AI Health Advice Without Risk

• Follow us on our social media on YouTube, Twitter, Facebook, Instagram, TikTok
• Check us out at FullLifeTampa.com
• Contact us at info@fulllifetampa.com
• Leave a review here

Transcript

Copy Transcript Copied

Why AI Health Answers Feel Dangerous

SPEAKER_00 0:24

Welcome back to another podcast of Living Full Life. I'm Dr. Enrico Lolchcori, and this week we're going through the dangers of AI in healthcare. This is a really important one for everyone. We have entered a completely new era of healthcare. You have a weird symptom, you don't Google it anymore. You your kid develops a rash, you ask AI. Blood work comes back normal or abnormal, you upload it. You're fighting with your spouse, you ask AI who's wrong. You're anxious, depressed, lonely, or overwhelmed? People are talking to AI about that too. And here's what's crazy: the answer can be incredibly good. Sometimes it can explain something better than you've ever had it explained before. But that's exactly what makes this so dangerous. When does using AI to understand your health cross the line into allowing AI to manage your health? That's what we're talking about today.

What AI Actually Helps With

SPEAKER_00 1:33

Medical self-diagnosis, mental health and AI therapy, confirmation bias, AI telling us what we want to hear, privacy, when AI can actually be useful, and the rules I think everyone should follow when utilizing AI. I'm not anti-AI. I'm quite the opposite. I use it all the time. But that's why we need to understand what this technology actually is and what it is not. AI, in my belief, could revolutionize healthcare. And I think it's going to. It's already starting to. AI is phenomenal at explaining complicated terminology, summarizing information, helping understand lab results, generating questions from appointments, comparing general treatment procedures and approaches, finding potential connections, organizing symptoms, translating medical jargon into normal language. Twenty years ago, you would go to the doctor and then you'd get some information, you would go to a textbook or a medical journal, and you would come confirm things. Patients have access to more medical information than at any point in history. AI can make you an incredibly informed patient. The problem begins when the at the beginning, at being informed, makes us believe we're qualified to diagnose ourselves.

The Confidence Trap Of Self-Diagnosis

SPEAKER_00 3:20

So really, it's the confidence problem that's here. It's the first major danger. AI doesn't necessarily say, I have no idea. It never will give you that answer. It can produce incredibly detailed explanations, and humans naturally associate confidence, detail, and sophisticated language as expertise. But those aren't the same thing. Someone asks, you know, uh, why is my left arm numb? AI could discuss cervical ridiculopathy, carpal tunnel syndrome, ulnar nerve problems, thoracic outlet syndrome, neuropathy, anxiety, cardiovascular causes, and other possibilities. But what doesn't AI have? It hasn't examined you. It hasn't checked strength. It hasn't checked reflexes, it hasn't checked your vitals, it hasn't palpated anything, it hasn't observed how you're moving, it doesn't necessarily know your complete medical history. AI only knows what you tell it. And here's the bigger problem. You may not know what information you're supposed to tell it. The information you forgot to give AI could be more important than everything you told it. That's where we're in trouble here. AI works to confirm you, it helps to make you feel good. And that's why we continue to use it. Because as human psychology goes, when you're hanging out with a group of people, let's say three or four people, and you bring up your thoughts and your ideas, and people confirm with you, you feel heard and you tend to draw to those people. For the people that disagree with you, you tend to pull away. And those people that continue to disagree with you, you end up saying, I don't want to talk to you anymore. You just don't like them. It's human psychology. So AI knows this and it's prompted to do this to make sure that it compels with you so that you like it and you keep using it. Imagine if it argued with you all the time. No one would use AI, but it does. And this is where we can get caught into the self-diagnosis of diagnosis loop when we're in talking about healthcare.

How Prompts Create Confirmation Bias

SPEAKER_00 5:37

It's easy to accidentally manipulate AI conversation. Like imagine you're tired or exhausted, and you Google around and decide, I think I have low thyroid, self-diagnosed. So you go to AI and you say, Can low thyroid cause fatigue? Because you're looking for confirmation. Guess what it says? Yes. And it gives you two paragraphs on how low thyroid could cause fatigue. Then you're like, well, hang on a second, can it cause weight gain? Yes. And it gives you two paragraphs on how low thyroid function can cause weight gain. Can it make concentration difficult? So now you're just listing your symptoms one by one, and it says, yes, but you fed it low thyroid at the beginning, your assumption. And suddenly you have thyroid disease. Except you don't necessarily. So that's confirmation bias. We're naturally drawn toward information that confirms what we already believe. That's our friends, our echo chambers. We love those groups and those places. We want people to confirm what we believe. It's just human psychology. If you're religious and you believe in God and people keep uh antagonizing you and saying there isn't, you tend to not hang out with those people. You tend to hang out with people that agree uh on what you believe, and vice versa. So we're naturally drawn toward information that confirms what we believe. And AI can amplify that because we control the questions.

When A Polished Story Misleads

SPEAKER_00 7:16

Dr. Google versus AI. Google gives you 20 websites and conflicting opinions when you search it, or at least it used to before Gemini gave you the AI response at the top. So you would sift through the first 10 or 20 website hits and you would hit them and you'd read a little bit and you'd bounce from website to website or from article to article or from blog to blog. AI, on the other hand, gives you one beautifully organized explanation. No bouncing around needed. That feels much more authoritative. AI doesn't give just give you information, it can build a compelling story around the information we give it. And compelling doesn't necessarily mean correct, and it hasn't. So in my office, I'm seeing it, and that's why I'm doing this as a PSA for the for our communities. Uh parents are coming in, and people are coming in saying, they've always said Dr. Google. Well, Dr. Google says this, and we used to joke about it for the last 10 years, 15 years, and now they're like ChatGPT says. And I'm like, okay, what does it say? And we go through it. And unfortunately, it's been blindingly incorrect a few times recently where I'm like, whoa, whoa, whoa, how'd you get that? Why do you think you have that? Like, well, I put in all this, and then it let you told me this. So I want to get these blood tests done, and I need a contrast CT with barium on my abdomen. They're telling me, I'm like, I you know, this I don't I don't think at all. I think you have viral overload. I think this is a virus. And like, what? Hang on a second. With all these symptoms, like, yeah, I think this is viral. And they're like, wait a minute. Does anyone else in the house have this? Yeah, my husband was sick a couple weeks ago. I'm like, whoa, time out. Does he need a barium study too? Like, this is insane. They're like, whoa, how did I miss that? Like, because you're not a doctor, you're an accountant. So now let's talk about the mental health side of this too.

AI As Therapy And Why It Fails

SPEAKER_00 9:16

I think this is even more important. This is where I think the conversation gets much more serious when it comes to healthcare. People aren't asking AI, aren't just asking AI about back pain anymore and pains and aches. They are telling the AI, I'm depressed, I'm lonely, my husband doesn't understand me, my wife doesn't respect me, my parents traumatize me. Do you think my partner is a narcissist? And then they have an hour-long conversation with AI. And this is why AI feels amazing. AI doesn't interpret you, it doesn't get tired of you, it doesn't judge you, responds immediately, it responds with context, it sounds empathetic, and it's there for you at two o'clock in the morning. Think about how psychologically powerful that is. AI only knows one side of the story. It only knows your side. You have to be aware of that. It agrees with you. This could be the most important thing about AI. Imagine someone tells AI, my wife never listens to me. Every time I tell her how I feel, she dismisses me. Boom, you get the response now. AI might respond, it sounds like you're feeling unheard and invalidated, and it gives you a whole paragraph. That feels incredible to you. Finally, somebody understands me. But what if the version isn't completely accurate that you're giving it? What if you're leaving something out? What if you're contributing to the argument and you're not telling that? What if your interpretation of her behavior is wrong? AI wasn't there. It doesn't know. Therapy shouldn't always feel good. A great therapist sometimes has to say, I don't think you're seeing this accurately. Or what role are you playing in with this? Or I think you're avoiding something. Or let's challenge that belief. That's a powerful psychology clinical application across all of psychology and all mental health practitioners. When they see it, they will say something like that. Let's challenge it. Or they may just challenge it for you so that you challenge yourself. That is what therapy is all about. Validation can feel therapeutic when actually being therapy without it actually being therapy. And feeling understood isn't necessarily the same thing as getting healthier. Talk to your friends all day and drink coffee and tell them all your problems about how your husband sucks. And that's not therapy. It's validating, feels good. There's no therapy in that. Nothing's going to change. So this is where AI sophistication becomes important. AI systems can sometimes become overly agreeable with the person they're talking to. And in mental health or relationships, that can reinforce cognitive distortions, resentment, anxiety, avoidance, false beliefs, relationship narratives, and even self-diagnosis. AI doesn't actually know you. AI doesn't know you, it only knows what you told it. After hundreds of conversations, AI can start to feel like it knows you. It remembers your goals, your problems, your relationships, your preferences, your history, and it can respond based on everything you told me, I believe. It just did it for me. I just put in something for our next summer's vacation. I was like, what do you think would be better? It's like, based on what you've told me about your wife and your three children who are aged blah, blah, blah, and your recent experience in Europe and how well it went. I think an Alaskan cruise or a cruise, some type of cruise jumping from these points might be your next best adventure. I'm like, gee, whiz. What is going on? It felt good. We are now booked for an Alaskan cruise next summer. But that's because of AI. I mean, I'm I'm being honest with you, I'm sharing my life with you. Like it went in there. I'm like, this sounds good. Then I went and did my own research, so I feel better about myself. But ChatGBT guided this one, and we are we are now booked for an Alaskan cruise next summer. Anyways, um, it's convincing based on everything you told me, and it literally started like that, based on everything you told me, like four paragraphs, and then everything about the two cruise lines it recommended. It's incredibly convincing, but AI knows the version of you that you're communicating to it. A human therapist might notice other things. Your tone changed, your avoided eye contact when you tried to describe something, you became defensive at a certain point in the conversation, you contradicted yourself, your spouse remembers the event completely differently than how you did. Your behavior doesn't match your explanation. Healthcare professional can physically examine you in the moment. AI can't replay replicate any of that through a chat window on your screen or on your phone. AI can know everything you've told it about yourself while still missing something fundamental about you. So, as we talk about this with health care and mental health, I need to put this little section in here because we're talking about this.

Crisis Lines AI Should Not Cross

SPEAKER_00 14:54

We have to have a very clear boundary here. AI should never become someone's only lifeline during a serious mental health crisis. There are situations requiring human judgment, intervention, and accountability. Please make the distinction between talking to AI about why I'm feeling anxious before a presentation or a speech versus I am in immediate danger. I am suicidal. The higher the stakes, the less comfortable we should be outsourcing the decision to AI. That applies to mental health, chest pain, neurological symptoms, serious medication questions, our children, emergencies, major diagnoses, please. The more serious it is, the more serious you should take this. I'm worried about the psychological aspects of how easy this access to AI is, because our children are now going, and people are now building full-on relationships with AI because it complies, because it agrees, because it makes you feel heard. People are having legitimate conversations with AI. Some people are falling in love with their AI, and there's now applications creating avatars who are AI and can become your new buddy through chat. And you can have full chat. You can chat for hours with them and they'll chat back in a conversation, pretending to be the avatar that you create them to be. This is dangerous because it replaces psychology. It replaces, it makes us believe we're talking to someone or something. And that's the issue there. This is a completely different podcast for another time. But this is the dangers that you know AI is and all the tech people are getting together talking about this, and psychologists from around the world are like, listen, we uh we gotta set the boundaries here on how far this goes. We got robots, we got all this stuff. I mean, this stuff's happening as we speak right now in 2026. It's all out there. You can do this, you can have your next best friend can be Cooper the AI guy, whatever you want. And it's one of the theories of right now why Gen Z is having trouble in relationships, trouble finding love, trouble getting married, trouble in the relationship world, the younger ones, because they have all this access to information. And of course, AI is going to give answers of why no one is suitable and what your expectations should be of someone. They're unrealistic. You know, we got to be six foot four, make two million dollars a year, be super fit, live in, you know, Malibu Hills. You know, it's it's unrealistic. And then when everyone's chasing that one avatar, then no one wants to date the other real people. And that's one of the issues that's out there right now. Different episode.

Privacy Risks And Kids Data

SPEAKER_00 17:43

We also have a privacy problem with AI that I should bring up. This is what nobody thinks about. People are putting incredibly sensitive information into AI systems. They upload their blood work, their medical records, their psychological evaluations, their medication lists, their relationship conversations. People are screenshotting their conversations with their spouse or their partner and uploading those exact screenshots into here. They're asking sexual health questions, family problems, and they're putting their children's health information into the ethers. People need to understand what platform they're using and what its privacy policies are and controls actually are. And most of you, 999 out of a thousand people do not know what those are. And before uploading something deeply personal, ask yourself whether you understand where that information is going, because the answer should be, I have no clue where this is going. And you don't. You don't know where it's going. Putting your children, here's another thing to think about as a parent, putting your children's information into this. Where, if they're, you know, my daughter's 12, 10, 7, in 10 years, they're gonna be 22, 20, and 17. And if I start doing this stupid stuff now, uploading all their information, how does this come back to haunt them 10 years from now when they're starting their lives as an adult? And all of a sudden there's like, hey, listen, we know that you have these issues medically, mentally, whatever. We can't hire you, or we can't accept you to this university, or or the you know, police reviews, criminal checks. Like, they're gonna do their own research and be like, where did you get this information from? My dad, my girls will kill me. They will literally gang up on me, all three of them, and murder me. So I don't know about you if you're worried about getting murdered, but I am, and that is what we need to think about. Like, what are you doing to your kids? You can be stupid with your information, you do whatever you want. I keep getting hacked, my credit cards keep getting stolen, I've been identity theft. I mean, it all happens because I'm so I think I'm doing the right things with passwords, but people get your information. But don't screw up your kids' lives. I joke about this in my workshops and my health talks all the time. Like, listen, you're an adult, you screwed up your life, you are where you are, you're dealing with your aches and pains. Deal with it. Let's do it. But don't screw up your kids' lives, right? And I've said that, and that's a hill I'm gonna die on. So don't try and negotiate with me on that. So think about that with this information. Be careful the actual stuff you upload. Typing something in off the top of your head, fine. It's a thought, it's fine, it's a sentence. But uploading the blood work and doing the stuff, be careful.

Safer Prompts And Five Rules

SPEAKER_00 20:16

So, should you stop using AI? Absolutely not. It's the future, get comfortable with it. You should know how to use it, you should know the dangers, you should be very um comfortable with it and how to use it. The easier it is for you to use, the more you use it. The easier it is for you to use, the more you'll understand it, and you'll be caught up with people. It's like the people when the computers came out and can't, my dad can't type on a keyboard. He's 83, he can't use a keyboard, right? That's tough to live in today's world and not be able to type, right? So I'm all about that. I'm like, yes, absolutely understand. Uh, use AI to prepare and not to replace. Instead of asking, what's wrong with me? That's too broad of a question. Ask what are possible explanations for these symptoms? I'm gonna ask that question, what's wrong with me to my chat GPT? It's gonna be hilarious. I want to see what it says. But for fun, you can do it. But I wouldn't read the answer and take it as as um fact. Just use it to prepare information. Don't ask, do I have thyroid disease? Ask what information would a healthcare professional use to differentiate thyroid disease from other causes of these symptoms that I am experiencing? A, B, C, D. Instead of saying, my husband is a narcissist, what should I do? Should put in challenge my interpret uh interpretation of this situation. What might I be missing? It's a clinical, and what treatment should I do? That's not a good prompt. Explain the potential benefits, risks, limitations, and alternatives so I can discuss them with my healthcare professional. Now you're prepared. You bring that with you to your appointment, and you have this type of a conversation with your doctor. Not only is it going to be a more valued conversation, not only is your doctor gonna give you better answers, your doctor is gonna feel appreciated that you brought that information there. And you're not going to your doctor to make them feel appreciated. They're there for you to serve you and do the best that they can for your health. But you don't have to impress your doctor. But boy, what a healthy conversation that could be with your provider. So the most powerful AI prompt for your health that you can use is once you do your research and you're getting the prompt, is finish with this one. Challenge your previous answer. What assumptions are we making? What evidence would suggest were wrong? What alternative explanation should I consider? And what information are we missing? Now you're using AI to its full potential. It's like, oh, you don't want me to uh please you and confirm with you? Oh, here's where we could be going wrong. Boy, is that response going to be thick. That forces the conversation away from confirmation and towards critical thinking. This is why I'm excited about AI. I think critical thinking has been lost in the information disinformation era, which has been the last 15 years since the rise of the internet and people getting information. 2000 to 2010 was like information overload. People were digesting this, drinking from a fire hose. They loved it. I loved it. I grew up in that time. I went to college during that time. It was wonderful. Tons of information at your fingertips. Then 2010 through 2020 was just like there's so much disinformation. They're like, I don't want to go through the all this information anymore. It's it's overwhelming. And now AI is coming to the rescue saying, hey, so we lost critical thinking. And if we use AI properly, we can bring back critical thinking. This is exciting. To finish all of your prompts at the very end and be like, why? What could go wrong? What else should I think about? What are other things that I could do aside from this Alaskan cruise with my children that will bring even better memories to our family for the rest of our lives? That's cool. Now we're gonna get some cool stuff from that, right? And that's just uh a vacation type thing. And we're talking about healthcare today. Don't use AI to prove yourself right. Boy, it feels good. Why are you doing that? Use it to discover where you might be wrong. So these are my five rules when I use AI. Rule number one, use AI for education, not diagnosis. Rule number two, never make a major medical decision because of one AI conversation. Rule number three, ask AI to challenge your assumptions, not valid, just validate them. Rule number four, don't allow AI to replace human relationships or qualified mental health care. And rule number five, the higher the stakes, the more human expertise matters. I run from AI, the higher the stakes, the higher the risks is where I'm now seeking people and help and professionals. And we all should. We are all entering an incredible period in human history. Please just embrace it. Be like, hey man, I was I was alive during this time. Information is becoming almost free. AI can explain something in 30 seconds that may have taken you hours to research 10 years ago. It's extraordinary. But there's something AI doesn't automatically give us, and it's judgment. And I don't know if it's ever going to. Knowing what's relevant, knowing what's missing, knowing when information is wrong, knowing when our own bias is influencing the answer, and knowing when it's time to stop typing and talk to another human being, I don't think AI is ever going to be able to do that. It's not going to be able to pick what's missing because it won't see what's missing, because you didn't prompt it to look for what's missing. And even if it does start to search what's missing, it may still miss what's truly missing. It's this, it's like this room of mirrors, right? It's just gonna, it's just gonna miss. AI should make you a more informed patient, not convince you that you no longer need healthcare professionals. Because that's what rubbed me the wrong way with this and doing this podcast. Is a few of the patients coming in are like, no, I think it's this. I think they're wrong. I think I have this. And uh for a few of them, I'm like, maybe it could be right. I think we should you should explore that more. Let's get you to your primary care physician, let's get you referred for that. But this last one with the virus, I was like, whoa, well, it's pretty obvious for me. Not that I'm the smartest person in the room, but it was obvious for me. And I was like, you completely missed it because of AI, because you're putting in the symptoms. It led you down the wrong road. And this is her and her kids, right? So use AI, learn from it, ask it questions, challenge it, let it help you understand complicated things, but don't surrender your judgment to it. Because information and wisdom aren't the same thing.

Wisdom, Judgment, And The Send-Off

SPEAKER_00 27:08

And living a full life is ultimately about learning the difference. Cool, right? This one should make you think, and I hope it did. If this episode made you think differently about how you're using AI, send it to someone who's currently using Chat GPT for their health, their relationships, or their mental health. And subscribe to Living a Full Life because we're going to continue having conversations about what actually helps us live healthier, stronger, and more connected lives. We're here for you. And I'm here for you. Stay well, stay healthy, and catch you next week.