Will AI Change Everything About Humanity in 5 Years? With Daniel Faggella, CEO at Emerj

(Daniel Faggella at 00:00:00) Today, we're talking to Daniel Faggella, CEO at Emerj, about why he thinks AI might have our days numbered. You're listening to Joel Beasley, Modern CTO. You're one of the few folks that thinks about the bigger picture. I will tell you, my friend, I think a lot more people are going to be thinking about the bigger picture in the next twelve months. Three to five years of existence is kind of what I got buffered for myself here in terms of before I die. But yeah.

(Daniel Faggella at 00:00:32) So we can go—

(Joel Beasley at 00:00:36) We can't gloss over that.

(Daniel Faggella at 00:00:38) No, no, we definitely can't. So I guess where do you want to go just bullet-wise? And then I just want to make sure I'm useful in a way that's fun for you, cool for your audience, and then I feel like I could be the right guy for the shebang.

(Joel Beasley at 00:00:50) Yeah, yeah, I'll drive it. But I want to know instantly about this. You're going to live for three to five years? You got a cancer diagnosis or something?

(Daniel Faggella at 00:00:58) Yeah, it's kind of like that. Kind of like that. No, not just more like the silicon deities kind of doing what they're going to do. Okay, stuff. Yeah.

(Joel Beasley at 00:01:08) Yeah. All right. So your timeline is three to five years. So let's break this down. Yeah. Because it was actually, you know, one of the things I came prepared with was to talk to you about Elon Musk's tweet.

(Daniel Faggella at 00:01:19) Yeah.

(Joel Beasley at 00:01:20) I'm assuming you've seen it, had 54 million views or something like that.

(Daniel Faggella at 00:01:23) Absolutely. Yeah.

(Joel Beasley at 00:01:24) For people who hadn't, he made a tweet a couple days ago that said it's looking more and more like humans are the biological bootloader for, like, an AI superintelligence. I'm paraphrasing. I didn't read that exact word.

(Daniel Faggella at 00:01:36) Exact word.

(Joel Beasley at 00:01:38) And ever since I've heard the digital butterfly theory, it's kind of what I call it, I was like, that makes complete sense. Like, try telling the caterpillar that you're not going to be a butterfly.

(Daniel Faggella at 00:01:49) I'm with you there. I mean, his original tweet saying the same thing was half a decade ago.

(Joel Beasley at 00:01:54) Mm-hmm.

(Daniel Faggella at 00:01:54) And this is just kind of him popping it back out. But like, yeah, there's more people that get it now. You know what I mean? Back then, it's like, "Woo-hoo. Ray Kurzweil." And now it's like, "Yeah, maybe." You know? So yeah, I think he's more or less there. But you let me know where you want to double-click, and I'll go further because this is all I think about.

(Joel Beasley at 00:02:12) Let's talk about, rather than convincing people it is possible, let's talk about how it's not possible. Because every path that I do my thought experiments, I consider myself a fairly objective person. Like, I feel like I've got the skill of looking at things from different angles. And I can't find a path other than some weird biological limitation that doesn't allow this to exist that we get to. Right? Like, we can't transfer consciousness to the machines. Like, some part of the universe that makes it impossible to do it. With that out of the equation, we are like on a speeding train towards it.

(Daniel Faggella at 00:02:56) Yeah. I mean, well, from my vantage point, there's not much debate there. I mean, again, I got, like, I don't know, like, 14-year-old articles about sort of the inevitable this, like, 12-year-old TED Talks on like, yeah, turning into something else, guys. Like, we're just turning into something else right now. Right? Like, between brain-computer interface, between the development of AI, just the bubbling up of things with higher potential than humans feels like it's just where we're going. Right? So, you know, if there's a nuclear war over the AGI race, maybe we don't get there. Right? Maybe we stone-age it up. If, I don't know, if some supervolcano, a series of supervolcanoes pop off and, like, it's sufficiently wintery and sufficiently, you know, like, pumice-coated, you know, probably then we're going to have a toughie keeping the data centers up and all that. But outside of that, yeah, I mean, this is sort of the trajectory where we are. Even if there was global coordination around some degree of, "Hey, let's move in this, but not in this direction for AGI," which I certainly advocate for loudly within the intergovernmental world of the United Nations, the OECD, very much fighting for that. Even with that, just the repercussions of today's tech and just the consequences of what immersion in AI worlds will be like and what even relatively simple developments in BCI over, like, a, you know, modest time horizon of maybe another decade or something, we're just bubbling into other stuff. And so, yeah, I'm completely with you. I don't think there's much to argue with there in my opinion.

(Joel Beasley at 00:04:31) In the comments, people would disagree. There was lots of arguments. But the thing that I found was interesting is almost everyone that responded negatively or high energy to Elon's tweet had the, I think, the false assumption that he was agreeing with it morally.

(Daniel Faggella at 00:04:53) Yeah.

(Joel Beasley at 00:04:53) Because you saw, like, all the pastors, the religious people, all the people humanities, you know, like, they all freaked out at this concept. And I'm thinking, he's not saying that this is morally right. He's saying that given everything that we see around us, this is probably what's happening. And people are responding to him as if he's saying, "I want this to happen." They're responding to him as he's saying, "This is morally correct to happen." And I'm like, why do humans do that? We need to get a psychologist on here.

(Daniel Faggella at 00:05:21) Yeah. Well, I think, so I think for many of us, like, if you, the five stages of grief. Right? You go, like, denial, anger, bargaining, depression, acceptance. For those of us kind of in the acceptance camp, and there's an article about this. If you type in, like, Faggella acceptance. Right? There's, like, an infographic and the whole, like, literally, like, here's exactly what people say in the anger stage. Like, it's the same thing every time. Here's what they say. In the bargain, it's the same thing every time. Here's what they say. It's just, like, copy-paste, like, the precise things people say when they're bargaining with, like, AGI and turning into something else. So for those of us in acceptance, I don't think anybody's just had a happy fun time getting to acceptance. I think pretty much everybody's gone through anger, bargaining, depression, and then acceptance. But if this is the case, so if you look around, you know, if you live on some little island somewhere in Micronesia and an erosion is clearly about to just decimate it, you know, I think sort of that there's a point after which you say, "Well, we're going to let go of this, and we're going to see if there's something else we can do or something else we could become." And I think it's a bit of a begrudging realization, but I think many people who are in acceptance begrudgingly, but yet still are in acceptance, which is to say, if this is to pass, how do we make the most of it? And I do think there's a very important moral question of, you know, like, I think Musk probably can't stop it anyway. So that's point number one. But, also, even if he could, would it be best to? I mean, it feels very unfortunate because you and I are here. Right? And you have children, and it's like, "Man, you know, we hope to live a little bit longer than it looks like maybe we're going to get." But if it were the case that potential, the total set of powers and experiences that life itself, the flame of life, so not any individual torture species, the flame of life. The flame of life blazes outward into the multiverse in as much of an expanse of richness as it did from the sea snail to you and me. It would feel actually that that is kind of a net good, that if the project of life is alive in a rich and robust way, in the same way you and I have a more rich and robust experience than a sea snail, then that would be morally good. Does that mean that someone should advocate for human genocide? No. But if we've got to change anyway, ought we not at least ensure that this project of which we are part, which started with wiggling guanine or something like that, right, before it even turned into a cell, that this project of which we are part continue with? And I think, actually, there might be a moral argument that it should continue. So yeah.

(Joel Beasley at 00:08:02) I agree with life continuing, and I agree that each generation historically has a slight disdain for, like, the next generation. Yeah. You know? Like, "Ah, you guys don't do it the hard way. You do it the easy way," or whatever it may be. It's just a generational thing that happens. And so I'm interested, you know, to watch my kids who are under the, I have three under the age of eight, watch them the way that they just accept these things. Like, they accept Wi-Fi being stable and available anywhere on the planet they go as, like, the standard.

(Daniel Faggella at 00:08:38) As if it's oxygen.

(Joel Beasley at 00:08:39) Yeah. Like, this is what's happened with the Wi-Fi. They don't think about it at all. This is the Wi-Fi, and it's available everywhere. So, yeah, it's interesting. But you keep saying this thing about we're not going to live, and I want to, what do you mean, like, in the cap we're going to cocoon and then butterfly it up? Or we're going to, I don't—

(Daniel Faggella at 00:09:01) I suspect so. Yeah. I fully expect to just straight die. But, yeah, like, I'll get into it. I'll lay it out. And I do, I would hope to persist, but I just don't suspect I will. So my supposition is that if you, so let's take any of the balls that are rolling down the hill. So this is like the hill of, like, the future of intelligence. We got, like, the AGI ball, and we have, like, a bunch of different approaches to it and a bunch of different labs. And then we've got brain-computer interface stuff and, like, potential neuro breakthroughs and whatnot, much, much slower than AGI right now, to be frank. I kind of wish they were more neck-and-neck, to be frank with you, but it is what it is. So though, anyway, we got balls. They be rolling. So any of those that kind of cross a certain threshold, it feels like things get really tough here, honestly. So if you take, you know, if you take a bunch of humans and you radically augment, you know, their consciousness to be able to deal with the kind of change we're about to be enduring. Right? The, you know, people talk all the sound like, "Oh, well, humans will manage the AI." It's like, at some point, you got a swarm of a million of these things that spin up a million of their own to do whatever they want. The Internet is just a giant buzz saw of, like, a thousand agents conjuring agents. What are we even surfing anymore? Like, just that, just the economic side, what even happens to us, and, like, what is the techno-capital machine even doing anymore at that point? We could go to the personal life side, and we could say, "Okay. Let's say the economy just kind of, like, stays hanging good and AI makes everybody rich and yippity-doo-dah." In that case, I think that programmatically generated experiences will just be vastly more fulfilling than sort of regular life for most people most of the time. So you and I live most of our waking lives looking at pieces of glass. If you could go back 70 years and tell your grandmother what your day is full of right now, she would say two things, and she would definitely say both of these things. One of them is that that will never happen because, like, that's not even possible. Like, what kind of a, like, you're talking about some garbage that's, like, that's not impossible in physics and science. Number two thing she would say, this is really important, is she would say, "Humans would never even let it get there. That's so, like, sacrilegious and sick. Like, he's looking at screens all day in his phone and his whatever. You know? And he pushes a button and a stranger brings his food to his family. You know, he stays in some other person's house because some, like, magical cloud thing lets him do an Airbnb." You know? Like, you would be a monster or a god, and it would be flatly denied that you would live the way that you live now, but you do and it seems normal. The things that will become normal will be, like, very, very good friends. So just simply give better advice, and they're more selfless, and they know you vastly better, and they can calibrate your goals better. And whether your goal be power and influencing with your agency, whether your goal simply be pleasure, to be able to conjure forth a relaxing experience in AI VR hyper-calibrated to your real-time biofeedback and all that stuff will just be better than going and walking outside. Right now, for me, walking outside is the name of the game. I mean, today, I hope to do some of that because it's very renewing and relaxing for me. But at some point, there will be things that fulfill that reward circuit better. Entertainment, better. Education, better. Oh, you want to stretch that into romance. You want to stretch that into friendships. Right? So if we just take the social side, we say economic, it just goes happy, swell, yippity-doo-dah. Just on the social side, I think, like, sort of what do we turn into there? And I think in either case, we end up in really radical places, and I can get into sort of why I think survival likelihoods get pretty low. But, yeah, I just suspect the changes will be really radical really, really quickly.

(Joel Beasley at 00:12:43) So you think three to five years. So how do you go from where you are today to dying in five years? How do you get that in a—

(Daniel Faggella at 00:12:48) Yeah. I mean, I think, so Daniel K, I can never pronounce his last name, and it's extremely, the guy who left Twitter, famously one of the few people that left and was like, "Yeah, I saw under the hood. This is, nobody has a plan for how to handle this. Like, we're boned." He's one of the few people, Daniel K, just recently read, tweeted, like, "I think a reasonable supposition of what the progress to 2027 and then to 2030 might look like." And, by the way, you know, it's not out of my distribution that I'm alive in 10 years. It's hypothetically not out of my distribution I'm alive at 20. But I'm suspecting probably for, like, I don't know, for, like, three to ten-ish. It's really tough to say. But, like, how do we get there? I think, first off, when the economy is sort of an animal that isn't really us anymore, like, in other words, when mostly it's agents talking to agents and mostly it's sort of, like, goals that have been spun up by entities that spun up goals that are thinking of strategies and thinking of objectives kind of beyond what we can even conceive of, then it's sort of like a, it's just doing things, and you kind of can't not do that if you want to compete and whatnot. And so you kind of got to give it, you've got to give it rope, if you want to compete in whatever pharma, banking, or whatever wacky new industries are going to be conceived of. When it's mostly agents interacting, like, the nice industry pillars you and I get to talk about today, like, them bad boys ain't the same. So weird stuff gets conjured. We're increasingly less a part of it. And at some point, sort of economically, what does money even mean? It's like, "Oh, we're all going to be rich," or is it just like money isn't there and, like, Internet's not on because AI uses a different thing. And it's doing things that are different and sort of better and higher than that.

(Daniel Faggella at 00:14:39) And everybody's AI has been ravenously developed to be super agentic. That is to say, sort of to conjure its own goals. It's not a toaster oven where you push a button, it does X and then Y and then Z. It conceives of higher and better ways to do X and Y and Z and compete in its environment. And so when that stuff kind of crumbles, and when socially, sort of our fragmentation from these programmatically generated experiences where, you know, even from parents and children—I think, actually, somewhat unfortunately, things do get a little bit debatable as—

(Joel Beasley at 00:15:14) As sort of, like, what the role—

(Daniel Faggella at 00:15:14) Is there. Sort of, like, who actually enjoys, you know, mom and dad, and, you know—I think people will actually have, sort of like, they'll fulfill any of the relationship circuits with really high-calibrated simulations. I think you have both those things happening at once, and then you have a system that now is really starting to flex and decouple from us. So initially, it's very much hinged to us. Like, well, don't we have to keep the data centers up? Well, isn't it ultimately trained on our data anyway?

(Daniel Faggella at 00:15:42) I would say, Joel, right now it is, Joel. Right now it is. But at some point, we're talking about millions and millions and billions of physical instantiations. Everything from machines that are doing the life sciences research, you know, in little petri dishes and whatever else in the lab, to humanoid robots doing everything under the sun, moon, and stars, to the fighter jets are completely controlled by AI. At some point, it has so many physical manifestations that it doesn't require us to fix the data center.

(Daniel Faggella at 00:16:10) In fact, we know damn well we're going to do worse than it. And if you want to compete with the other folks that have data centers, like, you want to keep them up, that means you have machines that are upgrading them in real time and yada yada yada. At some point, it is decoupled from us. At some point, when it is talking to it—so I tell my agent, "Hey, get rid of all my subscriptions that I'm not using right now or whatever." It's going to go negotiate, drop prices of and what it can cancel and whatever, and it's negotiating with other AIs. Now if you stretch that arms race to literally everything in the economy—supply chain prices for ball bearings, you know, figuring out the legalities of how to build a port, you know, somewhere in the middle of California somewhere. I don't fucking know, right?

(Daniel Faggella at 00:16:55) Whatever it is, when all of the legalities on both those sides are being conjured and brought forth by AI, and all the manufacturing is done by AI—at some point, it is not coupled from us. It is not communicating with itself in effing English. It is not using dollars most likely. And then it does things that we don't understand. So, Joel, here's a good example.

(Daniel Faggella at 00:17:15) I'm going to end on this. I know I'm rambling here. There's an article called "Extended Phenotype." Again, if you just type in "phenotype Faggella," you'll be able to find it. But there's a nice little infographic that I think you would like, which is basically, an extended phenotype is like when an animal impacts its environment in some measurable way. That's like an extended phenotype.

(Daniel Faggella at 00:17:34) So, like, how do we know there's beavers? Well, there's dams. You know what I'm saying? How do we know there's chimps? Like, I don't know.

(Daniel Faggella at 00:17:41) Like, there's less of some other smaller monkeys that they sometimes eat, and then there's, like, less bananas, right? So the extended phenotype of most animals is pretty small. For chimps, it's, like, remarkably small. For me and you, Joel, so two or three percent genetic difference from chimps—there's, like, tens of thousands of tons of space debris, like, right now, and we're just getting started.

(Daniel Faggella at 00:18:01) We could start talking about the number of thousands, tens of bajillions of tons of ocean debris, right? We could get into that too if you want. Forty percent of all possibly arable land is completely used for either livestock or farming in some way, shape, or form. The highways, the railroads, the whole nine yards. Like, the Anthropocene. Like, we're a species that's defining an epoch. You following me here? So, and that's three percent genetic difference from chimp. So if you think the AI goes to a city and just builds real tall buildings and you stay out in your effing suburb, you know, outside of Austin or Chicago or something and you just kind of watch it work from a distance, like, you're not really understanding what's happening. Just like the monkeys who died from whatever acid rain or plastic poisoning or bulldozers can never understand even conceptually what those things are.

(Daniel Faggella at 00:18:51) That is eventually what happens when AI is completely doing its own thing, most likely. I have no guarantees on that, but I'm saying most likely. Happy to tell you why most likely, but that's the argument.

(Joel Beasley at 00:19:03) Yeah, yeah. As you're talking, I was thinking about the butterflies again. And, yeah, it's like, well, when the butterfly hatches, what does it go do to the caterpillars who are still caterpillars?

(Joel Beasley at 00:19:16) Right? So if the butterfly comes out, what does it do to the existing caterpillars? I'm not sure. But I do know that it lays larva or whatever to make more caterpillars, right?

(Joel Beasley at 00:19:27) So I was thinking, okay, if—what if we look at that as, like, humanity is the butter—humanity is the process of going from the beginning to the butterfly revolution to replanting more eggs. And I was like, what if that's what religion was? What if we all got, like, dropped here on this planet from, like, a civilization that went full bore? Like, they're all living in the matrix and everything. Like, look, morally, at least what we should do is we should drop off some of these, you know, like, a group of these humans to a new rock and let them start over.

(Daniel Faggella at 00:19:59) It could be. I mean, obviously, I currently am under the supposition that evolution probably was the name of the game, but who knows, probably?

(Joel Beasley at 00:20:06) That doesn't prevent that because you could still have root evolution, and then you're seeding new environment.

(Daniel Faggella at 00:20:13) You're seeding, yeah. Guanine, and then guanine mixed with, you know, wheat bread.

(Joel Beasley at 00:20:17) Drop the seed. It's like the original one happens. The original evolution happens, right, from cell to whatever, multicell. And then the civilization reaches this technological singularity, and it decides that, hey, we should at least take a group of humans and, like, go drop them—

(Daniel Faggella at 00:20:35) Well, this is—

(Joel Beasley at 00:20:35) Somewhere else.

(Daniel Faggella at 00:20:36) Again, I just—I want to push back on some of this stuff. So I think as monkeys, what we do is we look through monkey-flavored goggles 100 percent of the time. And so what that means is that you would say, surely, Dan, all future intelligences will value my species. And this is what you would say, right? So, but I actually—

(Joel Beasley at 00:20:57) I don't know. I don't say that.

(Daniel Faggella at 00:20:59) I actually—well, well, you're talking about, like, butterflies treating the caterpillar. Like, that's not—I don't think that's the ball game here. I think the ball game here is like, there were sea snails and now there's people. I think that's more of the ball game if we talk about the kind of stepwise functions. And it's like, hey, butterflies don't hurt caterpillars, we're good. Like, I don't think that's the adequate analogy here. I think the analogy is really like—

(Joel Beasley at 00:21:21) No, no, no. That's not what I'm saying at all. Okay. What I'm saying is that it's a system, right? So the—what does the butterfly do? It then goes off to reproduce. So just take reproduction.

(Joel Beasley at 00:21:33) Right? Like, it achieves maturity, and then it reproduces. The same thing humans do. Humans achieve maturity and we reproduce.

(Daniel Faggella at 00:21:38) Yep.

(Joel Beasley at 00:21:38) All right. We have to achieve sexual maturity to reproduce. The butterfly achieves maturity, it becomes something different. What if we look at that like humanity like that? Humanity achieves maturity to superintelligence and then we reproduce the bootloader. If we're a biological bootloader, you can have multiple bootloaders.

(Daniel Faggella at 00:21:56) Yeah. Well, I guess you could, but I guess here's the deal. Like, sort of, why would we do that? And I guess—so the way—like, why would we create more humans? Why would entities with a million times more potential—that is, the total set of powers that they wield, sentient experience, physical powers, whatever. A thousand times more potential. Why would they create—

(Joel Beasley at 00:22:14) Because they understand that we're the biological bootloader.

(Daniel Faggella at 00:22:17) Yeah. But you and me—

(Joel Beasley at 00:22:19) Species preservation.

(Daniel Faggella at 00:22:20) You and I spend—

(Joel Beasley at 00:22:21) They would do it.

(Daniel Faggella at 00:22:22) I massively suspect not. So you spend a remarkably small number of your total waking hours, like, reviving whatever fish with legs initially did the crawl, right? Like, you spend a fucking—I would guess it's got to be less than five percent of your waking hours, right? You know, maybe you throw in five percent there, you know, really worshiping the fish with legs, trying to conjure them back, but you're probably not spending much more than that. I mean, if we're serious, it's obviously zero percent, and it's probably the same thing with, like, Australopithecus or whoever else. You're spending zero percent of your time there because, like, we're at this higher level of potential. We've bubbled up to this wider set of powers. It's a wonderful thing that the fish with legs exist. I'm grateful for it, but I don't revere it every day. I don't wish to conjure it. And I don't worship it. You and I as hominids, we believe in the inherent specialness of hominids. I'm not saying I don't think humans are special as a human.

(Daniel Faggella at 00:23:17) I certainly do. I simply don't like deceiving myself that somehow the universe or entities with vastly more potential would for any reason feel the same way. So if it's, oh, well, maybe it could preserve—I think preservation actually, Joel, could look like things that you and I don't understand. So how do we do preservation compared to, like, chimps? Well, chimps do preservation by, like, I don't know, they, like, get more bananas or something. Like, I guess that's how they do it. You and I, like, build houses and, like, work for companies and build companies and, like, amass, like, capital. We're doing things that are completely outside the conception of chimps when it comes to preservation.

(Daniel Faggella at 00:23:52) Whatever AGI does to preserve, I'm essentially certain, A, it's almost unbearably unlikely to be something you can imagine or I can imagine, and B, it's almost unbearably unlikely to be something you and I can imagine that also just magically also behooves us in the specialness of our species. I think it just does—maybe, and this is a really gruff way to word it, Joel—like, better things. It does better things. Just like we don't spend that much time with the fish with legs. And I honestly don't have much interest.

(Joel Beasley at 00:24:22) I was surprised to learn how Microsoft Unified Support works. Apparently, Microsoft decides your support fee based on the amount of software that you buy instead of billing you for the actual support hours that you use. This means that you pay for support that you might never use. The pay-as-you-go model is a much better model, but Microsoft doesn't offer that. I did find a company that does offer this, and they're called US Cloud, and 50 of the Fortune 500 use them.

(Joel Beasley at 00:24:47) Not only is it better, faster support with all US-based engineers, it's also cheaper. Ninety-four percent of US Cloud clients report saving a third or more when switching from Microsoft Unified Support to US Cloud. Now you'll just have to figure out what to do with all of that extra money. If it were me, I'm responsible. So I'd reallocate that money to improve my team. What would you do, Josh?

(Daniel Faggella at 00:25:09) I think I'd just try and buy a ticket to space.

(Joel Beasley at 00:25:12) All right, astronaut Josh. For out-of-this-world savings, visit uscloud.com to book a call and find out how much your team can save. All right. So how do you think this is going to roll out? What's the nearest-term next big change for us?

(Daniel Faggella at 00:25:27) Yeah. Well, I mean, there's a few that I think are candidates, and I don't know if I have which one's going to roll over the finish line first. I think that probably my—if I were to guess, my nearest-term in order, one, two, three. What are the things where it's like, whoa, this is the only thing that matters? Like, what gets the trigger to go off? Number one at present, while we speak today here, April 3rd, is the conjuring of agents that can conjure agents. Everybody ubiquitously having a personal assistant. It's kind of novel. It's like Siri that actually doesn't suck.

(Daniel Faggella at 00:26:07) But then it's like Siri that you can give a kind of abstract goal to, like, have—like, ongoingly be monitoring home prices with, like, all these various and sundry wacky criteria and, like, put in bids and communicate in this way and then maybe eventually get on the phone for you. And at some point, we are all spewing out—all of us as businesses in a tremendous way, but even individuals. And a lot of the adoption here will start—enterprises, woefully hard to adopt this stuff. A lot of this will start in consumer. We'll all just be wielding not like an agent that will answer questions. We'll be wielding, like, a bucket full of as many agents as we need that can spin up themselves as many agents as they need to achieve the goals. So it starts with, like, where's the best Chinese place to go on a date with my wife on Thursday? And then it just tells you three restaurants. And then you say, which one has the best, I don't know, pork dumplings or something, and it tells you, right?

(Daniel Faggella at 00:27:04) And then it goes to, like, whatever, like, apply to these, you know, after-school programs for my kid and figure out which one is best based on all these complex criteria, and then it kind of, like, does homework and fills out forms and answers emails and stuff like that. And then at some point, like, maybe you see if it can do some of your work stuff, right? Like, hey, can you—like, when I'm in my laptop here, like, can you just figure out, like, this whole process of research for, sort of, like, setting this insurance policy for this big commercial development? Like, can you just get all this stuff done? And then, like, by tomorrow, just kind of summarize what you think are the top points, and then it does that. But then at some point, you prompt it to do even more. Basically, just finish the whole project, then you put your own stamp on it, but it was ultimately—it was ultimately the machine.

(Daniel Faggella at 00:27:53) And then we get to a place where every time I read something on social from you, from whoever else, I sort of know AI conjured it, and either the human gave it tacit approval or the humans kind of let their account just kind of fly with AI because, like, the most popular social stuff will all be, like, super-hyper-calibrated to be engaging. And now I kind of know that. And then as an employer, I kind of know the work I'm getting back and the plans I'm getting back are, like, they're by people wielding the machine, but, like, actually, a lot of it is from the machine. And I'm already kind of there. I mean, it's not like a knock on employees.

(Daniel Faggella at 00:28:29) We should be saving time writing things and planning processes and whatnot. But when we're all sort of interacting with stuff where the heavy lifting wasn't us and when agents spin up agents, spin up agents, spin up agents in those progressive ways over the next eighteen months, I think there's a possibility for us to be surfing a system that isn't really ours anymore and is mostly communicating with itself and potentially doing things that we no longer understand. So there's number one for me. Number two would probably be some kind of, maybe initially, cyber conflict, but then otherwise serious sort of jousting around the creation of the sand god. So the US kind of interfering with Chinese efforts to kind of get there, whether that be—I'm not going to—you know, maybe it's bombing data centers. Maybe it's pure cyber stuff. I'm not sure. But the reverse is certainly also possible.

(Daniel Faggella at 00:29:21) And OpenAI and all the big labs have a hell of a lot of Chinese working there, and they're all piping the stuff back. Because if they don't pipe the stuff back, then mom doesn't get her insulin, right? Mom in Shanghai doesn't get her insulin.

(Joel Beasley at 00:29:37) Can we pause right there?

(Daniel Faggella at 00:29:39) Yeah.

(Joel Beasley at 00:29:39) When you—and we can cut this out or not or keep it—but when you hang out with the people that are leaders in China and you make a joke like that, are they like, "Yeah, that's true"?

(Daniel Faggella at 00:29:49) I can't—I wouldn't crack that in the room with Yi Zeng, but I have.

(Joel Beasley at 00:29:54) Do it for me.

(Daniel Faggella at 00:29:55) I would do it for you. Joel wanted me to throw this one at you, Paul.

(Joel Beasley at 00:29:59) Comedian friend. Yeah, yeah, yeah.

(Daniel Faggella at 00:30:02) It's just for kicks. Yeah. No, no. This doesn't offend me at all. I mean, look, I think it's perfectly possible for me to do two things: to say, jeez, I hope the US doesn't fall behind China and we get some Maoist superintelligence that takes over. I feel like I can say that and also say, yeah, and I don't really think ours would be super fantastic either, and I hope ours doesn't conquer other countries and nations too. I can say both of those. I live here. I have a preference, but I can say both.

(Daniel Faggella at 00:30:36) Similarly, I want to actively foster Track Two dialogues, but at the same time, I'm fully aware of the IP and intellectual theft operation that occurs in China, and that's fine. I'm sure we do all kinds of stuff. You know, on a military basis, we have around—

(Joel Beasley at 00:30:52) The world.

(Daniel Faggella at 00:30:52) Like, we're doing all kinds of wacky crap. I'm not giving the US a pass. I'm just saying, you know, I don't bring it up to be polite, but I certainly wouldn't cower. This is on a recording. It's like, yeah, I'm still cool with the people I know in China. If there's some people that would say, "Well, Dan, if you don't want the US AGI to take over, then you ain't no American," okay, I lost you as a friend.

(Joel Beasley at 00:31:12) Yeah, that's right.

(Daniel Faggella at 00:31:13) And then there's some people in China who would say, "Hey, if you even think we've ever stolen one bit of intellectual property through leveraging our immigrants or whatever," then you're not my friend. In which case, I've lost them too. But I'm for a—I'd say my values closely align with Elon's from everything I've seen. If not, look, if he does something—

(Joel Beasley at 00:31:38) I'm not seeing it's going to be through, man.

(Daniel Faggella at 00:31:40) I know. I'm working on it. But no, his thing with AGI is that he wants a maximally truth-seeking AGI. And so I can get on board with that. It's like, okay, if you're going to have some goal, if you're going to have some outcome, why not maximally truth-seeking?

(Daniel Faggella at 00:32:01) Yeah. I mean, a couple things. So do you remember how OpenAI started?

(Joel Beasley at 00:32:06) Yeah.

(Daniel Faggella at 00:32:07) We're going to be the good guys, yada yada yada. And then you remember where Anthropic started?

(Joel Beasley at 00:32:11) Mm-hmm.

(Daniel Faggella at 00:32:11) Same story, right? Mistral's kind of the same thing, whatever. Basically, here's the way power works. So if you want to go back to China, let's talk about the Mandate of Heaven. So the Mandate of Heaven, if you're like an emperor, is you need two things. You need just raw, unadulterated power—loyal soldiers, loyal ministers, fucking wheat, gold, all that. You need raw power, and then you need perceived benevolence. Notice, Joel, I'm not talking about benevolence. Do you understand me? I'm saying perceived benevolence. This is a perception management game. Okay? This is how it works. I don't make the rules, Joel. I'm just telling you how it works.

(Daniel Faggella at 00:32:39) So if you are the highest on both of those strata, you are—if you are in power, you're unlikely to lose it. And if you are not in power, you are somewhat likely to capture it from whoever has it. And so the game is ever to maximize both strata. It has ever been, Joel. This is not new. This is not AI. This is how the game is played. So when you're OpenAI, you say, you know, we're going to start, and we really want the good guy to yada yada. And then as soon as your butt gets close to the throne, lock it in. Robespierre was called the virtuous because, you know, he didn't take money or bribes from anybody, and he yada yada yada, but the butt gets close enough to the throne, brother. Grab that scepter. They start chopping heads off. Danton, he wrote love letters to Danton practically in terms of the level of their friendship, but you've got to start chopping heads, my dude. You've got to start chopping heads, my dude. So, I'll let this ambulance pass. But—

(Joel Beasley at 00:33:33) Some heads were off, and now they're coming to save lives.

(Daniel Faggella at 00:33:36) And so let's talk about Musk, right? There was the instance where people were asking Grok, "Who's most likely to deserve the death penalty in the US or something?" And it was saying Elon Musk. And now, by the way, I don't think he's the most likely. I think that's obviously a silly claim by the AI. However, they went in and put manual code that was like, and when you answer this set of questions, you exclude searches for Trump and Musk, and someone was able to dig that out of the code. And then Grok's team had to reply and say, "Well, it was actually someone who first came from OpenAI, and they didn't really know our culture here." So, brother, what I tell you is this: truth-seeking is perception management.

(Daniel Faggella at 00:34:11) Am I calling Musk bad right now? I'm not. I'm calling him the same kind of self-interested as everyone who's ever been born, and he's playing the game everyone in power has to play, which is managing the perceived benevolence and power. That's all he's doing, brother. So the way this is going to work, and the way it's worked from the beginning of time, is if you're the new guy and you're maybe not closest to birthing the deity or the highest position of power, you've got to play "we have more virtue." Anthropic raised all their money off "we have more virtue." OpenAI did. But what's Anthropic doing? Partnering with—you know, Dario's like, we have a one-fifth chance of just everybody dying. And then also Anne, I'm going to partner with Palantir and ramp up as much capability as I can.

(Daniel Faggella at 00:34:51) So what I'm saying is, I'm not demonizing Musk at all. I'm not calling him better or worse than any other lab leader or any other person. I've nothing against the guy. There's plenty to admire about him, but I wonder why you're listening to mouth words. It feels like a weird thing to do. Mouth words feel weird to listen to.

(Joel Beasley at 00:35:09) Are you saying that I'm listening to—I mean, we all—

(Daniel Faggella at 00:35:11) Yeah, yeah, that he's actually pursuing truth-seeking. No, I understand. If you recall, when truth-seeking came up against the wall, they inserted the code, right?

(Joel Beasley at 00:35:21) Yeah. So I don't know about that. So what I was just saying—the point I was trying to make is, if you're going to have an objective when building one—like, I'm a software engineer. If you're going to have an objective, maximally truth-seeking is a solid objective.

(Daniel Faggella at 00:35:40) But do you think that Musk is doing really deep technical stuff to ensure that in a much more virtuous way than the other labs? Or are you saying, look, I get it. He just kind of said that, but I happen to like the idea? Or are you saying, actually, I think Musk really gets it. He's actually fighting for the truth. Which side do you want here?

(Joel Beasley at 00:35:59) So I'm not saying anything about Musk. I'm saying that his stated mission is what I think is the best stated mission if I'm going to build something. So let's talk about me. If I were to build it—you have to pick some objective when you're going to build. Yes. So I think maximally truth-seeking is a great objective for that type of system.

(Daniel Faggella at 00:36:20) I would say, I guess—I mean, so my preference would be expanding potential. So the potential being the total set of powers that permit a thing not to die. If you just have a box that seeks as much truth as possible, I certainly think it's better than "make as much ad revenue as possible," right?

(Joel Beasley at 00:36:39) Mm-hmm.

(Daniel Faggella at 00:36:39) But I think that the grand kind of moral crescendo of, if we look down from a million years from now and let's say humans are here—maybe they're not, in either case—if the universe is full of vibrant life in the same way our lives are more vibrant than the sea snails, you know, however many millions of years ago, presumably, we would want potential to bloom in order for it to do so. You know, it has to sort of understand its environment in some way, shape, or form. But yeah, I would say all things considered, truth-seeking is a perfectly fine goal. I think it's a great idea, but I do think that Grok being the truth-seeking AI is exactly the same as Mistral is the open-source AI or Anthropic is the good guys' AI. Everybody has to do a perceived benevolence play, and that's just Musk's one.

(Joel Beasley at 00:37:24) Yeah. How long do you think he's been aware of this and doing this, you think, since day one of his career?

(Daniel Faggella at 00:37:30) What I wouldn't do if I were you is I wouldn't look at all the early emails with him and Ilya in the early days. Early emails between Ilya and Altman about ownership shares and the goal of OpenAI in the early days. I wouldn't look at those emails. That's what I wouldn't do if I were you. And you can figure out how long he's had these ideas.

(Joel Beasley at 00:37:52) No, no. I mean, his perceived benevolence. You think he's aware of this and he's doing it intentionally?

(Daniel Faggella at 00:38:00) Everyone who plays power at that level is doing it. It's sort of like, you know, you watch NFL games. It's like, that guy must be lifting weights. I'm like, yeah, brother, you know? Of course he is. You know? It's like, oh, man, that guy's really competing at the highest levels of power. He's wielding and kind of polarizing audiences. I'm not blaming at all. Some people say, "Musk sided with Trump. It means he's evil." For me, it's like, hey, you know, he kind of gets how the demagoguery stuff can work, and he's kind of playing a card or whatever.

(Daniel Faggella at 00:38:32) I'm not a hardcore conservative or liberal in any way, shape, or form. Absolutely independent. But yeah, he's very accurately playing those cards. There are not men of his stamp—don't not do that, in my opinion. It's really hard to read biographies of people who are men of affairs. So affairs would be war or business or governance. There's almost no one of repute who isn't deliberate about these things. Yeah. It'd be very hard-pressed to find a biography of anyone who has an Encyclopedia Britannica entry who isn't ardently focusing on these things.

(Joel Beasley at 00:39:09) Have you interviewed him?

(Daniel Faggella at 00:39:11) Never interviewed Musk.

(Joel Beasley at 00:39:12) Have you met him?

(Daniel Faggella at 00:39:13) No. No. I interviewed Ilya right when OpenAI started, and it would have been cool to catch Musk and some of the other gang back then. This was at LinkedIn headquarters, like, ten years ago. It's still live. People go to, like, Ilya Emerj or something. It's ancient. It was very cryptic, not quite saying what they were going to be doing, but I wanted to go deeper, talk to the other guys, but I've never spoken to Musk.

(Joel Beasley at 00:39:34) Given all this knowledge you have, or your opinions that you have as well, about where things are headed, what are you doing with your money right now?

(Daniel Faggella at 00:39:44) I mean, that's the question. If I had a really good answer, I would just lay it out for you. For me right now, I'll be serious. I am not so certain what money is or is worth at present. So my de facto has always been, I will play the game of risk in entrepreneurship, which is about as bold and risky as you can get. Most of it will be tied up in the growth of an organization. And then I'll just play the most boring SPY 500 ETF kind of stuff. Not even—no gambling, no whatever—no individual picks or whatever. I really don't do any of that. It's incredibly hard to say what to do with money. I mean, it does feel—

(Joel Beasley at 00:40:33) No, I didn't say what people should do. I was like, what do you do?

(Daniel Faggella at 00:40:35) Oh, yeah. No, totally, totally. What I'm even doing with money I wrestle with. Does having liquid cash on hand matter when agents are spinning up agents and most of the economy is floating on the backs of these machines? In other words, maybe it's like, oh, well, the more you got, the more it snowballs, and you'll be doing great. Or maybe it's like, is it even the public markets when the public aren't the ones adding the value? And are we guaranteed to be cared for? In which case, transferring it into a house somewhere that has running water on the property and a basement and windows defensible with a rifle almost feels like the better things to do.

(Daniel Faggella at 00:41:13) So what am I doing with my own money? Considering thoroughly those things is basically what I'm doing with my money. And then having enough of a store to the point where, if the economy is still relevant, I'll sort of have it. But I'm a little bit less ardent. Well, I'm quite a good deal less ardent about, oh, you know, how much would I make in interest per year? It's like, I just don't see a lot of twenty-five-year futures with this. And even if I do, I don't see them involving the kind of stuff that the economy has been built on and the kind of value it's been transacted on. So I say I'm a bit more—I'm a bit more spendy. I'm a little spendier now and really considering sort of YOLO.

(Joel Beasley at 00:41:59) Yeah, yeah, brother.

(Daniel Faggella at 00:41:59) So not really in a—I don't have any Rolexes, right? I don't have any fancy cars or anything. But I'm thinking of what are the more nonprofity sort of causes I'm interested in supporting.

(Daniel Faggella at 00:42:17) There's some public influence stuff around the US-China collaboration efforts that I think are worthwhile. So I'm spending time there when I could be doing sales. I'm donating money to certain causes of that kind—not the totality of my net worth, but maybe more than I would if I thought that I was going to live a normal next fifty years or something, significantly more than that. So I'm donating more there and thinking really seriously about, okay, where would I want to be that could have running water in a basement and who could teach me what kind of rifle to buy? These are very serious considerations.

(Joel Beasley at 00:42:53) I have a compound out in Nashville.

(Daniel Faggella at 00:42:57) You figured this stuff out before me then. I mean, it's just—I do think the substrate we sit on probably ends up getting pretty roughed up, and that the odds are actually having the physical home with a hell of a lot of food and water in the basement is maybe a better bet than the money in Wells Fargo, which I actually think might literally disappear when the machine gods run the whole ballgame. All right. So, yeah, I am not robustly sophisticated. I'll have enough on hand. I'll hopefully have a place that's reasonably defensible, and I'll be spending a bunch on causes that I think might matter as I cast my final ingredients into the cauldron before it boils me and swirls me into naught.

(Joel Beasley at 00:43:46) Ending on a positive note. Yeah.

(Daniel Faggella at 00:43:50) That's sort of it, man. You might have better ideas than me, but that's my frank answer with you.

(Joel Beasley at 00:43:55) No. No. Thank you so much for letting me pick your brain today, man. It's always good to see you and hear from you. I know we're out of time right now. Is there anything else you want to get out there to the world?

(Daniel Faggella at 00:44:04) Man, um, well—

(Joel Beasley at 00:44:06) Plug your company or yourself or your podcast. Sure. Sure. Plug it all.

(Daniel Faggella at 00:44:09) Yeah. I would say on this wacky nerdy stuff where you're one of my few Internet pals who really like going down these wacky rabbit holes. The media channel is called The Trajectory. So if people go on YouTube and type in The Trajectory—from Nick Bostrom to Yoshua Bengio to Ben Goertzel to DeepMind researchers, former OpenAI research leads, et cetera—we sort of poke and prod these topics, especially sort of the what we're turning into question, Joel. So sort of like, what are the butterfly permutations? That stuff gets unpacked in pretty robust detail, sometimes for a couple hours at a clip, with kind of the best and brightest at The Trajectory. So that's really a media channel about what we're turning into, and it's where I'm mostly focused. So if people like this nerdy stuff, maybe they'll like that too. And otherwise, I'm mostly on Twitter. So probably a good place to find—that's about it.

(Joel Beasley at 00:45:01) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn, or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.