The Future of Truth: How AI Reshaped Steven Rosenbaum’s Reality

00:00:30 Andrew Keen: Hello, everybody. In our age of hysteria over AI, there was an interesting piece in the New York Times a couple of days ago by a professor at Williams College called Joe Cruz, telling everyone that writing isn't as important as we think. He argues that whilst AI, of course, is a threat, teaching students to think doesn't necessarily involve writing that well. He thinks that we should be concentrating on how they speak rather than how they write. My guest today may or may not agree. He was on the show a few months ago back in April, an old friend of mine, Steve Rosenbaum. He had a book out back then called Truth is Dead [ed.: the book is titled The Future of Truth: How AI Reshapes Reality; Rosenbaum corrects this later in the conversation], and we talked about how AI is a spectacularly good liar. The book, was published by Matt Holt. But in the meantime, there's been somewhat of a controversy over this book. There was a New York Times piece few weeks after, the Rosenbaum book was published suggesting that the book was full of quotes made up by AI. And since then, Steve Rosenbaum has been critiqued in many ways, sometimes vilified in the media, everything from Wired to the Daily Beast about how he made up quotes. So we thought it'd be a good opportunity to talk to Steve and see how he's feeling six months later. Before we went live, Steve, you said you're doing better now than you were six months ago. How did you take all this? Sounds like, it was rather traumatic.

00:02:18 Steven Rosenbaum: Well, you know, it happened in waves. So first of all, you know, people who've written books know that it's a long and often laborious process. And so when your book finally gets to pub day and it's out in the world, there's this sigh of relief that it's come out of your inner workings and it's gone out into the world. And so for a couple of days after the pub date, we were getting great press, great buzz. You know, the debate about the future of truth was healthy and interesting. And then along came this New York Times reporter who had had the audacity, the temerity to do his job and actually read the book that he was gonna talk to me about because lots of other people opined on the book without having picked one up. And he found— You know, it's funny. It— It's been amplified. The book was right— He found three errors in the book.

00:03:19 Andrew Keen: Well, let's be clear, Steve. I mean, they were— I don't know if they have serious errors, but they were given the nature of the book, that's supposed to be about truth and AI, what the New York Times writer, exposed was that, you made three quotes up. It was Benjamin Mullin in the Times. So as you say, he read the book and must have occurred to him that some of the quotes— And some of the quotes were from very well known media figures like Kara Swisher. So it must have occurred to him that these quotes were made up. So these are serious allegations. I mean, they're not just minor mistakes.

00:03:59 Steven Rosenbaum: I wanna be clear about this, and I haven't ever said this in public, so I want you get to hear it first. I appreciate what he did. I didn't enjoy the process, and I had a, you know, couple of months of misery, which I'll share with you because I'm now past the misery portion of this chapter of my life. But if the choice was have the book remain in circulation with inaccuracies in it that nobody quite notices, and my work is damaged but not called out, or having, you know, Ben say, wait. There are errors here, and then, you know, say a book about truth needs to be truthful. The answer is I— I'm really happy with where we ended up. The process was miserable, but the end result is a better book that I'm proud of and that made its way— Is being published.

00:04:51 Andrew Keen: The public— Let's just be clear. How did the publisher respond? Matt Holt, a credible publisher. I went to the website before, we went live. I didn't see any evidence of your book. If they— Did they pull it? Were they angry?

00:05:08 Steven Rosenbaum: Let— It, again, hap— Happened in phases. Right? So the first question was, you know, Ben calling me on a Sunday morning and saying, hey. I have a question about this quote, and then we talked about it. Then he had a question about another quote, and then he did some research. And by the time we were done with Sunday, we were gonna— We were about to do an event at a bookstore, he and I. And he said, listen. I— You know, we agreed we would not do the event. And then the next day, he called me and said, listen. You know, I feel badly about this, but I gotta write about this thing. And if you read the article as opposed to the articles that came from it, the article is actually pretty complimentary of my work and the book overall, but says there are these errors in them, and it came from AI.

00:05:50 Andrew Keen: And that's ironic. I mean, I read the article. I didn't read— I can't remember reading it when it came out. What's complimentary about it?

00:05:59 Steven Rosenbaum: It talks about my work as a body of work. It clearly says it's not plagiarism, but they're mistakes. And, you know, he wrote me as he was reading the book, was enjoying it very much. So net net, the book had errors in it. And when he said to me on Monday morning, you know, tell me what happened, my answer honestly was, this book took four and a half years to write. You know, I have thousands and thousands of pages of research. I will tell you what happened, but I need to actually do the investigation and find out. And that became kind of an odd word that got amplified. He's investigating. But I was. So to answer your question, I sat down with the publisher. They were unhappy. I sat down with my literary agent who was initially not, you know, in the best of moods about the whole thing, and we talked about what the alternatives were. And after, you know, they also said what happened, and I said, look. I don't know. I know they're in there, and I know they have to come out. And we hired a new fact checker. And I spent the bulk of the summer, you know, the only harder than writing a book is rewriting a book, which I don't recommend to anybody. But that's what we did. And we—

00:07:16 Andrew Keen: But I mean, I've read— Sorry. I've written number of books as you know. I don't understand how— I mean, how long was the book? How many thousand words?

00:07:26 Steven Rosenbaum: It was almost 300 pages.

00:07:29 Andrew Keen: So it's, what, sixty, seventy thousand words?

00:07:31 Steven Rosenbaum: Yeah. Like, 70,000.

00:07:33 Andrew Keen: How can made up quotes, especially from people as prominent as Kara Swisher, Meredith Broussard, AI experts, they've all been on this show, Lee McIntyre. How can they get into a book? How can they get into your book?

00:07:49 Steven Rosenbaum: So—

00:07:49 Andrew Keen: I've written— Let me just be clear. I've written number of books. It's not quite an error. How are you getting quotes that don't exist in your book?

00:08:02 Steven Rosenbaum: So let's separate the AI— Handful of AI quotes from— So Meredith said, I said those words. The citation is wrong, and the citation was corrected in the new edition. I don't think that's a crime. I think it's a mistake and a human mistake. I wouldn't blame a robot for that. But one of the things we learned and that I would— Am deeply disturbed about was that in the process of writing the book and revising it and tightening chapters and retightening chapters, one of the things— You know, I use AI as a writing companion. I acknowledge that in the book. I told my publisher about it. It's in the book that I use AI, so it wasn't a secret. I wasn't being sneaky. Or as you said before we started recording, cheating. I don't think that was accurate. But net net, what we now know, and there have been now hundreds of public examples of doctors, lawyers, nonfiction authors, public— Poets all being called out for using AI in various ways. So I— I'm now part of a club of people that, you know, one way or another, been caught. Net net, nobody saw the mistakes. There was a great big team of really smart editors, proofreaders, copy editors.

00:09:19 Andrew Keen: Yeah. But you're the writer. Why? I mean, you— The book is under your name. [unclear]

00:09:28 Steven Rosenbaum: 100% responsible. But responsible doesn't answer the question of where AI fits in the author's process. And I would ask you this. So when you wrote your books, did you not have a copy editor?

00:09:42 Andrew Keen: Yeah. But most— As you know, most copy editors in the book business don't do anything.

00:09:47 Steven Rosenbaum: Well—

00:09:48 Andrew Keen: And they certainly don't question whether or not a quote is real. They'll ask you. I mean, let's go back to this question. Okay. So I take your point on Meredith Broussard, but the Kara Swisher quote—

00:09:55 Steven Rosenbaum: Yep.

00:10:00 Andrew Keen: What was the quote that got into the book?

00:10:04 Steven Rosenbaum: It's interesting. It was a science fiction quote about, the laws of robotics, Isaac Asimov. And, you know, what's interesting is, you know, her response, which The Times printed, was it sounded like she had a stick up her ass. And then she chuckled publicly about the fact that she was amused that she got The Times to publish that. To be honest, she says it didn't sound like her. To me, it sounded very much like her.

00:10:33 Andrew Keen: But how did it get in the book in the first place? You know— And Kara's— If Kara said she didn't say that, how'd it get in the book?

00:10:42 Steven Rosenbaum: So I listened to her all the time. I had notes and notes and notes. In that particular case, it was flagged by a copy editor. The citation was changed to another— I mean, we did the thing you do in writing a book, which was you say, hey. Check this date. Check this. There were almost 4,000, you know, track changes in the five versions of the manuscript. So this is the professional editors doing their job. And my job, once the book was written and handed to them and then back to me in this series of back and forth, was to read their notes, go into the citations, check it. And the simple answer to the Kara Swisher thing is, you know, I dropped the ball. Human mistake. That's it.

00:11:28 Andrew Keen: But you're, in another interesting piece in The Atlantic by Will Oremus.

00:11:35 Steven Rosenbaum: Yeah.

00:11:36 Andrew Keen: You say that you don't blame your copy editors or even your— Certainly not yourself. You said, chat, and I'm gonna use these words because they're a direct quote. I assume that AI didn't make them up. You said ChatGPT, which is the OpenAI product, fucked up the book. So what does ChatGPT have to do with making up quotes?

00:12:03 Steven Rosenbaum: So this is maybe the crux of the question. Here. Let me turn it back to you for a second. Do you use any form of AI in your daily life?

00:12:12 Andrew Keen: Absolutely.

00:12:14 Steven Rosenbaum: Has it ever provided you with a mistake?

00:12:19 Andrew Keen: What do you mean a mistake?

00:12:21 Steven Rosenbaum: It answers a question and you say, wait. Like, it tells you when something happened, and you look at the date and you say, that doesn't sound right. And then it comes back and apologize and says, oh, I'm so sorry. You're correct. And it corrects itself.

00:12:35 Andrew Keen: Well, I don't intend to use AI like that, but I'm probably— I mean, it's certainly not foolproof.

00:12:41 Steven Rosenbaum: So one of the things I've learned is I put some transcripts into ChatGPT and said, give me some pull quotes. These are primary source transcripts from interviews. And I said pull some quotes from [unclear]. It pulled some quotes. I just have to— You know, this is now early days of AI. This is within a year of it being released. And it never occurred to me that if you were giving it actual first person interviews, that it would do anything other than pull copy from those interviews and say, I think this is strong, and we look back and forth. When I went back to the audio, I found that it was, in some cases, determining that it could make first person interviews just a little better, which I was saying [unclear]. Right? Which makes me crazy.

00:13:31 Andrew Keen: Right. So the reason you blame AI is that you asked OpenAI. You asked ChatGPT for some quotes, and it delivered a quote from Kara that she didn't actually say. Is that right?

00:13:48 Steven Rosenbaum: Well, there are a bunch of different examples, but that's one of them. Yeah.

00:13:51 Andrew Keen: Let's just focus on that one because that's the one that the New York Times one exposed, and that's the one that, let's say, brought you down, but that's the one that's most associated with the book. But then—

00:14:01 Steven Rosenbaum: And by the way, Kara is in the new edition of the book.

00:14:05 Andrew Keen: I'm sure she's pleased about that.

00:14:07 Steven Rosenbaum: Well, I talked to—

00:14:09 Andrew Keen: That was a joke.

00:14:09 Steven Rosenbaum: It was a joke. But— Well, but here's what I did. I reached out to her. She didn't answer me. I emailed. She didn't answer me. I called her. She didn't answer me because people don't answer shit. And I then went and found the quote I wanted to replace it with. I emailed her again and said, this is where this came from. This is what I'm going to use. If you don't want me to use it, call me and give me something better. And we ended up, I think, texting back and forth, and she corrected it gently. She wanted the source adjusted, which we did. But, you know, when a— When you're talking about a book that has more than 400 citations in it, you know, I couldn't have done that with every subject in the book.

00:14:50 Andrew Keen: Just can't do it. I mean, I've written books. 400 is not that much. As you said, it took you a long time to write. If you're gonna quote someone in a book, and especially if AI is delivering, the quote, at the very minimum, you would check online. Presumably, and since it's an online product, it would be able to point you to where she said that.

00:15:12 Steven Rosenbaum: It— So what we've learned— And you know all of this because the— you know, the other people who are now part of this club of damaged authors who are repairing their work is not only will it point you to a DOI number and a link, but you have to click on it, find it, and increasingly a year ago, not so much now, but it would routinely point you to fake URLs and fake DOI numbers. And at first, you think, I must be doing something wrong, or I must have a bad— And, you know, it changes every 30 days. So today, I'm now much more careful about how I use AI, and it's actually much better. But it—

00:15:57 Andrew Keen: Yeah. But it comes back, Steve. It's— I mean, you know better than I do. The— You— You're writing a book about AI. I'm literally— And in terms of, our show, you said that AI is a spectacularly good liar. I mean, I'm literally quoting myself in terms of the subtitle of the book. And then you use AI in the book— In the actual book, and you didn't check the source knowing that AI is a spectacularly good liar. In fact, the book— That's what the book's about.

00:16:33 Steven Rosenbaum: So let— Let's use as an example this podcast. So somebody listens to the first three minutes of this podcast, they read the transcript, and they quote you saying, Steve Rosenbaum's book got in trouble, quote, the book was titled Truth is Dead. Well, I didn't correct you. We were recording, but that's not the title of the book. And then you put up the page from Amazon and showed the title of the book, which is the future of truth. So someone could read this transcript, read you saying that, and quote you on the title of the book and be absolutely wrong. That's the nature of this messy world we're in.

00:17:08 Andrew Keen: Yeah. But there's a fundamental difference between—

00:17:12 Steven Rosenbaum: Getting the title wrong in the book?

00:17:13 Andrew Keen: Hold on. Hold on. You accuse me. I mean, you're right. Of course, I got the title wrong. I'm often mistaken. But there's a fundamental difference between— You— Your last book was on curation and the importance of curation before this future of Truth book. There's a fundamental difference between a chatty podcast between you and I, where it really doesn't matter whether or not I make mistakes, and a book, which as you say is curated, took years, took a team of people to make where someone was paying you. So there's a fundamental difference between those two things. They're not comparable.

00:17:49 Steven Rosenbaum: I would— I'm not gonna argue on your show about that except to say, I think you're absolutely wrong.

00:17:56 Andrew Keen: That I have the same responsibility in terms of—

00:17:59 Steven Rosenbaum: To tell the truth, I believe you do.

00:18:02 Andrew Keen: Yeah. But I'm not lying when I mistakenly called your book what the future of AI.

00:18:08 Steven Rosenbaum: You said truth is dead, which is the exact opposite of what the book's about.

00:18:13 Andrew Keen: So I made the same mistake as you. Am I gonna get a New York Times piece about me?

00:18:19 Steven Rosenbaum: I think so. If—

00:18:21 Andrew Keen: What's his name? The—

00:18:23 Steven Rosenbaum: The guy who wrote—

00:18:23 Andrew Keen: The piece.

00:18:25 Steven Rosenbaum: I'll take him—

00:18:26 Andrew Keen: Ben Mullin. If you're watching or listening, Benjamin, you're more than welcome to write a piece about me.

00:18:33 Steven Rosenbaum: But here's, I think, what we all have to now deal with in the world we live in, which is, you know, when a nonfiction author makes a mistake, they have two choices. They can pulp the books and tuck it under the rug, or they can man up or woman up or human up and fix it. And the answer is fixing it is much harder, much more exhausting. And, you know, for me, the hypocrisy would have been to say I wrote a book about truth that had mistakes in it. They were call— I was called out, and I basically just hoped it would go away. And the decision that— You know, I mean, when I told, you know, one of my advisers that I was doing the podcast, he said, are you out of your fucking mind? That was—

00:19:15 Andrew Keen: Who is that?

00:19:17 Steven Rosenbaum: I'm not—

00:19:17 Andrew Keen: You have a— You're lucky. You have advisers? I don't have advisers.

00:19:21 Steven Rosenbaum: Yeah. I have advisers.

00:19:22 Andrew Keen: Adviser. You mean your life? Your— Not your life, your wife?

00:19:26 Steven Rosenbaum: No. No. No. No. This was— The question was, if you're going to put the book back out into the— Like, the reason I wrote the book was because I think truth is in great danger, and I ended up being a spectacular example of that. And had it been anyone but me, I would have dined out on the mistakes in the book just like every other media person in the universe did for a week and a half. That being said—

00:19:53 Andrew Keen: What do you mean that if it been— If it had been anyone but you? That you'd be the only one who would acknowledge guilt. I think most authors would actually be so embarrassed that they wouldn't do shows like this. They wouldn't give interviews to The New York Times and The Atlantic. They would just go away.

00:20:12 Steven Rosenbaum: Yeah. You know, I've been at this a long time. This is not, you know, I've written a couple of books, made a couple of movies, and run some companies. Embarrassed wasn't ever an issue for me. I will tell you— I'll tell you some things that make me angry. So, you know, if you look at the reviews, you know, on Goodreads, they're horrific and mean, and they almost inevitably start with or end with, I didn't read the book, but this guy is a lazy motherfucker. Blah blah blah. Just read the—

00:20:48 Andrew Keen: It's the same. I checked the reviews on Amazon, and they're mostly— 68% of them are one star, which is quite an achievement. I've never seen a book. Like, even my book's got better reviews than that on Amazon.

00:20:59 Steven Rosenbaum: But they didn't read it. So that— What the reviewing is me based on The New York Times. So I, you know, I went out to Amazon and said, your policy says you have to read the book to review it. Could you take down the ones in which people say explicitly, I didn't read the book, but this guy's an idiot? And they didn't really answer. So I'm hopeful that with the new book, which has two great big new chapters in it and a lot of corrections and some of them small and a handful of them big, that the next chapter of this book's life out in the world will be— They could be harsh reviews. They can be critical. They can call me names, but read the damn book if you're gonna criticize the book.

00:21:40 Andrew Keen: So coming back to what happened, so you went to your publisher, Matt Holt. They weren't very happy, but then you, what, you convinced them to do another edition that you would rewrite, check everything?

00:21:52 Steven Rosenbaum: No. It happened more slowly than that. They said, you know, if you wanna do a detailed fact check and come back to us and really, you know, explain what happened, we'll withhold judgment about whether or not we'll remain your publisher on the next edition. And then I came back and I said, here's the detailed fact check, and here's what we learned, and here are the changes that we proposed to make. And they said, based on that, we'll go ahead and publish a new edition. And— But it was, I mean, it was basically the entire— What was my entire summer? It was three months of working every day, both with the fact checker and with new proofreaders, and then it had to be laid out again, and there had to be a new index. It had to be— You know? I mean, it's, it was an extraordinary amount of work to repair the book so that I could say to you, it has been carefully fact checked. It is accurate, and it is worth a read.

00:22:54 Andrew Keen: Well, I'm not sure you were fixing the book. I mean, it sounds to me in many ways for better or worse as though you were fixing your reputation or trying to fix your reputation.

00:23:06 Steven Rosenbaum: It was more internal. It's a fair question. I would say it was more about me being able— You know, maybe I wasn't embarrassed, but I certainly wanted to be able to say, you know, in a year when people say to me, what, you know, what do you think about this? Well, here's the book I wrote. And I didn't want people to say, oh, that's the one that has, you know, six errors in it or however the number of errors are. And for me to be able to— I mean, what— Here— Here's the thing that I'm puzzled by, and I'm still puzzled by. If everyone has had an experience in which AI, whether it's Claude or ChatGPT or something else, has made provable, demonstrable mistakes and put them in quotation marks, you know, I've asked a number of very smart developers, you know, how difficult would it be for Sam Altman to say to his team, let's no longer put things inside quotation marks that we can reliably cite back to a source. And their answer, three out of the three developers I quote them this was, it would be trivially easy to do that. So then you say to, you know, Sam Altman, who I would like to someday be sitting across from, okay. If the change to not put lies in quotations is trivial, why don't you do it? And the honest answer, and this is what ChatG— GPT will tell you if you ask it, is, you know, we want to grow dramatically, and delivering confident answers that are occasionally wrong is a risk we're willing to take. And I think that's actionable. And—

00:24:53 Andrew Keen: What do you mean actionable?

00:24:55 Steven Rosenbaum: I mean, you know, if—

00:24:58 Andrew Keen: It's his product and do whatever he likes with it. Why should he bend over backwards for—

00:25:04 Steven Rosenbaum: To tell the truth?

00:25:06 Andrew Keen: For lazy writers like yourself who didn't check their sources.

00:25:10 Steven Rosenbaum: So you're saying automobile companies shouldn't worry about the brakes working because you should actually pay attention to the road and slow down.

00:25:18 Andrew Keen: I'm not sure that's a particularly good argument. I mean, if I— There's no reason. I mean, you can't blame Sam Altman. I mean, you can blame Sam Altman for lots of things. We can't blame him for not having some, check mechanism to be absolutely sure that if a writer like yourself asks OpenAI for a quote which is wrong or exaggerated or inaccurate, that there's some sort of what emergency break on it so that it will reveal that?

00:25:51 Steven Rosenbaum: So you're asking me exactly the right question, which is that little squib of text at the bottom of the screen on their interface that says, you know, ChatGPT makes mistakes. You should check your work.

00:26:02 Andrew Keen: Yeah. But you know that. That wouldn't have changed anything with you, would it?

00:26:07 Steven Rosenbaum: Well, it did. It did now, for sure. But—

00:26:11 Andrew Keen: Especially given the nature of your book. I mean, you knew that. So— Everybody knows that these AI— Especially when you were using it a year or two ago to write this book. They weren't accurate. There were— Every day, there was an article in the newspaper about how inaccurate they were.

00:26:25 Steven Rosenbaum: So in the broader world— Let's zoom back from my book for just a second. So when your doctor says to you, based on this thing that we put our information into, we think you should be on this drug, and you find out that, you know, prescription was at least in part generated through an AI interface, you're saying we— Shame on the doctor for trusting the AI. He should know better than to trust it.

00:26:55 Andrew Keen: Again, I mean, we're— Well, you— You're coming out with some fictional doctor—

00:26:58 Steven Rosenbaum: No.

00:27:02 Andrew Keen: —who relies on AI as opposed to his own professional judgment. And if that's the case, then they're probably not a very reliable or maybe even honest, professionally credible doctor because they should be making decisions on their own.

00:27:15 Steven Rosenbaum: So I added a chapter in the book, in the new edition, about hallucinations, which were basically ignored in the first book because of when it was written. And in it, there is a story about a researcher at Columbia University who's finding now that about 20% of all the medical research that's going out into the public and into professional medical journals is created by AI and with falsehoods in it. So the basis of what medicine is using as a research tool has about 20 errors that are AI generated. It's horrific. I mean, I didn't kill anybody. Nobody's gonna read my book and choke to death on it unless they—

00:27:55 Andrew Keen: Yeah. But there's no moral equivalent between the two. Those are entirely different. So you're saying that because some AIs give false medical information or advice that might result someone dying, that what you did was not a serious—

00:28:14 Steven Rosenbaum: No.

00:28:14 Andrew Keen: Or crime. The transgression.

00:28:16 Steven Rosenbaum: I'm saying the opposite. I'm saying that this technology that is zooming into our society and everything we touch, every car we get in, every lawyer that takes you to court, every doctor that does a physical on you is increasingly finding that their professions are being infused with AI. And when we say to the companies that are building these models, where does truth fit into your equation? They say, oh, we're not in the truth business. Well, you know, respectfully, what the fuck is the answer? I mean, how do you say—

00:28:51 Andrew Keen: Who says? I mean, you've made that up. Who says that they're not in the truth business? Give me an example.

00:28:56 Steven Rosenbaum: We could do it live. You can go on ChatGPT and ask it whether or not it's in the truth business, and it will give you a long winded answer. And when you boil it down, the answer is no. And they don't plan to be.

00:29:09 Andrew Keen: Well, if they're honest about it. I mean, what do you, Steve, I don't know if you read the latest biography of Roy Cohn. I'm not comparing you with him, although you're both New Yorkers. But, one of the things that Roy Cohn did very well, if that's the right way of putting it, which Donald Trump learned, is never admit error. Never admit defeat. Just move on and attack anyone who is accusing you of anything. Is there a Cohnian, if there's such a word, quality to your decision to come out fighting on this one?

00:29:52 Steven Rosenbaum: If you're saying that Roy Cohn and I are not, agreed on this, I will take that as a badge of courage, and you can put Donald Trump in the conversation while you're at it. I mean, you know, I think history will write that among his greatest skills was being a relentless, ruthless, driven liar and to be able to deliver mistruths with such confidence that some nonzero amount of the public goes well. You know? I mean, I don't know if you bought the Donald Trump branded phone or watch or silver dollar or any of the crap that he's manufactured, but, you know, I was very careful when scoping this book to not make it about Trump because it would make it in my mind, you know, a short term book rather than something that I hope will have a purpose and value in ten years in the future. But, you know, there's a— There's a— There is a chapter about lying and politics and the danger of misinformation being amplified by technology. And, you know, as we come up to this election, there's no doubt that, you know, we're gonna see a massive number of fake videos that look like a candidate or sound like a candidate, that call your phone, that are on your social media, and that are provably not made by them. I mean, that's— That used to be illegal.

00:31:15 Andrew Keen: One of the things you do very well, Steve, is get great, blurbs. You're very well connected. A lot of people have been on this show. Nick Thompson, the publisher of The Atlantic [ed.: Thompson is CEO of The Atlantic], called your book, at least according to Amazon. Brilliant. Kurt Andersen, who was just on, called it essential. Michael Wolff, who was on a few months ago, called it lucid. It came with an introduction from the Nobel Prize winning Maria Ressa, who called it, as, who called it, urgent, whatever that means. I'm always wary of that word urgent. What— What's been the response of all your network, people like Maria and Nick Thompson and everybody else to this? Have— Were people angry with you, or did they just sympathize?

00:32:09 Steven Rosenbaum: So, very quickly, I had to make a whole bunch of decisions about people who'd been very generous with their time. And I made the decision not to ask the question. What do you want me to do with your blurb, or will you blurb the book once it's repaired? Because I didn't want them to be put in a position where they have to say either yes or no, both of which I thought were, for them, uncomfortable. And so I called the publisher almost immediately and said, take the blurbs off the book, which they took immediately off the digital book, not so quickly off the print book because it was ink on dead tree. I took them down off the website. We took them—

00:32:52 Andrew Keen: They're still on Amazon.

00:32:54 Steven Rosenbaum: Yep. Because that's the original book, and they wouldn't edit the description of the original book. But they're not on the new edition. And not because they— I didn't wanna get into a horse race where Maria would have to say, this I'll leave up or Nick or, you know— And I have different levels of friendship with these people. Some are acquaintances and some of them are friends. But net net, the book has no endorsements on it, and I think that was the right thing to do.

00:33:25 Andrew Keen: What do you think, Steve, of products like Pangram, which I'm not sure if they existed when you were writing the book, but they become very popular, accurate AI detector. I'm not sure if they were designed or— To detect the mistake that OpenAI made for your book. But do you think these are legitimate tools for checking whether or not people are using AI? And I assume that there are products like Pangram that will eventually be able to pick out errors created by AI products.

00:34:01 Steven Rosenbaum: So it's great question. I would love to be in a betting pool of whether or not Pangram in any form or— Will exist in six months. You know, Anthropic has already determined that they're watermarking everything that goes through that tool. So if you go to Claude and say, find me any places where I made mistakes in spelling. So I'm a terrible speller, by the way. It's part of when you know you get an email from me. I have certain spelling atrocities that I commit daily. But, you know, Claude will not allow that. Claude— And so the fact that they watermark everything means if you presume that some— Maybe half the content that's out in the world comes from Anthropic, maybe it's not quite half. I don't know what the, you know, sixty forty. But net net, that makes Pangram useless. You know, everyone went around and tried to kill em dashes. And that— You know, everyone cried about the death of em dashes for, like, three months. And now the apps, you know, no longer use em dashes. So you're playing a game of whack a mole. But I think the larger question and the one that should really worry you and I and other nonfiction authors and publishers and less Amazon is, are we gonna find out that this idea that human authorship, big stamp on the front of the book, we're written by a human, no AI allowed, that readers actually don't care. They care about the reading. And if you're getting a book about how to use the new edition of Adobe Photoshop, do you really care whether it's a carefully worded AI created book or that it's written by a human? And in some cases— I mean, on to— In, you know, in the case of Adobe, for example, when I'm having a problem with Premiere or Photoshop, ChatGPT answers those questions with extraordinary skill, because it's read every possible version of that platform, and it knows which one I use, and it only gives me answers about the Photoshop version I have on my computer. I— There are places where AI is going to be a good writing companion. There are places where it's going to be a good write— Writer, and there are places where it's gonna be atrocious. And—

00:36:32 Andrew Keen: How do you spell atrocious?

00:36:34 Steven Rosenbaum: Couldn't even begin to guess. Something with an a and a t. But I would put it in spell check or not. But I think this larger question— And you started with the New York Times article, so let's go back to that. I mean—

00:36:49 Andrew Keen: The Times article that, so to speak, brought this whole issue up in the first place.

00:36:54 Steven Rosenbaum: No. No. No. I'm going back to the newer one where the professor said writing is—

00:36:59 Andrew Keen: Okay. And I did wanna return to that, the Joe Cruz one about—

00:36:59 Steven Rosenbaum: Yeah.

00:37:03 Andrew Keen: Writing not being quite as important as we think. And, of course, the headline is, [unclear] in the sense that thinking for this particular professor is more important than writing.

00:37:15 Steven Rosenbaum: Which made people very angry at him. You know, I will tell you, I work with a lot of young people. My organization, the Sustainable Media Center, has a board of 50 extraordinary Gen Z, folks, who really do, you know, leadership. And there are times when they speak using language and words and— That makes not a bit of sense to me as a relatively old person. I don't accept the fact that writing is going away, but there's more and more audio and emoji. I mean, do you know every emoji people send you in a day? I get all these things with emojis, and some of them, I'm like, what does that mean? I mean, I do know what some of the more graphic ones mean, and some just make no sense to me because I don't use them in my communication. Language changes.

00:38:08 Andrew Keen: From my wife. I don't know what they mean. What do you think, by the way, of Cruz's, thesis? Do you think it's right? Do you think that thinking is, so to speak, the thing in itself? I mean, he agrees, of course, with Socrates who himself was a critic of writing.

00:38:24 Steven Rosenbaum: So I don't. I've been writing since I was 15 years old. I write a lot. I write every day. I write long form. I write short form. I write on social networks. I write a lot. And there's never a time when AI and any interaction I have with it convinces me to soften my writing or blunt my writing or be kinder or gentler. It is— It has kind of a puritanical streak that I find infuriating. And so if I ask it to read something and give me feedback on it— Because the other thing about being a writer is it's a fairly lonely existence. Like, there's nobody in your world that's like— Unless you have, like, a really good writing [unclear].

00:39:12 Andrew Keen: You've got advisers. You're unusual. I don't have any advisers. I— How many advisers do you have, by the way?

00:39:19 Steven Rosenbaum: One, two, three, four, five, six.

00:39:22 Andrew Keen: Wow. [unclear] real? They're not digital. They're not bots.

00:39:26 Steven Rosenbaum: Well, they're not bots. But, you know, when I write for myself, the voice I hear in my head— This is one of the things about editing that I think is interesting, which is you write something, and when— If you make a mistake and you leave out a conjunction or a— The sentence is— Often in my reading of my own writing, I fill in the missing words because it makes perfect sense to me that I meant to say and even though it's not there. And so to have, you know, ChatGPT say to me, hey. This sentence doesn't connect to this. You should take a look at that. It's really useful. It's a useful kind of like watchdog. But when it says, as it often does, I think you've gone too far here. I say, fuck you. It's none of your damn business. Just tell me if I made typos.

00:40:18 Andrew Keen: Yeah. Did you tell ChatGPT, by the way, that it fucked up your book?

00:40:23 Steven Rosenbaum: No. I do curse at it, but not in that case.

00:40:28 Andrew Keen: Finally, Steve, what about the moral qualities of, the judgmental qualities? I— I'd put into— I asked Gemini before this whether what you did was a form of cheating, and it came back with a fairly nuanced reply. It came up with a core controversy. Said, and I'm quoting Gemini, whether this constitutes, quote, unquote, cheating depends on how the term is defined. Are these products good at making moral judgments? Can we trust them in that sense? Truth is itself, of course, a slippery word when it comes to moral judgments, which are more subjective than, objective. But should we be relying? I mean, if you— Could you rely on Gemini or Claude or ChatGPT to make a judgment on what you did and didn't do wrong?

00:41:23 Steven Rosenbaum: Oh, I love that we're ending here. It's delightful. So first of all, if you and I and five other people ask the same question of some kind of a chat or whatever, we'd all get different answers. Because among other things, what it's doing is trying to deliver an answer back to you that meets your needs. Right? So when it gets kinda squishy like that, it says, well, you know, it kinda depends on— I mean, that's— That always sends up a warning flare for me. I don't ask you questions like that, and I don't think— It will be very careful, all of them. It doesn't wanna place blame. It doesn't wanna— It'll say, well, there are two sides to that. I— To me, I think the answer is, I'll give you a human answer rather than an AI answer. I think I was responsible. I think a lot of great people worked on the book, but at the end of the day, if the book was a massive hit and I had, you know, won the Pulitzer and blah blah blah, whatever, like, no— None of those other people who touched the writing were gonna get the credit other than in the fine print on the back of the book. So I take the responsibility for it. But I do think that the platform companies— If somebody came out and said, our platform, our new AI platform, includes truth as a component of what we deliver back to you, and when we make mistakes, which we will occasionally, we are going to correct them. So, for example, when you yell at ChatGPT and say that's not the right spelling of that as I did, you're never speaking to the underlying basis of information. So you can yell at it all you want. It's not fixing anything. I think the source of the information that it's gotten historically was stolen, and I think the source increasingly is things like Reddit, which terrifies me. And— No. I think guilt and responsibility are different. I take responsibility. And that's— By the way, that's why I'm here. I'm here today because I want some nonzero number of your listeners to say, you know, this thing about truth in society really matters. And even if this guy is 80% right, this book is worth— If someone else has a better book on the state of truth, I'll read the better one. At the moment, I think I'm the leading inquisitor in the question of truth in society and that's a role that I'm gonna take seriously.

00:44:01 Andrew Keen: Hold on. You're the leading inquisitor on the role of truth in society?

00:44:05 Steven Rosenbaum: Right now, at this moment.

00:44:07 Andrew Keen: What, in the whole world?

00:44:09 Steven Rosenbaum: No. In the English speaking world. In the books I read.

00:44:12 Andrew Keen: Wow. And one thing we can say about our friend, Steve Rosenberg— he— Steve Rosenbaum, I almost did an OpenAI. I got your name wrong, as he's not shy, may not be Roy Cohn, but he has his chutzpah. Steve, congratulations on the new book. Is it out yet, this— The second version?

00:44:32 Steven Rosenbaum: It is out for presale, and you— If you order it today on any of— If you go to futureoftruth.com, all the various bookstores are available, among them Amazon, but also Simon & Schuster and others. And they will deliver—

00:44:46 Andrew Keen: And shouldn't you be, I mean, for people who bought the original book, shouldn't you be giving them their money back and giving them the new version, the one that's actually truthful or more truthful?

00:44:56 Steven Rosenbaum: The one that's actually truthful. The nature of the call— Of publishing is there— It's not something where the updated version get— You get for free. That's a business decision that is strangely above my pay grade.

00:45:11 Andrew Keen: Wow. Well, it was nice to see you, Steve. And as I said, I admire the fact that you didn't crawl into a corner. You didn't dig a hole and disappear. You came out fighting in good New York fashion, good Upper West Side fashion. So congratulations on that. Still continuing your work at the, Sustainable Media Center. Maybe we'll do a show in the next few weeks, about the new movie about social media, which I think is gonna be a big hit with Frances Haugen. As always, Steve, good to see you, and best of luck with the second book. I hope it's more successful than the first.

00:45:45 Steven Rosenbaum: Well, it won't make as much noise as the first, but it will sell a few more copies, I hope. Thank you. Appreciate the time.