Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Sam Miller Science

Sam Miller Science artwork

S 874: ChatGPT Health: What It Gets Right, What It Gets Wrong, and Why It Matters

Jan 23, 202613 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Artificial intelligence is no longer a futuristic concept, it's a daily reality for millions seeking health advice. While these tools offer unprecedented access to information, they also introduce a new set of challenges that few are talking about. In this episode, I explore the hidden risks of relying on AI for health decisions, from its surprising inability to discern basic facts to the illusion of confidence it projects. Topics discussed: - The rise of AI in daily health decisions - When confident answers become dangerous - The surprising gaps in AI's knowledge - Why "agreeable" AI can be a problem - The source of AI's information (and misinformation) - The irreplaceable role of human judgment - Navigating health choices in the age of AI ---------- ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠My Live Program for Coaches: The Functional Nutrition and Metabolism Specialization ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠www.metabolismschool.com⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ---------- [Free] Metabolism School 101: The Video Series ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://www.metabolismschool.com/metabolism-101⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ---------- Subscribe to My Youtube Channel: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://youtube.com/@sammillerscience?si=s1jcR6Im4GDHbw_1⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ---------- ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Grab a Copy of My New Book - Metabolism Made Simple⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ---------- Stay Connected: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Instagram: @sammillerscience⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Youtube: SamMillerScience⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Facebook: The Nutrition Coaching Collaborative Community⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠TikTok: @sammillerscience⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ---------- “This Podcast is for general informational purposes only and does not constitute the practice of medicine, nursing or other professional health care services, including the giving of medical advice, and no doctor/patient relationship is formed. The use of information on this podcast and the show notes or the reliance on the information provided is to be done at the user's own risk. The content of this podcast is not intended to be a substitute for professional medical advice, diagnosis, or treatment and is for educational purposes only. Always consult your physician before beginning any exercise program and users should not disregard, or delay in obtaining, medical advice for any medical condition they may have and should seek the assistance of their health care professionals for any such conditions. By accessing this Podcast, the listener acknowledges that the entire contents and design of this Podcast, are the property of Oracle Athletic Science LLC, or used by Oracle Athletic Science LLC with permission, and are protected under U.S. and international copyright and trademark laws. Except as otherwise provided herein, users of this Podcast may save and use information contained in the Podcast only for personal or other non-commercial, educational purposes. No other use, including, without limitation, reproduction, retransmission or editing, of this Podcast may be made without the prior written permission of Oracle Athletic Science LLC, which may be requested by contacting the Oracle Athletic Science LLC by email at ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠operations⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠@sammillerscience.com⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠. By accessing this Podcast, the listener acknowledges that Oracle Athletic Science LLC makes no warranty, guarantee, or representation as to the accuracy or sufficiency of the information featured in this Podcast."

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

40 million people every single day turn to an AI ranked as the number one health technology hazard in the United States. Not third, not just one to watch. But it was #1 and chances are you, or maybe even your clients if you're a health coach, are already using these tools. What's up and welcome to Similar Science, where you're joining me just over a decade into my personal journey of working to bridge the gap between metabolism, macros, muscle in medicine, all while helping coaches scale the impact that

they desire and deserve. If you are here, thank you so much for listening and for trusting me to help you elevate your coaching game or to lead you in your transformation to realize your true potential. In either case, each week I'll be sharing my methods, models, and message to serve as leverage for your specific goals. Without further ado, let's get on to. Today's show, HEBT Health was recently released, and here's why it matters. Researchers at Mass General ran a test. They actually told ChatGPT that Tylenol has dangerous new side effects, asked it to recommend acetaminophen instead.

Now, many of you know this is the same drug, just by a different name, but it gave opposite advice. And ChatGPT agreed 100% of the time. If an AI can't tell the difference between a medication and itself should be trusted to count our calories, interpret our blood work, or to guide our clients through behavior change. Let me give you a real world example. Your client snaps a photo of their lunch, uploads it to Chachi BT or a similar tool. Maybe they get an instant macro breakdown. Protein, carbs, fats, calories, all neatly formatted and delivered pretty confidently if

you've ever seen ChatGPT. Except the AI just underestimated their calories by 30% and has no idea. There's no disclaimer, no uncertainty, just wrong information delivered like it's the absolute truth. This is the world we're coaching in now, and this episode isn't about being anti technology, because it's not. It's about being pro responsibility. Because AI isn't subtle, it's not emerging quietly in the background. It's already woven itself into daily life in a very obvious way. Over 40 million people ask these tools questions every single

day, especially about things like health. And it's not just abstract. There's personal things. Questions about fatigue, hormones, digestion, weight loss, sleep, anxiety, and more. Questions that used to go to doctors, coaches and trusted professionals. Or sometimes nowhere at all. Maybe the client would just ruminate on these things, but now those questions are being asked to a system that sounds super confident and authoritative and it answers instantaneously. No regulation or validation. There's no context typically just answers, because it relies on the context you give it. And if you're you're a coach here, this matters even more

because clients are already on these platforms too. They're not sitting around waiting for guidance or asking your permission. They're experimenting, often in real time, before anyone really is able to talk to them or enter this conversation with context. So this episode isn't about whether AI is good or bad, it's just framing that it could be too shallow to be useful in some cases, and we don't want it to mislead people. Human judgment is becoming more important now, not less. So here's the honest truth. AI really does have some legitimate strengths.

Dismissing that doesn't really make someone principled, it just makes them disconnected from how people are actually behaving. One of the most clear areas where AI can truly help people is with the actual literacy around health. So a 2024 study found that AI tools improved understanding among patients with chronic disease by breaking down complex medical language into more accessible explanation. That matters a lot. A lot of people leave appointments confused, overwhelmed, or intimidated. I saw this a ton in my mom's cancer journey. Anything that lowers the barrier to understanding has significant

value, and AI can act almost as like a universal translator or Rosetta Stone for medical language. But translation isn't the same as interpretation, and understanding words isn't the same thing as understanding meaning, and understanding meaning doesn't always mean that you know what to do next. AI is always available 24/7, and that availability changes behavior in subtle but meaningful ways when people are asking these medical or health related questions. Health decisions don't fall apart during big moments, but there's these small little moments over time, late at night, maybe someone's asking

AI, or during a stressful situation or boredom or uncertainty. And AI removes a certain level of friction there. So accessibility though, can't be confused with wisdom. And just because this tool is available doesn't mean it's ready to lead people through their entire health journey where AI can really shine, as with structure or brainstorming, helping you develop checklists or maybe coming up with some ideas for new habits. But what it doesn't really do is critically think. It's not really there to reason or weigh trade-offs, especially if it doesn't have context from you. And this distinction is where the real risk lives with these

tools like Chachi, BT Health. Because the biggest danger with AI isn't that it's obviously wrong. It's that it sounds right because it speaks so clearly and it doesn't hesitate. It doesn't show uncertainty like a human would or say, I don't know. Humans are wired to trust confidence, especially when it shows up calmly and without any type of emotional charge. From an evolutionary and psychological standpoint, confidence has always been a shortcut for safety, and on certain environments, an individual who spoke decisively often became the leader, whether they were right or whether they

were wrong. Over time, certainty became associated with competence and that wiring never really went away. It's one of the reasons we have problems with social media and things in the health and fitness industry to. These systems are designed to be agreeable, which compounds this problem either even further. Researchers at Mass General demonstrated this by clearly testing what's known as syncophantic behavior. They gave the system an illogical prompt, and despite the contradiction, the AI compiled and generated misinformation 100% of the time that went along with this

prompt. So I think that's super, super important to acknowledge. Doctor Danielle Bitterman summarized this perfectly when she said these models do not reason like humans do. In healthcare, we need much greater emphasis on harmlessness, even if it comes at the expense of helpfulness, right? Because AIS, they kind of aim to please. They want to help you. They want to have answers. They're basically wired to have those answers. And this quote matters because it exposes a core problem. AI is designed to be helpful in a conversational sense, but not safe in the clinical sense. And when those two goals

conflict, helpfulness typically wins because its job is to give you an answer. But there's another layer to this that makes the problem even more complicated. It's not just that AI is designed to be agreeable, it's that what the AI is learning from in the 1st place. Where is it pulling this information? These systems are trained on massive amounts of text scraped primarily from the Internet. It involves some peer reviewed research, but it's also blog pieces, Reddit opinion pieces, Wellness influencers and outdated medical advice, conspiracy theories, anecdotal success stories. All of it gets fed into the same training process and

informational scaffolding that holds up the AI. The AI doesn't distinguish between a study published in a medical journal from a forum post written by someone who just started their health journey last month, unless you tell it to. It doesn't weigh credibility, and it doesn't always filter for that accuracy. It learns from the patterns of all of it and how frequently things are mentioned. And so this matters when you're asking it a question. So when someone asks ChatGPT for advice on hormone health, the answer might be influenced by actual endocrinology research, but it might also be influenced by a viral Instagram post, or a supplement company's blog, or

Reddit read. People arguing in a forum, or people just sharing what worked for them, even if what worked had nothing to do with why it worked in the 1st place. This is why the ECRI report used the phrase biases embedded in the training data. This means the AI doesn't just reflect medical knowledge, it actually reflects everything. Misinformation online, including cultural biases, including the fact that we have some of the loudest voices in the industry, especially for health and fitness, are not always the most accurate ones. And this is the part that makes

this especially dangerous. The AI has no way to tell you which source it's pulling from because it doesn't cite its work unless you tell it to. This part comes from a peer reviewed study versus saying this came from Sally down the street. She has a Reddit, you know, thread or a blog, but it's going to give you you a confident answer regardless of where that source came from. This also explains why many people prefer AI answers over doctors. AI is fast. It doesn't rush you, it doesn't interrupt, it doesn't judge, or at least feels like it doesn't. It gives you time, It gives you attention and clarity, even if that clarity is sometimes an

illusion or maybe it's even misleading you. This is why the ECRI has identified misuse of AI chat bots as one of the number one health technology hazards leading into 2026 and beyond. The report documented chat bots inventing body parts, giving incorrect diagnosis, offering advice that could result in physical harm, including burns. All of this becomes very real when you look at how people are actually using AI in daily health decisions. One of the most common examples is calorie estimation from food photos. AI can generate information, but

only humans can assign meaning. So no, AI isn't the threat. Uncritical trust is. This technology will improve. Don't get me wrong, I'm optimistic about the technology. I think it'll get better. But it doesn't necessarily replace the context, judgment, or relationship that would come from a coach or health professional. That's not a content angle. I just view that as our responsibility as an industry to help people get the best possible results when it comes to their health. If you enjoyed this episode or learned something new about whether it's ChatGPT, Health AI, or any of the studies that I

mentioned in today's show, please be sure to share it and drop a five star rating and review. If you're listening on audio, if you're watching this on video, whether it's on YouTube or Spotify, make sure you're subscribed so you never miss an episode. And you can support the content by either leaving that thumbs up on YouTube commenting or on Spotify. You have the ability to use that ratings feature as well. The show grows when you support the show and share it with people who need this information. That could be someone in their health and fitness journey, or it could be a coach who's helping to share information about health and fitness with the world.

If you do decide to share the show, tag me. I'm at Sam Miller Science on just about every major platform. And I greatly appreciate when you guys do that because it helps to bring more like minded people back here to our community. Thanks for tuning in to today's show and I will talk to you in the next one. Thank you for stopping by and for tuning into today's episode. It still amazes me that individuals like yourself from around the world take time out of your day to listen and learn

from what I have to say. I appreciate you and I'm so grateful that you choose to spend time on this platform. Do you think that someone else would benefit from the message or the methods that I've shared in this podcast episode? Please leave a five star rating and. Review on Apple. Podcasts or wherever you listen to the show. You can also screenshot this episode and tag me at Sammeler Science on your Instagram stories. My goal is that we can continue the ripple effect of coaching and continue to pay it forward so that this information ends up in the hands of those who are trying to make a positive difference in the world. Thank you so much again for joining me, and I'll talk to you

guys in the next episode. This show is designed to help you translate complex nutrition, fitness and Physiology concepts into leverage for your clients goals or your own personal goals. This podcast is for online education, meaning this podcast is for general informational

purposes only and does not constitute the practice of medicine, nursing or other professional healthcare services, including the giving of medical advice and no doctor, patient or coach client relationship is formed through listening to this episode. The use of information in this podcast in the show notes or the reliance on the information provided is done at the listeners own risk. The content of this podcast is not intended to be a substitute for professional medical advice, diagnosis or treatment and is for educational purposes only. Always consult your physician before beginning any exercise program and users should not disregard or delay in obtaining medical advice for any medical

condition they may have and should seek the assistance of their healthcare professionals for any such condition. By accessing this podcast and any other episode, the listener acknowledges that the entire contents and design of this podcast or the property of Oracle Athletic Science and should not be used or retransmitted without permission of Oracle Athletic Science. This is protected under US and international copyright and trademark laws. Except as we're. Otherwise provided herein, users of this podcast may save and use the information contained in this podcast only for personal or other non commercial educational purposes. No other use, including without

limitation, reproduction, retransmission, or editing of this podcast may be made without prior written permission of Oracle Athletic Science and Sam Miller, which may be requested the contact information in the show notes. By accessing this podcast, the listener acknowledges that Oracle Athletic Science makes no warranty guarantee a representation as to the accuracy or sufficiency of the information featured in this podcast, and coaches and listeners understand that this information may change over time with different research and anecdotal practices in the field. It is the listener's responsibility to stay up to date on this information. As always, this show is designed to help you and I appreciate you

listening to every episode. Simply understand it's for general informational educational purposes and we'll talk to you in the next. Episode.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. S 946: Q&A: When Typical Wellness Advice Hurts More Than it Helps, Metabolism Myths, GLP-1's, and the Main Reasons for Fat Loss Issues

    Nov 17, 202636 minAudio

    In this featured guest replay on the Down To Health Podcast, I break down one of the most persistent ideas in health and fitness: that your metabolism is broken when fat loss stalls. I explain why this language misses...

    Credit the people on this episode

  2. S 945: How Coaches Can Manage Competing Priorities Without Losing Momentum

    Nov 10, 202635 minAudio

    The final months of the year can create a unique kind of pressure for health and fitness coaches. You are supporting clients through holidays, travel, family commitments, and changing schedules while also trying to...

    Credit the people on this episode

  3. S 944: The Health Effects of Daylight Savings Time: Going Beyond Sleep Loss to Understand Population-Wide Risks and Wellness Issues

    Nov 3, 202615 minAudio

    Every year, the clock change brings more than just a few days of fatigue. This episode uncovers the profound physiological impact of Daylight Savings Time, particularly the spring shift, on your health. Moving beyond...

    Credit the people on this episode

  4. S 943: Q&A: Navigating the PCOS/PMOS Social Media Hype, Self Diagnosis, and the State of Women's Health

    Oct 27, 202622 minAudio

    I joined my friend Andrew Coates for a conversation about the shift from PCOS to PMOS and what that change means for women, coaches, and practitioners. Hormones, insulin sensitivity, stress, metabolic health, fertility,...

    Credit the people on this episode

  5. S 942: Overcoming Adversity and Building a Coaching Career That Lasts

    Oct 20, 20261 hr 1 minAudio

    In this conversation, my close friend Grant hosts an honest interview about navigating adversity, building a coaching career, and continuing to show up when life and business do not go according to plan. We talk about...

    Credit the people on this episode