You Can't Outsource Wisdom: Bestselling Author Ryan Holiday on What the Stoics Have to Say About AI

23:36Ryan Holiday:You have to want it. Yeah. Like I think people think like, oh, I can use AI and it'll write for me. Right. And I actually find the writing tends not to be very good. But what I do use it for is I'll upload paragraphs and I'll go, tell me what's wrong with this. tell me how it could be better tell me where i have errors you know tell me how it could be shorter tell it could be tighter which by the way is the same set of prompts that i give my human editor and my research assistants and my friends right yes like it's just what it really is is a way of scaling something uh that like you know i might hire an external editor on my books you know it's not cheap but it makes the project better and i can afford to do that i'm probably not going to hire five, right?

24:20Ryan Holiday:I'm not going to hire 10 external editors. But if you told me that by hiring one, I might get two or three for free, I'd be interested in that, right? And so I think what AI can allow you to do is scale some of this stuff. Again, you have to have the desire, you have to have the interest, you have to have the openness to it. And there probably is at this point, it's not currently as good as some of the best people in the world at these things. But sometimes you don't have access to the very best, right? Very few could afford the very best. And for most people, the choice isn't between the very best editor and a mediocre one.

24:55The choice is between no editing whatsoever, right? Totally. But this, I think actually it's emblematic of something I've observed, which is AI tends to amplify the underlying bias, right? So one of my obsessions in kind of the problem solving innovation space is this bias of satisficing or being satisfied by that which is good enough, right? herbert Simon. And humans tend to just to settle, right? Which is fine most of the time. But when you're trying to innovate, it's actually, that's a problematic cognitive bias. And what AI does is it gets somebody to mediocre much more quickly. And so if you're the kind of person who's going to settle for mediocre, AI is your enemy.

25:35However, if you're the kind of person who's going to push for exceptional, you've never had a better thought partner than AI, right? Which is to say it amplifies what's underneath. And I think cognitive offloading, I get this question all the time in keynotes. What about cognitive offloading? And to me, what I always say is it's going to make you more of what you want to be. If you want to be lazy, it will help you be far lazier than ever. Now, but your point, no, you don't, Henrik, you definitely don't. Ryan, it's your point. If you want a thorough teardown of where this paragraph sags, it will be more brutal than your most brutal human editor, right?

26:11Because in a sense, it's a mirror of your underlying desire. What do you think about that?

26:15Ryan Holiday:Yeah, I think that's right. And like, look, if you're going to be satisfied with the first thing that you found from ChatGPT, you're probably the same person who was just satisfied with whatever Google kicked up five years ago, right? So whether your base AI answer is better than your base Wikipedia answer, I suppose is interesting and you could compare them side by side. But basically the person who's going to settle or takes things at first glance or doesn't question appearances, they're going to be doing this either way. And so in a way, what's interesting for me, I started as a research assistant for a great writer and I watched him sort of throw out most of what I brought to him because it either it wasn't good enough or he had enough prior knowledge to know, hey, that wasn't accurate.

27:06Ryan Holiday:And watching him be so diligent, I worked for this great writer named Robert Greene, watching him be so exacting and diligent in how it worked obviously shaped me in working for him and then as a writer. And then it's something I try to implement with the research assistants that I use. And then now, as you bring sort of AI into the toolkit, it's the same thing. It's like, hey, this answer is interesting, but show me what it's from. I don't just want this quote. I want to see the context in which the quote is, oh, okay, actually it's from a book. And now I'm looking at the book and I trust the author and the law.

27:44Ryan Holiday:Okay, I'm going to use this. And then it's like, oh, no, no, this is actually a quote that you just found from A to Z quotes.com. And is it actually real or actually the author, the person saying it meant something very different than what the sort of perverted understanding of it means. So one of the things you just learn is that like the research that people can surface for you or the things that people track down for you, it's really just like the first in a several step process before any of that can actually be used. The funny thing is like I know this when I'm researching, you know, I'm writing something about Lincoln or whatever, and I'm very diligent.

28:22Ryan Holiday:And then I'm like, hey, I've got this rash and ChatGPG tells me that it's this. I'm like, okay, that's what it is. And you don't push for the level of depth that you did for Abe Lincoln, is that what you're saying? Yeah, because you don't know. You don't know what you don't know, right? And there's something, I think it's called the Gale Amnesia Effect or something like that, where you tend to question media coverage about things which you have some level of expertise about and then you're pretty deferential when it comes to topics you know nothing about. Can I ask on that, as I understand it, Storix warn against outsourcing judgment, right?

28:58And what I think a lot of people now is that they ask Chetubuti, should I do X or Y?

29:04Ryan Holiday:Yes. What can we learn from the Storix on becoming better or more comfortable in our own decision making? Yeah, it's a good question. It's actually funny. in the opening pages of Meditations, one of the things that Marks really acknowledges that he learned from his stepfather, who was his predecessor as the emperor, was his willingness to cede the floor to experts and to listen to them and ask probing questions, which, you know, again, talking 2000 years ago, it's interesting how sort of timeless some of the skills can be. But I think judgment is an interesting thing because everyone thinks they have it.

29:43Ryan Holiday:And then And on average, obviously we don't, right? Like most people don't have good judgment. And that's why most people struggle to do things or get themselves into trouble or make bad decisions. And I think even individually, our judgment, we can have great judgment in one area and pretty poor judgment in another. And so, you know, this idea of like, oh, I'm just a very gut-based person. I trust my gut. I'm always interested on like, has that person done the work to have that trust? will make these sort of snap decisions, which feels like it's based on a lot of experience or a lot of knowledge.

30:18Ryan Holiday:Then you kind of get under the hood and you go, actually, you don't really know anything about this or anything about anything like this. And that deference or that certainty in the judgment is really problematic. By the way, Ryan, you know what we've learned as hosts? We probably only released 75 % of the interviews that we have because we realize fully probably one quarter of that, which by the way is a better hitting percentage than most VCs, right? So I don't mind, but fully a quarter of the people who get recommended to us, Henrik and I get halfway through the conversation. We go, I'm not sure there's actually a lot of depth here.

30:55And this is a person who is a, you know, they're the head of blah, blah, blah. It'll be so awkward now, Jeremy, if we don't realize this. This will be the most awkward interview we've ever done. If we don't release it. You know, Ryan really didn't have that much to stay about, you know, stoicism and wisdom in the modern age. Speaking about our own egos being inflated, you were like, what did he know about stoicism? Okay, hang on. One thing I want to get at. I saw you say this the other day, which I thought was great. The essential skill of our time isn't prompt engineering or coding. It is having a finely tuned bullshit detector.

31:28Ryan Holiday:Yeah. How does one develop the detection ability? I mean, I wish I could give you a great answer. I do think that is sort of the skill of our time. And it was clearly a skill most people didn't have well before AI. We live in a time of conspiracy theories and misinformation and nonsense and sort of popular delusions of all types. And so to add in there, you know, again, a very savvy sort of technology that is ultimately a consumer product and consumer products do not want to tell you no or I'm not sure or I can't do that. And so you do end up getting these sort of hallucinations or these sort of fudging, hedging answers.

32:17Ryan Holiday:And to be able to separate something that seems true or is close to true is going to be really important. I would say, though, practically, one of the interesting things about AI is because it's not a human and because you can't hurt its feelings, just some simple habits of like, are you sure? Or what's the evidence for that? Or what I heard was X, Y, and Z. Just like one of the things I do when I'm messing around with this, not only do I instinctively check all the work, but it doesn't cost anything to make it do it again. Right. Like there's a there's a famous story about Kissinger where he he's he asked someone to give him a report and they give him the report and he sends it back and he says, you know, this is wrong.

33:08Ryan Holiday:Do it better. The guy comes back. The report's tighter, better research, whatever. He says, same thing. No, this is still wrong. I told you to do it better. and then he sends it back and then the guy sends it. And finally, the third time he says, okay, now I'm going to read it. Because humans, just like machines, don't always give you everything they've got the first time, right? Wow, that's so good. Some room for refinement. So that's kind of one of the things I think about is just, there's no cost in me calling bullshit. If I say like, no, no, that's not right. Worst thing is going to happen is it's going to waste some compute doing it again.

33:46Ryan Holiday:I think that's super interesting. I was having dinner with a friend of mine, and he's a movie director, and he was saying the same thing, but about actors. That sometimes he used to work together with a very famous movie director, and sometimes the actors would do it, and he'd go like, do that, but just better. And then he said, I often feel bad about doing that, and so I'll go in and act out the whole scene like I wanted, which obviously is dumb because he's not an actor. But I really like that idea of just saying to the AI, hey, I like what you just did there, but just make it better. Yeah. It was awesome.

34:25But there are tons of tactics like that. To me, this is critically important. I feel the question is often leveled as a conclusion, meaning this. what about cognitive offloading is not an honest question, meaning what do I, as a conscious, sentient, well-intentioned human, do I do knowing that this collaborator, like every other human I've ever worked with, is liable to trigger cognitive offloading? They render the question, it's framed as a question, but it's really a conclusion. Therefore, I'm opting out. And I think that is really dangerous to say. If I could just add something there, Jeremy, on that.

35:06Like I now have, I use all these different agents that run all time. So I have like whatever, 10 agents that do stuff. And my kind of, my Uber agent, my agents of agents called Iggy. And Iggy have this script now that we, that I run every week where it looks through all our transcript and it tries to kind of like answer that question. Like what do I always offload? Where's my biases without knowing it? And then sometimes it will prompt me back. So, for example, if I say yes to too many decisions, they'll basically like, ah, you seem to be just offloading decision making to me right now. Are you tired?

35:39Do you need to go have a break or whatever? And so, ironically, I think increasingly because it's so seductive, you need to create all these protocols for yourself to kind of avoid you falling in this super seductive trap because it is awesome. And that's kind of what I was getting at. To me, the answer is not opt out. The answer is get educated, right? There are techniques. I mean, like the Kissinger, right? That's one of a hundred techniques, Ryan. But it's like, if Kissinger always said, do better, that should be a part of AI training programs. And then every human that collaborates with AI who doesn't do it is uneducated.

36:19And every human that opts out is opting out of tremendous augmentation. But to me, it's very important to realize the answer is education, not opting out. And I feel like right now there's all of these issues, which by the way, have always been present with people. AI is just trained on the way we think and work and all that stuff. And there's a, there's kind of a class of people that are putting up this smoke screen. What about, you know, environmental impact, you know, blah, blah, right. And to me, those aren't actually the issue. It's just, it's a way to justify disengagement. And that I think is actually, I think it's dangerous.

36:53Yeah.

36:54Ryan Holiday:And I would say that even if you don't want to use it, right? Which I think there's some ethical reasons to not want to use it or you're worried about atrophy or whatever. You don't want to use it. I get it. That doesn't mean that everyone else isn't. And so to not have any familiarity with it makes you actually vulnerable to still the same sort of sloppy, crappy work product, right? So if I hadn't spent the time that I've spent sort of messing around with AI and trying to figure out how it works and using it in some cases and finding it insufficient in other cases, I would be a lot worse at recognizing, for instance, when I'm getting like an AI written pitch or when someone on my team sends me something that it's very clear, they offloaded this task.

37:44Ryan Holiday:So instead of sitting down and thinking about how to actually do this best, they had AI write the CTA for this email or this outline or whatever it is. And that's why it's problematic. And so you have to understand it both offensively and defensively, I think. And to not do that is to just, as you said, sort of opt out. Circling back to the beginning of this conversation, we talked about how the Stoics took change as a given, change as a constant. Recognizing that folks right now tend to think we're in a moment of particular change. are there any mindsets or behaviors, rituals, routines, et cetera, that you would recommend inspired by the Stoics in terms of how to successfully navigate change?

38:32How do we do it? What are the things that we can implement in our lives that the Stoics have known?

38:37Ryan Holiday:Well, so the basic idea in Stoic philosophy is we don't control what's happening, but we control how we respond to what happens. And so the idea for the Stoics, it wasn't like, hey, these are the things you have to do. This is how you're successful. Instead, what they tried to be is really adaptable and adjustable to circumstances. That's what Epictetus said he was trying to teach his students, was to teach them to be able to be the kind of person that whatever happens, they could say, oh, that's just what I was looking for, or oh, I can work with that. And so I think, thinking again about what are the sort of meta skills, what are the things that aren't going to change?

39:15Ryan Holiday:What are the things that are going to remain valuable? You know, even if some of these trend lines continue, that's kind of where I would go as a person is like, hey, what are the kinds of things that throughout historical moments, throughout moments of flux and change have sort of remained unchanged? And that's probably the kind of, you know, investment I would make. If I can make an add on to that question. Yeah. If it is about you know, basically saying there's things I control and there's things that I don't control and what I can change is how I kind of deal with it. But increasingly isn't AI, it gives you control of a lot, right?

39:55Like suddenly you can write the legal brief without being a lawyer, you can ship an app without being an engineer. And so the nicety of being able to say, I am not in control of all this. So therefore, I just have to deal with my own emotions. Is that being kind of like degraded now that you are in much more control? I mean, almost evasive or like burying your head in the sand. Is that what you're getting at?

40:20Ryan Holiday:I don't think the Stoics are saying, hey, just focus on your emotions. Nothing is in your control. I think when you look at the lives of who the Stoics were from, you know, emperors to generals and entrepreneurs and executives, of all these people throughout history, there are always high agency people. And so I think, you know, seizing the increased amount of agency that certain technology might offer us doesn't strike me as outside the bounds of stoicism. I think what they mean is like, hey, you don't control whether you're living in the middle of an AI revolution or not. You do control how you respond to that revolution, whether you decide to learn about it or not, how you deploy it inside your business or your life or not.

41:03Ryan Holiday:you decide whether you despair and give up and go, everything's pointless, we're all going to be replaced or not. And so to me, that's what the idea means. It's like most of the macro trends in the world are not up to us, including the weather and the economy and the political situation. We can have little bits of influence here or there on some of those things. But for the most part, the larger macro world is not up to us, but our inner emotions, our actions, our decisions, our priorities. These are all things that are up to us. Can we wrap maybe on this idea of agency? Because I'd like to, I'd love to drill into that.

41:40I happen to believe, Henrik and I were having a conversation earlier today, actually. And agency came up as kind of a defining characteristic, not bullshit detection, but agency is right up there and neck and neck with BS detection. You have kids similar to our kids age. Henrik and I both have kids as well. One of the things I've been grappling with personally, I'd love to hear your thoughts are how do we cultivate agency, both in ourselves and how do we cultivate an agentic attitude in our children? You have any thoughts on that?

42:09Ryan Holiday:Yeah. You know, one of the things I'm always really impressed with with my kids is even if I don't like the thing, like I'm not a big video games person. I'm always impressed at their ability to suddenly be playing a thing or get good at a thing that I certainly didn't tell them anything about. like the way that they heard about a game from one of their friends and then watched videos about it and then fiddled around in it. And then by trial and error, got really good at a thing. So my son was just, he just asked me for this new computer game and we were getting it. And then, you know, he was like, well, how do I do this?

42:44Ryan Holiday:And how do I do this? And how do I do this? And I was like, I was like, I don't know. And I'm not going to figure it out because I have no interest. I was like, I'm not going to help you with this. But I was like, I know, you know how to do this because I've seen you do it before. I didn't teach you how to play Minecraft. And so how do we get good at teaching them how to get good at figuring stuff out? To me, that's the main skill. And that's why one of the reasons I started, you know, sort of messing around with some of these things is like, I wanted it to be a thing that we could do together, but that in the process, they would take away the meta skill of being able to use these tools to figure out how to do things.

43:24I think maybe that's a good place to wrap. Really, really, really cool stuff.

43:28Ryan Holiday:This is fun, guys. I really appreciate it. Drop the mic. Professor Werdelin, what are your big takeaways from Ryan Holiday, Modern Stoic? I'm so fascinated by the topic and so was excited to talk to him. The thing that I got to think about is when he was talking about how, you know, there's been changes before. Everything inside me was like, yeah, but not this. And then I talked to my dad the other day, who's 87, and he was telling me about growing up here in Europe under the Second World War. And he goes, life was pretty changed, like a lot of changes there. And so I guess the maybe not necessarily like takeaway in how to use AI.

44:12I think for me, the big takeaway was like, this is going to be a big change. But there has been big changes before. And I think the second thing he said is that we can either sit and kind of like skulk about it or we can try to figure out how do we do something positive. And I'm reminded when I did work trying to get entrepreneurship into organizations, I often would say, well, if you want to see yourself as an entrepreneurial person, one of your jobs is to take anxiety on your shoulder for other people. And so when they ask you, are you sure that this is going to happen? You have to say, I'm absolutely 100 % sure.

44:45even though in your hearts of hearts you're like, oh, there's a lot of things that can go wrong here. And so I do - I'm sure about what? What was your dad's advice? Whatever it is, I mean, like entrepreneurship is always about thinking about a potential future that hasn't happened yet and then basically articulating to other people, right? Like that's often what entrepreneurs do. And so we are kind of like the imposters that go around and posturing all the time, right? But I think part of it is our job to say, give people comfort in saying there is a potential path. And while it's not crystal clear that we should walk it, let's try to walk it.

45:23And I think it's a little bit the same thing with AI. Yes, there can be many ways that AI could be used. And yes, there are things that could happen that is not ideal for society. And I think we can all worry about that and we do. But there are also a lot of positive path we can do. And so we have the choice right now to say, do I think and do I want to try to come up and use AI for something good? Or do I want to just sit and be very worried about all these bad things happening? And I think that is an individual choice that people will have to make. And so I think the big thing I took away is that that is a moment you can take every second, every day, right?

45:59And I guess I'm choosing to do the latter. Well said. Well said. I think the point about change, it's interesting that your remark is, well, but what about this moment? Like this moment does feel particular. And I think there is a sense in which that's true, right? That, that this is a different kind of a change. However, every kind of change before was relative to the changes that came before a different kind of change. And I am, I find it somewhat uplifting and heartening the opposite of disheartening. Can you be heartened? I don't know, but the opposite of disheartening, it's somewhat heartening to realize humans have been dealing with this.

46:43The future is unknown and uncertain, and there's great upheaval and uncertainty, right? For example, I take the realm of politics. If you think that we're polarized now, and there is data that suggests we're more polarized than ever, yet when you look at newspaper clippings, even going back to the 1800s, 1700s, whatever, there's vitriol there. And it's the most important election in the history of the nation then, right? It's almost like every election is most important because it's the only one that's unknown. And we tend to put excessive valence on this particular unknown. Like, are you, you aren't concerned at all about the deal that you even not forget the deal you won last year, the deal you lost.

47:23You probably don't spend any time thinking about why? Cause like everything worked out. Right. Yeah. But this deal now that's unknown, there's this anxiety, right? So to me, that was a, it was a little bit of a heartening message. I think that humans have actually been grappling with the question and the uncertainty of change for quite some time. And to know that there are traditions that have been developed to help humans grapple with that, I think is, as I mentioned at the top, somebody replied to my newsletter saying, you know, the Stoics have been grappling with these questions for a couple thousand years.

47:55And that's actually what kind of got us in touch with Ryan. And to me, when I got that reply, it was a realization, oh, wait, there are tools for this. Because right now what's being hyped in the media and everything is this is unprecedented. It's there's nothing like it. There's something reassuring about the reminder. Oh, no, actually, people have always been feeling that way. And they've gotten through it. You know, I find that uplifting. And also sometimes, I mean, I'm definitely somebody who defaults a worry. So, you know, like the concerns about AI is not lost on me. But then you start to look at some of the stats of different things, right?

48:35You look at when ATM was introduced and everybody was worried about the people that work in banks or when McDonald's introduced their self-serve kiosk. And then you read that basically both of those industries hired more people instead of less afterwards, right? Same with radiologists now, right? And so I think there is like this tendency to worry. And I do think if we look at that, I have another person the other day who made the argument that the market, like the financial market, which is like this very, very intelligent machine where everybody pours a ton of money in to basically predict how the future is.

49:12And it thinks it's going to be better than it used to be. Right. And so there's all these data points that we get that things are probably going to be fine. And then the but then you still kind of like worry about it. I guess his point was that it doesn't really matter. We are here today. The weather tomorrow is going to change. But make sure to look at and and check out the sun. You know, you know what I haven't studied stoicism much myself. And so I don't know some of these things, but I was actually I don't know how we could push the conversation. one thing I find myself wishing for is some more kind of practical, like tangible.

49:47Like for example, I think one stoic practice is kind of a daily reflection. I think that's probably a great idea for all of us, right? I think another stoic practice is something about kind of contemplating your own mortality, recognizing you are going, like the only thing that is certain in your life is that it will end, you know, and yet that's probably, we treat, as they say, right? We treat death as if it's uncertain in life, as if it's certain. whereas the opposite is true. Life is uncertain and death is certain. So to me, if nothing else, I want to pay Ryan the compliment that he has created in me a suction for, I'm curious to go and see what did the Stoics have to say about this stuff?

50:27Because we didn't get into too many tactical details. One thing I did want to mention that I absolutely, or one thing I did want to mention, wanted, actually, now that I say it again, it does make sense, both words. One thing that I want to mention, And maybe let's just leave it. Let's just see what happens. Drop in the comments, Jeremy Zadufus. The Kissinger anecdote. That was worth the price of admission for me. I mean, I'm a, I'm, and Ryan actually, you know, Ryan will read entire books. I know this because I've read that he does this. He'll read entire books for a single anecdote. So, and he'll, he'll, he'll read an 800 page book.

51:04So I was sitting in a one hour conversation that the Kissinger anecdote is worth the price of mission alone for me. I mean, that is such an exceptional practice for AI, whether he knew it or not. And I would say I don't implement it nearly enough, but just if you want the playbook here, if you want to write and maybe just have your AI take dictation for a second, whatever the AI says, say, this is wrong, do better without reading its response. And then when it comes back, say, this is wrong, do better. And then when it comes back, say, now I'm going to read it for the first time. That's funny.

51:46And that to me, it's not, I mean, it is funny, but it's so true. Yeah. It's so true. And insofar as - Do you think technically if you say to an agent, this is wrong, do better, that it would be so ready to please you that it might actually go the wrong way because you're wrong it, wrong it? Maybe what I would say is not this is wrong, because I agree, it might go the opposite direction, which could be interesting. But what I would say is it feels like you didn't exert sufficient effort and thoughtfulness to do an exceptional job here. To better. I think you can do better. Or even, like I've heard actually a lot of people in our friend group say things like, I think you can do better, and AI does better, right?

52:26But I would say I've known about that as a tactic, but I haven't incorporated it into kind of my standard practice. And granted, what does it cost? Basically, it's only tokens. And to environmentalists, perhaps that's not zero, but meaning to a human, it's another two weeks of work. To an AI, it's literally another 15 seconds. and if you can actually improve the output on the proportion of two more weeks of work for 10 more seconds and the only thing that's keeping you from getting that better work output is having the instinct of Henry Kissinger to tell the AI you think it can do better. You know as Steve Jobs used to do, have you heard this?

53:05I read this in Isaacson's book, he would often say when he had an audacious goal he would often tell someone he'd look the eye and he'd say, don't be afraid. I now have like a lot of people sitting listening to the podcast, just shouting us do better. So like, that's where I'm going. Maybe, maybe we should end it there. And then if people send us an email with do better, we will send them a book. Isn't that like, well, hang on. There's, there's one other thing that I want to call it just while we're rattling through, while we're rattling through kind of, there is maybe to just like to a hammer, everything's a nail, maybe to a professor, everything's a, you know, education problem.

53:43This became clear to me this week when again, for the thousandth time, somebody asked me in a session, I'm in a room of 400 people and there's a 10 minute Q and a and five minutes are devoted to what about cognitive offloading? And again, to the honest question asker, that is a legitimate, there are legitimate strategies for addressing that. My feeling is most of the time, that's a conclusion someone is leveling about why they're disengaging, not a question they're earnestly asking. And I think it's at the very least, it's a good kind of personal check. I mean, it's probably impossible that anybody listening to our show is in the one camp, but it is worthwhile for folks to realize, you know, listeners to our show are probably the evangelist in their company.

54:26When people ask that question, you have to realize they actually aren't interested in the answer. they're telling you or your approach will change maybe. The way you engage with that person is going to change when you realize what's actually going on is not they need tactics. It's a different mindset. I'll leave it at that. Amen. I think that is all for the show. So I guess we only really have one thing left to say and that is bye-bye.