Latest / Great Houses / 26. Cursus Honorum of the AI Future Part 1

Transcript

26. Cursus Honorum of the AI Future Part 1 · Great Houses

• 0:00
  All right.
• 0:01
  Hello everybody, and welcome to episode 26 of The Great Houses forum, where we talk about what it means to be a great house, what it means to be a institution that, that owns your own property, that employs your own children, and that, in general, is a pillar of society.
• 0:17
  That's at least, uh, what I'm, what I'm going for.
• 0:19
  Um, my name is Gregory Treat.
• 0:21
  I am an estate planning attorney licensed in Texas, and I do s- uh, family governance and succession consulting for high net worth families around the country, and, uh, soon… hopefully soon around the world.
• 0:32
  And, uh, so today we are launching a new series.
• 0:37
  Uh, the goal of the series is to answer a question that many of you have been asking since I first started framing AI as an aristocratic technology.
• 0:43
  And the question, of course, is I think, uh, you know, one, one of the guys that's on the call here asked me the very first time that I, I laid this out.
• 0:50
  He goes, "Greg, okay, let's assume you're right. What, what does that mean? How should we train our children if, uh, AI is an aristocratic technology?" And, and so that's kind of asking, you know, what, what, what is, to the extent that we can discern it, what are the broad strokes of how AI is going to reshape our world?
• 1:07
  Is this, you know, just totally new?
• 1:09
  Is this something that, that we've never seen before?
• 1:11
  And to some extent it, it will be.
• 1:13
  Um, but I, I think there will be, there will be things that rhyme with stuff that we have seen in the past.
• 1:21
  You know, one of the, one of the things I was having a discussion with somebody and, and they're like, "I mean, how, how can you compare a pre-industrial society with a post-industrial society?"
• 1:30
  And I said, "Well, I mean, let's, let's describe what we're talking about, right?
• 1:34
  Uh, like, by post-industrial, do we mean a society where 2 to 5% of the population creates essentially all of the wealth and, and, and has, you know, multiple wealth creation, right?
• 1:46
  So 2 to 5% of the society produces N or 100 times as much wealth as they need personally to survive, and the remaining 95%, 98%, whatever, of the population produces, you know, marginally more than they need to survive, 10%, 15% more than they need to survive in terms of economic output.
• 2:03
  If that's what we're facing, I mean, that's, that's a scenario that, that the human species has faced before and handled pretty well, you know?
• 2:11
  I mean, there, that, that, that's essentially feudalism, right?
• 2:14
  That's the, the, the way that the economics work in a feudal economy.
• 2:18
  And so, uh, you know, there's, there's things to like about that, there's things to not like about that.
• 2:22
  There's certainly ways to do that wrong, but it's not like we've never faced this before.
• 2:27
  So really, you know, what, what, what broadly could this world look like, and what kind of person will master that world, right?
• 2:34
  So as I've said, you know, in the past, though, though we've got some new people and it, and it bears repeating, uh, I frame an aristocratic technology is a technology that's expensive, time-consuming to train.
• 2:45
  It requires sometimes innate talents or unusual kind of psychological profiles, and once mastered, it gives its wielder a profound, I mean, effectively insurmountable advantage over non-users, right?
• 2:57
  Uh, you know, the, the classic example of this is the knight.
• 3:00
  You know, we, we actually don't have a word for the tech stack of the knight.
• 3:03
  The knight had, you know, h- horses and, and armor and swordsmanship and the lance and the, and all, all of these different things, the lance and the bow and all this stuff, right?
• 3:11
  But, but it was the person that we focus on, the trained person, because you could give all of those tools, you could give a war horse and armor and swords and shields and lances and bows to non-knights, and they would not succeed.
• 3:25
  They would not have the same kind of impact on the battlefield as a fully trained knight.
• 3:30
  You know, the, the, the kind of the, the last aristocratic technology that we see in, in the progression is the longbow, right?
• 3:36
  Which famously took twenty years to train a, a, a, somebody to pull a war bow, and then, you know, you, uh…
• 3:44
  As opposed to a firearm, which took about six weeks, right?
• 3:48
  Um, so you needed… And the other thing about the longbow is you, you really had to start them early, like before puberty, preferably around six years old.
• 3:54
  You had to feed them, you know, a lot of meat.
• 3:56
  They had to be part of the martial class.
• 3:58
  And so what that meant was you had to have somebody who was in the successful class, the elite, if, if you will.
• 4:04
  So you either had to be a child of an elite, or you had to be connected in some way to an elite, and that elite had to say, "Okay, I'm, you know, going to pay for this kid to go through this exceptional process that's gonna produce, hopefully, you know, a, a, a faithful yeoman who's gonna, who's gonna fire a cloth yard shaft at my enemies and not against me," right?
• 4:23
  Um, that, that's, that's the last aristocratic technology.
• 4:27
  It was… That was still disruptive.
• 4:28
  The longbow was, was a pretty disruptive technology.
• 4:31
  But then the crossbow and then the firearm kind of ultimately just became… It was so easy to train people that anybody that was willing to arm the peasants, right, with firearms realized, "Hey, we can just kinda take over these feudal societies," and that's basically what they did, right?
• 4:47
  But an aristocratic te… The, the idea that there are certain technologies that just naturally create hierarchy, that, that relatively small numbers of people… Like, in an industrial economy, you're talking thirty, forty, fifty percent of the population and maybe eighty percent of the male population is actually useful.
• 5:06
  Generates, you know, multiples of, of what they need to survive in terms of, you know, like the subsistence, you know, the, there's subsistence farming where you're just barely producing enough to survive.
• 5:14
  And then in an industrial economy, people are producing way more than they need to survive from an economic perspective.
• 5:21
  And, and, and that's great, right?
• 5:22
  That's what, that's what we had.
• 5:24
  That's kind of what created the modern, the modern democracies.
• 5:27
  That's why I call it democratic technology.
• 5:29
  One of the, one of the ways that, that you can think about a democratic technology is democratic technologies are, are that, are, are any tech stack that allows you to have a stable democracy.
• 5:39
  And I would posit to all of you as, and I made this case, you know, for this is for the new guys.
• 5:42
  I made this case extensively in past episodes.
• 5:44
  I do not think we have a tech stack that supports stable democracy today, right?
• 5:49
  Um- So, you know, one of the features of an aristocratic technology is, is the ter- the tools do not determine the output, right?
• 5:57
  In, in a democratic technology, when you're pulling a fire-- pulling a trigger of a firearm, a certain number of bullets come out.
• 6:03
  Now, you still gotta aim, and there's certainly some, some aristocratic elements to, to certain types of firearms stuff, you know, particularly, you know, sharpshooting.
• 6:10
  Uh, but in general, when we think of the types of infantry engagements that created the modern world, the, the…
• 6:18
  it's, it's a numbers game, and you're kind of massing people who are shooting.
• 6:22
  And, and, and aim matters, but, but, uh, what's that famous line?
• 6:26
  You know, "Quantity has equality all its own." When you're operating in an aristocratic domain, you cannot replace skill with capital, right?
• 6:33
  You, you, you… it's the high-quality men that are fundamentally scarce, okay?
• 6:38
  You know, Machiavelli once said, "Gold cannot always find good soldiers, uh, but good soldiers can always find gold." This is a, this is fundamentally an aristocratic statement and, and, you know, many people would say that politics is a fundamentally aristocratic domain.
• 6:51
  But Machiavelli certainly believes that.
• 6:53
  Now, the modern world relies on the fantasy that there is no lag, no development time between what capital wants and what it can get, right?
• 7:01
  So I mean, this is kind of the, the offshoring crisis, right?
• 7:04
  There's people that are like, "Well, we're just gonna, we're just gonna allow anyone in the world, uh, to give us the cheapest goods that they, that they can, um, and gl- we're not, we're not gonna be concerned with whether or not, you know, the Chinese government is propping up their local industries or undercutting us," and you're effectively, you know, competing with the state, right?
• 7:22
  Which is, you know, that's one of the reasons for the, the development of the, the modern multinational corporation is that is actually the scale at which they have to play.
• 7:29
  They have to be capable of contending with, at a minimum, you know, the, the, the minor states and, and on some, in certain respects with, with major states like China, okay?
• 7:38
  So but, but the modern, the modern concept is, well, well, if there, if there's a need, then capital will be allocated, and capital will always be able to find an efficient expression.
• 7:51
  So the, so when capital wants it, there will always be a car factory set up, running, and able to produce cars.
• 7:57
  We just gotta direct money to it, and we gotta dir- and then, and we gotta direct the, the, the steel and the other inputs to it, and then bam, cars will be being produced.
• 8:06
  And that's just not true, right?
• 8:08
  Like, that is, that is not how, a- as it turns out, the world works.
• 8:11
  There is a lot of lag time.
• 8:13
  There is a lot of friction.
• 8:15
  There is some really h- thing, things that are hard to do in terms of, you know, arranging people and property in virtuous order so that you can get high-quality products like cars and, I mean, even, you know, toasters and things of this nature, right?
• 8:28
  These are very, very hard things to do, um- And so I would say we are, we are surrounded by evidence that, that in fact the quality of the underlying human activity matters and matters a great deal.
• 8:38
  Under- underlying human activity is how my friend Mike Faulkner, who was on the podcast a couple weeks ago, how he, he puts it.
• 8:43
  Like we're all financial abstractions are effectively at some level a call on human activity, an ability to say, "I got a claim on this future human activity." And, uh, you know, so M- Mike has, has made a lot of money being able to, to identify, hey, you know, this, this human activity is actually superior or the, you know, the, the, the underlying human activity that this claim is on is much better than this other, uh, the, this other, uh, human activity.
• 9:07
  And so, uh, that, that's kinda the, the short story of how he's been so successful, right?
• 9:12
  But we tend to ignore this evidence that, that skill matters, that there are hierarchies.
• 9:18
  Um, and I think we're, we're ignoring, you know, the output of a level playing field, right?
• 9:22
  One of the, one of the, the great conflicts in modern life is, is, uh, you know, there are people that want a level playing field because they think it will lead to egalitarian outcomes, and then there is the non-communist position, which is we want a level playing field because it will lead to virtuous hierarchies.
• 9:40
  Hierarchies are gonna form, right?
• 9:42
  Um, you know, the non-communist position on these matters is basically things left to themselves in a true level playing field will sort themselves into durable hierarchies, okay?
• 9:52
  Um, now, there are … When we have a new technology, we don't know what the outcomes are.
• 9:57
  We have an information gap, right?
• 9:59
  So this always creates, whenever there's a new technology, it always creates a period of time where egalitarianism looks plausible, okay?
• 10:06
  Um, and so what we, what the, the, the, the, the success, you know, the, the initial success strategy is treat everybody equally or at least treat everybody well because we do not know which one of them will turn out to be relevant or, you know, super successful, right?
• 10:21
  Um, so And one of the, one of the points that I wanna make is that, that ironically, aristocratic technologies can involve a massive reduction in the cost of access to the tools.
• 10:33
  Uh, that's, that's because the main bottleneck in many of these scenarios, the scarce thing, is the people who have the raw capability and have put in the training to use the tools, not the tools themselves.
• 10:45
  So there's a relative oversupply, and I'm, you know, kind of… I got a bunch of libertarians in the audience here, so I'm over-indexing on kind of economic language.
• 10:53
  But there is an oversupply of, of the technology or the, the, uh, the tools, and not necessarily because the technology is, is cheap or even abundant, but because the, the scarcity is in the, the number of peak users, okay?
• 11:06
  And we'll, we'll come back to, to, to this idea of peak users.
• 11:10
  Um, I spent an episode talking about, you know, domain mastery and this idea that, that, uh, there are certain… Mastery in certain domains is transferable and to, to, to domains that you can kind of analogize or, or have a, have a, have a framework that says, "Principles that I learned over here work over there."
• 11:26
  Um, but you have to have some metaphor.
• 11:28
  You have to have some mental way of accessing the old skills in a new domain, and there's, there's domains that are, you know, relatively near and relatively far, and relatively near means there's a high efficiency, you know, when you try to access the skills in the old domain.
• 11:41
  And then there's other domains where when you try, um, when you try to access skills from another domain, you know, the famously, you, you try to move between engineering and politics, and you're gonna run into some real hard times because the, the operative principles of those two, two domains are so different, okay?
• 11:57
  But when, uh, when you have a new technology, frequently that new technology will create a bridge between domains, right?
• 12:04
  So, so the, the technology is an interface which, you know, by definition is an abstraction, a metaphor that allows you to manipulate all of the domains that the technology touches, right?
• 12:15
  Um, and so when we ask things like, "What, what are you doing?" Um, people tend to answer that, as they should, in terms of the purpose they are putting the tools to, right?
• 12:25
  There's that, that famous story about, you know, two men, you know, marching along in medieval Europe, but one, one guy is pulling rocks and, and the other guy is building a cathedral, right?
• 12:34
  And the guy that's building a cathedral obviously is, uh, sees a much more value in what he's doing and, and kind of is, is, is doing better, at least how, that's how we would frame it, uh, normatively.
• 12:45
  Um, but from the perspective of the outside observer, you are using the same tool, right?
• 12:49
  So pretty much everyone that's on this call is using a computer, right?
• 12:53
  That, that, that is the tool, the, the, the interface that allows… And, and when we say a computer, right, what we mean specifically is a screen, a, you know, keyboard, and some kind of mouse or pointer device, you know, trackpad, whatever, right?
• 13:07
  That's what people are using.
• 13:11
  Um, so, you know, we need to, we need to understand that, that, that, that when, when you have a new tool, it creates these new linkages, right?
• 13:22
  So, so you have a new tool, and then, and then, you know, returning to kind of the information gaps, right?
• 13:27
  You have a bunch of people that are using new tools, and the short-term strategy is we don't bet against any player.
• 13:32
  This is essentially a, a let's not offend anybody type of strategy.
• 13:35
  But long-term wealth is always created by a bet that X hierarchical structure can reliably produce Y outcome, not a simple let's open the floodgates to everyone and charge a fee, right?
• 13:47
  Um, it's, it's, you know, the, the there's these ideas of, you know, the, the pick-and-shovel guys, right?
• 13:54
  Um, so the, the old joke is that when, when you have a gold rush, the people that are gold mining, that are actually going mining gold, they didn't make very much money, but the guys that are selling the picks and shovels to, uh, the gold miners, those guys are doing, are doing well.
• 14:07
  And, and that is true in the context of a gold rush, right?
• 14:11
  In the context of you're selling to a bunch of people who are by definition making bad bets, right?
• 14:18
  At some point, the gold rush ends, right?
• 14:22
  And it ends basically because enough people recognize, "Hey, this isn't working for me. I'm not actually making money."
• 14:29
  Um, you know, this is, this is kind of the, you know, the, the thing that we're dealing with here, right?
• 14:34
  Um, so, so that's what makes the gold rush end, and at that point, the pick-and-shovel guys, they don't have anything.
• 14:40
  After that, it's organized groups of people who are doing real mining operations, who have real organizations, right?
• 14:47
  Again, you know, the, one of the other features of, of a gold rush type of environment is, oh, well, we're gonna go out, and one individual guy swinging his pick and gathering some gold or panning for gold himself, he can be an, an institution all himself.
• 15:00
  He can make money by himself, right?
• 15:02
  There's, there's some, there's some setup where labor is so under, under supplied, there's so much incentive for going and doing this work that one person by themselves can, can create, you know, that, that's enough efficiencies of scale, uh, for him to make, make it go, right?
• 15:18
  Um, and, and again, that sort of thing tends to, um- Um, that sort of thing tends to work itself out because there's an initial group of people that really do well, right?
• 15:31
  Then there's a bunch more people that flood in as individuals.
• 15:34
  They don't tend to do well, and that's the moment where the pick and shovel salesmen are, are, are successful.
• 15:40
  But that's a moment in time, okay?
• 15:42
  And, and, and that's actually not a stable business model.
• 15:46
  The stable business model is we create a hierarchical structure and that hierarchical structure cr-- you know, predictably, uh, creates an outcome.
• 15:52
  Okay?
• 15:54
  So, um, you know, one… Again, one of the things about aristocratic technologies or an aristocratic society is a lot of times, you know, these, these sort of things, it requires the adoption of a different mindset, maybe even an altered state of consciousness, which is difficult to attain or potentially psychologically damaging, right?
• 16:09
  Uh, you know, the, this is, this isn't, this is sort of a, a tongue in cheek example because this is kind of more of an institutionalized, uh, an institutionalized way.
• 16:19
  Uh, but you, you think about law schools.
• 16:20
  You know, traditionally law schools would take in twenty-five to thirty percent more people than they expected to graduate because they wanted law school to be really hard, and law school was really hard.
• 16:30
  Um, I remember my civil procedure professor first semester of law school and, you know, he said, "You know, we expect twenty-five percent of you to drop out in the first year, and it's my job to take the first ten percent." Right?
• 16:40
  Um, and he really did, you know, pick on somebody who had not.
• 16:44
  He went around the room till he found somebody who was not prepared, and he picked on her until he ma- she, he made her cry.
• 16:49
  Um, amazing stuff, right?
• 16:51
  Uh, I went to law school before twenty fifteen, so I, I, I pr- I assume that you can't do that anymore.
• 16:55
  But, uh, um, and, and, and when I, when I talk to older lawyers, it's like, you know, going back earlier and earlier, uh, you know, they used to do even more of that, right?
• 17:04
  So training for these types of psychological states is always, uh, closed source and frequently kind of deeply illegible, and that's because people, you know, it's just people are sort of uncomfortable with it, right?
• 17:14
  Um, you know, one of the, one of the things that, that, that, you know, going back to the legal stuff, there's a skill to writing legal briefs.
• 17:23
  There's a skill to briefing a case, and basically the way that we teach legal briefing is they force you to do it to a bunch of cases, and you read a ton of material, and you're very sleep deprived.
• 17:36
  And at some point after the first, you know, couple, six weeks, right, two to six weeks, your brain will just kind of lock in, and then you can brief a case super efficiently.
• 17:46
  Your brain, as you read through it, your brain is kind of already drafting the brief, and so you scan through the case and you write the brief and it just comes out.
• 17:53
  It's great, right?
• 17:54
  Um, and, and there's a certain amount of people that under that amount of pressure, they just kind of have a breakdown.
• 17:59
  And, and it's not a good thing for them, and they're not okay, and they, they wash out of school and, you know, some of them don't do very well after that, right?
• 18:06
  It kind of is a…
• 18:07
  It can be the kind of thing that puts somebody in a downward spiral and, you know, in our kind of modern culture, downward spirals, uh, they can go pretty far down, right?
• 18:16
  So, uh, when you're doing that, when you're doing something that has a risk of, of genuinely kind of harming people, uh, but it's really productive.
• 18:25
  Like, it's, it's super useful to be able to brief a case.
• 18:27
  But you have to acknowledge, hey, like this, this is not the sort of thing, at least the way we tend to do it, um, this is not the sort of thing that you, you kind of want to do.
• 18:37
  Um, certainly not, like, one of the things that, that I've, I've thought is, you know, how do I, how do I train my children?
• 18:44
  'Cause there's, there's some, some things about legal writing, it's very rigorous.
• 18:47
  Um, and, and, and I was very frustrated in law school 'cause I felt like I could have learned many of these skills earlier in life, um, had, had someone bothered to teach me.
• 18:55
  Uh, like I, I, I felt like I could have learned legal writing, you know, starting around 12 years old, right?
• 18:59
  Um, the, the, the, like the classical, you know, the, the, the trivium was, was, was, was waiting for me, right?
• 19:06
  Like I, I didn't know about that at the time.
• 19:09
  But, um You know, I, I, I, as I've thought about it, it, I understand now why it's so hard for people because you have to, you have to drive the kids that are learning this stuff.
• 19:20
  Like, it's, it's not fun, it's not pleasant, and it's not pleasant for like a month or a month and a half.
• 19:27
  And, and that's a really hard thing to do to your kids.
• 19:30
  It's a really hard thing to do, um, you know, as a, as a parent, as somebody who, who, who loves your children, right?
• 19:36
  So, uh, th- these, these things, they tend to be deeply illegible, right?
• 19:39
  No, you know, that, that… And, and increasingly in our kind of modern culture, they're, they're just passing away, right?
• 19:44
  The rigor is going out of, out of things.
• 19:46
  Um, so you know, those who fail the training are often actually less suited to normal life or other professions than they would've been had they not attempted the training at all.
• 19:54
  We kind of talked about that, you know, last week, um, when we, when we were talking about the, the architecture of trust and, and this idea that there are real opportunity costs, right?
• 20:05
  There are real… If, if you tell someone, "Hey, you should go in this direction. You should, you know, get this degree, or you should take this specialty," there are, you know, y- you're, you're constraining their options in the future.
• 20:20
  Um, and, and so for some people, that's not gonna work out very well.
• 20:24
  And, and if you're the person giving advice, what responsibility do you have to people?
• 20:28
  So, you know, one of, one of the kind of the corollaries to this is that in order to, to attempt the training, like as an individual, maybe it doesn't make all that much sense for you to do this training, or maybe it makes less sense, right, than we, uh, than we like.
• 20:41
  Um, so y- you have to figure out, okay, if, if we're gonna do this, if we're going to say, "We want you to attempt this training," um, it, it's gonna be hard.
• 20:52
  It, you know, people talk about kind of the, the military, um, the military aspect that there's, there's always risks whenever you train to do lethal things.
• 21:00
  There's, you know, potentially lethal risks that you're walking around with.
• 21:03
  Um, especially at, if, if you're gonna train to a high level, you have to put people under the same conditions or, you know, ideally, you know, I'm, I, I'm not a military guy myself, but I've, I've, you know, read books and listened to people and the, the people, all of the people that study that say you actually have to train harder than you anticipate the missions being, um, because if you don't, then, then you will fail in, in, in big, bad ways.
• 21:26
  So, um, so y- as an individual, you're looking at that and, and you're saying, "Well, why should I do this?
• 21:33
  This is a, this is a high-risk thing." So you have to be able to have groups, sort of intermediary groups, who then get cohorts of people and say, "We are going to incentivize all of you to go down this difficult and dangerous path, and we are going to reap the bene…"
• 21:47
  That means two things.
• 21:48
  Number one, we are going to reap the benefits of the ones that succeed to some degree, right?
• 21:53
  You don't, you don't want these to be slave contracts, but you, you need that insuring institution to be able to get some benefit from the, the limited amount of people who succeed.
• 22:02
  And you need some duty, some enforceable duty, uh, to take care of the ones who fail, right?
• 22:08
  Um, you know, sort of Evel Knievel style, right?
• 22:10
  You get paid for the attempt.
• 22:11
  So this is particularly true when the maximum number of people who possess the innate qualities is likely around two percent, right?
• 22:18
  Maybe, you know, five percent with, with the most generous assumption.
• 22:21
  So the, the combination of relatively rare hereditary traits and illegible training requirements means that rank in-- when, when you have a tech stack that, that, that bends this way, then, then rank is gonna tend towards households that can fund the training and create privacy, that necessary kind of illegibility around the training, right?
• 22:41
  Um, so, like just to lay out what I, what I'm, what I'm thinking here.
• 22:44
  I believe that AI will produce a hierarchy, right?
• 22:47
  I believe that that hierarchy will have an apex class, a, a relatively stable group of people who can wield their, the new technology at the highest level.
• 22:55
  And I think that that apex class, that apex group will be so valuable that entire structures will be designed, you know, to kind of designed around them.
• 23:04
  We're gonna say the most important thing we can do is find and/or develop more of these people, and if we kinda break a few eggs in the process, it is what it is, right?
• 23:15
  I mean, you think about, you know, all, all of the things that we have done to our educational system looking for geniuses, right?
• 23:21
  There's, there's, uh, one of my favorite conspiracy theories that I'll, I'll share with you for the moment is, uh, um, that, that we have designed our modern reading curriculum not to actually teach people to read, but to help quickly identify geniuses in even the first and second grade.
• 23:39
  This is especially relevant, um, after, uh, IQ testing became generally illegal after, uh, the, the Griggs versus Duke Power case and then the Civil Ra- Rights Act in the early nineties that, that Congress passed kind of reinforcing this idea, you can't use IQ tests.
• 23:53
  Though I'll note that the Trump administration is making noises about, uh, pushing back on that, and I think they, I think they have an executive order that actually says, "No, you can do IQ testing."
• 24:01
  Uh, I don't know if anybody's doing that right now, but, uh, but it's fascinating, right?
• 24:05
  So our-- but, but the point is it became very difficult at a broad level to do IQ testing and, and certainly at a, at a private level, right?
• 24:14
  So universities effectively can do IQ testing, and the US military can do IQ testing, and everybody else has to rely on those two institutions to kind of sort people and to give them, you know, credentialed, insulated results, which are al-always, you know, kind of less optimized.
• 24:32
  They're, they're optimized for the thing that the university is doing or the thing that the military was doing.
• 24:36
  Not necessarily the thing that that institution or business is doing.
• 24:41
  Um, you know, which is kind of-- that's, that's one of, one of the core reasons for the problems that we're having in our society is, is that basic paradigm, okay?
• 24:49
  Um-
• 24:51
  But if we have apex users of this new, uh, new dominant technology, they're gonna be really, really valuable.
• 24:56
  And just as we, you know, reoriented our whole educational establishment to find the geniuses, right?
• 25:03
  Because when y- when, when you take away the ability, when you, when you teach people to read in an inefficient way, which is basically what we do, right?
• 25:09
  Look, say, it's a very inefficient way to teach people to read.
• 25:12
  Then, um, the people that are naturally gifted and can kind of figure out for themselves how to read, they pop in the test scores, right?
• 25:20
  Whereas if you give, you know, like classic phonics, which is a very efficient algorithm that can teach a bunch of people to read at a competent level, well, that means you're kind of pulling up people that, that otherwise would not be able to figure out how to read.
• 25:33
  And so you're, you're creating clusters, which makes it harder to identify geniuses.
• 25:36
  And that's really, you know, in my view, why we don't teach phonics more broadly.
• 25:40
  Um, though, you know, it's, it's, it's fascinating.
• 25:43
  Phonics keeps coming back because it, it, it really does work.
• 25:46
  Uh, but we know how to do it.
• 25:47
  We know how to teach people to read.
• 25:48
  Uh, but we don't do it because we're, we're taking that piece of our society and saying it's more important to identify geniuses.
• 25:54
  And if we can do that right now with all of our, you know, hypothetical, you know, DEI, we, we, we care about everyone gobbledygook, then we will definitely do it in an AI future.
• 26:04
  We, we will have, you know, there will be orientations for all of these core institutions that say, "Hey, the thing that we're doing," right, just as the thing we were doing a generation ago is, "We're producing the new geniuses for the Manhattan Project." That's kind of, if you sum all of it up, what I, what I think the, uh, the, the educational establishment is chasing.
• 26:22
  They're, they're trying to produce as many geniuses as they can, right?
• 26:26
  Um, but in, in the future, right, if there's a new, there's a new technology that's more valuable than that, more valuable than, than like the STEM physics ge- and, and, you know, p- part of the problem in the modern moment is, is that we produce a bunch of STEM geniuses, and then we undercut them by, uh, bringing in, you know, physics majors and, and PhDs from, from outside the country and hilariously undercutting their wages.
• 26:47
  But enough about Er- uh, you know, the stuff that, uh, that Eric Weinstein has, has, has demonstrated.
• 26:51
  Okay.
• 26:51
  So I believe that your children will be successful if in this AI future, if they are, either are the apex users or know how to serve the apex users well.
• 27:01
  And in both cases, they must learn to work well within the new paradigm.
• 27:05
  And, and I'm gonna emphasize this is gonna be a paradigm that the existing culture will resist, right?
• 27:10
  There will be, there will be a bunch of people that don't like the outcomes, the natural implications of this technology and kinda how it works its way out, right?
• 27:20
  So why do I think this kind of… What, what do I think is a good model for, for understanding, you know, what we're about to do?
• 27:25
  Um, I think that, that, that CAD gives us a great example of, of a bridge technology, and it follows the progression that I outlined earlier.
• 27:32
  So we have a bridge between domains, which creates a new playing field, which initially produces kind of a bunch of egalitarian activity, but then durable hierarchies emerge, okay?
• 27:42
  So when you, when you talk to people about CAD, a lot of people will, will frame CAD is, um, you know, well, well, there was the CAD movement, and then the CAD movement failed.
• 27:51
  Um- And I, I, I don't actually think that the CAD movement failed.
• 27:55
  The… Now, what I think happened is a CAD aristocracy emerged, and so a bunch of left-leaning pu-publications that had formerly been really excited about the CAD movement stopped talking about them.
• 28:07
  But that's not the same thing as, as the technology not, you know, being super dominant and people having a, a great impact.
• 28:13
  It's just the, the people that were formerly publishing all of these pieces about them didn't like… Like, it, like, what was happening in the industry no longer supported their narrative, so they stopped talking about it, okay?
• 28:24
  Um, and, and there were a couple of kinda flagship companies that, that, that went under, and we'll talk about that in a second.
• 28:28
  But again, the, w-when, when you have this progression, initial egalitarian economy, a new playing field, initial egalitarian economy, and then a durable hierarchy that starts to emerge, the people that bet on the, the egalitarian framework just kinda being the way it is from now on lose their shirts, right?
• 28:46
  And, and, you know, I'm wanting you to think about the implications of this for, for the AI ecosystem, right?
• 28:52
  There are people in the AI ecosystem that are, that are really focused on, "Well, we just wanna, you know, sell picks to, to a bunch of gold miners, and we're gonna, we're gonna be rich, uh, you know, by, by, by being the, the pickaxeman."
• 29:05
  And, and that's not a long-term strategy either, right?
• 29:09
  Um, because most of the gold miners go bankrupt, right?
• 29:13
  M- And, and, and that is, you know, again, what we're seeing in, in the AI space.
• 29:17
  The, you know, uh, Uber famously spent, you know, like, billions of dollars on their token budget, and they have effectively nothing to show for it, right?
• 29:26
  So there's a bunch of big companies that i- you know, in, i- created AI work groups and said, "Hey, everybody's gotta join AI." Um, and then they didn't get anything out of it.
• 29:35
  They didn't get anything out of it.
• 29:37
  Um, and so now they're, now they're retrenching.
• 29:39
  Now, there are specific people who are being incredibly productive with AI, but they are the elite.
• 29:46
  They are the best of the best among the programmers.
• 29:48
  They are people who have deep knowledge of, of the, of the industry, deep knowledge of what they're trying to do.
• 29:55
  They have, they have incredibly logical, um, you know, incredibly logical thought processes.
• 30:01
  You know, one of the fun things that I like to do on, on X is go on and watch, you know, Eric, uh, Eric Raymond, I think his name is, uh, just bash other programmers, right?
• 30:09
  'Cause basically you can go on, you know, X on any random day, and somebody will be complaining about how this prompt that they tried doesn't work and, and Er- Eric Raymond will jump in and just be like, you know, "Skill issue," right?
• 30:22
  You know, "You're just not, you're not doing it properly." Um, which I… And, and, I mean, he's one of the great programmers alive, I think.
• 30:28
  Uh, I, I, I wonder if he appreciates how rare his gifts are, um- But okay, so let's, let, r- let's go back to CAD.
• 30:36
  So, um, the, the basically the, the story of the CAD movement, uh, is th- there's a couple of cool stories about it.
• 30:40
  So in 2011, a Steve Punk, steampunk prop maker in Bellingham named Ivan Owen posted a video of a giant articulated metal hand, right?
• 30:50
  So then a guy, uh, named Richard Van As in, in Johannesburg, a South African, he was a carpenter.
• 30:55
  He'd lost four fingers, four fingers to a table saw, and he couldn't afford prosthetics, right?
• 31:00
  You know, you know, South Africa was, uh, uh, as it is now, you know, kind of going through troubles.
• 31:05
  Um, so these, these two guys, they, they start figuring out how they could turn this giant mechanical hand that the prop maker had designed into an actual, you know, effectively prosthetic hand.
• 31:15
  Um, and then there was a, a, a lady who's, who she was sending Liam.
• 31:20
  He was a five-year-old.
• 31:21
  He, he was born without fingers, and the two of them, like, got really into, "Hey, we're gonna, we're gonna build this kid a mechanical hand." And they designed it, put it on a desktop 3D printer, and, and they were able to build it incredibly cheaply.
• 31:34
  Um, you know, famously Owen released the designs free.
• 31:36
  He didn't patent them.
• 31:38
  And, and then that, that created a network which, which is still, you know, still going pretty strong.
• 31:42
  E-NABLE, um, accounts for roughly 40,000 volunteers in over 100 countries, and it's delivered hands to more than 10,000 children, and these were two guys with no medical training, right?
• 31:52
  And they outperformed the entire medical industry.
• 31:54
  Yeah.
• 31:55
  So this is CAD, computer-aided design, um, and it's, it uses, you know, things like the, the desktop three D printer and the cheap scanner.
• 32:03
  I mean, it, it-- but it, it's a lot more than that, right?
• 32:05
  This, this technology links domains through a bunch of different, uh, industries, um, jewelry, dental aligners, aerospace, architecture, surveying.
• 32:16
  You know, and, I mean, it's essentially a file standard, right, with a couple of programs that, that, that, that can work in three D space, right?
• 32:23
  Um, and, and then, you know, as, as this, this thing was growing, they had Make Magazine.
• 32:28
  There was a Maker Faire.
• 32:29
  This is in kinda the mid-two thousands.
• 32:31
  There was a group called, uh, Tech Shop.
• 32:33
  Um, then, then MIT kinda got in, got on board.
• 32:36
  There was a group called Fab Labs, and it, it had, uh, forty-five sites, uh, and then it grew.
• 32:41
  At its peak, I think it had three thousand sites in s- a hundred and sixty-five countries.
• 32:45
  And the idea here was these maker spaces, right?
• 32:47
  Anybody can walk in and learn to use a tool that, that dominated the manufacturing space.
• 32:53
  And, you know, again, in this ear-- these early days, nobody knew which walk-in would matter.
• 32:57
  There was a lot of egalitarian expectations, and expectations were high, right?
• 33:02
  Um, so you had, you know, some incredible success stories, right?
• 33:06
  Uh, uh, Jim McKelvey, uh, St. Louis glassblower, he, he walked into Tech Shop.
• 33:10
  He prototyped a card reader, and that card reader eventually became, you know, the company we know as Square.
• 33:15
  And then, you know, Square's still doing-- St-Square's still here, and it's doing pretty well for themself.
• 33:19
  Um, Brett Pettis, who was a, a Seattle art teacher, a public, public school teacher, he co-founded MakerBot, which, you know, sold for four hundred and three million dollars.
• 33:29
  Um, so, you know, you have a guy that's a glassblower who has an incredible impact on the payments industry.
• 33:35
  You have an art teacher who's able to go into manufacturing, right?
• 33:38
  MakerBot has kind of become a, a thing that has allowed a lot of people to launch manufacturing businesses, small-scale manufacturing.
• 33:46
  Um, nearly every custom hearing aid on earth is now 3D printed.
• 33:50
  You scan the ear, you shape the scan, you print the shell.
• 33:53
  Uh, you know, millions, tens of millions of units are in field use.
• 33:56
  And then dentistry the same, right?
• 33:58
  You have… You scan the tooth, you scan the mouth, uh, you, you, you create a CAD image, then some kind of printer builds the image.
• 34:07
  Um, and, and the interesting thing is that if you are skilled in the basics of CAD, right?
• 34:14
  Or not, not, well, not the basics, I suppose.
• 34:15
  If you are really skilled at CAD, at CAD design, then you can go between these industries in, in… There, there's a lot of mobility between them, right?
• 34:24
  You can go from, you know, jewelry resin, dental resin, engineering resin.
• 34:29
  It's the same interface, it's the same printer, it's the same, frequently the same CAD program.
• 34:35
  And, and maybe you might have skins, you may have, um, you know, specific plugins for, for different applications.
• 34:41
  But the basic uses of how do I create something in this 3D space and then get a machine to make it in, in real life, it's, it's really similar.
• 34:51
  There's a lot of overlap there, right?
• 34:52
  So they're building…
• 34:54
  What you are building, right, is very different, but what these people are doing, right, I, I, uh, talked about earlier the, the interface, right?
• 35:00
  The tool.
• 35:01
  They're sitting in front of computers, and they're interacting with a really similar, um, a really similar Technology, right?
• 35:08
  Uh, the same tech- underlying technology, right?
• 35:10
  So now, as I said earlier, uh, the, the, the egalitarian models went, went bankrupt.
• 35:16
  Uh, TechShop filed Chapter, uh, Seven bankruptcy in 2017.
• 35:21
  Maker Media went insolvent in 2019.
• 35:23
  Um, and then Autodesk, Intel, and Microsoft withdrew their sponsorships from Maker Media.
• 35:28
  Uh, but again, what I, what I want to emphasize is that was these businesses.
• 35:34
  These were the people that were selling the pickaxes.
• 35:36
  Their idea was, "We're gonna have cheap access for everyone." And as it turns out, most of their customers, like the overwhelming majority, 99%, never produced anything, right?
• 35:47
  They never produced anything profitable, okay?
• 35:49
  And that meant eventually people were gonna figure out, "Oh, this is actually really hard.
• 35:54
  This, you know, there, there are some difficulties here." Um, it's not just, and, and again, many ways the, the story here is it's not just as simple as, oh, the rich people have access to the right tools, and so they can do things that, that, that I can't do.
• 36:07
  There, there's actually some skill and some, some, some technique and, and some very important things that, that, uh, maybe most people don't have, okay?
• 36:15
  Um, but access to the, to the technology was not the bottleneck, okay?
• 36:19
  But the hierarchy, and this is the point that I wanna make, the hierarchy that, that, that was established in the CAD world survives, and it's still here today.
• 36:27
  So CAD Mastery is a certified ranked portable credential, right?
• 36:31
  You can have an Autodesk user all the way up to a professional.
• 36:34
  You can have a SolidWorks associate all the way up to expert.
• 36:37
  Um, there's a magazine called Engineers Rule, um, they, they put in 20 se- 2015.
• 36:41
  According to those who do the hiring over at Disney, SolidWorks certification won't necessarily guarantee you a job, but it will land you an interview.
• 36:48
  And that's really interesting, right?
• 36:50
  Um, a certification, right?
• 36:52
  This is not a college degree.
• 36:55
  Um, it's, it's, it's a, a relatively short program, um, actually.
• 37:00
  But you have to be really good at this one thing.
• 37:05
  And so, and I, I would, I would frame CAD if, uh, if, if we're talking about, you know, technologies here, just as the longbow was the last kind of…
• 37:13
  It, it was, it was m- it was disruptive, but it was more like an aristocratic technology than the firearm.
• 37:18
  Um, I would say CAD is kind of like the last democratic technology, but it's trending towards an aristocratic hierarchy, as we see.
• 37:25
  So, you know, rank earned inside the tool's hierarchy is honored inside every industry, um, the, the hier- every industry hierarchy that the tool touches.
• 37:33
  Um, so I had a friend who, who took a job with a surveyor precisely because that got him a bunch of CAD training, and then once he was able, you know, once he had had a certain number of years of experience with a CAD program, then he was, he was able to, to quickly and easily get a certified, uh, certification in some- something, and he could move anywhere.
• 37:53
  He called it CAD space.
• 37:54
  Anywhere in CAD space.
• 37:55
  I've never forgotten that.
• 37:56
  That was really interesting.
• 37:57
  Um, and so you see this, right?
• 37:59
  So there are, there are four main vendors, Autodesk, Dassault, PTC, and Siemens, and they hold roughly two-thirds of the CAD market.
• 38:06
  These are our dukes of CAD, if you'll excuse the expression.
• 38:09
  But I would argue that their, their power, their, their moat, their c- the market capture is not in providing a software.
• 38:15
  They, they provide software, but I would say it's assisting peak users in providing a moat for their skill.
• 38:20
  It's the certification process.
• 38:22
  It's the intellectual property.
• 38:24
  And, and actually, I think when you, when you review the financials of those companies, the ones that are public, you, you, you see that they, they are, they are claiming a great deal of value in their, in their certification process, which is really interesting.
• 38:37
  The, the, again, to, to, to extend the, uh, the, the medieval metaphor, they, they issue patents of nobility for the skilled users who hold the real bottleneck in their brains and fingers.
• 38:46
  Yeah?
• 38:47
  So, you know, MakerBot bet $400 million on monetizing ma- mass access.
• 38:51
  The bet failed, and Stratasys pivoted industrial com, uh, cons- uh, customers.
• 38:56
  In 2024, you know, these days, Stratasys has sued Bamboo Lab over patents.
• 39:00
  Last year, Bamboo locked its firmware against the open source tool that built the field.
• 39:04
  So the, the, the, the open source egalitarian framework of CAD as a whole is clamping down.
• 39:12
  Um, those who bet on making money giving access to everyone lost their shirts.
• 39:16
  Those who bet on rank and on, you know, identifying and protecting hierarchies of skill, they're doing just fine, okay?
• 39:22
  So again, this is what I see happening, um, in this industry, and, and there's others.
• 39:27
  You, you talk about Excel, you could talk… There's, there's a bunch of different, uh, f- different times that this has happened.
• 39:34
  Um, so when a tool crosses what's, uh, a historical domain threshold, it bridges between two historically disparate domains, you have a couple things that happen.
• 39:43
  Again, the floor drops first.
• 39:44
  Newcomers in any bridged domain can collaborate through the shared interface.
• 39:47
  Skills which can be acquired cheaply in one industry become disproportionately valuable in another.
• 39:52
  But also, the ceiling rises.
• 39:54
  You know, masters become comparable across domains.
• 39:57
  And when you can compare between users of the same underlying tools, then you have a hierarchy where comparison is possible, uh, ranking follows.
• 40:06
  You know, one of the interesting things as I was doing the, uh, the research for this episode is, you know, a shared tool is, is a shared language, and a shared language permits honest evaluation, and honest evaluation permits rank.
• 40:19
  Um, you know, there's this fascinating thing, you know, when, when Latin carried European letters, Erasmus in Rotterdam and More in London could weigh each other across the, the channel.
• 40:27
  But when Latin ceased to be the working language of elite Europe, you know, the, the next major wars between the European states, the French Revolutionary and th- then the Napoleonic Wars, which are in some sense an extension of the French Revolutionary Wars, or you could say that the French Revolutionary Wars were the precursors to the Napoleonic Wars, and they were shocking in their scope and disruption.
• 40:46
  Um, their precursor, the French and Indian War, uh, was, was… saw the rise of George Washington.
• 40:52
  One of the things people noted about George Washington was, uh, that he didn't speak Latin.
• 40:55
  He wasn't Latin-educated, and he was important in Western politics.
• 40:59
  That, you know, and, and people were like, "How, how can this happen?
• 41:02
  How can it be that this happens," right?
• 41:05
  And, um, you know, that, that… it's, it's really interesting.
• 41:08
  Uh, Latin remained, you know, the, the… remained important to the European elite.
• 41:12
  It became kind of like a gatekeeping function.
• 41:14
  You had to learn to speak Latin.
• 41:15
  But it was, instead of being a functional language, it was increasingly a prestige symbol, studied in specialized Latin classes, but not used as a language of instruction beyond that.
• 41:24
  Um, and then within a generation of scaling back Latin education for their elites in the late 1800s, the European nations engaged in, in the world wars, right?
• 41:31
  The brutal kind of… You could frame the world wars in many ways as the, the European civil wars or the Western civil wars that created the modern international order, right?
• 41:40
  So CAD's capacity to bridge industries was essentially one file type deep, and it was across an important but limited part of the manufacturing sector.
• 41:50
  If we're, i- i- if, you know, the, the CAD space is, you know, maybe two, three dozen industries.
• 41:56
  And, and they're important industries, you know, all these kind of manufacturing and, and, uh, prop making and, and, and trialing, you know, prototyping, all, like, almost anything with plastics is modeled with, with CAD.
• 42:07
  The, the CAD is very important, but the CAD space is not the whole economy.
• 42:12
  W- the question is, what is the size of the AI space?
• 42:17
  Okay?
• 42:18
  Um, if, if we're going to, if we're going to say, "Hey, what, what, what is the AI space?" What industry is not part of that, right?
• 42:28
  Um, every industry is being impacted, segmented, reconnected by a tool that translates natural language into code precise implementation, right?
• 42:36
  I mean, we talk about wrappers, right?
• 42:38
  That, that we're, we're able to take a, um, natural language and have it translated into machine accurate code.
• 42:48
  Uh, there's this old video game that, that, uh, that a friend of mine sent recently, uh, you know, th- these, these, these… It's a text-based video game, and you have to get the text precisely accurate, right?
• 43:00
  Absolutely perfectly accurate.
• 43:02
  Um, and he, you know, he recommended it as a thing that I could have my, my, my children learn to play.
• 43:06
  And, and what, what struck me about it was- It's like interacting with a terminal.
• 43:12
  This, this old video game.
• 43:13
  It's like interacting with a terminal, and I always hated interacting with terminals because if you have one letter, one dot, one dash wrong, then it just doesn't work, and sometimes it tells you clearly it doesn't work, but it, many times it just fails and it won't tell you why, and it's very difficult, at least for me, to, to interact with.
• 43:30
  Whereas, you know, a, a, a terminal tool, a CLI operating in a terminal, I, I mean, even of a, of a relatively dumb AI, you know, something that you, you download, uh, you know, the open source weights, it makes my terminal wildly more powerful, wildly more productive.
• 43:46
  I, I'm able to, to get so much more out of the machines that I have, um, and that's really interesting, right?
• 43:53
  And so then that's kind of what drove, um, you know, all, all of the adoption, right?
• 43:59
  You have a bunch of people pouring in mass adoption, and everyone's like, "Okay, we're gonna, we're gonna, we're gonna sell AI to everyone." Right?
• 44:07
  "We're gonna sell AI to everyone 'cause everyone's now connected.
• 44:10
  Everyone with a computer can use AI.
• 44:12
  Everyone with a computer can benefit from AI." And, and to some extent that's true, right?
• 44:16
  Like, I think, I think, I think there really is going to be a, a incredible boon to the, the, the bottom rungs, um, and even the middle rungs.
• 44:24
  I think, I think everyone will, will benefit.
• 44:26
  But like with capitalism, right, you know, one of the, one of the great illustrations of, of, of capitalist social theory is, you know, i-i- in order to be a capitalist, you have to be okay with if you push a button, you're gonna get twice as much and somebody else is gonna get 10 times as much, and somebody else farther up is gonna get 100 times as much every time you push the button.
• 44:45
  Um, but you still objectively benefit, right?
• 44:47
  You still have twice as much.
• 44:49
  Um, and I think that's, that, that sort of hierarchy is, is what's happening in, in AI.
• 44:54
  And again, what is the size of the AI, AI space?
• 44:57
  What industries are truly outside of the AI space?
• 45:00
  Um- You know, I'll probably do an episode on, on that as a whole.
• 45:05
  Um, but the point being, AI is much, much bigger than CAD, okay?
• 45:11
  So, you know, the floor is dropping on schedule in randomized trials.
• 45:15
  AI closes three-quarters of the education-based productivity gap in, in task ed- uh, execution, but the ceiling is rising.
• 45:22
  The wage premium for AI skills hit 62% this year, up from se- 57% the year before.
• 45:27
  So I believe that execution will continue to flatten and command these top-level human skills will continue to stratify, okay?
• 45:35
  And, you know, what I, what I wanna say is I think that the, the rapper startups that sold access to the floor are already, maybe not dying, but, but they're maybe not already dead, but I think they're dying.
• 45:47
  I think that, that they are recognizing… You know, I was at a, at a dinner, uh, about a month ago, and, and the guy that was presenting on, on the financial health of our country basically said, um, if everyone on Earth bought a $200 ChatGPT subscription, it would not recompense them for the, the, the capital investment they're making right now with these hyperscales.
• 46:07
  So, so some… There has to be some other thing that they're planning to do, hoping to do, right?
• 46:13
  Famously, uh, you know, since the beginning of the Frontier Labs has sold intelligence below cost, so tokens are subsidized.
• 46:19
  You know, subsidized, like, literally above what it costs or below, excuse me, below what it costs to run the machines, to power the machines and to keep them going, right?
• 46:28
  Like there's, there's, there's indications that you're, you're not even paying for… Like you're not even paying for the electricity, let alone the, the computer units and all of that themselves, right?
• 46:38
  So what, what I wanna say is that that means that access to the models is not the scarce piece, right?
• 46:44
  Remember, in the CAD world, Autodesk withdrew its spon-spon-sponsorship of Maker Faire.
• 46:50
  It withdrew the subsidized access to these tools.
• 46:53
  And, and I promise you the token subsidies will end the same way, right?
• 46:57
  So- Uh, that's, that's, that's basically all for today.
• 47:01
  Um, next time we'll get into some, you know, uh, w- we'll start talking about what does the peak AI user look like?
• 47:09
  And the, the image that I want you to have in your head as we begin that discussion is that there's a guy in a tree line in Eastern Ukraine, a, a man crouching over a controller.
• 47:18
  He's got a battery, he's got a laptop running a model at the edge of the network i- on a, on an airframe, you know, computer that's cheaper than a used phone, and two kilometers out, a column of, of, uh, heavy cavalry of tanks worth a thousand times what, what this guy, you know, has in his kit, slows at a crossroads, um, one turns…
• 47:35
  A- and, and, and basically he flies a drone into a tank, and then the whole battalion has to turn away.
• 47:42
  Okay?
• 47:42
  So that… Actually, that guy is not what I wanna talk about, but I wanna talk about how do you defend against him?
• 47:48
  How do you defend against that, that operator?
• 47:50
  And basically, you know, w- we'll get into it more, but I believe that the working defense against a drone or a swarm of drones as, as these things scale up, is another swarm that has to be piloted by AI, right?
• 48:01
  Because the, the, the speed of these drones means that you have, like, a half a second be- w- between the time this person enters your true easy detection range and when they impact.
• 48:13
  So that means you have to have drones out there in the, in the air floating, um, and we'll get into, you know, there's, there's some really interesting news stories that this suggests this might already be happening.
• 48:23
  Um, have people having bubbles of drones out there that are already, already operating, and they're able to intercept drones, they're able to intercept missiles, they're able potentially to intercept even, you know, jets.
• 48:32
  And the people who can program, direct, and command those AI robotics fusions, I believe, are the paladins of the next century.
• 48:39
  So there are three core skills that AI is not going to erase, and we're gonna get into that probably two weeks from now.
• 48:45
  Um, code, which is understanding the, the core logic processes that AI-generated code must conform to to achieve your goals.
• 48:52
  Content, that's understanding the emotional and rhetorical processes that persuade people.
• 48:56
  And command, knowing when to make a decision and bearing the moral weight of it, right?
• 49:00
  When, when, when you have a decision that has moral weight, uh, where, where getting it wrong is a bad idea, it is incredibly expensive to have an AI make a decision that, that cannot be wrong, right?
• 49:13
  Um, and again, the, the, the, the ending, right?
• 49:16
  What about your kids?
• 49:17
  What I would say is if those three things are the things that are gonna endure, and, and there may be more than that, but, uh, I've got a, I've got a, a- I am actually attempting… My… Certain, certain people that, uh, that, that advise me and help me with content will, will be shocked by this, but I am actually trying to simplify things for, uh, for my audience.
• 49:35
  We actually have a lot of experience in training children in, in logic, right?
• 49:38
  Which is the, the… Basically, what it, what it looks like to me is the people that are still good at code, they're actually good at the underlying skill of logic.
• 49:45
  They, they can hold logic frames, you know, bigger discussions in their heads, and they…
• 49:50
  That's, that's what allows them to instruct the AI efficiently in, "Hey, do this, and then this, and then this, and this." This is the flow of the logic.
• 49:57
  That's their… That's what's really going on there.
• 50:00
  So we're, we're… We know how to train kids in logic, we know how to train them in rhetoric, and we definitely know how to train them in what our Founding Fathers called the habit of command, right?
• 50:09
  And so, uh, I look forward in, uh, in coming days to exploring that with you.
• 50:14
  Uh, so thank you, everybody.
• 50:15
  That concludes the, uh, the free part of this session.