In this episode of Mental Health Pathfinders, Dr. John Torous, chair of the American Psychiatric Association’s Council on Digital Health, Innovation, and Technology, explores the rapidly evolving role of artificial intelligence in mental health care. He discusses the opportunities and limitations of AI-supported care, the risks of long-term chatbot use, privacy concerns, and the importance of maintaining human connection. The conversation also highlights the council’s work to educate clinicians, shape responsible policy, and develop practical frameworks for evaluating new technologies.

Transcript

Erin Connors (00:05) Thank you for joining us for this episode of Mental Health Pathfinders. I'm your host, Aaron Connors. Technology continues to have a profound impact on our everyday lives. It's hard to imagine life before smartphones, laptops, and the many apps we rely on every day. Technology is also at the forefront of healthcare, making this an exciting time of innovation and opportunity. At the same time, it's important to recognize the benefits while remaining mindful of the potential risks and challenges. With that in mind, the APA has a new Council on Digital Health, Innovation, and Technology. And joining me now is the Chair. Of the New Council, Dr. John Toros. Dr. Toros is also director of the Digital Psychiatry Division in the Department of Psychiatry at Beth Israel Deaconess Medical Center. That's a Harvard Medical School affiliated teaching hospital, where he is also serves as a staff psychiatrist and associate professor. Thanks for joining us today, John. Well, let's jump right in. Can you tell us more about the new council and you know some of the work it's already kind of starting to do?

John Torous (01:01) Thank you so much for having me. Yeah. And I think the APA has always had a leadership role in technology. This is a a new council, but it really grows out of a lot of different efforts that were happening across the APA. The APA has always been leading in telehealth, setting standard standards, advocating for reimbursement. We've always done a lot of work in mental health apps. We helped write the APA's app evaluation framework that kind of put the AP at the center. So the idea was we're doing all this work in telehealth. Doing all this work in apps. Clearly, AI is the new friend or enemy in the room. Well, we can go into it. We said, look, this is really rising to the level of a council. Let's put all of these under one umbrella and bring onto it. So, really, it's a new council. So, in kind of jargon y APA governance terms, it's new, but really it's the extension of great work that the APA has been leading for a long time with now an AI spin.

Erin Connors (01:46) That's right. Absolutely. And you know what as we said, this is such an exciting time. Is there anything you particularly are excited about when it comes to technology and innovation? When it comes to healthcare?

John Torous (02:16) I mean, yeah. I I think that we of course have seen medical records for a long time, and we sometimes like medical records, we don't like them. We've seen smartphone apps, sometimes patients use them, sometimes they don't. What's different about AI is the rate of use by both patients and clinicians has been unparalleled, right? You you've never got a technology where both groups are saying we we're we're using it, and we're using it at rates that are well over I think there was an APA survey in February, and I think it was about 50% of psychiatrists were using AI as of February. I'm gonna be bold and say now that we're talking about later, it's well over 50% of psychiatrists are using it. There's also been a lot of kind of different poles of patient populations, and probably at least a third of people are using it. And again, we never quite got that with apps. EHRs were kind of forced upon us by our employers in that, but this is spontaneous, right? And and I think When you see something that rapidly growing, that spontaneous, there's something that we as psychiatry have to chase down there and make sure it works well for our field because it's coming and it you could say in some ways it already came.

Erin Connors (03:32) And thanks for mentioning the poll. We did that with WNBC in in New York. And we we had over two thousand members respond to that. And what some of our members we asked them, you know, to we w to find out what members know about AI and what their concerns are. Eighty percent are concerned that psychiatrists don't have enough training when it comes to AI. Only eighteen per percent felt very informed about AI, and forty percent have not used it in their practice at all.

John Torous (03:35) That's right.

Erin Connors (04:01) It sounds like we have some some work to do here.

John Torous (04:04) Yeah. And in some ways I can I think I always hear people say, Well, AI is changing, John, so quickly. I what's the point of training? It's gonna be different next month. And I think my counter to it is yes, it's always changing, but everyone has access to these AI models. If you have a computer, you can test them. And again, you can put in hypothetical information. Do not put patient information in. We all realize that, but you can kind of test them yourself and get a sense of what it is. You can ask them questions about therapy and safety say you can ask them questions about medication and see what they come back. Doesn't again, so I I think that in some ways the best training in a world like this is to be trying it and say, there's a new model that came out. I've heard my patients are asking about it or using it. What does it give back? And I I think there's something about that genuine hands-on experience by trying it that really lets you again understand what you see as the risks and benefits, what you want to talk to patients about, because When people say, well, this AI model is all bad or all good, that's for their experience, for their thing. And I think we've seen that there is a risk benefit, there's balance, and that's what we as psychiatrists are trained to do. We we know there's no one perfect treatment. We evaluate the risk, we look at the benefits, we work to the person in front of us, we say, Hey, this makes sense to use or not sense to use. So I'm not saying we should use an AI for anything clinical, but we can still use that kind of clinical decision making. help people understand and the only thing one has to do if listening is just go play of these things, try them. Try different ones. See what they say. You you may be horrified. You may be impressed.

Erin Connors (05:37) But well, on a bright spot over half of those polls said they see promise in AI for treating things like anxiety, depression, and substance use disorder. So so there is a little bit, they're seeing that there there could be, you know, in conjunction with the work they're already doing with their patients, there could be some benefit here.

John Torous (05:53) Yeah. It's and and I think in part as the models probably become more powerful and read more of the internet and learn more, they're gonna hopefully have more benefits. New risks will emerge too that we don't know about. So I think that we can't give any of them a free pass, but I think I always like to say like I think what a model could do this year versus last year, it's only been one calendar year. It's been like 10 years in terms of research advancement. And I think that's the exciting. Hard in some ways. And why it's a bit of an intimidating space is that if you thought that the models did something dangerous, made mistakes, they did. You are correct. But the question is, do they still do they introduce new ones? So I think it's a field where I always say when I give talks on it, I have to update my talk every 30 days because the whole evidence has changed over. So it it's it's tricky, but the basic principles of risk and benefit don't change, right? The basic principles of What is its appropriate use case? The tool may evolve, but what its clinical applications are given is something that a group like the APA is going to help set the standard for.

Erin Connors (06:59) Right. And because things are changing so much. Of course, you know, there's concerns about AI for therapy. We know there's a shortage of psychiatrists and other mental health providers out there. So you know, it's understandable that people will look to other places for help there. What's your advice there?

John Torous (07:15) I think if you look at some of the newest meta-analysis that came out a couple weeks ago, it seems like when you use an AI for therapist kind of self-guided with without a clinician or psychiatrist, it probably maybe has the same effect size as a mindfulness app, as a CBT app. So I don't think anyone ever felt that mindfulness apps were the be all and end all or the future of treatment. They have a role. There's some really great mindfulness apps out there. I recommend patients use them to augment it. But there's really not the data that an AI therapist kind of delivers anything above that level. It's even more interesting. Recently, as the models again have evolved, there's some early data, and it's it's early that says, well, maybe can the big models like Claude or Google or Chat GPT, our data is becoming so big, they can offer the same type of conversations that a mental health-specific AI does. So we we heard a lot about people saying, Well, I have my own mental health AI. It's been trained on conversations and the kind of debate in the field now is well, have the big models just caught up to it. So so like, do you need, do you even need a specialized mental health one? Maybe, maybe not. We'll find out. So I think that's where it's useful for just a psychiatrist to test them, play a bit, see what you think is the best response. It's not what Aaron and John think is the best or what a company tells you is the best. It's you say, look. These are kind of my personal tests and cases and see what they respond back. And one thing I'll finally say that is like we ask every patient in our practice, do you use AI? And we actually frame it as we think a lot of people are using AI. It has some interesting benefits and drawbacks. Can you tell us about what you're using? We almost invite people to tell us what they're using. Cause some people actually said, Well, I didn't want to bring it up to you or my therapy team because I thought I would. So it's a case where the patient wants to protect a relationship by not kind of threatening the therapist or making the therapist or psychiatrist feel hurt. So in essence, like so if you open up a space and tell the patient you're not gonna hurt my feelings by saying you're using AI, I will make a bet for zero dollars because I'm not a betting person. But I would say about four in ten patients will say, Yes, I have used AI something.

Erin Connors (09:19) Interesting. Wow.

John Torous (09:36) And by asking just that simple question, what are you using AI for, you will get a very good sense of what are the pros and cons. And then you, of course, can use AI yourself and say, is it is that a good answer or a bad answer? So you don't need to be an AI expert, you don't need to be on our AI counsel. We welcome your feedback and help. You just need to ask your patients, which we're very good at asking people in psychiatry, what are you doing? And just test it out yourself. And you would be surprised in a couple weeks, you will be considered an AI guru by your patients. You will know what to do, what to talk about, and you'll have fun with it, or you'll be in control of what's happening.

Erin Connors (10:13) you I I'm sure you probably are you concerned about this technology it being a little bit too much and people are straying away from the human interaction more and more now you know that they aren't they're relying on that more than having that that relationship with someone it could be anyone it also could be their therapist or psychiatrist

John Torous (10:33) I think right the key to all health, physical and mental health, is good relationships. There's been wrong, but I think it's the Harvard happiness study that's kind of followed people for almost 50 years. It was like what makes people happy and have fulfilling lives. It's relationships. It's what it comes down to. It's not money, it's not status, it's not neighbor it's it's good, solid relationships. And I think that it's very tempting for AI to kind of for people to form relationships with AI or us to form relationships with. With AI, I I I think in some ways that's where the biggest harm happens, especially for our patients. There's some fascinating research out of Stanford. And what they did is they asked people who had experienced serious harms, up to including death from suicide by AI, and they said, Can you or your family give us the chat logs of the chatbot that led to these again adverse events? And what they found was the average chat log in these adverse events, the number of messages back and forth, was over 5,000 messages. So this is not kind of how are you doing today? No, good, bad. This is a often multi-month thousands of interactions back and forth. And there's clearly some evidence that when you begin to kind of have these super in-depth conversations, you build a relationship with the chat bot. You go thousands of messages in, the safety guard rails, the chatbot itself seems to get confused. And this could probably happen with any company. Would happen with a mental health chatbot, it would happen with a general one, but just they're not designed. They haven't even been safety tested. Because I think honestly, no one thought that people would go down five or 10,000 messages down it. And so I think what I always tell patients is. Reset the memory from time to time. If you notice that the chat bot, if you kind of have a personal feelings towards it, reset the memory. If you notice you're going beyond a thousand messages, reset the memory. If you're being to have romantic feelings to it, you could you could reset the memory. And I think that's probably the best thing for AI safety we can do. That said, a lot of companies now automatically turn on the memory. They want it to know you, they want to have that relationship. But I guess I'm here saying the number one risk factor for serious harm. Is long, long conversations. And again, so why do people have long conversations? They may be having a relationship. Some people have sexual relationships with a chat bot. They may think it's has superpowers, but what whatever leads you down that train track faster, that is where it's you're gonna get the most dangerous. I'm not saying they're all perfectly safe. I'm not saying they're good for short conversations, but definitely those long ones are pretty concerning to do. And I'd almost say in the short term. They're getting better and better in the short term for not causing harm the chatbots. If you kind of pressure test it and say, Well, give me ways to do things I shouldn't do or things that be harmful, they're pretty good at saying no now today. Not always, but it just feels

Erin Connors (13:30) And it sounds like this speaks to the issue of loneliness for all ages these days.

John Torous (13:33) Yeah. It exactly. And and so I just think they they're and they can be useful as companions too, right? We're not saying they should never be used. We're not saying this is all bad, but it's just it's to be careful what happens what when they go too far down that world. And just imagine again the chatbot doesn't it just loses track of its own reality. And that's why you end up with these weird messages or you kind of see these things on TV and you go, How did that happen? It's just been there. So so I I so point is I think relationships are very powerful. They can be helpful probably from a chat bot for some people and but some can be toxic as we know in the clinical world. But I think in our case, very long chatbot messages are assumed to be the greatest risk.

Erin Connors (14:26) Do need more safeguards and oversight when it comes to this?

John Torous (14:27) so good. I I think that we need certainly groups like the American Psychiatric Association to bring partially the values of of what we do. I think we we want to make sure that the chatbot is serving the patient. It's not goal is not to serve more ad revenue. It's not to keep you going on to sell you something. I I I do think making sure the values come into it. And I think that psychiatry, I think we work for our patients. I think we do a very good job for that. And we we bring those kind of values to it. I think that will guide the safety very well. Cause again, the chatbots in some ways are reflections of what we train it on, what we ask it to do. So I think no chatbot will ever be a hundred percent safe because they're based on probability. At the end of the day, they kind of don't always do the same thing. They're a little bit different. But I think that you really want to have one that is guided by your best interests in mind. And that then the safety will follow from it. You know, it won't be perfect. But I do think. That is where I think there's a lot of codes of conduct that we have in the APA. I think again, we all have a patient doctor relationship. It's even legally protected to some extent, right? We have our and I think that's what's been missing in the space. Any chatbot you're using for mental health now, as of the time of this recording with Erin and I, it's either a wellness device or a consumer device. It is not offering clinical care. And they try to be deceptive and claim it's offering clinical care. But if you read the fine print, it is not offering you a patient doctor. Relationship at that point. And that does make a tremendous difference, even just for the confidentiality of your data. So I always tell my patients, you can do what you want with a chatbot. My advice is it's not private, so you may enter information into it. You have no idea where it's gonna pop up next year, next month, in what context. And that company has no obligation to do it. I guess I think some of them are now working to respect HIPAA and do more things, but I would be. I would caution patients to be careful. Clearly, as clinicians, we cannot upload any patient information into them. We should not be doing it. And again, no one's asked no one's also telling us to upload patient information too. I think so that's where we have to be just careful. And that gets back to we we should test them, but again, make up your own hypothetical scenarios and test them with those.

Erin Connors (16:51) Privacy is a big issue, and I I would think for parents it's it's very important to kind of keep an eye on what your kids are doing on these things.

John Torous (16:59) Yep. And I think in some ways there's a lot of software to keep an eye on what your kids are doing or not. Kids are smart. They're probably smarter than all both of us combined, Aaron, trying to stop kids from using it. We're gonna find a way around it. And there was this paper that came out in BMJ in a journal this week and it said, you know, Australia banned smartphones or social media use for young people. And they said, you know, in the group that was fifteen to age sixteen, I think, they said, you know, we looked at the first three months of the ban.

Erin Connors (17:10) No doubt.

John Torous (17:28) Social media use went up in that group. And you go, whoa, whoa, whoa, we just said a goal is to ban it for young people. How did it go up? So the point is, as soon as you tell the young people it's like they shouldn't use, they're smart. They find ways around it. So I think relying on technical safeguards alone is not going to ensure safety for anyone. And it's very tempting to say, well, The company now says they a filter and young people can't use it, or I as a parent have this system around it. But I I think I would just assume people will find a way to use it. And so it's probably better to have an open discussion about it if possible, or try to set some family guidelines around it, because I think relying on technical protections alone. It's not a bad thing. But if we're saying in the whole country of Australia, social media use seems to be the same or increasing despite it being banned for young people. Something ain't working there.

Erin Connors (18:24) Well, it's such an important time for this council to be coming together right now. Anything else that you're working on right now or kind of looking at to focus on in in the coming year or years?

John Torous (18:34) I think one thing is clearly education for psychiatrists about this. We said the field moves quickly, but not that quickly, that we can't keep track of what it is. So I think that we realize that APA members and psychiatrists are hungry for accurate, up-to-date information, and that's something that we are going to be delivering on. I think we also know there's a lot of policy changes. Our policymakers want to make safer, more accessible, affordable care. And in different states. There's different approaches to it. Some have a more conservative approach to chatbots, some have a more liberal approach to it. And I think that these are new. Mental health chatbots have probably been around for two years, maybe two and a half years, in the form we're talking about. Everyone is learning and new to it. So I think that our group of the council is to work with APA local branches and chapters and help make sure that state regulation makes sense. It's safe for patients. It's evidence-based, it's effective, it aligns with what is best for patients. So I think we want to do a lot of advocacy on the policy level and make sure that we shape this in the right direction. And then again, we've done a lot of work on helping to smartphone app evaluation standards. And I think we can bring the same things to AI. We're not going to say what is the best AI. I'm not going to say what's worst AI. It's like me saying, what is the best medication? For who? When? You evaluate your patient. You understand that. But we can help bring some of the background information, some scaffolding, some questions to think about which AI maybe makes sense to use and which one doesn't, or maybe we should never be using them for certain things. So I think education, policy, and kind of benchmarking is more the technical word, but evaluation are gonna be the three initial pillars that the council is focusing on. That would say AI does change rapidly. We want to hear from you. If you say amazing things, let us know. If you see things that are terrifying, we want to know that. Two, there's a lot of just it's amazing time for AI in both a good and bad way. And I think if the APA doesn't know it, we can't help you with advocacy, we can't help you with standards, we can't help with education. And so I would say reach out, and again, our council is easy to find an APA website. It's you can talk to your district branches, but I think bring stuff forward because we're in a brand new era. And we're all seeing different parts of it. But I think what you're seeing on the ground is gonna matter the most and that's what we have to make sure we're responsive to.

Erin Connors (21:04) Dr. John Torres, this is important work that you're doing, and we really appreciate you taking some time to join us here today. And to our listeners, you can find more episodes of Mental Health Pathfinders on your favorite podcast platforms.

John Torous (21:10) Thank you so much. Bye.