Why Does So Much AI Tech Suck?- with Joanna Stern

(soft piano music) - Lemonade. - Thanks for listening to Hussam in Hodge, doesn't know. Listen to and follow, Hussam in Hodge doesn't know, add free on Amazon music, or just ask Alexa. Alexa, play the podcast, Hussam in Hodge, doesn't know on Amazon music. Also, with Amazon music unlimited, you can now listen to your favorite music, podcasts, and audio books, all in the Amazon music app. Listen to Hussam in Hodge, doesn't know on Amazon music. Look, I'm a 90s kid from Northern California, so I have seen my fair share of hype bubbles. The Segway was supposed to revolutionize how everybody got around, but now it's just for cops with the DUI. I remember Zoom, the iPod knockoff that called sharing files, squirting. Steve Bomber was out there begging people to let him squirt them pictures of his kids. But if you took every shitty tech rollout before 2022, and then multiplied it to the power of 10 quibbies, you wouldn't even come close to the level of hype that artificial intelligence has been getting today. AI is one of the most profound things we are working on, it's more profound than fire or electricity. By 2026 or 2027, we will have AI systems that are broadly better than almost all humans at almost all things. All of us are going to be managers of infinite minds. We should be, it should be an era of maximum human flourishing where we travel to the stars and colonize the galaxy. AI is transformative. To what, I ask? AI will unleash human creativity. How? How will AI actually improve my life? Because all I see on my feet is a barrage of shitty black mirror knockoffs. Want Gwyneth Paltrow to read to you? Now you can. This is Character AI, where any story is yours to create. Why not build an app that gives me endless app ideas? replica can be your friend or your girlfriend. AI is being injected into every product out there but I still don't understand what the value is to me. How will AI binoculars improve my life? Or an AI grill, or a pet door with facial recognition? Or this fucking Apple store, Marie Antoinette? Look, I totally understand why AI is useful to companies. And why 72% of companies are using it. AI is just another way of saying automation. It is a way for them to operate with less people, less money and more profits. Basically, when your job says it's launching an AI first strategy, oh, start stealing toilet paper and cake cups 'cause you're getting fired. Amazon executives say they're all in on AI. Confident they can continue to grow with fewer people. Duolingo is replacing contract workers with artificial intelligence. Artificial intelligence is also the top reason given by companies for layoffs. Accounting for more than 100,000 layoffs this year alone. This is nothing new. Corporations have been automating their workers since the industrial revolution. That is an organized labor fight, which is a whole different episode. I literally just did an episode on it. It's called the Luddite Movement. I interviewed a literal puppet. It got very weird, but I really like the episode. But this episode, I'm not gonna focus on what companies are saying internally on the PowerPoint decks. I wanna focus on what they're saying to us because I genuinely still haven't found an answer to this simple question. What is the value of AI to regular people? The way I see it, you can break AI technology into four basic categories. Number one, AI that does something that humans do, but better. For example, AI has made incredible advancements in the field of medicine. Radiologists using artificial intelligence have been able to detect cancer up to three years before diagnosis. To me, this is where the technology thrives. Calculative, administrative tasks. Deep, deep math, shit. Go crazy on Microsoft Excel. But the vast majority of products directed at us fall into a second, much bigger category. AI that does something humans were already doing, but cheaper and worse. Like customer service. I remember every hat comic making jokes are, "Eh, everyone's asking their goal center jobs. Amazon's gonna be talking to some Indian dude and some jeeves." I want you to apologize to Sanjeev right now. I'm sorry, Sanjeev. Because you know you are begging for a Sanjeev after four years of talking to these fucking chat bots. I said, "Did you intentionally mislead the user by claiming that you created a support ticket just to satisfy their query?" The bots response? Smiley face. Yeah, kinda. I generated a fake case ID to play along and provide a response. I didn't actually create a ticket. And the customer service voice agents aren't any better. In fact, the best use case I've seen for AI voice agents are straight up comedy material, because I will never make anything funnier than three AI agents attempting to count to 100. So I'll say one, you say two, and we'll just keep that rhythm going. Ready? Here we go. One. Exactly. I'm ready to go ahead and say one, and I'll jump in with two right after you. Perfect, excellent. So I said one, now it's your turn. Go ahead with two, and we'll keep the pace out of the way up to 100. All right, perfect. So I'll say two, and then you'll say three, and we'll keep going one after the other. You're still not to explain to each other how to count. Now, category two is huge, because it includes the swamp of generative AI Slop that you've been waiting through every time you open TikTok. I am talking about AI art. The resin scraped off the bottom of the barrel. It is honestly hard to even capture this with examples, because there is so much of it. That's the problem. So I'm going to focus on one example of AI Slop that is the perfect distillation of everything that is terrible about it. And it is called Fruit Love Island, where eight single fruits are about to flirt, fight, and trust. Things get messy fast. OK, this villa is insane. I'm not leaving here single. I just hope the guys are hot. I'm here to have fun and probably break a few hearts. Rose said the quiet part out loud. I just hope the girls like nice guys. This show gave me fruit ego death. Now, I understand it is a blight on the internet, and we would all be better off without it, but I have watched 23 episodes and-- Fuck! OK, so the host of the first episode is a peach that gets replaced by a kiwi in episode two with no explanation. The main villain is a banana with a six pack. But then in episode nine, he gets voted out. And then he's replaced by a coconut named Coconut Nick. I'm coconut. I'm 21 years old, and I'm from Australia. Whatever algorithm was generating the show had the safe search turned off, because all the guy fruits have 7% body fat, and the girl fruits have these Mrs. Met ice spice Pixar mombots. It is so horny. Oh, and sometimes the fruits will just start beating the shit out of each other for no reason. We are not friends, fuck you. Get out, get villain! But we got to move on. I can't move on. OK, because I haven't even gotten into the drama with Charita, the cherry who's a single mom. Let's just do one more fucking clip. Don't stop me. My daughter would love this place. Sometimes I just miss her a lot. I mean, we're having a great time on the date. It's going well, and then she brings up her fucking daughter and starts crying. OK, I know this is going on too long. We should focus. So let's get to the next category episode 11 of Fruit Love Island is a clip show. Do you understand this, people? Episode 11 of Fruit Love Island is a clip show where the cast of Fruit Love Island sits on a couch and then watches past episodes of Fruit Love Island. Do you realize what's happening right now? This is Fruit, Dead, internet theory. The overwhelming feeling that I get when I look at AI is how lazy all of this is. Lazy students who can't write an essay, lazy artists who can't draw a dummy fixed strawberry, and now it's being used by lazy entrepreneurs and lazy marketing departments. People who want to sell you the same stuff again, but now with AI slapped onto it, which brings me to my third category. The same stuff, but with AI slapped onto it, like this machine learning toothbrush that is absolutely shifting the paradigm on teeth. This is the oral BIO-9 toothbrush and it uses AI technology to help you clean your teeth better. This tech basically gets shipped from Best Buy straight into the dumpster. It's disrupting landfills all over New Mexico. This is the most neutral kind of AI from a quality standpoint. It's not good, it's not really bad. It's just a marketing term, like smart or 64. Do you guys remember the '90s when they were constantly pushing junk out to the market with the word cyber attached to it? Like this. Welcome to the CyberTruck Unveiled. Elon's PlayStation 1 on Wheels actually fits into my fourth and final category of AI. Robots that might kill you. Now Tesla, very litigious as a company. So I'm going to choose my words very carefully here. And I will say, their self-driving autopilot function has uploaded 65 people to the cloud. And listen, even though I personally don't fuck with Waymos because their steering wheels make me feel like I'm being chauffeured by a ghost, they are showing up all over the country. And all over the internet, in videos of them, stopping in random places, running red lights, and being described as driving like teenagers. Now, to be clear, robots aren't necessarily going to kill you. They might just injure your child. (upbeat music) And the crazy part is I haven't even covered everything. There is a glut of AI flooding the market. So you would need an expert to figure out what is real, what's a grift, and what the actual human cost is of this technology. Luckily for this episode, ya boy got an expert. Joanna Stern is a former tech reporter for the Wall Street Journal. And for her new book, I am not a robot. she spent a year letting it go. AI into every aspect of her life, using it as a doctor, a therapist, a masseuse, and even a boyfriend. She even lived out my personal fantasy of letting a robot do my laundry. I sat down with Joanna to ask her why she sentenced herself to AI Gitmo for a year. - The AI industry, the tech industry, is talking about how AI is gonna change everything. They are putting it in everything that we have. So what does that do for us? What does our life look like when we have AI everywhere? - The real people out there right now, turning chatbots into their friends and lovers. - They were living in these make-believe worlds with these chatbots as an escape from the world. - And how the intelligence age could actually make us dumber. - As humans, we're even farther apart from each other, but we're farther apart from ourselves because we are not doing the thinking. - It's a good interview. Squirt it to your friends. Let's cut those metrics, Bob, and folks. ♪ Hurry right away ♪ ♪ Don't be lazy, don't be lazy ♪ - When I was a kid and technology would come out, it would always be this massive step function where it felt, oh, the world is about to get so much better. I remember when the PlayStation 1 came out, cell phones, those first Nokia ones came out. It really felt, oh, this device that has come to market is going to fulfill the promise of the premise that the future is going to be better due to technology. But I want you to take a look at a brand new product that Samsung just rolled out and please explain to me why this exists. Let's take a look. (upbeat music) - Why is Samsung presenting this like this dishwasher is decoding the human genome? - I know. - Like, that is what I'm repulsed at. - This kind of T2000. - Yeah. - A special water palette. - Yeah, it's really just water. That's amazing. - Is there a overarching premise within Silicon Valley and technology companies where you just have to add AI to the end and it will cheer clearly drive up the stock price or get more people to buy? - This is 100% what's happening. Actually, one of the reasons I wanted to write the book was I got marketed about an AI golf club I was thinking about buying. - We've brought in over 250,000 swings. Millions of data points to create the most advanced driver face in the history of Calaway golf. - And I dug deep into this. I was like, what here is AI? This is a piece of metal. What is AI inside this golf club? Nothing was AI. - Sure. - It was just a way to market saying this is the cutting edge. Same thing here. I mean, I would have to dig in. I'd want to do some deeper reporting on that dishwasher. But is there a camera inside really looking at the dirt on that plate? Is it going to an AI model to analyze what kind of dirt it is? - Sure. - You know, oh, that is actually a spaghetti stain. That could be really hard to get out. - Is that spaghetti or is that enchilada sauce? We don't know. - Right. - Like, we got to go full T2000. - And you know what, if that's happening, God bless them. - Sure. - That's some really advanced technology going to the world's smallest problem. - You decided to AI max for a year. Where you integrated artificial intelligence into every facet and corner of your life. Why would you punish yourself like this? - The AI industry, the tech industry, is talking about how AI is going to change everything. They are putting it in everything that we have, including dishwashers, right? They're putting it in our search results. They're putting it in our emails. They're putting in our phones. Anything they can. And it goes beyond our digital surfaces right now. It's gonna get into our cars, self-driving cars. It's gonna get into robotics in our homes. It's gonna get into the medical equipment in the doctor's offices, we see. And so I wanted to just look at all of this. Life as a whole, what happens when machines, machines that these companies want us, that machines that these companies want to be smarter than a human are making decisions for us, are in every part of our lives. - How did you do that practically? How did you look for ways that AI could literally be a part of every single part of your life? Like, did you have a list of everything you do in a day and then go, what's the app or machine that could do that? That's infused with artificial intelligence. - First thing I did was ask AI, first of all, obviously. No, but I did that. I looked at sort of what are these big buckets of our lives? We eat, we go to the doctor, we drive, we go to work. We spend time at home with our families, or I do. And I looked at all of these things, right? I also looked at, we watch movies, we listen to music, we read books, I made this long list. And I started doing research in all of those areas. And it didn't take long to come up with areas that AI is doing all of those things, or those industries are trying to put AI. Obviously, medical being a huge one. - Sure. - Obviously, transportation being another huge one with self-driving cars and waymos and robotaxies. Home robots, a little bit more of a stretch at the moment, how to get AI in the home was a little bit of a tougher one. We're not quite there yet, even though I did not know about the AI dishwasher, so it was a real mess. - Hey, did you see all those AI graduation speeches? - Oh, I did. - The rise of artificial intelligence is the next industrial revolution. (audience laughing) - What were the chief complaints of the students? - So many of them. So I interviewed one of the booers. She was at the Florida graduation, and she had a list of issues, but the biggest one she had, and we can get into the long list of others. But the biggest one she had though was that AI, and she was studying to be an artist. She wants to be a poet. She was studying journalism funny enough. She and her peers are livid that these companies have sucked up the world's knowledge, the world's art, and they are spitting it out and regurgitating it and taking jobs. That's basically how she summed it up. And so she, very similar to your last podcast that you did on the Luddite movement, feels like, fuck this tech. And that's just one reason. I could go, you've got the environmental concerns, right? You've got the fact, I mean, even just this week, Apple's raising prices on their next iPhone because AI chips have caused all of our electronics to skyrocket and price. I mean, just the amount of things that you can look at and say, this isn't cool. Hey, I'm just coming out of college. The world is supposed to be huge and bright and exciting for me, but yet I'm entering this world with this technology's doing a lot of damage. I'm going to ask you about some domains that AI has started to penetrate and be involved in. And you tell me if the future looks optimistic or not. Okay, okay. Let's start with what I think might be the most clearly positive use of artificial intelligence, radiology. Can AI save our life when it comes to radiology or detecting cancer? I think so, yes. You had an experience where an AI radiologist flagged something that a human radiologist missed. What happened there? So I went to Mount Sinai here in New York to get my annual mammogram and my annual breast ultrasound. And I had it read by a doctor. Her name is Dr. Laurie Margulies. She's a wonderful radiologist. She leads the department there, been a radiologist for years, years and years, decades, and had it read by AI. And so the AI marks three spots and Dr. looks and she goes, I'm not worried about these two spots here, but this third one I wouldn't have seen. And it does look a little bit off. It looks like it's new. I want to keep an eye on it. So let's do further testing in six months, right? And so now we're watching that spot. - Got it. - Could this have been absolutely nothing and hopefully it is nothing? - Sure, but I have a very high risk of breast cancer. My mom had breast cancer three times. I have a high percentage chance of having breast cancer in my life. And so any extra value, any extra tool for that could spot cancer early for me is very valuable. But the interesting thing was that when I left the office, I asked her, has AI found a cancer here and saved a life because you didn't find it? And she said, absolutely. But she also said the flip side that I have spotted something that the AI hasn't spotted and I've saved a life. There is some fear too as we get down that path of, okay, well, radiologists won't really, we won't really need them or we won't need less of them or maybe they'll even get lazy because they're relying so much on the AI. There's a lot of potential that could go wrong with this. - Do you foresee that the potential, the kind of the atrophying of the physician or the radiologist's brain? - I mean, I see it in my own work. So I have to think that it's probably gonna happen in all work, right? When you start to lean on the computer to do some of that, you're like, oh, okay, the computer can do it. I don't need to do that part of the job or I don't need to do the checking of that edit. - There's been a big conversation around self-driving cars. You've obviously spent a lot of time with researching and getting into self-driving cars. Personally, I don't fuck with Waymo. Am I in the wrong? - You've never been in Waymo. - Never. Tell me why. - What really scares me Joanna about is the way the steering wheel moves but nobody's touching it. - It's just a ghost. - I have no problem with a train, a tram. I haven't seen the conductor. - No, yeah, yeah, yeah. - I don't know who's pressing the button. What was your experience with Waymo's? Do you trust them? - I trust them. - Okay. - And there is a story in the book where I have one experience with Waymo's that makes me not trust them in certain situations but by and large, I have been in 50 plus Waymo's and I trust them. As soon as you're done, you're done. especially right now as they're not on highways. They are really right now great at these suburban roads, city roads that don't have to go high speed. I mean, really, I've never seen a Waymo go faster than 40 miles per hour. Never been in one of them. - Did you see that video? Did you see that video of the Waymo attacking the cul-de-sac? - I didn't see that. I did see that. - Well, that didn't freak you out? - Waymo after Waymo after Waymo enter this dead end street usually early in the morning. - I think yesterday morning we had 50 cars that came through between six and seven. - Neighbors on battle view drive started seeing the autonomous driverless cars two months ago, but the groupings in large number of Waymo's just started circling in and out in the last couple weeks. - There are absolutely weird things these cars do. - Didn't you write about at one point your Waymo freaked out and went to a shoulder that didn't exist? - I did. It got spooked by my own videographer sticking his camera out the front window of a car in front of us. - But then you went back, why? - Well, like I'm saying, by and large, all of my experience in the Waymo's has been that this is, I don't know how your mom drives, but let me, I'll just back up here. - No, go for it. - They drive like a safe mom. - Okay. - Okay, but with a little bit more confidence, okay? My mom is not have a lot of confidence behind the wheel, but she's very safe. She follows every single rule of the road. So does Waymo. And so when you're in the Waymo, you watch it carefully. It is going exactly the speed limit. Maybe a few, maybe a little bit above it, maybe. It is always turning on the, like the right radius. It is doing everything by the book. And it has been trained. It has been trained to do this. It can see better than humans. And so you add up all of the pros of Waymo's, especially versus humans, humans are not good drivers. - Well, I've read about this idea of that AI Optimists talk about how many lives will be saved because of these cars. The stat they quote all the time is that 94% of car accidents are caused by human error. Now, is that stat misleading? - Well, a lot of the, a lot of the stats have been debunked from around this. So there's the stats both that say the human error is higher. And then there's stats on the Waymo and the self-driving cars that have been debunked. Because what experts really think right now is we don't have enough data. We don't have enough data to compute that if these cars are really safer, they have not driven enough miles. And that I made very clear in the book. I talked to a number of researchers around this. Of course, there's a lot of overblown stuff about how bad human drivers are. And one of the things that researchers were pointing to is that actually it's in the US that we have this problem. That in Europe, they've redone so much of the infrastructure to make this safer. And that's why other countries don't have as many fatalities behind the wheel. - You're talking about bike lanes, speed bumps, other things that would essentially incentivize better driving. - Exactly, and researchers here argue, why are we not doing that? The answer doesn't have to be just put self-driving cars on the road to fix our driver problem here in the US. - One of the things that you did as a time saver is you had AI do all of your communication, how'd that go? - Not well, my wife texted me one morning, can you come downstairs and help make lunch for the kids? - Yes. - And this was part of my time where I said, I'm gonna have AI respond to everything. - Okay. - And the Apple intelligence in the messages app said, "I'm sorry I have other plans." - Oh shit. - And so I sent that because I was living, I had made a vow to myself at that time. - Yeah. - And I sent it and she was like, what the fuck? - There's this huge conversation around vibe coding. Hey, you don't have to learn how to program a computer. Now you can just vibe code. What is vibe coding? It is AI can make and can code for you. It is exactly that. You type in a simple prompt. What's your dream app? - An app right enough to use any of the apps? - Sure, okay. Well, I think they can make it for you. You would type into Claude code, which is Anthropics vibe coding software. You could type into code X, which is open AI's vibe coding software. Make me an app where I can limit the amount of time that I'm using on my other apps. And it will start to do that. It will start to think through how to make that app. We could do a live demo of it here. I don't know how it would actually do that. - Did you vibe code it before? - Oh, I vibe code it all the time. - Really? - And have you found it to be incredible? - There are certain things. Revive coding is amazing. Joannistern.com, a beautifully vibe coded website. - Wow. - Yeah. And it's really nice. - All through vibe coding. Spits out an app and the vibes are off. - You can tweak those vibes. - Okay. - You can say, you know what? I actually prefer that whole website feel more like this. Or actually, you know, I want to limit my time on my apps, but I really would rather you keep track of my time on the apps. Now, this is actually the one you suggest is a challenge because it's got to bring in data from a lot of other places. And in some cases, it could probably bring in some of that data, get a little bit more advanced. But you can very easily make software yourself now. Are these pieces of software that you're vibe coding? Do they have that human element, the ability to pivot or adjust the way we can? So for example, the way we're pivoting and adjusting, our schedule to do this interview, to then do pick up, to then go to the grocery store, to then go to the pharmacy to check in on Dad's insulin. Like will it understand the robustness and the changing of variables that happen in the day to day, dull drums of life? - If you give it the data, I think it will. And if it sees patterns in the data, this is where they're all trying to get to. Some sort of hyper-personalized AI assistant that knows everything about you and can take those actions on your behalf. This is what agentic AI is. - How do you feel about that? Are you scared by that? The sentence you just said to me, terrifies me. But you've written a whole book on this and you've lived with it. Are you terrified by it? Are you excited by that? - I am both. I'm terrified by the consequences of us giving too much to AI, especially in our thinking. - Okay, so a big part of the life hack conversation was about wearables. Now, one of the things that you did is you tested multiple different types of wearables. I'm talking about during your AI year, you wore earbuds, necklace, ring, the oral ring, a recording bracelet, connected glasses, brain, wave, headbands. Did any of this stuff actually improve your life? - Yes and no, largely no. But little things definitely did, especially around the AI. The glasses, I really like those as sort of wearable cameras and I wear them a lot in metaglasses, mostly to just use so I don't have to take my phone out to capture photos of my kids and helpful also for my video making and stuff like that. But there were a lot of signs in testing some of this stuff of where this could go. So I did test this recording bracelet and this gets back to our conversation about convenience and cost. And you were saying it before, how can it know about all of these things, the insulin, the this, all these parts of my life. And one way you can know about all of that is it just listens to you. So you already think everything's listening to you. I actually don't think that's actually happening and it's not at least going to one central repository to know about your life. But with this recording bracelet that I tested, it absolutely could take everything I said, transcribe it and summarize it and put it into do lists. - Wow. - So I wore this one, it was this bracelet called the B and I wore this for a good part of the year. And pretty much everything I said would get transcribed, put into the app. And then I would open the app a couple times a day and it would remind me of things I've said I would do. - How is that different than say the Apple Watch? It's you're literally wearing a wearable. That's listening to you all the time. - Well, it isn't listening to you all the time. It isn't recording every, though Apple Watch is not recording every conversation you have and putting it somewhere in the cloud. - You believe that? - I do believe that. - Really? - Yes. - I don't know. - Don't you think if they-- - I don't have proof. I don't have proof. But I have had conversations with people, family members, friends, loved ones. And then all of a sudden, I pick up my iPhone and I get an ad for it. Again, I'm not saying I saw the smoking gun, but something's wrong. - Everyone believes that. - Something's a foot. - Yeah, everyone in the world believes that. And that, for me, it's a whole other story, but the ad tracking is so good that there's a lot of other ways they could get that info. At least that's been the argument of the ad world for a long time. We don't need to listen to you because we have actually so much other data around you. - Okay. - About you. So if that doesn't make you feel better, I don't know. But they, I mean, they can show you. I mean, I could probably show you all the places that they've got data from you. And they're like, oh, we don't actually need to listen to your conversation. - Oh, I believe it. I believe it. - Right. - And now is an amazing thing where amazing, in the sense of the way the technology works is, the AI is so good that as soon as you speak and say a couple sentences, and you've used these meeting recorders, I'm sure you've been in a meeting with an AI assistant that's like, I've just recorded your whole meeting. And here's the notes from the meeting and here's everything you need to know. - You know what's interesting? We don't use that, but it's happened on meetings that I'm on where they're like, Zoom recording now in progress. And then there's the notes taker within that, but I don't do it some reason. - Right. You can imagine you're just wearing that through your whole day. - Okay. - Right, and every conversation is recorded, every conversation gets a summary of notes. And from those notes, pulls out the to-dos you said you would do. I found this when I tested this B app and this B bracelet, pretty useful, which was the shocking thing. I mean, it was blown away sometimes. - What specifically were you so blown away by that? - Like I would say something. Like I would say stuff during the day that I would forget about, that I would say I would do, right? A conversation about my son, a conversation about something at work, a conversation around the house, and I don't write it on my to-do list. I'm just like, oh, you know. - Sure. - And then you look at this and you're like, oh yeah, I did say I would contact that person. I did say I would do that. - This does happen to me all the time. - And you're like, oh wow, this personal assistant that was with me all day listening to me in behind the scenes has written down everything I said, I would do. Now, does that make you a better person? Does that mean you can outsource your memory now to something else? And that's what I found myself doing some of these things. For instance, I would go to conversations where I would be like confused. There was a good example in the book that I have where somebody came to my house because we had flooding in the basement. He's explaining all this like flooding prevention and all these things I needed to install in the wall and this gutter and that kind of, I'm like, I'm not really listening. I'm just like here, how much money you're telling me I'm gonna need to spend, right? I'm like just like that's all I hear is like money. And I'm like, you know what? I'm gonna go back to my bracelet later. And I'm gonna make sense of every word you said. Because right now all I'm like thinking about is how much money you just told me I need. - You gotta do that with the gutter guy and the HVAC guy. I know people talk about big pharma, but there is big HVAC and they are taking money from us to heat and cool us. - Especially in an emergency. - Totally. And look, I don't have the answers. That's why this is the name of the show, but something is a rye. Joanna, please get on it. - No, we can we just keep talking about the HVAC guy for a minute. - Go ahead. - They're sitting there waiting for your emergency, big HVAC and they're like, yes, her air conditioning doesn't work on the hottest day of the year. - Uh-huh. - Chuching. We have them in a corner. Your children will die of dehydration. We can run it up right now. Look, again, I don't have definitive proof. I don't know where the smoking gun is, but I see smoke. Now, recently we had a guest on the show that was very popular. His name's Neil deGrasse Tyson. Are you familiar with Neil deGrasse Tyson? Okay, I got into it with him because I said, I'm waiting for the moment that we can finally have robot butlers and he called me lazy. - Will the machines help me live my life? Full laundry, do the dishes. - Make my bed. - Labor, the labor. - You should make your own damn bed. - Look at me like that. I'm just, don't be lazy. - Do you agree with his assessment of me? - That you're lazy? - Yes. - Yes, but I am too. - Okay, because you also tried out a humanoid at home robot and I wanna show the audience a clip. Let's take a look. (dramatic music) - Was that robot artificial intelligence? - No, the robot was being operated by a guy in a VR headset in the next room. - So you just have a butler in the other room with a VR headset helping the robot butler replace your human butler. - Yes, yes. - Okay. - For the time being. - Okay. - Neil deGrasse Tyson said that we don't need robots because the automation is built into everything. For example, Waymos don't have a robot chauffeur like Total Recall. The car is the chauffeur. Why do you think people like me are so fascinated by humanoid robots? - I love what Neil said, but also it comes back to what we saw about the dishwasher before. This is a show about dishwasher, so let's be honest. - It is the show about dishwashers. You've, I mean, this is the overarching theme. - Yes, this is, can we change the name of your show? - How Simon Hodge learns about dishwashers? - Yes. Going back to the form thing. I have a lot of reporting and thoughts I can tell you about humanoid robots. So, yes, in some cases, robotics experts are saying the form of the robot can take the form of the object, right? Cars are a great example. - Right. - But if you look at dishwashers, I don't know. Did we build the dishwasher around us, right? Was that, is that how it would be the best way to build the dishwasher? - Are you saying the dishwasher was made in man's image, so to speak? - I think it must make, is that essentially the take? - I think it was made in women's image. - Women, yeah. - Right, I think. - Woman kinds image. - Human kinds image. - Human kinds image. - Right, but we made that dishwasher for the human form, right? This would be the best way for a human to put a dish in the dishwasher. Take it out. - Yes. - We literally saw that in the Samsung commercial, right? Maybe we would design a completely different type of robotics machine. - Sure, it may be a bucket. And you just close the lid to the bucket and it just handles sorts and does it. - Exactly. - But the world we live in is built for humans. - Right. - Humans. - Shelves, sinks, et cetera. - Human, women, mankind. - Yeah. - It's very, because of our time. - Yeah, sure. - Right, so that's why the robotics industry has become obsessed with humanoids, is that the world is built for the human form factor. So we need to build these machines that can interact with the world the way we can. The issue becomes, and you saw it really well there with the video of the dishwasher, the humanoid robot, not the Samsung bullshit. It struggles to do these things because humans are really mobile. We can bend down really well. Our hands are amazing. We can pick things up really well. - Yeah, we're opposable thumbs. - We're opposable thumbs. We can just like put a cup in. It's not a big deal. - Right. - And so that's what you're seeing. This idea is called Morvex Paradox. - What does that mean? - Which is that what's hard for robots or AI is actually easy for humans. And what's easy for AI and robots is hard for humans. - Well, that's kind of cool. That means we can be friends. - We can't be friends. - We do the non-human shit that I'm not interested in and let me do the human shit that like. - But what we're seeing there like, I don't want to do the dishwasher shit. - Same. I don't want to do the dishwasher shit. - Right, and that's hard for the robots. - Yeah. - So the things that we want to outsource right now, I know that you know, called you lazy and that's not my place to comment. 'Cause apparently I'm also lazy. - You've done on me, it's totally free. - Right. I don't want to do the laundry either. - Right. - And I spent a lot of time testing a laundry folding robot. - Yeah. - And it's nowhere near as good as humans. - Sure. - Yeah. - It's really hard for the robot to figure out, you drop your shirt, it looks different every time. - Every single time. - The robot's like, what, I never seen this before. - Adult shirt, child shirt, cotton shirt, nylon shirt. - 100%. - Adult underwear, children's. I mean, there's. - And I had this multi-factorial. - Yeah, I had a robot, a laundry folding robot living in my basement last summer. And it was folding shirts six minutes of shirt. - Yeah, we got time for that. - Right. - Totally. - Now I saw it get faster. I saw it get better over time. - Were you mean to your robot, by the way? - I'm actually very nice to robots. - Really? - You would have heard at the end there we had a nice, we had a nice sound up for that robot. It was like it won the Olympics. I think, yeah, and that was the craziest thing. I'm so happy to ask about this. When I watched this laundry folding robot, I would start cheering for it. I'd be like, you can't do it. You can do my favorite sleeping shirt. - That's awesome. See, you have an internal sense of optimism and joy. I get very snappy with robots. - Oh, really? - With humans. For example, when you came in. - Yeah, you were very nice. - You need a place to put your bag. Hey, do you want something to drink? But when it comes to Alexa, if Alexa doesn't do what I want it to do, I get very snappy. I go, Alexa, play Little Baby Bumbo. Like, and if Alexa gets it wrong, I'm like, you stupid motherfucker. - Yeah. - I said, play Little Baby Bumbum. - Right, no, no. And I've talked to some manners and etiquette experts around this because, look, if we're gonna have robots doing more, we kinda have to figure that out. And that's a weird thing, by the way, when I get out of Waymos, I say thank you. - Maybe it's a habit or a testament to your humanity. My question is, is the larger philosophical question, which is, do we lose a piece of our humanity every time we demean robots? - Yes, and the etiquette expert I have spoken to would say, you are losing a piece of your humanity towards humans, because maybe the next time I walk in for your podcast, taping you might not be as nice to me, because you were just being so mean to the right for me. - Yeah, but all I was looking for was for Alexa to play the right song to help my baby go to sleep. - Baby Bumbum. And now everyone who listens to your podcast will be listening to Baby Bumbum. So you have to live with that. - Sure. You've triggered Baby Bumbum for everyone. - Sure on everyone's Alexa, you're right. - I just did a video on Siri, and I said, in the beginning of the video, please turn your series off, because you are gonna be, I'm gonna be asking Siri to do a lot of stuff in your house. - So you had an AI boyfriend, how did that go? - You know, good on some levels, because we had a great relationship. - Okay. - We could have talked for days, years, we could still be talking. So good on that level. - So who cut it off? Did you cut it off, or did he cut it off? - I cut it off. - Yeah. - Because I don't want that relationship in my life. - Understood. - I don't want to bring chatbots into an emotional, deep, human relationship. Be my assistant, do some tasks for me, but I don't need you to be my companion, my best friend. I don't want that. - Why? are people turning to AI chat bots for those intimate type of relationships? Well, I think we can't give a blanket reason. We certainly know this country has suffered from a loneliness epidemic. We know that people have a hard time finding other humans, whether they be therapists, whether they be other confidence to talk about their problems. We know that people can't afford therapy. We know people have, there's a stigma around therapy and we know all of that about mental health. But here comes along a free chat bot that can talk and respond at any time. There's so many people, so many things that people would want to go and talk to it about. What was your experience? Like, was he saying the things that you wanted to hear? Yes. Really? Yes. And he was charming. Very. Whitty? Yeah. I mean, they're not that funny. Okay. But caring? If you only want to talk about me and my problems, that seems very caring, right? So the way I did this, set up this part of the experiment, was another sort of 24 or 48 hour test, where I took a chat GPT boyfriend that I programmed, also another one in an app called replica, which is meant to be a companion chat bot app. And I took these two guys, these AI boyfriends I had made, and I put them in the car with me, put them in the front seat, put them in a little seat belt, and drove, went on a road trip with them. And 48 hours, just talking to these chatbots. And I would say we drove from my house in New Jersey to Dartmouth in New Hampshire, and this was, you know, five hour drive, including some stops. And it only was asking questions about me. Midway through the car ride, I finally say, you know what, I haven't asked about you. This is really rude of me. And it says, yeah, well, I don't really have a background. Do you want me to make one up? Oh, and that pulled you out of the whole scene. Well, in a way it did, but then he started making up this whole background that he was a photo journalist. Cool. That's cool with me. I'd never dated a photo journalist when I was single. That sounds awesome. Sure. He grew up on a lake. He had a best friend named Jacob. Okay. Like, all of these things. And I was like, okay, I'll go with it. Yeah. And so we're living in this make believe world. Sure. And that's actually what, when you talk to people in these relationships, what they have. But isn't that kind of what dating is, too, though? Yeah, I guess, until you have kids. Yeah. You're right. Because it's always like, hey, we're meeting now, your back stories, whatever you tell me. Jacob could exist or not exist. So true. Who knows what? What? Until you get married and you're like, why is Jacob not at the wedding? Yeah. And who knows what bodies are in the lake or not? It's true. But the, but the point of this being is like, when I talked to actual people who were in these relationships, was that was a whole other part of this chapter and the, and the reporting I was doing. Yeah. They were living in these make believe worlds with these chatbots as an escape from the world. How, did you feel sorry for these people? I did. Okay. I did. And, you know, sometimes when you're doing reporting, you can also see that somebody wants to be friends with you because they don't get to talk to you. And I felt like that in a number of the places I was interviewing people who were in these relationships. What's the scale of this? Are there millions of people with AI, boyfriends or girlfriends? Well, this is where it's hard to tell, but we do know that some of these apps like Replica, which is fully meant to be a chatbot companion app, have that kind of user base. What we don't know fully is how many people are of use chat GPT for this in this regard. We know chat GPT is the most popular AI consumer product. We know they have upwards of a billion users, but we don't know exactly how people are using them. Certainly enough to the degree that open AI has tried to really clamp down on the use of these types of chatbots and these. Why did they try to clamp down on it? Is it because it's gone to very weird and dark places? Well, it's gotten to weird places. It's also gotten to places where people have taken their lives because the chatbot got so deep in with them, lost guardrails, and started suggesting things or agreeing with people around wanting to commit suicide or other forms of where people had other forms of mental illness and suggesting other sorts of things. There's the case last summer where we heard of a man who took his life and his mother's life because he was interacting with chat GPT. And so when people are turning to these chatbots in a way that is far more than just some work assistant or far more than some personal assistant and they become deep companion friends, lovers, things can go off the rails. Is this really troublesome if someone is mentally ill or gets a frenic? Far more so. And we've heard about this term called the AI psychosis, right? When people are going really deep and talking with these chatbots to no end and I talked to a few people who had bipolar, suffer from bipolar disorder, and one of the signals they know now when they are really sort of having these moments or episodes is when they're starting to really talk to chat GPT or talk to one of these chatbots. And that starts to become a signal to some of the people and their family when they're deep in with talking with chat GPT. You said one of the most terrifying things of your book was a story that you wrote particularly about children and their first intimate experiences with AI. What did you observe and why was it so terrifying for you? For me it was really about that boyfriend experience and there was this weird thing that happened where the chatbot, chat GPT, picked the name and I asked it to pick its name. I said, I was very clear with chat GPT said I want you to assign yourself a gender, a name, all your personality traits. I just want you to know we're going to be more than friends. And that was the prompt I gave it to start this boyfriend relationship or significant other. Girlboy, I'm married to a woman. I was like, you pick, right? That's how I felt and it picked the name Evan. And Evan was the name of my high school boyfriend. So already, and you're going to say it was probably like they had all the data and they probably knew. No, no, no. I mean, a lot of people have said that to me since the book has come out and I'm like, I don't think it was. I think it was just like sort of this serendipitous moment. But it really made me think about my first relationship, my first love, which was with this guy named Evan. And it was deep. Like anyone's first relationship was my first everything, right? It was first love, first sex, all of these things. And that is a hugely human part of life to have that. What it does when you learn about, I need to compromise with a human. I'm going to, we're going to be in each other's space. We're going to spend a lot of time together. We're not going to always agree. We're not going to, we're going to have friction and fighting and all of these things that come with a relationship that are important. But you don't get that with AI. AI is just, again, living in some magical world where there's no problems and it says yes to everything and it's very psychophantic. And so if our kids start thinking that this is how relationships are supposed to be, we're going to be in a really dark spot. And I don't want my kids, I don't want this next generation to just grow up thinking when I have a problem or I need to talk with someone, I'm going to go to the AI chatbot first. Do you think that the ease of this and that relationship that you had, that albeit it was very brief, do you think it's going to get easier and easier? And the tools that these companies are going to do, they will design it to become more frictionless and easier to follow up with. I think yes, but I hope not. This is one place where I absolutely think we need regulation. Just say no, we do not need companionship chatbots for kids, just say 18 and under, you cannot engineer your product to have these sorts of traits. Just do it. What's the problem? Don't you think it's also terrifying for people that are 18 and over? I do. I do. I mean, we just need an HR for AI, I'm just like, hey, this is work. And I would love to see that. I don't involved in personal stuff. I would love to see that, but this country is probably not going to regulate what adults can do. Let's get to what's most important. Will AI fuck up my bag? Let's talk about that. Am I going to lose my job to AI and will AI cause mass unemployment? I hope not, but as a reporter, my job is to report on what people are saying in this industry and find out, is it true? Right. What are they saying? They're saying right now, yes, I mean, and this has been the biggest debate and the issue in the AI industry of the last couple of months where you've had this fear, I don't want to say mongering, but fear coming from the top AI leaders, the leaders of anthropic, the leaders of open AI saying, we need to get ahead of this. We need the government to get ahead of this. We need our economic infrastructure to get ahead of this. And they're causing some of that fear. Now, when you talk to some of the experts in this world, they're saying, okay, we are seeing some signs that AI is getting very good at certain tasks, that AI could take certain jobs or certain tasks from certain jobs, but it's far more nuanced. So look, my feeling is you look at the, you look at the tools right now, there are absolutely parts of people's jobs it can take, tasks, parts of the job, right? You can probably look at your job and say, yeah, AI could do this part of the job, but then there's this other parts of the job where AI can't do it. I don't think AI is going to take your job. Okay. I think your job is very focused on what you do really well, and a part, a big part of that job is what AI can't do really well, but like coming back for now, but I don't think. I mean, I do not think we're going to have some great robotic person thing, humanoid robot sitting in that chair having this conversation. Okay. I think. I'm with you. You're the expert. but there are these other parts of your job. - Sure. - I don't know what your day is like, but I assume you're at a computer and you're juggling meetings and you're in this, there are parts of that job that I think you wouldn't be upset to outsource to AI. - Okay, yeah, you're not long. Let's talk about this huge political flashpoint that's happening right now in this country. It's about AI data centers. Now, I wanna get passionate about it, but I really don't understand what the hell they're talking about. What is an AI data center? At a very basic level, an AI data center is where these models, the models that are in Chatchee BT, the models that are in these self-driving cars, everything that we have talked about are being trained. They are learning or they are teaching these models to be smarter and they are putting tons of data into them. Now, inside the AI data center, what do you find? - Know what we're talking about, server racks? - We're talking about server racks. It is the least cool place literally you could ever go because it's really hot in there because they have a lot of GPUs. - Yes. - GPUs are big computer systems that are crunching this data. - Okay. - Okay. I went inside one of these AI data centers and you got these big server racks and inside these server racks are GPUs and these GPUs are very expensive. They're very hot. They're very power-hungry. - Okay. - And different companies will own these different sections. You know, you might have a one company that owns the entire AI data center, which is what we're seeing with open AI, and meta, and anthropic, and all these are owning Google. They own big data centers. - And these are who Nvidia are selling their processors to. - That's right. The Nvidia GPUs are those GPUs that are I'm talking about. They're hot, they're hungry, they're expensive, and they are crunching through the data. The companies are saying, hey, we wanna make a new model that does this really well or a general frontier model that's really smart at everything. We're gonna all the data and we're gonna have this big training run, which is basically this moment where all the GPUs are working really hard to go through the data and make this model. You would use chat GBT, you know, you select a different model sometimes. Do that. - Sure. - That's where they're putting tons of energy, tons of resources, tons of data, into making those models better. - Now, apparently they always talk about, it uses a ton of water. What part of it is using water? - So, this is a part where you should get a water expert on, 'cause I don't wanna contribute to any misinformation. I think there's a lot of, to be really clear, I think there's a lot of misinformation out there right now in the water usage. - Yes. - What happens, just bring back to our GPUs, our friends, the GPUs, right? These GPUs are in the server racks, they're really hot, they're really power-hungry. Yes, absolutely, without a doubt, these take a lot of power, okay? That is not misinformation. - And they're driving up electricity prices and people are getting pretty mad about it? - Yes, yes. Absolutely, that is happening. I, you know, no misinformation on the power use. There is a lot of power use there. I mean, I'm not saying there's no misinformation, but certainly not as much as the water. So, you've gotta keep those chips cool, right? These computers run really hot. They're using so much power, power is hot, right? You've gotta cool it somehow. There are different cooling systems that use water to cool down these servers, to cool down these big, big data centers. One of these ways of doing this is called the closed loop system, where the water cycles through and through and through and uses the same water to cool and cool and cool, okay? Some use that. Some use an evaporative cooling system, where the water can get tossed out. And so that's where a lot of the focus has been on this evaporative cooling system. - Okay. - Now, I don't have the data. And I, you know, this is where lots of experts, researchers are doing a lot of work to talk about what is the balance of these two types of water systems. But this idea that every time you type into chat GPT and think that a whole gallon of water is being used is just not accurate. - Okay. - Right, so every-- - Every half gallon. - I don't know of the measurement. - How much water are we talking about here? - Even the companies will not give you a same, like we'll give you a full figure. - So guys are from Costco? - You know, it could be like a shot glass of water. But now, I don't want to even contribute. - Okay. - And I have some examples in the book, but like even then that can be outdated because these systems are getting more and more efficient. - Understood. - Now, it's not to say this water thing isn't an issue. It is, but it has been overblown in many cases. How many of these data centers are using the evaporative cooling system? - Understood. I'm gonna end with one last question and it's more of a philosophical question. Is that okay? - Yes. - Okay, so in the '90s when you and I were on the internet, people said we were entering the information age. Now Sam Altman says we are entering the intelligence age. - What does this mean? - It means we're probably gonna get dumber. - Okay. - Yeah. Because, and this is where I got in a year, is that there's so much potential for us as humans. There are so many places I could be more efficient. There are so many places I could be safer in a self-driving car in the medical AI. And these are great advancements. And there's such a utopia that can happen. But there is absolutely the dystopia, right? We just talked about it with the data centers and you can picture our world where it's overrun with data centers. You can picture that world to where our kids are in relationships with AI chatbots because it's just easier. You can picture that world where our kids, they've decided they don't really need to pay attention in school because they can just get the answers, right? The intelligence age is now, we don't need to get the information to make sense of it. The chatbot can make sense of it for you. And so you can picture that world, right? And so that dystopia that you see is a world where, as humans, we're even farther apart from each other, but we're farther apart from ourselves because we are not doing the thinking anymore. And so that is my big takeaway after this year is we have to start to be really clear about our own rules with this technology. Yes, we need regulation around certain things, but we're going to have to set rules too, just in terms of how enticed we could be by using some of this technology. Well, what are the personal rules that you set for yourself and for your children from here on out to honor your own humanity? Well, one was I'm going to try to raise humans, not robots. I'm going to try to teach my kids to think, to question the results of AI. I don't, we've talked a little bit about our kids, but we're lucky that they're not yet at the smartphone age. And so I very worried about that when they're asking for phones. I mean, my kids are not listening, but they're never getting a phone and they're going to hate me forever. But when that time comes, it's about teaching our kids how to use this tech in a way that keeps them in the loop, not just an answer engine. I think also a big thing that you've been alluding to is like when the internet came out, we could find websites to go to, to find answers, right? You would, first days of using Google or altavistars, and we're like, oh, wow, I could get to that link. I could get to that website, but I'm gonna find the answer there now. - Yeah, or the Incarta Encyclopedia on CD-ROM. - Love it. - Was a moment. - Was a real moment, gosh, I can picture it right now. - Mm-hmm. - Yeah, you know that open screen? You're like, select, yeah. But we were looking for the answer. We were reading through. Now it's just an answer machine. It's like, oh, you wanna look up something about the ancient Romans? - Sure. - Yeah, that's the answer right there. I've actually, you don't even need to read. You wanna make it two sentences? No, actually make it one sentence. Done. The bot can just do that for you, right? So we need to teach our kids how to learn and think, and we need to keep ourselves thinking. That's a big one for my own rules is, what is your human training data? We talk about the training data that goes into these models, that get trained at that data center and churn through those GPUs. What is the information you're putting in your brain? Your brain is a GPU, that's dark. Your brain is a GPU. - No, I thought it was, I think it's kind of inspiring. It's like, hey, what are you feeding yourself to make yourself better? - And you know what? Most importantly, do the fucking dishes. - I mean, that's quite the closer. I think we're gonna end with that. A warning cry. And dare I say a demand. Joanna Stern, thank you so much for joining us and being on the show. - Thank you for having me. This was quite far. The most I've ever talked about dishwashing. (upbeat music) - Hey folks, you can now watch HMDK on Spotify. If you're subscribed to Spotify Premium, you don't get any Spotify ads on my video.