Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Psyched to Practice

Psyched to Practice artwork

Beyond the Bot: Should You Trust AI More When You Know Less

Sep 11, 202634 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Are we more likely to trust AI when we know less about a subject? Dan and Adam explore the surprising gap between how experts and novices use AI, and what that could mean for mental health and school psychology. They talk about the risks of relying on AI when you do not know enough to spot mistakes, the importance of using multiple sources to check AI-generated information, and why AI should support professional judgment rather than replace it. The conversation also covers major AI developments discussed at the APA conference, including AI report writers, electronic medical records, school data systems, AI agents, and recent changes to Google NotebookLM. Along the way, Dan and Adam consider a bigger question: How can professionals use increasingly powerful AI tools without losing the skills and critical thinking that make them experts? Website: https://www.learn.psychedtopractice.com Social Media: https://www.facebook.com/PsychedToPractice https://www.instagram.com/psychedtopractice/ https://www.linkedin.com/company/psyched-to-practice/ “Be well, and stay psyched"

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

SPEAKER_01The following is a site to practice production.

SPEAKER_00Hey, welcome back to Beyond the Bot as we delve further into AI and mental health, uh, specifically more school psychology. I'm Dan Florell from Eastern Kentucky University, professor there. And with me I have Adam. Adam, you want to do a little introduction to any new lessoners we have today?

SPEAKER_01Um, I'm Adam Lockwood at Kent State University. Uh the only thing I wanted to plug was in our episode we talked about AI agents. And um I got so interested in the topic that I'm doing a series on using AI agents. Um I think it's gonna be four or five parts. Uh it's um I hope it's really useful for people. Using agents is much different than prompting the chat bot. So there's examples, there's screenshots.

SPEAKER_01Um hopefully it's of use to people um who are interested. I've also found that it really is touching off a lot of um strong feelings, uh some strong negative feelings on social media too, that people are like, I do not want uh to use it for this purpose, and to those people, that's fine.

SPEAKER_00But for people You mean just because there's been headlines like rogue AI, hacks, uh other AI agencies and things like that?

SPEAKER_01Because there's a lot of risk involved. You know, anytime you let AI go off and do things autonomously, there's risk involved. I talk about how to do that safely, but that's why you gotta be careful.

SPEAKER_00Do you bring up the genie and their three wishes? You know, the genie where no matter what you ask for, it's gonna end badly because you it like finds the loophole and literally operates, you know, like what you want, but that's not really what you wanted.

SPEAKER_01I do not.

SPEAKER_00I can almost do anything, but you got to be really careful about how you specify what you want it to do. And of course, unlike the genie, of course, you have a little bit more uh control over the AI agent with your instructions and monitoring, but I think that would be a word to the wise if people just think you set it and forget it with an agent, an AI agent.

SPEAKER_01Agreed.

SPEAKER_00Well, I uh look forward to reading the rest of those blogs. I actually, Adam, our first episode that did go over that uh inspired me to start looking into opening it up, and then I ran out of time and then I got distracted from a little thing called the APA conference. But um I tell you, you know, one of the things I've also learned to do is when dealing with AI, please admit your own ignorance as to what you're doing, and then ask the AI to kind of help shepherd you through.

SPEAKER_00And if the first initial instructions of the AI are still too high, then say, hey, look, look, I need you to break it down even more. It will do that. Uh you have to you have to check your ego just a titch when you're using some of this AI to really get it working as best as you can, because none of us are total experts in AI. Like there's too many new things that pop up. And it's okay to admit your own ignorance and continually breaking it down for you so that you ultimately can understand or get to the the destination that you're looking for.

SPEAKER_00Uh we would you agree with that, Adam?

SPEAKER_01I think AI is a wonderful tutor. I think um whenever I don't know how to do something, I definitely reach out to have AI help me. If I'm setting up a new agent, I help it, I have it help me do that. Um it's actually what the people at uh OpenAI have told me that they do as well. So yeah, and you shouldn't have an ego with AI. It doesn't care. So don't make a fool out of yourself or sound stupid.

SPEAKER_00I mean, yeah, until you run a personality check and it's like, this guy's an idiot. He doesn't know what he's doing. You know, you're like, because you've anthropomorphized the AI so much, you're like, I I don't want you know, it's like going up to your hero or something and saying, Oh, I so admire your stuff, and you don't want to look like an idiot in front of them kind of thing. Uh I imagine there's got to be a little bit of that for some people, uh, you know, when you're thinking about like, I don't want to look bad in front of the AI.

SPEAKER_01I mean, that's never crossed my mind. I've never cared. But you know, I'm not so surprised when I hear people are proposing to AI and all sorts of weird stuff. So it takes all kinds of yeah.

SPEAKER_00Well, um, switching topics a little bit back to my APA conference. Um, that happened in the first week of August. And uh let me tell you, if you thought that uh AI has been ignored in psychology, it has not. And I think we are uh at least currently at peak apex AI saturation. Uh, there were a ton of AI-oriented programs uh being offered at APA this year.

SPEAKER_00And uh my main takeaway, Adam, is that you and I seem to be about at least uh where most of the other experts that they had presenting were. Um when you go to a conference, you never want to appear like you know, as the expert and then seem lost about what was going on.

SPEAKER_01Well, I gotta say, with this technology, I always feel like I'm just chasing, trying trying to catch up. And so I'll myself, but I don't feel like uh I'll ever be on top of this technology.

SPEAKER_00Well, it was interesting. I think um we're still in the philosophical phase of uh adoption, which means I think there are a lot of people that haven't really used AI a ton, but they are opining a lot about what we need to be worried about regarding it. Um, so you know, and I think that's very wise. Anytime that you're gonna go into a real powerful technology, it's pays to go ahead and take a pause and think about the ethics and uh what are the big issues and regarding our own skill levels and things like that.

SPEAKER_00And I think that was coming through. I, you know, there were a lot that were labeled like ethics of AI, implications for, um, and those types of things. But I didn't take away like new techniques like your AI agents. Nobody was presenting on AI agents. Uh, there was uh also in the exhibitor hall, as opposed to the National Association of School Psychologists Convention in February, where you probably had about seven or eight AI report writers. Um, there was really only one AI report writer, the one from PAR, and that of course wasn't their whole focus.

SPEAKER_00And then the rest of the AI-oriented products were for the most part um electronic medical records or EMR systems. Uh now, obviously, the APA conference is much more clinical-oriented, though. Um, to be fair, the APA conference is probably one of the last omnibus psychology conferences out there that covers everything from experimental uh to psychophysiology to counseling clinical school. And so you do get kind of an eclectic uh group, but the EMR systems, there must have been at least seven or eight of them there, you know, all touting similar things, um, transcription services, of course, help with the note-taking, um, insurance billing, you know, a lot of the background productivity kind of oriented things.

SPEAKER_00Um, and I think that part of the field has really started to mature. You're beginning to see some leaders. I think our AI report writers, we're still seeing a bit of a wild, wild west period before those begin to coalesce a little bit. And of course, uh, and this isn't even our topic for today, uh, Adam, but Pearson uh has started to make noises about their report writer coming out, which of course in the publishing world, huh?

SPEAKER_01Should be launching right here uh at the beginning of the school years, what I have been told. Yeah. What so what EMR systems are these smaller? Because you know, I think of like Epic, you know, so big hospital ones that that are enormous and and and I think have the vast majority of a physician and hospital clients. So these must be more geared towards you know, small practices, private practices, things like that. Is that right?

SPEAKER_00Yes, I believe so. And more clinic oriented, so you know, community mental health, um, larger clinic organizations, private practices, uh, very mental health oriented, though. You're right. I mean, and Epic is so huge, and um frankly, I think Epic works fairly well. I mean, I don't I don't use it a ton. I actually listened to a podcast with the CEO of Epic, um, and uh it sounds like they're trying to come from the right place. Um I don't know how much that influences in school psychology.

SPEAKER_00You know, school psychology, we have these more school-based systems that uh alarmingly don't seem as focused on the security aspect as the AEMRs of the hospitals, though I imagine uh they will start rolling that way. But have you been struck by these school systems and kind of their their lack of security in a lot of them?

SPEAKER_01I've been struck by the clunkiness of them, by the lack of integration. You know, people are cutting and pasting two, three times. And the thing I'm always shouting to any company that that listens is the first one that creates a hub that can integrate report writing and all the different documentation and have it automatically populate um the you know, whatever IAP Pro or whatever sort of other systems they have is going to be positioned really well.

SPEAKER_01It's like an epic for school.

SPEAKER_00Yeah, no, I I totally agree. The field's still ripe. Um, they really haven't done a good job with that. Actually, I wrote a communique article on uh a system that did help kind of cross all of these little silos to integrate information. Because you and I both know, having worked in the schools, they have reams of data that just sits there wasting away. Uh schools are awesome at collecting data and they are horrible at analyzing it from a kind of composite standpoint.

SPEAKER_00I think they do fine at maybe a grade level or a classroom, but when you're looking at a whole district and trying to see trends, I think that they have the data that could show them that, but they don't have the systems that enable them to do it well. And I think maybe that's where AI can even um come into the fore in regards to it.

SPEAKER_01Yeah, still a lot of districts that are they got paper files, you know. You go into the file cabinet and you you start there, which is just crazy to me.

SPEAKER_00Well, it doesn't feel too uh too bad for me. That's uh that's kind of the reality of what I um grew into the field with. And then it's gradually gotten into this more. And I'm not claiming that the paper system is the system that we wanted to go with. They were so much missing data. And of course, if that student had changed schools, it was a disaster in regards to um trying to put together like where's this kid been? Where are they at, you know, as far as academically? It was quite the challenge.

SPEAKER_01It's it yeah, it still is to request out-of-state records. I'll tell you, I've moved from state to state, and they go into my chart and they go, Oh, we have everything, your we know your medication, they know everything instantly. Can you imagine if we had that in school psychology? You got a new student and you didn't have to fax and do all this craziness to to get their their records and see if they qualified in your state, and uh that would be amazing.

SPEAKER_00It would be, but uh, you seem to be striking me as a nationalist, don't you know? Schools are a local thing and we do our own thing, no matter how inefficient it is going to be.

SPEAKER_01Well, you still could move those across. You can still have your own states' rights, but the way to share that information uh more seamlessly would make sense.

SPEAKER_00Yeah, I I just like poking you every now and then, Adam. I I totally agree with you. Illinois has like a million districts uh in there, and they don't use a unified system even within their state. So I can only imagine how clunky it is for those poor school personnel in trying to track students. Whereas Kentucky, I will give them this a long time ago, they uh invested in an infinite campus, so everybody has infinite campus, and so at least if you transfer within the state, stuff is pretty seamless in regards to the student being transferred.

SPEAKER_00Now, that doesn't help when people go back and forth between the river in Indiana and or Ohio, and those systems, of course, do not cooperate with each other, so there's still some of that missing child information going on. But if you needed the utility of it, if you look within the state, yeah, you can really do some decent stuff. Um, it's not an endorsement of Infinite Campus, it has its own flaws, but uh anyway, just an example of what you're talking about, huh?

SPEAKER_01Checks.

SPEAKER_00Right.

SPEAKER_01No endorsements without their sponsors.

SPEAKER_00Yeah, I know. We need more sponsors. Maybe too much to expect by episode four. So um getting back to APA conference and what I wanted to talk, at least in our remaining half of our episode this week, uh, is the idea. I went to a talk. It was a kind of symposium where they had two or three people come out, TED Talk style. That was the other thing I didn't tell you about, Adam. They've adapted the slick TED Talk kind of style at APA for at least their big ones that they show on virtual.

SPEAKER_00I don't know how I quite feel about it because it feels almost too overproduced for a conference. Because a conference, to me, a scientific one anyway, needs to be like ultra geeky, a bit awkward, a little bit in the weeds about stats or something. And these were these were finely produced. And the reason I guess I have a problem with the finely produced is I don't get a sense that you can get any depth from it because it's bright and shiny. There's no there uh in some cases. But all of that being said, uh I went to one presentation called The Largest Behavioral Experiment in Human History, and it was by Kieran Snyder, who was with Microsoft.

SPEAKER_00And she is not a psychologist, but she was really talking about a variety of issues. But one of the big ones was are we experts or novices and how that influences our view of AI? And so what her point was, and she had some really nice little stories about her daughter writing some little novel when she was like 12 and heard about the possibilities of AI's writing and getting very upset, saying, uh, they don't understand how hard it is to do this work on your own and the process of it, and you should never use AI for it.

SPEAKER_00But then she goes on that her daughter went by two weeks later and had happily uh put in graphics to her novel using the image generating AI. And when asked about this kind of duality, she just responded in a very matter-of-factly 12-year-old mold, which is, well, I can't draw, but it can. And so uh they went on uh and made the point that we, as experts in whatever particular area of expertise that we have are dismissive of AI.

SPEAKER_00And for good reason, uh it has its flaws, right? It talks about hallucinations, uh, maybe some inaccuracies, lack of awareness of cultural values of an area, uh, you name it. And uh APA had plenty of presentations on those drawbacks. I totally agree. I feel the same way. However, I also see the point of the novice argument because it's in all of its flaws, it knows a lot more in all the areas that I am not an expert, in which I am a novice.

SPEAKER_00And that's what the data that Microsoft had shown was is that people are very leery in the area that they have expertise with AI, but as novices, they're much more accepting of using it. So uh I kind of sprung this topic a little late on, Adam. So I'm gonna let you chime in, though, Adam, on what you think about that kind of premise that we are more accepting of AI if we are novices in that area, and that whether that has an impact in the field of practice, particularly like school psychology.

SPEAKER_01Well, I think there's a number of layers there. I I think one thing we have ownership over our jobs, and so if we see that, you know, we're gonna be extra protective of that, like that doesn't do as good a job as me. Um but I think of the implications for parents going to AI um and asking questions and not really knowing how good the output is, um and what sort of issues that will that will bring that people are already seeing, I'm sure.

SPEAKER_01Um because they have no problem um asking questions and they don't really know what they don't know. Um and then we're pretty critical of it, I assume, uh, for good reason. But that's gonna create some tension, I think.

SPEAKER_00Yeah, I agree. I know that it's frequent now that people will go to AI for second opinion, say, for medical or even psychological diagnoses. Like they'll upload the reports or products that they've gotten from the professionals and then ask it, the AI, is this correct? Do you think it's this? Um we always have sought second opinions, especially with news that we aren't necessarily agreeing with.

SPEAKER_00And I think that's healthy. You know, there is no one expert that's going to give you the definitive answer for most of the stuff regarding humanity. And so it's good to check. Uh, but that over reliance and assuming that AI is the equivalent of the dictionary and has the end-all, be-all definition, and that it's like the the great arbiter. Uh, I'm a baseball fan, and I don't know if, Adam, you're aware that this year they have the ability to challenge called strikes and balls with umpires, which in in my mind is the second opinion.

SPEAKER_00And yet, the way it's presented is well, this is actually the computer is the expert, and that is the default perfect diagnosis of a ball or a strike.

SPEAKER_01But they do not have it has to be challenged by a human to go to that system, which is what they didn't do in minor league baseball first, which I like that it yeah.

SPEAKER_00I like keeping the human aspect in in the game like that, that adds some flavor. Now, I don't know how I would feel though if it was my medical diagnosis. Uh but I don't have you um in the medical field, there is a startup called Open Evidence. I don't know if you're familiar with that, but for those of us uh that you haven't heard of it, if you have an MPI or national provider ID number, you can sign up for this program, AI program called Open Evidence, which um is a relative closed system that relies on a lot of journal articles from a number of high prestigious journals, and it gives advice to physicians regarding possible diagnosis given certain symptoms, possible interventions, what's the latest and greatest.

SPEAKER_00Uh, there's a lot to be said for it. Uh, one of the big things that we know for a long time is the um gap between research and practice, and something like open evidence promises a fairly uh good outcome in regards to closing that gap. However, uh it gets once again into this idea of expert versus novice usage and some of the other things we've talked about in this podcast as far as de-skilling and other things.

SPEAKER_00Uh, but what's your feeling about this whole expert versus novice AI usage? Uh, I can even see one concern is if I'm in graduate school learning a field, I am a novice. So do I start to rather than default to my professors and what they're telling me in class, that I default to what the AI has come up with.

SPEAKER_01Well, one thing is we talked about open evidence, I think, on the last episode. And if you remember that there was a study in nature medicine that open evidence wasn't as good as the other frontier models, which is which is interesting. Um as far as what's the thoughts of Novic is using it. Um, I'm really I'm really nervous that people aren't going to build the skills they need if they just rely too heavily on AI.

SPEAKER_00Well, I think the field, our field, is ripe for something like an open evidence equivalent. Uh there's some entrepreneur out there that's got to be pursuing something similar for a clinical psychology field like there is with the medical field in open evidence. Um that's just seems like it's too big of a pie to have left. Of course, there's not as much money associated with it compared to, say, a physician um level, but it's not inconsequential either.

SPEAKER_01Uh I don't think that's money has enough utility in it. I mean ChatGPT 5.6 on high will do a better job than any um fine-tuned model, any model that's specific to school psychology. And then if you use something like Notebook LM, that's a Chival augmented generation model that pulls from source documents only. Although, did you notice that they updated it now that it will do a web search as well?

SPEAKER_00Yeah, yeah. I'm preparing for a presentation exclusively on kind of Gemini and Notebook, and it's called Google Notebook, I think, now. Yeah, I was using it and I was very chagrined that it uh had the web capability. And I I have to do a little deeper dive myself to see if there's a way to exclude that. Because to me, the whole reason I was using Notebook is that it had a you know a captured environment. And if it's going out rogue on me and also bringing in stuff from the web, uh now I I'm back to I might as well just use the general model uh and uploaded stuff to it.

SPEAKER_01Just to explain to people who don't know, um Notebook is a product by Google, and you can upload hundreds of documents, websites, YouTube videos, etc., to this model and then ask questions of those specific documents. You can create flashcards, um podcasts, videos, and it until now had just pulled from those materials. So it was very reliable, and you it would point you to where the information came from.

SPEAKER_01Now it's also integrated this web search uh capability, which to me I do not understand why they did that. Um but yeah, maybe before the next time we meet, I'll look and see if there's a way to turn that off. But when I when I saw that, I was like, what?

SPEAKER_00Yeah, when I first saw it, I specifically said in my instruction to it, do not only look at the uploaded resources, do not do the web. But once again, then I'm in a world where I'm why am I using notebook? I might as well just have used the general search models, you know, and I'm just use Gemini or Chat GPT, that's essentially what it is. And so uh I also am already ahead of the game, Adam, on you, on that question because I've I've been having the AI doing some deep research. Uh, and my Gemini notebook is going to be the next thing I'm delving into because I've got to train some school psychologists on how to use this with report writing, and I want to give them the latest and greatest and not uh misinformation because they've changed it.

SPEAKER_00What would you say? This changes probably just within the last month or two. Uh it certainly something I just noticed.

SPEAKER_01I'm pretty sure.

SPEAKER_00Um yeah, yeah. So that's some breaking news that that we have for you on that. Um, so what's your feeling about uh your findings about distrusting if you're an expert, but then wholeheartedly embracing if you're a novice in an area? Do you have any kind of final thoughts on that aspect of it as we're getting close to the end of our podcast time?

SPEAKER_01I mean, that's not a new phenomenon. It's true of any source of authority, you know, like so many people will trust whoever has good marketing and doesn't know what they're talking about at all. And it's only the people who are truly experts who will be like, wait a second, um, this is BS. I'm calling BS on this. So it's it's not anything new as far as from a social psychological standpoint. It's just um this really capable, always on, always accessible, um expert.

SPEAKER_01Um so it just kind of takes it to another level.

SPEAKER_00Yeah, I just I'm curious as to how all of us novices, which is everybody, because we're more novices in more areas than we could ever be experts in, how do we tap into the hesitancy of experts in their particular area to use AI to be helpful, but not necessarily all trustworthy? Like, in other words, how do we spot those deficits? And would it be better to have the AI when we're wanting it to do a task, say first, what did the experts say about your deficits and being able to tell me about this?

SPEAKER_00Would that be a good way anytime that you start in an area that you just know nothing about, so that it at least makes you aware when it starts talking to you? Uh, you know, the sycophansy part of it really gets to me every time. Every time I start something out, it's like you're a little kid again and you master everything and you get real upset if you have any negative feedback. You know, it just gives you positive, positive, like that's a brilliant insight. But then when you contradict it, it goes, Well, yes, you should that really is a great thing that you notice. And I'm like, no, I shouldn't have noticed it, you know.

SPEAKER_00Um, would that be logical, or is that just making things too complicated when people usually aren't looking for that level of complication? They just want answers.

SPEAKER_01Well, you know, it's all it comes down to how harmful is it going to be if it's wrong? If it's something high stakes, you want to be very careful. But in general, I always use one, I pit one model against another, and there's been studies looking at the ability to reduce hallucinations, which is what we're talking about, just generation of misinformation. And you can reduce it by like 96%. If you take, let's say you got output from Chat GPT, and you're like, I don't know if this is correct, and then you put it into Gemini and say, I want you to be a skeptic, look for poke any holes in this that you can, you can really bring down those hallucinations and be a lot more certain.

SPEAKER_01And that's the approach that I take.

SPEAKER_00So, really, just kind of a critical thinker as a user and bouncing the models off from one another. I know that's certainly what I engage in too when I'm preparing to write something. I'll have both Jim and I and ChatGPT do deep research on a topic, and then I'll have one go ahead and help me, you know, with a draft, and then I'll bring it back to the other and say, this is something I drafted, but what am I missing in this? And between the two, you know, I figure if they're giving me the same information, maybe I'm the one hallucinating in what I think.

SPEAKER_00It's a lot uh like our training as psychologists to when we do assessments. I mean, it's always ideal to have two perspectives on, say, someone's behavior before you would want to make a generality of it, if you can. Uh, we aren't always blessed with that type of information, but that's what we're taught. And then we're also taught to try to get at the same constructs in different ways. So maybe you observe them and then you also have the rating scales. And if all that coincides, you have a lot of more confidence that you have accurate information.

SPEAKER_00So if you've been trained as psychologists and you're listening to this, um, you know, just tap into what you've always been taught and regarding the AI models in the same manner. And I think that's going to steer you to more satisfying outcomes than um otherwise.

unknownYeah.

SPEAKER_01That and use the most updated models you can. Because they just keep getting better and better. And turn your reasoning up to extra high if it's really important. Um because if you use Chat GPT Soul, for example, which is their best model, and you crank it up to extra high, it will really think and it'll take a long time, and it's more likely to get a good answer.

SPEAKER_00Yeah, that's good advice. And I've noticed when I do the Gemini versus ChatGPT on even the deep research, chat takes at least double the time uh to get out the answers. Uh, but it tends to be a bit higher quality than what I see with, say, a Gemini Deep Research. Both provide good service. And once again, like you've made earlier the point, you know, what are you really looking into? If it's some uh who has the best chicken sandwich, I'm not really too worried about if it's the most validated versus uh what is the most empirically supported IQ test or something?

SPEAKER_00I'm gonna want a lot more depth if it's gonna give me an answer like that.

SPEAKER_01And so that's where we're at in the in the AI field. Is most that for most things it's good enough. If you look at benchmarks, ChatGPT, Claude, their better models are killing Gemini. But you know, are you having them do new math proofs that have that have never been solved before, like ChatGPT is? No, you're having it uh help you determine where to go out to eat, or you know, fairly low stakes like writing tasks that the best the models a year ago could do pretty well.

SPEAKER_01You don't really need you know this this super um intelligent and air quotes model to do.

SPEAKER_00Right, it's the age-old thing, like uh, you know, if you're commuting to college, do you need to have the Lamborghini to get there? I mean, the Lamborghini looks really nice and probably gets you there faster, but you know, a good old Volvo, used Volvo or something will also get you there, and it's a lot cheaper economically, and there's all sorts of factors to consider. Um, you know, and I know one of the big concerns is a theme that we continually come across about the environmental impact. Well, if you're worried about environmental impact, you know, some of these things, uh, think about what you're asking it.

SPEAKER_00Good enough is going to be good enough, and it's going to make less of an environmental impact than these larger, kind of more extensive um queries. And so if you're kind of one of those holding your nose but still using AI because you're worried about some of the environmental aspects, uh, that is one compromise that might work. And not tuning things up always to 10 when you only need a five in regards to the level of thinking that you're requesting.

SPEAKER_01And I actually have a planned uh blog about about that too, and kind of letting people know how much electricity it takes to write an AI write report, and it's not much. Um now running agents, yeah.

SPEAKER_00But yeah, so everybody can make you the bad guy with your agent focus at the moment.

SPEAKER_01There's definitely real concerns there, for sure.

SPEAKER_00Yeah. All right, well, uh, I see we're about out of time here, so I'm gonna go ahead and wrap up on our fourth episode here, uh, as we've been talking a little bit about expert versus novice and the use of AI and kind of updates on the trends of AI regarding AI agents and what has been going on at the APA conference. We'll be coming back in our next episode talking a little bit more about the APA conference, but stay tuned. We are glad that you're with us, and uh, we'll have episode five up here in a couple of weeks.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Beyond the Bot: AI in Supervision- How Much Help Is Too Much?

    Nov 20, 202632 minAudio

    AI is quickly becoming part of everyday clinical work, but what happens when it starts shaping how the next generation of mental health professionals learns their craft? In this episode of Beyond the Bot, Dan Florell...

    Credit the people on this episode

  2. THROWBACK EPISODE: The Difference Between Feeling Gratitude and Performing It

    Nov 13, 202653 minAudio

    A quick note for ABCT attendees: Paul and Ray are getting ready for the ABCT Conference in Baltimore, Maryland! Stop by the Psyched to Practice booth at Booth 305 at the Baltimore Convention Center and Hilton Baltimore...

    Credit the people on this episode

  3. Beyond the Bot: Lonely, Anxious, and Talking to AI

    Nov 6, 202636 minAudio

    AI chatbots are becoming more than homework helpers. For many teenagers, they’re turning into confidants, sources of emotional support, and even substitutes for human relationships. But what happens when young people...

    Credit the people on this episode

  4. Practice in Action: Turning Play Into Progress

    Oct 30, 20261 hr 12 minAudio

    What if the games already sitting in your therapy office could do more than help kids pass the time? In this episode of Psyched to Practice, Paul Wagner and Dr. Ray Christner explore how therapists can use everyday...

    Credit the people on this episode

  5. Beyond the Bot: From Research Overload to Real Understanding: The AI Tool Worth Exploring

    Oct 23, 202634 minAudio

    What if you could turn a collection of research papers into a podcast, create practice questions from professional guidelines, or build an infographic from dozens of sources? In this episode of Beyond the Bot, Dan...

    Credit the people on this episode