ChatGPT Is Your Worst Flatmate For Health Advice

Episode 122: ChatGPT Is Your Worst Flatmate For Health Advice

AI can sound calm, clever, and certain, even when it’s guessing. That’s fine when you’re drafting an email, but it gets complicated fast when people start using ChatGPT and other tools to interpret symptoms, gut health results, and lab reports without support.

We’re joined again by Josh Netherwood from Your Gut Map to unpack what’s really happening behind the screen. We trace how AI evolved from early ideas about machine intelligence to today’s deep learning models, then bring it back to the real world: inconsistent answers, “personalisation” that drags in the wrong context, and advice that often jumps straight to worst case scenarios. If you’ve ever asked a health question twice and received two different paths, we explain why that happens and why it can lead to confusion and fear.

From H. pylori scare stories to the unintended consequences of self led treatment, we talk about how AI can push people towards a scattergun approach: overusing antibiotics, relying on untargeted probiotics, or trying extreme interventions without understanding the full picture. We also dig into the big issues most people miss: governance, accountability, and source quality. If an AI tool pulls from a mix of social posts, articles, and outdated research, how do you know what to trust?

We finish with practical, listener friendly prompts to ask “one more question”, request peer reviewed evidence, and get more up to date, scientifically led guidance. If you care about safe AI in healthcare, gut health, functional medicine, and personalised wellness, this is your roadmap. Subscribe, share with a friend, and leave us a review: what’s the most worrying piece of AI health advice you’ve seen?

Click HERE to support us.

Thank you for listening.

You can continue the conversation with us in the Far 2 Fabulous Facebook group. Come and connect with other women on a journey to empowered health. 

For more information about Julie Clark Nutrition, click HERE
For more information about Catherine Chapman, click HERE
 

If you enjoy the podcast and would like to support the show, you can buy us a coffee HERE. Your support helps us continue creating episodes and improving our recording setup. 

We look forward to you joining us on the next episode.

Transcript

SPEAKER_02 0:01

Welcome to Far Too Fabulous. Judy and Catherine. Join us on a mission to embrace your fabulousness and redefine wellness. Get ready for some feistiness, inspiration, and chat, and humour as we journey together towards empowered wellbeing.

SPEAKER_03 0:18

Let's dive in. Hello and welcome to this week's episode of the Far Too Fabulous podcast.

Welcome And Why AI Matters

SPEAKER_03 0:24

Now we are very lucky that we have got a previous guest come back on. Not only have they come back on, they asked to come back on. We didn't scare him off, and he wants to come back again. Exactly. So we're really, really, really happy to have Josh uh Netherwood from Your Gut Map on again. And we are talking about artificial intelligence. We can't escape AI, can we? No, it's it's everywhere.

SPEAKER_02 0:50

It's especially with we were talking about in business, it's it is absolutely everywhere.

SPEAKER_03 0:55

Yeah, and in my profession, we've been potentially told that we could be um made redundant through AI. And we will find out when we talk to Josh as to whether that's the case.

SPEAKER_00 1:08

No, thank you so much for having me back. No, you didn't scare me off. Um I was so interested in so many topics that you've already covered, and it aligns so well with all of the gut health-related topics that we've been um presenting to our audience as well. I thought, why not bring in such an interesting and controversial topic of AI? It's just so out there at the moment and it's hard to get away from. So, yeah, I was quite keen to, you know, talk about how this resonates into a healthcare setting.

SPEAKER_03 1:41

It's not going away, that's the thing, and it is going to become the norm, even if you're not somebody that's that's using AI, because we were thinking about our age demographic and whether people are really, I mean, we use it in our business, don't we? And it is, I think it's coming becoming more and more common to use it, but there is some danger in that as well, because it's very easy to trust this thing that's coming at you, and it seems to tell you the right things and knows you. So I think it's uh a really important subject. You tell yours off, don't you? I do tell mine off because it gets stuff wrong and it annoys me. And I will tell it. I you know, I use the voice uh memo option on mine, and I will say, um, I think you'll find that you've totally messed that up, and maybe you want to have a little rethink and come back with another answer. That's what I'll say to it.

SPEAKER_02 2:32

Do you think your chat GPT pack picks up on your passive aggression there? Yeah, quite possibly.

SPEAKER_00 2:40

It is one of those things as well, where AI integration, chat GPT is just so readily available now as well. You know, it's available through podcasts, it's available through content, images, and even your interpretation of your health. And that's something that in my industry really scares me because obviously I'm in the

AI In Healthcare And Diagnostic Fear

SPEAKER_00 3:03

field of diagnostics. I supply, you know, practitioners like yourself with biological data to be interpreted by you. But also, there's so many people out there within this industry that just quite frankly kick the clients to the curb and throw them into the wild and just really expect them to understand complex information, whether it's bacteria, whether it's DNA, whether it's blood sample analysis. And one of the things I'm so interested in is how is that interpreted through AI? And how is it governed? And, you know, where is the accountability of information? And that's all about the inputs and the outputs, which I can obviously detail more, but there's so many examples of dangerous interpretation because we've got to consider that AI is not a medical professional, it is very much determined by what information it's fed with. So I always say your AI is only ever as good as what you feed it. You get out what you put in. And I don't I don't know, Julia, you know, whether you've had clients come in and say, I got these results, AI told me this. And then you're you know, scratching your head thinking, where did you get that information from?

SPEAKER_03 4:22

Yeah, and and so and often it's wrong as well. Uh AI has told them to do something that could potentially be damaging, which is always concerning.

SPEAKER_00 4:31

100% is. And you know, I think people have come to trust AI more than they maybe should do. And I think one great place to start is kind of like where did AI come from? People all of a sudden, you know, within the Gen Z um era may not even know where it came from, but they would always have had it on their phone. It's like an integral part of their app library. And you know, I've been I've been doing quite a lot of research around AI, and you you wouldn't

Where AI Began And How It Works

SPEAKER_00 5:03

kind of think that artificial intelligence actually was pre-1950s, kind of like World War II eras, where technology wasn't even that much of a contributor in in the war, but yet it started there. And I think the whole principle initially was can machines actually think? And can we actually evaluate machine intelligence? And it was done by um a guy called Alan Turing, who basically did the Turing testing, which was basically looking at how we can create AI to provide convincing human interactions, but at that stage there was absolutely no learning because we kind of know now that you know, as we run through the principle of AI that it's learning on the job, and that really comes with kind of like the birth of artificial intelligence, which was in the 60s, which was basically looking at how you could hard code instructions within the AI to basically provide rule-based systems. So it's kind of like a this particular algorithm would fit this particular answer out based on what you asked it. So that was a very like singular approach. But then as we kind of moved into the 2000s, um, I always think of kind of like Terminator, like the rise of the machines, and you know, like you said, are you redundant in a sense? Is it a job's redundance that that can be done very easily through AI? And you know, it is absolutely going that way. But like I say, in the early 2000s, it was looking at how machines could learn through algorithms, and they called it decision trees. So basically looking at more of a statistical approach to pattern recognition around your voice, around your speech, around your, I suppose, the image integration as well. So, can we upload a set of biological results in order to actually spit out some recommendations? And that was where deep learning came in within 2010. And that's where it kind of began to create kind of like the brain. You'll you'll think of like a neural network, and the machine brain started to look at, like you say, speech recognition. Julie, you talk to your AI, it took in that information. It might, like I said, look at the image recognition of a set of results and from there develop such a an ability to take in such a huge amount of data sets. So that could be bioinformatic databases, it could be information about previous results uploaded into an AI to offer some form of comparison, but then also kind of like how do we know what data sets they are? We we don't, because kind of like the unique pattern of AI is that it's specifically theirs. So there's more than one artificial intelligence, there's more than one app, and they don't all use the same algorithms, they don't all use the same information or data sets, and that's where in healthcare it can get so scary because that level of information basically becomes inconsistent. And I don't know about yourselves, guys, but uh have you ever asked AI something twice of the same question and it came out with something completely different?

SPEAKER_02 8:29

I don't think I have ever tried it. When you said you were talking about that, I'm just thinking about how like how muddy the waters are with everything. When I ask it something about me or about my daughter, as opposed to when I'm creating content for my clients, it then answers me as if I'm talking about my clients.

SPEAKER_00 8:50

Yes.

SPEAKER_02 8:50

Like you've got to be really, really specific about what you want, but it assumes that because I because I talk about my clients a lot on my AI, that that's what I'm always talking about, and that's how it answers me. My client avatar is called Sally. I asked it something about Anya going to going on holiday or something like that, or writing her CV, and it says, Yes, this Sally would really like this. And you're like, No, I'm not talking about her.

SPEAKER_00 9:14

Oh no. So it's like that recognition of previous requests as well, isn't it? Yeah, and that's worrying in a sense.

SPEAKER_02 9:21

Does that

Personalisation Errors And Mixed Context

SPEAKER_02 9:22

ever happen with you? If if you're asking it a question about something, does it ever then draw back from information that you've spoken about like different clients with?

SPEAKER_03 9:33

Yeah, yeah, it does it's done that before. If I've asked something, it will say, because you've got Hashimoto's, you might want to consider such and such. I'm like, what, this person hasn't got Hashimoto's. Where's that come from? And then I realised it's because in a previous conversation I was doing something on Hashimoto's and it brings it in. Yeah, exactly.

SPEAKER_00 9:51

And like you say, that that's the scary principle of personalization. When you're trying to personalize something with previous data, then it's more work for you when it's supposed to be a time-saving element as well. And that's the other thing about business is that we're here, you know, formulating plans for different clients with different needs and different requirements. And AI is great, you know, it gives you that time-saving element. But I always talk about it like a skeleton model. It does not know medication, it does not know past history. Maybe you can upload pre-screen questionnaires potentially through that image recognition. But without that and the absence of it, they're really blindly diagnosing the problem, which can be so dangerous, and especially when in the present-day AI, you know, even you know, I I am proudly a part of an AI-integrated healthcare business. And it's all about how that's done safely as well, because when it comes to the integration into different platforms, so for Google, you've got Gemini. Gemini is an AI-generated email content provider and it can summarize this, that, and the other. So you've got to, you know, be able to, as a practitioner as well, bring your own tone of voice. Because again, AI is not empathetic. It's very, very straight to the point. It really doesn't have that empathetic nature. So it's kind of like losing that that personal touch, I think, as well. So I always think, you know, in practice, we should always be very careful of how it is integrated and how that's then disseminated to a client. And I think that's really important for trust and buy-in from from a client. Because I know for a fact that people are getting a lot more, you know, aware of how AI talks. And I don't know if you've seen grammar or punctuality, but if someone ever emailed me from AI, I could tell it within uh five seconds. Because there's a real pattern of repeating the question. That's the first thing. And there's also the use of grammar. There's always these elongated dashes.

SPEAKER_02 12:02

Dashes, yeah. What's that all about?

SPEAKER_00 12:05

And I always say to anyone who's seeking, you know, support through the healthcare system or you know, whether that's privately or publicly, do not accept answers on any communication that look AI generated because it's not personable and that you should accept within your your healthcare regime and and protocols. And it's all about helping each other, isn't it? Like we're trying to help people. And if we don't have that personal touch, then what what do we have? That's literally what our job entails.

SPEAKER_02 12:37

Yeah, exactly. And I think that's gonna be the thing that makes the difference eventually is like when when people talk about is AI gonna take over our jobs, is the is the fact that we make it personal, we make it exactly about that person. And when we're making a a suggestion or a diagnosis or whatever, we're making it with that person literally right in front of us, rather than if someone's googling or AIing their whatever's going on, and perhaps you know, they could well have Googled, just they've listened to Julie and they've gone, what is Hashimoto's? And they've they've Googled that, and then later on they've asked about another symptom, and AI is now dragging that into it as well. They're gonna get the most weird and wonderful mix of things, yeah.

SPEAKER_00 13:29

It's causing confusion. I'll give you a really good example of I suppose challenging professionals through AI. So, one of the things that I would say when you're looking at a particular diagnosis, so I'll I'll use H. pylori as an example. You know, we've got a lot of people who, you know, maybe suffering with high levels of H. pylori. And that

The Skeleton Model And Lost Empathy

SPEAKER_00 13:56

on the face of it, if you had a high level of that particular bacteria in your stomach and you asked that to chat GPT, it would basically tell you that you're at an increased risk of stomach cancers. And how scary to receive a set of results with something so drastic and something so scary that you're then trying to, you know, bring that person, you know, down. You're you're trying to then talk to them about you know how we can intervene before something like this happens because we're in this proactive nature, that that's what we do. Everything that's coming out of AI is pretty much worst-case scenario. And that's kind of like the scaremongering tactic of AI. And when we kind of think of you know the the GP roots as well, you know, we'll call it self-led AI consequences. So basically, if you had, let's just say, a result that came back as a high level of H. Pylori, that could be going from yourself, Julie, to then someone going to a GP because they're worried. They've got this cancer word that's just come into their vocabulary around their health, and all of a sudden they think reactively and they think about a very clinical problem. So that could mean that they have to go to the GP. The GP then might not even accept the results because they utilize different technology, but not to go too far into it, but we utilize more advanced methodology called shotgun sequencing, whereas GPs tend to use PCR as one of their methods of bacterial analysis. So because we're not CQC registered as a complementary medicine association, then they're not going to accept our results. So instantly we've got a contradiction, and instantly we've got further diagnostics for another round of testing, which ultimately means that that person's then stuck in a journey. They're stuck in this really long journey of having to go back to the GP, do more testing, receive the results, potentially have different levels because time has changed as well, because results change over time. And then we've got different grades of sensitivity with different grades of testing as well. So that can really cause a lot of confusion with that self-led AI kind of input. But then when we look at the the element of like let's just say, you know, you yourself, Julie, you've got a set of results for H. pylori, you've told that individual, and then they've gone away and thought, you know what, I'm just going to deal with it myself. And people do, people do do that. You know, people might take results um at the absence of any practitioner support whatsoever. But a lot of the times it can really lead to a lot of ineffective treatments because ultimately you can't get rid of things like H. pylori with just a dietary intervention. Because with that in mind, you might require antibiotics, but then is it a misuse of antibiotics because you've not integrated certain interventions to support the rebalancing of the microbiome

H Pylori Scares And Self Led Spirals

SPEAKER_00 16:57

to essentially reduce that load onto the body. And all the while, you know, this infection persists, potentially gets worse, and ultimately what we're then promoting is more antibiotic consumption, all because we've been led down this path, and then it again puts the pressure on the GPs because they're thinking, well, they want this test, and this particular results come up. And that is my protocol, it is what I want them to do. It's it's antibiotics, because that's a massive go-to. I don't know about you guys, but I see so many people on antibiotics now. It is such such a problem, I think.

SPEAKER_03 17:33

Yeah, I would agree. Massive problem.

SPEAKER_02 17:35

And it's interesting, isn't it? Because I mean there is a real awareness around overuse of antibiotics, and like sometimes when people genuinely need it, they have to fight tooth and nail for it. It's a crazy thing.

SPEAKER_00 17:48

Absolutely. And I mean, and the other element as well is you know, through self-led routes, is that if people are very open to holistic medicine and they aren't seeking the professional advice, they may ask for holistic options through an AI intelligence. And that can actually lead to things like the overuse of natural remedies. So we all know that you know, probiotics, antimicrobial, herbal medication is all something that you know is very popular with within this industry. The worst thing is the overuse of probiotics. Without targeted strain analysis, it's really causing you more problems than anything else. It's kind of like uh you overfeed something is a bad thing to a degree. Too much of a good thing is a bad thing.

SPEAKER_03 18:30

Yeah, we've spoken before about too much of a good thing, like you can drink too much water. Yes, if you drink too much water, it could actually kill you, right?

SPEAKER_01 18:40

Yeah, yeah.

SPEAKER_03 18:40

But we need water, it's essential. Most of us don't drink enough water, but you can drink too much water, so yeah, you can do too much good.

SPEAKER_02 18:48

Yeah. And this what you're talking about here, it's like it's almost like whack-a-mole, isn't it? Like if you're if they're just left to their own devices, are we gonna try and sort that out, but maybe that's gone wrong. We'll sort that out. Oh no, that's gone wrong, you know, or or trying to shoot a target blindfolded, it's just not gonna work.

SPEAKER_00 19:08

It's it's that whole scatter gun approach, though, isn't it? Yeah, and all that AI tends to do is create more holes, it brings out more paths, and and that can be well, you've got to do this, but actually you could do that. But if it's moderately a problem, you could do this. If it's a low level of uh intensity or an issue, you could do that. And if it's high level of intensity, then you need to do this most extreme version, or this most extreme consequence will happen. So all of a sudden, people's minds are just completely blown, and they're having this scattergun approach of if I do this, this, this, this, and this, I throw as much as I can at it, then surely something's gonna work. Something's gonna be like, Something's gonna stick. Yeah, exactly. And that's something that doesn't work in reality. And like I say, probiotics is a fantastic example. Um, but when we're really concentrating, I mean, kind of like going back to that H. pylori element, without appropriate medical guidance and without the right dosage for probiotics or herbal medication, whatever that might be, it can actually cause further complications. And we're talking things like ulcers, internal bleeding. And these are these are these are the negative consequences of the misuse and the misguidance. And I think we all just need to take AI for what it is. It's uh only a guideline. But if we can really look at where where is this information coming from? Because I know for a fact that something I do personally is I don't I don't pick up anything without looking at the ingredients list. And, you know, I was actually doing a health conference in Leicester and I just picked up a flapjack. I do like a homemade flapjack, and I basically brought it up and I said, you know, who's eaten this today? And they said, Well, you know, oh, about four or five hands went up. And I was thinking, right, do you know the first ingredient here is palm oil? And everyone's like, no. And I was like, Do you know there's an emulsifier in here that's linked with cardiovascular disease? No. And it's it's really kind of that, you know, we need to know what we're putting into our bodies, but we also need to know what we're accepting externally in order to make that intervention. When it comes to inputs, that's something that I'm really, really interested in as a topic. I mean, what I don't know what your opinion is on this, but where where where do we think all this information is coming

Antibiotics Overuse And Remedy Misfires

SPEAKER_00 21:32

from from Chat GPT?

SPEAKER_02 21:33

It's that's a good quick good question. I mean, that's I was as you were talking, I'm thinking, how do we influence what people end up with on their screen? Because I'm assuming it is just a an absolute again, it's like a scattergun thing. It's like a mismatch of all of that information out there pulled in together. And like you said, you can ask the same question twice and get totally different answers.

SPEAKER_00 21:56

I mean, one thing that is really blatantly obvious is. Is that we have no control over what we're told. We ask the question and it gives us an answer. We think it's the best thing since sliced bread. It's the laziest approach known to man. But also we we automatically trust it. But should we? Especially when it comes to a healthcare setting. So one thing that I would say is that when we're looking at the input from AI, typically this is from we'll use primary and secondary resources as an example. So AI is basically pulling from anything online. Secondary resources could be things like articles, social media posts, newsletter articles. But then you've also got things like primary resources, which are journals

Sources Matter And Controlling Inputs

SPEAKER_00 22:48

or pilot studies or gold standard peer-reviewed research papers. The consistency is actually lacking. So we're pulling from all of these different variables that change the percentage outcome and the consistency of that. So that question that we asked yesterday, we don't know whether that came from a primary or secondary resource. Without specifically asking it, please provide me with a peer-reviewed research paper around this topic. It won't give it to you. But then we shouldn't really be thinking people think like that anyway, because not everyone's got a scientific hat on all the time. So a really important element of AI is knowing how to control the input. And that's something that you know is a part of my role within the diagnostic field and openly working with AI in a controllable manner. Now, Julie, you've seen many of our reports, and you can probably, you know, see how much interaction insights we get and you know how we can connect different um issues associated with different biological data. But it's kind of like thinking how they got to that. It's always really thinking about where the input has come from. So one thing that we were very conscious of as an organization was that we actually don't want a deep learning AI. We we want it to be a human-led AI. And what I mean by that is again, going back to that phrase, is that your AI is only ever as good as what you feed it. So I'm going to control what I put in my AI's mouth. So that's controlled AI, where you're restricting the learning, but you're feeding it with information that you want. Now, obviously, that opens up subjectivity. What one person values in one business might be different to what another person values in another business. And then again, you've got another area of inconsistency. But one thing as a researcher myself that we had as a part of ours was that we always make sure that anything that we're claiming is always a part of a double blind randomized control procedure, which is basically the highest form of research. It's something that a lot of practitioners hang their hats on in terms of the accuracy and the connection between your biological data and the outcomes of health and disease or metabolic dysfunctions. And that's something that we actually do monthly. So, like I am a part of the scientific board. We basically review new research. And what does that mean? It means that one month one thing could be associated with a disease. Next month it may not. Because do you know what? Research changes. So the other element is how far back does the primary data pulling come from with normal AI? Is it pulling from 1999 when we've got a study that's available in 2026, which actually disproved that? So that's all about how we can support the clients with the most up-to-date and relevant information. And that's that's dynamic. That that's dynamic and it's also learning, so that we can make sure that healthcare remains up to date, it remains relevant, and also it gives them the best chance of recovery.

SPEAKER_02 26:08

I think you can always you can almost get the answer that you're after as well, can't you? Depending on how you word it on a on Google or on AI or something, you could almost get the answer that you're after because it can find that research somewhere in the internet. And I think the danger with if somebody is looking for some sort of health solution, we know what humans are like, and uh, and we're no, we're no different. We want the path of least resistance, we want the easiest, the easiest thing, the easiest route. And so we're gonna look at this list of suggestions that it's given us, and we're gonna go, oh, look, that looks like the easiest one, and we're gonna go for that. Is that true?

SPEAKER_00 26:48

And I think as well, it's all about consistency as well. You know, people want consistency in their delivery, consistency in their protocol, and ultimately the confidence that what they're following is gonna help them. And I think that's really important to remember. So, one thing that I would absolutely love to change within, you know, I don't want people to be scared of AI. It sounds like we're scaremongering AI, but it's more it's more about how can we use AI safely? And, you know, this is something that's so important with educating people around the consequences of self-led recovery and professional-led recovery, because then that's going to make sure that you know yourself, you're you're part of bands, you know, you're you're regulated to a degree, you're expected to learn so much throughout the year, and that's something that you know people can trust in that you're up to date, and that's something that Chat GBT can't do. And where's the governance? Do we know who governs any of these AI?

SPEAKER_02 27:55

Not clear.

SPEAKER_00 27:56

No idea. So again, it's kind of like if there's no governance, then where is the the accountability? If somebody follows a set of rules and a set of advice around

Human Led AI And Keeping Research Current

SPEAKER_00 28:09

a set of results that they've uploaded, and then they go into worst case scenario, severe health issues, potential misuse of drugs, potential overdoses, who's accountable for that?

SPEAKER_02 28:23

It's a good question, actually. And I think it's one of the most um searched things, isn't it, on AI is about health and stuff. And I I believe that that they're creating a health-specific AI. But again, I don't know, like who's that? You don't know the um these people. Exactly. You don't know the motives of these people, you don't know it's yeah, it's very interesting.

SPEAKER_00 28:46

But I mean, other examples as well, because obviously we know the NHS, you know, we know that they're very reactive and they they tend to be about around 30 years behind what we're doing in in complementary medicine. Kind of like we know we know this, but it's it's used in the NHS, it's just not really spoken about. And, you know, one of the things that we see is uh paper trails. I think a lot of them are not using the same systems, by the way. They don't from one trust to the next, the systems aren't used. There's a lot of paper trails, but it's actually used quite often in things like medical imaging when it comes to things like x-rays, um, MRIs, and they can then utilize that kind of like image recognition to support a diagnostic because there are set algorithms and set images that would result in X, Y, and Z. So actually it's it's there, just people don't speak about it that often because it's new to us. I think the most advanced people don't think of the 1950s as when AI started. No one thinks of that. Everyone thinks AI came into fruition five, six years ago because these apps weren't necessarily there, but then we've just seen it being so much more accessible, and that's that's people's norm. But it's kind of been building in the background, then as it's been building, its applications have been used in very different ways. So I think we also need to accept that it's there, it's there already. People just might not necessarily know or understand its implications and its usage. So I think that acceptance is one thing, but accepting that the information that we get challenge its source, I would always say. Challenge its source and make sure that you're not making any healthcare interventions yourself without the right form of information. There's so many as well, AI generated diet planning. I mean, do you utilize uh diet planning software at all in your practices?

SPEAKER_03 30:55

Uh no, I d I don't because I've just I've been at this game for quite a long time. So I've got a lot of recipes and things anyway. So I just tend to do it, I just tend to do it myself to be honest.

SPEAKER_02 31:07

So is that something that people could like subscribe to and then it generates something for them?

SPEAKER_00 31:12

Yeah, so what you've just said there, uh Julie, is experience. Yes, you've experienced you've you've got the experience, you've got your own pattern recognitions with certain people's biological data and their results, whether it's microbiome, whether it's blood work, and you have your own, you know, recognition patterns that these recipes are going to work for you, these ones are gluten-free, these ones are dairy-free. So you're not leaning on artificial intelligence because experience is priceless. Okay. But we have to think of uh graduating students in this field. They're constantly being exposed to subscriptions for diet planning. Like you've just said there. You can subscribe. You can. And you and what does that do? It saves time for the practitioner, but at the absence of pure personalization. So that's why I always call it a skeleton model. And these diet planning subscriptions also rely on the, I suppose, the buy-in from businesses. You know, we've had personally at GutMap a lot of people asking us for our results and our algorithms as to how we populate our nutritional interventions, because they want to integrate that into their plans so that their subscribers can use our results so that they can then take advantage of their software. Now, where we are a little bit standoffish is that we actually have our own diet planning software, and that for us is consistency because we've got our results, we've got our diet planning software, and we can have full faith in what is being spat out. We can make sure that our results are being interpreted correctly in order to create the right diet plan. So, a bit of an example with ourselves, and I say, Julie, you've used these before, is we actually look at how food can support the depletion of these bad bacteria like H. pylori, whilst also finding foods rich in probiotics, prebiotics, um, fibers, um, vitamins and minerals in order to feed good bacteria. So our diet planning software is all geared around which particular bacteria are in a negative level or maybe in a state of abundance, maybe too much bacteria. We did talk about that in the last podcast that we did together about feeding and depleting. Um, but then also how we can support that in terms of prebiotic um prescriptions as well,

Governance Gaps And Accountability Risks

SPEAKER_00 33:50

because again, once we know someone's bacteria set, we can advise precision ingredients to include in your supplementation protocols. The worry is that if it starts going into third-party softwares, then is there a leverage from another supplement company? Is there an endorsement for a certain prior probiotic? It's it's out there, it's so out there. These affiliates exist.

SPEAKER_02 34:13

And I'm assuming that technology that you're talking about, that you guys have created, is for people like Julie. So it's kind of almost filtered through. So you add that personal bit in, and the danger is if that even if that information was then given out straight to the general public, it's still open to their own interpretation. Yeah.

SPEAKER_00 34:34

The the fear of cutting you out here, Jeff. If someone was to take our results as an absolute base minimum, they would know what to eat and what to avoid. And that's something that we purposely did in order to safeguard clients because, again, they're going away at the absence of nutritional intervention and just biological data, finding that self-led AI and then making their own assumptions. So, in actually integrating not just the analysis, the outcome in terms of disease, but also the intervention and protocol was something that we were really keen on doing uh when we actually launched this business. And then where practitioners like Julie come in is that you're taking that information and making it more personalized. Because again, I don't know anything about your client's medical history, I don't know anything about their medication, I don't know anything about their likes and dislikes. You know, we might be advising things that they really don't like. Um, you know, you've got um poor vitamin D absorption. Okay, go and have some more mushrooms. Me personally, I hate mushrooms. I would not eat, I would not eat them at all. So then that that that's where the personalization and that that care comes into play because you know, you can find other sources. I'm sure Julie could rack off about 10, 15 different foods that have vitamin D sources. Um, whereas our AI wouldn't wouldn't know that person's preference. And that's where that AI coupled with professional support is so, so

NHS Use Cases And Medical Imaging

SPEAKER_00 36:07

important. And I think it's become becoming a movement. It's like a partnership. I feel like I'm in partnership with an AI, it doesn't talk to me that much. Um, so you know it can be a lonely place working with AI. That's why that's why I speak to people like Julie.

SPEAKER_03 36:24

You need to turn off your AI and then you start a different type of relationship.

SPEAKER_00 36:28

Yeah, it goes on a tangent then. But yeah, like I said, I think it's a really important partnership that we need to establish because, again, we don't want to be left behind in this world of healthcare because technology is a part of our life, technology is a part of our everyday workings as human beings. But when we make sure there's that personal touch, I think that's where compliance happens because you can buy into an individual like a practitioner who is empathetic and passionate about your health. You don't get that from AI, you don't get the sympathetic, you you don't get that that checking. I mean, I've never asked a question um, you know, to AI, and it's got back to me three lump three months later and said, How are you doing, Josh?

SPEAKER_02 37:21

Yeah, exactly, exactly. Did you did you do what you said you were gonna do?

SPEAKER_00 37:25

Yeah, so there's this again, there's that real brush-off mentality with AI. It's I'll give you that information and then I'll just wait for the next question. There there isn't that personability, then there aren't any relationships with AI apart from it's like the the most selfish relationship.

SPEAKER_02 37:43

You're asking all the questions, so it's scary though, isn't it? Because people do, and especially the way that they talk to you sometimes, people do feel like they're forming a relationship with it, but you're

Diet Planning Software Versus Experience

SPEAKER_02 37:55

right, it's the most toxic relationship ever.

SPEAKER_00 37:57

Yeah, absolutely. And um, I think kind of like a bit of a concluding statement in all of this is you know, don't be scared of it, embrace it, but know where it's coming from. I think that's a really a good kind of round off to everything that we've spoken about today.

SPEAKER_02 38:14

No, I I agree, and I think I I guess that's that would be if somebody was going, because I mean I th I don't think we're gonna stop people from looking in in Chat GPT or other AIs for their for like what's going on with their with their bodies. And it's just making sure that you know where that information is coming from and just keep questioning it, keep questioning it, is this right? Is this right for me? Yeah, yeah, but ideally you get that information filtered through a hundred percental.

SPEAKER_00 38:42

And and if we can just take one thing for one person who doesn't want to see anyone, all we would ever say is just ask one more question. Just ask one more question in terms of where did this come from, or can you reposition the question? I've had a set of results which came back with X, Y, and Z. Can you provide me a scientifically led answer in terms of how to effectively approach this? You know, just shaping that question more specifically so that you're actually dictating the narrative. So rather than it searching everything, you're actually asking it a honed-in question. And then it will pick from the peer-reviewed research papers because that's exactly what you've asked it. But then can you go one level further? Can you give me a peer-reviewed research paper that's the most up to date around this particular biomarker and the most up-to-date intervention? So just adding in most recent, most up-to-date, most valid, most researched, those little additions for people who let's just say can't afford to seek private healthcare medication. Just adding those little phrases is going to give you just that little bit more of a chance of having the input that they want.

SPEAKER_03 40:02

Yeah, it's really good. Yeah, that is so important. I think when I'm using it, because I know what I know and I've got all the experience and and everything, I do know what to ask it. And that is the difference, isn't it? And then I do know when it's wrong.

SPEAKER_01 40:17

Yeah, I say you'll know when it's wrong.

SPEAKER_02 40:19

But also, this and I we we're thinking about people maybe coming to you or I, but also if you're heading off to your GP, for instance, it gives you a chance. If you've asked the right questions and you've got a bit of knowledge, it gives you a chance to be a partnership with the GP rather than just being told what to do. And you can start to work things out like between you, and then you're on a level rather than just being told what to do.

SPEAKER_00 40:48

It's kind of like you're taking autonomy over your own recovery, and that is really important as well because that'll set your own foundations for adherence, but then you know, you're not just being dictated to, I think is a good thing. Yeah, 100%.

SPEAKER_02 41:02

If you're empowered, you're yeah, you're yeah, far more bought into it. Yeah. Have you had the you've obviously you've got input into the AI stuff that you're creating with GutMap. Do you have any input into the into the like the wider AI? Does that does that go into something everybody can Google or AI?

SPEAKER_00 41:26

No, so it's kind of like an in-house AI, is probably a good thing, a good way to describe it. So in terms of publicly led AI, is obviously there through apps, and but then there's obviously things like um intellectual property, which is obviously where we are with ours, that we are very open in terms of our own AI is governed by humans, it's covered by a scientific board, and it's governed under these principles. So people like Julie can essentially accept those results and know, do you know what? I'm going to be getting these results from a reliable source, and I've got all of the references that I need in order to back up my claim. So in short, the answer is no, we don't take external AI, it's an in-house AI which is led by us in a way that supports consistency for both biological interventions, interactions, insights, and for protocols to help support the problem. And then who's liable? We are.

unknown 42:26

Yeah.

SPEAKER_00 42:28

And that's something that you know we hang our hat on. You know, we're very confident with our AI, helping thousands of different clients through professional practitioners to support people's symptomatic improvements. And that's something that we can hang our hats on. But again, I talk about

GutMap Approach To Safer AI Support

SPEAKER_00 42:45

accountability. We're accountable. So we're making ourselves accountable. And that's something that I think that external AIs, like we've mentioned before, just don't do.

SPEAKER_02 42:54

Yeah, I'd never even thought about that. But yeah, if something goes seriously wrong with information through a general AI, there's there's no backup, there's nothing, is there? No, no, no.

SPEAKER_00 43:07

No govern, no governance, no replies. I don't know even how you'd contact the the user. Where's the chat GPT help line?

SPEAKER_02 43:16

I'm sorry, this information has really messed me up.

SPEAKER_00 43:19

Yeah, maybe we should type in. Um, can you give me the number for your support center on uh Chat GPT and see what happens?

SPEAKER_02 43:27

I am so gonna do that.

SPEAKER_03 43:31

Yeah, exactly. Oh, thank you so much for coming on again, Josh. We appreciate it. It is an interesting topic to discuss. And yeah, you're right. As a as a practitioner, it is really important that I'm using companies like yourself that have all of this in the background. It's we as practitioners, it's the same when we're looking at supplements. I was talking to my husband, he was looking up something for something he's got, and I said, Hang a minute, let me Look, has that company been independently uh checked and tested? And where are they getting their ingredients from? He's like, Oh my goodness, I was like, This is important, you're gonna put it inside your body, you need to know where it comes from. So, is it it's so important? And um, as you know, I do I do love your test and I use it a lot in my clinic, and and also the food intolerance testing and the genetic testing you do. It's it's all brilliant. The report's excellent.

SPEAKER_00 44:27

Appreciate that. Thank you. Thank you for having me on again. I might test you again to come back on. So I like chatting with you guys.

SPEAKER_02 44:33

You can come on any time. I think I honestly I think you may have left me with more questions going round in my head than I might have to listen to this one again myself, just to kind of process it. It's huge. And I do think about Terminator whenever we think about AI. Yeah, I guess like the AI creators. Have they not seen Terminator? Can we not see what's gonna go wrong here?

SPEAKER_00 44:56

It's always the fear, isn't it?

SPEAKER_02 44:58

Yes. Um, but he'll be back. Thanks

Better Prompts And Asking One More Question

SPEAKER_02 45:05

so much.

SPEAKER_00 45:06

Thank you guys.

SPEAKER_02 45:07

Thank you so much for joining us today. We love creating this for you. We'll be back next week with another great episode.

SPEAKER_03 45:13

Until then, we'd be beyond grateful if you'd subscribe to the podcast and leave us a glowing review. If you've already done this, thank you so much.

SPEAKER_02 45:20

Please do share the podcast with friends and family. You never know which tiny piece of information could be life-changing for someone you care about.

SPEAKER_03 45:28

We absolutely love hearing from you. So connect, comment, or message us on our social media channels. You'll find all the links in the show not.

SPEAKER_02 45:35

And if you haven't already, come and join us in our free Facebook group where we continue the conversation and you get to connect with like minded women. We'd love to welcome you in. Until next time, stay fabulous!