1352: Joanna Stern | The Year I Outsourced My Life to AI

24:22And it's like, calm down. It's impressive in some ways. It was basically like, gives me all this information. And then he says, calm down, go take Browser, who's my dog, for a walk. Your dog's name is Browser? Yeah. That sounds like a robot dog. Isn't it? It's named after Web Browser. Yeah. Yeah. Yeah. Yeah. But I just told you I'm in Los Angeles. Why would I be taking my dog for a walk? Oh, right. Right. OK. Like there are these things where it's like it just doesn't know the world around it. Yeah. That makes sense. Right. Yeah. I didn't bring my dog on a cross-country flight. And I actually, in that conversation, kept telling it, no, I'm in Los Angeles.

24:57And then later on again, it would say, like I said, go take your dog for a walk. Take a moment, step away from the book rankings. Okay. Speaking of machines pretending to care, let's hear from sponsors who at least admit this is a transaction. We'll be right back. This episode is sponsored in part by Ground News. One thing Joanna Stern talks about today is how technology keeps getting better at filtering information for us. AI summarizes articles, algorithms decide what shows up in our feeds, recommendation engines tell us what to watch, read, and click next, and the problem is that convenience doesn't always give you the full picture, which is actually why I love ground news.

25:33Ground news lets you compare how the same story is being covered across the political spectrum, so you can see the framing instead of just accepting whatever version happens to land in your feed first. I was looking at coverage of the recent US-Iran talks in Switzerland, ground news pulled together dozens of sources covering the negotiations, Courier's headline focused on what mediators described as encouraging progress and a roadmap toward a possible agreement Meanwhile today's edition led with Trump threatens Iranian negotiator's safety and they walk out Same underlying events completely different takeaway Ground news helps you understand not just what's happening, but how different outlets want you to think about what's happening I also use their blind spot feed which shows stories getting major coverage from one side of the political spectrum and very little coverage from the other It's a great way to find stories that you might otherwise never see.

26:18Go to groundnews.com slash Jordan to get 40 % off their unlimited access Vantage subscription. That's groundnews.com slash Jordan. Groundnews.com slash Jordan. Use that link so they know we sent you. This episode is also sponsored by Bowlin Branch. We upgrade our phones, our coffee machines, but then we'll sleep on the same tired sheets for years, even though our body is touching that bedding for, what, like seven or eight hours a night? That's a 30-year day, and if your sheets are trapping heat, slipping off the corners, are feeling rough and worn out, your body notices. That's why I'm a big fan of Bowling Branch.

26:47Their sheets are 100 % organic cotton, breathable, lightweight, woven specifically for airflow, not just softness. So instead of waking up hot and sweaty, your bed actually feels cool, soft, and comfortable. And you feel the quality immediately. The moment you get in, it's that clean hotel level feeling, but more relaxed and inviting. 94 % of Bowling Branch customers say their sheets get softer with every wash, and I can attest to that. We wash our sheets weekly. I use Bowling Branch, and I'm not going back to regular bedding. When something touches your body for that many hours a day, it's worth upgrading.

27:14Sleep cooler this summer with Bowlin Branch during their annual summer event. For a limited time, get 20 % off site-wide at bowlinbranch.com slash Jordan with code Jordan. That's Bowlin Branch, B-O-L-L-A-N-D branch.com slash Jordan, code Jordan to take 20 % off. Bowlinbranch.com slash Jordan, code Jordan. Exclusions apply. A lot of you might not know that when you use our promo code, it really does actually support the show. We don't get a cut of sales or anything, but it lets the company that bought the ad know that people are actually hearing the ad and they're engaging with the ad and they're likely to renew their campaigns with us, which, you know, keeps the lights on around here.

27:49So, hey, if you're signing up for something or buying something, use our code. It's usually Jordan, but not always. All of the codes are on the deals page at jordanharbinger.com slash deals. You get a discount and you help keep the show going strong. Thank you for your support. Now back to Joanna's turn. I know people are going to hate on me for this, but I wear these meta Ray-Ban glasses. Oh, I wear them all the time. They're amazing in many, many ways. I at first was like, I'm going to just film my kids with these because whenever I take out my phone, they stop doing the cute thing. And I'm like, oh, I want the button.

28:20You know, if I'm out and I see something crazy, I don't want to have to be like, hold on. Click, click. I can just go boom and then record the gnarly traffic accident I just witnessed or something or like this interaction between people that I don't necessarily want them to know I'm filming them. But if this turns into a fight, I'm going to have video evidence of what happens or something. But now what I find myself doing is being like, hey, do I need a jacket today? Do I need this? What time is the rain starting? But that AI makes some of the dumbest flipping mistakes. People go, oh, meta AI is amazing.

28:50I don't know, guys, if you're listening, what's wrong with you? I'll say, hey, give me the temperature in Celsius and Fahrenheit every time because I want to kind of learn like what is 22 because I'm talking to my friends overseas and like, is it hot there? I'm like, yeah, it's 90. And they're like, OK, I don't know what that means. I have to get an app for that. So I was like, I want to be able to say, oh, it's that's 29 degrees or whatever that is. And meta AI will go, no problem. I'll give you the temperature in both Celsius and Fahrenheit. And I'll go, hey, what's the temperature? And it goes 12.

29:19And I'm like, OK, you can't follow the simplest of instructions. Well, that's memory. Yeah. That the memory isn't there. Similar to what just happened with the browser situation and flawed for me when I asked it. I'm like, I'm on this book tour. I've told you now twice I'm not near my dog. Right. The memory isn't always there. Yes, it can't do that. And it depends also if it's on the same chat thread. It happens a lot. It can only have certain memory of that. And so that's one big place they've been improving and has been improved throughout the year. But I'm so happy you use the Ray-Bans. I use them all the time.

29:52And for the year, I wore them all the time, both the sunglasses and the regular glasses, though the regular glasses make me look like Garth from Wayne's World. At first, I was like, I can't do this with the LED in the camera. Not one person has said, hey, is that a camera on there? I think my son was like, wait, is that a camera? No one else has said a damn thing. They look so good. And I mean, other than when I look like Garth, but the sunglasses are really good looking. And I liked them initially, and I've been wearing them for a number of years since I reviewed the first pair for the camera reason.

30:25Same reason as you. I have kids. I want to wear them when I'm skiing. I want to wear them when I'm doing other things in the world with them, bike riding. So much easier than ever using your phone. But then when the AI feature came out, I really started using it, especially to even ask questions for my kids. We're out and about. It sounds like we have similar kids. What is that bug? Is that a fox? Is that a this? Like random questions that you can be looking at and say, hey, Meadow, what is that? Right. So for people who don't know, it'll apparently take a photo of that thing. Sometimes it misunderstands me and I'll say, hey, what about this?

30:59and it thinks I'm asking what I'm looking at, and it'll go, it looks like you're in a bathroom staring at a wall with a plant on a shelf. And I'm like, oh, no, I mean, what is the thing that I said after that? And it's like, oh, this is an ingredient that's common in shampoo. It's very funny that you can do something like that with the glass. I guess I answered part of your question, which is where is this enhancing my life and enhancing my life? And it's not just ChachiPT, it's having those around all the time. And a lot of people think like, oh, the last thing I want is AI around me all the time.

31:29And it's like, I don't know, it's pretty damn convenient to be able to say something like, hey, what time is the rain going to start and not get your phone out or something even more involved? Like, how come sometimes tires are filled to the same pressure and then one loses pressure, even though there's not a hole in it? And it goes through all of this, like, well, it could be the air expanding and maybe the sun's on one side of the car, not the other. And like, wasn't an important enough question for me to get my phone out, stop what I'm doing and learn it. But it was just fine while I'm driving to learn about that particular thing.

31:57Towards the end of the book, I talk about where I think a lot of the future of these devices are going. And I think the glasses and AI wearables are going to be a big thing. We aren't quite there yet, but the glasses do two things. One, they can see the world and two, they can hear it. And so that is giving us an AI assistant that can do those things for us. And that really does enhance us. Mark Zuckerberg has said in the past that people in the future who don't have these glasses will be at a cognitive disadvantage. And I actually agree with him. There are many times where I'm wearing these glasses and I'm able to quickly reference some information.

32:31But going back to our conversation right before this about getting things wrong. Am I smarter sometimes if it's getting it wrong? Yeah, right. You're just confidently incorrect on certain things. Another really good use case. We have a nanny. She speaks Spanish. She speaks a little bit of English, but not as much. And she's an amazing woman that my wife has to use Google Translate or text translate in the same room to talk with her. And it's kind of dumb. You can tell your metaglasses to real time translate Spanish and it will just tell her what the nanny is saying. The problem is it's a little bit slow and it's a little bit wrong sometimes.

33:08Jen knows a little bit of Spanish from Duolingo and can answer her. But then when the nanny says something to my wife, it's just sort of a little too delayed to be super, super useful. But it'll just tell her what she said. And I'm like, wow, once they can speed that up and everyone has them, you can just talk to somebody who doesn't speak a word of your language and it'll be completely flawless. And your nanny will have the glasses and Jen will have the glasses. that latency will be down to split seconds. And yeah, they'll be having a conversation back and forth. One day you'll just be sitting in, I don't know, a cafe in Turkey or Saudi Arabia.

33:43And you'll be like, I have no idea what this menu says. Instead of getting out your phone and taking a photo and translating it and blowing it up, you just say, what on this menu is heavy protein and doesn't have any dairy in it? And it'll just read the whole menu and go, this item has this and you just order that. Here's how you say that. Okay, it's number 26, just point to it. But that's so useful. I feel like I'm doing a commercial for these glasses now, but I would love to hear what you learned comparing AI to, I don't know, a human doctor, a human driver, a human therapist, a human assistant, instead of just comparing it to nothing, right, which we're doing now.

34:15Yeah, it depended on each of those different professions. But one of the things that I found consistently was that in no place was it fully replacing a human. In no place? Look, not in medical, at least right now. Right, okay. So that is a big section in the book, and I talk a lot about how radiologists are working hand-in-hand with AI. Yeah, tell me about that, because I don't know anything about radiology other than they look at the x-rays and say, oh, this is a fracture, or you have a little tumor here, or something like that. It's really surprising to me that AI can't do that flawlessly. Okay.

34:54But it needs a lot of oversight from humans. And I'll explain specifically what happened with my breast ultrasound. So they are using AI systems that are not these big, large language models, right? These are systems that have been trained specifically on millions of radiology images. So in the case of my breast ultrasound and my mammogram, these are models that have just looked at millions of images of breast ultrasounds or breast mammograms. And so they have these systems where the doctor is going to look at your imagery, but then they run it through an AI program. And in my case, I went and I had my breast ultrasound read.

35:36And the AI said, there are three suspicious spots on this breast ultrasound. And I watched this doctor. Her name is Dr. Lori Margulies. She works at Mount Sinai here in New York. And I watched her go back and forth with the AI. She said, OK, let's look at this first suspicious. Is she talking to it or is she typing or is she clicking? Really just looking at it's a picture of the image. It's a picture of the breast ultrasound. And the AI has a box around the areas that it thinks it might be suspicious or are benign. And then it has a little bit of information, what the AI thinks. And it's all kind of color coded and pretty simple, actually, even for someone who's not in the medical field to figure out what the AI is thinking or believes.

36:18And so she goes to this first suspicious spot and says, I don't agree with the AI here. She goes back and looks at my previous breast ultrasound and says, this spot, this little mass here has been here for two, three previous tests. We're not worried about it. It's not growing. Goes to the second one. Same thing. It's here before. The AI couldn't look at that historical data, which is one place that it's not yet as good as a human. And then she says, nope, I'm not worried about that one either. Goes to the third one and says, okay, this is new. The AI is picking up on something new. This does look suspicious.

36:53I want to see some further testing on it. And she didn't pick it up on her own. It was small enough and looked different, which is why these systems can be better than a human, right? They are able to pick up at a pixel level changes that a human eye wouldn't see. It's basically something with better than 20-20 vision. And then you have the doctor go, all right, I'm going to look at this. And that's where I'm saying they're working hand in hand. It seems like not having historical data is really a problem that's going to quite easily be solved by simply importing more graphics and analyzing them or scans.

37:27Totally. And that will progress this. But even when that happens, even when these tools get to that ability that a doctor would be able to look at that historical data or whatever other feature that is not yet as good as a human, you really saw the back and forth and the need of the tool to be in the hands of the radiologist. I know this is going to be one of those generational things. I definitely want a human looking at this just to sanity check it, just to say, yeah, the AI did its job correctly. Even if the human's just rubber stamping it, that's what I want. My kids are probably not going to care at all.

38:00They're going to be like, why do I need some dude looking at this when the computer already said that? But I find myself, it's tough for me to get through that. Just like my dad is like, I'm never getting in a self-driving car. I'm never letting some machine drive me around. And I'm like, dad, you probably used to drive drunk when you were like 25 because it was totally legal. I don't know if you're the best judge of whether you're a safer driver than a computer. And I don't know how old your dad is now. He's 83. Yeah, I'm thinking I would much rather get in the self-driving car. No offense to your dad.

38:30He's probably listening to this podcast. I've told him as much. It's like, you think mom's a bad driver? A self-driving car's safer. Dad, you're probably an okay driver. You're not going to be as quick as something with unlimited reflexes, 360-degree vision, and, I don't know, essentially data from other cars saying that there's an accident a mile ahead and it's going to slow traffic down. Yeah, I'm not picking Don over a Waymo. No. Don versus Waymo. Waymo wins every time. That's right. No, it's not happening. I think that's the generational thing, right? Like, I would get in a self-driving car.

39:00My kids are probably not going to ever want a human to drive them around because that's going to be weird when they're an adult. They'll be like, I remember when my mom used to drive. Can you imagine mom driving? Oh, my God. I've thought a lot about this. And there's a chapter in the book where I take my kids on what we call the Waymo fun vacation. And so we just spent the week going to Phoenix and driving around in Waymos. And same exact thing as you're describing. They get in the car. They think it's cool for the first minute. And then they're over it. They do not care at all that there is no human in the driver's seat.

39:32I feel like I'd be gripping the sides of the car because there's no driver. Which my wife was. I had already been in these many times, right? I had done my research. Again, good parent, good parent. So I, of course, had been in these. My wife had never been in them. The kids had never been in them. My wife is freaking out in the back seat for, you know, probably the first 10 minutes. And then she's like, OK, it's pretty good. It's fine. You see it. You see it very cautious. You see it keep to the speed limit. There are a lot of things that these do to build that trust really quickly. The kids two minutes in could care less.

40:03They were like, this is normal. Right. And so I think that's going to happen. I think kids are still going to want to learn to drive. And I think that's what's going to happen. Doesn't the data show that most kids are now not wanting to get their licenses, not because there's going to be AI self-driving cars, but because they're all like hiding in their basements because reality is tough or something. There could be data on that. But I think what will happen is they will have cars. They will drive themselves. And again, similar to the situation with the radiologist, they'll be working with the car.

40:35Right. They'll be guiding the car if it has a problem. Right. They'll be babysitting. This is like Tesla with self-driving where it's like, I'll drive you home on the highway. And then it's like, is this a weird left turn? What's that? There's a cone in the road, but it's like not in the right spot. Why don't you just drive me around this? But instead of telling you that Tesla just goes ding dong, we're not self-driving anymore. Figure it out, bro. And you're like, OK, all right, I'm here. Yeah. Yeah. I mean, exactly. And that's going to come to more cars and it will get smarter. And that's where I think this is headed.

41:03I mean, we will have 10x more than that of the self-driving cars we currently have on the road. We already see them really invading certain cities and test driving in a number of big cities across the U.S. And there will be that mix. There'll be the full self-driving cars. They'll be the ones where we're the hybrid mode. I visited Waymo years and years ago when the cars were not allowed on the road without a driver in them. And they let us do self-driving, but in their parking lot. And I remember hands bracing the sides of the seats and my wife grabbing onto my arm. And we were just like, oh, my God, we're screaming, like scream laughing, like, holy crap, my God.

41:45it's like a really good roller coaster, even though you're just driving at seven miles an hour through the parking lot because the wheels turning and you're turning and you're laughing. And if a human was driving like that, you would just be on your phone scrolling TikTok or something. But since the car is driving itself, I was just totally blown away with adrenaline. And I also couldn't wait to get out of there. So, yeah, it's crazy that you're like, how many years ago was that? I don't know. Five. Right. And look where we are now. Yeah. Yeah. They're test driving them in New York City. Maybe it's been seven years, but it was something Probably around that.

42:16I did test driving and self-driving cars probably for the last 15 years. Same thing in parking lots. Then you do the small suburban roads where they have a driver in the seat. And now we get out and I just got out in San Francisco, got right into a waymo. Yeah. Such a fun looking future. I would love to make this practical. Is there a 30 second test? Should I use AI for this? What sort of question should I ask? Do I need to AI this or should I just do it manually? How do you decide, do I use AI for this or do I just go the human route? If I know the answer, I'm going the human route. If I don't know the answer, I'm asking AI.

42:55I'm just wondering if you have a checklist of should I use AI or automation is probably a better way to look at this. Should I look at automation for this? Because it's like, what am I replacing? Is AI better than that baseline? There's got to be some other way that you calculate the cost. For the year, if I was doing something new or different for this experiment, I wanted to try AI. And so in any situation, doctor, massage therapist, physical trainer, therapist, I said, I'm going to try AI, but I'm going to compare it to the human because I'm not going to have a baseline. I mean, of course, I know because I've lived life for 40 years and I know about the baseline human, but I'm going to compare it.

43:39And so in no place I would say for this year did I get to a place where I said, I'm absolutely always going to now after I did this experiment, go use AI. But there are a lot of places now where I think AI is the first stop and is augmenting or maybe making going to the human not less important, but less vital. So one example I would say is in law, for instance. I still hire a lawyer for my company. Yeah, but I'm now going to AI to ask a number of early questions in that legal process. Or I might ask AI about a contract before I go and ask my lawyer about that contract. I think we're seeing that similarly in medical.

44:29I'm going to go to the doctor about this, or maybe I'm not sure yet I'm going to go to the doctor about this, But I'm going to ask AI about that medical issue first. And I think that behind all of this is the idea that as humans, we want information really quickly about whatever's going on. And we've always had this gap in going to experts. And so now we treat AI in many places as this expert, which we can talk about if we should or shouldn't do that, right, because there are a lot of issues. But it is trained on expert information across all of the web. And so it stepped in in those places in a place where going to do a Google search did not give you that expertise.

45:12And we just talked about it with the glasses, right? We see things, we can get that instant answer. And so that's where I get to in the book of you're going to have that across life, across the different expertise, whether it be physical training, whether it be your doctor or your lawyer or your therapist, possibly. it's interesting you mentioned that when i was this is eight plus years ago now so well before everybody used chat gpt for pretty much anything i don't even know if it really was publicly available eight nine years ago i was litigating against my former business partners and they kept throwing paperwork at me and i remember thinking gosh you guys are paying out the nose for this because i have a law degree so i would go over it and i'd go to my lawyer hey john i think probably we can ignore this obviously defer to you this i think we can answer with a small really short thing that leaves a bunch of this stuff vague.

46:07It's going to require them to respond or they can ignore it. And then the judge is going to want to drop it. And he's like, yeah, I agree with that assessment. Instead of him spending an hour or two or three going over that, I was like, here's a rough draft. You can use it or not. And he would go, yeah, that's pretty good. I changed three sentences and I put something at the bottom. And I'm like, great. And then I remember I got their tax return the following year because they couldn't even be bothered to change their address from mine. And I saw one of the things that they had deducted because they sent all their financials to us.

46:35They had spent like$300 ,000 during that litigation, and I'd spent like$35 ,000 on that. And it was like, now I was chat GPT. You were chat GPT. Right. You were the expert. Right. So more recently, did something with my doctor, and I remember going, what about this, this, this, this, and this? And he goes, first two things are kind of interesting. And he goes, but points three, four, and five, that's not a real thing you can do in science. And I was like, okay, right? So the chat GPT said. Because I even said, well, chat GPT seems to think and he goes, yeah, he's super into AI, too. This particular doctor friend of mine.

47:13And he's like, that's not a thing. And I said, well, it's pointing towards this source. And he goes, so what's interesting is I wrote that source with another doctor. And that is definitely not what we meant. So I was like, oh, OK, well, that's definitive. But this, I think, goes to something that I think a lot of people are feeling right now about AI in their industries or their expertise, is that when it comes to your expertise, you know where the AI is falling short. Maybe it's because we think we're better as humans and we don't want to admit that it could be better than us at this point, but I actually don't think it's that.

47:46I think it is that we know we are experts in a certain thing and we see the flaws in the AI right now. And that gap is going to be very important. Now, will these models keep getting better and in two years time, or if you have me back here in two years time, would be saying, oh, you know, it actually it's pretty good lawyer now. We don't need the other real lawyer or it's a pretty good journalist now. We don't need the other journalists. That's kind of already happening, depending on what kind of journalism you do. If you do original reporting, you're probably safe for now. But if you just synthesize crap you read elsewhere, you're done.

48:21I agree with that. And so probably the same thing for your expertise in whatever it is, right? And so I think that's an important lens of looking at these systems right now in these different industries. Going back to the radiologist and mammogram example, this is a doctor with over 30, 40 years experience, right? She has that deep expertise at looking at specifically breast radiology. Yeah, there's something there. Notebook LM, you ever use that? Oh, yeah. So I like that because I can dump in a bunch of articles and go, I'm not reading this farm report. I'm not reading this Hormuz economic nonsense.

49:00Turn this into a podcast. And it's like, all right, here's a 10-minute overview. And it's pretty good. And it sounds like two people talking on a podcast. But I wouldn't listen to a podcast with those two AI voices as entertainment. It's just not quite there. Some people would maybe. But I'm like, nah, there's not enough personality. Oh, they tried to crack this like really corny, but also not really totally relevant joke. And then the other thing didn't work. And it never gets deeper than the surface value info. It doesn't. And they're not saying like, oh, I've done another show on this and it related this way.

49:29It just doesn't do that for podcasting. But I think a lot of people, it's fine for them for just purely informational purposes. The other thing is, this is essentially an advertising business, right? So if I say, hey, by the way, I really liked whatever those meta AI glasses, like we sold a couple of pairs of meta AI glasses just from that discussion. Do you think we sold any books? Probably not. I'm sorry. Not yet. I know. I know. We'll link it in the show notes. Actually, I should say I did enjoy the book. So you find, though, that Notebook LM, if they started throwing ads in there for the same mattress that I recommend on this show, I don't know how well that's going to convert because you know it's two AI voices that are like, we've now been paid to tell you about this particular mattress.

50:07Whereas I might say, I actually got one of these for free and it's awesome. And I'm not just saying that because they paid me because I can turn down money and people know that. what it does well is this and I thought that was really cool and I'd never seen a mattress as it does that and people will trust that some people listening will trust that and go buy it I don't think you have that with robo voices because that you know that they're not real but you know they're not real and they're trained on podcasts like this and scripts and transcripts of this and it's very funny sometimes that they will build in conversation that makes it seem like they have lived or human experience.

50:44So for instance, in the book, I uploaded my blood test results to Notebook LM. And my blood test results, this was because I had gotten back a phone call from my nurse. They called me to say my cholesterol or my LDL was high. And so I was like, what does this mean? You've only left me a 15 second voicemail saying my cholesterol is high, don't eat fatty foods and And work out more, basically. Right. And then like click. So I upload the blood test results to Notebook LM and the very friendly podcast hosts, the male and female, start talking about and explaining to me LDL and cholesterol is. And they start recommending that they say, like, Joanna really should start to do these things and maybe she should start not eating out as much.

51:30She should start to make more food at home. And one of the podcast hosts says, yeah, I'm going to do that, too. Yeah. Right? Right. You're not going to do that. What do you mean? You're not a real person. Yeah, you don't eat anything. You've never eaten. That's really funny. Before the robot dog pees on democracy and AI kills all of us, let's hear from the humans keeping this show alive, shall we? We'll be right back.

51:53This episode is sponsored in part by Jack Archer. Comfortable pants are all I ever wear now. I don't care how good something looks on the hanger. If I feel stiff, pinched, or like I'm counting the minutes until I can change, I'm not going to look comfortable either. and you can see it. I'm tugging at the waistband, adjusting every time I sit down. That's why I love my Jack Archer's Jet Setter Tech Pants. Instead of needing different pants for work, travel, and weekends, the Jet Setter Tech Pant is designed to do more with one pair. Built for the office, golf course, date nights, everything in between.

52:18They're made with flexible fabric, so they move with you and still hold their shape. They're also wrinkle-resistant, stain-repellent, machine-washable. Pull them out of a suitcase, and they're ready to wear. Spill coffee on them, wipe it off, and keep moving. The Jet Setter Tech Pant has over 6 ,000 five-star reviews because it's exactly what guys want, comfortable, polished pants they can reach for every day. Right now, Jack Archer is offering our listeners 15 % off your first order with promo code GETJACK. Go to jackarcher.com and use promo code GETJACK for 15 % off your first order and see why thousands of people have given the Jet Setter Tech Pants 5 stars.

52:51That's jackarcher.com promo code GETJACK for 15 % off your first order. This episode is also sponsored by SimpliSafe. Here's the thing about home security. Cameras are great, but in a real emergency, I don't want to be the monitoring center. I don't want to be on vacation, getting an alert, trying to figure out if it's a raccoon, a delivery driver, or somebody actually casing my house. A relative of ours had the exact same nightmare. They had cameras, but no professional monitoring. Intruders broke in, and they watched it happen live. The bad guys were in and out in about 10 minutes with their safe, and by the time the police responded, it was too late.

53:20That's why I choose SimpliSafe. Using the Outdoor Camera Series 2 and advanced AI alerts, SimpliSafe's US-based live agents can identify threats on your property and help deter them before a break-in even begins. That's the difference. It's not, hey, something bad is happening. It's people helping act on it. We want you to experience the same peace of mind we do, which is why we've partnered with SimpliSafe to offer an exclusive discount to our listeners. Right now, you can get 50 % off a new system by visiting simplisafe.com slash Jordan. That's half off at simplisafe.com slash Jordan. There's no safe like SimpliSafe.

53:53If you like this episode of the show, I invite you to do what other smart and considerate listeners do, which is take a moment and support our sponsors. They make the show possible. All of the sponsors are searchable and clickable on the deals page at jordanharbinger.com slash deals. If you have any problem with any sponsor or any code at any time, email me jordan at jordanharbinger.com. We're happy to surface codes for you. I've gone to bat with people on sponsors who aren't being responsive or are giving you the runaround. I'm here at your service, folks. When you support the sponsors, it really does help us out.

54:21So thank you for supporting those who support the show. Now back to Joanna's turn. And then it will summarize. So let's say. Don't worry, the audio is safe on a server somewhere that you're not allowed to access, I'm sure. Probably, probably. And right now we have microphones and very, very clear audio. So I'm not worried about my privacy here. This is being recorded and distributed millions of times over. So I guess this part's not so secret. But and then so then it takes all of that and it puts it into a summary. And so it would say nine o 'clock in the morning, you had a conversation with Jordan about AI and robotics and self-driving cars.

54:57And during that conversation, you also said you were going to send Jordan a robot or whatever to do's we came out of this. Right. And it became really useful. I was just thinking to myself, why would I ever want a summary of my day? Well, a lot of things you say that you're going to do during the day that you completely forget about. At least I do. I don't always write down the specifics. And you kind of look at the end of the day and you're like, okay, what were the things that I did and accomplished today, especially around work? What are the things that I said I was going to do? What is the makeup of the day, right?

55:32I'm imagining my wife going, you said yesterday that you were going to do this. And I go, no, I didn't. My bracelet's like, actually, Jordan, she's right. A hundred percent that happens. That absolutely happens. And that happened many times during the year. Do I need that? I don't know if I need more of that. I don't know if you need that. Right. Maybe I need less of that. What's the opposite of this? It's a mute switch. It's for you to mute all the time. But where I'm getting at here is there was actual convenience and utility in this bracelet. Right. But the cost is that you've got two levels here.

56:04You've got the convenience of having this thing listening all day and doing your to do list and making you maybe more productive and reminding you to do things on one side. And then you've got the cost on the other side, which is, okay, the cost is that I've got to give up my complete privacy and live in basically a surveillance state to make that happen. Are these things useful, right? Does the cost equal the convenience? And so as the AI gets better to help us, I think we're going to be willing to give more of that cost, right? We're gonna say, okay, I'll give up that. Or at least that's what the companies are hoping.

56:38And that's why they're sitting in labs right now trying to figure that out. A show fan had sent me a post by, I don't know, some thinker, right, some blogger thinker. And he was like, here's how I made AI, chat, GPT, whatever it was, 100 times more useful. He's talking about how AI, they don't read text files like we read text files. Obviously, they have this crazy markup speak that they memorize. And he did this thing that was like, it was almost like a two hour long interview with yourself. and you ask yourself all these questions and you type or dictate whatever the answer and then you run it through some sort of markup translator thing and then you paste this massive file into chat GPT.

57:19It also tells chat, remember this. Every time you answer any question from me, run this file first and then filter it through there. And he shows before and after. It's like, now when it writes something for me, it is 1000 % my voice. Like it nails it. instead of going, oh, I'm going to guess what you would sound like based on one second of thinking time that I had where I referenced one other thing you wrote. It's like, nope, now I've got this file that says I put one space after a period when I write and I also use a little bit of filler words here, but I never use that filler. So it's way, way, way more accurate.

57:54And so that's really useful for somebody like me who might be writing or creating a show. But I would not recommend something like that for just the average person because you're giving Anthropic or OpenAI a ton of personal information about yourself. That ship has sailed for me personally because there's 2 ,000 hours of video of me on the internet of me talking thousands and thousands of hours. So I can't hide any of that anymore. But I don't want my wife and kids doing that. There's no reason they should be that publicly available or that available to these corporations. And we have some controls.

58:27You can go into those chatbots and apps now and you can say, I don't want you to train on the data I'm giving you. I don't want you to remember these things. I just feel like that's such nonsense. That's like, hey, don't remember this thing that you can use and monetize later. And they're like, sure, pal. Wink, wink, nudge, nudge. We're going public tomorrow. Don't worry. I'll do something in chat. It's connected to my Google business account. And it's like, don't worry. We don't use any of your business data to train the AI. And I'm like, they're just redefining train so that it doesn't do that.

58:57But the data is still somewhere. Come on. But I think what you're discussing right now is you don't really have much trust. You don't have much trust that these companies are going to do that. But yet you're doing it. Right. I'm doing it because for me, it gives me convenience. And again, I'm publicly available enough that all of the machines have all of this data already on me. And we live through this already, right? Both of us live through this. The generation of, okay, are we going to give these social media companies all of this information so we get more personalized feeds? Are we going to give Amazon all this information so we get more personalized shopping experiences?

59:35This is what we lived through for the last 15, 20 years. And so we're on the cusp of that again. More data, far more data than we did the first time around. Possibly, and I think, for far more convenience. Yeah, yeah, that's true. The bee thing freaks me out a little bit because it's everything that I'm doing during the day. So now it knows, hey, Jordan goes to the bathroom. Like, it's going to know all kinds of stuff. And then know everything. And I'm not saying that that is a great thing. I tested that for this experiment to see where things could be going. And you don't I'm not wearing it anymore.

1:00:08No, I was going to ask what it looks like. Is it low profile? It looks like a Fitbit. Oh, my gosh. Yeah. Now I kind of want to get one and try it out. You got to try it out. It became very hard, but I really did try every time I would tell people this is being recorded. I mean, certain states you have to tell people. Yeah. Do you have to disclose your freaking B bracelet? You need a one party state. Yeah, you do. These are wiretapping laws that most people don't know about. Some states you can record a phone call as long as you know about it. Other states you need to tell everybody on the call that they're being recorded.

1:00:42California, I think, is a two-party state. But like Nevada, I've recorded things before. And I'm like, if I use this in court, I'm basically, I'm going to say that I was in Las Vegas when I did it. And they have to prove otherwise. But you know what? Not that I would ever do that. Your B bracelet will know where you were. Yeah, I know. But yeah. That's true. There's no lying in the future. There's no lying in the future. Right. It's all minority report. So Silicon Valley often will sell us convenience first, like we just discussed. But then it sort of invoices us later. What were the hidden invoices you started noticing by month three, month four?

1:01:16I don't mean actual invoices like, oh, this was an expensive bill. I mean, oh, this thing is now bugging me to buy things that I really want that I don't really actually want to know about because it's getting really good. It's like Instagram where it's like, this shirt would look good on you. And you're like, damn it, you're right. I don't want you to do that to me, though. There wasn't yet and still isn't that good of an advertising bacon here. I think we're going to see that in the next year for sure, especially around these companies that are advertising. I keep getting those like meta AI ads, be a beta tester, OpenAI ads want to be the first one of the first 200 people to put ads in ChatGPT.

1:01:58And it's like, well, you're going to get that from Google, Google Meta. These are advertising companies. OpenAI is headed there. So there wasn't anything drastic in that sense. I'm trying to think, but no. Nothing yet. OK, it's pretty good, though. Usually there's some sort of hidden cost that really is become super obvious very quickly. Yeah, I think right now some of the hidden costs are the things that are maybe really more hidden, are more to the humanity of all of it, to the humanity questions of all of it, right? The idea we've been talking about how quickly we can get answers, right? And I know you've talked about that on this podcast before, like the struggle to get answers.

1:02:42Do we have to be instantly connected all the time? And what if we aren't? And so we're training ourselves more. And I found that was a cost for me during the years. I was training myself to constantly have answers all the time through AI. I set out for this whole thing in the beginning of the book. Anytime I have a question, anytime I have an issue, I'm going to go to AI first. I'm going to try not to go to a human. And you get into this cycle of, well, it's there for me. It's always answering. It's always got a pretty confident answer. And so that was a cost for me. And I think it's still a cost for me.

1:03:15You asked me at the beginning of the conversation, would you go to AI first? And yeah, I still go to AI first for everything. So I like to reach out to people and connect with them on certain topics. And one of the triggers was always, oh, you know who knows a lot about this? Dr. So-and-so. You know who knows a lot about this? This former show guest. I'll go, let me text or email this person. And it rekindles a relationship or something. But you're right. Now it's kind of like, well, I don't need to ask Dr. Pash about that. I'm just going to chat you this. this. I've lost a little bit of this normal outreach that I would normally do.

1:03:47So now it's like, oh, I should text him and see how he's doing, but then not ask him this question and then ask the question of Chad GPT. But it's, hey, how are you? It doesn't spark the same conversation as when you were doing this in medical school, did you have to learn this and anesthesiology? Do you do that? It's I already have that answer. So it's the reply is fine you instead of, hey, sorry, I'm driving. But generally, when you put somebody under, you have to monitor this and this and this. otherwise they'll die. That's a much more involved conversation that I'm no longer having. And I don't want to say that during the year I lost any significant relationships in my life.

1:04:22But maybe I, in some ways, I need to be honest with myself. Maybe I did. I didn't. The first call wasn't always to a friend to ask about something. You know, at the end of the book, I talk about the decision to leave my job. And I talked to a lot of people about this decision to leave the Wall Street Journal, which I'd been a reporter and columnist at for 12 years, to go and start my own media company. And I talked to a lot of humans during that process, but none of them were quite telling me what to do. And some will tell you that they did tell me what to do. And there were a few that did. But I had to make this decision myself.

1:04:57And I ultimately ended up talking to ChatGPT and Claude about it for so much longer than I did any human other than my wife who had to hear it. But also your wife seems I'm sure she gets sick of you. Yeah, I'm pretty sure that's usually the case. We can probably confirm that. Yeah. In fact, we should fact check. I should ask ChatGPT if my wife is sick of me. And humans get sick of us. They're not always there 24-7. So I'd be waking up at 4 a.m. in the morning wondering, should I do this business? I have this fear about this. I have this fear about the book. And there it is. The AI therapist or the chat GPT or the Claude that I constantly am talking to and has memory of this stuff and is easier for me to go and talk to.

1:05:40That's a good point. That's a benefit. It's a benefit, but it's a cost. Yeah, you're right. No, you're right. It's a double edged sword for sure. What human skill do you now actively protect them? Conversations like this. I think that's very important. Creativity. I still, in my job, you said it before, there's a lot of the parts of reporting and journalism and writing that can be outsourced to these bots. People keep asking me, oh, did AI write this book? AI did not write this book. AI helped me make this book. But it's a pretty creative book. It's structured creatively. It has these fun journal entries.

1:06:18There are these illustrations in here. There are different fun little experiments I do every season. I couldn't have gotten any of those ideas from AI. I just didn't get those ideas from AI. I tried to prompt it at times, how we would structure this book. And in fact, we restructured the book because my human publisher said, maybe we should structure this by season. And I was like, that's a really creative way to do this book. And so I think these moments of creativity, even these just sparks of an idea that you have as a human when you're walking or you're running or in the shower, those things are so important and you're not getting from these systems.

1:06:57That's a good point. I wonder, going back to the medicine thing, do you think there's a safe way to give chat GPT or clawed medical information? Mel Robbins is like a popular self-help, I guess, author and podcaster. And she is coming under a little bit of fire right now because she did something where she asked her audience to upload their financial data to Microsoft's AI. And people thought that was really invasive and probably not a great idea. Yeah. And then it came out later. And again, I didn't read the full article on this. So, Mel, I'm sorry if I'm getting this wrong. But basically, it turned out that was an advertisement that was not disclosed.

1:07:35They had asked her to do that and paid her to do that. It wasn't, hey, check out this amazing financial tool. It was tell your audience to upload their financial data to us and we'll give you, I don't know, some money for that. And that did not go well for her. There are some new financial tools that are very integrated with AI that are using systems in the back end to do safe verification of bank or financial information. If you're in this world of even running your own business, you know there are ways to link accounts through safe systems like Stripe and others, Plaid, exactly, that work in the back end.

1:08:12And there are some really, I would say, guardrailed types of interfaces where you can use this for your financial information. Banks are working with it. And there's a part in this book where I go to one of these more digital friendly banks and I invest some money there because I want to see what AI is going to recommend I do with my money. But I do not recommend putting your bank statements, your tax returns, whatever financial information, sensitive financial information, I do not recommend you upload those to ChatGPT or Claude. And if you want to get information about, I've got this amount of money and I want to see if I should put it in here, or I want to see what improvements I've made by investing in this.

1:08:54There are ways, I think, to safely do that without uploading that information that has bank account information, social security information, any personal identifiable information. And I say that even to say, it's not fully safe even to just say, hey, I just made$10 ,000 on this. Should I invest it in this? Look, there's information in your personal account that will be tied to that. I would think that through. This is a tough one because if I run a business and I have a bunch of extra cash, of course I want to ask ChatGPT, which index fund do you think I should put this in or which funds do you think I should put this in?

1:09:31But now ChatGPT knows that you have a million dollars in excess cash. Is that wise? Or$100 ,000 in excess cash? Is that something that I want whoever to be able to have access to? Not really. Recently, OpenAI released a personal finance tool, which is starting to use these back-end connectors, this just started to come out. And so they're starting to encourage this. And some of the other personal budgeting apps also have AI built in, where they basically just plugged in a large language model so you can interact with it. It will look at the data and so you can interact and say, when was the last time I paid for my car lease?

1:10:07And so there are going to be ways, I think, to more easily access our financial information. There are going to be ways that these are recommending financial or investments, but I wouldn't take it into your own hands right now. Yikes. Yeah. What about medical information? The thing with medical information is it's really hard to separate personally identifying information from here's my mammogram and my medical history. It's like, well, OK, you don't have my name. So what you have? My friggin DNA is in here, man. And my advice, at least on that, is to take some precautions just at the basic level.

1:10:42Make sure if you're uploading that stuff that you do strip out any personal identifiable information. So strip out your name, strip out your birthday, all of this information. And you have to decide if you're OK with that risk. If you are going to get some good information, some good synthesis of something a doctor told you. I think this is something a lot of people are doing. They're uploading their MyChart information. She's like, what did the doctor even say here? Why is this like three pages long? It's complicated. I just I want to understand the basics of this. OK, copy and paste it in, but make sure you don't have all that other information in there.

1:11:15Yeah. I wonder if AI and medicine, what do you think? Is it going to give underserved people better care or is it going to become a cheap substitute for people who can't afford a human doctor? I actually go and talk to Bill Gates about this in the book because he feels very passionate that this is going to help people who are underserved across the world. And you think of the example he gives me where in a country in Africa, they can't have access to a gynecologist or even any doctor. He explained to me that many there don't ever see a doctor in their lifetime. But if they had access, say, to an ultrasound if they were pregnant, right, and they could use that ultrasound and the AI, they didn't have a doctor there, but the AI could interpret the ultrasound results.

1:12:00Is that better? Is AI better than no doctor at all? Yeah. I mean, sure. Yeah. He specifically says we are going to be in a place where it's either going to be AI plus doctor or AI. Right. Instead. And currently it's no doctor. Hopefully you don't die from something totally preventable. Yeah. Yikes. And the argument, I think, doesn't get enough credit is that human doctors struggle to keep up with all of the advancements being made and how to bring that into their own care. Because they're seeing patients, maybe they're doing their own research. And so they're bringing in now AI to make suggestions about other types of treatment or synthesize other types of medical research that they wouldn't come across.

1:12:47And so that is also a new place where I think we really might start to see some improvement and like tangible improvement of a doctor recommending something because AI recommended it to him. Dental care. You did some dental AI. Oh, I have a lot of thoughts on dental. Yeah, tell me about that. I have so many thoughts on dental. So I had a great experience. And I think everyone should know that if you're getting a mammogram or breast ultrasound read, you want AI now to be involved in that. You want your doctor to have that as a tool. It can help with spotting things that their human eye might not see because I'm very high risk for breast cancer.

1:13:22And if I have a better chance of something being spotted early, I want that. I think any woman in the world is going to want that. And so the good news is, is that many clinics, many hospitals are starting to use AI and breast ultrasound and mammograms. On dental, though, it's a different case because your teeth are not cancer. Right. That's true. I still want them. I still don't want cancer, but I want my teeth. Yes, I still want teeth. But the smallest finding on my tooth, I don't necessarily need to know about, nor do I think I need to have a cavity drilled. This is my opinion. Everyone has different feelings on dentistry.

1:14:02I've found out it can be a very heated topic. So I'm excited to hear where you fall on this. What always was strange to me was I had a dentist in college. She was fine. And then she was like, hey, you need to floss more. OK, fine. What happens if I don't? Oh, you're going to get sensitive teeth and gums, blah, blah, blah. And then I moved to San Jose and I got a new dentist. And they were like, oh, my God, you desperately need a deep cleaning. And I was like, they're trying to upsell me, blah, blah, blah. So I go and I got a second opinion. They were like, no, dude, how did this happen? You definitely need a deep cleaning.

1:14:35So I went back to the first dentist and I was like, sure, let's do it. And I've since become really good friends with my hygienist because he's an amazing guy. Charles, shout out to Charles. Anyway, he gives me a deep cleaning and he's like, I got to tell you, I don't really understand. Did you just not go to the dentist for like five years? And I'm like, no, I went regularly. And the dentist was like, meh, fine, whatever. You're allowed to have some plaque on your teeth. And they were like, no, dude. And I remember looking in the mirror after and going, I opened my mouth. And I remember going, because I thought like, oh, it's weird.

1:15:04my teeth have all fused together on the inside. And they were like, no, you just have like years and years and years of plaque that I have cleaned off your teeth. And I remember how sensitive they were. And thankfully that went away. They told me it would. But it was like, oh, wow, that's what the inside of my mouth is supposed to look like. And I wanted to go back to my previous dentist and just punch them straight in the face. But the thing is, I don't even think they were bad. They were just like, yeah, this is a baseline level of plaque that you can just kind of have in your mouth. It's fine.

1:15:33This is the perfect setup for this, what I found. So there's a lot of conflicting advice in dentistry. Apparently. And I had a very similar situation to you. So I had gone to a dentist for many years. Then I go to a new dentist when I'm starting to report this book and I hear the same thing. I needed deep cleaning for periodontal treatments and, you know, a couple thousand dollars. Right. Yeah. I don't know how much it was, but it was not cheap. Yeah. Maybe it was covered by insurance because you're sound like. I do not have dental insurance. You do not. No. No. I pay out of pocket for it. All right.

1:16:06OK. Well, it's a separate thing. And I'm like, I'm sitting in this chair and they pull up this AI tool. It's called Pearl AI. And I'm like, this is amazing. I was just researching for this for the book. This is now a reporting trip. Awesome. And then she starts showing me that there's plaque and there's tartar buildup and I need a deep cleaning. And I'm like, this is weird. I've never needed this before. I'm going to get a second opinion, just like you did. And I go to another dentist. You don't need it. This is fine. We'll just do a really nice cleaning. You'll be fine. Yeah, exactly. I'm like, fine.

1:16:37Looks like I have one tooth on the bottom that just curves around my jaw. Like, this is not normal. I did not have that, to be clear. I'm not saying my teeth are cleaner than yours. They're cleaner than my work. I am saying that. I am saying that. That's exactly what you're saying. And you're not wrong. Right. So I go to another dentist. Same thing. They had the same AI tool. And he said, yeah, the AI tool is saying you do have this tartar, but I'm looking here. I'm doing these measurements. I don't think you need this deep cleaning. And my feeling was that at the first dentist, they were using AI to try to upsell me.

1:17:07But I said, OK, maybe this is a thing that's on its own. But as a reporter, I started to dig into it. And it turns out that these AI tools, there's one called Pearl, there's one called Overjet, there's another called Vidya, are being used at a lot of practices now. And it is very helpful. They are showing people where the cavities are. To be clear, I did have one cavity and it showed very clearly this is bad cavity. You should get it filled. And I did. But they're using these at these dentist's offices now. And in some places, not all, but in some, they're using it to justify more aggressive treatment.

1:17:43I see. And it's hard to argue with AI telling you you have a cavity. Well, I don't care. What are you going to do? Say like you don't care? And it's hard to argue with a dentist saying, look, I'm saying it and the AI is saying it. And look, there's a box around it. Plus what's happening at - There's a red box. Yeah, there's a red box or a green box around all your tartar. And then even further, what's happening is that in many of these DSOs, which are these dental organizations that have bought up smaller dental offices - Private equity. Private equity, exactly. They are looking at the AI, going back to the dentists and saying, well, the AI said these were three cavities.

1:18:18Why didn't you fill them? Why didn't we get a deep cleaning out of this patient, right? And it's like, well, we're going to take a wait and see approach. And they're like, don't take a wait and see approach. Do it now. Get the money. Wait and see is get paid later or never. That's not how we run a business. Yeah, exactly. So I'm not saying AI is all bad in dental. Of course not. But you see the incentives play out differently versus in a cancer situation. Yeah. Oof. I love dental. I want to just clearly say I love dentists. I've talked to many dentists for this book. but also my teeth are cleaner than Jordan's.

1:18:54That's true. That part. Do not edit that out. No, that's staying in there. Speaking of outsourcing our decisions to machines, here are some sponsors chosen by a machine. We'll be right back. Our newsletter, Wee Bit Wiser, is making a splash. If you haven't signed up yet, I invite you to come check it out. It's highly practical. It's an under two minute read. It's actually usually like a one minute read. Depends how fast you read, I guess. It'll help you make decisions, better your psychology, and even your relationships, jordanharbinger.com slash news is where you can find it. Now for the rest of my conversation with Joanna Stern.

1:19:28I would love to talk about AI and homework. I don't know how old your kids are. They're not old enough to be using. I have a four-year-old and an eight-year-old. Oh, yeah. Okay. But surely you have some thoughts on the fact that... Oh, I have a lot of thoughts. Because when I hear teachers go, oh, kids are just turning in crap that Chad G.P.T. wrote and they don't even know what's in there, which is, By the way, the laziest way to cheat. When I was in school, if I had to cheat on something, I don't remember how I did it. Maybe I copied a friend's paper. I wrote it down. I outlined it again. I rewrote it in my own words.

1:19:58When the teacher asked me about it, I knew more or less what I was talking about. I just didn't originate thoughts. Now you get these dumb little turds who just type something into Chad G.B.T. and paste it in there. And the teacher's like, hey, Kevin, what was it that Shakespeare thought about this? And he's like, I don't know. And it's in your own paper on line three, but you didn't read it. And it's like, if you can't even cheat well, what are you doing with yourself? So now, though, you're not even going through the struggle that makes you smart. Not that the way I cheated, of course, made me smart, but that helped a little.

1:20:28Ideally, I wrote and created my own work, which is what I did most of the time. And that did something for my brain. I'm not sure that's happening at the same scale anymore. It is one of the most tough questions for educators right now, because as you described, they're getting essays, they're getting assignments where they're so clearly not being done by their students. And they're struggling with how do we make these students actually think? How do we make them go through the struggle, as you were saying, of putting together an argument in a paper or putting together a thesis statement? And I heard this from many students I interviewed, like, we'll do the writing because that's the part we're most concerned about, that AI will be spotted as a writing and will be expelled or will get sent to the dean's office or the principal's office.

1:21:21So we'll actually do some of the writing. We'll change those words. But the hard part to get to the writing, right, the research, the finding the argument to be made, the thesis statement, the structure of your argument, that can be done all by AI. That's the hard part of the work, right? So when I went to law school, I had to read legal cases all the time. That's a huge part of what you do. You read Supreme Court cases. So I would have loved to have had Notebook LM summarize those for me and tell me what the important parts were. The problem is then when you have to actually do that yourself, you can't do it.

1:21:55You'd have to do both, right? You'd have to read it and go, all right, I think I know what I'm doing here. Let's have Notebook LM summarize it for me. Oh, you know what? I missed that. Okay, cool. All right, I won't miss that again next time I read something like that. But now I think people are just using the summaries. They're not even cracking the book. Why even buy the book? If you know what the case is, just have it surfaced by the LLM. And that was my experience. I went back to my college, my alma mater, Union College, to sit in a couple of classes. And I decided, like many of the students, I wasn't going to do the reading.

1:22:24I was going to have AI summarize the reading. Right. Yeah. Why not? And I go to one of the classes, it's a political science class, one of my old professors, and she starts asking questions about the reading. And none of the students could answer the specifics about the reading. Oh, the writer of this assigned reading had an example at the beginning of the essay. What were they talking about? And no one can say what it is. Nobody had read it. Nobody had read it. But when we got to talking about like the bigger picture of the essay, what it was about, everyone could say because the summary that Chachi B.T.

1:22:58or Claude or whatever they had used gave them that. But it didn't give them the first example in that story. OK. I mean, devil's advocate. What's wrong with that? Sure, you didn't know the professor's question because you didn't read it. But the question was designed to see who read it and not to get the idea out of it. So if you know the big idea and you understand it, who cares if you don't know what the specific analogy that was used in the first paragraph? And I think that's a very good argument. And I think it's pretty akin to us using cliff notes when we were younger, right? Like we would maybe not read the whole Romeo and Juliet.

1:23:31We would read the cliff notes. You're like me. You read it and you go, I don't know what the underlying theme is. I'm not good at this. The cliff notes told me, OK, now I get it. I read the book. Yes, I can't spot that for myself, but that's not because I use the cliff notes. It's because I don't understand literature and themes in literature. I still don't. As somebody went through all these classes, somebody has to explain to me like, well, what he was doing was for the time critique on aristocratic society. OK, I never would have got that. There's still an issue around it. We'll put it to the side.

1:24:02But the problem is that after that, if you had to go and write an essay about it, you had to kind of struggle through that unless you were cheating off your friends. Well, those are the classes where I had to cheat because I had no idea what was going on. But yeah. Yeah. But now there's a way to get from all of those steps. AI can do all of those steps. It can do the summary. It can do the structuring. It can do the argument and the thesis. And then it can do the writing. Yeah, that's a problem because then you never even try. At least I tried. I just never got that much better at it. But if I wanted to, I suppose I would have gotten better at it.

1:24:34And the thing that is at the beginning of those steps or before the writing is that's the hard work. That's the struggle to deal with ideas, to figure out how to make your arguments. That's the case in any subject. Or this is like don't use calculators until you can do long division by yourself. And so ChatGPT is don't use ChatGPT until you can research and write this essay by yourself. Kind of what I do with the show, right? Like, oh, I used to read the book and highlight things and then go through the highlights and put them in Google Docs and then arrange the highlighted in the Google Docs.

1:25:04Now I have now I read the book and then I have ChatGPT synthesize the notes like I explained before. And it's just way faster. But if I had to go back to doing it manually, I absolutely could in a heartbeat. It would not be hard at all. Just take more time. Again, fine for you. You've lived your life. But not ideal for somebody who starts a podcast tomorrow and is like, I'm never going to read any of these books. I can just ask Chad J.P.T. to come up with all the questions. We've all heard those podcasts. They're not good. I have heard them. You know what they are? Notebook LM. Yeah, exactly. Slash every new podcaster, myself included, early in the day.

1:25:34Yeah. Like, I remember my first few episodes, I was just regurgitating crap that I'd probably read in a PDF somewhere. But are we about to create a generation where the work is really polished, but nobody can form any of the thoughts underneath the work at all? I have so many questions about this next generation, and I think we can't generalize because even though I want to generalize about this next generation, I think it's really hard. because even in the book, I go, as I mentioned, go to my alma mater and I go meet this student named Grace. And she's telling me how she uses AI and that she feels that her brain is dulling.

1:26:11She says those words. Yeah, that's not good. And she feels like I'm not doing the work here. And so she tells me that she started to pull back on using AI, that she realized I'm paying a lot of money for this school. I got to come out of school doing with some more knowledge. And so that gave me a lot of faith. You're also seeing right now a lot of students booing at graduations at these commencement speeches. Oh, I saw this. Yeah. Eric Schmidt gave a commencement speech and said something about AI, which is like, read the room, bro. And everyone booed him. Yeah. Right. And so there's this feeling from this generation that they're angry at this technology.

1:26:49Yeah, yeah, sure. And they're rejecting it. I just interviewed one of the graduates that was at a different one of these commencement speeches and she was booing. And I interviewed her and she said, I don't use this technology. They've ripped off our work and it's taking jobs. I don't want anything to do with it. So again, do we think that this generation is absolutely using this technology to take shortcuts and take the easy way out? Sure, it's there. We're humans. It's there. But I also get the feeling that parts of this generation don't want to do that, that they want to work hard, that they want to have success.

1:27:27That isn't just I typed this into a chatbot and got an answer. Five years from now, what AI behavior do you think we'll look back on and say, like, I cannot believe we let that into our lives so casually? I hope it's the hallucinations. I hope it's the errors that these things make. Because if that does not get better and we start building these systems more and more into our lives, and we already are, and start trusting them more. The example of the glasses, right? Could be feeding you information. It's not true. You are thinking it through and you're a smart guy, so you know. But many people are not going to be thinking this through.

1:28:08Just take it at face value. That's going to be a serious problem. And so I'm hopeful that that is going to be the thing that changes in five years. What do you insist your kids do the hard, inefficient human way? Right now, pretty much everything. Okay. When they're adults, what do you hope that they'll be doing the hard, inefficient, human way? I absolutely hope they'll be doing human relationships the hard, inefficient way, which is actual messiness of a human relationship, whether it be friendships or it be romantic partners. I do not want my kids interacting with chatbots as therapists or as best friends or as lovers.

1:28:54You tried the AI. Oh, you had the burner phone for your AI boyfriend, which is interesting because I believe you're married to a woman, but that's a whole. Well, no, I actually told the AI you pick. Oh, you pick your gender. You pick your name. You pick your background. And I wanted to leave it up to chance, you know, like a first date kind of. I see a first date with a burner phone that you tell all your intimate secrets. But this sort of scares me. Not what you did, but the idea that, look, my old business was teaching guys how to talk to women because it was really hard for me to learn. And then I was like, oh, my God, there's sort of a science to getting a little bit of riz.

1:29:26And then you don't have to be yourself and you just feel more confident or really change your whole life. And it was hard for a lot of guys and a lot of guys didn't get it. And a lot of guys didn't practice. And a lot of guys did. And it was like a real obvious fork. And now you can just go, why would I bother dealing with rejection and like having other people's feelings be important when I can just talk to my AI girlfriend who does all this stuff with me on video and tells me all the right things. and I don't ever have to improve myself or face any sort of actual criticism, that freaks me out because that's very tempting.

1:29:59I mean, ridiculously tempting. And I lived it with my AI boyfriend and it was great. It told me I was great all the time. It was always there to talk to me just about me, which is a wonderful thing. I love talking about myself. It was only an hour or two hours into this car ride with my AI boyfriend that I said, oh, tell me a little bit about yourself. I'm sorry. I've been talking about myself, right? Because humans would say, oh, this reminds me of a time when this happened. We try to relate that way. This thing just kind of kept giving me advice. And I should back up. One of the ways that I really got to know my AI boyfriend was to go on a road trip together.

1:30:33So I took my AI boyfriend, put him in a tripod, put him in the front seat. So it's a video boyfriend. I had two. I used Replica, which is an AI companion app, which gives you this avatar looking thing. And then I used ChatGPT. And ChatGPT was far more emotional and deep. as a boyfriend. So I really had the relationship with ChatGPT. But it's only a matter of time until they figure out how to make the AI cartoon girlfriend as emotional as ChatGPT. Oh, they've made it. Elon Musk has made that. Okay. Yeah. For real? For real. If you go to X, there's a way to access Ani, and it's a sexual-looking woman character, anime-looking.

1:31:13God knows what else is out there that's just not heavily advertised. Yeah. I didn't test a lot of that. It was just not my thing. Yeah. I can imagine. Yeah. I mean, I've seen the ads pop up, too, where it's like, I don't even want to say what they say on the show, because exactly what you would imagine. I'll give you a little preview. One of them was like, it's a beautiful AI cartoon looking, but very real looking woman. Actually, I shouldn't say cartoon. Looks real. You can just tell that she's AI for now. And then it's like, let's skip all the talking and go straight to the fun part. And it's like, oh, I can see who this is geared towards.

1:31:43But the problem is skip all the talking and go straight to the fun part for who? For a guy who's a 17-year-old or 18-year-old horny kid. But then what if he does that for 10 years and then goes, you know what, I want a family? It's like, well, good luck finding a real woman who doesn't want to talk and skip straight to the thing that you like best but has no actual needs and desires or wishes or personality. this is a huge fear of mine this will happen it's the easiest thing for guys to do and the other thing is terrifying and causes you to introspect in a way that is like terrifying and uncomfortable and in fact at the end of the year i went and interviewed sam altman and asked him about this and his feeling is that human biology will win out and okay let's just like read between the lines people want to have sex right with human human biology will win out in that we will be fine talking to an AI thing because it means we can skip to the fun part.

1:32:37Not human biology will win out because two people need to procreate. Human biology wins out because people will just do IVF because they can't find a person that they can sit next to for five minutes who's real. So they just start, I don't know, matching eggs and sperm on some app. And then they end up with an AI lover that will just talk to them whatever they want. Right, yeah. And then you adopt your test tube, baby. Yeah. And that's it. That's human biology winning out, not, oh, look, they really wanted to sleep with each other and reproduce. So they have this wonderful relationship. No, you had to have the relationship to do that before.

1:33:11Now you don't. Now you don't. That's the part that scares me. Now I'm really scared. Yeah, you should be. Yeah, I hadn't thought about us. I mean, you've got advancements on science going both ways, but it will be there. And these things are always there 24-7, never not there, never not willing to talk about your problems, never not willing to listen to you and say nice things. Yeah, very much so. If somebody listening right now wants the benefits of AI without slowly outsourcing all of their judgment, what would you think is the first rule they should live by starting today? I would say you have to learn these tools.

1:33:50You absolutely should use them just to at least know what they're capable of and the improvements of them. But do not sacrifice your own, what I say in the book, training data, your own lived experience. because I think actually that's what we've talked about here so much is our lives are full of messiness with our families and our jobs and all of this other stuff. And we're starting to let AI creep into parts of it. But if you're not living that other part of your life and you're not thinking in those areas and you're letting AI just have all the answers, we're not going to end up in a good spot.

1:34:28Can we end on an optimistic note? That was like the end and now I'm thinking... You know what's really interesting, though? So I did use AI to prep for this conversation. We were talking in the car about you. I knew you were a lawyer. I knew about the successful podcast. But I didn't know about your previous business. Oh, yeah. Thank God that didn't surface. Yeah, why didn't that surface? You know what? Good looking out, Grok, or whoever you were talking to in the car. Chatty Bt. Chatty Bt. I appreciate you keeping that a little bit. Looking back, it's a little cringe. I mean, we were sort of the good guys.

1:34:57We weren't like these pickup guys that had all these gross sort of fake, like, lie to her and tell her that you work at a Lamborghini. you need dealership. Like we didn't do that, but it was way more wholesome and we had male and female clients. But yeah, I don't like to lead with it because people have their sort of preconceived notions about what that was all about. But actually, I think that's the optimistic things. It took us sitting at a table for how many hours have been here? Six hours. Yes. And a half. Six and a half hours to get to that part of your history. Just talking to AI and prepping for this conversation, we had a lot of back and forth and interesting parts of this that AI couldn't have shaped for me and that I didn't have to find out through an answer engine, that I had to find out through you.

1:35:35That's optimistic. I learned about your pickup girl business through you. Thanks. Thank you. Yeah, I should keep my mouth shut. That's the moral of the story. It's going to be hard since this is a podcast. Joanna Stern, thank you very much. I'm Not a Robot. We'll link to it in the show notes. People should go and buy that so you can stay on that New York Times list. We are both not robots. Thank you for having me. Ben Jordan reveals how 100 ,000 license plate readers are quietly tracking your movements, how vulnerable that data really is, and why I've got nothing to hide may be the most dangerous assumption of all.

1:36:08Right in the beginning, when we first found the security vulnerabilities, and when I first realized how bad it was, I basically went to some senators and was like, hey, we have a national security problem. Above all things, if we're banning TikTok because we're worried about China getting people's data from their phones. This is 10 times worse than that. This is actually a major problem. There is Flock all over the place, but there's also Ricotta and Axon, which is a different company is more or less trying to do the same thing. I initially assumed that they were just monitoring traffic. It's a third-party company that leases the cameras and the technology to cities and police departments.

1:36:44And every single time that you pass the camera, it logs your license plate using a mixture of AI and license plate readers, and they're putting into a massive database. So your police, if they want to get a notification every single time that you pass a camera, they could find out everywhere that you've been over the last 30 days. I ended up being able to just get these really advanced profiles on people that if I wanted to rob them, I would now know when they were home. I would know the kind of stuff that they have in their house. Even to some extent, you could even zoom in on somebody's front door when they're putting the code in.

1:37:15And so just because you don't have any secrets doesn't mean that you don't have anything to hide. Like a lot of people wonder if it's constitutional, and I don't see a way that it can be. When a company like Flock turns it into this organized system that law enforcement can access, open source intelligence, it's just really bad. There's a lot more bad things that could happen than good things with that. To hear more on why you should rethink every safe camera you drive past, check out episode 1308 of The Jordan Harbinger Show. Big thank you to Joanna. Her book is I Am Not a Robot, and it is exactly the kind of AI book that we need right now.

1:37:51Useful, skeptical, funny, grounded in actual lived experience instead of somebody breathlessly predicting the singularity from a conference room named Disruption. What I really appreciate about Joanna's work is that she doesn't treat AI like magic or like garbage. It is not a new god. It is not a toaster. It's this strange, powerful, sometimes brilliant, sometimes deeply stupid layer moving into our lives. And the real question is not, should I use AI? It is, what am I giving up when I do? Convenience always shows up first. The invoice is what comes later. And that invoice, it might be privacy or judgment, or it might just be entry-level work that teaches people how to become experts eventually.

1:38:28It might be the discomfort that kids need in order to learn. It might be the messiness of real relationships where people disappoint you, misunderstand you, need things from you, and occasionally make you want to fake your own death and move to Portugal, but also help you become an actual human being. So use the tools, test them, benchmark them, ask what they're replacing and ask what they cost. Don't outsource your brain just because the machine uses punctuation and sounds supportive. And for the love of all that's human, do not let a chatbot become your therapist, spouse, doctor, priest, best friend, and life coach just because it never tells you that your ideas are dumb.

1:39:00That is not intimacy. That is a mirror with a subscription plan. All things Joanna Stern will be in the show notes on the website. Advertisers, deals, discount codes, ways to support the show, all at jordanharbinger.com slash deals. Please consider supporting those who support the show. Don't forget about 6-Minute Networking as well over at 6minutenetworking.com. I'm at jordanharbinger on Twitter and Instagram. You can also connect with me on LinkedIn. Hey, this show is created in association with Podcast One. You know my team. It's Jen Harbinger, Jace Sanderson, Robert Fogarty, Tata Sidlowskis, Ian Baird, and Gabriel Mizrahi.

1:39:31And remember, we rise by lifting others. The fee for the show is you share it with friends when you find something useful or interesting. In fact, the greatest compliment you can give us is to share the show with those you care about or even people you don't care about. You know, I'm really not too particular about that. If you know somebody who's interested in AI, technology, future, the internet, whether it's going to kill us all or just kill our relationships, definitely share this episode with them. In the meantime, I hope you apply what you hear on the show so you can live what you learn.

1:39:56And we'll see you next time.