The AI Dilemma in Mental Health: Balancing Administrative Efficiency with Human-Centered Care

Richard Zwicky: Hi everyone, it's Richard Zwicky with High on Healthy, the show where we know there's no one path to healthy, so we explore them all. Today we're going to look at AI and the question is basically can AI fix mental health care before it breaks and before it actually breaks mental health care itself is part of the paradigm of that and paradox that we run into because AI may transform a lot of the infrastructure of mental health long before it transforms therapy itself but where the lines are and where they cross will become blurred. Joining me today is William Shredder, who is the co-owner of one of the largest independently owned mental health practices in Austin And Billy oversees more than 50 therapists and thousands of therapy sessions annually. Welcome aboard the show. Billy,

William Schroeder: Thank you It's nice to be here.

Richard Zwicky: So right off the bat, one of the things I noticed is you're overseeing 50 thou therapists and tens of thousands of patients every year. How many patients does your clinic deal with on an annual basis or how many sessions do you hold? Because an individual therapist if they're seeing somebody once a month means they see somebody maybe they can see 130 different patients a year. What are you able to do?

William Schroeder: So it varies depending on a variety of things, but somewhere around 35,000 is what we're looking at it currently. And we're always looking at how to be able to expand that in kind of a qualitative way, meaning there's been a lot of pressure in the field to push metrics. So instead of having full-time being 20 clients per week that bar is continually moved to being 24, 28, 30, all the way up to 40. And that's part of the reason why we're seeing a lot of burnout and less quality like it be an issue in the field quite honestly.

Richard Zwicky: You need time between sessions to decompress. You're not just sitting there passively, and

William Schroeder: There's a lot that goes into that,

Richard Zwicky: It's not regular job is the wrong term, but a 40-hour a week job. You can't see patients for 40 hours. You need time to record the notes and to prepare and to decompress.

William Schroeder: Whether it's dynamics, coordinating care. I mean, there was a period in my own life where I was doing 30 to 35 client hours per week and I joke that I was sponsored by Red Bull or five hour energy because it very much felt that way. It's just I mean you think about any kind of job where you're having 30 to 35 hours of conversation where you really have to be very deeply aware attentive and working on planning in between there just because I think part of what makes good therapy is doing your own work doing research taking time between sessions to reflect on cases doing consultation so there's a lot that goes into that and I think that's the challenge in our field right now there's been a lot of private equity that's come into the space because of that they're like wait we can totally push these margins further and further because there's such a need let's see what we can do by pushing this number a little bit higher and it's led to a lot of burnout and people leaving the Build. Yeah,

Richard Zwicky: You made me smile when you mentioned about Red Bull and 5 hour energy shots because the people who've worked with me for the longest used to know that whenever I have weeks packed with meetings, I'll need four or five meals a day because you burn a lot of energy just thinking and consuming that. And the energy shots alone don't do it. And we underestimate that impact.

William Schroeder: There's a lot of cognitive load that goes to the work and emotional and stuff of course too because I mean I think there's no one client that you're going to get consistently it's like there's a variety of different needs and things that are coming up in everybody's life every week.

Richard Zwicky: There is. And that brings us in an interesting way to AI because people think the AI revolution in mental health is about the tools becoming the advisers, the tools who are, always on and always at the ready to answer my question as if we could just dial, where we can dial 988 or the for assist in certain circumstances and critical, but that these tools are always there for us when we need them and they have the right answers. But that's really not necessar it's not necessarily true right off the bat, but it's also fundamentally a part of the conversation we're missing where it's not just about the AI becoming the robots that are basically functioning or filling in gaps. It's a lot more. How do you look at it?

William Schroeder: I try and balance the perspective because there's one perspective as I think the clinicians in the field are very worried about kind of the coming robo apocalypse of stuff for therapy.

00:05:00

William Schroeder: I guess my perspective as a clinician is that we'll probably see a future of where there's AI that's involved and we're already starting to see this as Talkspace and Headspace and some of the other products have EB and these different things that are going and doing kind of chatbot therapy. I think the challenge within that is the liability piece because they aren't licensed and there is the concern of what do you do when something goes wrong and right now they're trying to deny all culpability for that and they're also trying to get around some of the stuff with it hallucinating by having shorter sessions. I guess so that's the tricky side. The positive side that I can say as a practice owner is I feel like there's a lot of room for growth within this of helping of dealing with we have 50 clinicians a bunch of insuranceances that we work with things like that so helping with compliance testing helping with rate renegotiation with re insurance companies for anybody that doesn't work with them day in and day out they are just a very tricky beast and they will make themselves compliant but make it very hard for everybody else. For instance, they put out all of their different information about all the rates that they have anywhere like in any market, but they put it out where it's like you need a supercomput to be able to go through all of this stuff just because the amount it's so many terabytes of data. No normal human can be able to go through this because you're needing PowerBI and all these different tools that are in, having it all hosted in AWS just because there's no way that your normal laptop or home setup is going to be able to go and manipulate this data. And so there's a lot of things where I feel like AI on the application layer for practice owners is really helpful for claims dealing with denials. And even like going through and looking at outcome data. That's something that we're continually paying attention of looking at what are our clinicians doing best and how do we continue to match qual care and also as well as what are clinicians bad at that they may not be aware of. It's something that's really important just because inevitably I feel like they're all things that we know maybe that we're less comfortable with, but there's also things that they may not recognize that there's a deficit. I was talking with somebody right recently that was like, " for me that area was substance abuse." And it took me a little while to recognize that's something that I shouldn't be doing. And so it's helpful to be able to go through and have something to help. It's not about replacing the person in the room as much as it is about helping with the bureaucracy of running a practice s And that's something to know that I think that there's the compliance piece and things like that can be really helpful with AI because currently a lot of our clients as we even introduced AI being a part in our paperwork of it's part of our EMR but we're not currently using it. And it immediately it raised a lot of red flags for clients where they were like, I don't want an AI scribe in session. I don't want this thing recording anything that I say just because I'm worried about how this may affect me down the line. I'm worried about the security of this. And to be honest, there is that. I mean, if you look at any of the cyber security conferences that are happening right now, it's like Windows 95 right now where all of a sudden we plug this thing into the internet and it's a bit of a wild west. The security stuff is a big concern around this. It'll get sorted in time, but there's just a lot of that's, still very new and there's a lot of concerns with that.

Richard Zwicky: . So, one of the ways that people are using AI today is to self diagnose, to prepare themselves, to start asking the base questions before they go see the therapist or to ask the questions between therapy sessions both as a second opinion to validate And of course, the AI tools for the most part aren't really, even the ones that are AI focus are mental health focused don't have all the guardrails in place to protect people. There's going to be some conflict between what the AI systems tell people and what the therapists tell them. But what I noticed when I built my prior analytics company, we learned that people type their darkest secrets into the search engines. They self-reveal. Because they don't believe they're being judged by the technology. They do have to get over that hurdle with the therapist where they do innately believe another human will judge them when they're revealing things. How do you help them overcome that challenge? But also, how do you help them use that leverage and explore that behavior that self-disclosure in the same way? Or how do you use both sets of data?

William Schroeder: So, one thing that I'll say is trust it is so important and finding a pairing or what have you. And when it comes into the therapy session, a lot of times there's actually different signals that people give off unconsciously, I remember I worked with a kid that was sitting on their hands a lot and was very shut down. And on its surface, this client came in and it presented as anxiety and maybe some depression. And having done this a little while, it absolutely flagged for me as there might be something that's a lot bigger that's there. And with any teenager it's not going to come up very quickly or easily. Trust is earned in inches not in miles. So sure enough there had been sexual assault that had happened. And then there had been some other stuff that came up from a family member. And so it was a really challenging case and that was something that wasn't something that somebody might say something to a chatbot around the edges or, maybe they disclose something like that there. But ultimately, it's like I was able to go and do EMDR with this client and help them to be able to go and work through something that had been really truly a hard, scarring, traumatic life experience that they had experienced at a young age. And it's something that it's I think that there more tools that are coming, but there is something to being a clinician and working with skilled clinicians that are trained and have seen these different things before. And so much of it can come through the body language that you're not going to get transmitted on a layer, through chat. And there is the different stuff that you get. It's kind of like when you think about it when you walk into a room and everybody's taking a test and you're like, " my god." You feel that wave that shoots through your body. And it's that reminder of our nervous system. It's constantly it running and it's running even faster than our brain can process that stuff. That's why you'll feel it in your gut and in your body before your brain's like, "Oh crap, we forgot about all of that stuff."

00:15:00

William Schroeder: And so it's that reminder of we get so many signals from people and that's something that the LLMs are missing at this point and maybe they'll be tied into more biometric data. Maybe we'll find some more different stuff with that. But currently I see more people coming into care because of they've had some back and forth with this. We're seeing sometimes even giving them summaries of this is what I'm seeking care for because they've gone and had these conversations back and forth. And to me this is just the evolution of what we saw in the 90s and 2000s of where people go into different forums anonymously and talk about what was there and unburden themselves about their experience and trying to be transparent. And it's hard just because there are a lot of different things that people don't want to say publicly, whether it may be, around sex and sexuality or gender and they've been in a marriage that has never been allowed to question that stuff maybe they're in a professional field where that's something that if exposed could have concerns about consequences. You think about the different scenarios that have come up in past history of polit politicians where it's come out around some sexual behavior and all of the sudden it was the downfall of their career. So these things can have real consequences and my concern as a clinician is absolutely the safety of that client but you're right these tools they give you anonyomity. I think the challenge is continuing the challenge of the different LLMs is they're designed in such a way to be able to encourage interactions and so it's the question of are they also doing their job to try and make sure that they're providing quality care and as a clinician I think that's the challenging piece. It's like we had somebody that came into our office recently that was really in the throws of a bad episode of disordered eating and it had been really reinforced by the LLM that they had been interacting with. It had not been seeing that this was a problem. It continued encouraging and basically seeing in their conversations with it what was revealed to the clinician was something where it was just continually being like no then you need to go ahead and try eating less and eating less and eating less if you're not getting results with this and so it's the challenge of this stuff because I think it's very powerful I think over time these things will improve and I do think that clinicians it's the most challenging thing in our field to have trust with clients and to earn trust with clients Because you're right, it's not going to happen with everybody. It's, almost like going on dates. Sometimes you walk into a room with therapist and it's like walk watching paint dry. That person could be the most boring person in the world and yet they may have every degree in the world. So, it's trying to make sure that things are a right clinical fit. And I kind of wonder if in the future we won't see something of where for the people that have good insurance or can afford insurance, they will then have a human and maybe there's some AI that helps to try and make sure that different things don't fall through the cracks around medication compliance, things like that. However, I think it's possible for the people that have less good insurance and things like that, they may then be faced with something that's more automated. And it's not to say that's bad. I could see a critical role with helping with training different people that are out in the field like providing services to those that are unhoused or unsheltered or churches as they're trying to go and improve the training maybe it's around CBT or things like that or learning different skills to be able to help distressed couples and certainly it's something that could be helpful in roleplaying that going through that and going ahead and assisting in some different training that they otherwise wouldn't be able to afford or access.

Richard Zwicky: I mean, there is an aspect to it where it will democratize access. I've spent obviously a lot of time in the space. And even with the platform I built with the guardrails, it's possible to give people self-awareness and insights, but therapeutic practices delivered through an AI tool are still far off from being able to come close to what a therapist can offer. But there are alternatives and there's ways to bridge that gap without the AI being the one that creates the therapeutic practice. But I'm not going to go into a long sales pitch about how to do that at this point. But one thing that as I did develop things and ties into something you said earlier is what we've ended up building is a relational intelligence platform which can match people in various contexts among other things. And one of the things you mentioned earlier is one of the biggest bottlenecks in therapy is getting people through the right door but to find the right clinician. You don't want to be sitting down with somebody with whom you just don't click. Because you're never going to get the help and AI could become extremely good or is extremely good as I've proven now at matching people matching clients to therapists.

00:20:00

Richard Zwicky: How will those types of tools and then using the tools to monitor and help people between sessions impact and help alleviate the problem because in so many areas if you're going through the public channels you've got a 10 to 20 month wait to see a therapist and somebody who's in critical care can't wait that long. So how do we mindfully take that understand one we can speed that up by matching the right people but using AI to shorten the outcome time so that we don't have that 10 or 20 month right?

William Schroeder: It's a math problem ultimately as part of it. Like we were just talking about we're a site with UT School of Social Work and part of what we were talking about on the last call is there's a shortage of 33,000 professionals in the state and so it's like they can go ahead and they are doing their part trying to train as many people as possible but inevitably it's a challenging game just because every year you have people that also leave the field and it can be things like they get tired of fighting with insurance companies about denials. I was talking with one practice

Richard Zwicky: There.

William Schroeder: Who was telling me they had an insurance company go because they implemented AI, it lowered their denials down to 1% or 2%. And so then that insurance company went and required them to send a paper copy of all of their clients for the past two years to them. So it was pallets and this is a huge practice that they have 30 locations across the state that they're in. So they had to send pallets of paperwork and print out all of this junk just to be able to send them all of the case history, all of the notes for hundreds of thousands of clients. So I think it's that reminder that there is this burden that happens in the field. So anyway, getting to your problem of access, I do think that it's hard. I don't have a perfect answer to that. I do think that ultimately there is going to be this need of a two-pronged approach within this, of how do you go and have more one to many solutions that can help people and there are certain things in the field clinically for instance there's a form of EMDR that can be used and they used it after the earthquakes to be able to go and help this one to many approach. This guy everybody calls him He's a trainer out of it's his nickname, I'm totally forgetting his last name. But Nacho trains people. He's from Mexico City and he trains people all over the country and all over the world and the protocol that he trains people on is one that can help that. I think things like that are really great. I think also getting cities to look at how they can be involved. For instance in Austin they went and they created a program that requires clinicians for the clinicians going through eye movement desensitization reprocessing training. It's something that can help people with trauma and PTSD. And so they actually fund this of around 50 clinicians per year and they require them to give back a certain number of hours to different people through victim services for people that have been affected. This program has been so remarkable in its outcomes and being able to help people because the standardized protocols that they've also started looking at how do we implement this for the people in the 911 call centers because those people are regularly being traumatized with the different things that they're going through. How do we potentially expand funding for this to go ahead and help more people more quickly here in Austin. So there's protocols that I think also can be part of it. But no doubt it's going to be the challenge of how do we implement this stuff at scale and keep in mind privacy. So if there is something there's a Chad GBT of therapy that really does truly come out and show a lot of promise is trying to make sure that this stuff remains private. We've seen what happens when it doesn't. We talked a little bit about the security side, but I think it's important to reiterate in Finland they had an EMR where it got exposed and initially they got ransomware because of a bad security protocol. It was something very simple that they messed up on. They didn't pay for the ransomware and then all of a sudden they went after each it was 20 or 30,000 different people that had been in this EMR that they went to directly showing them, copies of their session notes from the therapists that were in there and saying, if you don't pay this ransomware, then we're going to expose you.

00:25:00

William Schroeder: And so that's something that's like an important piece of this is trusting that the different data that is being fed into these different things is being kept private and secure because all of a sudden I don't want my chatbt therapy session to then all of a sudden started influencing my Amazon list or potentially I'm kind of being a little bit joke joking.

Richard Zwicky: But you may say it facitiously in some degree, but the reality is when people enter information into chat GPT it's not private. And their terms and service say they're allowed to mine it for training, but training is undefined.

William Schroeder: And so it's the question too is what are all these things being trained on which is another thing being like is this being trained on all populations or some populations. And so I think that's the challenging piece with this too is to try and make sure that what's it's the trust piece. It's like if people can go in and have these conversations with the LLMs and I get that as we're talking about ones that are more therapy specific. There's the con making sure that all of the PHI is stripped out as it goes to the different model and then it's reinserted as it comes back to whatever or just making sure that it's secure because there is the big concern about how does this maybe affect somebody applying for life insurance? How does this come up in a divorce? How does any of this stuff come up in all the different variety of ways that I'm not even thinking about at this moment of where it could affect somebody's future life and overall their health and safety because that's something as a therapist we're constantly thinking about is making sure that we have litigation that inevitably comes up as parents go through tricky divorces and it's something that we're very careful about in our notes and our protocols and everything that we're doing that it's something where we're following the letter of the law of what's being required for us in our ethics as well as our boards, but also trying to be careful about, are we documenting something that could later be a liability for this person depending on how that's expressed. And so those are the things that are challenging and can certainly be worked around. But as we go through this, I think it's that reminder of I think trust is lost in miles. And so as we have different things that may go wrong within the AI space because all of this is scrambling so fast and having come from and don't to sound like I'm some keragin that totally doesn't like technology because I started out working in the Silicon Valley. I worked for technology companies. I worked in IT. I worked as a systems engineer and sales engineer. I get how all of these things go and I'm the one that goes and I'm using claude code regularly. So I am totally the type of person that embraces technology, but I think it's something of being cautious in the approach and how you're doing stuff. And I think the challenge is all of these things are scrambling with funding and trying to be kind of the one that dominates the field. And so I think a lot of the guard rails that should be there are missing at the moment. And then the worry is what's going to make it so that makes them cautious enough to do that. So

Richard Zwicky: It's critically important and the word guardrails comes up a lot and it's something which always incredibly mindful of as you're developing and using anything with AI is setting up the right guardrails. And it's do no harm. It's much more than that and we can chat GPT and all the different things. We can go in at nauseium about the different aspects but one thing that does it does bring forward is many people don't know what they actually need when they seek help. They go to and they turn to their AI system with specific questions that the systems answer but not a holistic view on themselves because we're the only normal person in our universe. So, we don't realize and recognize what our differences really are. We're only aware of certain ones at certain times and the differences are the ones where we haven't figured out how to build the bridges to everybody else yet. It doesn't mean that everything else is the same as everyone else. But when AI is designed to make people feel good, because they do have a certain amount of psychopantry built into the way they respond to keep people engaged, that also means they don't tell people the hard facts. How do you have your therapists when they're speaking to patients guide people that, hey, I know you're going to go use an AI tool to ask questions about XYZ. Here's how to use it properly to learn useful information not just reinforce what you already know.

William Schroeder: So I think honestly there are a lot of people that are using their reading for things like that. They're going and they're reading Bassel Vanderolk's latest works or things like that. We're certainly finding a lot more people that I think are coming in with a higher education and

00:30:00

William Schroeder: Awareness of kind of what may be going on. Then there are also the other people that may not truly know to be honest. It's like, and it's not about a lack of education, but there are things that let's say somebody was sexually assaulted as a child. The brain has a brilliant way of blocking out that stuff is called disassociation. And so by dissociating, all of that stuff is completely it's like you can think about it like the fuse box in a breaker blowing. And it keeps them safe from having that be something that they're having to constantly process dayto day. But as an adult, they may notice certain things that come up, whether it's around sex and sexuality, maybe it's sleeping in with the lights on, maybe it's not sleeping in their bedroom. It can be so many different things that come that way. And when they come in with a skilled therapist, there'll be these different things where it's like, as they're talking about these different things, it may be slight body, signals or cues that come up with that. Maybe it's some piece of their story that just doesn't add up with kind of your learned or lived experience of this person. And so, it's a process of trying to then close those different gaps. So as far as using chat GPT and things like that to be able to go and continue education I mean personally I can see number one I think there's a lot of good podcast and resources that are out there and I still encourage that with clients. I think there's a lot of different good literature and stuff. I think, it's possible that people could go and try and utilize those different things and be like, how would Bassel Vandel, look at this problem that I'm having or how would John and Julie Gottman look at this issue that my wife and I are having. I think the challenge may be what it's missing because for instance in John and Julie Gottman's work which is based on 40 years of research that LLM also isn't doing the several hour intake that you have to do to work with a Gotman couples therapist and that has the wife sretto and all these different inventories that are in it that are giving that clinician when they're meeting with them a ton of baseline data and it's not to say that chatbt may not have some of that data that's there or whatever they're working with. It's still stripping out, I think, oversimplifying different things because it's not talking about the white sretto for instance, the thing that I mentioned there. It's a inventory that looks at basically how toxic is this marriage is ultimately what it's looking at. It's looking at how many times this is divorce, mentioned, how many different things that have happened that are basically on bad healthy and they're bad health indicators for relationships. And so certainly CHBT may get some of that different stuff or whatever somebody is using and certainly it can be helpful. However, I think it's just that reminder of the importance of different things that therapists may be using that's more unique and over time we may see a leveling of that. Maybe John and Julie Gobin, I believe, sold their institute. And I don't know who the new owners of it are now, but that may be something that they're thinking about of how do we go and make this assessment be augmented by AI so that somebody could go take this stuff and we pull in the data that, model knows about them to then have something that's a little bit more personally tailored and closes some of those gaps. Maybe it also takes in some of the data from their biometric devices to know all right, the spike that they had at 9 p.m. Of their heart rate, was that a fight that they had with their partner or was that something of where they had sex? Was that a negative cue? There may be more data that's helpful. So I mean, I guess ultimately I would say in my mind, it's still trying to look at what are those different things that are the high authority sources? And to me, I think that there's still a little bit of a barrier that exists with that. It's not to say that it's bad. I mean, certainly you can go back and forth with it. Have I ever used it? Yes. Ultimately I think I view these different things primarily as do I use it often? No. I have my own therapist as somebody that, she has 35 years of experience and she's trained at the Wazoo and she's incredible and I've been able to make more progress with her than I had in the past 20 years. So there are those people that are really skilled that can help you and make progress more quickly and there are protocols that can help you with that EMDR I keep mentioning as one because it's one that's kind of blown my mind as a therapist and that in six sessions for people that have had a single incidence of trauma 99% of people no longer meet the criteria for PTSD. So, there's things that are like that have a lot of good effic efficacy behind them and very good effectiveness. And will we see some of that stuff potentially augmented with, I'm sure. But, certainly at this moment when I look at the tools that are out there for people so much of I think is around education and trying to pull in those different things that can be helpful. If it's around marriage, listening to different podcasts, that can be helpful for that.

00:35:00

William Schroeder: Reading different books that can be helpful for that because no doubt the more that you're aware of the different resources the more that you can implement those different things and the more that it helps you in your awareness of improving yourself in an intelligent way.

Richard Zwicky: No, it certainly is and it's fascinating and we are running out of time but one last question regarding it which isn't a short answer so don't worry but across all these conversations in our conversation today as well there's one tension that keeps surfacing and that AI may become incredibly good at delivering information, but therapy isn't really anformational problem. It's a relationship problem. And people are anthropomorphizing their AI tools. People name them as opposed to, anything else. They start thinking of them as an adjunct to them, a help. And we've explored with previous guests and how AI in a sense becomes an attachment figure. And yet for so many people it's the attachment issues that are at the root of so many of their mental health challenges. U AI though is always available. It's always patient. It's validating through psychopantry and it's responsive. From a clinician's perspective is there danger of creating an emotional system and layer that never challenges, disappoints, frustrates or requires reciprocity when it comes to mental health. I think it's art in general, but when it comes to

William Schroeder: I think it may not challenge them, It can go and create an insulating barrier between connecting with others in the outside world. Personally, I'm somebody that's done a lot of work with people on the autism spectrum and absolutely this is one of those things that can be a very attractive thing to a variety of different people, but definitely people on the spectrum for getting, feedback and being able to feed a lot of data, study a lot of data. I think the challenge is ultimately personal connection. We're wired for it and it may be hard and so we've got to look at those different things that help encourage us to do those things that are hard and to bridge those different gaps because I think it may practice some different skills which is great for instance with a lot of my clients that I've worked with over the years they've done different things whether it be D\&D groups or improv comedy and things like that and it's the process of practicing getting yourself out there in front of others and putting yourself in uncomfortable situations. So certainly chat and stuff like that may help to practice some of that different stuff. But if it's always in that same protected layer, then I think that's the challenge is, getting people out there and taking those different risks. I have a buddy of mine that talked about making this movie and I wish he would do it. The short synopsis of it's called the diving board movie as a working title but of there's all these people in the Austin community that are kind of these renegade divers and the short synopsis of this is the importance of taking different plunges in life and trying these different things that may make us uncomfortable and continually trying to level up within that and that's a really important skill that I feel like maybe being slightly debased by all of these different things that allow us to live in a comfort zone without the real challenging that we may need. And as I think all of these things are further and further pushing towards monetization, I wonder how that ultimately weighs out and at what cost, whether it be the data that's been harvested, or, the pricing models that may be put behind it, what may be there. It's like we've definitely seen a lot of different people where they've come into therapy because of different things like going on where maybe they've been in a bad way and they've been posting too much on Facebook and then Facebook shuts them down. So it's like at what point do things like that happen with chat GPT and all the LLMs of where people are going, to a dark place and then it's like here's resource. All right, we're shutting down. Cuz that stuff is also going to happen just because at the end of the day, these things need to monetize. And if you look at the latest stuff that's come out from ChatgBT, they're like, hey, we're going to be clawed because we realize this chat stuff isn't truly the total focus of how we make money is we got to be able to do stuff. So now everybody's trying to follow anthropic and so I don't know. I think I have concerns as a professional in this space just because these things don't have a license and they fought in a very enduring and durable way to try and make sure that they are not held liable

00:40:00

William Schroeder: If something bad happens. And so that's something that in my mind should be an alarm bell for people as they're looking at this. So I don't know hopefully I answered your question with that.

Richard Zwicky: You did and the whole liability issue could take a whole show and segment just on that because there's a lot to discuss there and there's a lot that people overlook but sadly we are out of time for today Billy but this has really been a great conversation Thanks for joining us and

William Schroeder: Thank you, Richard.

Richard Zwicky: Thanks for joining us on High on Healthy. And thanks to everybody for listening. I'm Richard Zwicky and we'll be back again with you soon.