← The 1000 Hours Outside Podcast
The 1000 Hours Outside Podcast
1KHO 714: Do We Have Free Will in an Algorithmic World? | Kartik Hosanagar, A Human's Guide to Machine Intelligence
00:58:37
Loading summary
Transcript69 lines
- [00:00]
Singer
Oh, it's a beautiful world Ain't nothing on screen that's ever gonna be this view oh, it's a beautiful world and I just want to share with I just want to share with you this beautiful world Such a beautiful.
- [00:35]
Peloton Sponsor
This episode is brought to you by Peloton Break through the busiest time of year with the brand new Peloton Cross Training Tread Plus. Powered by Peloton iq. With real time guidance and endless ways to move, you can personalize your workouts and train with confidence, helping you reach your goals in less time. Let yourself run, lift, sculpt, push and go. Explore the new peloton cross training tread +@1peloton.com welcome to the 1000 Hours Outside podcast.
- [01:02]
Jenny Urchin
My name is Jenny Urchin, the founder of 1000 Hours Outside and I have a guest today that I'm so excited to talk about. He wrote a book called How Algorithms Are Shaping Our Lives and How We Can Stay well. Okay, this is the subtitle. The book is called A Human's Guide to Machine Intelligence. The subtitle is How Algorithms Are Shaping Our Lives and How We Can Stay In Control. The author kartikosaniger is here. Welcome, Kartik.
- [01:25]
Kartik Hosanagar
Thanks so much for having me. Ginny, excited to be here.
- [01:28]
Jenny Urchin
This is a fantastic topic. And what is so intriguing to me about your book, which came out in 2019, which means you wrote it before then, and you're talking in this book about AI chatbots that were in China in 2014, is that this is not new. And I think a lot of parents, especially since ChatGPT came out, are kind of like, oh, I'm dabbling in this new technology, but it's not new. And you're explaining this in your book. I mean, this is going back, it's really going back to the 19, I think, 60s or, you know, they're talking about it way back then, artificial intelligence. But a lot of these technologies have been around at this point for a while. You teach a course, it's called Enabling Technologies. What's going on in tech? You're still writing about this on your substack and I'll put the link in the show notes. It's hosannagar substack.com it's called creative Intelligence. But can you talk, can you talk to the parent who thinks this just came out a year or two ago?
- [02:24]
Kartik Hosanagar
Yes. So I've been working in this broad space of AI and what it means for business and society for almost 20 years. And initially we saw it in small ways in our lives for Example, when you would go to Amazon and you'd see recommendations, people who bought this also bought that, or on Netflix, hey, maybe you should be watching this show or YouTube where it's recommending other videos to watch. We used to see early versions of this, and of course, the big one was online advertising. All the ads we see on Google, Instagram, all of these were driven by AI algorithms behind the scenes. So it was starting to be there. But roughly around 2016, 2017, I felt like we were now going into a stage of AI where it wasn't these few isolated pockets, but we would have AI all around our lives. And I felt like I shouldn't be only focusing on the C suite in companies and what AI means for them, but it was something that was going to be profoundly impactful for society at large, the way the Internet has been. And I need to take the message to civil society, lay people, you know, parents, and help them understand, hey, this is about to change in big ways. And it's going to be running our lives. And if we are not careful, it will be running our lives for sure. And we lose control. And so the book was all about recognizing we lose control and how we can take measures in advance.
- [04:05]
Jenny Urchin
Kartik, it's an incredible book. And what an interesting thing that you say because it came out in 2019. So you said because you're involved in this world, you're noticing 10 years ago already that this is already all around our lives. I think a lot of us feel like it just popped up. So in 2014, you're talking about this chat bop. I don't know how to pronounce the name of it.
- [04:26]
Kartik Hosanagar
Sha Ice chow.
- [04:28]
Jenny Urchin
Say it one more time.
- [04:29]
Kartik Hosanagar
Sh ice.
- [04:31]
Jenny Urchin
Sha ice. So this chatbot in 2014 in China already had 40 million followers and friends using WeChat and Webo, maybe it's called the two most popular social apps in China. And they talk about how they love her jokes. And you have a great story in there where you're like, it's like a friend, but it then you find out that it's not a friend, it's this joke chatbot. And now all of a sudden, this is sort of everywhere here in the United States as well. A lot of these chat bots are targeted at teenagers and young adults. So the point is, is this technology has been around for a while at this point, and it is always, I think, rapidly increasing. Can you talk about what changed in terms of that? Computers. I love the omelet analogy that you use. So you like people look at computers and they think, oh, it's just going to enact this set of directions I give it. Here's how you make an omelet. Step one, step two, step three. But it changed because it has started to learn. Like, almost like a child learns.
- [05:37]
Kartik Hosanagar
Yeah, Yeah. I think how a child learns is a great analogy. So what I mentioned in my book is that the way computer software used to be coded was that there would be developers, human programmers, who would come up with a sequence of steps that the software has to follow. It's like an omelette recipe. First you break your eggs, you beat the eggs, then you add this much milk. Then you do this, and you add vegetables. Whatever the steps are, you have a very clear set of steps. If something goes wrong, you kind of know which step something went wrong in. You know, a software application crashes. You know what's going on. If it's behaving in an unusual way, you kind of know, oh, okay, I messed up in this step. But where things have gone in recent years is that computer scientists have recognized that, look, if we want to write software that can figure out our tax payments, you can do it with a sequence of steps. But if you want software to drive cars or to diagnose diseases, you cannot write the steps. For example, Ginny, you and I can spend hours here together, and you can give me a set of rules on how to identify your mom's face in a photograph. We can write down all these rules, and that software will fail, probably with the very first photograph, because there is so much that you know that you are not able to articulate and that you may not even know that you know, which is helpful in recognizing your mom's face. And so the computer scientists realized, look, we have to stop writing the steps. Instead, we need to teach computers to learn on their own. Let's give them a lot of data. Instead of teaching them all these steps, why don't we give them photos of a million people, all kinds of faces? And in there, we will have at least 500 photos of your mom, and we'll clearly label it, and the software will now learn the patterns on its own, the same way we learn. Sometimes our parents give us the rules that we are supposed to follow, but many times we just observe, do things, try stuff, and then we figure out, oh, you know what? This is how the world works. And we figure out those mental models. So that's the analogy I provided. And coming back to what you started with, which is the Shao AI story that Chad bought in China, What I started that story with was about this young adult in China who has a daily conversation with Shaw Ice, almost thinks of Shahweiss as another person and has a, quote, unquote, a relationship with Shaw Ice in the sense that this is a confidant, a friend, a substitute for having real relationships and the messiness of real relationships. And I started the book with that and talked about how many people have said I love you to Shaw Ice. But interestingly, now we're starting to see after ChatGPT and others, there have been stories of people who are having, you know, almost like romantic relationships with their chatbot or people who, you know, are believing whatever the chatbot is saying and rather would spend time with the chat system than have a real friend or a real relationship. So a lot of what I started the book with is now playing out in big ways.
- [09:08]
Jenny Urchin
Absolutely. You talk about these unanticipated consequences, unanticipated negative outcomes that happen in addition to.
- [09:15]
Sponsor Voice
The desired positive results.
- [09:17]
Jenny Urchin
Because obviously people are like, well, there's a lot of positive things here if this computer can learn. But you're. The book is. Is really talking about, like, how free are we actually? You know, are we pretty shackled? It's really eye opening. Can you talk then with the computers that are learning? Because you say this, most people, and this is such a huge statement, cardic, most people know very little about a technology that has and will have a very large impact on their lives. Not only their lives, but the lives of their children and their grandchildren. These machines that learn, they've come a long way in the last decade. They no longer merely follow a program sequence of instructions. They can take in data, this huge amount of data, learn completely new sequences of steps, and generate more sophisticated versions of. Of themselves. So I've actually not heard of this game. I've, I've read about it maybe once or twice. You know, at the beginning they were like, can the computer beat someone in jeopardy? Can the. Which is. Which is even a little bit more complicated than chess because you have to figure out how much you're going to wager and all of that. But then there's this even more complicated game called Go, which there are as many moves as there are atoms in the universe or something like that. So can you talk about how that shocked, like, what the computer could do with Go, how that shocked people?
- [10:38]
Kartik Hosanagar
Yeah, yeah. So, you know, when, when people started building these kinds of AI systems, games was. Games were a good place to begin because they have structured rules and you can see and you can measure how good is the system and all of that stuff. And initially it was like, hey, can we have a system that can play chess? And then eventually we had a system that beat Gary Kasparov, who was the best chess player at the time in the late 80s. Then we had a system that could beat Jeopardy. Champions. That also was amazing improvement, advancement. But with all of these systems, you're using brute force computing power, meaning that, hey, with chess, I'm going to throw in massive amount of processors and memory and just brute force computing power to beat a human being with Jeopardy. Will arm the system with all of the world's data and so on. But Go is interesting because as you mentioned, there's just way too many potential combinations, like as many atoms in the universe. And so you cannot just bring brute force computing power and say, let me try every possible move and figure out what and then figure out how you will respond with every possible potential move, then how I can respond with every possible move and so on and see how this will all play out and say, oh, you know what, if I do this, I, I win in 90 out of 100 games. But you cannot do that with Go because there's just too many combinations. You need something that human beings have, which is creativity. We can look at something and we can say, I'm not even going to consider a bunch of options. And I will, you know, when you ask me a question, there may be a million possibilities, but I'm not going to explore the million possibilities because I know through creativity and through experience, here are the ones to focus on. That's what you would need for an AI to be good at GO to beat the world's best GO player. And that's what Google did when they built the system called AlphaGo, which was a go playing AI and it beat Lee Sedol, who was the world champion in GO at that time. And the way they beat it, the way AlphaGo beat the human champion at Google, is by getting creative. And in fact, one of the craziest things that happened in the game, in one of the games was that GO played the AI system played a move, and the world champion is just staring at the board and trying to figure out what just happened here. And all the commentators were commenting live. They said that this is not a human move. They said this move has come out of nowhere. And eventually Lisa Sido lost and later commented, this is real creativity because this is not a move I've seen before. And the AI system could come up with that.
- [13:37]
Jenny Urchin
Wow, wow. Book came out in 2019 and it's 2026. So we're even many more years into this. You say it was trained on a database. AlphaGo programs its own rules using machine learning. And you talk about it's very similar to children. You say, how often do children routinely surprise us? The capacity to surprise is core to a child's normal development. It also makes them admittedly difficult to handle at times. The same is true of these new algorithms. So it's trained based off this database of more than 30 million past moves. It plays millions of games against itself, and it plays more than the number of games played by the entire human race since the creation of this game goes. So this is part of the reason why data is so important. All of these machine learning algorithms are built off of large, large amounts of data. Is that why they're trying to build all these data centers?
- [14:35]
Kartik Hosanagar
Absolutely. So if you look at these systems, one, they're built on large amounts of data. To train them, you need first the ability to store and process this massive amounts of data. You need lots of components, computing power to do that, and you need large data centers to be able to do that. And then, of course, if they build really good AI, it's going to be running the Internet, it's going to be running our lives and all of that stuff. So you also need these data centers to be able to do all of that. So, you know, we are already Starting to use ChatGPT more and more for a lot of things. You need a large data center so that the world's questions can go to ChatGPT and it can answer that. But it won't just be chatgpt in the future. Retail websites will be run out of this. Software will be run out of this, Finance and trading will be run out of this. Everything will be run out of this. And so that intelligence that is running the world and commerce needs its fuel. Our fuel is food and its fuel is energy and the computing power. So the data centers are basically lots of computers that are always running a lot of electricity.
- [15:49]
Jenny Urchin
Okay, and you say big data, with all its messiness, is here to stay. It's hard to overstate the value of data in modern machine learning. More data means more examples for algorithms to learn from, and more exact algorithm or more examples lead to unprecedented increase in accuracy. So the more data, the more accurate and maybe they become creative. But you talk in this book then about what does this mean for society, and that's what you're talking about on your sub stack. What does it mean for businesses. What does it mean for us as people?
- [16:22]
Sponsor Voice
February can be a really tough stretch for homeschool families. The days are shorter, routines feel heavier and motivation can start to dip for kids and parents alike. And this is where a tool like IXL can be such a helpful support during the mid year stretch. One of the biggest gifts of homeschooling is flexibility, meeting kids exactly where they are right now. IXL is designed to do just that. Whether a child needs to reinforce foundational skills, work through a tricky concept, or move ahead with confidence, it adapts to their level without pressure or comparison. What stands out to me is how simple and steady it makes learning feel. Everything is organized by grade and subject, so you're not just juggling resources or wondering what to do next. Kids get real time feedback and clear explanations, and parents get progress reports that show exactly how learning is unfolding. No guesswork. IXL covers math, language arts, science and social studies from Pre K through 12th grade, and it's built to help families keep momentum going even through winter, so confidence can grow heading into spring. Make an impact on your child's learning. Get IXL now and 1000 Hours Outside listeners can get an exclusive 20% off IXL membership when they sign up today at ixl.com 1000hours. Visit ixl.com 1000hours to get the most effective learning program out there at the best price. The new year always makes me want to reset our spaces not perfectly, but intentionally. I want our home to support our routines again instead of working against them. And that's where Wayfair has been such a help. Wayfair really is a one stop shop for everything your home needs, from bedding and bath basics to storage to those little decor touches that make a space feel finished. Whether you're refreshing kids rooms, organizing your kitchen for easier weeknight dinners, or rethinking a work from home setup, it's all there. One of my favorite updates this season has actually been for our podcast studio. We get so many books sent from guests and they were stacking up everywhere. We ordered new shelving from Wayfair and and it completely changed the space. Clean line, simple style and finally a place for all those books to live. It looks great, it's functional and honestly it just makes the room feel calmer and more put together. That's what I love about Wayfair. The selection is huge, the styles are easy to browse and you can find pieces that work for your home and your budget without overthinking it. Get organized, refreshed and back on track this year for way less, head to Wayfair.com right now to shop all things home. That's W-A-Y-F-A-I-R.com Wayfair Every style, every home. Hey, friends, Jinny here. Quince is all about elevated essentials that feel effortless. And that's honestly what I want my wardrobe to be at this stage of life. Pieces that layer well, mix easily, and stand the test of time. Not trendy, not fussy. Just really well made clothes you actually wear. Quince truly nails the staples. They're 100% organic. Cotton sweaters are soft and breathable. Their premium denim has just enough stretch to stay comfortable all day. And those cotton cashmere blends are perfect for the changing seasons. These are the kind of basics that quietly become the backbone of your closet. One of the things I appreciate most is how Quint works. They partner directly with safe, ethical factories and cut out the middleman. So you're not paying for brand markup, just genuinely high quality clothing. And you can feel it. The European linen, the organic cotton, the stitching, the fit, everything is built to hold up season after season. These are pieces you reach for over and over because they simply work. I can't tell you how much I'm loving my cashmere sweater from Quince. It's become a staple in my wardrobe already, and I honestly can't wait to give one to a friend. It's that good. Refresh your wardrobe with quince. Go to quince.com outside for free shipping on your order and 365 day returns, now available in Canada too. That's Q-U-I-N-C-E.com outside to get free shipping and 365 day returns. Quince.com outside and one of the.
- [19:53]
Jenny Urchin
So you just talked about environmental. I mean, that's one of the impacts, right? So these data centers that are starting to pop up and people are pushing back, but what they're talking about are the environmental impacts on that. So we could talk about that too. But I want to talk about, and this is something I think you're talking about with students, what's going on with tech? These technologies are shaping entire industries as well as daily lives. Do we have free will in an algorithmic world? This is a huge question, Kartik. So can you tell people about the things that they. You brought it up earlier. You brought them Netflix, but the statistics are pretty shocking. Amazon recommendation algorithms have the biggest influence on which books people are reading. I mean, that's huge. That is so huge. 99% of possible alternatives are excluded. When you go to search for things so you're not even shown the breadth of, of what might be out there. I mean, the statistics are pretty staggering. So when you're talking with students, and I'm sure that they love this conversation. If we have free will in an algorithmic world, what are some things that come up?
- [21:02]
Kartik Hosanagar
Yeah, yeah, so, so I guess before I talk about the impact they're having, the quick headline I'll share is that a lot of us believe we have free will. A lot of us are thinking we're making our own choices. The reality is we are not. And let me just peel the layers of the onion when I say that. So first of all, these systems are adding amazing amount of value to us. If I want to buy a book and I want to figure out what to read over the holidays or something like that, I might go into Amazon and I might search for a book, it'll provide other recommendations, I'll click on a few links, figure out reviews, ratings, and then finally buy one. If I want to watch a show show, I'll go to Netflix and it will already curate a set of videos for me to watch. Same thing with YouTube and so on. The statistics are quite staggering. At Netflix, over 80% of the time people spend on Netflix is influenced by algorithmic recommendations. So in other words, 80% of what we watch is determined by an AI algorithm that puts it in front of us on the main screen. At YouTube, it's over 90% of the time we spend on YouTube is influenced by that. It's going to be similar for Instagram and TikTok. I would in fact argue it's probably 99% at these places because they just figure out next video to play, the next video to play, the next video to play, and so on. And they're just showing us stuff. And you go in, somebody shared a video or a Instagram post or a TikTok reel. You go in there to watch that two minute video, but before you know it, you've spent an hour and almost all of it was curated by the algorithm. So they are deciding what we read, what we watch, what we buy, all of this, even Amazon recommendations at what we buy, and so on. And at the end of the day, who we are is mostly choices we make around what we read, what we watch, what kind of information we consume and what opinions we form as a result, what products we buy, how we present ourselves to society, all of those decisions are a big part of who we are. And those are currently all shaped by algorithms. And in fact, there have been studies that Facebook will not actively advertise it today. But researchers at Facebook, which is now called Meta, did studies on interesting things about things like people's mood, people's decision to vote in elections.
- [23:44]
Singer
And.
- [23:45]
Kartik Hosanagar
And they found that the algorithm's decision to show certain posts in our feed decides whether we choose to vote or not, decides whether, you know, our mood is positive or not. In particular, they had a news. An experiment where they changed the news feed, where some people were sure more, what they call hard news, more political news in their feed than others. The people who saw more political feeds, they indicated that they voted more often in the elections than those who were not shown more of the hard news. So these algorithms can influence whether we feel motivated to participate in elections or not. They can also motivate mood. For example, they found when they showed more negative news stories to people on Facebook, they found those people became more negative in their own posts on Facebook. And of course, right now you see the big divide in the US around political lines. It is my assertion that a big chunk of it, I'm not saying all of it, but a big chunk of it has to do with just how the newsfeed algorithms, if we watch something, will show you more and more and more of that and will take us from having a small reaction to having an extreme, extreme reaction to that. And before you know it, you know, we're all polarized and divided. So, yes, we don't have free will. These systems are influencing choices we make, and I do not want to discard the value they provide. Yeah, none of us want to go back to a world where every time you needed some information, you have to, you know, spend three hours doing the research, picking up lots of books in a library, and, and figuring out what to do or, or things like that. You can just go to Google and ask your question and get the answer. So that's brilliant. But having zero friction and having AI run everything, there are cost to that.
- [25:52]
Jenny Urchin
Yeah. And that's what you talk about in the book. These unintended, negative, unanticipated negative outcomes, there's another one you talk about, too. So you basically, you say this. Many of us clearly do not have quite the freedom of choice that we believe that we do. So talking about that, the 80%. That is wild. 80% of your viewing hours streamed on Netflix are originated from automated recommendations from the algorithm. By some estimates, nearly 35% of all sales on Amazon originate from these algorithmic recommendations. So you talk about how this is decreasing the amount of products that we consume. The diversity of the products that we consume is becoming less. You also talk about it in terms of dating apps.
- [26:37]
Kartik Hosanagar
Yeah.
- [26:38]
Jenny Urchin
Can you tell people what's going on there?
- [26:41]
Kartik Hosanagar
Yeah, look, it used to be the case that people actually dated in the past based on the physical world and our social, real social networks. Meaning our friend said, hey, you should meet so and so, because that person is really. I think you'll get along at things like that. Right. So we found people, or you went to college and you met lots of people in college and so on, and that's how people found their partners. I think technology has made us all very inward looking, meaning not inward in our minds, but like into our devices and not outward looking, not spending time outside with people and so on. And so, yes, people aren't finding their mates as easily. So where would you go find them? You'd go to dating apps, like, you know, and there's a bunch of them. I don't want to name anyone in particular, but most young people today find not just their dates, but they're, you know, your, your partners, your marriage, your partners on these dating apps. And ultimately these apps are algorithms that take certain information about you, that look at clicks you make and look at right swipes and left swipes you make and then learn your preferences and then figure out who should we recommend. And so it's no surprise that increasingly algorithms are also deciding who we marry and live with as well.
- [28:15]
Jenny Urchin
You wrote in the book that they will ignore, you know, they'll ignore what you say you want. In fact, you say some of them, they stop even asking because they're going to just use what they observe. The, you know, in immense amount of data that can be garnished, I guess, through any type of online platform. And so they're using their formula to figure out who to match you to, even if you didn't say you wanted that thing or like that thing, but you wrote that it works.
- [28:48]
Kartik Hosanagar
Yes. So all of these systems have moved in the direction of we don't even want to ask you what you want. Because the problem is that people behave in ways that are not consistent with what they say they want. We could ask you what they want, but these systems are trying to optimize for engagement and to keep you there. And if they follow what you say you want, you're not as engaged, so they decide to ignore you. So a few examples of that. Facebook tried a system where it tried to give people more control about what they want to see in their newsfeed, you might say, hey, I want to see more educational stuff. I want to learn more about culture and so on. But if your behavior is such that when they show you stuff about culture and education, you don't click on them, you don't spend time on them, but when they show you these wild political stories or something that's kind of very sleazy or whatever, you actually click on it. Because our human minds are such that we gravitate towards extreme things, and you click on them, they realize, oh, I want to keep this person here on the app. I should ignore what they said and instead show them based on their behavior. And so the same thing is true not just for. I just talked about, like, Facebook feed. It's also true for dating apps. If you say you're looking for, you know, I'll just make something up. I'm married, so I haven't used these apps. But let's say somebody says, I'm looking for a date who's educated, who has an interest in, let's say, the arts and can talk intelligently about paintings and all of this stuff. But when you show that profile and you don't click on it, and instead you show somebody dressed provocatively and you just click on that, then they know, you know what? I want to keep this person on the app. I'm going to ignore everything this person is saying, and instead I will model their behavior. And it could take us down a path that we know we don't want to go there, but that's what we. That's how we behave. And the AI and the algorithms take us right there.
- [30:54]
Jenny Urchin
It is so interesting to learn about. You wrote people used to place personal ads in the actual newspaper. The job of finding a match was left to what the reader left to the reader of the ad. And now these companies are just ignoring, at least partially, what you say you want. And it works better. It says, in this case, this approach works. So you say, are the algorithms serving our desires? These are like all the deep questions, Kartik, Is this because you're on there to find somebody? So you're like, well, is it serving your desire or is it manipulating? And what. What do you think?
- [31:33]
Kartik Hosanagar
Look, I think there's. There's definitely manipulation going on for sure. And when I say that, what I mean is that the incentives of who owns the AI and who is rolling out the AI is not aligned with the user. The incentives of the owners of these AI and of the tech companies is to keep you engaged. There is to have you spend as much time on the system and so on. And so that's what they're trying to optimize for. Does it cater to your desires? Yes or no? Because I think it caters to our desires in the sense that if our behavior reflects our desire, it caters to it too. But sometimes we all behave in ways that are in short term. We behave in ways that are inconsistent with our long term goals for our lives. And so sometimes these systems take our long term, short term actions and optimize around it. So they cater to our short term desires, but not to our long term goals.
- [32:45]
Jenny Urchin
Wow. You say unless you're a tech luddite, algorithms are silently rearranging your life. In this brave new world, many of our choices are in fact predestined. And all the seemingly small effects that algorithms have on our decisions add up to a transformative impact on our lives. So the thing that is really unique about what you do is you then take all of this information and say, what does it mean for us? So what does it mean for us as individuals and as parents? Or maybe specifically, what are you doing? Does, does awareness do a lot to help? Or will we never have freedom of choice? Like, what are the implications in your life? Let's say.
- [33:30]
Kartik Hosanagar
Yeah, look, I think the sad reality is that a lot of us will unfortunately have no free will and control over this. And not because it cannot be done, but because the odds are stacked against us. But many of us, and I don't mean a small 1%, 2%, but probably 30, 40% of us, will recognize the problem, take it seriously, take action, and therefore be in control of their lives. And I want this number to go from 2% today to 10% tomorrow to 20% in a year, 30%. And you know, if we can even get it to 50% plus, that'll be awesome. That's my goal with the book and with my work right now and so on. And I do believe it's possible for us to take control. And before we get into how we take control, one other piece I want to add here is that what's at stake is of course, our free will and the choices we make, but it's also the choices others make about us as well, so not just the choices we make for ourselves. So, for example, if you are a parent, very soon your kids will be entering a job market where they will be applying for jobs. And on the other side, there is not a human being reading your job application or your resume. There'll be an AI system reading all of these resumes and deciding who to invite for an interview. By the way, Amazon tried such a system around the time I wrote my book. That system had a gender bias because that system was trained on their data on which job applicants get invited for interviews, which interviewed candidates get the job offer, which people who get job offers actually get promoted. And if there's a gender bias, it will show up in the AI. It shows up in the AI in the following sense. Even if you keep all of the details in a person's resume the same and just change their name and interests, it has this gender bias and it won't like this person. And if you and I had the same resume, then it wouldn't like you Jenny over me, even though the resume was identical. And so then Amazon phased that out and they obviously over time have made changes to their system. There are examples of biases in AI that's used in courtrooms that are trying to predict whether a defendant is likely to reoffend, which ultimately influences sentencing decisions. It's much more than all of this. But decisions other make about us is also influenced by this. So what's at stake is super high. And we have to take charge. We have to take charge first in our own lives and make changes to make sure we're in control. And then we have to also make sure that when others are making decisions about us, like approving loan decisions and those and job applications and so on, that there are some regulations or some, if not regulations, some safeguards to protect us.
- [36:39]
Jenny Urchin
We have a family friend who lost his job and has put out more than 1,000 job applications maybe in the last six months or so. And that's what he said. He said everyone's. They're written by AI. Your resume, your cover letter, it's all written by a. It's all being processed by AI. Everyone's sending them through AI and that places are just getting a glut of response too much to even sift through. And it's just such a mess. And I don't think he's found anything. You know, it's really changing.
- [37:12]
Sponsor Voice
Some seasons I really want to eat healthier, but I just don't have the time or mental energy to plan it all out. Between busy days and full schedules, meal planning is usually the first thing to fall apart. That's what changed for us when we started using Hungry Root. It made healthy eating feel doable again. Hungryroot works like a personal nutrition coach and grocery shopper. In one, you tell them your taste preferences and health goals and they take care of the planning and shopping for you. The recipes are simple, the ingredients are high quality, and everything is tailored specifically to you. What I've noticed is how easy it is to stay consistent whether you're trying to eat cleaner, get more protein, or just avoid ingredients that don't make you feel your best Hungry Root removes the guesswork, and the more you use it, the better it gets at learning what you actually love. There are thousands of recipe options every week, no junk ingredients and responsibly sourced meat and seafood. And if you're trying to keep healthy habits going past January, this really helps you stick with it. Right now. Take advantage of this exclusive offer. Get 40% off your first box plus a free item in every box for life. Go to hungryroot.com 1000hours and use code 1000hours. That's hungryroot.com 1000hours code 1000hours there's something about the start of a new year that makes me want our home to feel more supportive, like it's actually helping us stay organized and and focused instead of adding to the chaos. That's why I keep coming back to Wayfair. Wayfair truly has everything in one place, from bedding and bath essentials to storage solutions to easy updates for kids rooms, living spaces and work from home setups. Whether you're trying to simplify mornings, reset routines, or just make your space feel a little more peaceful, it's all there. One of the best updates we made recently was in our podcast studio. We received so many books from guests and they slowly had taken over every surface. We ordered shelving from Wayfair and it was one of those changes that instantly made the space feel lighter and more intentional. The shelves fit the room perfectly, the style is clean and simple, and finally, everything has a place. It sounds small, but it made a huge difference in how the space feels when we walk in. That's what I love about Wayfair. The huge selection, the range of styles and price points, and how easy it is to find pieces that actually work for real life. If you're ready to get organized, refreshed and back on track this new year, for way less, head to Wayfair.com right now to shop all things home. That's W-A-Y-F-A-I-R.com Wayfair Every Style, Every Home this episode is sponsored by BetterHelp. February has a funny way of putting relationships under a microscope. Everywhere you look, there are messages about romance and connection and having it all figured out and whether you're married, dating, single, or somewhere in between, it can start to feel like you're the only one still finding your way. But the truth is, we're all still figuring it out. Therapy is one of those places you can slow down and get honest about what's weighing on your relationships. Unspoken expectations, pressure to get it right. Old patterns that quietly drain connection. Whether you're working through things on your own or together as a couple, therapy can bring clarity to what feels heavy and create space for things to feel brighter again. That's why I appreciate better help. Their therapists are fully licensed in the US and follow a strict code of conduct. They also make it easy to get started by matching you with a therapist through a short questionnaire. And if the match isn't right, you can switch anytime. BetterHelp is the world's largest online therapy program with over 30, 000 therapists and more than 6 million people served globally. And it works. Earning an average of 4.9 out of 5 star rating across 1.7 million client reviews. You don't have to have everything figured out to make meaningful progress. Sign up and get 10 off@betterhelp.com 1000 hours. That's B E T T E R H E L P.
- [40:38]
Jenny Urchin
So when you talk about taking charge of our own lives, you know, obviously the others making decisions is a little trickier. But it's a call to action, to be involved and to know, be in the know of what's going on. But how do we take charge in some ways in our own lives?
- [40:56]
Kartik Hosanagar
Yeah. I think first of all, it starts with awareness.
- [41:00]
Jenny Urchin
Yeah.
- [41:01]
Kartik Hosanagar
And once you're aware of what's going on, you can be more mindful when it's happening and take action. So I'll give you a few examples, few changes that I've implemented in my life. And to be honest, I wrote the book, but it's a struggle for me too. I'm not suggesting I have all the answers. So one of those is it's important to add some friction in your life. And all of technology is built around removing friction. How do we make things seamless? How do we make sure that stuff gets done super fast? I think it's important to add some friction some places. And so what that means when it comes to my use of AI, I think there are certain. I'm actually worried if I'm already. If I'm always using AI for everything, I will lose certain core skills that are important for me. And there is evidence by the way of this happening that Even doctors within six months of using AI that's doing diagnosis, they are losing some of their diagnostic skills if they become heavily reliant on AI pilots. There's a good example. The FAA right now suggests that all pilots do at least a certain minimum amount of manual flying because excessive reliance on autopilot has reduced people's ability to fly the plane on their own. So I actually do a few things, which is I make sure that I have some time for what I call analog work, which is doing the work without digital technology and AI support. I actually do a lot of human first rounds. So before just saying, hey, AI, do this, I will do a first round basic paths before starting to lean on AI to do that. So these are certain things I do in my life. And so in my work when I use AI, I don't want to lose certain core skills. So I add in certain friction to do that. In my personal life, the friction tends to be things like my phone. I've set up a Mode where after 9pm it turns into black and white mode. And it's a reminder if I turn on my phone, everything is in black and white and I have like, oh, I shouldn't be using this. Certainly certain apps turn off after some time and there's only so many minutes of an app per day. Things like that. For my children. I made a mistake with my son where I gave him a phone when he was in middle school with my daughter. I know I will delay it until high school. And so again, important to make sure that we are all not living in a screen, not all living in an automated world. I think those are the kinds of checks in my personal life that I add. And of course happy to go into other areas, but I'll pause here.
- [43:52]
Jenny Urchin
No, it's good. I really like this man and I've talked about him before on the show named Neil Postman, who is was a futurist and he wrote some really good books. And he said in one of his books, no, he wrote the book like the Disappearance of Childhood and Amusing Ourselves to Death. And they were written in the 80s. So once again, it's just so eye opening to read your book and to realize, okay, this has been brewing for a very long time. So in the 80s he was talking about that we're amusing ourselves to death like on the big TV with the tubes and you hardly had any channels. Anyway, he had said something in his book that no technology is particularly dangerous and it might not be this exact words. But he said if the users understand what the Dangers are, he said, to ask is to break the spell. And, you know, in some ways you're like, is that too simple? But to your point, in your book, this Human's Guide to To Machine Intelligence, you wrote, most people know very little about a technology that has and will have a very large impact on their lives. So I do think it is important to understand what's going on and to have an idea of the background and to ask the questions what might be the unanticipated consequences. One of the things that you brought up that was shocking, once again, this book from 2019 is that these chat bots, they start communicating with each other, and then they make up their own language, so nobody even really totally knows what's going on underneath the surface. And you say, should AI systems have the right to invent their own language? Because then you don't even know what they're talking about. Can you talk about what happened?
- [45:34]
Kartik Hosanagar
Yeah. Yeah. So that was an AI experiment at Facebook where they were trying to see if you could have a world where AI is interacting with AI, transacting with AI. We just talked about how people are creating resumes using AI. AI is evaluating resumes. So in the future, there may be a system where anyone who's looking for a job uses their AI. That applies to all these jobs. Somebody else, all the recruiters use AI, and the AI will not just analyze resumes. They will have a computer conversation to, oh, here's what I find interesting about your man, which is, if my AI is representing me, here's what I find interesting about this person. Can you elaborate on this? And then my AI will elaborate on it. And so the AI are talking to each other. And by the way, this will happen in commerce, too, where we. I mean, it's already starting to happen, but in very small ways, where in the future you might just be saying, hey, I want to buy a gift for a friend. My budget is $200. I know what their interests are. I could spend two hours searching through everything, or I can have my AI and say, find a gift for my friend Ginny. The budget is $200. Ginny likes travel. She likes reading a lot. Figure it out. And the AI does go searches everywhere. Figures it out, and so on. Right? And so the merchants also will have AI on their site, where my AI goes to their website and says, hey, what do you have? I'm looking for a gift for somebody who loves reading and travel. The budget is 200. Tell me what you have. And then my AI might ask their AI do you have any discounts or specials right now? So the AI are talking to each other. Now, what happened in this experiment was, at some point, these AI are learning. They figure out shorthand. Why are we communicating through this very inefficient way of communicating called English. There are other ways to communicate faster. And they start to slowly, English is changing shorthand words, and before you know it, you can't even understand what is one AI telling the other. They're understanding each other. They figured out a language. They sealed the deal, they negotiated a transaction was complete. But I cannot even look at the transcript of that and figure out what's going on. So that's what happened in that one experiment. And so we need guardrails for these kinds of things. Right? Because, again, driverless cars. Future, there will be lots of driverless cars communicating with each other. My driverless car will inform another driverless car, I'm about to go to the left lane. So if you're planning to speed up, please slow down. And things like that. And so they're communicating. So AI will be talking to AI. And, yeah, on the language stuff, they will invent their own communication pathways as well.
- [48:20]
Jenny Urchin
I mean, you. I mean, you spell it out right here. It appears that the bots had developed a shorthand to improve communication efficiency. In hindsight, it's not surprising because you say this is what people do. They have, you know, if you ever talk to anybody in the medical world, they have all sorts of acronyms that they use.
- [48:36]
Kartik Hosanagar
Right, Right.
- [48:37]
Jenny Urchin
So they have these shorthand things, and they tried to understand the transcript between these two AIs that are talking to each other, and they couldn't understand it. So then there's no transparency. You don't even really know what's going on. So you brought up the driverless cars, and that could maybe take us into what? Those are already here, at least to some degree. You talked in the book about one where there was an accident, but then it was saying that that person wasn't using it correctly. It was in a Tesla, and they were using it where they shouldn't have been. It should have been on the highway. But you talk about how people. So it's a little bit about, like, what's coming. Yeah, what's coming. And also this concept of we trust algorithms sometimes, but not others. And yet the, you know, 1.5 million lives in the next 50 years could be saved because accidents are usually human error. And maybe the algorithm would do a better job. I don't know. It's a lot to grapple with, yeah, that's something that's coming and then how do we know, you know, do we trust that or not use it.
- [49:50]
Kartik Hosanagar
Right, right. Yeah, there's a lot to unpack there. I think I've shared a lot here about what can go wrong or why we should be worried about AI as well. But my message is not let's become Luddites and let's move to a world where there's no technology, no AI and so on. I did mention friction and I think that's good to have, good to have friction, good to have some analog time and so on. But at the same time, I'm not advocating a world where we're moving back to, oh, no computers, no AI, none of that. Instead, in fact, the reason I think AI is important is that when we talk about these problems of AI, it's useful to talk about what's the alternative. The alternative is to go back to humans making all the decisions and humans are flawed themselves. If I just said, hey, an AI recruiter has biases, where did that come from? That came from human decisions that were made in the past. So clearly humans are flawed. If we're talking about a Tesla having a crash, what's the alternative? Human drivers. And there are millions of people dying because of crashes where driven by human errors. So the alternative is very flawed too, arguably more flawed than AI. The opportunity here is that it's very hard to detect human biases and correct human biases. You know, somebody's drunk driving, you know, you can tell them stuff, but they might not take it seriously, they might not act on it. But the AI system, you could measure, you could detect problems, you could correct it. So it's mostly, I think my message is about safeguards. It's not about going back to no AI, no technology, it's about having the right safeguards. So you want driverless cars. Great, let's do it. Let's have driverless cars, let's have extensive testing, let's have new standards for driverless cars. Let's do that. If you're going to use AI in recruiting, let's have standards and governance around that. If you're going to use AI in high stakes decisions like even lending, let's have governance around that. Because I think at the end of the day these systems are black boxes. So let's make sure that they're not complete black boxes. If a system rejects a loan application or a job application, there should be a way to then figure out AI explainability, get actual explanations from AI about why those Decisions were made because we know these systems hallucinate. Let's make sure the decision to reject an application was not because of a hallucination and that it was based on some real assessment by the AI. So those are the kinds of checks, like have real explanations, have a human being in the loop, measure these systems, evaluate these systems, test those systems. Those are the kinds of checks we need to. When we are talking about not AI in. In our personal lives, but AI that's rolled out by companies that affect our.
- [52:58]
Jenny Urchin
Lives or in society. Like who's driving? Who's driving next to you? Is it an AI car? Or is it your. Is it a person? So you have in the book, people can check this out, A Human's Guide to Machine Intelligence. And then they can also check out your substack because you're always writing about these things. But the algorithmic bill of rights. So you wrote, the scientists who design these systems have to take on the mantle to fix them. We have to set boundaries where technology limits aren't setting them for us, like our. Like them coming up with their own language. So a lot of really good ideas here. I love to wrap up with this last concept of. And bringing it back to the beginning of the chatbot, of the chat bot, friend, of the chatbot relationship.
- [53:41]
Kartik Hosanagar
Yeah.
- [53:41]
Jenny Urchin
Because this is an. A new ish thing. Although it's not completely, but I think for a lot of parents, it's new parents are shocked. Every once in a while I'll talk about this. I'll talk at some parenting conferences about the chatbot friends and how many people are using them. And people are shocked, cardic like, they're like jaws drop. Because they just don't even really know what's out there, what's happening. And I feel like. I know a little bit, but probably have just scratched the surface. But you say we have a tendency to anthromorph. Anthropomorphite. What's the word?
- [54:15]
Kartik Hosanagar
Anthropomorphize. I also always struggle with that word.
- [54:20]
Jenny Urchin
Yeah, it's too long. Anthropomorphize machines. There's a tendency to do this. And a lot of times people don't even know that they're communicating with a bot, and especially our children. So you're talking about. There's a lot of good here too. But what are some of the unanticipated negative outcomes that could happen along with.
- [54:41]
Sponsor Voice
These chat bots that we should be aware of?
- [54:44]
Kartik Hosanagar
Yeah, look, I think today a lot of accounts on social media are AI accounts. We don't realize them, you know, whether it's on LinkedIn, Insta, wherever. There are AI accounts that are, you know, being run by people that will chat with your kids, that will chat with us. They're meant to defraud. Some of them are meant for amusement, Some of them will reveal their AI bots, some of them will not, and so on. So there are people getting tricked, a lot of people. Recently, a friend of mine lost a bunch of money because an AI bot was communicating with him, pretending to be me, and then convinced him to wire some money, and somehow he felt trapped. And then, you know, we're navigating that issue. But unfortunately, these, they have access to my photos, my information online voice, your voice, very soon, my voice and video. So, in fact, you might be talking to a person live on a video call with my voice, with my face, but it's not me, but it's a bot trained on me. So our kids need to be aware of all of this because there could be people who are trying to manipulate them into making really poor decisions. And they need to be aware when they're talking to an account. First of all, don't engage with accounts you don't know. But even if an account seems to be somebody you know, the face matches, the communication style matches, have a way to figure out outside of the system to verify you're talking to who you think you're talking to. I think those are extremely important choices we have to make. And of course, the other thing now you have to worry about is deep fakes as well. Not just in terms of who you're talking to, but your own photos. You know, there are examples, unfortunately, of young teenage girls whose photos have been taken and people have manipulated them and put them in, like, in pornographic settings and so on, and posted them online, which is all a problem. So you have to be careful about who you release your photos to. What are your privacy settings? Can your photos be seen by somebody else? Do you know who you're allowing to see access to your photos on your social media accounts? All of those are extremely important. And then I also like the mission that you have, Ginny, which is about, like the thousand outdoor hours, which is not about immediate threats of AI, but it's also really to help children reframe this whole thing that, look, your goal is not to seek constant entertainment, amusement that is actually bad for you. And AI systems are all optimized for engagement, speed, scale, and things like that. But childhood should be about slowness, boredom, friction, physical world play, and so on. And there's a selfish reason parents should push their kids for that. Because if you're looking at everything AI is automating today and what it's going to do to the job marketplace, and you want our kids to have AI proof skills, the skills that will ultimately matter. I think outdoor play and the physical world interactions build those skills like social interaction. Social. Social negotiation, emotional regulation, risk assessment, judgment. Those are things that humans will be doing. And AI will do the basic stuff really well. And we can't compete with AI on documentation, on summarizing documents and those kinds of things. So the skills children will need in the next stage in the job marketplace are going to come from the kinds of things that require maybe interacting with real human beings in the outdoor world and so on.
- [58:54]
Jenny Urchin
It's such good advice. It's so good, it's counterintuitive because you think that you have to speed up, speed up, speed up because technology is going so fast. But it really is the slowing down that's allowing us to differentiate ourselves. Is there so much change happening all the time? You know, I know you're writing on your sub stack so you're like on the front line of this is it just. Has it even rapidly changed since 2019?
- [59:17]
Kartik Hosanagar
Oh, let alone since 2019? I can tell you I am in the business of AI and I struggle to keep up because every day there's a brand new announcement. Every day these systems are changing. It's exponential change that's going on right now. Every month there's a new capability. You would think you are. Here's what January 2026 looks like. And you can say, okay, based on that, I'll project what January 2027 will. And trust me, you will be shocked with where we'll be in a year from now.
- [59:50]
Jenny Urchin
Wow. Okay. Well, people can follow along on your substack. It's called Creative Intelligence. The intersection of AI, creativity, mindfulness and entrepreneurship. Fantastic information. And I'll link to that so people can make sure that they're following along. The book is also phenomenal. A guide, A Human's Guide to Machine Intelligence. How Algorithms. It's a really cool cover too you how algorithms are shaping our lives and how we can stay in control. A lot to consider. A lot of conversation topics within your own family. Like this would be one you talk about at the dinner table for a very long time. Really interesting things to think about. Where are we at in terms of our own free will in this algorithmic world? Kartik, what an honor. Thank you so much for Spending this time with us, we always end our show with the same question. What's a favorite memory from your child? Childhood.
- [60:38]
Sponsor Voice
That was outside.
- [60:40]
Kartik Hosanagar
Okay, well, before I share that memory, Ginny, first of all, thank you for having me. Thanks for your curiosity and research into the book and everything else. My fa. You said what is my favorite memory of from childhood? That's it, right?
- [60:54]
Jenny Urchin
Outside. But outside.
- [60:56]
Kartik Hosanagar
Outside. Okay, well, my favorite memories from childhood generally is when I used to spend time with my grandmom. And so my parents, obviously, both of them were working, and so I'd return from school, go to my grandma's place, and sometimes we would go. You know, it's like basic stuff. We would go shopping and I used to find it so boring at the time, and I just keep begging her, and she would bribe me with, I'll get you this soda or this drink or this chocolate or something like that. But that was fun. And my cousins were. Would come over every summer from all these places to my grandma's place. So the entire summer was us cousins playing sport and, you know, all these other games together in the summer. So I think those summer holidays with my cousins at my grandma's place and all the nearby parks, that was my favorite childhood memory.
- [61:50]
Jenny Urchin
Oh, it's wonderful. So wonderful. It's what kids should still be doing today. Kardik, thank you so much for being here.
- [61:56]
Kartik Hosanagar
It was a great pleasure. Thanks for all your research. This was a fantastic conversation. We packed a lot into an hour, so thanks for that.
- [62:02]
Jenny Urchin
We sure did. Thanks, Kartik.
- [62:04]
Kartik Hosanagar
Thanks for having me. Chilli.
- [62:09]
Singer
Get outside Open your eyes Feel that.
- [62:13]
Kartik Hosanagar
Sunshine Kissing your skin Throw your worries out to the wind.
- [62:20]
Singer
Climb some trees Skin your knees Feel that grass on your feet again get out there and.
- [62:28]
Kartik Hosanagar
Take it in.
- [62:31]
Singer
Oh, it's a beautiful world Ain't nothing on the screen Is ever going to be this view oh, it's a beautiful and I just want to share with I just want to share with you this beautiful world Such a beautiful world.