The Liz Moody Podcast - How To Make Sure AI Changes Your Life For The Better: Brain Health, Environmental Concerns, The Future of Jobs, & More

Episode Date: August 6, 2025

I sit down with AI expert Harper Carroll—an artificial intelligence and machine learning engineer, advisor, and educator—to answer all of your burning questions about AI. Is it destroying our brains? Will it take our jobs? What about the environmental cost of asking ChatGPT to write an email?  But we don’t stop at the fears. Harper shares actionable, real-world strategies to help you harness the power of AI in ways that will transform your life and the lives of those around you. Whether you’re looking to boost your career, improve your mental and physical health, or simply stay ahead in an AI-driven world, this episode is packed with insights you can start using today.  We get into:  Practical ways to use AI in your daily life  How AI can revolutionize your business  The surprising ways AI can support your health  Ethical and environmental considerations of AI  How to prepare your kids (and yourself!) for an AI-driven future  What AI means for job security—and how to stay relevant  And so much more For more from Harper Carroll, find her on Instagram @harpercarrollai or online at www.harpercarrollai.com. Ready to uplevel every part of your life? Order Liz’s book 100 Ways to Change Your Life: The Science of Leveling Up Health, Happiness, Relationships & Success now!  Connect with Liz on Instagram @lizmoody or online at www.lizmoody.com. Subscribe to the substack by visiting https://lizmoody.substack.com/welcome. Use our discount codes from our  highly vetted and tested brand partners by visiting https://www.lizmoody.com/codes.  To join The Liz Moody Podcast Club Facebook group, go to www.facebook.com/groups/thelizmoodypodcast. This episode is brought to you completely free thanks to the following podcast sponsors: Seed: Go to Seed.com/LizMoody and use code LIZMOODY for 25% off your first month.  Maui Nui: Head to MauiNuiVenison.com/Liz to secure your access to a limited collection of Liz’s favorite nutrient-dense wildly delicious meat cuts and products.  LMNT: Go to DrinkLMNT.com/LizMoody to get a free LMNT sample pack with any order. IQ Bar: Text LIZ to 64000 for 20% off all IQBAR products plus FREE shipping. The Liz Moody Podcast cover art by Zack. The Liz Moody Podcast music by Alex Ruimy. Formerly the Healthier Together Podcast.  This podcast and website represents the opinions of Liz Moody and her guests to the show. The content here should not be taken as medical advice. The content here is for information purposes only, and because each person is so unique, please consult your healthcare professional for any medical questions. The Liz Moody Podcast Episode 353. Learn more about your ad choices. Visit megaphone.fm/adchoices

Bill Gates said that over the next 10 years, AI will replace the majority of jobs and that humans aren't going to be needed for most things. How do you think that AI is going to shift the career landscape? The scientific discoveries that we will see with AI will enable jobs that we have never seen before. How do you think about the ethical considerations around something like that, AI being trained by people's work? How is AI going to impact our health or our lifespan in the future? I mean, the health implications of AI are astronomically wonderful. I mean, we're going to see cures for many diseases, potentially cancer. How do you think about the environmental implications of using AI?

There's a few thoughts around this. Hello, friends, and welcome back to the Liz Moody podcast. Today, we're talking about AI. Some people say that it's destroying the environment, taking our jobs, potentially making us all go extinct. Other people say it holds the key to curing cancer, democratizing education, and solving climate change. So what is the truth? Is AI a utopian dream or a dystopian nightmare? And more importantly, how should you, a normal person just trying to live your life, be thinking about it? Should you be

encouraging your kids to use it or protecting them from it? Is it destroying our brains? Will it take our jobs? And what about the very real environmental cost of asking ChatGPT to write an email? I am so excited to say that today we have the perfect person to answer every. every single one of these questions and so many more. Harper Carroll is an AI and machine learning engineer, advisor, and educator, who has made it her mission to make these complex topics accessible to everyone. With nearly a decade of formal experience, she holds two computer science degrees from Stanford specializing in AI,

and she worked as the founding engineer and the head of AI at a startup acquired by NVIDIA. Her background combines work in the field of AI with teaching experience at Stanford and now online to the masses where she has over 500,000 followers on social media. So she is the perfect person to communicate the science, but in a way that we can all actually understand. This episode gets into a lot of the fears that many of us rightly have around AI and also gives a ton of practical action steps that you can use to mitigate the risks and even turn AI into something that will transform your life for the better.

Harper Carroll, welcome to the podcast. Hi, Liz. Thanks for having me. I'm so excited to have you here. I have so many questions for you. This is like the type of thing that I'm talking about at dinner all of the time these days that I'm asking everybody for their thoughts on. So I'm really excited to get into this. Let's start with just though like in a nutshell, if you had to say how AI is different from Googling something. And I know Google is now using AI for its responses, but like five years ago, what's happening that's different? Yeah. So Google is actually searching through sites, whereas AI models, at least the base models are learning from text data. So it grabs everything on the internet. And it's actually learning the probabilistic distribution of words. So it's a word generator.

Whereas Google is looking up your words and it's looking at kind of a page rank system and seeing what sites are most relevant and actually looking at concrete data. AI models learn from the data, but it's just a word generator based on learning the probability distribution of words. So by learning patterns and text, it outputs new text according to those patterns as opposed to Google. But now AI models are being enhanced with the capability to search the web and to access concrete data as well as being this word generator, which is very important because we do need to tackle the issue of hallucinations, which is when models make things up. And that is an inherent property of models because they are word generators.

Okay. I'm just trying to like make sure anybody listening is like, if I'm talking about this at happy hour, like do I actually understand what is going on here with AI? Are they generating new ideas? Like are they putting words together that's not been there before together? So that is actually possible. Now that's something called emergent behavior. When a model shows behavior that is not shown in its training data, that's called emergent behavior. And that is something that's, that was never really the case with large language models. So models like ChatGPT or these tech generation models, those are called large language models. In the past, historically, they would really just output, only really output based on their data. So they wouldn't come up with anything new and novel. But what we're seeing now is that these models are actually able to draw

correlations between distinct fields of, you know, science or math or, you know, certain types of science that were never found before. So they were obviously not in the training data. One thing I like to talk about is polymaths, the concept of polymaths. So polymats are those who are experts in more than one field. And in history, we haven't had many true polymats. Some of them were like Nicola Tesla or Thomas Edison. And they were able to become experts in more than one field and then draw correlations

between those fields and then have these outsized impact on the world based on those insights. AI is the ultimate polymath. It's able to become an expert in thousands of different fields. And so now that it's showing this emergent behavior, which means it's being able to draw insights that weren't in its training data, we're going to see so many scientific discoveries and advancements. But I guess back to your original question. Generally, it is outputting text based on what it's seen before with this caveat of having this new kind of emergent behavior. When you say training data, what? What do you mean? What is it trained on specifically? Most of the large language models, the massive large language models that you hear about like chat GPT, meta, slama, those are trained on the internet. So they have these bots. There's just code that scrapes the internet and grabs massive amounts of text. And so then the model learns. I went to a website once that let me put in my book titles and see whether or not AI had been trained on my books. And it said yes. What does that mean? Is somebody, uploading all of my books onto some sort of model that it's, what does that mean? It's such a good question. You know, I wonder about that as well. My guess is that someone

uploaded a PDF of your books online and the scraper found it. Okay. You know, and added it. And is it trained on a lot of different books? Like most books that exist. If those books are on the internet, my guess is it's found a way to grab them. I mean, I think there's ethical considerations of this. Maybe I'll ask you about that too, but I have a smaller quam, which is that everybody online thinks that if you use an M-Dash that AI has written, whatever you've written. And I'm like, and writers love M-Dashes. Like, we love them. And AI loves them because it's been trained on our work.

I know. And so I feel like I have to go through my work now and delete M-Dashes so people don't think it was written by ChatGPT. Well, you can have the normal. I just keep doing it. But I have noticed that chat GPT dash is a little different. Oh. Like it's basically, it's a little longer and it connects the actual, the two words are actually connected.

That's how I do my M dash. Oh, that is how you do it. That's like a proper M dash. I do no, a proper M dash. You should go word, no space, long dash, and then no space and then word. And I love it. It's my favorite form of punctuation. And I feel like AI's taking it away from me.

Oh, you should just keep doing it. I know. But then everybody's going to like, AI about this. That's so darn. Oh, I'm sorry to hear that. How do you think about, we're going to get into a lot of like how does AI impact me? How should I be thinking about AI just for the normal person living a normal life? But how do you think about the ethical considerations around something like that,

AI being trained by people's work? It's a really challenging discussion. I think it needs to be treated with nuance. So training on specific people's work, I think is very questionable. And it's a challenging question because like you could argue that when we go about the world and we're taking in art and, you know, images and books and talking to people. And we are creating our own probability distribution in our minds based on what we've learned. We are and I, natural intelligence, right?

We take in what we see and then we kind of act that way. And the same way, you know, you kind of hear you're the sum of the people you spend the most time around and, you know, you act like your mom. And like we're picking up these things from other people in the same way AI models are. Or rather, they are picking it up. in the same way we are. And so you could say then that when it generates a piece of art, if you were to say, generate a picture of a cat, it's going to generate a picture of a cat, but it's going to be so broad, it'll have seen thousands and thousands of images of cats. Like if you give it a term that is so broad, I think where we get into sticky territory

is when you're saying, you know, generate a picture of a cat that is in the style of Van Gogh, you know, or asking about specific artists. That's where I think it gets sticky. and I'm not an expert at the ethics and law that is not my domain, but I understand why people would be upset about it because it does have this kind of, you know, you could ask a person generate a cat in the style of Van Gogh and they do their darned best, but it would not be as good as an AI doing it. So there is that difference there. So it's a challenging question and I definitely care about the nuance. But I think I also want to emphasize that AI is so much more than just large language models

and image generation models and video generation models. Like so much more than what we see and play around with as the layman on the Internet. Yeah, like the insights that we're going to see from AI in terms of health optimization and technological advances, those kinds of insights. AI is so much broader than these generation machines. like generation is just one area of AI, but there's so much more. And there's so much promise in the generation models, but also in the rest of the forms of AI.

I feel sad sometimes when I think, when I see people throwing away AI, it's like throwing the baby out with the bathwater. You know, there's so much more there than just generation. When I think so much of it is about like what gets attention. We live in an attention economy, and I do think that it's like, what are people paying attention to? and the memes and the silly stuff and asking AI to help you manifest your future and stuff like that, that plays better on social media, I think.

But we're going to talk about a lot of the exciting developments of AI. But first we're going to go the other direction, which is that Professor Jeffrey Hinton, he's called the godfather of AI. He was given a noble prize for his work on AI. And he said there is a 10 to 20% chance that AI would lead to human extinction within the next three decades. There are a lot of other AI safety researchers who've painted even grimmer pictures. Do you think that we are being too casual about the potential for our own extinction when we are talking about AI? I agree that there's potentially a risk.

I will say that Jan Lecon, who I think I'm saying that right, French people don't give out of me. He argues that there is basically no extinction risk. Oh, this is interesting. Yeah, he's the lead researcher at Meta. He's brilliant. He's working on all these cutting-edge models. and he's completely on the opposite side. He thinks we're really far from AGI.

What is AGI for somebody listening who's not familiar? Artificial General Intelligence. And it's the idea that it's AI that basically has the brain capabilities of a person in the sense that you can put it into an environment and it can navigate it. It can really do anything that you ask it to. If it doesn't know how, it can learn very quickly. It's basically a person. And that is much harder to train for one of the reasons.

being, you know, text generation models seem very human, but that's because we have copious amounts of data for it, and it's just a text generation machine. Whereas this is actually navigating an environment and taking action and developing that data set or just training it in an environment. There are different ways you can train AI models, both of which are very challenging to do for something like an AGI. And so Yahn argues that we are quite far away from AGI. He also thinks that, you know, in general, even if we we were to have an AGI. He doesn't see intelligence being correlated with the desire to dominate and for power. He said, you know, if that were the case, then Einstein would be, would have

been rich and powerful, but he was not. Most of the top scientists that we've ever seen have not been rich and powerful because that is not correlated. Those two aren't necessarily correlated. He also argues that we will be able to handle problems as they come up. We underestimate our own ability to handle problems we always have been able to. He also says that people kind of see AI through a lens of like the Terminator based on the media that they have received. And he thinks that that's just not actually going to be the case. So again, those two figures are both considered godfathers of AI and they're just, they differ.

So it's just hard to know. A lot of really top researchers are not concerned about extinctionism. What do you think personally? Well, I think I see the benefits of AI being so extraordinarily great that, well, first of all, I think the risk is low. I'm kind of on yon side with this. Second of all, I think the benefits are so extraordinarily great that I think it would be such a shame to not move forward because of a hazy fear that may or may not come true. So especially since there are so many top researchers who are not so concerned. Would it impact your thoughts on like having a child or something where you're like, I'm too nervous about the future of the world with all this stuff to want to bring a child into this world?

I want to have a child. I'm so excited. I'm so excited about AI for kids. This is something I get really excited about. I would love to have a kid. I'm like, I'm more excited than ever. Interesting. What are you excited about for AI for kids? Well, I think kids. have, they're born with this kind of insatiable curiosity. And so many kids grow up with a single parent or a parent that's working two jobs. And, you know, parents are busy and they spend time with their kids. I'm not saying they should replace it. But instead of putting the kid in front of a screen, they could be talking to an AI and asking questions or using the AI with their parents and going into something they're really excited about. Like some kids have

obsessions with dinosaurs. They can learn all about dinosaurs or, you know, more. advanced kids or less advanced kids can practice talking and, you know, they can talk to the bot. So the AI would have guardrails around it. We would have AI specifically for kids. And then the kids could even, you know, talk to the AI, see what they're saying, show up on the screen and learn to read and write faster. I mean, because that correlation between speaking and seeing it on the screen or hearing it and seeing it on the screen, it's like you can learn to read so much faster, learn to write, You can go down whatever topic interests you the most. So I think even just a basic large language model,

and all kids would have access to this because you don't need a very advanced model for kids. These expensive models are mainly for adults doing highly complex tasks or need PhD-level research resources. But for kids, you could have a standard model available for free to every kid. So it would democratize education in a sense. Exactly. To everybody in the world, which is actually kind of crazy,

because we talk about making education more accessible here, but what about somebody who doesn't live anywhere near a school or something like that? In the way that the internet has helped democratize it, AI is going to be that so much greater. That's interesting. That's really interesting. I have a girlfriend and her kid goes to the school

where they have AI-assisted learning for like three hours a day and because it's individualized for every kid, they fit all of sort of the government, you know, the learning that I would have gotten in my normal public. school into a much smaller period of time because you're not having to normalize for 40 kids in a class. And then they do other like group projects or play outside or learn in the real world for the rest of their school time. And her kid loves it. I have chills. That's so amazing. Right. Like kids can learn at their own paces and it will, it's just exactly. Okay. We're going to get into more

scary stuff in a second. But since we're on education, let's talk about education. I think that there's a lot of fear from teachers that I hear. There's a lot of concerns about what is the future of AI look like in the classroom other than what we just said when you picture how AI is going to impact education in the future. What do you picture? I picture kids having their own personalized model working with them on learning lessons. They can ask questions. They can go in person and be with the teacher. And if they need to work through something, if their personal model is struggling with something or they want to do group work with other kids, then that will be there as a resource. But kids will be able to learn at their own pace through their own examples that

they find interesting. So if one kid's like, one kid likes baseball and a kid likes cooking, you'd be able to teach them in a way that excites them. I also see kids potentially going down certain routes that they would have never been able to. So for example, if someone has a really niche interest, a kid is born with a gift for some very very, very, esoteric interest and no one in their community knows how to talk about it. None of their teachers know anything about it. They grow up in, you know, potentially a small town. AI can help them flourish in this kind of passion that they've been born with. So I think we're going to have a generation of gifted kids because these kids are going to be enabled rather than shut down. And

their curiosities, you'll be able to stoke that flame rather than, you know, have parents put it out. I actually had a few parents reach out to me. One example I think is interesting is they have two kids, a nine-year-old and a 12-year-old, and they said that AI is kind of like their cool older sister who talks to them about things that they would never listen to their dumb parents about, quote-un-unquote, dumb parents. They were talking them about exercise and nutrition and screen time and things that they were like, no, mom and dad, but their cool 20-year-old older sisters like, hey, you know, you shouldn't be using screen time. it's bad for your brain and they're able to actually listen.

I thought that was a really interesting use case. Oh, another one, I learned about this bot Norby. It's a little robot, a physical robot, that was made to help a kid with a speech impediment. And I know about this because my little sister had a speech impediment until she was eight. She couldn't say her ours. And so she would go to this speech therapist and they're super expensive and they're talking with the kid for two hours or three hours. and that's all the time you get, whereas this little bot is able to sit with the kid for infinite amounts of time,

help the kid practice, talk to the kid about what he's interested in. The child feels comfortable with this bot. And again, it's unlimited access in a personalized way to help overcome that. And then, I mean, you could have the bot work. I think now it's been expanded to helping to learn languages. The possibilities are really kind of infinite. The mattress that you sleep on is one of the highest, exposures that you have. Like, if it's off gassing, you're breathing that in for basically a third of

your life and most conventional mattresses are loaded with synthetic foams, flame retardants, microplastics, and more. Plus, if it's not comfy, you're going to be tossing and turning, you're going to be really hot. And sleep is the foundation of health. It is so important. Because of all of this, if I were not going to invest in any other part of my house, no other things in my house, I would invest in my mattress. And in fact, I have four years well before I had much money to spend on any of these things because it is such a big needle mover. The birch mattress is incredible. It is made with organic cotton, natural latex, and ethically sourced wool. So it has literally no off-gassing. You can not smell anything right when you unbox it. It has no microplastics,

no synthetic foams or flame retardants. And the wool makes it so breathable, which if you've listened to this podcast for more than five minutes, you know that I run very, very hot. Like I am a furnace. Zach has basically accepted that sleeping next to me is like sleeping next to like a little fire. And the birch mattress has been a game changer for that. It's made me sleep so much better. Like I can see my sleep score going up because I'm not hot all night long. It's also hypoallergenic, which is really worth paying attention to if you find yourself waking up stuffy or congested.

Dust mites in a conventional mattress can actually impact your breathing and your sleep quality without you even realizing it. And again, we need to be able to breathe to get good sleep. and getting good sleep is the foundation of our health. And then comfort-wise, Zach sleeps on his back and then I sleep on my side and my stomach. So we have two completely different sleep styles. And we both wake up feeling really good, no aches and pains. CNN actually named the Birch mattress, the best mattress for side sleepers. And Wired named it the best organic mattress period.

So there you go. Birch has options for every kind of sleeper in your household. They have three firmness layers. I like the Lux Natural, which is the meat. medium one, and then parents, they even have a Birch kid's natural mattress. Birch ships right to your door. It sets up in minutes, and it comes with 120-night risk-free trial. So if it is not for you, they will pick it up for free.

You have nothing to lose here. I want all of you to enjoy a deep, restful night sleep with a new mattress from Birch. Go to birchliving.com slash Liz Moody for 20% off. That is birchliving.com slash Liz Moody, and you're going to get 20% off. birchliving.com slash Liz Moody. I'm genuinely confused how masterclass gets literally the absolute top people in every single field to teach every single one of their classes. I use it when I want to learn things directly, like the cooking class from Thomas Keller

has all of the wisdom that you would normally have to go to culinary school for. But also, I'm being honest, this is like a use case I don't hear a lot of people talking about. I'll just watch it for entertainment when I want to do something that's far more interesting than scrolling. Christina Aguilera taught me to sing. Dan Boudrum's Art of Mastering Confidence and Sex Appeal class. It's 10 out of 10. There's menopause classes with leading doctors.

There's script writing with Mindy Kaly. Literally, you name it. They're on masterclass and it is such a good way to get off your phone but have something that's like not quite as long or hard to get into as a TV show or a movie and that it just keeps you entertained and interested. And you are learning. There are over 200 classes from the world's best all for just $10 a month when billed annually, and you get unlimited access to every class on the platform so you can learn at

your own pace whenever you want on your phone, your computer, or even in audio mode like a podcast. If you're looking to stop scrolling and start consuming entertaining content that makes you feel excited and helps you learn, Masterclass is it. And the best part, every membership comes with a 30-day money-back guarantee so you can start learning risk-free. Right now, our listeners get an additional 15% off any annual membership at Masterclass. com slash Liz Moody. That's 15% off at masterclass.com slash Liz Moody, masterclass.com slash Liz Moody. When you think about strength and resilience, like your ability to feel energized, to recover well, to stay strong as you get older, what do you think that actually comes from?

Most people say working out are good nutrition, and yes, of course that matters. But there is a biological foundation underneath all of that that most people are completely overlooking. I have been diving deep into this lately with the team at timeline and what I've learned has genuinely shifted how I think about my own health. Every single movement that your body makes every step, every workout, every muscle contraction depends on energy produced at the cellular level. And at the center of that is your mitochondria. Here's the thing that nobody tells you, certainly nobody told me, starting around age 30,

our mitochondria naturally become less efficient, more get damaged, more become sluggish, and over time that impacts your energy, your strength, your recovery, and your resilience. Most of us respond by pushing more. We're like noticing these things and we're adding in more protein. We're trying to fix it with more supplements. We're trying to do harder workouts. And those things do help. But timelines research suggests that we also need to be supporting the cellular machinery underneath. And that is exactly what their supplement mitopure does. It contains urolithin A, which helps your body clear out damage mitochondria and support health. healthier ones so that your cells can produce energy more efficiently. Because this is happening to

your cells, it's going to impact your entire body, your immune system, your muscles. One study found that taking mitochondria increased muscle strength by 12% in four months with no change in exercise routine, it's going to impact your energy, your sleep, your skin, your cell health impacts all of this, and urolithineg keeps your cells healthy. Timeline has done over 15 years of research in testing on this one product, urolithin A, which, by the way, most of us lack the gut bacteria to synthesize naturally. That's why many of us need to supplement it to get the benefits. This has become a staple supplement for me. It is my top way to support how I want to look and feel as I age. Support yourselves and how you age with mitopure gummies from Timeline.

Visit Timeline.com slash Liz and save up to 39% on your mitochondere gummies. That is timeline.com slash Liz. Where do teachers fit into all of this? Do you think being a teacher is an AI safe job? I do think that we will need to have teachers. I do. And again, it's so hard to know what's going to happen. Like no one knows what the world is going to look like. We're going to have so many more industries made and the scientific discoveries that we'll see will have jobs we can't even imagine yet. We'll enable jobs we can't even imagine yet. But I do think there will still be there. and need for teachers and to facilitate classrooms and to have kids come together and, you know, be social, right? But I think there will be a lot more personalized education as well. But I do think kind of facilitating that, being mentors for kids, having like a human mentor and being in person

will continue to be really important. How would you suggest that teachers approach AI now? I know there's a lot of struggle because a lot of kids are using AI to write all of their assignments. teachers are getting flooded with just like AI generated material. They're like the kids aren't learning. I hear from teachers who are like, I want to help kids learn to use AI because I know this is going to be so important in the future and teachers who just want to protect the kids that they're teaching from AI so that they can, their brains can develop in a way that we know and we're comfortable with.

How do you think teachers should be thinking about AI? That's a great question. I would have them use AI tools in the same way in the past. they were concerned about when we were introducing calculators, we were concerned kids wouldn't be able to do long division and they would become stupid. That's not the case, right? It expanded much more that they could do. So I think teachers should teach kids how do you learn these tools, get them excited about learning these tools, but get them excited about learning and actually doing the work. And this is something that we can talk about more with, you know, our brains kind of on AI.

you do need to consciously think when you're using AI. You have to choose sometimes to put in the effort to think. And so thinking of thoughtful assignments where the kids think of something first and then augment with AI and see how AI can help it and see how it will integrate. That's something that was shown in the MIT paper, that your own brain augmented with AI actually lights up certain areas of your brain that don't light up in other ways,

like with only using your brain, I think teachers should help them use AI, but in a thoughtful way to the teachers listening. There are ways that you can do that, and I think that's the best way to go forward. And I'm happy to talk about it as well. How do we keep kids from just putting in their assignment and asking chat GPT to do it?

To do it. I mean, yeah, it comes back to the love of learning. And it's such a scary time, right? Because how do we actually teach kids to want to learn? Especially like I think our brain is so wired to how can we do this in the easiest possible way. How can we expend the least amount of effort for the most results? So why would we put in the extra work to think if we can just get the output that we want? I mean, it's actually interesting because to me it actually, if we dive back, back back,

it's about our productivity outcome obsessed society. And it's about how can we teach people to fall in love with process when we have told people that outcome is all there is for years. and years and years. Totally. And I actually wonder if this will cause a shift in focus on the process rather than necessarily the outcome. Because when the outcome, you can have it in a second, what really matters is that you are kind of working out your brain, like brain health, making sure that you actually put in the mental effort and trying to like train your brain. And I think one of the ways, as we're talking about this, I'm thinking about it,

is perhaps just making the homework assignments more fun. How do we do that? Make them more customizable. How do we do that with AI? Like instantly, you could have homework assignments that kids all find interesting that they want to do as opposed to sticking something in front of them, one size fits all, super bland. Let's make learning fun. Let's make them excited about using their brains. A lot of kids are super curious. They want to use their brain when they're young. And then it gets kind of beaten out of them because their things are put in front of them that they're not interested in and then they're forced to do it. You know? That's interesting. You mentioned an MIT study. Are you talking about the one that talked about our critical thinking skills? Okay. So there was an MIT study recently that found that we had a loss of critical thinking skills when we outsource tasks to AI. It sounds like you had some different takeaways from that study. So I would love for you to share them. They had them write an essay and they put them into three groups into a brain only group, into a search the web only group and then into an large language model group where they just used AI to write it. Then they had them do a few essay, a few rounds of this. And then at a certain point, they,

had people switch over from the brain-only group to using a large language model and from the large language model to using their brain. And so what they found was that the people who were using their brains only had more interconnection in different areas of their brain and show different types of brain waves and showing different kinds of brain processes that were not visible in the other groups, particularly in the large language model group. They also found that people, there was a sense of ownership over the essay that people in the large language model group didn't have and there was also a recall,

a strong recall issue where they asked them to directly quote from the essay and the people with a large language model couldn't and the people with their brains only could. So this is an interesting study and they do bring up some important thoughts, but the way that I kind of saw it right away was that let's think of an analogous study, right? Okay, say you're ordering a book, just any book.

and you decide that you are going to either go to the store and buy a book there, or you're going to go to Amazon.com and buy a book there, or you're going to use Alexa and buy a book. If you go in person, you'll have much more choice. You'll browse the shelves. You'll have brain connections getting to the store, talking to the people in there, finding the book, choosing what you want, thinking about it, going home. If you go to Amazon.com, you don't have to drive to the store, you don't have to talk to the people,

but you're kind of browsing what's popular. It'll show you a bunch of things as opposed to, you know, massive choice. It'll show you a subset. With Alexa, it'll just order you the most popular book that week. Would you not have Amazon because of this study? Would you get rid of Amazon? I love bookstore so much. So I might be like an interesting person to ask this. But that's what I'm saying. Okay. For the people who want to go to the bookstore, That still exists. But for the people for whom this is a really tedious task, they have to pick up their kids

from school, they have so many other things going on. The fact that they can just order a book online. Makes them much more likely to read, essentially. Much more likely to read, but also they're filling their time with other things. It's not like with their free time. They're like sitting like, you know, I mean, maybe they're doom scrolling. That's kind of on them. Like, I don't mean to be flippant about it, but what they choose to do with their

extra time is up to them. You wouldn't want to get rid of the ease of Alexa in the same way with an essay. Some people, they're writers. They want to write a really good essay. And that matters to them. But for some people, maybe they just have to get something out or, you know, they have to have a few words around something. And it's done for them in a moment, right? And then they can go work on something else that excites them. You can't really look at the task in isolation. Another interesting finding from the MIT study was that they found that the people who moved from the brain-only group to the AI model. So they were integrating AI's insights with their own essay. As I said, had greater connection in certain areas of the brain.

I see AI as being most helpful to us as humans when it is used as a tool, but not as a replacement for the task. So for those who just had AI write their essay for them, it was very generic, kind of talked about the same thing. It had similar word patterns, which is obvious it would have similar word patterns, right? If you just tell it to write an essay and it's a word generation machine and it's outputing text according to patterns, then it's going to output text according to a similar pattern. But if you are using it and you come up with notes and you're thinking about it and you have your idea, it's going to not be generic because it's going to take your notes. and you can then synthesize it and you can go through the output and make sure, does that look good?

Is that you actually end up using your brain a lot? And I find when I'm using AI as a tool where it's, you know, putting structure to my thoughts, my brain is working a lot. It's not so black and white as if we use AI, we won't use our brains. How do we overcome, though, again, this human desire to conserve energy? Like I picture the end of Wally where we're all like blobs. basically end up on a spaceship and they're blobs and they move around in motorized chairs and drink soda and like entertain themselves 100% of the time because they don't need to do anything else.

So if we don't need it, we don't actually do it. How do we overcome that? Gosh. It's a big philosophical question. Yeah, I'm like, oh, well, I think I would say instilling in ourselves a love of learning of generation. Paul Conti, he talks about this thing called the generative drive, where we as humans have this drive in us for certain things. And it's kind of mystical, but it's in us. And so allowing that to cultivate. And one thing about AI is that you will actually be able to follow it. Because so many of our needs will be satisfied with AI, we won't have to worry about pure survival. We can actually follow that. I think so. so many people are stuck in jobs that they don't like, that they're just trying to make ends meet.

It's hard. Life is really hard. And with AI being able to help us take a lot of these tasks out of our hands, we can follow that generative drive. And so I'm so excited about a future where people are doing what they were born to do. The thing that I'm hearing over and over and over from you is that if we're going to have a takeaway, it's how do we cultivate a love of generation, a love of creativity, a love of learning in ourselves and in our kids. Yes, and I have to say, if you, if you've ever had the luxury of taking like three weeks off of work or more, you'll notice that you start to want to do stuff again. You have that in you. You're just so exhausted by upkeep in your life that you're like, I don't want to do anything. I just want to watch TV and

veg out and go to sleep. But I think we all have that. in us. Like when kids don't have to worry about survival and they don't have all these problems, they're always creating. That's interesting. So when you think about our brains in 50 years, you don't picture like a Wally mush brain. What do you picture? I picture people who are building what they're excited about. I think we might have a lot more artists. I think maybe art will be easier to make a living on, which I think is a counterintuitive to some people. with all these generation models, but I do think people will be able to follow their artistic passions

in a way that makes them money. So I don't know. I kind of have a really optimistic view of AI. How do you think that AI is going to shift the career landscape? Let's say like in five years and then maybe further out in the future? That's so hard to know. I don't know. The scientific discoveries that we will see with AI

will enable jobs that we have never seen before. So it really is hard to know what those jobs will look like. What I will say is that we will have so many more solopreneurs. So entrepreneurs that are solo working by themselves, build the product end to end and are able to make money off of that. So more people following their passions. Because AI will enable that. Exactly. Interesting.

It will enable businesses that were never available to individuals before, that were only available to larger corporations. That's interesting. So that's like a democratization of entrepreneurship as well. A hundred percent. Have you heard of vibe coding? Yes. I think so. I mean, yes. It's when you can like tell code what you want it to write, right, instead of actually having to write code. Yeah. Okay. So it's enabling people to have these app ideas and then create these apps without any coding experience and build it end to end. I know people who are making tens of thousands of dollars off a very simple app idea that they wished existed and now it exist and they're running this company just having built it with vibe coding. Rick Rubin, who is a very

famous producer who's worked with so many of the great musicians, he is all about, you know, energy and feel for art. He has likened the vibe coding movement to the punk rock movement. Because before the punk rock movement, if you wanted to be a musician or make music, you had to go to school for it and have practiced many years and right, there was so much there. With punk rock, you could learn three chords in a day and suddenly you're on the ground running, making music. If you had something you wanted to say in your heart, you could do it suddenly. And that's how it is with vibe coding now. If you have something that's in your heart, you can bring it to life. Where would I find the ability to vibe code? So there are some websites like Loveable is a really

popular one that's gaining massive popularity very quickly. So if you have no coding background, I would check that out. I know Bolt is another popular one, Replit. And then if you're a coder, cursor is very helpful. Or if you want to learn basic coding skills cursors, it just the code is more kind of front-facing when you work with cursor. Just these different apps that you can use. Bill Gates said that over the next 10 years, AI will replace the majority of jobs and that humans aren't going to be needed for most things. Would you agree or disagree with that? I think my view is nuanced here because I think a lot of the necessities will be accounted for,

enabling us to pursue things that we really enjoy. I don't kind of see that sentence as necessarily a doomsday sentence, right? Because it would be a utopia if that were the case. If there's another way to get money, I think the thing that freaks me out is we've circled around the idea of UBI, universal basic income. how more the government gives us, you know, a few thousand dollars a month or something like that, but only in the fringiest parts of government that's not even close to being passed. So I'm like, it would be great if there was another way to get our needs met,

but we're not even close to talking about that. Well, I think we will still need people for jobs. And I think, as I said, there will be more jobs than we can imagine now. Like maybe AI will take many of the jobs that exist today. and but we will see more jobs as a result of what AI is doing now. You know what I'm saying? And so I think there will always be kind of a role for people. And I think another thing I'll add is that these forecasts are not necessarily true.

So Bill Gates. I love how you're like, when we predict the future is not 100% true. We don't know. We are like, oh my gosh, you know, it is true. Like none of us actually have the ability to predict the future. future. None of us know, even Bill Gates. I mean, Bill Gates didn't think the internet was going to be a thing. Like, he did not think there was any real commercial value in the internet. He wrote a book called The Road Ahead in the 90s, and it didn't mention the internet. So, again, and it's a brilliant person. We don't

know. We don't know. We don't know. And he might be overcorrecting now, right? Like, so we don't know. And even the most brilliant people, especially with something like AI, it's, as I said, we're going to have to see so many discoveries that we're going to see a world that we can't even imagine now. And I think it's going to go in the best possible way. What advice would you give somebody who's in college right now and they're trying to think about what career path to go down that feels the safest? That is a very hard question. What I would say instead, it's kind of like what we were talking about moving from being outcome focused to process focused. learn to love the process of learning. Learn to seek exercising your brain. In the same way,

some people go through their teens or start working out in their teens and then keep going and then they're healthy for the rest of their lives because they build that into themselves, that that is something they're going to do. They're going to work out every day or every other day. And then they're set. If you prioritize that kind of workout for your brain, instead of necessarily what degree you get, it's more, are you showing up and using your brain every day? Are you excited to work it out? Those people, I think, will go the furthest. What about somebody who's sitting in a corporate job and they're worried that in five years their job is going to be outsourced to AI? How can they protect themselves?

I would say, learn the AI skills that are available to you. I've talked with a lot of people in corporate jobs who have said their CEOs have introduced certain AI tools to them and say, please start using this. Use those tools. Learn how they work. Once you feel good with those, do some research, see what else is available to you. Start to, as I said, love the process of learning. Start to see what's available to you. Don't let yourself be a victim in this. I think one of my biggest fears in why I'm doing this, this education thing, is that I think AI is now available to everyone. It used to be really hard to access. You know, as I, like, I've been doing this for 10 years. It used to be very confined to the select few who really understood how it worked, who were able to use it.

Now, anyone can use it. Anyone can understand it. I'm trying to help bring it to people. Use it. Do the work and learn the tools because one, I think your life will be enhanced. Yes, learning a new tool has a learning curve and it kind of sucks. But try to get excited about the learning curve, you know? and once you learn the new tool, you will really empower yourself. And things like the vibe coding tools, it's so empowering to be able to say, hey, I have an idea, I'm going to build it end to end, and I'm going to make a business out of it. How empowering is that? That's what I would say.

Learn the tools. Learn the tools that are presented to you and then potentially expand and learn vibe coding tools. I very rarely get genuinely excited about skincare, but this is one of the most innovative products that I have come across in years. and I'm so obsessed with it, and I've been telling all of my friends to get it, so now I need to tell you guys. Here's some science first. Your skin isn't just getting older. It's being actively broken down by something called senescent cells. These are cells that have stopped functioning but refused to die. They sit there, releasing inflammatory signals, breaking down your collagen, degrading your skin

barrier, and accelerating every visible sign of aging. Scientists call them zombie cells, and as they accumulate, they are one of the primary drivers of how old your skin looks and feels. The team at one skin, a group of female longevity researchers and PhDs, spent five years testing over 900 peptides to figure out how to help reduce the accumulation of senescent cells. And they finally landed on it. OSO1, the first peptide scientifically studied to reduce skin's biological age at the molecular level. OSO1 goes in and it clears out the synestine. cells so it helps skin function like healthier, younger looking skin. It is not masking the signs of aging.

It's not targeting one thing. It is actually rolling the clock back at a cellular level. I've been using the face moisturizer for almost six months now and I love it so much. It feels amazing. It goes on really smoothly. It's not tacky at all. And I actually see a difference, which I just feel like is never the case with skincare. You want to always like see a real difference. And you're kind of like, do I? Do I? And this I genuinely do. do. Because it's clearing the senescent cells, it doesn't just target one thing. So my skin looks firmer. It looks glowier. The texture feels dramatically smoother. And I feel like you can see that too. I also love the body moisturizer. It dries down really quickly, which is always a pet peeve of mine

with moisturizers. I hate that like sticky feeling when you go to put your clothes on. This does not do that. But it does moisturize really, really well. And then again, I'm reducing my skin's biological age. I am not making it just look younger. I am making it just look younger. I am making it actually younger. OneSkin has four peer-reviewed clinical studies in over 10,000 five-star reviews. The data backs everything up. For a limited time, get 15% off with code Liz at oneskin.c.co slash Liz. Again, that's 15% off at oneskin.co with code Liz. If you have headaches, allergies, brain fog, skin irritation, or you've been dealing with hormone issues, you are especially going to want to listen to this ad. Every single

thing that I just listed has been linked to chemicals and household cleaners. One study found that regular use of cleaning sprays was associated with a lung function decline comparable to smoking 10 to 20 cigarettes per day. Synthetic fragrances are a well-documented migraine trigger and a number of studies have linked ingredients in common household cleaners to fertility issues, thyroid dysfunction, mood changes and more because of their impact on our endocrine systems. Luckily, you can just make one simple swap and you can eliminate all of those risk factors that you get from common household cleaners. Branch basics makes cleaning products that are plant and mineral based, completely fragrance-free, and made safe certified, meaning they are screened against

thousands of known and suspected harmful chemicals. And best of all, you literally use one product, but you mix it with different amounts of water to make your kitchen spray, your bathroom cleaner, your laundry detergent, your produce wash, your glass cleaner, literally everything. It does not smell. It does not give me headaches. It does not give me rashes or itchiness from clothes because you are. Remember, we're using this for laundry. We're using this for literally everything. And it works so well. Everything gets so clean. And I know that I'm protecting myself from all of those long-term issues too. And I'm protecting my kittens who are much tinier than me. So anything they're exposed to has a much greater impact on their little tiny systems. And they're running around on their little paws and then they're licking them. So I'm protecting them too. branch basics is now available at Target, Target.com, Amazon, and branchbasics.com, so it's easier than ever to make the switch. And if you want to grab their premium starter kit, you can use code Liz Moody for 15% off at branchbasics.com. That's code Liz Moody for 15% off the premium starter kit at branchbasics.com. And when they ask you where you heard about us, please make sure to mention the show.

That is branchbasics.com and use code Liz Moody. The number one rule of habits is to make the that you want easier and the things that you don't want harder. Yet so many of us want to eat healthier, but so few of us actually take the steps to make eating healthier easier. That's where Marley Spoon comes in. What I love about this company and what's different than all of the other companies out there that are doing like stuff in the same arena is that you can customize your choices based on the effort that you want to put in. So if you want them to send you ingredients, so you can make your own 20-minute meal and like get into your chef energy, they'll send it to you It'll all be in perfect portion so you'll eliminate waste.

Great, that's sorted. But they also have meals that you can just heat up. They have ready-made breakfast, which is always such a tough time of day to get a healthy meal in. They have grab and go snacks. Everything is made from farm fresh produce with high-quality proteins, and you can select by dietary preferences, including Mediterranean diet, which is the top diet that doctors on this podcast recommend. Also, things like gluten-free, dairy-free, low-sodium, anything that you need. The food is so good and it's so gourmet feeling like you feel like you're at a nice restaurant. We're talking like chicken milanese with a crunchy cucumber arugula salad or everything bagel salmon with truffle chive

potatoes. My favorite recent meal was the creamy lemon chicken tray bake. I had one of those moments where I looked at my plate and I was like, wait, I made this. And so quickly, like so easily. It's just so little effort for so much reward. Marley Spoon just makes eating well feel easy instead of stressful. And honestly, that is everything. This new year, fast track your way to eating well with Marley Spoon.

Head to marlyspoon.com slash offer slash Liz Moody for up to 25 free meals. That is right, up to 25 free meals with Marley Spoon. That is Marley Spoon.com slash offer slash Liz Moody. So remember to get the offer in there. Marley Spoon.com slash offer slash Liz Moody for up to 25 free meals. This is a fear I have is that I hear from people in my audience all the time. time who don't want to use AI for reasons we've talked about, reasons that we'll get into later on in the episode like environmental concerns. And I know we can't predict the future,

but I don't think anybody is disagreeing with the fact that AI is happening. And I think that is one of the things we know almost 100% for sure. Like the cat is out of the bag. The Pandora's box is open. And I really, really worried that people who are taking a personal stand are not going to change that fact, they're just going to be left behind. Right. You're right. Yeah. I don't, what do you, I mean, I'm sure you encounter people who have that feeling in your work. How do you talk to them? Do you address their, I mean, we're trying to address individual concerns in this episode, but like, yeah, I think I try to address those concerns. Yeah. And, you know, I have a weekly Q&A where people can send me their, sorry, biweekly, it's about every two weeks, send me their questions,

and I try to answer them. I appreciate you inviting me on the podcast. It's, it's things like this. where I think once they learn more about it, a lot of people who know more about AI are less afraid of it. Obviously, there's some people at the far end like Jeffrey Hinton, who is not that. He is more afraid of it, knowing more. But a lot of people, before they start using it, think that it's kind of something there, it's not. And like most things in life, if you're afraid of something and you don't know much about it, once you gather information about it, it becomes much less scary. And you can find that, just take that sentence,

and you'll start to notice it exists in your life with all types of things. The number one thing that I hear from people is people are concerned about the environmental impact of AI. The Washington Post just did a whole article feature thing that was like individual actions around AI matter. It's not just government or corporate onus. I think they said that one 100-word email from GPT, uses up a little bit more than a one-liter bottle of water. Most of the data centers are powered by fossil fuels. When you see people do like little memes with AI or isn't online,

if you look in the comment section, everybody is like, I cannot believe you like wasted water to do this. How do you think about the environmental implications of using AI? So there's a few thoughts around this. If you actually compare AI to other actions or foods that use energy, it is potentially less than you think. I know you just quoted that study, but I was doing some research recently, and they kind of have this number for a chat chit query where they say it uses 0.003 kilowatts, kilowat hours per query.

So then I was like, okay, how much is that? So let me do some research. How much energy does buying a liter of almond milk use? and that was 1.1 kilowatt hours. So let's say, okay, so there's four cups in a liter. So that's a little bit less than, what is it, 3.3. It's a little bit less than 0.3 kilowatt hours per cup of almond milk. That is 100 times.

0.003. And so you could have 100 queries for a couple of of almond milk. Okay, let's look at a burger. I like burgers. They're probably my favorite food. I love burgers. How much energy does that consume? The number I got was 2.5 kilowatt hours. That's about 800 chat GPT queries for one burger. So these numbers that we're hearing, if the research that I'm doing is true, kind of, I don't know. It's not as bad as it seems. and I do, I mean, we have heard this environmental, all the environmental impact of almond milk versus meat versus, you know, that's, we always kind of have this wave of that thing that's popular. It's bad for the environment. And I think this might just be one of those examples where if you actually look, it's not as bad as you think. And also, so now that's moving on to my next point, which is recent research came out that showed that because of new compute, increased efficiency, in the compute that runs AI models, the queries are using about a tenth of what they used before.

So it used to be about 10 times a Google search. Now it's a tenth of what it was before, which is, you know, maybe a normal Google search is now also a tenth of what it was before. But the Chatsubit query is now a tenth of what it was. So now it's potentially 0.003. Right? And so then that would be a thousand cups of almond milk if these numbers are checking out. There's that, although I posted that and then there's were some people who were upset because they said the source was questionable. Apparently the source is connected to open AI. So it's potentially questionable research, I understand.

However, that moves to my next point, which is that with science progressing in a way with AI and with AI being this polymath, drawing these correlations between different fields like Nikola Tesla, like Thomas Edison, we're going to find advancements. in science and energy that we would never be able to find before because of AI. And it could solve climate change. It could solve, you know, our carbon dioxide issues. It truly could fix everything. And so it's worth it, in my opinion, especially since the numbers don't seem to be as bad as we thought. Also, you know, the sun has more energy than we could ever need. And if we find a good way to harness the power of the sun, we could be set.

And you know what is really going to help us do that? AI. It's going to help us create the technology that we need to harness the power of the sun. There's also nuclear. One of my best friends, Isabel Bimicki. She teaches about nuclear. She actually has a book coming out called Rad Future. It might have just been released or about to be released.

And she goes in depth onto about nuclear and why it's deep. humanized unfairly and how it's, you know, a lot safer than a lot of people believe. And so there's that as an option as well. That's clean. Yeah, that's my take on the energy issue. So a lot of that is like in the future, this might get better in terms of both the types of energy we're using and also AI's potential to help us with our environmental crisis that we're in. What about right now? How do you think about if I want to use ChatyPT to make a meal plan for me this week? Am I personally part of the problem by doing that? How much do you think people need to be filtering the fun queries they're using these things for so that they can make their little bit of a dent

of difference in the energy consumption? Maybe don't eat a burger. You know, like, I don't know. I don't know. I'm guilty of this because I love to eat meat. So you're like, you're making other choices in your life. You're driving a car. I don't even drive. I have food. Yeah. Not you, but like the general person. Like we're making choices all day every day that are about our energy consumption, that are about our environmental impact. And this is one of those choices. But I also think for me, there's a benefit, like weighing the different risks and the benefits, a lot of different things that we do as human beings flying, huge risk to the environment. But we decide the benefit of going to Europe is worth that risk. And having these adventures and shaping our brain in that way is worth the risk to

the environment. So I think it's about for all of us, recognizing that everything sits along that we do because we're not living out in the forest and weighing those things accordingly. Exactly. It fits along with everything else. And if you making a meal plan helps you be a better, healthier person and it's really augmenting your life and you are increasing your quality of life because of it, do it. It is so worth it. You know, if you just are making thousands of images and you're just throwing them away after, maybe you don't need to do that. Yeah. You know? I think especially, it is interesting because it is like that is where the focus is right now. The thing that I worry about is it's like the plastic straws thing where we were like, oh my gosh, we're going to get rid of

plastic straws and the onus was so put on the consumer when actually the polluters are not the plastic straws. We need government oversight. We need corporate oversight and we need to make products differently in a way that's not just contributing massive amounts of plastic to the planet. It distracts from solving the real problem. That's like my concern around this type of thing is the individuals are going to be falling behind while really rich people get even richer because they're taking themselves out of the game because of these considerations. That makes me very, very nervous. A hundred percent. I think people should absolutely be using AI. Taking yourself out of the game is harming yourself. And once you see the benefits of AI, your life is going to be so much more

enhanced by it and, you know, your loved ones like your kids, don't harm yourself, you know, for what, right? Like the environmental impact, like we were looking at those numbers. It's, it's comparable to many other things, a hundred glasses of omelm milk for a query, right? Or a thousand. Which doesn't mean, I think, that you shouldn't be conscious. Yeah. But I think, again, right-sizing the decision amongst the many other decisions that we're making. single day and continuing to push for government legislation around the environment and around climate change and those types of things. I think all of those are important. It's a big equation and there's a lot of nuance in it. And I think we like the simplicity of being like AI is bad.

I'm not going to use AI. Totally. Sometimes I wonder if subconsciously, and this is something our brains can do, is if the resistance that some people are facing trying AI and they're attributing it to, you know, it's bad for the environment. I have this feeling around it. I don't want to do it. I wonder if some of that is an unconscious, aversion to learning the hard thing or a fear of that looks really difficult. And I don't even want to try because I don't want to fail. And that is very normal. That's a brain thing. And I think on many levels, it could be subconscious. But I wonder if what percentage of people who are fully averse to it are experiencing that at their subconscious level, just the fear of kind of trying it and failing.

But what I'll say to that is it's so much easier than you think to get started. These tools are so accessible now. And just try it. Like just download the app and use it. Which one? What app are we downloading? So I think a very accessible one that most people use is chat GPT. It has access to sources now, which is really important because, you know, hallucinations.

We talked about how AI models mess things up. It's important for you to have access to sources, especially if you, you're using it as kind of a browser to search the web and get a quick fact. So chat chattchipt is good in that sense. You should definitely be fact-checking things with the sources that they provide. But yeah, kind of use it as a browser. Use it instead of Google search. You could use Google Gemini instead. I actually love Gemini. I feel like Gemini came out the gate and it was really terrible. And then everybody's like, I'm not going to use that. And then they course correct it. And now I actually really, really like it. Yeah, it's interesting. Would it have been better for them to just wait?

Yes. But do they need the training data? I don't know. Oh, that's interesting. I have no idea. But is chat, like when people are talking about like Claude, Chatsy BT, Gemini, like, do you have a favorite, especially for people like the more laymen using it? Yeah. So I use them for different things, different ones for different things. I like Claude a lot. Claude is built by Anthropic. They do a lot of AI alignment research. They put a lot of effort into making sure that their AI models are safe and effective for people to use. they spend a lot of money doing that kind of research. The model also just has kind of a nice personality. So I use that for anything where I don't need it to be a fact. So if it's like writing or if I'm, you know, talking to it about interpersonal matters,

it has access to, you know, expert knowledge. For example, I made this video on AI for dating advice. And, you know, it's aware of attachment styles and all these kinds of expert therapeutic modalities. But it doesn't need to be like 100% fact-checked, you know? So I use Claude for things where I don't need facts. So if I need actual facts, I'll use chat chabit or perplexity because they can search the web and they provide sources for you with their statements that they output. So you always want to fact-check those because even if it provides a source, it's not

necessarily true. It does mess up. It does provide a source and then it's still hallucinated. It's still made something up and it's in its output to you. So you do want to cross-check. So I use those models for fact-based tasks. Can you say in a nutshell what AI hallucinations are? Yes.

We talked a little bit about how large language models are probability machines. They output text according to the patterns that they found in their training data. Because they don't have access to actual facts and they're just word generators, they can go down a path that is made up. So one cause for this might be, I like to use a. this example. Say you have the input text, the cat in the. The model for the next word is going to take that and then provide a probability distribution for all the possible next words. Perhaps hat, cat in the hat, has a high probability distribution and cat in the bathtub has a relatively low

probability. But because it's sampling, a low probability event can occur and it could choose bathtub. And then suddenly it's going down the cat in the bathtub route. And so now the next word, the probability of it being, you know, a word like Dr. Seuss or the next 10 words, that's completely shifted now. It's chosen this route. It's kind of like I like to liken it to those essays or stories. Like the choose your own adventure. Or like the stories we would write in school where you would write one word or one sentence and then pass it to the next person. And then you just take what you've gotten and you're like, all right, here I am, and you go with it. And one word choice can completely change the direction of the entire thing. And it's kind of like that. Okay, it's chosen bathtub.

Cat in the hat is out of the way, like is out of the question now. So that is one major reason that hallucinations happen is because it's a probability sampling machine. That's so interesting. I always thought it was like trying to give me the answer that it thought that I wanted. That's also. Okay. That's also, there's many reasons why it could be training data issues. So for example, like training on Reddit, it might have false data, right? So there's that. But then it could also be, it's trying to echo you. So these models are trained, often explicitly trained, to kind of validate you and mirror you. Like if I think about it for my work where if I'm like, tell me why raw milk is good for me, it'll be like, oh, here's why raw milk is good for you. And if you're like,

tell me why raw milk is bad for me, it'll be like, here's why raw milk is bad for you. What you're getting on is leading questions. A hundred percent. 100% they are meant to kind of echo you. And if you think about it, your input sets the probability distribution for the next words, right? In the same way, cat and the hat, you're the first line of like the story you're writing. Yes. So the way you write that first line is dictating the next line, the next line, the next line. Exactly. And so you have to be really careful, you know, something I talked about recently was these psychosis issues with people using AI because these AI models are trained to either trained or, you know, put in a system prompt, which is like kind of a hidden

prompt by the company that they send to the model before it gets to you. So either through explicit training or through prompt, they're kind of often told to validate the user. And so users are then going down these deep rabbit holes and potentially, you know, entering psychosis from being validated down a kind of conspiracy that is not real. And so those kinds of things are really scary. Is there any way to tell if the AI is hallucinating or do we just need to be fact-checking every single thing that it's spitting out? I would fact-check every single thing it spits out. If you need it to be true, do it. Especially things like case numbers and things. If you're looking up certain sources, it could so easily make that up. That is like textbook hallucination.

I do think it's funny. Sam Altman said that people have like this wild trust in AI. And he's like, we never said that it was accurate, but people are acting like it's definitely accurate. Totally. He's the founder of Open AI for anybody who's listening. So he's like, the AI guy. And he's like, I think it's the funniest thing about AI to me is that people trust it so much. And we did not tell them to trust it so much. Totally. I mean, people will ask for resources for something. Like, oh, give me a source that says, you know, this thing. Like if you want some kind of science-backed, you know about this. Like, show me your sources that say that going out in the sun is good for me or whatever.

It will often, if you're not using one that can actually search sources, it will just make up a case number. It'll make up a source. And it will present it to you and you won't have any idea. And it's actually gotten some people into trouble. Does AI store our personal information? Like, how careful should we be about telling AI stuff about us that we wouldn't want somebody to be able to find out later?

There are ways that you can use AI that you can keep it on your own machine. They're becoming increasingly available where you can download a model onto your computer and run it locally so you're not sending anything over to company's servers. I would assume that if you are using a model like OpenAI or Claude or any model that is owned by a company, I would assume that your data is stored there and kept there and everything you've said is there. So, I mean, use it at your own discretion, like with everything else, right? All your Google searches are tracked. So just be aware. Okay. What do you think of using AI for therapy? I think it can be really great. AI has access to all different types of therapeutic modalities. It's extremely unbiased. So I think I have had lovely therapists in the past. But being human, they approach therapy from their own language. of their own life experience. And so something I'm going through, they will project their life experience onto it or the experience of other patients that they've seen. We're all fallible as humans.

We only know what we've been trained on. And the AI model has been trained on everything. And we've been trained on a very small subset of everything. There's a very important aspect of human-to-human interaction, which we need with therapists and friends. I think having friends is really important. Having therapist is really important. Someone that you can kind of trust and look to. You do need it beyond AI. Like, AI don't think truly suffices in some moderate cases, perhaps, which is great because we also have a shortage of therapists. So it's great. In the moderate cases, it's available to 24-7. It's not going to judge you. So I think it's really great in a lot of ways. It's very low cost. I mean, you could use a free model, right? Even to just get some skills.

Like, hey, I'm having a panic attack. Like, what do I do? You know? know, it can really help with the low-hanging fruit. That's what I feel like it's going to help. I come from a family of a zillion therapist, and even they are pretty optimistic about AI because they think that it's going to help with the educational component so that in session they can focus more on the relational component. So it actually will free up the time to do even deeper work. I love that.

Right. It's available 24-7, give you a quick tool, help you understand yourself better. Yeah, like it can analyze things you've told it about yourself or interact. actions and help you understand yourself better, you know. How is AI going to impact our health or our lifespan in the future? I'm so excited about this. It's going to be, I mean, the health implications of AI are are astronomically wonderful. I mean, we're going to see cures for many diseases, potentially cancer. We're going to see treatments, customized treatments for different types of people. I think we will be able to

expand our lifespans, you know, the longevity research is pretty amazing. I think, yeah, we'll live a lot longer, we'll be a lot healthier, we'll be able to cure illness. Why? What's an example of like why AI would help with any of that? AI is able to see very, very nuanced features of things that we can't see ourselves. So by analyzing mass amounts of data, they will be able to discern like, oh, that one molecule is associated with all of these rats that survived this thing, this tiny and little thing. And so because it's able to analyze massive amounts of data and then see patterns, we will draw insights about diagnosis, about treatment. And again, it goes past large language models. This is what I was saying at the beginning about AI is so much more than chat models.

Being able to analyze data is kind of the heart of AI, and that is what we will be able to enable with AI with health. How far out are we talking? Like, do you think in a year we're going to have cancer treatments that we don't have now? Do you think in five years, we'll be able to live until we're like 150? Like, what are we talking about? So this comes down to education and why I'm doing what I'm doing. We just need AI in the hands of more people, researchers, people learn. learning how to use AI for their specific use case and applying it to their problem. So it could be a

year if we had people working on cancer research using AI. But they just need to learn how to use it and how to design an experiment that would help them use AI, right? Like analyzing mass amounts of data, gathering that data. So it's learning about how AI could be used for your specific use case or for your research and getting that information out to those doing the research. or enabling other people who, you know, are interested in the topic and want to do research themselves, like not necessarily in a lab, but this tool is available for everyone. Are there any specific health advancements that you've, like, seen coming up or seen already have happened that you're really excited about?

Yeah, there was recently one with glioblastoma. So glioblastoma is the most deadly brain cancer. I think if you get glioblastoma, I think your likelihood of surviving five years is about 10%. So it's very, very grim. And they found that using AI at USC, they were able to discover a treatment that raised survivability by about 75%. Things like that, we will continue to see more and more. Again, it's just getting it into the hands of these brilliant scientists. I'm also really excited about getting into the hands of doctors because right now, with the way that insurance is structured,

doctors only get very small amounts of time with patients. and there's only so much you can do in terms of looking into their whole plan and their history and all of that. And so I feel like AI could enable them to make the most of that like 15 minute they have with every single patient and create personalized health plans that actually would move the needle for people. And also hopefully to like communicate because I feel like right now like this doctor knows this about me, this doctor knows this about me. And I'm like, come on. Like what year we living in? I completely agree. I feel like there should be a centralized database with everything. I don't know. I don't understand how that doesn't exist. And also, I mean, that would enable, again,

massive amounts of data analytics. What we would have then is we would have people with their backgrounds, so their socio-demographic backgrounds, their health metrics, you know, height, weight, whatever, any diseases that they've accumulated, their historical, you know, family lineage of disease, and then health outcomes. Yeah, their geographic location. Like, we're like, okay, so in this cluster, we're already seeing, like, there's clusters of areas where you're much more likely to have neurodegenerative diseases because environmental exposures when we have. have that information, we can really press the government to make the changes that they need to take care of people, et cetera, et cetera, yeah. If we had this centralized database, we could have

unbelievable diagnosis, prediction, like disease prediction, disease treatment, customized treatment for different people. Again, it's AI analyzes data and draws these patterns and insights. That would enable extraordinary health advancements. And this is already happening in what you're seeing. And what I'm saying, it is already happening, but we need more people to move the needle and to learn AI and apply it to their use cases. And I agree with you. I really think there should be a centralized health database. I don't know why there isn't. You said there's all these different ways to use these tools where people will begin to see real enhancements in their own lives. Can you live us with like maybe two or three very specific ways that somebody could start using these tools to see real benefits in their lives today? I would start by using AI as your new browser.

So whether you want to use Google Gemini, you want to use ChatGBTGBT, BT, use AI as your browser. Anytime you are about to, you're thinking, oh, I'll Google that, or I'll start to do some research on that. Or, you know, I want to, anytime you have a question or you're planning on doing some research to develop something, use AI. Frame it as a human question to an expert. Whether it's personal, whether it's with work, what have you, health, anything. Imagine you have this expert with you 24-7 that knows everything about everything. That's where I would start, because then you'll become familiar with its capabilities, and you'll be able to see how you find it most useful.

Another fun use case and exciting use case for AI is if you have a business idea, and you're like, you know what, I have this dream of I had a friend who wanted to make a nightgown business. It's like, how do you get from zero to one? I have this dream, but how do I even get there? AI can help you break that down into steps. So you can tell it, hey, I want to create a nightgown business. What are the steps I would need to do for that? And it can break it down into step-by-step procedure for you.

And then it can do research on each of those steps. So it can actually help you do those steps. And then it can break it down into timelines. You know, create a timeline for me over the next eight weeks. Break it into month segments. Break it into, you know, the next year. And it can help you with all these tasks of structuring these abstract ideas that you have and making it a reality. And also I like to use it if I have an abstract

thought and I can't figure out how to word it properly. Say like, you know, I'll just talk to it. You know, it has voice mode and I'll say, I have this idea. You know, what I'm trying to say is this. This is what I'm thinking. Can you help me make this into a concise, kind of pithy statement or paragraph? And it will help me do that and it will gather my thoughts and put it into something cohesive. Also, it'll make bullet points for me. So if I have an idea of, you know, something and it's just abstract, it will be able to take that cloud of thoughts and make bullet points about it and maybe make an outline. And it's just amazing. What about for health? Like somebody told me that they said, use the principles of the Liz Moody podcast to structure like a meal plan and a

workout plan for me for a week. And it did like an incredible job of creating like finding recipes that use the principles that we talk about on the podcast and incorporating mental health and things like that. I love that. Are there any other like fun uses for health that you've heard for like the normal person, not the researchers, curing cancer? Yeah, well, I would use it for kind of all your health questions, you know, if there's something serious. And it will tell you if it thinks you should see a doctor. But I have a friend, her mom, is kind of a hypochondriac. and she would go to the doctor and she would ask these questions and the doctor would completely discredit her and, you know, just be like, okay, like, bye. But she's sitting there then with all this anxiety and she wants to ask someone something.

AI can actually take in your specific facts about yourself. When you're asking those kinds of questions to AI on, say, a Google browser, it's very hard to make it customizable. You're like, oh, I have this symptom and I have this symptom. And there's a special thing about myself. You know, I'm prone to this thing. It's like impossible. to put all those things together and do that kind of research in Google. AI will take all of your customizable information, run it through its brain, and then give you its own opinion on whatever your specific question is. So it really listens. It considers all of the data that you give it. Oh, another one is uploading your blood test results. Get some really good insights from that. All the medical questions that you have,

it's available to use. And these models are caught really good. Harper, thank you so much for joining us today. This was absolutely wonderful. If people want more from you, can you tell us where they can find you? Well, first off, thank you for having me. It's been such a joy. And my channels are Instagram, YouTube, TikTok, Harper Carroll, AI, 2Rs, 2Ls, and Carol, like, carrolls put together. Harper Carol AI. And I have a course called The 10 Days of AI Basics, both in short form on my TikTok Instagram and YouTube shorts, and then longer course, longer form videos.

with those 10 days on my YouTube. And I'm also coming out with a self-paced course that takes you through my videos kind of from start to finish as an AI course, because I realized recently I have all these videos and I actually have a cohesive AI course, but how do you navigate it? So I'm coming out with this kind of self-paced navigation. So if you're interested in that, then find me there. I also have a newsletter that you can sign up at Harpercaryl.a.com and I do a weekly news update with AI, just all the most important AI news because it's moving fast. It's moving so fast,

and I think it's really easy to get overwhelmed. This was something that I wanted to say is that I think following people on social media or news sites are really useful to keep up with AI, so I recommend doing that. However, there's a lot of hype. Things are constantly coming out. It's really easy to get overwhelmed

and to have brain overload for a lot of things that actually won't matter in two weeks. Well, and knowing that. And the journalist in me feels compelled to say these people are looking for titles that are going to make people click, which is usually going to be stoking heightened emotions for people. Right. And so, yeah, you'll see like research papers. You'll see tools that come out that are suddenly all changing everything. And then in two weeks, like no one talks about the tool. So one thing I try to do is just distill it down to the most important things once a week. And then on my own channels, you'll notice I don't talk about a lot of news on my channels. And that's because I think a lot of news, it just doesn't pass my check of is. is worth talking about. So just being aware of what kind of you're following and also staying up to date though with things, but not getting overwhelmed. Totally. Yeah. Well, thank you so much, Harper. Thank you, Liz. Thank you for listening. Make sure that you're following the podcast because on Monday,

I'm going to be sharing 39 health and happiness hacks that I have learned in 39 years. These are going to be really quick, actionable steps that will make a huge difference in your life. And if you are following on Apple, on Spotify, on YouTube, that will show up right in your feed. If you loved this episode or you feel like it contains valuable information that people should know, take a really quick second, just copy and paste a link and send it to a friend or a family member or a coworker, maybe a parent who's worried about their kid's future with AI, maybe a friend who's worried about the environmental impact and you want to start a real conversation about that. It's an amazing way to help the people in your life to spark really interesting conversations. And it is hands down the best way

to support the podcast. I'm always so grateful and so appreciative that you guys send so many links to everybody that you know in your life. We're also going to be sharing the best prompts and hacks to get the most out of AI, like specifically what you can put in to get the best results for like meal plans or workouts or financial plans or other life hacks. That's going to be on our substack. So go to Liz Moody.substack.com, put your email in, and then that will come right to your inbox when it comes out. We're able to bring you this show completely for free, thanks to the amazing brands that we partner with. If you want to try any of the products that you heard about in this episode, or we also have codes for so many other products from like grocery shopping to skincare to

supplements, you can find all of our current discount codes at Liz Moody.com slash codes. Okay, I love you, and I will see you on the next episode of the Liz Moody podcast. Oh, just one more thing. It's the legal language. This podcast is presented solely for educational and entertainment purposes. It is not intended as a substitute for the advice of a physician, a psychotherapist, or any other qualified professional.