know, bibliotherapy is when we recommend self help books or novels or poetry or anything to clients. And I think at least for the aspect of you know clients really caring about learning more about themselves and growing outside a session that can be celebrated–now we can talk about how to use it well. It’s a practice, but at least there’s, there should be room for like, hey, good on you, for like, caring about this and not waiting every other week to talk about it.

 

Liath Dalton  

That’s a really excellent point Evan, and I think kind of goes to how to, like, lean into this and use it as something that is, you know, grist for the therapeutic mill, right? And where it doesn’t just have to be a fearful reaction or avoidant reaction, the bringing curiosity, finding the threads within what clients report that are positive and that could be harnessed in in other ways, within the therapeutic process, and your work together is an awesome approach. So you know one one suggestion in terms of practical guidance going forward for clinicians might be to incorporate an AI related question into intake forms. 

 

Evan Dumas  

Definitely 

 

Liath Dalton  

Right? To include some provision like psychoeducation about AI risks and limitations. But something that I think is really, really important, more so than those other two points, is encouraging clients to bring AI content into sessions for review, right? 

 

Evan Dumas  

Yeah.

 

Liath Dalton  

If they’re interacting with it, we don’t want them interacting with it in a vacuum and separate to distinct from the therapeutic work they’re doing with the clinician, or feeling like they cannot disclose or process what is brought up in that use of it with their therapist. I think that is, like the most important point about all of this. So check out the resources section of the show notes, because there are going to be two items in there. One, a clinician conversation starters–little prompts, basically for talking with clients about AI use. One section is looking at kind of how to invite disclosure and exploration without shame or defensiveness, of course. And then some follow up prompts to deepen reflection and examine impact, you know, looking at how AI use affects emotional well being, self concept relationships, etc. And then some effective therapist responses and guidance. Ways to really respond when clients share about their AI use that balances how you address risks, cautions and constructive guidance. And this is very much just a starting point template, we highly recommend that you use it basically as a rough draft, and modify the questions to reflect your practices, therapeutic approach, or kind of tied into what your modalities are, etc, but it is a starting point so that this can be framed and defined and then incorporated into practice. And then the other resource is basically a leadership checklist for how any clinical supervisor or director in your practice can guide your clinicians on client AI use. So addressing team awareness, providing them with tools, integrating all of this into the practice infrastructure. And of course, always goes back to culture right, fostering a culture of curiosity and safety, and just making this a very intentional and explicit part of both the therapeutic work that clinicians are doing, but as part of that, it needs to be something that the clinical leadership is overseeing and has kind of a systematized way of addressing it. So check those resources out. We hope you find them helpful and that this conversation spurs more conversation within your practice, so that those important conversations with clients can be had and can be had in a way that is really constructive and generative, that reflects, again, that all important curiosity, rather than the kind of fear based response that I know understandably came up for a lot of folks when we just started seeing stories about how the AI might be just replacing therapists. And as a last note, because that’s not the note I want to end on, interestingly enough, there was a big Washington Post article last week talking about how all of these folks who are in professions that have been really diluted, shall we say, by AI, are becoming therapists, including, like screenwriters and actors who were becoming therapists, and in that, one of the stats that it included was that there is expected To be an 18% growth in jobs for licensed mental health counselors by 2033 right? And the end of the article was, there just some things AI can’t replace.

 

Evan Dumas  

Yeah, definitely, yeah. 

 

Liath Dalton  

Which is so true, but I’m glad we’re starting to see more of that, and hopefully that helps address some of the concerns and some of the like initial reactivity that was was generated, yeah, yeah. So check out the show notes, because it’s going to be packed with resources and links to all these other interesting stories. Thanks for listening, and we’ll talk to you next week.

 

Evan Dumas  

Yeah, talk to you next week everybody.

 

Liath Dalton  

This has been Group Practice Tech. You can find us at PersonCenteredTech.com. For more podcast episodes, you can go to PersonCenteredTech.com/podcast or click podcast on the menu bar.