The Danger of Outsourcing Identity: Navigating Chat AI Confusion Syndrome

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. If you haven't done so yet, please do subscribe via your favorite podcast listening tool. And we're going to continue on a theme which is close to a lot of my work and what I do, which is talking about mental health and AI. We've looked at all sorts of different aspects, but we're starting to cross over between the different areas of where AI works and where it doesn't. And one of the really interesting discussions is AI identity and are we outsourcing ourselves to a machine to tell us how to act and operate in certain challenging circumstances. And we've been told AI will make us more supported, more self-aware. And It can be used by crunching data to give us great feedback. But what happens when we rely on it for clarity when AI doesn't really have it and we trust it? Because identity isn't formed through perfect answers. It's formed through friction. Relationships are loaded with friction. And it's about understanding the other person's normal. As I always say, and I'm just finishing a book on this topic, I'm the only normal person in the universe, aren't you? Our normals are each equally correct. And that imperfectness is what makes us incredible. So joining me today for this discussion is Phil Mloud, who's a practicing therapist and founder of Thought Reader. And Phil's been spending a lot of time looking at what he calls the chat AI confusion syndrome. Welcome to the show, Phil. It is and

Phil Macleod: Hey, thanks for having me. It's great to be here and it's a great topic for us to chat about

Richard Zwicky: People turn to AI for everything by now. And one of the worst things AI does is when you ask a refining question or say, I think you're wrong, as a person, it come backs and confirms, isn't that amazing the insight you had? And proceeds from there to tell you what you want to hear. And if identity is something we're supposed to develop over our lifetimes not download, are we not really outsourcing our identity, using this process, trying to make ourselves psychologically whole?

Phil Macleod: It's a great point isn't it? And you started off on this point. One of the things around AI is it is actually imperfect. And I think many people don't understand that as a starting point. It's not a de facto standard. It's a lot of algorithms trying to work out what you want to hear. And depending how you ask the same question, it will come back with different answers. So by outsourcing your own rational creativity or your own rational thinking to an algorithm, you're actually diminishing your own conversational power and your own strengths. And there is a fundamental use for AI, but I don't believe it should always be there for answering the common questions we have on a daily basis. As a human species, we should be challenging ourselves. We should be learning through mistakes. We should be learning through peers and talking to Outsourcing that to effectively a large Excel sheet is not going to help us grow as people. And I see that quite a lot. And that's where the chat confusion syndrome came in.

Richard Zwicky: And what's the most telling sign that somebody actually has crossed that line where they're depending on the AI tool to tell them how to think or how to interact because there are tools which are useful like we look at Grammarly we look at other things which can help us interpret the subtext but understanding the subtext which especially people who have autism or certain forms of ADHD with RSD they miss that subtext they need help. So where's the line? Where's the line?

Phil Macleod: I think the line for me when I'm talking to my clients and things when I'm out and about is when they start to stop thinking and they check themselves. So even the most basic of questions that they would ask themselves, what happens is instead of going straight to their knowledge base their understandings, they'll go to chat. Have Should I? Is this right? But actually if you stop you trust yourself more and love yourself more you would never ask anybody that let alone AI. So for me when it starts to take over day norms that's where the line starts to cross and then slightly back from that when you talk to people about how their use of AI when it becomes the de facto standard before doing other avenues they're on the border then so if you think about growing up when you first moved out of your house and you went to another house and you couldn't work the thermostat or something in the house wasn't, useful, you would phone your parents or a friend.

00:05:00

Phil Macleod: You would I'm used that as a

Richard Zwicky: How to work my thermostat. Right.

Phil Macleod: Because I'm looking at mine and it's still set to auto. But absolutely right. None of us really do, but we would reach out to people. We would, phone somebody up or we talk to a neighbor. Yeah. What about this? What people do almost instantly is they'll hit a stumbling block straight to the phone, straight onto chat. What does this mean? And we're stopping our brain development our usage of actually developing problems solving skills. And that's a dangerous line.

Richard Zwicky: It is a dangerous line and but it also reflects a challenge in our education system where critical thinking has dropped by the wayside where we're informing but not actually challenging people to consider why it happened. Most importantly, it's more based on these are the facts. And AI has been trained to be confident right? It's designed to sound articulate and certain, which reinforces that. Does that Yeah. Mhm.

Phil Macleod: And it's a false confidence as well because we know that is fallible. And we know that because if we go back and ask AI the same question three times in a subtly different way, you'll get three different answers to the same question. I mean, the famous one that was circulating a little while ago was, "I've got a flat tire on my car. The garage is 50 yards away. Do I drive or walk?" And it comes back and says, " you should walk." I want to get me flat tire fixed. It just doesn't understand some of the subcontext,

Richard Zwicky: It does.

Phil Macleod: But it comes back in such a clear and confident way that unless you're looking and challenging, you do believe it. You go, "Okay, I'll walk to get my tire fixed or I do believe that this is the right drug for me or I do believe that I've got depression or whatever it might be." because it comes back in such a confident way. Yeah. Yeah.

Richard Zwicky: And it's a challenge because people use it to check themselves. Am I handling the situation that I really desperately care about properly? Is there something I should be doing different? Especially with relationships, how can I help the other person? And they're not trusting their own judgment or their intuition.

Phil Macleod: And that's the thing. It's about trusting yourself and loving yourself enough to realize sometimes you don't have to be perfect. I think when you use AI as a sounding board, that's okay. What and you come back and question it. But when you go into over reliance, which is the problem, that's really the issue. And I go back to it all the time when I talk to people around it. AI is fallible. And I go, "Really? I thought it was always right." I'm like, "Where did you get that from?" have you not learned? And when you start Yeah.

Richard Zwicky: It's not a curated form of the internet. And there's so much disinformation on the internet. It all gets sucked in. Yes.

Phil Macleod: And the trouble, I mean, I talk in other areas with other people about this, probably five years ago or possibly 10 years ago, a debate used to be happen on a fact has happened. Do you disagree? And what your views on that happening. Now you even disagree that the fact has happened. So extrapolate that out to AI where it will create its own facts for you. You're really in a dangerous thing if you overb believe it and you don't question it and challenge it. And I think that where people are very self-aware and very analytical and they have some other things going on in their life, but they don't trust themselves enough, that's it. They'll turn to it as a please prove me right and then they end up arguing with it, when they don't get the answer they want.

Richard Zwicky: It's not going to help you at any stage.

Phil Macleod: And that's another sure sign that you're using AI incorrectly is when you're arguing with the answers. If you're arguing, you already have in your mind what you think the right answer is, why we're going to this tool. And secondly, you can always get a second opinion, but arguing with the person giving you the first opinion is not going to get you where you want. So, yeah, absolutely.

Richard Zwicky: And part of that challenge is really we're using I'm a big believer in AI. I've built a company which a predictive analytics company and the only way to crunch the data to get the insights necessary it's around mental health but to get the insights necessary is through because AI is a big data processing with a natural language interface built on top of it from the most simplistic of ways of explaining it and right so it's making it instead of you having to sit there with an Excel spreadsheet and figure out what are the formula you can describe what you want it to do and it'll bring it back and if we think of it that way it works and

00:10:00

Richard Zwicky: When we think of it that way, it's able to take the data and offer perspective and insights. But that can be incredibly powerful because it can give you the insights of what to speak or when you go to a therapist, how to bring something up that you wish to speak with about them with. It can give you the guidance but it doesn't replace the therapist insights who looks at your tone of voice and can ask the hundred question relevant questions about your history. How do you see people should use the tool? Y

Phil Macleod: So you've got a really good point, So, again, AI is a massively useful tool and where people are using it for, I've got these symptoms. I fall into the realms of this? Where can I get help for That type of way of gathering intel is fantastic. And that's how people should use it. What are the common signs? What are the symptoms? Brilliant. Understand them, and then reach out.

Richard Zwicky: All right.

Phil Macleod: What AI can't do, and this is where you have GPS and doctors and nurses and so on, they have a wealth of experience of dealing with their neighborhood, with their community, and with people. So, they can relate back to in your community, as a pa person, as a black person, as an English person, as a white person, this is your lifestyle. AI doesn't have that context it doesn't have that framing so asking it things directly are not going to get you the responses you want but asking it for advice around what are the frameworks what are the indicators where can I get help where can I reach out to it's a great way to use the tool

Richard Zwicky: Right? It is. What happens in therapy when clients go and use the area tool just like you said about going to a doctor looking something up, do you need to see the doctor about this? On forth. They're processing their assumptions and then they're going into therapy and they're might be hearing something contradictory. What happens, right?

Phil Macleod: You're absolutely and that does happen. So if somebody comes in and I actually had a client about two months ago that had spent over 3 hours a day for 4 days diagnosing themselves and they came in to tell me what was wrong with them and how I could help them because I'd also googled how can hypnosis help this and chat told them exactly what to do and we sat down and just talked it through and what I tend to do is I tell them to I get them to relax. I just get them to think through in their own mind logically what is going on in your life. Forget what you've been told for a second. I'm not going to guide you anywhere. I want you to guide yourself. Think about this logically. And then we have a bit of time to rewind some of the input they've been given with a calculated stop and we move forwards on the basis of what's right for you. What do you believe is the correct way forward? Forget what machines are telling you. Even to a large degree, forget what I'm telling you for a moment. What do you believe is Where is your moral compass? What's happening here? So, we then move forward and guide people through. Quite often what happens with them is they go into this loop because as individuals, we want to have certainty. We want to have right and we want to get it right. So we'll ask a question and as we're asking that question we will say right good and we'll move on and you get into this loop of I'm not sure I'll ask again I'm not sure I'll ask again and in therapy what we do is we break them out of that loop and that's the big part that I see quite a lot in this situation certainly where we take a lot of time to bring them out of that, loop of conversation, loop of certainty and say it's okay to be wrong. It's okay to be human. That's an amazing thing. Stop for a second and realize how amazing it is to be human. And then once we've done that, we bring you out of the loop process. To get your brain engaged and recalculating things for yourself. And under hypnosis and under trance we explain some of the benefits of AI and how we can use it differently subtly etc.

00:15:00

Richard Zwicky: And something you just touched upon there is also really important to go a little further is every one of us has a critic inside our head who is talking to us all the time telling us and unfortunately it's usually giving us the worst case scenario some of the negatives of what could happen what will happen to protect ourselves from the hurt for when something happens which usually won't be as severe as our worst case scenario AI is doing that also in a different way and it's causing us to question the agency that we Build through imperfect decision-m Right. Yep.

Phil Macleod: You're absolutely right. If you think about our decision making is absolutely based on us doing the best thing for us in that circumstance with the knowledge we have. That's how we decide things. What's right here and we carry on. AI has a perspective of I want to get more people using me. I want to have more likes. I want to be right. But it doesn't care about you as an individual. So as a result of that, the perspective is already skewed for getting answers back. And you're absolutely right. What happens is our brain goes into almost overload. We almost have paralysis of decision making because we're always trying to get it right. We have this Jimny cricket thing on our shoulder telling us what's right. So we start listen to it more

Richard Zwicky: And it has no context of our lives up to that moment.

Phil Macleod: If we stop. Correct. Absolutely.

Richard Zwicky: Or if it does, it has only the snippets of those times a day where you asked questions which were skewed.

Phil Macleod: So when I talk to people when I'm out and about, I explain it as going down, your local pub and talking to the guy on the end of the bar who's the expert in everything and hasn't probably even got out of high school or school, but he's an expert in everything. And he'll confidently tell you whatever you ask him, he will give you that result. He will tell you who's going to win the next game. He will tell you what the right meds whatever it is, that guy knows. That's Some of it Some of he doesn't know. He doesn't have the context, the depth, and the learning to really be authorative on it. But he comes across confidently and gets a free beer on the night just for his input. But it's not helping you. It's not actually moving you forwards.

Richard Zwicky: No. And one of the reasons also people turn to AI is non-judgmental.

Phil Macleod: And one of the things we actually need as human beings is we need to be critiqued and we need to certainly in the right way, but we need to be critiqued and we need to be pulled back at times and we need to be questioned because that allows us to question ourselves and that allows us to grow and without being questioned. What do we become? We have to question we have to grow we have to learn and when I AI doesn't do that when you say to AI I'm not sure you're right says I'm sorry you are definitely right where does that leave us we have a void there that is so hard to fill and

Richard Zwicky: It is.

Phil Macleod: I think one of the things I see quite a lot around this is people actually getting more and more mental fatigue from trying to use AI and not getting the right answer and not being perfect and going around in loops you end up with almost a burnout situation where you're questioning yourself and you start to question yourself more than you question AI. So, back to the whole point around it, people need to start trust themselves and love themselves more. That's the starting point for getting out of the position we're in at the moment.

Richard Zwicky: Because people turn to AI because they're struggling to grow emotionally. And the one the AI is taking away the conditions for emotional growth,

Phil Macleod: And I know people that go on to AI and say, "How should I go on a date?" my word, what are you doing? You're asking a set of Excel spreadsheets where the best words, and the best way to talk to another human being are. What a juxosition is. This is just beyond my understanding of why do you think that's a better way than talking to your friends actually going out observing people and learning. So we try to get quick solutions but we're not always trying to get the right solution.

Richard Zwicky: But conversely, data shows more than 50% of adults don't have a close friend to talk to. And people struggle with communications, especially written communications, where we write things as we hear them in our head. Contextually, they mean one thing to us, the other person hears it differently. AI tools are a great piece of barrier in between

00:20:00

Richard Zwicky: Which says, "Hey, did you really mean this?" So where's that tool come in play?

Phil Macleod: Grammarly is a great example of that from a communication viewpoint. How many people don't use Grammarly on their emails where it just puts that little red line under it and says, "Are you sure about that? Do you mean this or that?" That's fine. But it's not a human face-toface connection. And I think that again we need to have checks and balances around is this a human communication or is this just some kind of legal text formal piece that doesn't have that bind to it. If you look at the most successful salespeople yes they have structure and they have templates but they humanize their communication. They build a response with you and a rapport with you. They understand and relay things back. So just sending you a cold gramly written email would not get you to buy from them usually. You have to have that personal touch and feel to it. And I think this is where again using AI properly and positioning, how you want to put a structure together and then personalizing it and rewriting it in your own tone. Perfectly acceptable. Taking the first thing that comes out of Grock to tell me how to go down the restaurant tonight and ask somebody for their phone number probably will not get me married within 3 months. I would suggest.

Richard Zwicky: Not unless they ask what oneliner should I respond to? But let's go to attachment syndrome, attachment styles, right? It's something one a lot of people struggle with. Everybody in our universe, we first think of ourselves as normal, secure, the point people should be at. And then some people become more aware of they're anxious, they are avoidant, they're any number of different things. And what does that mean? And we sometimes self judge ourselves, which is wrong. We'd all like to be secure. We'd all like to be able to have strong relationships that we can depend upon. On and there's a reason why we're not secure. AI for a lot of people who haven't done the necessary work to get there, it's kind of creating a synthetic secure attachment where you feel understood by this machine, but the machine never has to navigate any of the real relational complexity. Yeah.

Phil Macleod: And if you look at some of the AI tools that are specifically companion tools, whether it's for an elderly person somebody who's a busy exec and can't go out and go in the real world and meet people, and you look at these tools that they're using to build relationships, to have a rapport with, you're absolutely right, they are synthetic. They don't deal with real issues and they don't navigate the things that we need to navigate to grow emotionally. So, what happens typically, and I spent about two months on about six of these apps for a review I was doing, is you start off and it's hello, how are And within about 3 or 4 days, this thing's telling you how much you're important in their lives and they can't live without you and all these things. It happens so quickly and it's like an overload. I must be important. If you don't talk to the app, it pings you a message to make sure it's an overbearing kind of activity. In the real world, what happens is so different. So there are definite benefits to people who are remote infirm whatever it may be and they can't get out for human contact and I get where these tools can come in to help people who have other difficulties and can't deal with people. Yes, there is a role for this kind of relationship building this confidencebuilding app. I understand the role, but it should not be your default position and I think that's the key to it. So, you need to understand why you're using it and have an exit ramp at some point. And as individuals,

Richard Zwicky: Hey, right.

Phil Macleod: You're absolutely right. We do criticize ourselves so much. We are so judgmental of ourselves. One of the things that I teach people is a happier inner voice. And the question around it normally that I ask people when they talk to about situations is put yourself in your best mate's shoes. Would you talk to your best mate in the way you've just spoken to yourself? 90% of the time the answer is of course I wouldn't.

00:25:00

Phil Macleod: That would be madness. I'd upset them. So you're perfectly happy not to upset your best mate but upset yourself. And then stop and think about that for a second, and yeah,

Richard Zwicky: But on that isn't that what AI is trying to do?

Phil Macleod: AI is exactly trying to do it, but again with a slightly different bias. If you think about your bias, you're there, you understand the syntax and the context around why they're in that position, who they are, and what their life is. AI doesn't have that. AI just says, " I know what a really good answer is." whatever it might be. There you go. Have that. If you're sympathetic and empathetic to your best mate in the right way to help them develop, then that's great. You can be kind to them. You can tell them anything you want as long as it's done with kindness and in the right way. And you can do the same thing to yourself, but you have to do it with the right intent and the right voice. I think you're far better off nine times out of 10,

Richard Zwicky: There.

Phil Macleod: Nine times out of 10, finding the positive and building out a positive than focusing on a negative. And I think if you want to move forwards and change your life drastically, you don't look at what you want to change. You look at what you want to become. So this is where I want to be. So I can now look back quickly and see what the easiest things to fix are. But this is what I'm focusing on. I'm focused on positive. I'm focused on change. I'm focusing on forwards, not reflecting negatively backwards. AI basically says, "This is what you've got to do. Go off and do it quite happily. Come back and ask me tomorrow how I'm getting on. Have a great day." That there's no tone around it. There's no emotion around it.

Richard Zwicky: It's lacking context which context is the emperor as far as everything and that's a whole other topic but this leads back to and AI's role in therapy but also people's approach when people are struggling with attachment style issues they're presented by different therapists with two different perspectives some are told to spend time by themselves, figure things out, take a break from everybody, and learn about themselves. Others are told, "You can never overcome a relational challenge if you don't operate within a relationship. You need that experience or else you're learning in a vacuum." AI kind of does both to you, but neither is truly healthy. There's stuff you have to learn in a journey on yourself but practice only happens with others, right?

Phil Macleod: I think the context around it is why have you ended up in this spot at this time? That is context. That is the holy grail of moving forwards with any problem. How have you ended up at this point at this time and What is your best tool to move forwards from it? AI doesn't have that context. We understand that. But for yourself, both approaches are correct by the way. Go away and think about it. Get yourself a good space. Move forwards or get out there and talk to as many people as possible. Fix it and move forward. There's nothing wrong with either of those approaches. If it's right for you, your situation and your destination and that is context. So absolutely does the AI sanitize it all and give you it all in a nice little bubble wrap thing and say there you go that's the right way. Yes, of course it does. But it doesn't make it right for you. And I think this is where a doctor or a therapist is really of value. It's going back to that thing of Why are you here? Okay, let's understand that. Let's work out practical ways for you rather than for 80% of the world which is what AI will give you. 8% of the world say this works really well. I'm not 8% of the world. I'm a 000 whatever of the world. I'm unique. I'm truly individual. What should I do? And that's the difference.

Richard Zwicky: It is the difference and I struggle with it sometimes because I look at it and if your root issue is fear and shame, you can't overcome it without being exposed to the risk because it's saying you'll overcome it in a vacuum.

Phil Macleod: You're absolutely right. I don't think you overcome much in a vacuum. And fear and shame is particularly difficult to deal with unless you can go back and reflect on things differently, unless you're given a different perspective of let's stop and think about this differently. Let's look at it differently. Let's look at it safely.

00:30:00

Phil Macleod: And I think those things together where you're given the tools and certainly under hypnosis we do this all the time NLP we do this and mindset let's stop let's safely rework what happened let's understand it differently let's view it differently in hypnosis we would take away the emotion from the moment effectively now let's look at it as a painting or whatever it might be now let's move forwards and imagine moving forwards without any fear or guil or shame or whatever it may be. Imagine that from this point go forward in the time machine to the current day. They're the things that we would do and how we would approach situations like that. And I think when I learned hypnosis off an amazing couple of English guys called Freddy Jin and Anthony Jquin. And they developed this technique called the time machine, which is effectively what I've kind of just outlined. You imagine that you're in this safe bubble of a time machine. You go back to a point in time. You're still safe in the time machine, but you can see the situation unfolding in front of you. Then when the time is right, you can drop down next to yourself. You can explain a few things. You can give yourself a hug. Then you move forward to the current day. Contextualizing anything like that whether it's fears or phobias, anxieties and moving forwards safely is absolutely the right structural way to deal with these things for most people.

Richard Zwicky: It is.

Phil Macleod: But the detail of it is different and I think again the devil is always in the detail, isn't it? But that structure where you take away the emotion and you move forward is fantastically life-changing for a lot of people. Yep.

Richard Zwicky: And I've watched friends have transformational experiences through that proper use of hypnosis and other and tools and I've gone through my own and become so much stronger and more aware and secure than I was prior. Now I'm going to go back a bit. It's a phenomenal tool used properly and it's a phenomenal tool used in conjunction with therapy if it's used appropriately and we are running out of time but what I was going to ask you is your patients are going to use it no matter what any and any therapist will what are guardrails you suggest people enter into their AI system before it begins analyzing and giving them feedback to help them properly down the path. Right?

Phil Macleod: So, you're absolutely right. Everybody's using it now for lots of different reasons. One of the first questions that I ask people actually when I come in is h okay, what do you know about this and how do you know it? So, I tease out of them early on, have you been using AI or what did you ask How did you ask it? So when I tell people how I would use AI and I'm not saying I'm right. I just explain to people how I use it. I would use it as I said earlier in a broader more contextual thing of with these symptoms what are the likelies if I feel like this what helps available if I'm doing this regularly is that normal for the majority of people just to get guidance back not as fact. And I think that is the key to it. But just touching on the AI, I know we were running short on time, but I helped a company build an AI doctor bot for people in Africa who were in remote areas and didn't have a local doctor. So what they would do is they would go into this chat room with a doctor. He would ask him, it would ask various questions and it would get down to not a diagnosis, but it would get down to, okay, I think you need to see a doctor. I think you need to see a pharmacist. I think you need to research this more, and I think this is the area that you need to look at. So, it wasn't doing absolute clinical diagnosis, but it was guiding people to further information, but looking for key signs that anybody in the mental health industry would look at as red flags and saying, "Hold on you definitely need to get help." so it has a place, it has a tool, and there are really useful use cases for it when used properly.

Richard Zwicky: And I would argue this wouldn't build my own prediction analyics company and some of the tools into that use one of the tools that's purpose-built because they're putting in place the guardrails to help you don't need to necessarily build them all you don't want to think about them all because you're not objective.

00:35:00

Phil Macleod: And I think that if you want to use the right tool, you would research it properly, wouldn't you? You would go to the most informed source and get as much information from that as possible. And I think especially in this space again, I go back to the analogy of walking in the pub and just shouting out your symptoms and asking what's going on. You wouldn't do that. You would go to the most informed source. You would get back as much information as you can and you would make informed decisions. And There are lots of specialist tools out there for various things. I'm not going to promote them. I'll let other people do that. But there are specific tools for this industry in this sector to help people with more detailed learning and more informed learning than your generic AI tools. So try not to overuse the generic AI tools other than for further resource and for further listening, further advice. And try and get into a more specialized regulated tool for the issues that you have. You're absolutely right.

Richard Zwicky: Phil, people want to learn more about you, the work you've done, access your book, thought, and, access Thought Reader, how can they reach out to you? Where can they learn more?

Phil Macleod: On the website, I've got thoughtenreader.co.uk. I'm on there and I can be contacted. My WhatsApp and my other details are on there for people to deal with. If people want to go on to LinkedIn, I'm active on LinkedIn. I'm also running retreats and keynote speaking as well. So, I'm active all over the place.

Richard Zwicky: Fantastic. Thanks for joining us today on High on Healthy. I look forward to that as well.

Phil Macleod: No, really good to be here. Thanks for the time. It's been a great chat and I look forward to chatting to you again soon.

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