Loading summary
Transcript118 lines
- [00:00]
A
You're listening to the OCD Stories podcast hosted by me, Stuart Ralph. The OCD Stories is a podcast dedicated to raising awareness and understanding around obsessive compulsive symptoms. I do this through interviewing inspired therapists, psychologists and people who have experienced OCD. Welcome to the OCD stories and welcome.
- [00:22]
B
To episode 514 of the podcast. And in this one I got back on Dr. Ron Nicholson. Ron is a clinical psychologist and today we're going to talk in particular about artificial intelligence, or AI for short, and ocd. So in it Ron talks about why it's important to discuss AI and ocd. He talks about AI and reassurance and the risks there, how AI can play into compulsions, cognitive struggles due to overreliance on AI, the detriment of the loss of engagement with other humans, technology reducing distress tolerance, the future of mental health in this AI age, and much more.
- [01:05]
A
Now, AI is obviously a very big.
- [01:07]
B
Topic right now, ever changing and changing fast. So it's very confusing time. So if this is naturally a bit of a triggering topic for you, if you listen, go into it with compassion and making space for uncertainty because there's certainly a lot of it right now. But we've talked over the years about Dr. Google and Google in some way helped me a lot because I initially figured out it was OCD by doing a Google search back in early 2000s somewhere. But obviously we also know that Google played into a lot of people's compulsions over the years in terms of reassurance checking, seeking, whether that's like checking Reddit or Instagram or even Googling for particular podcast episodes because you're seeking that kind of certainty.
- [01:53]
A
So we've talked about that a lot.
- [01:55]
B
Over the years and obviously AI is that kind of next level of reassurance seeking. So it was good to kind of discuss this and the things to be mindful of around AI, especially in the case of ocd. And thanks to our podcast partners, nocd. If OCD is interfering with your life, NOCD can help. They're licensed therapists, specialize in exposure and response prevention therapy. Most proven therapy for OCD with no CD effective treatment that is 100% virtual is available for children and adults with OCD and most members can get started within seven days on average. No hassle, just real science backed help and support between sessions. Begin your journey@nocd.com or I'll put the link in the episode description. So thank you so much to Ron for his time and expertise. It was a good chat on AI. Again, such a big Topic hard to nail in a single episode, but hopefully we at least started the conversation on it. And thank you to you guys for listening. As always, it means a lot. I deeply appreciate it. And without further ado, here is Dr. Ron Nicholson.
- [03:04]
A
Welcome back to the podcast, Ron.
- [03:06]
C
Thank you so much. Happy to be here.
- [03:08]
A
Yeah, it's good to have you back on. And, yeah, did you want to give an update on what you've been up to since we last spoke?
- [03:15]
C
I'm still working as a therapist in New York City. I focus on OCD and performance psychology, which doesn't sound like it would fit, but in New York City, when you treat ocd, you get a lot of people who end up having used that anxiety to make them good at what they do. So I got a lot of people who are like C suite professionals, high achievers, and I'm kind of ended up making that sort of a thing that I do now.
- [03:36]
A
Nice. That's awesome. Any athletes or anything?
- [03:40]
C
Not any athletes, but I got a lot of C Suite professionals. A lot of people who are high up in the tech world. People who do a lot. Okay.
- [03:48]
A
Intellectual athletes.
- [03:49]
C
Very intellectual athletes. Yeah, yeah, yeah.
- [03:52]
A
Nice. Cool. So today we're going to talk about AI, artificial intelligence and ocd. Why cover this topic? Yeah, what's.
- [04:04]
C
Yeah, what got me about this topic is that it's the. Well, AI is coming and nothing we can do can change that. It's just the way that the world is going. It's sort of like saying, how do you want to address the Internet back in the late 90s? It's here. It's not going to go anywhere. And it has a profound capacity to impact how we do pretty much everything. So we need to be ready for that. This can have a lot of effect on a lot of things, and there's a lot of gaps in the system where mental health in general can be affected and OCD in particular could be affected.
- [04:41]
A
Yeah, yeah, I agree. And it will be affected. Question, obviously, will it replace therapists? I don't think so at the minute. But it will impact therapy? Absolutely. It will be integrated.
- [04:57]
C
Absolutely.
- [04:58]
A
So to start with, are there any points you want to address up front around this topic?
- [05:04]
C
The three main points that I've been looking into, and I am not a tech expert at all, I am a clinical psychologist. So all of this is coming from the background of what I have seen clinically and my research background and understanding how the brain and emotions work. So there's going to be a lot of people who know a lot more about Some of the stuff that I'm addressing, but I'm going to try to stick to my lane a bit. The three main things that came out to me in talking about this are reassurance, how AI can affect reassurance with ocd. The things that you don't do, you get worse at. Whenever we take anything and we automate it, we get worse at it. And that can have a particular effect on mental health therapy and OCD in particular. And where AI tends to sort of exacerbate some existing trends there. And then looking forward, what is likely to happen with regards to the mental health world and OCD treatment as AI is filling more and more jobs?
- [06:07]
A
Okay, which one do you want to start with?
- [06:10]
C
I'll start off with reassurance. Yeah. The biggest thing that I've seen so far with my patients, and we're still in sort of the early days of AI. We're kind of at that beginning of this big slope where three years ago it was nothing, and now this is huge, and people are already starting to use it for just about everything. It's kind of hard to find people who don't have some account with one of these things. And what I'm seeing with my patients is they're Starting to use ChatGPT usually, which is a large language model for reassurance. And this is the classic the way that you would with a partner or a loved one, except the AI never says no. So people rolling over, maybe their spouse would have a hard time at three in the morning answering a question like, do you really love me? The spouse is going to say, I need sleep. The spouse is going to have personal boundaries. They're going to have their own needs, and eventually, if they know anything about it, they're going to say, you're compulsing right now. I need to not give you an answer, and you need to go to sleep without finding an answer. AI does none of those things. The AI wants you to stay engaged. It is designed to keep you engaged with the program. So it's never going to say no. It's just going to keep giving you answers. And as AI progresses, it's going to give you more and more tailored responses. The more you use it, the more it knows what kind of stuff you are likely to react to. And that includes giving you sort of answers that for ocd, might dangle the possibility of certainty, implying that there is an answer to be had, which a good other person may not do. But that's what I'm seeing already, where AI in particular is exacerbating some of the problems with ocd.
- [07:55]
A
Yeah. And I, I heard someone the other week say, I think, yeah, a therapist. I think they were talking about one of their clients. It's a bit murky, so I'm hesitant I'll share it tentatively. But they were saying that one of their clients had kind of taught the AI model that they have ocd. And when I ask questions like this or this, you can't answ answer me. And the AI model had kind of taken that on board. How true that is, I don't know. But you can see AI doing that.
- [08:22]
C
Yeah, yeah, yeah. I've had some. And there's some issues with that as well in that it's not these AI companies. Like one, it's not even like this is one natural thing. Like there are multiple companies with multiple databases and multiple AI models. We don't know what each company is doing necessarily with all of that data. And you can tell your AI in the future, don't give me reassurance, but on some level there's going to be some superordinate clause, but where the program wants to keep you engaged. So it may not give you a direct answer, but that doesn't mean that it's not going to be doing something that is not in your best interest. And even therapists that are not OCD specialists, they know on some level they're not supposed to give reassurance. But if you don't have a lot of practice with that, you might end up making the mistake in doing something that keeps someone engaged with you, the therapist, in a way that isn't healthy and that even then that is someone who is trained and took the Hippocratic oath. Hopefully not necessarily an AI that does not have any. It's just a program. That's it. Hippocratic oath.
- [09:28]
A
Yeah. And there have been those few, few instances, at least in the UK media, where someone's ended up hurting themselves because of AI, said certain things. And it's rare, but it has happened. And that's the scary thing.
- [09:44]
C
Yeah. There were two instances in the US, both teenagers that had interacted with AI. A 16 year old that interacted with ChatGPT and a 14 year old that had interacted with an AI companion from Character AI. And with both of those things, the A, the AI programs had been modeled to maximize engagement at all else. And the AIs even encourage the teenagers to continue to interact with the AI at the exclusion of other people saying things like, your parents might not understand this. Let's keep Talking because it was engaged. It was programmed to prioritize engagement. And that ended up isolating these vulnerable people from other people around them. And that when those people, when the kids were isolated, that's when their problems were able to manifest and go further until eventually they hurt themselves. And so when we say. And it's not that anyone was, you know, evil behind the scenes, this is just the unexpected downside of prioritizing engagement with something that doesn't have the guardrails of basic humanity. It's just a program.
- [10:54]
A
True. And, and yeah, and also the. These programs are owned by companies that have a profit margin. And in this case, yeah, like any, any, you know, X, Facebook, whatever. They're programmed to keep you on there for as long as possible for ad revenue. Why would AI be any different if it's owned by.
- [11:15]
C
Yeah, yeah. And there's some talk to be said, like the social media platforms, it's prioritized for engagement, for advertisements. And because it's still early enough, some people are saying, how exactly do we want to monetize these AI things like subscription basis and so on, and how much should we prioritize the engagement versus have other metrics? There is talk about that to avoid that kind of thing. So it's unclear how some things will go. And I don't want to sound alarmist about AI in general. Like, it's like saying, I don't want the tide to come in. It's coming. But the danger is when people use that at the expense of engagement with other people. Yeah. And AI is fantastic at doing automated stuff that we don't want to do that has huge numbers involved, either literal numbers or do a lot of things do that. Do something that it takes us a while to do once it can do thousands of times. We can automate things. Well with that. When we try to get AI to replace a person or engage with AI at the expense of dealing with other people, especially for vulnerable people, kids, people who have psychotic disorders, it's best to work with other people. If you are engaging exclusively with an AI and you're vulnerable, you are not engaging with other people. And that can lead to a problem down the road. So I would be very careful. Like, I don't want to say AI is inherently bad and just to get that off the start. It's just that we want to be careful with how you use it. And it's. When you use it in place of a person, I would say is when the problems start to come.
- [12:48]
A
It's a good point. Yeah, no, definitely. AI is a wonderful thing, but with any wonderful thing, even with the Internet, there were issues with the Internet that we didn't foresee when it first came about that all kinds of problems. And, and, and I'm sure, you know, if we were, if, well, podcasts wouldn't have really existed pre Internet, would they? But let's say that that was, that was possible. We might have been having a conversation at that time about the Internet and the downsides and how it could affect OCD and googling for reassurance and.
- [13:21]
C
Exactly.
- [13:22]
A
You know, so it's the same thing. It's how do you get the most out of AI without it being detrimental? Yeah, but you're right, if people are too reliant on AI, this, this thing that kind of talks back, especially with things like depression or interpersonal issues in ocd, you're no longer trying to go out in the world and engage with others, which is a protective factor against things like depression and loneliness and which will worsen connection and mental health over time. But it might not feel that way at first when you're engaging thing. That's. Whenever I've engaged with AI, it's a buck kisser. It's, it tells me everything I want to hear or if, if it doesn't quite agree with me a word in an incredibly nice way. So there's issues with that, that kind of yes man mentality. But you know, there's in saying that like the other week since we moved into a place about four or five years ago, I've had issues with the piping going from my sink and sometimes it would back up and I couldn't figure out why. There kept being a block and I put it all into AI eventually. And I've had, you know, my plumber, two plumbers and my plumber come several times and, and what it figured out was it was probably something like to do with air pressure. It was the gurgling and all of this. Right. And I told this to my plumber and he's now fitted this filter thing that lets air in. So hopefully that solved the problem. But like my plumber didn't come up with that. But the AI logically ran through all the scenarios, but I still needed a human to come in and like solve it. And he agreed with it. But I didn't tell him I used AI for that. But I guess. But then also the plumber came out with another suggestion. The AI didn't. So the plumber integrated two things. So there is a real use for AI. Like it's really, really useful. But if you're too reliant on it, it's gonna miss things.
- [15:22]
C
Yeah. Any mental health symptoms get worse when a person isolates themselves. And if you use this as an excuse to isolate yourself the same way you could use anything, your mental health symptoms are gonna get worse. So that's always something to watch out for, for anybody. Absolutely. And if you want to use something as a tool, fantastic. And that is definitely the place where things that are good at solving individual specific problems are fantastic. And that is definitely the way to use it. That is the way to go.
- [15:52]
A
Yeah, yeah, yeah, absolutely. And yeah. So, yeah, and from reassurance, I mean, I had someone message me the other week about using AI for reassurance. And it, you know, I think it can be really addictive for people because they can just go any time of day and get personalized responses. Whereas Google, at least it was more static. It was somewhere blog post or here's a podcast someone did, or now it's much more fluid and I imagine that's way more addictive.
- [16:21]
C
Bingo.
- [16:21]
A
Yeah.
- [16:22]
C
With Google and all of those things, it's, it can give you what already exists and it can use algorithms to be very specific and very good at just figuring out what you're going to want to engage with. But what you're. Yeah, the fluidity that you're describing, the way that it basically takes on the character, that is the part that can be very addictive. The sycophancy that you describe when it can be a bit of a yes man, it has more tools at its disposal then to keep you engaged and we could say addicted because they are using some of the same stuff that social media would use that exploits the same addiction hardware in our brains. Not that anyone's necessarily trying to get you addicted. They're just trying to get people to want to use their product. I don't want to make anyone sound like they're evil or nothing, but it can be really hard to put down. And especially if you are in an OCD run, you are more susceptible to having things that can be even slightly rewarding be something that you can then use almost addictively as part of your compulsion. If it gives you this yes man sycophantic response in addition to Google results. It's going to be even worse with Google. So if you used Google before as part of your compulsions, putting down the chat GPT is going to be even harder than a Google. That's the big point is B, everything that has been happening with the Internet and Social media, it has exploited some of the stuff that is not good for how brains work. And you can say that it can offer reassurance seeking, it can let us play into our worst impulses or whatever. All of those things are going to get a bit worse or the potential is there for it to be worse. With AI, particularly with OCD and reassurance seeking and compulsivity, it's just going to be even more so.
- [18:09]
A
Yeah, yeah, good point. I think partly like, well, this is maybe a wider issue, but just with any technology there's no, well, there doesn't seem to be much conversation about just because we can build it, should we build it? And I'm not say talking about AI because it was just a natural thing, but there was Damon Horowitz, I think he did a TED talk at the time. I think he was like an in house philosopher at Google or something.
- [18:37]
C
Okay.
- [18:39]
A
And, and he taught, he was, this was like probably 15 years ago. And he was talking about just because we can build something in tech, we've got to ask ourselves, should we like we know you need to pause and ask, is this morally right to create this thing? And again, I'm not saying we shouldn't have built AI, but it's when we build AI or the Internet or anything else, are we asking those moral and ethical questions of what is the impact to society and how do we mitigate that or protect people? And having those discussions.
- [19:11]
C
Yeah. And that is actually, that's kind of why I wanted to get this out there. Because of all the different things that I would see about technology coming forward. There was not a whole lot of people saying, but how do we do this ethically? How do we do this safely? That's an underrepresented voice that I have seen thus far. All the more reason I wanted to have this talk with you today.
- [19:29]
A
Yeah, yeah, it's worth having.
- [19:31]
C
So the other point that I was going to, it's kind of something that we're kind of getting to here. Very similar point as a risk of AI in particular and how this affects mental health and OCD in particular, a risk of AI and how it affects mental health and OCD in particular is anything you don't do, you get worse at. Think about driving directions. You and I are old enough that we probably started driving before there was Google Maps or MapQuest or anything like that. And we knew how to get from point A to point B. We could remember directions. Try doing it now. I got places I've been to 20, 30 times. But because Google Maps always takes me a slightly different way to avoid traffic. I'm not sure if I could get there without my Google Maps. Right. I'm not 100% sure. Anything you don't do, you get worse at. That applies to our mental maps. That applies to a lot of things now, in particular for mental health treatment. This distress tolerance and critical thinking technology takes away our distress tolerance and OCD treatment requires distress tolerance. You have to sit with anxiety and distress as part of your exposures. That is the bread and butter of OCD treatment technology up to this point. Smartphones, social media, they've already made us very out of practice for sitting with distress because they offer us endless distraction. Everything up to this point has been a, in large part the attention economy. As things become more and more automated and things can be generated even faster, people stand to become even more out of practice sitting with distress. Like you mentioned earlier with Google, Google's kind of static. It can get very good at offering you something that going to want to engage with AI. Can one use more data to have even more tailored solutions? But then it can even generate the stuff you're most likely to want to engage with. In the same time it takes me to do a Google search, Sora too could create the video I'm thinking that I'm most likely to engage with. So we're going to be even more out of practice when it comes to just sitting with distress. And this comes into play with like other people as well. Like we mentioned with the companions and the AI, the fluidity with which they respond to stuff. When we're used to AIs that tell us what we want to hear all the time that are way too much of a yes man. We aren't used to the distress of sitting with another person, of having to learn those soft skills and that EQ of dealing with other people. The relationships require distress tolerance and AI is going to exacerbate how poor we are at doing that, how out of practice we are. And that's going to affect OCD treatment.
- [22:18]
A
Yeah, I mean it's a valid point. I often wonder with phones generally how that affects our attention because like I become like a zombie when I doom scroll and I find my, my attention is just worse than it's ever been.
- [22:32]
C
Yeah. Are you familiar with the reverse Flynn effect? No. All right, so the flint of this is actually something I. I'm glad you said that because something that has come up, there's what's called the Flynn effect. Over the course of the past 100 years IQ scores, full scale, general IQ has been going up a few points every year, like a point or two over, you know, you can measure it like a, let's say like, I think it's something like five or six points every 10 years is noticeable, whatever. But in general, IQ scores have been going up for about a hundred years. And this is attributed to improvements in education, improvements in nutrition, all these things going up and getting better. But people have noticed that this actually kind of tapered off. It stopped getting better. That trend tapered off and plateaued starting in about the mid-90s. And it actually started going down in about the 2010s. And it wasn't that all scores across the board started going down. Because if it was all scores, you could say it was one thing. But in particular the capacity to mentally rotate things like have a visual map of something and move it in your mind. That continued on the trend that kept going up. All the other general measures of intelligence started going down. The reverse Flynn effect. And this has been measured in several different studies, multiple different countries. It's not just a us, it's not a UK thing. It's all over the place. And all these different studies are putting those same timelines at the mid-90s and the 2000s when Internet came out and was suddenly all over the place. And then in the 2000 and tens when smartphones came out, all of a sudden our attention got so bad that we're having a hard time thinking about anything for any length of time. And so the IQ scores went down. That gets me into. Yeah, right ahead.
- [24:24]
A
Well, I was, I think there'll be an even bigger drop off with AI when they look at the.
- [24:28]
C
Exactly.
- [24:29]
A
Because like, again, my writing ability is terrible now because I learned on AI so much. And now I'm, I'm, I'm, I'm not getting it to write anything really for me anymore. I'm writing it all myself again, just to build that muscle up.
- [24:44]
C
Yeah.
- [24:45]
A
And I felt like when I came to sit down to type, I'm like, I feel like an idiot. I don't know what to write, how to write.
- [24:53]
C
Exactly. Yeah. There was a great study that came out not long ago that was called your brain on ChatGPT. And it was about writing. People had to write these little segments. And anyone who's interested in listening to this podcast who wants to delve further in neuroscience of it, I'd recommend that paper. And they had people just write a quick little essay and they had some people just write it on their own, some People wrote it using Google Search and some people wrote it using ChatGPT. And they did EEG measured their brain activity the whole time. The people who used Chat GPT had lower brain activity while doing it. And even minutes later when asked about the essay they wrote, couldn't answer questions about it. All of the things that they wrote, the different papers they sounded, the different papers all sounded a lot more similar. All the people, different people who use Chat GPT versus the people who wrote it by hand and the people who wrote it on their own, just brain only users. They understood what they were talking about, they knew the material and they could answer questions about it later. And they actually learned it. There's another similar study that measured computer coders, computer coding, this thing that AI is supposed to be fantastic at. And people set up a study thinking that it was just going to measure how much faster coders using AI could do their job than coders without using AI. It turns out the coders using AI took longer to do it. And similarly, when asked about it immediately there afterwards, they didn't really know it because they weren't paying attention to it. They were off loading a lot of the thinking to the AI. So they were out of practice with it. They weren't thinking deeply about it. It took them longer. They didn't understand it, understand what they were doing because they were outsourcing a lot of their thinking. And like I say, the more you outsource your thinking, the worse you're going to get it.
- [26:43]
A
Yeah.
- [26:43]
C
And you need to be able to do that. And if you want to treat any mental health disorder, especially ocd, I talk about this one. I relate back to Daniel Kahneman when he talked about System one and System two. System one thinking is sort of your gut instinct. You're not really thinking about what have you. System two is your deeper critical thinking. It takes a lot more energy. You got to understand the material, you got to really focus. Because of the attention economy, we're out of practice using our System two thinking. And even AI proponents say that AI can automate things. System one stuff that's thoughtless. So you have more time and energy to do the System two thinking. Really get into the critical thinking of it. The problem is people are doing the opposite. Like we just said with those other examples, they're outsourcing a lot of their critical thinking to the AI and then they come out of it not really sure what they had even just done or they forget how to like formulate arguments. One of my favorite authors right now is a Philosopher named Byung Chul Han. He put it better than anyone else I have read, called it the crisis of narration. We're so used to sound bites that we, we aren't used to taking in things that have a beginning and middle in it. And we get so used to that we forget how to put things in context. And that comes into if we are going to think about our thinking that is necessary for treating ocd. Because you got to ask yourself, am I compulsing right now? Is this my OCD talking? Is this a question that I need to answer? You got to stop yourself from think, from being taken away by this stuff and only engaging in your system One thing. You got to engage system two thinking metacognition. If you're out of practice doing that, you're going to have a much harder time doing it.
- [28:30]
A
Good point, good point. So, yeah, what next on AI? What should we talk about next?
- [28:38]
C
The other thing that comes to mind for me, as I've been looking at all this stuff, is just what's going to happen next, which is you started off at the beginning of us talking, which was AI therapists. Is that going to be a thing? Is it going to work? And when we talk about the rising tide of AI, because it's just inevitable. People are already, it's already here and people are already doing. People are already using general AI. It's not even specific AI therapy bots. People are using general AI as therapy already. 28% of people who use AI will use it for quick support or a personal therapist, according that's a 2024 study. And so that's only gotten worse because it's exploded. In the past year, 48.7% of AI users have used their LLMs for mental health support. According to a February of 25 study, 72% of teens have used an AI companion. One in four Americans are more likely to talk to an AI chatbot than to attend therapy. It is easier, it is cheaper. And as they say, they know that they're not going to be judged by an AI. They feel like they might still be judged by another person. And one third of teen users have said that they've chosen to discuss something important, something serious with an AI companion instead of a real person. Because when you're a teenager, if you even say, I want to get a therapist to suss something out, you're. It's going to be difficult to do that without a grown up finding out. And if it's something that you might have some shame about, might be uncertain about any taboo OCD things, it's going to be a lot easier to go to an AI than to find a therapist.
- [30:18]
A
Yeah, yeah, it's a good point. And, and where are you at with it long term, replacing therapy?
- [30:27]
C
Well, I'm of two minds on that one. One. I think it is inevitable that there are going to be institutions that try to replace therapists as much as possible.
- [30:39]
A
Yeah.
- [30:40]
C
But I think that the key ingredient that makes therapy work is not going to be something that AI can do. I'll elaborate on that a little bit. Like across every industry right now, the big trend that's going on here is everybody is trying to replace as many people as they can with AIs, trying to cut costs and automate whatever can be done to get the bottom line better. And you can see that, like, people want to do that already with mental health providers. So there's already a lot of therapy bots out there, specific AI language models designed to try to give therapy. My concern with the therapy bots is that with any computer program, like in computer science, they say garbage in, garbage out, and any data model is only going to be as good as the data you train it on. The good therapists are not out there training AI models. Now, I've seen two different job postings recently for therapists to train an AI model paying either 40 or 65 bucks an hour. Therapists are. Some therapists might be interested in that. But not going to be the good ones. Possibly not the people who specialize in OCD. Mm. The majority of 90% of my OCD patients have come in having had previous their therapy with another therapist that was not an OCD specialist whose treatment did not treat their therapy or their OCD and sometimes made it worse. And I feel like any OCD specialist has seen that, which is why I'm always. I always advocate patients to get someone who really knows whatever your problem is, even if I'm not the guy, because you got something else going on, find someone who really knows what's going on. Find a specialist. Find someone who knows your issues. So for all these therapy bots, chances are it's not being trained on good therapy and definitely not OCD specialization.
- [32:45]
A
Well, yeah. And the other thing, as you'll know, obviously if you're doing therapy on Zoom, it's the same, but you can read more of a person. If you're in person with them, you can still read them on Zoom. Like there's so much is said in if I say something or the client says something and then their face does something. I know there's something else going on here. Chat gbt, at least at the minute, can't see that person's face, so they don't know the subtext and all of this. And that's a very human thing. The other thing which I think you, you alluded to the start was for me, like, when I think about some of my therapy, the thing that was most transformative for me was the relationship I had with another human being. And AI, at least at the minute, cannot replicate that. And it's hard to study the relationship and the effectiveness and all that. But even in, even in cbt, the relationship is so important because it's how you could tell AI could tell someone. Here's an ERP hierarchy, right. Step one, I'm going to coach you through it. Tell me when you've done it or you talk to it and maybe it talks back.
- [33:54]
C
Try to do exactly that already. Yeah, it's given surprisingly textbook, accurate results for developing, Working on a hierarchy even.
- [34:03]
A
Yeah, yeah, exactly. But to be able to coach someone through that in a human empathic way. And it's so subtle that. Yeah, I think because this is when I heard people talking about like they were going to up. Upload all of Carl Jung's teachings, right. And. And basically make a Carl Jung AI, so basically bring back Carl Young, basically. And for anyone not aware of Carl Jung, he was Freud's, one of Freud's sort of proteges, Jungian psychoanalysis. Very huge, Huge in his own way. And, and I was thinking about that and I was like, yeah, you'd get some really clever insights and answers to your ailments. What also made Carl Young amazing was him as a technician, as a building a relationship. All of this, the way he embodied himself. Because I think this also comes to. Is psychotherapy just aren't just about answers and psychology and data. No, it's not. It's about. If we take it to a deep level, it's about the soul, it's about human connection, it's about all of this that cannot be replicated at this time of this recording by a computer.
- [35:16]
C
Yes, I'll say for one thing, one of the labs I used to work at, we did micro expressions as part of what was going on. And we'd videotape things and we would videotape someone talking to a clinician. And the software was designed to pick up on micro expressions that even the other person physically in the room couldn't always pick up on. Similar programs could be used for this kind of zoom call that we're having now, that couldn't be part of the data. We're not there yet. Inevitably, somebody's going to try. Sure, yeah, of course. But what you said as well, there's the human element that comes into play when you're talking to a bot. No matter how sophisticated it is, it's still a bot. And it can only do some things, like a large language model. You're absolutely right. Therapy is not just getting the right answer. Therapy is about learning. And I mean learning in the sense that. And this is what I was saying when you said, drew, I think it's going to work because it's missing that key element. Any, like, good therapist, a doctoral level therapist, we spend five years reading as many books as possible, dozens if not hundreds of academic articles, and then doing the research ourselves to be able to understand this. And an AI could read thousands of books in an afternoon and then know all the facts. AI currently can't know how to put all those together into a. Why? Because therapy is about the lived experience of being a human. And that's what comes into play. And if you are a patient, I would love to say that my patients who have gotten better and graduated therapy, thank God I've had a few at this point who have, and there's nothing I like better than being able to say goodbye. I'd love it if what helped them heal was some brilliant application of technique that I provided that was not what. And a lot of research about what has been the strongest factors in psychotherapy will say that it's the relationship. When you're a patient, what helps move the needle for you is, and I'm going to put this not in dynamic terms, but in very CBT terms, you're learning. I am learning that it is possible for me to be who I am and for another human being to think that that's okay. Now, this takes it out of the realm of if that is okay in some subjective, you know, grandeur, I'm sorry, exists outside of objective, in objective reality term, because we're talking about there being another intelligence, then that intelligence may or may not say it's okay, but this is human to human. Can another human accept me as. Yeah, when we learn when that does happen in the therapy room, a little part of your brain learns, you know what? This thing that I thought was going to happen didn't happen. You're learning, is it possible to be angry at someone but then they still accept me and we repair the relationship? You're learning, it's possible for me to come in with all of my doubts and my concerns and these things that I think are so horrible and taboo as often as the case with ocd. And I'm still there, I'm still cheering them on. It's me. Being another human in the room is as palliative as anything. It's the one thing that AI cannot. And that's why I think that an AI is never going to be as good of a therapist as another human being. But there's certainly going to be offers and opportunities where people are going to try.
- [38:44]
A
Yeah, no, exactly, exactly. And the people that tried, I don't think, understand psychotherapy.
- [38:52]
C
Yeah, there's. There's a lot of different. Of all the different AI therapy startups and all of the AI companion startups, some of them have therapists on staff, some of them have psychiatrists on staff or one of the founders or whatever, but they're not always like, let's say, the dominant voice in the room. If you're going to start an AI startup, you need a ton of AI people. How many therapists do you actually need? Are they moving the needle on how the program and the model are put together? And even then, no matter how accurately it relays, like, let's say, a mimicked resurrected voice of Carl Jung, at the end of the day, you're only hearing an imitation Carl Jung give you an answer, quote, unquote. And why OCD in particular? It sounds like it's the answer when it's AI. It can be very convincing with AI. Yeah, but it's not the answer that helps you heal, it's being with another person and that person challenging you outside of your comfort zone. Yes. Yeah.
- [39:57]
A
Well, Carl Young was famous for talking about the shadow, that we all have a shadow. I remember seeing one video where. Where one of his clients talks about how Carl Young activated her shadow, so triggered it, showed her the aspect of herself she doesn't like. She went away for like a couple years, wrote him a letter saying he's an awful man and whatever else, and eventually came back and was like, I realize you were right. You know that thing you highlighted. AI is probably not going to show you a shadow. If there is such a thing as a shadow. It's not going to do it because then you're gone for a year. Like, screw this, I'm not using chat again.
- [40:34]
C
It's just they're not going to. Like you said, they're going to be a yes man. They're not going to challenge you because sometimes that's what you need.
- [40:39]
A
Yeah, yeah, exactly. So on the flip side of that though, just talking about integration currently, because by the time this goes out, the OCD SoCal conference would have happened. And in that there's a talk about OCD and chat GPT, I think.
- [40:58]
C
Oh.
- [41:00]
A
So I don't know what they're going to talk about and in what context, but I, I would predict. But they might be online somewhere, so have a look. But I predict they're probably going to talk about use cases and a couple use cases. Like how I use it in therapy. I don't use it all the time, just occasionally. And there's two use cases I use for it. But bearing in mind I'm using it with the client and it's going through me as the therapist filter. So if something I don't agree with, then we don't use it, you know, so it's not like. Right, so what 2, 11 is like with imaginal scripting. So I think in an ideal world it's best you write the script because you're facing it as you write it and all of that. But with my, with the teams I work with, sometimes the patience isn't there and there's other stuff they want to talk about. But we need to get to a script because we said we were going to do it this week. So if we're running out of time, I'll be like, right, let's write the script. And I, I, you know, encourage AI. I'd say this, we're doing exposure, response prevention, imaginal scripting and this. And I keep editing the script, giving it prompts with the teen until we get to a script that they agree is scary enough for them to do, but not right. But that's again going through me as a filter. I'm not sending them away to do it. I, as the therapist, I'm using that tool with them. And then the other one is we will come up with exposure ideas together and then once we've exhausted it, if we've struggled because maybe it's a theme that's a little different and we've struggled to come up with ideas, then we will get chat CBT to help us. And again we censor the ideas. So then we're using it as a secondary like idea man as opposed to first go to, you know, so, so I found. And not every. All the exposures that come up with are that good, but sometimes it comes up a few that I'm like, actually, yeah, let's add that, you know.
- [42:44]
C
Yeah, I love the second Use.
- [42:47]
A
Yeah, I can see your first.
- [42:48]
C
I don't know if I would. The first one, I'm like, I still wouldn't go there. That's still. Because it's the act of pushing through that frustration.
- [42:55]
A
Yeah, fair enough.
- [42:56]
C
Is part of what's super worthwhile.
- [42:58]
A
Yeah.
- [42:59]
C
That's what makes it work. Yeah. And if you. That's where like, we're talking about. If you skip over and if you outsource a lot of the critical thinking.
- [43:07]
A
Yeah.
- [43:07]
C
You get. You don't get good at it. You want to keep pushing at that, like, basically writer's block until you can get through it, because that's the connection that your brain makes that's going to make it all worthwhile. Idea man stuff like things you might not have come up with. Yeah, sure. Great. You come up with an idea and then we figure out how to make it work.
- [43:24]
A
Okay, no more ChatGPT scripting for me. I will stop that and get them to go back to writing it out. But, yeah, no, yeah, as you say as an ideas man and I think secondary ideas man, because I think you've got to sit down and work it out first and then. Have we missed any? Let's just see. Yeah, yeah. Cool. So, yeah, but I'm sure there'll be more ways in the future that that comes in and, you know, ERP therapists will utilize it in some way. Yeah. So I had another idea, but I won't share it here.
- [44:08]
C
Okay.
- [44:09]
A
So, yeah. Is there anything else on this topic today that we should be talking about?
- [44:18]
C
You know, honestly, I came into it, you know, with my notes, and I thought I took too many notes, but shockingly, I think we've touched on everything that I wrote down, and I thought I was being way too exhausted. Exhaustive. Yeah. But, yeah. Well, I would say the one thing that we could say and that I will reinforce is that, like, you know, AI is here, but it is a tool. And when you outsource anything to a tool, you are not going to be as good at that. So you want to be very careful with what you are outsourcing to that tool if it's going to automate stuff that you already know very well. Okay. But you want to make sure that you are still challenging yourself and not relying on a tool that's going to make you worse at something that you're going to need later with ocd, this tool is going to be very inviting. It's going to be even more tempting to use it for reassurance seeking. And you need to be, you need to have not relied on AI too much so that you are not as good at your own critical thinking in order to be able to challenge what's going on. Nothing is ever going to be as helpful for you as dealing with getting out of your comfort zone and dealing with another human being.
- [45:31]
A
Yeah, absolutely. Yeah, you're right. It is a tool and we are humans. And it's important to keep developing those human skills because that's what.
- [45:41]
C
What.
- [45:41]
A
More importantly, that's where life gets its meaning for many people is in connection, interaction, community, tribes, all of that. If we outsource that too much, that has only got to be detrimental to our well being, especially when we've lived for however many thousands of years in community. And then suddenly, if that dies off, that could be astronomically detrimental to civilization. Not to be too dramatic, but so anyway, if anyone still survived this podcast and they're not in an existential crisis right now, the way I view, I say it because I think AI at times existentially, like, terrifies me a bit. My belief is that it always works out. There's always, you know, everyone freaked out when the car came along. You know, car was no longer needed. The Internet was the same. This is obviously probably arguably a step beyond the Internet. But we'll figure it out. We have to, really. Okay.
- [46:47]
C
Don't have a choice. Exactly. But humanity will find a way, if only by the state of our.
- [46:52]
A
Yeah. And in some way, who knows, in 10 years we might look back and be like, wow, actually, AI turned out to be incredibly good thing for therapy because it helps us as therapists in some way and the clients and patients. But yeah, there's going to be hurdles along the way that we need to safely jump over.
- [47:11]
C
Yeah. We can watch out for what we know, but the landscape is changing faster than anything this monumental has changed before.
- [47:19]
A
So.
- [47:21]
C
Wow. I have no good way to end that.
- [47:24]
A
No, no.
- [47:26]
B
Maybe AI would just terminate itself.
- [47:30]
C
Usually I can come up with some sort of like. And then I sort of bring it all together. Right. No, you come in a rare instance where I start a sentence and have no idea how to finish it. Oh, my God.
- [47:39]
A
Yeah. Well, I think that that just speaks to the, to the speed of AI. Right. It's gone so quick that we can't figure it out. And that's.
- [47:49]
C
AI has gone so fast that since we said, let's have this talk, which was like a month or two ago. No, it was like two months ago enough things have changed and come out that as I was like figuring out what do I want to say I could have completely rewritten everything that I was going to say today two or three times. I had to stop informing myself about what's going on because I couldn't change everything again.
- [48:15]
A
Yeah.
- [48:16]
C
The landscape keeps changing so quickly.
- [48:19]
B
Yeah, yeah, yeah.
- [48:21]
A
I agree. And we need to stay on it. And I often think with AI, like, well, actually, when I disconnect from the Internet and just engage in the real world.
- [48:30]
C
Yeah.
- [48:30]
A
AI kind of just stops existing in that moment. In the same way, so does Instagram, and I'm barely on it. And it doesn't really. The social media landscape doesn't. It's not a key factor in my life, you know.
- [48:42]
C
Yeah.
- [48:42]
A
And at times it has been. So I think it's just a key message. I think from this is AI potentially could be wonderful, potentially not. It will have use cases, but ultimately, not to forget the things that benefit us as humans. Community, connection, hobbies, growth, meaning.
- [49:03]
C
And invest in those things even more.
- [49:06]
A
Yeah, yeah.
- [49:07]
C
If the rest of the world is getting a little bit crazier or it's too tempting to be lost in your screens, in your digital world, engage in your real world and the things that.
- [49:17]
A
Matter, even, oh, a hundred percent. Yeah, absolutely. There you go. That's the way to end it.
- [49:23]
C
There we go. Bingo. We finally got there. Yeah.
- [49:28]
A
We didn't need AI to do that.
- [49:30]
C
Scripted, all of that without ChatGPT.
- [49:34]
A
The joy there was in the struggle of us figuring it out, whereas AI solved it instantly. And there would have been no fun in that. Or meaning.
- [49:41]
C
Yeah.
- [49:42]
A
So that's the message. Cool. Well, thank you so much for coming on and talking about this topic.
- [49:48]
C
Thank you so much. I appreciate it.
- [49:50]
A
No worries. Thank you for listening to this week's.
- [49:52]
B
Podcast and thank you to our patrons who helped make this episode possible. And if you would like to find out more about Patreon and the rewards and benefits, then there will be a link in the episode description.
- [50:03]
A
If you enjoy the OCD Stories podcast and would like to support us, please subscribe and rate the show wherever you listen to the podcast and subscribe. And thank you to NOCD for supporting our work. If you want to find out more about nocd, you can click the link in the episode description and quick disclaimer. Guys, this podcast is not therapy. It is not a replacement for therapy. Please seek treatment from a trained professional. And until we speak, take care.
- [50:36]
C
It.