Loading summary
Transcript275 lines
- [00:00]
A
The line that got me, that cinched it for you, was, I will be your last human chief of staff.
- [00:06]
B
There's a few parts of the job that is just kind of like bullet holes through your attention during the day. It's essentially just nuking your ability to concentrate on one meaningful thing. An AI that can just take all of that sort of meaningless cognitive weight, you know, off of the human. There's no point for me or anyone else to be doing that.
- [00:29]
A
We are an AI first company. Anything that can be done by AI should be done by AI. Welcome back to Infinite Loops. I'm Jim o'. Shaughnessy. Today's guest is Jean Marc Decius. He is an amazing artist, he is an amazing programmer, and he is one of the highest agency people that I've ever met. Please enjoy my conversation with Shawn Martin. Let's do your superhero backstory, because when we were chatting before we decided to record, we both agreed that your cv, unless you had a nutcase like me being able to not pay it. I didn't even look at your cv, to be honest. But when I learned a bit after hiring you, I'm like, huh, wow, that's pretty cool that you got this job.
- [01:27]
B
Yeah, it's, you know, it's kind of like hyperstition. But I never even would have gone for a job like this, to be totally honest. But the situation here is much different than what I think most chief of staffs would be dealing with. So it's. I think it. It dovetails. Well, not nicely into the, you know, the fact that I don't have a traditional CV for this kind of position. Well, the position is also a bit different than what I think most chiefs of staff might. Might be accustomed to.
- [02:00]
A
Yeah. And, you know, although we use the terms, we are essentially in real time trying to create a completely new sort of company, you know, AI First. In all of our verticals, everything that we're doing, the line that got me, that cinched it for you was, I will be your last human chief of staff. Elaborate on that a little bit.
- [02:24]
B
Well, there's a few parts of the job that, you know, like, any job is, like, this is a little annoying. Like, particularly with email. Like, there's. There's so much that is just kind of like bullet holes through your attention during the day when it's like you. You check five emails, and then all of a sudden, by the time you're done with the fifth, you have like, six more, and it's like, all right, I'm not getting anything done. Like, yes, Email is a useful communicative tool, but it's. There's a huge trade off there because it's essentially just nuking your ability to concentrate on one meaningful thing that actually has ROI for a given period of time where you can actually make progress on it. I think AI can handle a lot of that shenanigans that happens in emails where it's like, all right, I guess I needed to read this. I guess I needed to be updated on this. I would much prefer, you know, at the end of the day and at the beginning, or at the beginning of the day, I check a little app that I've. I've built, and it's like, here are, you know, a little ticker tape of things that I need to know that have been aggregated from emails that I haven't actually gotten to since I last looked at it. And here's like a, you know, a short list of meetings and then my highest priority items. So, like, integrate a to do list into it. And so then an AI that also has the context of all of my priorities, my personal priorities, my work priorities, and then the priorities of the company can crunch all of that and then give me a short list of things that I should work on. It's like, all right, you have a lot of meetings today. You have a couple of emails that will require some time. So you don't actually have much time for any deep work. So your to do list has been reshuffled to reflect that. And so it's like, all right, here's a few little easy things that have been on the back burner that you can just knock out with a little bit of available time that you have. And then the next day it's like, all right, I got one meeting in the morning. Emails are kind of light. Block off this huge thing and tackle this next big phase of a coding project that I'm working on. So an AI that can just take all of that sort of meaningless cognitive weight, you know, off of the human would be great, because there's no point for me, me or anyone else to be doing that kind of thing.
- [04:46]
A
And you definitely avoided my question about your superhero backstory. So we can truncate it. What, sing and like, you've done so many different things, which I find super cool. What one of those things that you did do you think was the most helpful for you, the role you occupy now?
- [05:07]
B
Oh, man. Playing with Legos. When I was a kid, maybe actually my first entrepreneurial endeavor was actually accidental and slightly illegal. I was living in Vancouver in a tiny studio apartment. And I remember I'd gotten a decent paying job and like two months in, I looked at my bank statement, I'm like, why do I have no money? And I look at my credit card statement, I was like, whoa, booze is like really expensive in Vancouver. And my reaction to that was, well, how hard can it be to make whiskey? And so I spent the next nine months studying separation sciences and built a six foot fractioning column complete with a gram condenser in my tiny studio apartment and distilled my own vodka, gin, rum and whiskey. Even barrel aged the whiskey myself. And it wasn't until I was doing my first run and I like actually tasted this, got it up to 94% ethanol. I tasted it, I was like, this is good. And I had a friend over at the time and they were like, is this legal? And I was like, I didn't think about that at all. But I'd also spent quite a bit of money building this copper still. And so I wanted to make my money back. So I may have traded a few boxes of booze for money with some friends.
- [06:41]
A
So wait, so the headline is, Jim being chief of staff of o' Shaughnessy Ventures, the best prior skill I got was from bootlegging.
- [06:53]
B
Well, I mean, it was essentially a complete business that was founded on an experiment of curiosity that required a lot of tinkering and a lot of hands on building in order to get it to work. So it was this initial question of, well, how hard can this be? And of course Hofstadter's law, it always takes much longer than you think it will, especially when you take into account Hofstadter's Law. But, but then unintended, it developed into this business experiment that, well, I'd had all of these costs and it's like, well, now I want to recoup these costs a little bit. And so I was young at the time and I never really thought of starting a business at the time. Obviously it wasn't a legitimate business, it was just really like a hobby. But it was the first time that I made something that earned an unexpected dollar in that sense. And so I think the lessons involved in that entire little hobby experiment extrapolate really well to pretty much every product or service or business experiment that I've run since and have run since I've come on board here. So it's like, all right, what, this is a cool idea, but what will the ROI on it be? Is this actually going to move the needle? Is this actually like a Decent business experiment or is this really just fulfilling some curiosity? Fulfilling? Curiosity can be great. It can be useful because it can, it can lead into unexpected unknown unknowns where you can find, you know, a little bit of gold and then it's like, all right, we, you know, kind of like the Bell's lab, where it's like, all right, a whole bunch of scientists just do cool stuff and then they uncover all sorts of things that eventually do have higher roi. But that all being said, I think it is good to every sort of like proposal or idea go through that, you know, litmus test of like, can we see a viable ROI here?
- [08:56]
A
Yeah. And you know, basically what we believe is there is gold in them there tails. But a lot of people aren't operating there because of a variety of reasons. Like for example, with AI Lab, we think that we're going to be able to develop some very commercially successful software that people will use and love. But we have kind of an edge in that no VC would touch it because the market isn't big enough.
- [09:32]
B
Right.
- [09:33]
A
They, I think one of the troubles that traditional enterprises get into is they get captive of their own jargon. Right. And so what's the TAM to the total addressable market? Oh no, that's not big enough. And plus they're constrained by the fact that every investment that a traditional VC makes, essentially they, they needed to return the fund. And we are not under any of those problems.
- [10:01]
B
Yes, it's kind of like they have, they have incentives that create scope paralysis. So they can't, they can't zoom into something that it would actually be very much worth their while because of those incentives that you outlined. And then because of that, that kind of becomes inhabit. So if they move into some other area, they might still have those like, you know, their zoom is basically stuck. Whereas here there's just so much freedom in terms of like, well, how high could that go? You know, how low can we address it and still like a meaningful roi? Even, you know, even if something's like just an internal tool doesn't have immediate roi, but it has like, you know, once you compare it to an industry standard. So, you know, and then you, you scale that, you know, across time. It's like, oh, actually this is going to save a ton of money over time. So yeah, I do think about scope paralysis quite a bit. And it's often the very mistake that I'm making when I'm trying to work on something and I'm stuck. I often have to be like, all right, you know, I got to Shake my head out of the grass and, like, stand up and be like, all right, bird's eye view. What's going on?
- [11:10]
A
And I think also we're a fairly interesting crew here in that, like, if you are not a highly agentic person, you're not going to make it here. Right? Like, I, I, I'm not running a kindergarten, and neither are you. And so I, there's a lot of people who think that, like, writing furiously about something and posting it on social media, that shows them, whereas that's value of that is precisely zero for most endeavors. Like, what's it feel like? What's it feel like to. You've worked at a bunch of different places. One of the things that we try to do here is, you know, show me. You're very good at that.
- [11:57]
B
Yeah, you. I had an idea that I bounced off you based on something that Jimmy said in one of the monthly meetings, and I bounced the idea off you, and your response was, sounds really cool, like, write up a proposal for it. And so I went home that weekend. I was like, I thought, I've never written a proposal for anything. I guess I'll take a shot at this. And I was like, oh, it'd be really cool if I could show what the UI of this thing would look like. And so I whipped out my iPad, I started drawing it, but then I was like, oh, you know, actually it might just be faster if I just code up the UI real quick. And so I did that. And once I had, like, it on the browser, I was like, well, you know, I'll take a shot at the back end. Just see how far I get. And so Monday morning rolled around. I was like, I don't know if you remember. I was like, jim, so you asked me to write a proposal and I didn't do it. You were like, okay. I was like, but I built a working mvp. Do you want to check it out? And you were thrilled. So that developed into a tool that we use almost every day now. It's meeting about a meeting in order to think about a proposal, and then another meeting for someone to decide who's going to write the proposal.
- [13:13]
A
Proposal.
- [13:14]
B
And it's like, when are we eventually going to get to, like, the interesting part, which is, like, someone actually does the thing. So I, you know, that's why I've always been a little bit, I guess you could say allergic to corporate culture, because it just seems like a lot of meetings and nothing gets done where I'm much more of the, okay, that's great. Leave Me alone, I would like to actually take a crack and get something working and see how useful it is and then we can get feedback on it and, you know, we can actually build something meaningful. That's the number one thing that I think I love about being here. But perhaps maybe not having the traditional CV also, these go hand in hand because I'm. I just never wanted to go through all of these, jump through all these hoops in order to like build a CV in. In order to get this kind of job. It's like, no, I was always much more concerned with like. No, I. I just want to actually do things that are interesting and give me real feedback as opposed to more gold stars.
- [14:19]
A
Yeah, well, even when I was in a more traditional environment, I think, you know, 80% of internal corporate meetings are masturbatory and a total waste of time. And I think the challenge with that is that people think that they're accomplishing something. You know, I had 10 meetings today. Who the fuck cares? Did you do anything today? And you know, as part of the series that we never really kind of got off the ground. The great reshuffle, like I think all of that. And I started that, I don't know, 2018, where I got the sense where we are going to go through the biggest transformation in the way businesses are run. Who wins, who loses. The old Playbooks were going to stop working and we've already seen that. And then aligned our verticals around things where I saw the greatest arbitrage ability. For example, Infinite Books. A lot of people hate the book business and think that's not a good business. And the way it currently exists, they're probably right. But the way we are reinventing it kind of. We toy with the word, not even calling it publishing, but an author's operating system. And it amazes me. A simple thing. We a b test all of our covers, right? And you and I just had an experience where the COVID that both you and I were not happy with won. The whole team hated it. And yet again, maybe it's my empirical quant background, but if the market says something and you're saying something else, market is right. But I wonder what. Okay, that's the easiest thing in the world for like a Simon and Schuster or one of the legacy publishers to do. Do you think it's this corporate culture that is just endemic to most big company legacy companies where, you know, it's kind of like having to. Each upgrade needs to be compatible backward. Right. We don't suffer under that problem. But is it a Mixture of things is it? What do you think it is?
- [16:55]
B
I think maybe two parts from a human psychological point of view is people have way too high of a tolerance for bullshit. You were saying meetings are masturbatory. I feel like that's just watering water water bordering yourself with bullshit. And it's like if you have such a high tolerance for that, I don't think that's a good sign. And then I think people also just lack imagination where if someone tells you, oh, this is the way that it's been done, you know, the iconic class is like, well I don't care about that, let's find a different way. Most people don't react that way. Most people are like, oh great, someone just told me what my job is and told gave me a rubric for what I need to do. That's great. If this is the way it's always been done, then it must be tried and trued and I'm golden. I don't have to think anymore. If you want to find a new way of doing something, it's cognitively taxing. And so I think that goes hand in hand with people who can tolerate tons of bullshit, are perhaps talented in the ways of not wanting to think too much.
- [18:09]
A
Yeah. Didn't Bertrand Russell or maybe someone of his ilk say that what most people call thinking is simply rearranging their prejudices? And, and like if you really want to do good work, it's taxing, right? Because you got to think deeply and hard about things, but you also have to be willing to just say, oh well, I fucked that one up and what, what can I learn? I think there's also that very strong in traditional corporate culture the COVID your ass mentality. And it's just not a good structure. It was designed for a completely different era for industrialization and Taylorism, which you know how little I think of Taylorism. But to be fair for the time that Taylor came up with all of these ideas, they made sense, right? Because you were supposed to do one thing. You pull that crank right on a big industrial assembly line. And yet it took over what we call corporate management or those types of things to an extent that people didn't realize that it was just insurification of everything.
- [19:29]
B
I guess. I guess constant meetings are like the doom scrolling of the corporate world.
- [19:34]
A
I mean honestly. And I think that's one of the reasons why I like LinkedIn. They feel fine over there. It's like, oh, he got a promotion, wow. I guess I'll now mimetically become envious of him and try to bring him down. I think that we are very fortunate in that this isn't exactly a startup. Right, because we don't have to worry about funding and all those kinds of things. And your average startup founder spends what, 60% of their time on raising money or keeping investors informed, monitoring the situation and. And so we don't have to do any of that bullshit. And we also have a culture that, like, bullshitting is not appreciated unless it's like, shit posting or having fun. I mean, how many people do you really think right now could thrive in an environment like this? So, for example, we'll talk about this right now. You just did a post for our substack, which is fascinating. You told me about it months ago, and I rode you relentlessly to write it. But you're. You're a good writer too. And, and you know, how many places are you chief of staff? But hey, I also want that substack from you on AI and ag. Talk a little bit about that because I think it was a fascinating idea.
- [21:00]
B
Yeah. First, thanks for letting me do that. It's a lot of fun to have it broadcasted on the os. Verse some really nice feedback on it too, so far, which is great. So the basic thesis is that the same argument that you're making about Taylorism and, and how that's infected the corporate world in this overly bureaucratic, bloated, meeting heavy way.
- [21:23]
A
Voguns, Vogans. If you're. If you're a Hitchhiker's Guide to the Galaxy fan, you'll know the Vogans are the most bureaucratic race.
- [21:30]
B
Yeah, and they're. They're what? They're blowing up the world in order to build a superhighway straight through it because they don't have the dexterity to, like, I don't know, question the order. Question the order, or just like, build around a planet. Solar system, like, come on, guys. So very similar to these. Like, if you take that as an analogy or a metaphor for the current way that we do agriculture, which is essentially just industrial mono crop farming, there's something very similar there because we have these endless rows of the exact same crop, and they require like a Vogon, like, contraption or to harvest all of it. Because, you know, for example, I think it's the Colmer 1357 or something is like 32 rows of corn. It's basically just this long, you know, hammerhead type device that you put on the front of a tractor and then you run it through 32 rows of corn. And it can harvest all 32 rows simultaneously, which is amazing. And from, you know, from the point of view of like the Industrial Revolution, like, this is a triumph. But it has come with serious downsides. You know, one for example, is that in order to have monocropping work, we have to dump like loads of pesticides on these plants in order for them to make it to harvest. In the U.S. we, the U.S. applies over 1 billion pounds of pesticides to food every single year. Year. That trickles down to about 3 pounds of pesticides per person per year, which trickles down to about 11 aspirin tablets of pesticides on your food per day. And even after you wash food, you still have 75% of the food still has pesticides on it. And many of the pesticides now aren't even something that you can rub off. They're called systemic pesticides. So, so they basically infuse into the fruit or the food so you can't wash it off. It's literally inside. And this has health consequences which I could get into that even relate to the GLP1 craze and obesity and whatnot. But back to what I think is probably going to happen with the future of agriculture is we're going to get something that's far more dynamic. Charles Mann is a writer, awesome writer, who, who published this book called 1491, which details what his best thesis, what north and South America probably looked like prior to the arrival of Europeans. And one thing that stuck with me from that book is an agriculture technology that we don't use today. It's called the Milpa and it's where, it's a system of intercropping where you plant like a dozen different species together. And like, for example, corn lacks lysine and tryptophan, but beans have lysine and tryptophan. So if you plant them together, you, they basically complement one another. And what one plant takes from the soil, another plant provides. And so you essentially get this like fully regenerative system in this little collection of, of species that can be harvested basically forever without depleting the land, which, you know, it's kind of like an infinite food hack. And the, the Aztec empire, the Inca empire, all use these, these systems of intercropping. And then we basically bulldoze that system. And we invented monocropping. We had to because it had to cater to the machines that we had available as a result of the Industrial Revolution. We didn't have, you know, a tractor that could plow through a milpa and catch 12 different varieties of, of plants that are all getting, you know, ripe at or ready for harvest at all different times. Like, we don't have a machine that could do that, but we will. Once we have human robots, we could transition huge monocrop fields to essentially food forests that have dozens, if not hundreds of different species of food. And these fields, these food forests, would be attended to by essentially a small army of robotic gardeners. Because the human hand, once we have that in fully automated form, it can differentiate between, you know, a ripe tomato and a ripe avocado. And you know, all these other things that it's going to interact with can also, using AI, basically run hundreds, thousands, perhaps millions of concurrent experiments in order to, you know, find a little bit more arbitrage in these systems. Like, oh, this species does a little bit better for whatever reason. If it's 2 meters away from this species, we don't know yet, but it seems to work. And then like this species of insect and you know, next to this species of, you know, tree with this type of mushroom. And dandelions here are necessary because dandelions hoover up zinc and selenium and all that sort of stuff. We'll be able to run those experiments concurrently, but then also rebuild the topsoil. Because another thing that mono cropping does is it depletes the topsoil. And topsoil, an inch of topsoil takes a thousand years to build. And then in the Midwest we are losing, what was it? I think in an average field we lose 2.2 tons of topsoil per acre. And then in, in like the Midwest, we lose substantially more. It, it comes down to basically like in an average field, we lose 11 years of topsoil in one year. And then in the Midwest where it's, you know, much more heavily farmed, we lose 69 years of top sale soil growth in one year. And so with these numbers, you just project them forward. Things are going to look really bad in about 50 to 60 years. So if we don't figure out a different way to pivot away from mono crop culture, we're going to run into problems. The other thing about this Milpa style food forest is that you won't need pesticides because you have all of this intra species competition. It alleviates basically all of the problems that you get with a huge field of corn which can be just destroyed overnight by one particular species of bug. You just don't get that if you have this diverse ecosystem of plants and insects that are all interacting together, together. So Then you can do away the pesticides, the topsoil comes back, that means that the, the, the ground can hold more water. Like there's just so many benefits here. So I, I outlined all of that in this Future of Food.
- [28:52]
A
You know what I love about it too, is that it, it kind of speaks to our overall thesis brilliantly, right? Monocultures are bad just in humans, in any variety. When you slip into a monoculture, bad shit happens. Using food as the example. Yeah, for when that was the way it had to be done. That was the way it had to be done because we were limited and constrained by the technology and the hardware and rigs available to us. But that's no longer the case. And part of our thesis is that everything's going to become customizable and so can food, right? And then I guess with a tilt to my fierce anti authoritarianism, you've got to look at a system that it goes at things top down. And one thinks of course of Mao and the sparrows. And for people who aren't aware of this, Mao decided that he was going to collectivize all of the farms in China and he fucking hated those sparrows too. And so they literally killed all of the sparrows in China with disastrous effects because the sparrows were eating the bugs that were destroying the crops, right? And it just to me makes me go crazy. Like people don't think of second and tertiary effects of anything. And, and then take an ideology that thinks it has the answers to everything and one of them happens to be, yeah, the supreme leader doesn't like those fucking sparrows, so we got to wipe them out. And like millions of people starve to death. And so one of the reasons why I'm so incredibly bullish on this new era is it doesn't have to be like that anymore. And your article is a perfect example of how it doesn't have to be that way anymore. And one of our themes is that we are an AI first company, meaning that anything that can be done by AI should be done by AI because why should I want you, who's incredibly creative, incredibly productive? Like what? Why should I want you doing like horribly boring and tedious jobs that AI can do much more brilliantly and quickly and all of that. And I sometimes think when I talk to people who are like normies, who aren't like looking at this stuff and why should they, right? I get that those inside kind of the AI bubble, you know, we've drunk the Kool Aid and everything and so, you know, we think, oh, man, they're just promoting the hell out of this. And maybe they should tone it down a little. I think we not promoting it enough because there's a huge class of people. This just isn't their thing. Right, and that's a shame because they themselves, like Steve Jobs very famously said, a computer was a bicycle for the mind. AI is a rocket ship for creativity, for leverage for all of these things. And that was one thing we aligned on, like, immediately give our listeners and viewers some more of your. I don't want to steal the credit for your ideas, so talk a bit about more of some of the workflows that are just unlocking time, talent, and the need for huge overhead just to.
- [32:52]
B
Comment on the normies. I think there's something that could be said about, like, the school system which, you know, indoctrinates this high tolerance for bullshit and this, this.
- [33:03]
A
Oh, it's totally fucked. Fubar. Fucked up beyond all recognition.
- [33:07]
B
Like, a good example is, I think, you know, traditional advertising is just. It's. It's very much like mono crop culture, where you're just spraying and praying at, like, you know, thousands, millions of people with the idea that, like, oh, it'll work for some of them, but, like, for most of them, yeah, fingers crossed. But for most of them, you're. You're actually wasting time and attention, even if it's just like a tiny little bit of a scroll.
- [33:37]
A
And it's irritating.
- [33:38]
B
It's irritating, but you add all of that up over a lifetime and it's like someone. You know, people are wasting just ungodly amounts of just tiny little slices of cognitive attention that could be devoted to, you know, more fulfilling life. And so, like, what's the. What's the humanoid robot, food, forest gardener version of. Of advertising? In that sense, it's. It's very much like a really great recommendation from a friend where like, you know, like, liberty or comes to you and is like, all right, this movie, you gotta watch it. You're gonna be like, he never misses, so I'm gonna watch this. Essentially. That was an advertisement. He just advertised this movie to you. But it doesn't feel like that because it feels organic, it feels human. It's customized to you because a human is behind it. Whereas the majority of advertising is just like a bad billboard, which is like, why did I even just look at that? Oh, I had to, because it was on my screen and I had to click out of it in order to get the thing out, obviously. So I think in the age of AI, we can use AI to figure out if someone is a good fit for a recommendation for something where advertising doesn't become this sort of waterboarding the population with something that we want to sell, but more of like a, oh, Here are the 385 people in this pool who've expressed interest in similar things. That would be a good fit for this. Let's contact them. That just seems like a much better use of everyone's attention and much less waste.
- [35:35]
A
Well, and you're being modest. We really drill down and we find not only the nearest neighbor audience, we also look at their work and they're like, oh, wow, look, he just wrote an essay that he's going to love this book. But then we bring up the essay, right? And a lot of that. The reason I feel so comfortable with that is I would welcome that email if somebody, like, took the time to read what Works on Wall street, right, and they found a book that was really similar, but different, with a lot of new information. And I got an email, you know, kind of citing the findings from what Works. Wow. They actually read the book. Most people simply use it for personal protection or as a door jam because it's this thick, but I would welcome that, right? And. And the way we're doing it is, I think, somewhat similar. You're probably not going to get an email from Infinite Books, Infinite Films, Infinite Media, if we haven't done our homework on who you are as a person. It's all about really trust. And I would trust the company that I never got a bullshit spammy type email from. And, you know, I hate. I hate the ones, especially the ones who, like, seem to do it intentionally. If you want to unsubscribe, hit unsubscribe, and then you're fucked because it's just like they keep coming in and there's just so many of these annoyances from the previous industrial era that we don't have to have anymore. And, you know, the same is true with the app that we were just crawling on it right now, but the recommendation app, because I do believe that like, like anything, AI is going to bring a tsunami of slop and it will also bring brilliance and masterpieces and everything else, but you need. People are going to need help, right? And. And so the recommendation thing that we're going to be working on, it won't be in the form that I want it in until maybe 2030, 2032 too, but it will be something that I want, right? How cool would it be to be able to open an app first thing in the Day. Here's the books that you might like, here's the movies you might like, here's an essay. Here's a substack that you haven't discovered. Now, granted, this is the opposite of a mass market, right? We are using a very targeted market of people, but it's still a big, big market. And, and I anticipate that a lot of our stuff will appeal to not a huge group of people, but that's fine, right? That's the other thing that people don't seem to. To understand. It's like you live under a system for so long. So, right, like American TV, the. You're familiar with the standard bell curve. 68.5% are within one standard deviation of the median. And guess what? I love Lucy has 79 million viewers because it fits right into that distribution of 68.5%. Now, we've already seen this, of course, but there's going to be much more of it. There are shows, podcasts, substacks, movies, books, et cetera, that can be enormously profitable, but with a much smaller audience. I mean, what was Game of Thrones? I can't think what 18 million watched the finals season. It seems that that's a huge arbitrage point that people are just ignoring.
- [39:36]
B
I've often thought of, like, the. The human species as like a single entity. And what we're doing with all of this scrolling, we're kind of like liver cells that are like, processing these free radicals and whatnot. And all of the spam email is basically inflammation in the collective body. And it's like, all right, we need to do something about this. Where is the immune response to all of this spam? And then you get the arms race between Google, Gmail filters and whatnot. And then the spammers, like, level up their, their arsenal. And you essentially have, you know, like this immune battle between these two things that are going on. And I think AI can be become like a, you know, an antibiotic for a lot of this nonsense in our culture. And the way that we've set up our systems, it can just like, debloat a lot of this inflammation. Like, like imagine, like, if I could, you know, very similar to your career curation app. I would love to have an AI just curate like, Twitter for me and so that like every day I could just like, open it up and I could get like the 10 posts that would take me 15 minutes to find and, and, you know, like, and whatnot or comment on, and I could just see those and be like, all right, that Was actually great.
- [41:00]
A
Yeah. The risk there, of course, is that confirmation bias. Right. And I don't know that we want a society where everybody has their own personal bubble. So as part of the curation app, just to stick with that one for a minute, we're intentionally going to add things. Well, not noise, but it might be noise, but it might be something. They're like, oh, I never thought about that. Maybe I should check that out. What other areas do you think like? I personally, you know my views on the leverage component of AI, but I take it much further, as I usually do most everything. Like, I think you can learn a lot with like the book I'm writing right now. As you know, it's a thriller, it's a novel, but a lot of it is based on real world things that really exist in the real world. And literally we never formally calculated it out, but I think that conservatively, it would have taken 10 years to do the amount of research that I did in nine months.
- [42:12]
B
Yeah, I mean, it depends how many people you were employing exactly during that 10 years.
- [42:17]
A
Yeah, I mean, you would literally need an army because they'd all have to go to libraries around the world. Right. Because it's not all centralized and I wouldn't have done it. And so I think that the research part, but I also think the writing part, I write everything longhand and then I put it on the thing, but then I put it into AI and say, tell me how bad this is. And then I'm like, okay, you think you can do better?
- [42:49]
B
Show me.
- [42:51]
A
And so I think that there's always a moral panic when a new technology comes on which crushes the gate that the old talent used to have up, right. And so it's nothing new. It's like all the way back to the novel. Those manuscript monks were pissed off, man. And calculators I experienced when I was a kid, they were very expensive. Everyone thought they were the coolest thing in the world. And then they banned them from all the classrooms.
- [43:27]
B
You're not going to have a calculator in your pocket.
- [43:30]
A
Exactly. Well, yeah.
- [43:31]
B
Are you sure about that?
- [43:33]
A
But so the, er, pattern of it is the same. And I think we're kind of in the moral panic period right now. The pearl clutching. Oh, was AI used? Well, come on, get used to it. Right? Like it shouldn't be. And that's the other thing they always do. The, the ones who were the kings and stars under the old system, when they see their gate being crushed, right? Gate gets nuked, then everything becomes about process Right. Whereas it should be about outcome, especially in a creative endeavor. Did this movie or book or podcast change your mind? Did it, you know, inspire you? Did it piss you off? Right. It should have some effect on you. And, and so I just think it's really funny seeing everyone, like the process. Can you imagine if, if we did that with everything.
- [44:27]
B
Yeah.
- [44:28]
A
Like, we'd still be living in caves.
- [44:29]
B
Yeah. You don't need to take organic chemistry in order to talk to ChatGPT about a particular health condition and then go to a doctor with like a really informed opinion and then you get like this smug sort of gaslighty, like, rebuke from the doctor. It's like, did you chatgpt that? It's like, well, yes, I did. But, like, am I, am I wrong? Like, is, is my premise here that I'm proposing to you incorrect? And it's like, oh, but that's not the thing that you're focusing on. You're focusing on how I got to this information.
- [45:00]
A
Right.
- [45:00]
B
Yes, I did use ChatGPT and I want some actual answers because now I have like a little bit of context around this, which previously most people didn't. And then they were just at the, at the behest of like, whether they had a good doctor or not and whether that doctor was like, well informed about whatever the issue was. That's, that's one of, that's been one of my favorite and you know, most eye opening examples, you know, you were talking about, like, using AI to educate. Like, the amount that I've learned about health since the, since chat TBT dropped has been unbelievable. Like, the things that I've been able to do personally, just like, you know, taking blood tests and like plopping those into various AIs and being like, all right, give me a read. Like, it's, it's been incredible. It's like, oh man, I wish I had this like a couple decades ago.
- [45:54]
A
Well, as you know, you've seen the actual journal from 1983 or whatever in which I spend page after page talking. I didn't call it AI begging and pleading for it to come. Now, we're enthusiasts, right? And so I doubt that a lot of people out there are going to be doing their own blood work and sending it to a lab and having it processed. But even in that old 68.5%, so many lives could be made materially better by just embracing the technology for simple things like learning. I mean, we talked about the school system, yet another relic of the industrial era. Right. It was like industrialists saying to the imagine industrialists had influence on governments. The shock, basically, they're saying to him, yeah, what we need you to do here is we need the kid to not be curious, not be controversial. We want them to have a correct answer machine installed in their head. And there is only one correct answer. And the. That's the one we are supplying. Right. And. And you know, it's like, there's high tolerance for bullshit. Yeah, High tolerance for bullshit. There's a reason why I liked Pink Floyd's song about, hey, teacher, leave them kids alone. But rather than just go on and rant about it, there are now alternatives. And. And I think. I think it's really easy to root against things. And so I always look for things to root for. And I think that in the field of AI and across disciplines, Right. There is so much to root for. I mean, I look at my grandchildren and the way that they are using AI and it's really clever. And, you know, it's one of those unknown unknowns. Right. And this idea that, you know. Did you ever see the movie Matilda?
- [48:00]
B
Oh, yeah, that's.
- [48:01]
A
Yeah, right. Like, all my kids love Matteo, especially my girls. And. And I just remember that scene. Why? Because I'm big and you're little. I'm. I'm. I'm right and you're wrong. I'm the adult, you're the kid. It's like, that's just so much. And we can unlock, I mean, really what we've done. And I get a little extreme here, but, like, school and a lot of current curriculum is designed to crush creativity, crush that way of thinking. And, and it doesn't have to be like that anymore. With AI, we're investing. We've invested in several AI companies for education. What other areas do you think are non. Kind of intuitive for where AI could be like, an amazing unlock?
- [48:56]
B
Well, before answering that, just to address something that you said about, like, people criticizing it, like, they're so quick to criticize. And when I hear a criticism of AI, the first thing I always ask is like, well, how much do you use it?
- [49:10]
A
Yeah.
- [49:11]
B
Because if. If Picasso was in the room and he had a criticism about a particular paintbrush rush, I would take that criticism very seriously. As would I. Dude probably knows what he's talking about. Whereas, you know, someone who spent, like, who's only done like three or four queries on chat. Gbt.
- [49:31]
A
The free version.
- [49:32]
B
Yeah. And like, oh, this sucks. It's like.
- [49:35]
A
Well, they're right. The free version does suck.
- [49:37]
B
Well, yeah, it's like, all right, you're clearly an authority on this. You're, you're kind of like anyone on Twitter, Twitter who's like, whatever the, the subject of the day is, you're now an expert in when you actually haven't gone deep on it at all. Like, what's going on here. We hear a lot about it, but it's almost invisible, which, which I think might count as not intuitive. Like just Tesla's full self driving AI.
- [50:05]
A
Amazing.
- [50:06]
B
Unbelievably good. Like, and.
- [50:09]
A
Sorry, I will interrupt. You were the guy who converted me. I still don't drive a Tesla, but the firm has one and it's fucking cool. And like, if you're driving around midtown Manhattan or downtown here, the last time I was with you, I was blown away and you even turned to me and said, if I was driving, we would have gotten into an accident. Right. And so the proof's in the pudding.
- [50:38]
B
There's been at least twice in the last, in the last like 10 months where I'm like, oh my gosh, if FSD hadn't been operating, like I would, I would not.
- [50:49]
A
Would have been bad.
- [50:49]
B
Things would suck right now.
- [50:51]
A
It would be bad.
- [50:51]
B
Yeah. But the thing is, it's invisible. Yeah, you can't. Like, like, for example, when, when, when cars first were first invented, it was super visible and obvious. Like, you know, 1901, a picture of a Manhattan show street was like almost all horses and then like one car in the distance and you know, 2.5 million pounds of horseshit. And then like 15 years later it was the complete reverse where you had like a sea of cars and you know, one horse in the, in the background. Whereas with, with AI taking over driving, it's completely invisible because you can be in a regular gas powered car going down the highway and then someone right next to you is in a Tesla and they are. The Tesla is driving itself. But it's really hard to tell unless you're like, oh, they don't have their hands on the wheel and they don't have their foot on the pedal.
- [51:46]
A
Yeah.
- [51:48]
B
So I think that's going to be, I think that's an example of something that's non intuitive but which is actually going to have this enormous effect. You know, Tesla's starting to ramp up their robo taxis in, in Austin and in San Francisco and I think those are just going to demolish. At that point it might become slightly more obvious because then you'll see a car driving around with no one in it.
- [52:12]
A
Yeah.
- [52:12]
B
Because, you know, like, if you do if you do summon on a Tesla in a busy parking lot, like I've done this a few times where like the car is driving itself to me and like it drives past a pedestrian, they're like, there's no one in that fucking car. But some other non intuitive ways, I.
- [52:33]
A
Just want to make a point about that because I think something people often miss is that my thesis is at least that for your everyday person they will be beneficiaries of AI because it is invisible. It. I think that 10, 15 years from now I will be like electricity. It'll just be there and, and it will be doing yeoman's work. Right. It will be doing heavy lifting. The self driving car being a great example, but people will just get so used to it that, you know, won't even be remarkable anymore.
- [53:17]
B
Oh man. That's kind of the sad thing about technology is that once it becomes a protocol, it ceases to be magical.
- [53:24]
A
Yeah.
- [53:25]
B
Even though if you like went like a century before it was invented and introduced that now protocol thing to those people, they would be like, what kind of devil sorcery is this?
- [53:37]
A
You'd probably be burned at the stake depending on how far back you went.
- [53:40]
B
Yeah. Oh man. It's just how you would. It's just strange how, how humans have this like immense capacity for becoming inured to wonder and something that's like truly magical while simultaneously having like very crippled imagination. Huge tolerance for bullshit. We really are such a strange species.
- [54:05]
A
But we're actually quite pro human here at. Oh, but the, and of course the classic is Louis CK's thing about Wi fi in an airplane. He had this great bit where he's like, do you remember the first time you went on an airplane and they had WI fi? How you were like, oh my God, I can't believe it. I'm 30,000ft above the earth and I can waste my time scrolling Twitter, Twitter. And then he goes. Then the minute it goes out you go, what the.
- [54:41]
B
This is.
- [54:43]
A
But you're absolutely right. But I think that over time that that has been a positive for our species, that we are incredibly adaptive. And yeah, it does suck that something that is almost magical in its properties gets taken for granted.
- [55:03]
B
Right.
- [55:05]
A
But I think it's also good for the species.
- [55:08]
B
Yeah. I mean it could just be like an artifact of like the, the life cycle of a given innovation because you know, when it's in the.
- [55:15]
A
They're all speeding up.
- [55:16]
B
Yeah. Well, when it's an MVP stage and you know, Louis CK is sitting on a, on a plane and Then the WI fi router gets, like, overloaded. It's like everyone reacts negatively. It's almost like an immune reaction. And that. That's actually a signal to the people driving that innovation forward. Like, all right, well, we need to improve this. Clearly it's wanted, but clearly it's also failing. And so that's sort of like the, you know, the MVP stage is like, you get feedback real fast on whether it's working, whether it's not. And if you get a high signal on either of those, it means that people are using it. And which means, like, all right, this is useful. We should keep pursuing this.
- [56:00]
A
One of the downsides of a lot of the innovation you and I were chatting about before we started to record. And I admire your point of view about this because I didn't experience it. And it's this idea that there are secondary and tertiary effects to ubiquitous cell phones. And the way you put it is much more direct, but you're under continual surveillance. And I made the comment that if cell phone cameras or the ability to take a film had been around when I was a teenager, I would be totally fucked. Because. Because. But talk a bit. Talk a bit about that.
- [56:43]
B
Yeah. So in the 1920s, the Hawthorne Hawthorne Works factories in Chicago ran a study which unveiled something that later became known as the Hawthorne Effect. They were trying to figure out how to make their workers more productive. And they tried all sorts of things, like turn up the lights. That seemed to increase productivity. But then they, you know, to sort of ab test the experiment, they turned the lights back down, but productivity remained high. And they were like, this doesn't make any sense. Did the lights help or not? Eventually they realized that it was the presence of the people doing the study and the effect of the workers knowing they were being watched that changed their behavior. And so the Hawthorne effect is just like, when someone feels like they're being watched, their behavior changes. Like a sort of structural incentivize, or if we were going to take that form and those incentives and put it into, like an actual structure, you essentially get like a panopticon where, you know, it's essentially like a, you know, a circular prison with a guard tower in the center. You can actually put like one way or two way mirrors around that guard tower. And the guard tower can be empty, but everyone in this prison will think they're being watched just because there's like that eye of sauron there.
- [58:07]
A
Yeah.
- [58:08]
B
And so. So what inadvertently happened in the late, you know, around 2010 when camera phones started getting good is Suddenly we entered a social surveillance state. And that's particularly pronounced for, you know, high schoolers where they can't really have fun in the way that other, you know, previous generations did because now they're, they're hyper aware of the fact that like, oh, everyone has a camera in their pocket and if I act weird or if, you know, God forbid, I do some underage drinking or something like that.
- [58:51]
A
Underage drinking? People drink when they're underage?
- [58:54]
B
Actually, no, they don't anymore because they've.
- [58:57]
A
That was my generation, my bad.
- [59:00]
B
They've stopped. And you know, my thesis is that it's because the, the youth now lives in this social surveillance state where if someone did get drunk and like acted silly at a party, you know, everyone's going to whip out their phone. It's going to be, you know, there's the threat of the, the ever present looming threat of it getting posted online and then they wake up the next day with a hangover, open up their phone and realize that their life is like quite literally over. I mean, you think about like the, you know, mean girls and like all of the weird, yeah, you know, brutal social dynamics that happen while, you know, humans are figuring those things out between like age 5 and 18 and then you just magnify that with them being able to be like, look what he did kind of thing. I mean, I think it's, I think it's quite sad and, and potentially really bad. And I, I don't really, I, this makes me hope circling back to AI Strangely, I think that the, the answer to this is when AI gets so good with regards to deep fakes that it frees everyone from the social surveillance state. Because if you do do something silly and someone films it, everyone who sees it on social media the next day is like, it could be fake.
- [60:25]
A
Yeah, it's A.I.
- [60:26]
B
Yeah. So like suddenly it'll, you know, free those younger generations from basically being beholden to this social surveillance state again.
- [60:37]
A
I have six grandchildren and I would not want them to have to suffer that fate. So maybe something we should look into is like a blocker. Well, actually we funded a fellow who is, is, is blocking be surreptitious recordings. Wouldn't it be kind of cool if your phone emitted, it emitted something that made it impossible for the phones close by to, to take a video of you or whatever. I, I, I, I like your answer though because like when everything's fake, like what are you gonna do? I, I do think that that is a real danger because, you know, you, you look at Certain ideologies today. And, and they have zero contact with reality. And that's why you see so many of these idiotic things being done.
- [61:33]
B
Well, that's always been the case.
- [61:34]
A
I agree, I agree.
- [61:36]
B
We've always had like myths and religions.
- [61:38]
A
That are like, yeah, but they didn't have the potential to scale like they scale now, and they didn't travel as quickly as they travel now and they weren't nearly as ubiquitous. Right.
- [61:51]
B
Well, perhaps another counterpoint is that like for example, when, when film was first invented. Like moving pictures.
- [61:59]
A
Yeah.
- [62:00]
B
I mean there's the famous story of the, the first one that was shown in a theater is like a train coming towards the camera.
- [62:05]
A
And everyone freaked out.
- [62:06]
B
And everyone freaked out because they thought it was real.
- [62:08]
A
Yeah.
- [62:09]
B
And then eventually everyone learned like, oh, it's, it's on a screen.
- [62:13]
A
Yeah.
- [62:13]
B
It's actually fake sort of.
- [62:16]
A
It's a, it was real train, but it's a movie of a real train.
- [62:21]
B
Exactly.
- [62:21]
A
Not, it's not coming at me.
- [62:23]
B
So like, compared to that first audience, human behavior adapted and learned.
- [62:27]
A
Yeah.
- [62:27]
B
So I, I, I think that maybe something very similar will happen with, you know, there, there will actually be good in this avalanche of AI slop because in some of it will, will come like the redemption and the freeing from the social surveillance state which I, I.
- [62:44]
A
Am 100% in favor of. I remember I was horrified when my youngest was a, a tweener, I guess, and we were at a cocktail party and all of the other parents were talking about how they had spyware on their kids phones. And I'm like, that is horrible. Why are you doing? Why? And they get really defensive, right. Like, I want to make sure they're safe. And I'm like, you just want to, you know, lord it over them that they have zero privacy. And, and so I, I think, I hope that actually happens because like, what a bummer, man. Like to not be able to like a friend of mine and I were having lunch up at the Core Club many years ago, pre pandemic, and he's of my vintage and we kind of agreed that one of the problems that younger men specifically were facing today was they never got punched in the face. Right. My generation, I can guarantee you, if you, if there's a hundred guys 65 years old and you ask them, have you been punched in the face? 99 will say yes. And, and, but again, what does that lead to? It leads to the dispute is over. Right. Somebody says something you don't like is an asshole to you. Whatever. You get punched he gets punched and then you're done.
- [64:11]
B
It also puts other things in perspective. Like if someone calls you a name, it's like, yeah, words are not violence. I've been punched in the face.
- [64:22]
A
Yeah, words are not violence. Violence is violence.
- [64:25]
B
Exactly. Yeah.
- [64:26]
A
And, and when. But it's, but it's even bigger than that. I, I think that that was the way you adjudicated disputes between young teenage boys and. But then they were done.
- [64:38]
B
Yeah, it's part of a learning experience.
- [64:39]
A
With mommy and daddy calling out the kid and make. Getting them expelled and then, then that kid's parents suing mommy and daddy because they were mean and it just seems like such a waste of time to me.
- [64:53]
B
Yeah, I mean, let kids grow up.
- [64:55]
A
Exactly. You know, like, and that's the most, that's where we do the most learning. Talking to a neuroscientist once and he's like, yeah, your average baby is tripping balls 24 hours a day. That's how the brain develops. Right. And, and, and so this, this, this desire for control and this desire to think that you as a parent can, can make your child this perfect thing, it's just wrong.
- [65:27]
B
Well, I mean, I don't have children myself yet, but I've spent a lot of time with friends who do and I think it's probably like they're trying to prevent the worst case scenario. And given current technology, trying to prevent the worst case scenario also requires doing all of this other Big brother Orwellian stuff. And so going back to your question about other non intuitive ways that AI might help, I can very easily see a way where this will affect this issue. There's been a few devices like the friend device, where all of these sort of lanyard things where there's an AI that you can talk to.
- [66:16]
A
I have one of those, don't I?
- [66:18]
B
Yeah, you do.
- [66:18]
A
I've never used it, no.
- [66:21]
B
But there's these proto attempts at AI companions and they're just not very useful right now because it's essentially like a chat bar that has a microphone that is in a device and there's not anything like terribly innovative and because of that, super useful. But you know, I think back to one of my favorite games when I was younger, Zelda Ocarina of Time. You know, you're like this little kid with like a sword, but you have this little fairy that's following you around and, and her, her, I think her name is like Navi or something. And whenever you are stuck or you needed help, you could like ask her a question or she might prompt and tell you something. And so I imagine in the future, you know, you might have to, you might be able to like not worry about your kid at all because it has like this little AI companion that's falling it around which like 99% of the time doesn't do anything. But then it sees like, oh, the kid is getting peer pressured to jump off a 35 foot structure onto concrete. And it's like, actually this is a bad idea. That's when the AI speaks up and it's like, guys, no, this is bad. And it's like, actually given the density of your bones, if you jump off there and you hit at literally any angle, you will have a minimum of about seven fractures, that sort of thing. So maybe there's like an avenue where like essentially like the, the tracking tech that parents are using to try and, you know, prevent worst case scenarios. It's just another instance of like, the technology is too young. So we're forced to like, we haven't.
- [68:08]
A
Gotten to the iteration that would actually work.
- [68:10]
B
Exactly. We're still like using a hammer on a screw because we invented the screw screw before we invented the screwdriver. And it's like, well, this doesn't work very well, but like, it's the best I got right now. Yeah. So, yeah, I'm just gonna ruin this wall by, you know, hammering a screw into it.
- [68:27]
A
Something I have done.
- [68:28]
B
Yeah, I think we've all done that at some point where it's like, you just don't have a screwdriver and you're.
- [68:34]
A
Like, well, one of my big things when I got married, I got married very young. I was 22 and had my first child, Patrick at 24. So pretty young. But I quickly learned that I needed to selectively underperform in certain things that my wife wanted me to do. And so I became a master of selective underperformance. And she wanted me to do a bunch of work around the house. I'm all thumbs. And so she bought me a drill. And so I'm like, this is very dangerous. And so what I did was even more than I, I really underperformed. I basically took the drill and drilled holes into a basement wall until. And I did it, hoping she would hear me. And she did. And she came down and go, what are you doing? And I sometimes testing out the drill. And she's like, give me that. Let's bring it back to business. And the areas where we're seeing just tremendous leverage from AI. And one of the reasons, by the way, I think that this whole solo founder thing is a phenomenon that might have some legs because much of what we do in our various verticals, film books more so, medium, even more so is leverage AI workflows so that we don't need 100 people on staff. And the people we have on staff are using their time and thoughts for, for much higher level ROI potential coming. When do you think that that will be completely normalized in even like standard traditional corporations, if ever?
- [70:45]
B
I don't think for a while. I mean, I. Someone in the family who's, who's been a software engineer for, you know, his whole life now, two decades plus. And whenever I'm with that part of the family, I always ask him like what. I always like ask questions to try and suss out how much they're using AI and what for. And especially with something like coding, where AI has just made coding like so, so much more fun, right? Because you can move so much faster to the point where like the feedback loop in the personal agency, you know, algorithm there is just, it can be addictive to the point where it's like, like I was describing that, the proposal I was supposed to write and I was like, well, I'll just code up the ui. And then like in, instead of seeing the back end as like previous to AI, it would have been like, oh, this is going to take so much time. I got to like plan this out. I got to, you know, it just feels like this huge, enormous chore that I'm going to have to like slowly chunk away at with AI was like, I'll see how far I can get. Like, why not? And then I had it working within like a few hours.
- [72:01]
A
Yeah, it was amazing.
- [72:02]
B
Yes. You know, so the fact that like that exists on one end of the spectrum, which is very close, close to the, the solo founder thing that you're talking about. But then on the other end of the spectrum you have like a very big company that is employing coders and they're essentially just not really using AI at all.
- [72:27]
A
Self preservation.
- [72:29]
B
There's definitely an element of that because if, if they, if this, if, if certain power holders within this company were really going to face the music with regards to like, what actually is on offer with regards to these tools. Yeah, they might be like, oh, well, they might have two reactions. One might be like, oh, we can actually cut a bunch of people. Another reaction might be like, oh, our existing people can actually do like 10, 100x more. Yeah, like, wait a minute, maybe, maybe we're thinking too small. Yeah, maybe we Have Scope pro paralysis. Maybe we need to like zoom out and be like, wait a minute, what else can we do?
- [73:07]
A
And most of it is performative.
- [73:09]
B
Yeah.
- [73:09]
A
It isn't real.
- [73:10]
B
Yeah.
- [73:11]
A
It's like, sorry guys, but I. One of his things that I've told him is you guys got to get AI'd out. Teasing, teasing. You are artists of the highest order. We would never. Well, we might. We have the best producers in the business. So you guys are probably solid for a while. But you know, I say that jokingly, but that's the central fear that is driving a lot of people kind of crazy and they can't get beyond. What's the great quote? You can never convince a man of something when his salary. Yeah. Upton Sinclair. When his salary depends on him not understanding it.
- [74:04]
B
Yeah, that's exactly what's going on here.
- [74:06]
A
And, and, and in general, I am a. Look, I believe there will be a dislocation. I am not Pollyannish about this at all. I think we have to do something to take care of, of that cohort of society. I've said this multiple times. I'm willing to like throw everything against the wall and on, on fixing that. Primarily because people who through no fault of their own get, get disintermediated and you know, like the think of the 55 year old guy or woman who's been doing the same thing for 30 years and you know.
- [74:51]
B
They'Re really a product of the school system.
- [74:53]
A
Yeah. The point is that talented people. I'm not worried about talented people. Right. You guys will be just fine because you'll figure out a cooler, much more interesting way to do what you love to do.
- [75:08]
B
Well, they'll just start using the tools.
- [75:10]
A
Right, exactly right. But I am worried about the people who aren't cool and good at what they do. And I shouldn't use that. It's a pejorative term who just through. Again, through no fault of their own. Just that isn't their thing.
- [75:27]
B
And what, what are you afraid of happening here?
- [75:29]
A
Well, I would hate to see a bunch of pitchforks down in Union Square. You know, I don't, I don't want any Ned Ludds to become the, the COVID story of. Is Time even a magazine anymore? Do they still have that? Oh really?
- [75:51]
B
Okay, so, so funny.
- [75:54]
A
The magazine Time I, which I didn't know even exist, still existed. They used to have the person of the year. We do not want Luddite to be the person of the Year.
- [76:05]
B
So I asked you the leading question in order to get to that exact Answer.
- [76:09]
A
I, I didn't get. Oh, you fooled me.
- [76:14]
B
Continue. In order, in order to bring it full circle back to the, the future of food thing. Because once you have food forests with automated, you know, humanoid gardeners, what starts happening to the price of food because.
- [76:31]
A
Oh, it's incredibly deflationary.
- [76:33]
B
Exactly. Like once you have the hardware and like, like once you have that transition period, I mean, all you're really paying for is like the occasional replacement part and then solar to power all of these things.
- [76:49]
A
Yeah.
- [76:49]
B
And so like the, the parallel that I like to draw is, you know, why don't we pay for oxygen?
- [76:59]
A
It's because no doubt someone has tried.
- [77:01]
B
Someone has tried, but it's actually impossible to charge for oxygen. I mean, well, maybe not impossible.
- [77:08]
A
There are oxygen bars in Tokyo that are very popular.
- [77:12]
B
True, true. But the distribution system for oxygen is perfect and it's already built into the system. Yeah, the distribution for food sucks by comparison.
- [77:23]
A
Right.
- [77:24]
B
And so like once you automate the growing and harvesting and distribution of food, it will start to trend towards the price of oxygen, which is, and the.
- [77:35]
A
Other thing that's cool about it and you know how excited I get about this, it's not only does the price go down, the quality goes up. And we have never existed in that kind of world. Right? Tech. Yeah. But you know, it speaks to the, for almost all of human history, the, the central story has been scarcity. Right. All of economics is based on the theory of scarcity. This, the allocation of scarce resources. Right. Like in the world we're going into, scarcity becomes a lot less important for a lot of things. Of course there will, you know, there'll be enforced fake scarcity because of that old mindset.
- [78:25]
B
Right.
- [78:26]
A
And because we're status monkeys and. Right.
- [78:29]
B
And well, it's kind of like fasting today. Like fasting is enforced self, self imposed scarcity.
- [78:37]
A
Yeah, yeah. And, and, and so listen, there will always be an elite, right? It'll change as to what that elite is. But I mean that's human nature, right? The pecking order is real. I used to believe that, you know, you could stand outside the status hierarchy and tell like Will Storr and, and my friend Rob, who wrote a great book about it, convince me. No, no, we're all signaling all the time. And, and, but the people will figure out new ways to signal. Right. Like it would not be at all surprising to me if we see a strong counter trend where everything's in person, where, you know, the, the most exclusive ticket you could get is not filmed. It's Just a, a one time event won't be filmed. And oddly you'd be able to make a lot of bank on that.
- [79:41]
B
I think Netflix is actually going in the wrong direction because I think like, I find myself craving movie theaters even more and more and more. In fact, last, last night, what was I. I was trying to watch something with someone and the UI on Netflix was just like annoying me so much because, like, I couldn't automatically skip back to the beginning of the episode.
- [80:03]
A
Oh, yeah, no, don't get me going.
- [80:05]
B
And so I had to like, Apple.
- [80:06]
A
Is the biggest criminal. Yeah, they, they fucking ruined itunes because they let engineers who, you know, want all these cool features, you know, and your average user doesn't want any of them and they make it so complicated that you abandon the app. Yeah, that we went, we. That is what moved me to Spotify. We had here, we had. This was all Apple music, we had wonderful playlists, everything else. And then it just got to the point where they made it. They so inside ified it that it was unusable. And I see it happening to Apple TV now and I'm like, it's. It's one of these things where like, is the purpose here to continue to make money and make more money?
- [80:56]
B
Like, I'm a capitalist.
- [80:57]
A
I can get behind that. Why are you doing this?
- [81:02]
B
Yeah. Last night I kind of came to an unexpected realization because the person I was watching it with while I was trying to like rewind and it was like going at like, gosh, this is taking a while. And the comment that someone I was with said is like, oh, it's like rewinding with a vcr. And I was like, this is ridiculous. But it sent me on another tangent of thought when I was like, oh my God. You know, and this is, this relates to the sort of in person movie theater experience which led me to think that I think Netflix is absolutely going in the wrong direction. I thought about the blockbuster ritual where you have to drive the blockbuster and then you would do the.
- [81:49]
A
Requires an investment.
- [81:50]
B
Yes. You had skin in the game and then you did the perusal where like you had to spend at least 20 minutes just like looking around and you know.
- [81:57]
A
Exactly.
- [81:58]
B
By the end of that 20 to 20 to 90 minutes, you would finally end up with one. You'd be like, all right, I'm confident this is good.
- [82:05]
A
This is the one.
- [82:07]
B
And it really actually didn't matter how bad the movie was.
- [82:10]
A
Right.
- [82:10]
B
By the time you actually get back to and put it into the vcr, you are so sold on this experience that you're like, you're gonna watch the whole thing.
- [82:19]
A
Yeah.
- [82:20]
B
And I think that changes the quality of attention that someone brings to the movie. Whereas, like, Netflix, you know, you can be watching a great movie, you're like, four minutes in, you're like, I'm just not in the mood for something serious. And you just press.
- [82:35]
A
That line sucked.
- [82:36]
B
Yeah. You just press a button, it disappears, and you spend, like, most of your time scrolling. But for some reason, you know, the time spent scrolling is not at all equal to the time spent browsing in a Blockbuster. They're both. They're both time spent. So the impetus is to think like, oh, there's skin in the game on both ends. But, like, no, because you don't have a sense of commitment at the. The end of that time spent with Netflix, you do with Blockbuster. And so that all just came into my head last night. And so I think it speaks to this point where, like, I really think that, like, in person is just going to continue to, like, grow and grow and grow because of that.
- [83:21]
A
Yeah, I. I think that probably for the bespoke part of the market, you're absolutely right. I think even though Barnes and Nobles is opening bookstores everywhere, for example, and, man, I love bookstores. It used to be a family ritual. We would take the kids there every Sunday, give them as much time. It fits right into your Blockbuster thing. But I think 80% of books are moved online and will continue to be. And I think probably 80% of films will be consumed online.
- [84:02]
B
Actually, I'd put a caveat that. I 100% agree with you, but I would say that 100% of book sales where you know exactly which book you want. Yeah, it's probably going to be online.
- [84:13]
A
Yeah.
- [84:13]
B
Or it's going to be like, 99.9.
- [84:15]
A
Yeah, yeah, yeah, yeah.
- [84:16]
B
But then if you want to go browsing.
- [84:20]
A
Yeah. The explorer function, they don't do that well on Amazon. Does not do that well.
- [84:25]
B
And I think it's. It's just more. I don't know if you can capture on a screen what happens in person when you just, like, you have a space that you can walk around.
- [84:34]
A
The Strand Bookstore, which is a few blocks over. I love the place because. And in fact, I would get so guilty because I read most things on the iPad, the Kindle for iPad, back in the day when I used to still have to travel as the primary sales portfolio manager for osam. Like, literally, my bag was nothing but books and always got flagged on weight. Right. And so the minute Missy, My wife gave me a Kindle. Like, I was done that was it. Because I could have a thousand books on this thing, and it was just perfect. So I developed that way, but it never was. We, by the way, we always buy a hard copy of a book that we really, really love just to support the author and also just so that we can have the physical version, too.
- [85:24]
B
This also makes for a good gift.
- [85:26]
A
Yeah. But I got so guilty going over to the Strand that I ended up having to buy things, even though what I would do is just take pictures of the books that I was interested in and then buy them for Kindle. And I'm like, I feel like a cad here. I really have to buy. So generally what the fix was, we love art, as you know, and they've got a great art book selection, and art books are expensive. So I go and buy an art book. Usually they give us a gift.
- [86:03]
B
Yeah, I was just gonna say it sounds like you're speed running a museum when you're spending. You just like, run through and take a picture of the Van Gogh, take a picture of the Rembrandt. You're like, oh, I'm just going to study them later.
- [86:16]
A
We are funny creatures. Let's tease another thing you're writing that came out of a conversation you and I were having that I thought was, like, really hysterical.
- [86:24]
B
Oh, gosh.
- [86:25]
A
This was your insight that at first I thought you were bullshitting. And it has to do with. With the star ratings on books on either Goodreads or Amazon or. Or wherever. And you, you told me, like, the ideal that you want for stars is like, what is it?
- [86:48]
B
4.4.8.
- [86:49]
A
4.8.
- [86:51]
B
Yeah.
- [86:51]
A
Because you got to have some haters. If you get five, it means you span that you.
- [86:56]
B
It looks like a curved.
- [86:57]
A
And you're looking at a SP scam. And obviously if you get one, not optimal. But we started talking about the star rankings on some of the world's greatest literature. Lay some of the really hysterical ones on our listeners and viewers, because I. I was like, you have got to be kidding me.
- [87:22]
B
Okay, so the one that I gravitated towards at first, I ended up looking up a whole bunch, but was the Road by Cormac McCarthy. Now, amazing book. Is it his best book? Like, probably not, but it is spectacular.
- [87:40]
A
A true work of art.
- [87:41]
B
Yeah, it's amazing. And on Goodreads, I think it's a 4.2. And what really, what really, what I ended up gravitating towards is that the top. This is weird, but the top liked Review of this book gave it a one star, and it was long. This review is 3,000 words long, and it was quite well written and quite elegant. He's describing all of these reasons why he thinks the book is bad. And I was like, he's doing a great job describing all of the reasons why this book is actually really good, and yet he gives it one star. And the disconnect of, like, this is such an elegant moron. Like, how is this possible? How could you miss the point?
- [88:39]
A
We found an outlier.
- [88:41]
B
Exactly. He's in a weird quadrant of the graph that we haven't seen before. And so then I started looking up other things like Moby Dick and Lolita, the Bible, Harry Potter and the Sorcerer's Stone, Shakespeare, and the one that, like, out of this.
- [89:05]
A
I gotta say, when you first started telling me these, I'm like, holy shit. Idiocracy was a documentary.
- [89:15]
B
Maybe out of the ones that I just listed, plus like, a handful of others like Don Quixote and whatnot. The highest rated one was J.K. rowling's Harry Potter. And I think it got like a 4.7 or something like that. But, like, she beat Shakespeare.
- [89:36]
A
I think it's hysterical. But I choose to reframe it. And the way I reframe it is, look at these idiots giving Shakespeare lower ratings than J.K. rowling. Although my kids loved her books. They're great books, and they're great books. That's right. But if you're an author or creator or anything, do you need a better example of, you know, the fact that most of the people out there are not terribly bright? You don't have to be afraid. If Shakespeare is getting 4.1 stars, don't sweat putting anything out there and don't care what the majority think. And. And like, that's one of the things that we are trying to really emphasize now. Granted, you know, taste is the final moat, and we are putting podcasts and books and films and everything else out there that we think are in really good taste. Well, I'm sure we'll miss a lot, but the. The notion that you should be afraid of being judged, it just. That's what I loved about this article. Like, these are the greatest literary lights in. In history and, you know, 3.9 stars.
- [90:58]
B
So the, the, you know, you could say the digestif to this story is. After I'd written the article, I was thinking about it a little bit more, and I went back to that review. 3000 word review, one star for the road. And I look at how Many people liked his review. And it was like 3,000 people. And then I looked at his profile on Goodreads and he had like thousands of followers. And I was like, oh, this guy figured out a way to like, Engagement Farm. He's like, I'm going to write an incredibly well thought out review of a book that's going to get a lot of eyeballs. And then I'm going to do the incredibly controversial thing of giving it one star. And I was like, oh, shit, it worked on me.
- [91:49]
A
Yeah.
- [91:49]
B
Because I saw and I was like, well, I gotta, like, I gotta read this. Like, what is this guy talking about? I read all 3,000 words. So he succeeded and that like, essentially he, like, the incentives of this platform are probably misconstruing the reality of what these books actually mean to our species.
- [92:11]
A
Yeah.
- [92:12]
B
In general. And it's all because he found a formula that worked.
- [92:16]
A
Loophole.
- [92:17]
B
Exactly. It's at the detriment of, like, the truth about these books.
- [92:22]
A
Right.
- [92:22]
B
I mean, truth might be maybe lower, lowercase t here.
- [92:27]
A
Yeah, yeah.
- [92:27]
B
But it was clear after that, I went and looked at a few of his other reviews and I was like, son of a bitch. Like, yeah, he found an arbitrage. It's clearly working for, for some reason, I don't know how he's like actually benefiting it from it. Other than clout on Goodreads, which like, buddy, aim a little higher. But okay.
- [92:51]
A
I'm a good reads influencer.
- [92:53]
B
I mean, maybe there's benefit to that.
- [92:55]
A
Are you killing it?
- [92:56]
B
Yeah, I, I, you pay me $10,000, I'll give you five stars.
- [93:01]
A
I'll give you five stars. And I've got 100,000 followers. Maybe that's what he's doing.
- [93:06]
B
Yeah, maybe it works. Like, I don't know, turns out the.
- [93:09]
A
Guy'S making 100 grand a month. We just look like idiots.
- [93:13]
B
Got to respect the arbitrage.
- [93:15]
A
Always respect the arbitrage. All right, I'm going to put you on the spot and say, and I've never asked you this. So it's 2030. What does OSV look like? What breakthroughs or things that we might not even be working on them right now, but 5. And don't make it OSV. Make it a super innovative AI first company. What do you think are going to be the five differentiators that let them soar? And you know, if we want to have really fun, a lot of fun, let's make it the first billion dollar single solo person company. What five things would make that human? Be able to Be a billion dollar corporate.
- [94:08]
B
Okay. One of them, which I think is the, maybe the most pressing and most important, which Misha is actually working on, is the context problem. So right now, the way that AI's work is with retrieval, augmented generation, which has huge downsides. I built a rag system for an early MVP that I, I did in my, like, second month here. And I remember showing Misha and I was like, this is like, so clunky. Like, I hate this system. And he's like, no, this is exactly how like, companies worth millions and millions, hundreds of millions of dollars are doing it. It. I'm like, this is awful. And he's like, I have just the thing for you. And so he's working on that. If we can solve that context, which is really like a memory issue, if we can solve that, that really unlocks just like a tremendous number of things. Because the way that the AI field's been dealing with that first is just trying to enlarge the context window. You can only do that for so long before you start to get diminishing returns. I don't want to have to dump in the entire Encyclopedia Britannica in order to have a query, because then it's just going to take a long time. It doesn't matter how.
- [95:37]
A
Spoiler alert. We do have a looks to be solution for that.
- [95:43]
B
Yeah, I'm very excited to put this thing through the wringer. So that's definitely number one. Number two, there's clearly some data problems. Like, for example, whenever I hear an artist bemoan or begrudge the fact that, oh, AI can create art, but it can't, like, do my dishes for me. And I try to explain, like, why that's the case. It was like, all right, how many pieces of human art are. Were there on the Internet before AI became a thing? And the answer is like, well, all of it, obviously.
- [96:20]
A
Yeah.
- [96:21]
B
And I was like, how many examples on the Internet are there of first person point of view, Hands washing dishes? Like, maybe there's one weirdo YouTuber who did this, but like, the answer is like, almost nothing.
- [96:36]
A
Yeah, yeah.
- [96:37]
B
And so you can't train a system on something unless you have like a huge amount of data. So there are many areas in which we're like, very data poor. Yep. So there are some, like, really interesting solutions that are popping up. There's a really cool robotics model that just popped up where you can essentially wear like meta glasses and like, you know, film your hands doing some particular task. And it translates really well into the system where those robot hands can mimic. So essentially like mirror neurons, I guess you could say very loosely. But there's definitely like a whole bunch of other sources of data that we just, we're a little bit blind to. And the analogy that I like to use in this case is, you know, why are Lego manuals not text only? Is because it would be so obnoxious to read and it's so difficult to describe certain things of like oh, the, the two by four by one block.
- [97:45]
A
The limits of linguistic description.
- [97:49]
B
Exactly. So like I do a lot of woodworking and I've primarily done this like by myself because I enjoy it. And there's been a few instances where like a friend will come along and be like, oh, can I help? And I'll be like, sure. And then I'll start trying to like push what I've been doing through the language part of my brain and it's just like, you know what? No, you can't help because like my, I'm just shape rotating over here and it doesn't fit through my brain. So there's this huge like 3D area of reality which language doesn't capture very well. And so I'm using that as an analogy to say that there's all sorts of data that we're invisible to the fact that we don't even have it, which we can't even give to an AI. So I think the next thing there will be AI plus robotics that is simply for gathering data of all sorts of different types of number three people anthropomorphize AI a lot. And why doesn't AI prompt? Like there's all sorts of little things that after like a lot of deep use, I'm like, it's because it's not, it's like not curious that it, it didn't figure out like 10 chats ago what I was really trying to ask. And it took me like a whole bunch of like, you know, chiseling away at the granite in order to get to the shape, in order to ask the right question. And it's like if it had had some basic curiosities and basic drives, it would have been able to do what a human would do is like, oh, what you're really trying to get at. You just don't have the context and the language and the experience in order to get there. It's like, oh, you actually want to pay attention to variable X, Y, Z, but you're stuck on A, B, C, D, E, F in order to get there. And so I, I think that kind of probably requires something like a limbic system, which makes me nervous. I, I don't want a limbic system like created and attached to AI because then you could get some really weird incentives. But you would get some.
- [100:01]
A
I'll be back.
- [100:02]
B
Yeah, you would probably get something that's like way more agentic than chat.
- [100:08]
A
Yeah.
- [100:09]
B
So I think, I think like the words age, the word agents in AI is a bit blown out of proportion in terms of like what AI agents currently can do. You know, really from my experience and my understanding, you're really talking about like a recursive set of LLMs that are referencing basically a scratch pad or several in order to like go through these different things.
- [100:41]
A
Yeah.
- [100:42]
B
Is that really agentic or is that kind of like a Rube Goldberg machine that's made of jello? I'm not sure, but I think like true agencies going to require something that is going to be like an analog to the limbic system, but hopefully very different from the limbic system because ours was developed under very different circumstances with evolutionary pressure which gave us a whole bunch of bad os.
- [101:13]
A
Yeah.
- [101:13]
B
Which we don't want. And this is where like the, the what I think is a pretty silly debate about like the alignment problem actually comes from this area. Like, you know, I mean, we can't even align ourselves. Like, I don't think alignment as a word is particularly meaningful in this debate. And I agree it seems to stand as like a very solid flag in either camp. But yeah, I think that the development of an AGI appropriate limbic system is going to be necessary in order to really spark life through things like LLMs that are not human input. Because right now it's human input that gets the. Because I picture an LLM as essentially like a giant Galton board or like a bell curve with all of those with a bunch of pins and then you drop the marbles and then they create the bell curve. So I kind of. If you take out all of the balls and you just have the pins in the grid, any one input or query is essentially you're like dropping a marble at the top and most of them are going to fall straight down as part of that bell. And then every once in a while you're going to get a weird like, oh, that was strange. It told me that maybe I should end this relationship kind of thing. Thing. So yeah, I think an AGI appropriate limbic system would. Would be the thing that would breathe life into.
- [102:51]
A
Into that might be worth more than a billion. I would like an MVP of that a week from Today.
- [102:57]
B
Okay, yeah, yeah.
- [103:00]
A
You set yourself up for. What haven't I asked you that you feel passionate about in terms of anything that we're working on or trying to change.
- [103:17]
B
I am droolingly awaiting the day when we have an OSV os, which is essentially a system that has context of every little thing that's happening in osv. This is so necessary in order to combat the desire to bloat from a bureaucratic standpoint. Because I've witnessed in, in many times in my life that there is, there's this perennial desire to discuss something, come up with a solution, and then hire a third party that is that solution. And I'm like, no, don't want to do that. Like, let's just, let's just do a quick and dirty version of it with what the tools that, that we have. Because then at the very least we'll just like learn what's possible. If it really isn't possible, it's like, okay, let's, let's either build something more intense or look at other third party options if we need to. But I want to learn at the very least. So if we had. And I've noticed in a few instances at OSV where someone's working on a problem and then like a week later I discover that this person actually has fairly deep domain knowledge on set problem. And sometimes this person is me. And I'm like, oh, I didn't realize you had all this knowledge. Like, we need, I need to like hoover up a bunch of information from you because I'm building something in a domain that I've never built in before. And if there was simply instead of when I type into a chat bar, I'm talking to an LLM. What I really want to do is I want to talk to osv.
- [105:09]
A
Yeah.
- [105:09]
B
And I want to be like, do we have. Where does this piece of information exists? Can you like pull it up for me? And like a really cool thing that like, like the, the beginning sort of like ember spark of this was a little while ago. I wanted to pull some information from one of the CRMs that we do. And I was like looking at it, I was like, like, I just had enough of a headache. I was a little too tired. I was like, I don't want to try and figure out how this works works. And so I went to the settings and I pulled an API key and then I threw it into Rai Lab and I was like, here's the API key for this product. This is the information I need. Can you do this? And it just did it. And I was like.
- [105:56]
A
I remember you telling me, I was.
- [105:57]
B
Like, oh my God, thank you. Just saved me. So much like frustration.
- [106:02]
A
Yeah.
- [106:03]
B
So, yeah, I want that. That just like, you know, a spider with a leg in every corner kind of thing.
- [106:10]
A
I love the OSV os. I think that that is another thing that also fits in to the. The billion dollar solo practitioner. Well, this has been amazingly fun. I always, of course, get to have these conversations with you daily. But as you know, you've got your chance to be emperor of the world. Can't kill anyone, can't reeducate anyone, can't kill any sparrows. But what you can do is talk into the magical microphone. Incept all 8 billion or whatever we're at right now, people on the planet, whenever their next morning is, they wake up and the two things that you are going to incept are front and center in their brain and they're like, you know what? Unlike all the other times when I ignored all of these great insights I had, I'm going to actually act on these two things today and continue to do so. Dazzle me. Okay.
- [107:10]
B
I would incept them to try and foster two things simultaneously. Foster they sound mundane on their own, but I think it's in the combination where the magic happens. Foster a bigger imagination, just like always. Try and break scope paralysis. I love that little clip from inception. From inception when the guy's shooting with a pistol and then Ed Hardy comes over and he's like, you have to dream a darling dream a little bigger. And then he holds up a rocket launcher kind of thing that sort of enlarge your imagination, you know, coupled with have as low of a tolerance for bullshit as possible. And the reason I say these two things together is because I think there's this assumption that like, people with a huge imagination are just like kind of full of bullshit. And then people who are like, no bullshit, like, don't have any imagination.
- [108:16]
A
Right?
- [108:17]
B
Where the, the real arbitrage is the asymmetry of like having zero tolerance for bullshit. But like a huge imagination, which means that like you can go out into cognitive space and, and take something that is truly innovative but then actually translate it into reality. Because you don't. You don't. You can't put up with any bullshit about the fact that like, oh, it's just an I, you know, oh, it's a great idea. It's like, no, it's not a great idea until it's here and real and you can do something with it. Yeah, that's what I would accept the world.
- [108:54]
A
Love both of them. Thank you for taking the time.
- [108:58]
B
Thank you.
- [108:58]
A
And everyone watching and listening. We have these conversations all the time and sometimes they go off on tangents. Thanks for being on.
- [109:08]
B
Thank you. Appreciate.
- [109:15]
A
Sa.