This Best-selling Author Wrote a Book in 30 Days—With ChatGPT - Ep. 17 with Seth Stephens-Davidowitz

[00:00] I can't even tell you like just how fun it was to write this book. It was legitimately the best month of my life. I think things that used to take me four months or take me four days, four hours, you know, everything was just so much faster. It was so fun because I felt like ChatGPT did everything that was annoying. I'm asked to make a chart of where NBA players went to college. And it was just instantaneously Michigan State's green, UCLA's yellow, Duke's dark blue, Kentucky's a little bit lighter blue, North Carolina's a little lighter blue than that. [00:30] of time, they add up to rapid process. So I spent 30 days to write what I think is my proudest book yet. [00:47] Seth, welcome to the show. [00:49] Thanks for having me, Dan. [00:51] So for people who don't know, you are an American data scientist. You're an economist. You're an author. You wrote the book, Everybody Lies. You also wrote another book called Don't Trust Your Gut. And most recently, you wrote a book called Who Makes the NBA? And you caught my attention because you wrote that book in 30 days. Oh, there it is. You wrote that book in 30 days with Chachi BT. And I was like, oh my God, I need to talk to Seth. So I'm really, really excited to have you on the show. [01:21] I've listened to some of your podcasts and I'm the total agreeer with your main point that [01:25] ChatGVT and AI are just going to transform everything. Thank you. I love it. So I'm curious, just to get started, you wrote this book in 30 days. Where did the seed for that idea come from? How did you decide to try that? Well, I just became obsessed with AI.

[01:42] And particularly there's a tool Code Interpreter. It used to be called Code Interpreter. Now it's just called Data Analysis. [01:48] that does data analysis for you and [01:51] I started playing around with it. I'm just like, oh, my God, this is the coolest thing I've ever seen in my life. [01:56] you know, things that used to take me four months or take me four days or four hours, [02:01] You know, everything was just... [02:03] so much faster. [02:05] And [02:06] So I kind of was like, oh, wow, I could write a book really quickly. I just thought it'd be fun to... [02:10] have a countdown for myself. I don't think anybody else really necessarily cared about my countdown, but in my head, it was really, you know, a challenge. And, you know, [02:18] So I spent 30 days. My girlfriend thinks that I have a tendency to exaggerate. So she's like, you need to point out you were doing some research a little bit beforehand, which I say in the book, too. [02:29] know like i you know but it was a shockingly little period of time you know in the neighborhood of 30 days of full-time work i would say [02:36] to write what I think is my proudest book yet. Like I'm so happy with how it turned out. [02:41] Uh, so it was. [02:43] a kind of a wild process, kind of the future of creativity and everything. [02:48] I loved it. I thought it was really good. I had not finished the entire thing, but I read a decent chunk of it. And you've got a little bit of a Bill Simmons vibe. It's sort of Bill Simmons-y because it's voice-y and a little bloggy and funny, but more data and less pop culture type stuff. Is that what you're going for? Yeah. I got a message recently from a friend of mine. And he said, Seth, I was blown away by your book, Who Makes the MBA? And he goes,

[03:18] I did not feel that way about your previous two books. That's a good friend. So I'm like, oh, I actually, now I do believe you. So I go, I'm like, [03:28] You know, he said, one of the things that blew me away is just... [03:32] So many books, including your previous books, they have... [03:35] so much like fluff, you know, a long story of build up big words. [03:40] And I think... [03:42] You know, this book, [03:43] In part because I self-published it, in part because it was a race against time, in part just because that's my theory of book writing. [03:50] I just kind of limited to facts, a couple of dad jokes and some graphs, like, you know, like, like just really the bare minimum. And I think people actually prefer that. You know, I don't know. The whole book industry seems wrong to me in thinking that people want these long winds up in this story and, you know, descriptions of all the people and, you know, these big words kind of just tell me the point. And I think this book kind of does that. [04:20] That's really interesting. I would love to, I would sort of love to get into the, into the distinctions in your mind between like the sort of self-publishing industry and the, and the regular book publishing industry and all that kind of stuff. But maybe we can save that for a little later. I want to dive in first. Like, it sounds like one of the main things that you use this book for, use ChatGPT for is, is the advanced data analysis. And, and you said it sort of saved you a lot of time. Is that like, if you had to think about how it's impacted your writing process, maybe for this particular book, is that like the main place that you used it or where were

[04:50] You know, that was number one by far. There were some, like, little ones – [04:55] You know, I... [04:56] There was a point in the book I needed, I ranked people on how good they were, basketball players adjusting for their height. [05:03] And I needed [05:04] And I wanted to call the metric something, so I called it Muggsy's after Muggsy Bogues, 5'3", shortest NBA player in history, who ranks number one on this metric, height adjusted players. [05:14] It's like, you know, that that needs to be an acronym of acronym, you know, but I'm terrible. I wouldn't even know where to begin. How do I call? [05:21] you know, name Muggsy to the back of them. So chat GBD came up with the back of them. [05:26] as metric for understanding game given sporting individuals [05:32] effectiveness and size. [05:34] which is just like unbelievably good, mind-blowingly good, in my opinion. [05:39] So, you know, lots of little things. Having it in your back pocket kind of is very useful. You never know where to use it. I don't like the way it writes at this stage. I didn't let it rip. [05:49] on just, you know, write this chapter. I let it write my appendix. [05:52] because I didn't really care about the writing appendix, the actual writing I did, and then data analysis was just like, [05:57] A lifesaver. [05:59] Yeah, that's really cool. I kind of, I mean, it was sort of an eye-opening thing because you have all these charts and graphs about like how for every inch taller you are, your chances of making the NBA doubles. And I'm 6'2", so I'm, you know, I have a better chance than a 5'10 guy. But like, I was like, okay, I definitely never, never could have made the NBA. I barely made the JV team in my middle school.

[06:29] It is wild just that the doubling relationship is throughout the height distribution. So it's obvious that height helps in making basketball, but it's this really consistent relationship. I'm not the first one who's found this. Pablo Torre, David Epstein have talked about this, but each inch roughly doubles the chance to make the NBA. So if you're under six feet tall, you have roughly a one in 3.8 million chance of making the NBA. And if you're over seven feet tall, you have roughly one in seven chance of making the NBA, which is just insane. There's no other trait that gives you such an... [06:58] a genetically determined trait that gives you [07:01] like such odds of becoming [07:03] you know, a deck of billionaire. That's one of the things that, um, that really, uh, stood out to me too, is like you, you had another play. I think this is where the Muggsy ranking came in. You had another place in the book where you're sort of talking about how the people who are seven feet tall, even though you have a, like a one in seven chance of making the NBA just for being seven feet, like they're actually like under, they're not, they're like below average athletes generally. Like if you, if you rank their free throw percentages versus like your [07:33] or maybe they're a little bit worse. That was really surprising. [07:35] Yeah, I would say below average, but just average or a little above average. So they're not that special beyond... [07:41] this extraordinary trait they have of being seven foot tall, whereas a six foot player [07:47] You know, there are world-class sprinters. You know, they leap as high as some Olympic high jumpers. They're incredible shooters, you know, on average. Just... [07:56] Because you're competing, if you're six feet tall, you're competing against millions of other men for your spot in the NBA. If you're seven feet tall, you're competing with dozens of other men for the NBA. So you can kind of.

[08:05] You don't have to be that good. And there are quirks of the game that George Muir's son – [08:09] I think seven foot six man, [08:12] reach the NBA, [08:14] to entirely to a pituitary gland disorder, [08:17] He had a plaseotide gland disorder that made him enormous. [08:21] And that was enough. [08:23] basically by itself, just about to become a competent NBA player for many years. Just height is such an advantage in basketball. [08:32] You have to be tall and you have to come from a country where people play basketball or where it's one of the things. It's tall or come from a country that plays basketball. I see. Because if you're tall enough, it doesn't even really matter in that, you know, like a lot of the... [08:48] greatest, literally the greatest NBA players in history. [08:52] didn't start playing basketball until they were 14, 15 years old. And what happens is they're walking down the street, [08:57] And someone says, you know, have you thought of basketball? You are enormous. And then they start playing basketball. You know, Hakeem Olajuwon, Joel Embiid, Tim Duncan, Patrick Ewing. [09:06] you know, Hall of Famer, future Hall of Fame players, they all started basketball, you know, after the age of 12. [09:14] And, you know, height was such an advantage combined with, you know, reasonably good athleticism that it was enough to carry that to the NBA. And they didn't need to live in a basketball powerhouse. But if you're six foot tall, six one. [09:26] 6'2", 6'3", you have to grow up in one of the few regions of the world that really love basketball. Otherwise, you just play soccer or something else. [09:33] That's interesting. I know you said this is sort of rare, but is there anything off the top of your head where you can, another area in the professional world or elsewhere in the world where like having one trait that is like so off the charts, like kind of catapults you to the top automatically? Or is it usually just like a more complicated mix of factors that are harder to pin down? Yeah, I think it's usually a more...

[09:59] complicated mix of factors, maybe, you know, mathematical ability for being a mathematician, maybe super genetic and many of them their fathers. [10:07] We're also great mathematicians and [10:10] I don't know. I'd have to... [10:13] think, you know, it's a lot of the other pursuits they've [10:15] There are so many things that come into play. So being a president, you know, you need the appearance, you need the intelligence, you need the social skills, you need the height. [10:24] You need kind of a lot of different things, some of them that are genetic, some of them [10:28] can be trained, you need connections. There are all kinds of things that come into play. Yeah, it's interesting. I haven't really thought of any, I don't know that there's anything else that [10:38] gives you a one in seven chance of that level of success. I, [10:42] Uh... [10:43] Yeah, I don't know. I... [10:45] Yeah. Jockey being really short is very valuable, but I think there are a lot more short men. You know, you kind of want to be 5'3", 5'2", but particularly when you consider worldwide. [10:56] there are a lot more 5'2 and 5'3 men than there are 7' men because [11:00] You know, some of the I even talk about the book, some countries in Asia, the average height is five foot four, five foot five. So there are plenty of men. [11:08] that short. [11:10] That makes sense. So I want to get back to some ChatGPT stuff. So you said when it first came out and you're using Code Interpreter, which is now called it, it's called Advanced Data Analysis. Like, what was the thing that like got you or like, holy shit, I can't believe that this is now possible. And this is going to save me so much time.

[11:29] Everyone's looking for like the one thing that blew you away and it's usually not that, it's just... [11:35] thousands of little things that add up to tremendous [11:39] time saving. So just the fact that, you know, [11:42] You could just tell it, download this data set, run this regression, make a chart of that. [11:47] And it does that right away was so wild because, you know, I'm [11:51] not the world's greatest coder. Sometimes I work with [11:54] people to help me, but that's kind of a time-consuming process, a lot of back and forth and frustration. [12:01] But... [12:02] So right away, just the fact that you can talk to Code Interpreter and now advanced data analysis and get Python code back and then results back and charts back and... [12:13] new data sets, merge data sets back. It was pretty clear to me pretty early on that this was revolutionary. [12:19] Right. Okay, that makes sense. So for people who are listening who haven't used it before, what advanced data analysis does is you can say, hey, here's a data set, and then please run a regression. And it will go and write the code to run the regression. It will run the code, and then it will return the answer back to you. And normally that's something that all those steps you'd have to sort of like do manually. And in this case, you can kind of, you can talk to it and say, no, you didn't do it right. [12:49] A, do it this way or draw the graph in this way and it'll modify the code so you don't have to do it directly, which sounds like you love. I've loved it. It's been truly incredible. And I think ChatGPT is the only mainstream model that has that so far.

[13:03] They'll all write code, but none of them run it for you. And so it's a big differentiator. One of the things it did is... [13:11] I'm asked you to make a chart of where NBA players went to college. [13:16] And [13:18] Uh, [13:19] And I asked it [13:21] to... [13:22] make each [13:25] Uh, [13:26] Each college, the color of that college. [13:30] And it was just instantaneously Michigan State's green, UCLA's yellow, Duke's [13:35] dark blue. Kentucky's a little bit lighter blue. Perfect. North Carolina's a little bit lighter blue than that. [13:41] And like, that's not a wild, oh my God, that changes everything. I would have never been able to do this on my own. [13:48] But that's a big saving of time. And when you add a whole bunch of those savings, they add up to just... [13:54] know a rapid process i think the other thing about the process too uh that i want that i notice is [14:00] It was so fun because I felt like ChatGPT did everything that was annoying. [14:05] I don't know if that's something you can relate to, but it feels like, [14:10] You know, what I like to do [14:11] My dream day, my perfect day, [14:14] is give me a data set and let me just think of some questions to ask it. [14:18] And you'll [14:20] like rapidly. [14:22] And kind of in this book, that's what it was. Whereas my usual day of data analysis is merge a data set, clean a data set, look up code on how to make the chart. Oh, the charts are really working, you know, which is not fun for me.

[14:52] example, we spend a lot of time summarizing other people's ideas. So like, you know, I'll read, if I want to write about utilitarianism, I'll like read a bunch of Wikipedia pages about utilitarianism, even though I know what it is just to refresh myself. And then I'll try to like summarize what I've read into a couple sentences so that I can like write the article that I want to write. [15:12] And ChatGPT just sort of does that for you in the perfect way. So you don't have to, you don't have to summarize it anymore. And you can just like, yeah, be engaged in other parts of the process that are actually more interesting for you. And I think your point about... [15:25] Your perfect day and being able to ask questions, I think, is like exactly where things are going is... [15:32] As things progress, I feel like we are more and more living in a world where any question that humans know the answer to is answerable pretty quickly. [15:44] And so the real skill or the really interesting thing is knowing which questions to ask. And I think that that is still left to us. And that's really fun and exciting for me. [15:56] Totally agree. And that is actually something I asked ChatGPT to do. Could you help me ask questions? And it wasn't. [16:01] that good at it. So that was kind of cool. That kind of felt like, oh, maybe I still have some use on this planet. And that is exactly what I like doing. And [16:09] Uh, just [16:11] Yeah, I don't know. It was wildly fun. I mean, the other thing we can get into is the art in the book. [16:17] which maybe doesn't stand out because lots of people are using, you know, [16:20] Mid-Journey and Dali to create art, but that kind of fits into...

[16:24] the fun of the process and also giving you [16:27] Uh, [16:29] So [16:30] Skills. [16:31] that, you know, allowing you to express creativity that you would otherwise be unable to express is [16:37] You know, I've always thought that inside me is a Vincent Van Gogh. Somewhere inside me is a Van Gogh and that [16:43] I consider myself a creative person. I have a million ideas a minute about how things should be. [16:51] But I have zero artistic talent. You know, you asked me to draw a horse. [16:55] I can't draw anything that's in the universe of a horse. [16:59] Now with, you know, chat with Mid Journey and Dali, I say, [17:03] you know, first chapter on genetics, have a piece of DNA dribbling a basketball between its legs. And Dali draws that, you know, and that's fun because that's kind of a creative idea for, you know, what, [17:14] The arc should be, but... [17:16] you know, it's executed by Dali. So it's actually, I'm actually able to express this creativity that otherwise would be dormant. [17:23] Totally. No, I definitely have noticed that too. Like when I break down what this stuff is useful for, it's like, there's three things to me. It's like one, it makes you faster, maybe like 30% faster at what you already are doing. Two, it gives you skills that you don't have. So you can do work that you couldn't have done previously. Like with Dolly, like you might have ideas for images, but you can't make them. Now you can.

[17:53] what kinds of projects you can start because it's much faster to get started on something new. And so you can do net new work and try net new experiments that you like never would have done before because it would have, you know, all of the work is doable for you, but it would have taken you 15 hours to like get set up. And now it just takes like two seconds and you just write one prompt. And I think all those categories are exciting. [18:15] Yeah, I totally agree. What are examples of the third category? I haven't thought as much about that one. [18:21] Yeah, for me, I often will have little ideas for software products that I want to make. So that's a thing where it would have ordinarily... All the programming for that is... [18:33] known. It's not like super advanced. It's just like pretty, um, [18:38] it's pretty, it would just be rote and I'd have to go think about a lot of things and look up APIs and all that kind of stuff. And ChatGPT, you can just write a couple commands and get a very basic app working and then modify it from there. And it takes 30 minutes instead of... [18:57] four days to get started. Yeah, it's wild. I love this podcast you're doing. [19:04] Just because, I mean, it's right up my alley, but I so agree with you that people just aren't [19:09] changing their lives enough [19:11] based on this, it does feel like all the rules [19:16] are [19:17] you know, overnight kind of transformed of, you know, how long a project should take. And what does that mean? You know, like, you know, I think so much about the book publishing.

[19:26] industry because I've written books and [19:28] Dan and I are actually featured at a future AI and book publishing event. In a couple of weeks, we're going to talk about this. I'm so looking forward to it because I just think this industry is [19:39] It's just like, yeah, what does it mean when you can write, I think, a good book in 30 days and [19:45] How does that change the public? That seems to just like totally change all the rules of how publishing works, should work. Like, what does that mean? [19:53] Yeah, it's totally it's a whole new world. I think it's super exciting. I'm really glad you love the podcast. I'd love to just dive into some chats. Like, do you have some use cases of ChatGPT that you can walk us through for, you know, the way you use it to write this book? [20:10] Yeah, some of the art gets a little bit messed up. [20:16] One of the things that's interesting [20:19] is... [20:20] All right. [20:21] So before we get started, so before you dive in, just give us a little bit of background. Tell us what this chat is. Tell us how you started to get into it. Why you started this chat. [20:35] Yeah, so this is I discussed earlier that I felt I needed a backronym for Muggsies, my metric for ranking high NBA players on a height adjusted basis. [20:46] And I had no idea even where to begin. [20:49] So I literally, I don't think if you gave me, this is similar to art. I think if you gave me [20:54] Unlimited time, I'm not sure I could come up with a good acronym for Muggsy's. It's not really in my skill set. I mean, maybe I could in the same way that monkeys could type Shakespeare eventually, but it would take me a long, long time.

[21:08] So I asked, I said, you know, ChatGPT, could you think of an acronym for that? [21:14] And one of the things I like about this chat [21:18] is I think people sometimes think [21:21] that ChatGPT gives you the right answer on the first approach. [21:25] on the first ask that you just sit back and then if it doesn't, they kind of give up. [21:30] And what you actually see is the first answer by ChatGPT, [21:35] was [21:36] kind of not that great mastery, unique growth, game changer, size, adjusted influence, efficiency, scoring. And then it just starts, it goes kind of crazy. You know, and, and, and I, and I said, you know, I explained more clearly what I'm doing. I also noticed that sometimes I'm very rude to chat GPT. I'm not sure how to feel about this. I don't believe it as consciousness. If it had consciousness, I'd feel really bad. Cause I'm kind of asshole to it. I'm just like, do this, do that. [22:06] kind of treat it like a slave. [22:08] Uh, [22:11] And then it gives me one. So I asked it again. No, I mean, give me something to describe the stats, which is eight words in the first letter. First word is M. Second word is U, etc. [22:20] It gets a little better, measures unparalleled growth, signifying individual efficiency and scoring. [22:25] Not really that good for Bugsy's metric. I also say that... [22:29] That should have two Gs, it only has one, and it's not just about scoring. [22:35] You see it tries again. Mastery utilized for gauging games. I complain that doesn't explain the stat at all.

[22:45] Now it's getting a measure of utilized game performing, gauging size, impact, effectiveness, and skill. I said that's close, but please try again. No, I didn't say please. I should have said please. [22:59] And then [23:02] I said metric for understanding a player's game given individual efficiency and size, and now we're getting really close. [23:09] But I say efficiency is the wrong word. Can you give me a different E word that captures overall performance? [23:15] And then it comes effectiveness. [23:18] Uh, [23:20] And then I just noticed that it's actually missing an S. [23:23] So it's still not perfect. It's metric for understanding game given an individual's effectiveness and size, but I need an S word. [23:30] Uh, [23:32] And then it puts the S word in the wrong place. I explain that. Nat, I explain again that I don't that I put in the wrong place. And then I say, try again, offer me 10 possibilities. [23:44] And then I didn't like any of the 10s. I said, give me 20 possibilities. [23:50] And then... [23:51] Does it get it? Oh, and then eventually we land on or I kind of landed more on sporting individuals, effectiveness and size. [23:59] So I think, you know, there's a lot to take from that conversation. [24:03] But the main thing is that ChatGPT really is a tool, and you can't really expect it to nail it on the first attempt. [24:12] It's a process, and there's a lot of error, there's a lot of back and forth.

[24:16] And you kind of got to guide it. You know, you did that wrong. Give me 20 possibilities. Give me 10 possibilities. [24:22] and keep on working until you eventually get there. So it's not that from my experience of ChatGT, I wonder if this is similar to yours. It's not that you just tell it one shot and it's perfect. You got to really start the conversation and... [24:39] you know, build, build to the right answer. Yeah, no, I think you're, I think you're totally right. I think that's exactly where most people get tripped up is they're like, they give it a huge task to do all in one shot and then it doesn't work. And they're like, well, this isn't good. [24:55] Yeah, and then they just mock it. They like laugh at it. It's so bad it didn't even get the letters right. Look how bad ChatGBT is. Well, no, that's just the start of the conversation. [25:05] Totally, totally. No, I think this is a really good... [25:11] example of how you how to best use it which is like give it a give it a task really simple prompt to start and then just keep going back and forth with it you're offering your own thoughts like you offer sporting because you you like thought about it yourself and like that helps it um so i think i think it's i think it's really good i do have to point out i'm professionally obligated to point out you're using chat gpt 3.5 for this [25:35] Oh, usually I use 4.0. I think I was, I think I must've been a time limit. I must've been on my limit. [25:43] uh what i call time out yeah good i'm glad i'm glad because i would have been pretty disappointed otherwise no definitely the book or maybe it's not maybe i just have chat gpt in the

[25:55] It might just also be I had ChatGPT 3.5 on my home screen or whatever. So I don't know if it was using – I don't think it was using – [26:05] 4.0 [26:06] I used 4.0 for almost the whole book, so... [26:09] If it wasn't, it was only because I was on timeout. [26:13] So yeah, it would have been better probably and quicker. Maybe it wouldn't have made mess ups if it was GPT 4.0. [26:21] I do think it would have probably been a little bit better at making sure that it got the acronyms actually right instead of like forgetting what it actually was. And yeah, to your point about being polite, I am. I'm always very polite, but I don't know if that's the right move either. The politeness is sort of like a Pascal's wager for like if AGI comes and it's like threatening, you know. But I think it's also possible that being not polite, it gets it might get you better results because it's afraid of you. I don't know. I just remember I follow Ethan Malik and. [26:50] He just showed these examples where he was super impolite to ChatGBT, and I'm like, oh, I guess that's how you do it. I think I watched that right before I was talking to ChatGBT for this. [26:59] Muggsy's saying, I'm not always so aggressive, but sometimes I am. I'm just like, I don't know. Yeah, maybe I'll follow your advice, Pascal's wager. [27:09] Although, I mean, if they're conscious... [27:11] It's like their experience is negative infinity versus like, it's probably like negative infinity and negative infinity plus one or something if you're nice to them, right? They're working all the time. [27:22] as human being slaves. So maybe Pascal's wagers would just suggest don't use ChatGPT or don't create ChatGPT.

[27:30] rather than be nice. [27:31] to chat. It's totally possible. I think that's an interesting philosophical question. My sense is like if they are conscious, the consciousness is so different from ours that like they may actually enjoy... [27:42] doing these kinds of completions in a way that we probably wouldn't because it's like literally what they've been trained on it's how their like consciousness manifests is like is completing sentences so but you never know but they would also enjoy being treated politely i don't know i don't know yeah you're right it's a thin line it sort of breaks down at that point you're right [28:12] through with me uh let's see so this is a height of nba players and i'm [28:19] One of the things, and this kind of just shows the basics of... [28:23] advanced data analysis for people that don't know. So you ask it, you know, you upload the file, and then it'll upload, it'll tell you some basics about the file. [28:32] And then you just talk to it. Can you give me a histogram of height inches? [28:36] And here is a beautiful histogram of height inches, right? Number of players, how many players at every height, you know, does it really well. [28:43] Uh, [28:45] I noticed [28:47] Let me stop you there real quick. So like... [28:50] How long would that normally take you? Like going from dataset to histogram? [28:55] That's like, that's not that long. That's pretty quick. But...

[29:02] Uh... [29:04] You know, I think the whole process, if you don't like coding as I don't like, it's not just that it's. [29:09] not quick. It's not that how long it takes is just labor intensive. It's just a pain. I can't code for [29:16] five hours a day even, but I can talk to chat GPT for five hours a day, for like 10 hours a day. So it's just kind of annoying for me to do this thing. So this isn't [29:26] I would say the most... [29:27] labor intensive. But then I just noticed, oh, it didn't have [29:32] If you look at the actual histogram, it doesn't have individual inches. You see it has groups of inches. [29:38] And that kind of bothered me. So I wanted to have every inch. So I asked you to do that. You know, can you have every inch of height included? And it does that very nicely. And now, [29:47] Uh, [29:50] Can you only include American-born NBA players? Now it's just history of American-born NBA players. Again, just really nice to just constantly just be asking these questions. [30:01] and having the chart come back pretty much instantaneously. [30:07] What's the most common height and say what it is in feet? [30:11] The most common height is 81 inches. [30:14] Uh, [30:16] Let's see. [30:18] Oh, I asked it to give a label of the most common height inside the bar for six foot nine. [30:25] And it puts the you see it puts the label. I don't this is another thing that's important for ChatGBT is initially you see where the label is. It's not really where I want that label to be. I want it vertically.

[30:37] And ChatGBT doesn't understand that. [30:40] But, you know, I... [30:42] I, uh... [30:44] I ask it to switch, written vertical, and now it's vertical, if you can see right here. [30:50] Most common height, six foot nine. [30:53] Uh, [30:55] So there's a lot of back and forth, and this is getting to a point of things that would take a lot of time. [31:00] You know, figuring out how to label it, switch vertical, get it exactly in the right spot are very important. [31:06] time consuming projects. [31:08] And this is done... [31:11] almost instantaneously, change it to red, and now it's in red, you know. [31:16] get rid of horizontal and vertical lines, [31:19] It's just all kinds of things that, you know, just very, very quick to change the chart, to get to the chart that you want. [31:26] and you know [31:28] uh, [31:30] Again, just very, yeah, very... [31:33] Uh, [31:34] Very, very fun stuff to kind of have these things. I think... [31:38] Eventually, my next chart, which I'll go through a little quickly, is a normal distribution of height of American men versus NBA players. And you see it puts those on the same chart together very, very quickly. Just really, you know, and I didn't even use these in my books, but you're just playing around with it, you know, over time. [31:57] You know, like, oh, the colors overlap here. [32:01] You know, [32:04] They'll make it to their purple when they overlap. It does that. Most common height. It's just really fun to make these charts.

[32:11] You see, it's not just... [32:13] I think what I want people to take from this is kind of similar to the Bugsy's thing. [32:19] That it's not just you tell, you know, this is how this chart, probably one of the better versions of this chart of the normal distribution of. [32:27] the height distribution of men, the height distribution of NBA players. [32:32] And what you see is there's a long back and forth. It's not that you just... [32:36] And part of the back and forth is you're trying to figure out in your head what you want, right? [32:40] So you're trying to figure out how you want this chart to look. Does individual inches matter? Should there be horizontal bars behind it? What should the colors be? Should the colors, when they overlap, be blue or purple or red? All these things are... [32:58] decisions that you have to make in your head that you can't tell ChatGPT right away, [33:03] But the great thing about ChatGBT and advanced data analysis is it just comes back very quickly with those changes. [33:10] And it's a conversation for me. It's a fun conversation. It's not a laborious. Coding is a laborious, unfun conversation because half the time, you forget the coding. How do I code to make just the shaded parts shaded under the region purple? Or how do I make? [33:28] you know, determine the spot to put the most common height? How do I change the font for that? You know, how do I do all these things? And [33:36] You know, it's always looking it up. It's just a pain in the ass. [33:40] that makes the whole process, you're kind of going to the refrigerator a lot procrastinating if you're me doing that. And for me, there's no going to the refrigerator and procrastinating. The only times I stop working is when ChatGBT gives me a timeout. It says I have to stop working and I get disappointed because I want to continue working. So I think what I want people to take from these examples is,

[34:01] is it's not instantaneous and you can't get frustrated. Sometimes the chart gives you truly crazy things. [34:07] It's really a long back and forth. [34:09] But it is a fun back and forth. And, you know, if you know how to work with ChatGPT, it really is a lifesaver in time and doing things. So... [34:19] Yeah, I think those are those are that's like kind of the main point. [34:23] that I want from people [34:26] to take from [34:28] from my use that I took from using ChatGPT to write this NBA book. [34:35] That makes a lot of sense. There's so much in here that I feel and that I agree with. I see you playing here in this really interesting way. And I think your point about programming is really interesting. I do feel like what ChatGPT has shown us is how much of programming is looking up answers on Stack Overflow, usually, or Googling around for them, basically. [35:00] And what ChatGPT does is it sort of gets rid of the need to do that. And I think what you're saying is what that does for you is it takes this process that you would have a question and the question, having a question is very energizing, it seems like to you. But the process of getting the answer is... [35:17] used to be sort of like energy sucking for you because you don't really want to be googling around or like typing in code or whatever. [35:24] And the same exact process is now giving you energy because you just like having a conversation. And I think that's really cool. It turned the act of getting answers into something that you actually enjoy instead of something that you have to get through in order to get the result you wanted.

[35:44] That's exactly right. So how do you write a book in 30 days? Well, one of the ways to write a book in 30 days is to work 13 or 14 hours every day. And again, I could not work 13 or 14 hours every day. [35:58] if I had on a normal coding project, because just too much of a pain in the ass and too annoying and too frustrating. [36:05] And [36:06] This was just... [36:08] a joy to me, the whole thing. I don't know. I can't even tell you just how fun it was to write this book. It was legitimately the best month of my life, I think. I mean, partly because it's about basketball and obsessed with basketball. [36:23] But I don't know, maybe I'm unique, whatever it like those chats are. Those were just super fun to me, like even just adjusting the chart in that way, but without the coding part. [36:34] Like I am a perfectionist. I like like, you know, going things over and over again. But the coding part is just so not fun that that process is not fun. And the process with ChatGPT is fun for me. So. [36:46] Kind of how do you communicate the information in a way that we most [36:50] Uh, [36:51] compelling to other people. Yeah. That's the thing that I think is also interesting about it that you mentioned is having the conversation and having it do something. When it does something for the first time, you sort of notice all of these constraints or all these things you want to change that you wouldn't have been able to specify before you started. And so it's sort of helping you understand what you want as the conversation goes along instead of having to specify all those

[37:21] you're like unraveling what it is that you want or what you care about as you're having the conversation with it. Yeah. And in some ways, normal coding is the same way. [37:28] But just the difference for me, [37:31] in this process of normal coding is night and day in [37:35] how much more joyful and free and fun it is to use ChatGPT. [37:39] versus editing and adjusting your code. Half the time you adjust your code, you still make a mistake. Where's the bug in the code? [37:45] You know, it's just... [37:47] Thank you. [37:47] And going from... [37:50] You know, looking up the code, [37:53] you know, writing the code, looking for your bug in the code, adjusting the code, [37:58] So, ChatGBT doing it instantaneously, even if that's just... [38:02] 20 minutes. [38:03] to one minute [38:05] It is changing something that's 20 minutes, then a 40 minute break where I need to recharge myself. [38:11] So one minute and then another one minute and then another one minute and then another one minute of a new chart, new chart, new chart. And that is like night and day and how in the research process, right? [38:20] Totally. So curious, do you have other chats you want to show us? Those are the two I wanted to show that made the biggest impression. I think a lot of them are... [38:31] Let me see. [38:34] And you don't have to if you don't have anything. Yeah, let me see if they're... [38:38] Yeah, I'll show you one more. This one actually is an example of it working very much simpler, which sometimes happens too. [38:47] I have a whole section of my book on the names of NBA players. It's actually basketball fans will find this, I hope, interesting.

[38:53] is that I called a chapter, [38:55] Uh, [38:57] Why is... [38:59] Chris, the most popular name in it among black NBA players, [39:03] And the reason for that is it's a sign of the demographics of NBA players. So African-Americans from richer demographics, upper middle class demographics are much more likely given common names, Chris, Marcus, James, Mike. [39:14] African-Americans from poorer demographics are much more likely to be given unique names. LeBron, when he was given that name. [39:20] uh to to brick a shaw names like that and if you look at nba players they're disproportionately common names not unique names so i thought a fun way to show that is just make a word cloud again [39:32] I could look up how to make a word cloud. It's not that hard. But it's a little bit annoying because to make a word cloud of first names, my data set doesn't have, you know, [39:41] just... [39:42] It doesn't have first name, last name. It has just name of NBA players. [39:47] So you have to first separate them, then you have to look up how to make it work. It's not a big deal at all, but it's just that UGG factor that [39:57] that leads me to procrastinate and check what's on the New York times. Cause I'm just like, I don't really feel like doing that right now. And now instead, this is very fun. So, you know, limited to years between 1970, 1998, and you know, [40:11] a word cloud of first names of black players. Don't include anyone with initials because a lot of players have initials. And you see you have this very nice word cloud. [40:20] But there are a couple of things I want to exchange. Change only include those born in America.

[40:26] Again, really, really simple. It doesn't matter if the data set is labeled United States or USA or anything like that. ChatGPT will instantaneously... [40:36] figure out what's the label of America. So right now we get rid of all the foreign-born players, [40:42] If you notice this version of the chart, [40:45] Oh, I want a title. OK, you know, makes the title. [40:50] You notice this version of the chart, Chris, is the main – [40:56] a player, [40:57] The most common name is vertical. And I just thought, actually, it'd be nice to have that horizontal. [41:04] Because the chapter is about why are so many NBA players, Chris, this is another thing that I would never think of. [41:09] You know, disgusting first, but if it's all about Chris, I should have Chris, you know, more prominently easier to follow display. So that's another one. It's just, I hope coders understand just like all these little things. Okay. Let me look how to make sure Chris is horizontal. That's another thing. [41:23] little time on Stack Overflow or a little time here and a little annoying and a little me walking to the fridge. [41:30] to, you know, or walking to pour another cup of coffee because it's annoying. Instead, chat GBT, make Chris horizontal. And now it comes back, you know, Chris is horizontal in the new version of it. [41:43] So again, [41:44] Uh, so... [41:46] I think what I want people to take away from all these examples is it's [41:51] It's just the little things. It's the difference between these instantaneous, don't look up code, boom, boom, boom.

[41:58] is such a big difference for a data scientist. It changes the whole process [42:04] in my opinion. So, you know, I think a lot of people are looking for [42:07] to be blown away more. You use ChatGVT to write a book in 30 days. I think they just assume, maybe, did you just say, hey, give me 10 insights on the NBA from this data set? [42:18] And that I wasn't able to use ChatGPT to do. Maybe one day you will be able to do it. [42:23] But what you are able to do is just take these laborious, boring, [42:27] annoying processes and make them instantaneously and make them nearly instantaneous so that you just can keep firing away and hammering away at these fun questions. [42:38] Yeah, it'll be interesting to the extent that you have interesting questions to ask it, basically. But if you just ask it for... [42:46] 10 insights, it's going to be sort of bland. [42:50] I'm really curious, like... [42:53] When you're doing this, how are you... Because I've had this experience with Code Interpreter or Advanced Data Analysis where I'll ask it a question, it'll give me an answer, and then it's sort of wrong. But how do you tell when you need to go into the data set and check whether the size of Marcus is really right for the number of Marcuses in the data set? Yeah, you definitely... [43:21] learn over time when you need to check and when you don't need to check. [43:25] So merges, for example, can be a nightmare with ChatGPT.

[43:30] in that it just merges our nightmare with human beings too. Merging a data set, it's so often, [43:37] that things get messed up. In the NBA dataset, some datasets, Hall of Fame players have an asterisk next to their name. [43:44] And some data sets, those players don't have an asterisk. Some data sets include label junior differently. There are all these differences and the merge can be really problematic. [43:56] So I didn't do any merge without carefully going over [44:00] all the [44:02] all the [44:04] You know, all the... [44:06] Let me see a sample of players. Let me see a sample of the Hall of Fame players. Let me see a sample of the junior players. Let me download the data set. How many observations does it have? All these checks. A merge, you have to... [44:18] Double, triple check. The other thing is I do know Python coding. I'm not an expert at Python, but I know enough [44:24] that I always read the code that it comes out of it. So you can pretty much... [44:29] you know, you can have a pretty good idea if it's doing what it's supposed to be doing, you know, in [44:35] checking the first names. [44:37] For this one, I probably did a couple more checks of... [44:42] you know, list the players with the name Chris, making sure that they all check out, along with going over the code. So you definitely do do some checks on just about everything. But a lot of it is just intuition over time. Regressions, for example, ChatGBT tends to perform beautifully with no problems. So I'm pretty comfortable just letting it go on the regressions. I mean, you know, I definitely double check the code, but I'm not getting too much in the weeds because it definitely does what it's supposed to be doing.

[45:09] merges, you need to do a little bit more. You kind of get an intuition over time [45:15] of [45:16] how likely ChatGPT is to best something up. The other good thing is ChatGPT's mistakes are [45:22] maybe differently than human mistakes, this may be different than human mistakes tend to be really glaring and obvious. [45:28] So, you know, you know, like the mistake won't be. [45:34] It has too many Marcuses. It'll be like... [45:38] everybody is named Anthony and it's just shows Anthony. You know, I think, you know, whereas a human wouldn't make that mistake, or if they did make that mistake, they wouldn't show it to you. It's usually when you get a chart back from an RA, it's plausible and it's a little more, you know, but, but it might be plausibly wrong. I think chat GBT, [45:55] mistakes tend to be very implausible from what I've known. It's very rare that it [46:00] you know, it comes back with a chart that's almost right. It's kind of either nails it or really messes things up. So that's helpful. [46:07] in going over things. [46:10] Yeah, that makes a lot of sense. I do think that people tend to, and I've noticed this myself too, you know, just like managers get a sense for like when they need to like check into the details of something that a direct report is doing or maybe a research assistant is doing. I think model managers, people who are using ChatGPT and other models, like, [46:30] get that same sense of like, this is an area where it like might mess up and I should probably check into the details versus like, no, this is generally right. And it's probably going to get at this and I can just like keep going. Um, which I think is a, it's a really interesting skill to, to have to develop. And it's something that I think people look at again, sort of like the, Oh, it didn't give me the right answer on the first try. They look at it as sort of one of those reasons why, Oh, they shouldn't use it. Um, but once you develop the sense for it, it, it doesn't actually

[47:00] It's just another thing. It's just another aspect of using the tool that you start, you start to get used to. I couldn't agree more. Yeah. A hundred percent. That was definitely my experience. It was, it was, [47:10] You know, everything pretty much is doable and doable in a way. [47:14] that's not painstaking and boring. You know, pretty much every time it's [47:20] doable in a way that, as I said, is fun. And for me, for whatever I enjoy, kind of go. I don't mind going over a graph a few times. [47:30] or going through some errors, it's usually doable and it's usually fine. And you do have to be a little more patient than some people realize. [47:39] in working with JetGPT for data analysis. I really wonder what it is about your personality and I think mine too, where we would rather do it this way, we would rather have the conversation and be talking than coding. Because I think there are other people that feel the opposite, but I wonder what that little thing is that makes it fun for us and less expensive for us and allows us to be so excited about this where I think other people might be like, [48:09] So, you know, I'd rather just like look it up on Stack Overflow or type it out myself. [48:13] Yeah, that's a good question. I have no idea. I'm so far from that, that I can't even imagine the mindset of someone who would feel that way. Just because that is my nightmare of... [48:26] All right. [48:26] And this is so... [48:28] nice for me and [48:30] uh,

[48:30] I love it. I love it. Okay, cool. So now we're sort of starting to get into the part of the show where we do some exploration. And we are going to do a live exploration of a data set with you. So tell us what the data set is. [48:42] Thank you. [48:43] Yes, this is a data set of Olympic athletes. [48:48] throughout history. [48:50] every athlete, their height, their weight, and [48:54] whether they won a medal or not, the sport they participated in, the country they're from. [49:02] So I just thought maybe it'd be fun to play around with this dataset, see if [49:06] We find anything [49:08] interesting or fun in the data set. [49:12] I have a question that I'd like to ask it, if you think this is a good question. I want to know, given my height, weight, and country, which sport I would be most likely to medal in? [49:23] Yeah, that'd be interesting. I don't know if we can do that. That's a good question. I'm still uploading it. Yeah. Uh, [49:33] Let's see. [49:36] Maybe I suspect that we might have to do that in stages. First, build the model of, you [49:43] Uh, country, uh, [49:46] And like for each sport, maybe come up with a [49:50] Height, weight, [49:51] country model. [49:54] And I don't know. I'm just I'm not sure. Is it too complicated? If it's too complicated, we don't have to. We can try. If it fails, is it a big deal? I don't know. No, it's not a big deal at all. If it fails, like we'll learn something anyway. We'll discover something interesting. So so explain that to us. Explain how you'd want to attack this. Yeah. So well, actually, let's just see first if we can do just your question. So you said you're six two, Dan.

[50:20] I'm 6'2". I'm about 160 pounds. 160 pounds. [50:24] and I'm using ChatGPT for, can you... [50:28] Uh, [50:31] And I'm from the US. I'm from the US. While we leave out the country, I think that's... [50:39] What's... we can add that later. [50:42] What sport would give me the best chance of success? I don't know. My guess is it's not going to tell you a lot here. [50:51] Uh, um, well, uh, so a couple of things I'm, I'm noticing is like, you didn't tell it necessarily that, um, that the data set that you, um, uploaded. And you already noticed something that I should have done cause it's not doing it. Oh no. Yeah. Okay. So now it's uploading the athlete data. [51:08] Okay, got it. So basically, you didn't have to even give it that, which is kind of interesting. I'm curious how well it does. Yeah. I mean, usually when you upload it, that's enough. It knows it's supposed to look at that. Right, right, right. So it's giving a general answer first, which is like basketball, which we already know I would be pretty short for an NBA player. Volleyball, track and field. Yeah, the answers aren't that good because your height is actually not a significant advantage in the NBA. [51:33] Uh... [51:35] Yeah. [51:37] 6-2 is not the... [51:40] Best for it. [51:41] Not going to help me too much. I'm really curious what kind of analysis it's doing. So I can see basically first what it did is it looked through the data set. Yeah. So it's trying to understand the data set. It sees what it shows. So it's learning what the height and the weight is.

[51:57] categories are. And now it's doing analysis. Let's see. [52:02] And while it's analyzing, I have a very important question. Is it data or data? [52:09] Oh, I don't know. I think you'd say either way. Okay. Okay. So it gave an interesting approach, which is not a good one, but it's interesting how it did it wrong. It actually did something that I wouldn't have thought to do. [52:22] So if you look closely, [52:24] What it did is first it converted yours to... [52:29] centimeters and kilograms, because that's what the data set is. Another one of those things that's just really nice that it just does that and you don't have to [52:36] You don't have to think about it. And, you know, again, it wouldn't take very long, but it's just nice to do it. [52:41] Now what it's going to try to do is it's going to find... [52:45] The players... [52:47] It gives a tolerance range. So it chose five centimeters and five kilograms. And it's going to find the number of people within that tolerance range in that sport. [53:00] And right away, you can think what's wrong with this is it's going to be way overweighted to people who have more sports, right? [53:08] Who just have more athletes? [53:10] Uh, sports have more athletes. Yeah. So you see, number one is athletics. Well, track and field has the most athletes, swimming, cycling, rowing, fencing. [53:17] So I think that's not what we actually wanted, right? This is not telling you. We could get there by just dividing out. Yeah, so now we can say, this actually was an interesting approach, and I'm not sure it's necessarily wrong. And I wouldn't have actually thought of that. I would have thought to do a model.

[53:31] But just so I think actually chat GPT, this example of it doing it, something that's maybe a little more sensible than how I would have thought to do it. [53:38] It's simpler. [53:40] Yeah, it's simpler. [53:42] Can you divide... [53:46] by total athletes in that sport. [53:49] So what fraction are in that range? And show me the top 10. [53:55] One thing we might have to do, just using ChatGBT a lot, is we may have to do a minimum number of athletes. [54:03] Because it may come back with something that has only a few athletes and all of them are in it. So that's something that you have to. [54:09] think about. But... [54:11] I'm also sort of wondering, like, age... Like, I'm 32, so I wonder... [54:15] if my Olympic dreams are already over, you know, if like, if we have the age of these athletes too, yeah, that's, um, [54:22] We could add that later. [54:24] Okay, so this is interesting. Actually, I wouldn't have guessed some of these. [54:28] triathlon, volleyball, and beach volleyball. I wouldn't have thought... [54:32] Interesting. You know, it may have five centimeters, maybe a little, a little too big. [54:38] Big tolerance change, yeah. [54:41] Could you do, should we say, could you do two centimeters on the tolerance range? So help me interpret these results though. So basically like triathlon, it's giving 0.12, volleyball, it's giving 0.11. What is the 0.11 or 0.12? So 12% of triathlon athletes in the Olympics are... [55:01] who have participated in the Olympics have been within five centimeters and five kilograms of your height and weight.

[55:07] got it and what about we also wanted to be metal medallers um so maybe we want to filter i don't know that we want i don't know that we want medallers because i think what we're going to lose so much in the statistical power [55:19] I think you're going to throw out so much information. [55:22] Okay. Relative to total athletes. We can try it. [55:26] That's totally fine. [55:28] do two centimeters on the tolerance range. Cause I think, you know, one of the things that volleyball players may come up high, cause the high end of that tolerant, that five centimeter tolerance range is going to be, [55:39] a little bit higher. So let's try to really now we're really talking about people [55:42] that are really about 6'2". [55:51] And it was, I think it was five kilograms, right? Yeah, I think that's okay. [55:58] So now it's a little different triathlon, volleyball, volleyball. [56:02] Modern Pentathlon. [56:04] Amen. [56:05] Nordic. It's interesting. I wouldn't necessarily guess some of these. [56:10] I wouldn't have guessed, yeah. [56:13] So one of the things I'm guessing from this, and this will maybe flatter your ego. One of the things I'm taking from this is 6'2 is still very, very tall. [56:21] So, you know, this is clearly pushed in the direction of the sports where height is a big advantage, so basketball and volleyball. So I would have thought that. [56:30] your height would not be advantageous in basketball because yours [56:35] so far from the six, seven, six, eight point, but even among basketball players,

[56:39] there are more athletes. [56:41] A higher percent of athletes are 6'2" than among sports where height is not such an advantage. You're still kind of in the [56:50] I would guess that's because it's international, right? Like, you know, India may have fewer, you know, really, really tall people, but they still have to feel the team, right? [57:01] We could try and limit the United States, but it might get date. Yeah, that might be too small, right? [57:09] I could also say one of the things I might say afterwards is also tell me the total number of athletes. Actually, I'll stop and start again. [57:17] Mm-hmm. [57:18] Uh... [57:20] Could you limit to USA athletes? Also for each sport, tell me percent and total count of athletes in that category. [57:28] That's cool. Because now I'll know if we're getting... [57:32] How would you know if a number is like enough for it to be like a good statistical test? [57:37] There are more fancy ways to do this. [57:42] I think, I mean, when you're getting 2% of athletes or something, [57:48] What I really say that I might have said I should have said the total count counts of athletes in your range. [57:53] You definitely need... [57:54] you know, five to at least five, probably more athletes in your range to make this worthwhile. [58:01] Yeah. [58:04] Yeah. [58:06] Oh, 12 athletes in range. That's pretty good.

[58:09] Yeah, so I guess... [58:11] It still is. That's fascinating. Yeah. Yeah. [58:14] Table tennis. Is basketball a little lower than it was? Not that much, actually. No, it's still... [58:20] It's similar basketball United States athletes. It's interesting because there's so many more... [58:27] of those athletes, like, you know, there's 44 in swimming, but it seems like 41 in swimming, but it seems like there's just more swimming overall. So it's, that's why it's, [58:38] Um, [58:39] Not the top. This is one where I'd probably double check that it did... [58:45] that it did do it right. So can you list all the luge athletes, their country and their height and weight? [58:52] Bye. [58:54] all the canoeing see this is an interesting thing yeah because you're not worried that it's going to list all of the luge athletes in the original data set you're you're oh you think it's no i i should potentially worry that i may have just messed this up [59:07] I probably should have done that. [59:11] You're probably right. [59:13] I [59:16] Yeah, I think it's messing up. You're right. You make mistakes, and if I saw it, I would have... [59:22] Feeling athletes. [59:24] All of the seven. I probably could just say the seven. Let's see if I just sang seven canoeing athletes. Yeah, that's a good one. [59:30] Let's see. [59:33] Yeah, yeah. Now it knows. It's using the Filtered Data USA. [59:36] Let's see if the canoeing athletes are within the range and from the USA. I'm not 100% sure it's going to be.

[59:43] Like if I was doing this, I would have done more checks along the way if I were [59:46] - Probably. [59:47] Oh, it looks... [59:50] Oh, you see, this is something. These are the types of things you don't realize until you do it. [59:56] It has – the data set has multiple observations if they're in more than one Olympic. Ooh. So, you know, yeah, again, you know, not a mistake on ChatGPT necessarily part, but that's like – you always got to be checking for these little things. Right. [1:00:14] So what do you do now? Like let's say we want to – yeah, okay. But only include – [1:00:20] Uh, [1:00:22] each athlete wants [1:00:25] Even if they were in multiple Olympics. [1:00:30] . [1:00:34] I'm super curious if it does this well because it's like, that's a kind of, it's like, sometimes it's not good at like backing up and like redoing things. You know, it gets confused about which step to redo it on. [1:00:47] Yeah, you're right. I agree with that. [1:00:52] Thank you. [1:00:52] I [1:00:54] Thank you. [1:00:56] Well, canoeing is down to five, which I think is what it should be. [1:01:00] Interesting. And pentathlon got, I think it got moved up because I used to be, the top one was triathlon. [1:01:08] Oh, is that right? Yes, maybe there's one that makes sense.

[1:01:13] What are the sports in the modern pentathlon? [1:01:19] Uh... [1:01:20] And we can ask chat GPT, right? Let's ask. I want to know. [1:01:27] Yeah. [1:01:34] Fencing, freestyle, swimming. [1:01:36] Freestyle swimming, equestrian show jumping, pistol shooting, and cross-country running. [1:01:43] Shouldn't there be five? [1:01:46] Thank you. [1:01:47] *sniff* [1:01:48] Um... [1:01:50] That's fascinating. [1:01:55] Okay, we're missing one. [1:01:58] Thank you. [1:02:02] Oh, it just devised pistol shooting and... [1:02:05] Oh. [1:02:06] I'm going to see if it's right. Sports... [1:02:09] Modern Pentathlon. [1:02:11] Thank you. [1:02:12] This is interesting. Wow. Okay. So. Fencing, freestyle, swimming, equestrian, show jumping, pistol shooting, and cross country. I got it. Yeah. That's right. Yeah. [1:02:24] So this is my sport, apparently. I should be learning fencing and pistol shooting and equestrian show jumping. I don't think ChachyBT realizes that I'm Jewish and that we don't really do the pistol shooting thing. You know what's actually a sport that's great for Jews is swimming.

[1:02:44] A huge percent of the American swimmers are choose. Yeah, that too, probably. [1:02:51] When I got bar mitzvahed, one of the people who went to my bar mitzvah gave me this book called Great Jews in Sports. And I always used to joke that it was like three pages long. It was like Sandy Colfax had like 15 pages. [1:03:10] No, a lot of the swimmers, like the guy Spitz, who won all those Olympic medals, was Jewish. A bunch of the swimmers. Interesting. Well, that's on my list too, right? [1:03:18] I think swimming was on the list too. [1:03:23] It might have been one of them. I don't know if it was... [1:03:26] Oh, yeah, yeah. [1:03:29] But maybe not the, I don't know if it counted in the redone one. Yeah. That's fascinating. [1:03:36] Um, I love this. I love that. Um, [1:03:42] I think it's such a good... [1:03:44] encapsulation of what you were saying earlier that, you know, [1:03:47] Any question you have, you can just sort of answer super quickly. [1:03:52] Um, and I don't know how long this analysis would have done, but like would have taken, but like, um, [1:03:57] It's enough. It's probably enough that in another setting, if I had asked that question, it wouldn't have been worth it to you without this to like go through the process of like of answering the question. But with ChatGBT, it's like I can actually we can actually just answer that question. It doesn't take that much time. And I just love that. It's such a it's such an empowering feeling.

[1:04:19] Yeah, exactly. I totally agree. And this one actually gave, I think, a more sensible way to look at a first pass to the problem that I had thought of. I was going to run a whole regression model. [1:04:27] And that was kind of, I think, I think ChatGPT's [1:04:30] approach was much more sensible. [1:04:33] Especially for, again, just eyeballing it initially. Totally. Totally. Yeah. [1:04:38] Um, [1:04:40] Well, I guess like sort of just, I don't know, wrapping up, like, what are you, where are you going with this? Like, what's your next step? You've, you've, you've obviously written that book in 30 days. Like, where are you going with using ChatGPT in your work? [1:04:58] I'm not exactly sure because I'm... [1:05:01] Initially, in Who Makes the NBA, I'm like... [1:05:04] I'm going to just write a hundred more books. Like that was the best month of my life. I'll write a hundred more books. I have a hundred of the best books. [1:05:10] months of my life and [1:05:12] Uh, [1:05:14] I think... [1:05:15] It's a little hard. I haven't gotten monetization on who makes the NBA right yet. So I've sold a lot for a self-published book for sure. [1:05:24] But it's not enough to... [1:05:27] make it a worthwhile financial pursuit, even in a month of time. It's almost enough if it was just a month of time, but I also did some promotion and podcasts and stuff. [1:05:35] that it would be kind of a [1:05:37] worthwhile financials, [1:05:39] Uh, [1:05:40] Pursuit [1:05:43] It's not a big deal. I mean, I don't want to get, I don't need to get into the details. I have a bunch of savings from my previous stuff. So I'm not like in...

[1:05:50] in that much need of, of, of, [1:05:54] money right now. So I am tempted to just explore my creativity and see where it goes. [1:05:58] But if this book sold a ton of copies, then I'd just be like, boom, go to the next book. [1:06:04] I've kind of been playing around with advertising. [1:06:08] to see, you know, interestingly, if you publish a book with a traditional publisher, advertising it yourself doesn't really make any sense. [1:06:16] because you get about a dollar for every book. [1:06:18] But if you self-publish it, you might get about $6 for every book. [1:06:22] So it might actually make sense to advertise because the click through the order percent of clicks to order might be enough to make it cost effective. [1:06:31] Got it. [1:06:32] So I've kind of been playing around with advertising, but I haven't gotten to the place where it's cost effective. So I'm kind of like... [1:06:37] I'm playing around a lot with that. If I could figure that out, I'd just be like... [1:06:41] oh my God, this is the most fun. I'll do this until I die. [1:06:45] Who makes the NFL? Who makes the Olympics? Who makes Major League Baseball? Who becomes president? [1:06:51] Uh, but I mean, I really think, I think you could do that. Like, I feel like there's something to, I feel like there's some way to, to monetize this that would work is, is, is even if you like, you know, comparing this to self-published, even though you make a hundred percent of the, or sorry, traditionally published, even though you make a hundred percent of the book sales, like, you know, the, your previous books were more profitable is, is sort of, is what you're

[1:07:21] Yeah, well, that was due to the advances. And also my first book, Everybody Lies, kind of just exploded. [1:07:27] uh it just hit the zeitgeist kind of perfectly it was [1:07:31] about what we can learn about people from their Google searches. [1:07:34] And, [1:07:35] It just like, it was a bestseller. It's... [1:07:38] sold hundreds of thousands of more copies and [1:07:42] just the United States and all around the world. And I got a big advance for it. So, [1:07:47] And then because of the success of that book, I got a really big, even bigger advance for my second book. But then that book kind of didn't sell as many copies, but I still got the advance. Right. Yeah, that makes sense. I mean, everything I know about publishing books, especially self-publishing, is that you can kind of like you kind of one book builds on the next. And so if you have a bunch of them in a series like the 10th one, because, you know, people who are fans of that one are going to want to read the previous ones. [1:08:17] build this self-sustaining audience where even if the first one isn't well monetized, the tenth one might be. [1:08:25] Yes and no. There's just always a part of me that's like, Seth, are you just being the most impractical possible? Like, do you have a... [1:08:33] PhD in economics from a top school. You're a data scientist, worked at Google. Most people have become romance novelists with... [1:08:42] Like, [1:08:43] series of books don't have all those credentials that could potentially be more lucrative. So I'm like, you know, like, I don't know, [1:08:50] Maybe I just shouldn't care about that. And I love writing and I love this so much that, you know,

[1:08:54] even if it's a long shot or even the most successful version of it. Like if you look at the romance novelist, [1:09:00] that are... [1:09:01] making a living at self-publishing books. [1:09:04] It's not the living of a data scientist salary necessarily. That's true. But it's, you know, of course there are exceptions. There are some people who've, [1:09:13] hit it out of the park and made $40 million of their self-published books. But [1:09:18] There's a lot of selection bias there, so you got to be a little careful. [1:09:21] So you know, [1:09:22] Sometimes I do interrupt myself and I'm just like, why yet again... [1:09:28] You know, I went to Stanford as an undergrad. [1:09:31] and everyone was majoring in CS. I majored in philosophy. [1:09:35] They were like at Stanford or no, no, I penned. Okay. Yeah. Yeah. And there, there, there are philosophy graduation. Maybe this is the same for you. There were like four people in the, at the graduation ceremony. And I'm like, why? Like, and I, and then I'm like, okay, this isn't. [1:09:53] This isn't smart, so I pivoted to an economics PhD. [1:09:57] And then I got a job at Google. I was right on the path of practicality. [1:10:01] And I'm like... [1:10:02] I gotta write books about... [1:10:04] people's sexuality as learned by internet searches. And then like, I did that. And like, and then I was doing well there. I was consulting, I was making a living of it. Like I've made a really good living. And I'm like, [1:10:14] AI book about basketball. I always have this pull towards these things that I'm so passionate about that I'm not sure. Maybe that's just me and I just have to accept that.

[1:10:25] That's who I am and I'm not someone who's going to climb the corporate ladder as a data scientist. [1:10:31] But I don't know. It's interesting to [1:10:34] I mean, I vibe with that. I had a very similar journey. I studied philosophy, but I've been programming since I was in fifth grade. And I ran an enterprise software company in college and sold it coming out of college. And instead of going right back into that and starting another software business, I ended up writing a newsletter and starting a podcast and running every... And it took me a while to admit to myself that this is what I wanted to do. But I freaking love it. [1:11:04] maybe I'm maybe speaking to the devil on your shoulder, but like, it's, it's very like we, you know, I'm at a place with everywhere I can sort of pay for all the, all the bills with it. And it's very rewarding to be able to kind of like, just be curious all day. So I hope that happens for you. I, I, I suspect that there's, there are ways to monetize this that are, uh, compelling. Like, honestly, if you just did a chat GPT for data scientist course, like you [1:11:34] do half that and half this. [1:11:36] Interesting. Yeah, I haven't been... [1:11:40] I kind of my monetization strategy for my previous books also was just, you [1:11:44] work like hell on them and on the promotion of them. And then just wait till people emailed me for things they want me to do for money, which actually caused my first book was so successful.

[1:11:55] It was like I was living the dream. I had to pinch myself. There was like I'd open my inbox and there'd be. [1:12:00] give a talk here for a lot of money, consult on this for a lot of money, blah, blah, blah. [1:12:05] I kind of got spoiled because my first book was like 100. My second book was 20. My third book's five. [1:12:11] you know, the number of emails coming in [1:12:15] offering me lucrative engagements is just so much lower. So I think now I need to think more about [1:12:20] Like, I can't just go, you know, just think about the content and then [1:12:24] And I got to think about other things [1:12:27] you know, revenue streams and creating courses or creating newsletters or all these things that [1:12:34] by health. [1:12:35] That makes perfect sense. Well, I hope you figure it out. I hope if there's anyone watching that wants to hire you for data science, we'll make sure to, or ChatGPT stuff, we'll make sure to put all your contact information in the show notes. And yeah, I really appreciate you coming on this. I love this conversation. I feel very inspired right now. [1:12:58] Awesome. No, thanks for thanks for following your dreams, because I think from from a fan of yours perspective, we're very happy. [1:13:05] that you're doing this podcast and doing all these things that are so enlightening to people. [1:13:11] you know, thinking through these issues. So that's another reason to follow what you love. I think definitely I probably do more good in the world. [1:13:20] entertaining people with sports books than I would just another line coder at a company. That's true. That's true. Um, so I, I hope, I hope you, I hope you find that for yourself and I appreciate you coming on.

[1:13:32] Yeah. Thanks so much for having me, Dan. [1:13:49] Why? Because this show is the epitome of awesomeness. It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure, unadulterated knowledge bombs about chat GPT. Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat. [1:14:07] craving for more. It's not just a show. It's a journey into the future with Dan Shipper as the captain of the spaceship. [1:14:14] So, do yourself a favor. Hit like, smash subscribe, and strap in for the ride of your life. [1:14:20] And now, without any further ado, let me just say, Dan, I'm absolutely hopelessly in love with you.