Conversation with Mustafa Suleyman — The Proliferation of AI & the Next Wave of Technology

24:53And so it gets cheaper and more efficient. I mean, it sounds kind of like an obvious statement of fact to point that out, but it bears repeating because if indeed that is a law of technology, if it is like a fundamental property that things that are useful get cheaper and easier to use, it means that they spread far and wide. And the rate of change in the last, you know, sort of five decades, first in microchips and then in software itself is truly unprecedented. I mean, that word gets overused, but it really, really is exceptional. We're now seeing massive exponential increase in the total amount of compute used to train cutting-edge AI models.

25:3710x, in fact, compute every single year for the last 10 years. Moore's Law, which everybody is familiar with, is a simple doubling every 18 months of transistor the density per dollar. And that has been consistently going since the 60s, which is a remarkable achievement in itself and has led to, you know, all sorts of incredible benefits that are unquestionable. But it actually pales into insignificance when compared to the trajectory that we see in software. And I predict that we're about to see the same mass proliferation in a highly meritocratic way of intelligence and tools as we did with access to hardware.

26:22You know, today, whether you are a billionaire or you earn the minimum wage, broadly speaking, you get access, if you're living in the first world, you know, to a smartphone and a laptop, which is just as good as everybody else. That's an unbelievable, positive story. And I think over the next decade, many hundreds of millions of people, if not billions of people, are going to get the same equal access to an intelligence, like as I described earlier, the ability to synthesize, plan, and predict. And I think that's going to be one of the greatest boosts to productivity that we've seen in human history.

27:06You said that we're not prepared for the proliferation of these technologies. How so, and what is the manifestation or the downside of not being prepared for them? Well, in some ways, this is a compression of power, right? So if knowledge is power, then the ability to act on knowledge is like the ultimate source of power. To take accurate, precise, reliable actions in some environment is actually key to being effective as a decision maker of any type. And if these AIs now get good at not just distilling knowledge, but taking actions, which they will, they will learn to use APIs. They will make calls into third-party knowledge bases, third-party databases to retrieve information, to book things and buy things.

28:05they will learn to use the front end of a website. So, you know, instead of even needing an API, they'll be able to enter text in the correct text box. They'll be able to click on the right thing. They'll be able to read the pixels on the screen. They'll be able to phone other AIs and have a natural language conversation between two AIs about price, about availability, about negotiation terms, about, you know, some creative endeavor. And we would want them to do that in plain English so that there is an auditable record of that interaction. They could well do it in embedding space, but we don't really want that from an explainability perspective.

28:43And then, of course, they could phone humans and collect information and clarify what's required in any given setting. So that kind of is in some ways a compression of power. It's the ability to reason over information and take action in multiple different settings. That's what these AIs will be capable of doing. So that's great from a productivity perspective and for those who really want to do good with those tools. But clearly, there are going to be many people who, you know, want to broadcast their agenda, their agency into the world. You know, in that sense, AI is going to be the kind of manifestation of the best and worst of us, just as the social media age has been exactly that, right?

29:29You know, we've eliminated the barrier to entry to be able to broadcast in social media. Anyone can now do it. You don't have to have a license and a good education and be part of an established newspaper like you did 20 years ago. Everyone can now broadcast. And on the face of it, that's an amazing thing. It's produced unbelievable benefit and we should praise it to the max. At the same time, it's also caused, you know, quite a lot of chaos. And I think that what we're headed towards now is not just the ability to say things, right, but AIs with the ability to do things. And so I think then, you know, sort of up to your own imagination to imagine what kinds of people might get access to a cheap tool for getting stuff done in order to, you know, spread misinformation.

30:17So instability, you know, amplify polarization, take your pick of whatever concern you might have. Do you, and I'll reflect a bias here, I consider myself an AI optimist, and I find that there's a certain level of techno-narcissism where the founders or the people central to a technology either see it as a singular point of saving civilization or destroying it. And I find a lot of the AI, the original gangsters in AI, are like, well, I'm Dr. Frankenstein, and I'm worried about Frank, which I don't find that helpful. Do you think there's catastrophizing here? Do you think it's overstated or we should be more worried?

30:57Where are you on the catastrophist AI scale here? so i find the kind of are you an optimist or are you a pessimist frame of the the question sort of almost not that helpful because both are biases i mean what i'm trying to do is do my very best to cleanly describe the capabilities that are about to arrive um and try to to sort of offer that prediction to other people to interpret for their from their vantage point given their expertise, what you think the consequences are for the world that you know best, you know, whether it's academia or politics or media or commerce, whatever it is, these capabilities are coming.

31:40So my goal with my book and with a lot of my work is to try to just describe what I see coming. And that then sometimes makes people feel very uncomfortable, makes people then say that I'm a pessimist or I'm an optimist. I agree with you that there's a lot of catastrophizing here and and there is a bit of panic. And that's natural with the arrival of any new technology. Like, we've actually had that through most of the waves in the past, you know, from, you know, even the motor car or the arrival of the railway or the, you know, the first aeroplanes. I mean, you know, initially, it was very scary, the idea of going up in a plane, and now it's one of the safest modes of transport.

32:22So I think we do adapt. And I think that, you know, that's the process of desensitization and adaptation that we're going through at the moment. I think we've all just got to be patient with everybody wrapping their heads around what does that mean. And so I can see that, I'll use Google as an example. I always thought Google's genius from a shareholder perspective is that everybody has to use it, but no one can develop competitive of advantage using it. So it turns from a point of differentiation to a tax. And there's one company that has garnered, as far as I can tell, the majority of the shareholder increase from this unbelievable technology called search.

33:04Whereas, and I don't know if this is the right analogy, something like mobile or GPS, it seems like a lot of companies have sprouted up and made a lot of money. Do you see these types of technologies as great levelers that will create a lot of different ecosystems of value or that there'll be a small number of players that'll recognize the majority of the gains? I think that this technology is going to be quite different in the sense that many, many players are going to be able to use them for huge productivity gains. And the reason I say that is because, you know, we're seeing these compounding exponentials, both in terms of the size of the models, but also the models are able to achieve a fixed capability, but they get smaller.

33:52So they're shrinking as well as expanding. So let me just give you an example of that. In June 2020, when GBT3 was released, that had 175 billion parameters. And roughly speaking, the number of parameters gives you a sense of how expensive it is to run a query against it, to serve it, right? And today, you can achieve the same capabilities measured by a whole suite of independent benchmarks, question answering and knowledge synthesis and reading and comprehension and so on, factuality, with a model that is 2 billion parameters, right? So that's a remarkable achievement because it's that much cheaper to serve.

34:35So what's going to happen over the next 10 and 20 years is that the models are going to get smaller and smaller while still being able to achieve the same sort of capabilities performance with the quality of their generations. And that means that those models are going to spread far and wide, which means that anybody will be able to take advantage of a model like that. You know, no matter how strong the lock-in, you know, the threshold of capability that we are crashing through is human expertise. You're actually measuring it against can a human do a better job of reading this passage and rewriting it in the style of X, right?

35:14Or can it do a better job of answering a couple of questions about this passage or giving you the perfect answer to something? So we're getting more and more efficient at producing smaller and smaller models that can perform as well as a human or as the really big models. and that is going to continue that efficiency trajectory is one that we've seen for you know hundreds and hundreds of years and is going to continue with respect to ai which is a great story on the face of it because it means there isn't an obvious way there is going to be lock-in in my opinion over the long term over the short term there will be over five years to 10 years the big companies are going to be able to produce the highest quality in my opinion and people will turn to those companies first.

35:59So they'll get a compounding effect because they'll get the, they'll end up with, you know, sort of control of distribution and they'll control the market in the way that Google has with, with search. In some ways, Google's technology is, is cool and is valuable, but what they've really got is, is lock in on the both sides of the marketplace, right? They've got the brand. So people are very comfortable coming back to it and they've got the advertisers who are comfortable advertising with them. So it's a sort of lock in It's a three-way lock-in between the technology, the users, and the ads. And that comes from having a first-mover advantage.

36:34And when people talk about AI and they talk about the fears, it generally, I think it's sort of one of four things, that this thing becomes sentient and decides in a nanosecond that we should all die, or two, misinformation, three, some sort of income inequality or concentration of power, and then that it could be self-learning weapons. that you have little AI-informed drones that start, go rogue, and decide that, you know, just get better and better at killing people. Do you think, do you see, or any of these being the, what do you think is the biggest threat across these things, or am I missing one of them?

37:15So I think the good news is that the bigger these models get, the more controllable they are. And that was not obvious to a lot of people three years ago. When GBT-3 first came out, a lot of the panic was about, oh, this is going to be toxic, it's going to be biased, it's going to constantly produce misinformation, it's going to give you reassuringly inaccurate information and so on. Actually, that's turned out not to be true. So I think when predicting how things, what the implications are likely to be over the next few years and decade, it's important to sort of hold in your working memory the idea that these models are going to be really, really good and really accurate, and they will behave as intended.

37:57So I'm pretty skeptical that there are unintended intelligence explosions where the AI gets out the box, it manipulates you, it takes over. I think all these fantasy sci-fi things are actually a complete distraction. I think it's much more likely that the traditional threats we face, where bad actors who want to use existing tools to destabilize our world are suddenly going to have an easier time of it. I mean, that is for sure true. Like one example is that, you know, some of these models, all of the models are trained on, you know, vast amounts of open web data, which includes information on, for example, how to manufacture a biological chemical weapon.

38:38And because the models are pretty good at coaching you, you know, you can get into an iterative back and forth, you know, it actually reduces the complexity of trying to engineer a potentially dangerous biological weapon. And we've actually observed this in multiple models, and we are coordinating across the labs with Anthropic and OpenAI and DeepMind and my company Inflection on how to suppress these kinds of generations and share best practices among us. So those, to me, are the kinds of threats that we have to practically work on in the near term. And they're actually relatively easy to take care of.

39:14And if open source, you know, providers and open source, you know, producers of these models also participate in that, then I think that just like we did with spam back in the 90s and 2000s, I think we'll just make progress on these problems one after another, we'll just tackle them. And generally, you know, I think we'll be able to mitigate a lot of the downsides in order to get, you know, the full benefit out of these things. We'll be right back.

40:07frustration. Visit sierra.ai to learn more.

40:37the climate, and more. Read, watch, and listen to The Guardian for free at theguardian.com. So there's a pattern to new technologies, and it roughly goes, there's an innovation, there's a lot of short-term concerns around job loss, and there usually is some job loss in the short run. And then, you know, automation did destroy jobs in the factory floor in the automobile industry, but we couldn't envision heated seats and car stereos, and we end up with more jobs. it feels to me that it's the same here, that this is going to spawn a ton of new startups and new opportunities. Am I missing something?

41:15Again, I think it depends on the time horizon. I think I find it harder to believe that over a longer time horizon. And I find that almost certain over a shorter time horizon. So over the next decade, I don't think we're at any risk of large-scale structural disemployment where there are millions of people who cannot contribute their labor to the market, right? I think there's going to be a transition issue where people will have to train, adapt and find new opportunities in the workplace and so on. However, over a 30-year period, as I squint, you know, I think that is a, I think it is harder to see how we end up with full employment.

41:56Obviously, it's hard to know, it's hard to imagine what new demand we're going to have. But what I can see is that this is a very general purpose technology. I mean, if intelligence really does become a commodity, then we really will be able to do radically more with less. And I think that, you know, to have hundreds of millions of more intelligences in the world, albeit AIDS, you know, research assistance to creators and scientists, is, I think going to increase the likelihood that we make real progress towards significant abundance. You know, I think you mentioned energy production earlier. You know, I also think about synthetic food.

42:41I think about water desalination. I think about, you know, carbon sequestration. And we have these huge scientific challenges that we have to tackle over the next 20 years in order to flourish as a species, you know, as we move towards 10 billion people over the next century. And what we need more than ever are tools and aids to help us to solve those tough social problems. And so I feel very optimistic that that's where most of our energy is going to, you know, be focused. And that's where we're going to see huge productivity gains. But that doesn't necessarily mean full employment for everybody on the planet, right?

43:21And that's where I sort of maybe am a bit more concerned over a 20-year period that there will be significant unemployment, people who can't, you know, compete in the workplace. Have you given any thought around policy recommendations around if you were advising, you know, someone on the center, you know, the head of the Commerce Committee, what we should be thinking about? Because I would argue the robotics have been good for prosperity, but we've missed out on progress because we didn't think much about the people who got left behind. What are your thoughts? Yeah, I mean, I think that the trajectory that we're on is that we are sort of converting value that has been created by labor into value that is held in capital.

44:11so capital is going to compound far more quickly because capital was previously you know intelligence was previously labor now intelligence is going to become capital and that's going to compound far more quickly because it is a commodity that can be traded and exchanged and so that on the face of it is a good thing because it accelerates our productivity and i don't think the goal of society is to have people work for the sake of work, right? The end state is not work. The end state to me is productivity and freeing people, purpose, yeah, exactly. And freeing people from the obligation of working in ways that they don't want to work.

44:56And I think that given the choice, people will find creative ways to stay productive and enjoy life. And that will be a challenge. Like we will have to move many, many people out of the basic, you know, rudimentary life of just, you know, turning up, you know, without having to think about it. There's going to be a big identity shift in what it means to be alive in 20, 30 years, because, you know, things are going to get radically cheaper. You know, food will get cheaper. Transportation will get cheaper. You know, energy is the bedrock of the price of absolutely everything in society. And we are making radical progress on that front.

45:36And And if we are successful in really reducing it to something near zero marginal cost, the price of everything comes down, which is great for everybody. So that, I think, has got to be seen as the end state. And the question is, how do we manage the transition? So today, we tax labor more than we tax capital. So we all know that taxation is a tool for speeding up or slowing down some things. so if we want to make decisions like that rather than talk about UBI which is too politicized and hand wavy let's just talk about the current levers that we do have for slowing down adoption of some things and speeding up adoption of other things and in my mind that means reducing taxes on labor so that people can stay competitive in work for longer against the machines or against capital increasing taxes on capital that means land that means stocks it means you know obviously software software, you know, particularly AI.

46:34And I think that's the natural course of things over the next sort of 10 to 20 years. When you think about economic growth and as it relates to tax policy, do you think that we're doing a good job of inspiring the right? I mean, there has been, UBI is now a conversation. It used to be totally dismissed. People actually talk about it now. Do you think the public policy is adapting here? Do you think we're up to it? Do you think we need more young people in Congress, elected officials? What is the gap between public policy and how fast this technology is moving? Do you have any thoughts around how we should be thinking about our elected representatives and the way we approach public policy here?

47:22you know it's funny that you mention age because it is a sorry state of affairs everywhere you look um you know it is i do think that people are sort of too old and not able to adapt with the times and we need public uh officials in every area who you know really get it and aren't just technical but are living and breathing amongst this new digital revolution um you know i don't have an obvious prescription as to how to correct for that. But I think everybody agrees with that part of the diagnosis. It's a huge problem. And, you know, I think that we've got to figure out how to get people to have more confidence in the state and more care for and love for the state.

48:06Like, if anything, I've tried to frame my book as a love letter to the nation state. Like, I I really believe the nation state has to survive and we're going to need it more than ever. We're going to need a good, well-functioning mechanism for redistribution. You know, so some obvious things that we should be doing is that every cabinet should have a VP of engineering or a CTO. I mean, it's criminal that there isn't a technical decision maker, you know, at the top of governments and in, you know, public departments of work and so on. I think the second thing is, you know, we have to be prepared to pay public sector wages commensurate with what the open market, you know, sort of declares the price of that skill.

48:49Right. I mean, the fact that we expect people to take a five to 10x salary cut just for quote unquote public service. It makes no sense that that makes no sense from a completely self-interested perspective. You want the nation state to have the best quality people. And that means that you do have to pay for them. I mean, there are some other issues there with how you hold that power accountable and so on. It's not completely straightforward. But this idea that you'll get the best people in the world, you know, by paying a fraction of what they can earn in the open labor market is, I think, misguided.

49:21And that's obviously a really sensitive topic. But we have to confront these sort of difficult challenges if we want really high quality civil servants running our countries. So you said something, and I will get to Inflection AI, but something that is a passion point for us here, and that is you said that we need more respect or need to kind of a love letter to the nation state. And you roll with a pretty fast crowd. My guess is you're on speed dial to every major player in tech. And I find it heartening. I'm a big fan of Reid Hoffman, Bill and Eric Schmidt. And I find that they both generally have an appreciation for government.

50:03And then there's an entirely different sector of innovators who I feel like are growing, at least in terms, I don't know if they're numbers, but they're voice. They, for lack of a better term, are just in the business of shitposting government. And basically kind of position government is the bad guy. And I find it very disheartening. I would argue every major technology has been funded by the middle class and that the most valuable companies in the world build this thick layer of innovation on top of investments in technology made by middle class taxpayers with the world's most successful venture capitalists, the U.S.

50:36government. So, one, do you agree with that? And two, what is it about the tech community or components of the tech community that seems so hostile towards government and try and portray government and regulation employees as just, you know, you should just get out of the way and you don't get it? Yeah, I mean, I think that the broader problem isn't just the tech problem. I think it's an elite problem. I think that once people have been successful, they have the natural bias that leads them, confirmation bias that leads them to conclude that it was their success and their hard work alone. You know, I did it.

51:12It was me. And when I failed, it was the market's fault. Absolutely. Yeah. And so I see it across the board in, you know, in finance, in media, in all the sectors are skeptical of governments. And it's probably because they haven't actually had the chance to work in those governments or seen how hard it is to get things done. Like the number of times I hear people say, oh, I was, you know, fixing the housing crisis. I would do this or these idiots can't do that. It's like, dude, you've never run, never been involved in a really difficult, thorny social problem where you can't easily be measured by dollar and, you know, dollars and profit and the basic equation of a P &L.

51:50There's just a lack of humility there. And I think we all need to be a bit more forgiving and respectful and kind to civil servants who are choosing to, you know, try to make our country work. I mean, in some ways, one thing I really like about the U.S. is that there's real respect for the military, the fire service, the police department and, you know, first responders. I think we need to, you know, we have that less in the UK and in Europe. And I think that we need to extend that same respect that we have for first responders to, you know, members of the State Department and people who work in education and, you know, other policymakers and decision makers.

52:35Talk to us about Inflection AI. So at Inflection, we're developing a personal AI. It's called Pi. It's our personal intelligence. I think that over the next few years, everyone in the world is going to have their own personal AI. So all businesses and brands and celebrities are all going to have their own AIs. And just like in the last wave of technology in the last decade or so of social media, those commercial AIs are going to try to sell you stuff. They're going to try and persuade you stuff. They're going to re-rank. They're going to filter. They're going to prioritize their sponsored content.

53:10And I think that you want a personal AI that is on your team, that has a fiduciary connection to you, and that knows you intimately and is helping you to synthesize and sort for credible information online, to learn new information, to interact with other AIs, and generally function as your chief of staff. you know, helping you prioritize your day, plan your weekend, stay true to your, you know, New Year's resolutions and your goals for the year, et cetera, et cetera. And your personal AI, your pie, will interact with other AIs, with your friends, with your family, you know, with other sales AIs, other brand AIs, and really help you to be productive, be efficient, have fun, learn, think and do.

54:03And I mean, this year, your startup is, I mean, you have arguably the most blue chip investors ever. My understanding is you just raised at a$4 billion valuation. How does a company like Inflection, AI, what are the applications that ultimately result in some type of business model that can support that kind of valuation? Yeah, it's a fair question. um i'm committed to a business model where the user pays and um that i think is the right thing to do it is not the maximally profitable thing to do um we are a public benefit corporation which means that first and foremost whilst i am focused on returning value to my shareholders as a number one priority my adjacent equal number two priority is to make sure we do the right thing And I think that means resetting the business model of attention and distraction and creating this fiduciary relationship between you and your chief of staff or your personal AI.

55:05Just as you wouldn't want someone else paying your lawyer or your accountant or indeed your doctor that might not be you that might have their own agenda, you certainly wouldn't want your personal assistant and your chief of staff to be funded by ads or funded by some other random business model. that does mean that you're going to have to pay. And so our goal is to focus on a subscription approach and really create a personal AI that's on your side and aligned with your interests. I hope and I expect that many people will be prepared to pay for that kind of experience. Where do you live, Mustafa?

55:45I live in Palo Alto, of course, in Silicon Valley. So as someone, I went to Berkeley. I lived in the Bay Area for a decade, started companies there in the 90s, and then started heading east. Now I live in London. I'm curious to get your kind of compare and contrast living in Palo Alto versus London or U.S. for not versus U.S. contrast, compare and contrast U.S. and U.K. You know, I'm sorry to say it is night and day. I felt I've been traveling to Silicon Valley since 2010. You know, we were acquired by Google DeepMind. I co-founded DeepMind were acquired by Google in 2014. So I've been traveling there for a very long time and I have a sense for, you know, had a sense for the culture.

56:28But when I moved there full time in 2019, it's just a complete, you know, culture shift. The confidence, the optimism, the encouragement of failure, the kind of hopefulness, the fearlessness about what we can make and produce. No one looks ridiculous or silly. No one's trying to take you down a peg or two. And I just got ground down in London over the years with the kind of lack of self-confidence of the country and the kind of petty sort of bickering and snarkiness in the media and in the kind of, you know, just in the London scene in general. You know, I'm a real believer, like, you know, in like making things happen, making mistakes and picking up and moving forward.

57:15And that's just not a very british default attitude it's like embarrassing if you if you know if you screw up or if you get it wrong or if you make a mistake or if you have a failed startup and so on um you know not not that d mind was a failed startup it was incredibly successful but just the culture just wasn't encouraging of craziness and risk-taking and i love that and i love trying to i'm a natural futurist you know you know i just by default think about how things are likely to play out in the future. And what I found in London was that I was sort of running into people who just looked at me like I was, you know, smoking something.

57:52And then when I'm in Silicon Valley, people, you know, challenge me. Exactly. They're too busy on acid and mushrooms. That's right. Yeah. I think you're right on. The way I describe it is in the US, most business meetings start from position of yes. And in the UK, the business meetings, in Europe generally, business start from a position, I'm like, you know, probably not. It just not. It starts from a much more skeptical viewpoint. And just from a lifestyle perspective, and I'm interested, I don't know if you have kids or, I mean, you're living in two entirely different cultures right now. The way I would describe it is the US is still the best place in the world to make money, but Europe's still the best place to spend it.

58:37Your thoughts? you know um i find myself a kind of you know visiting lots of different places of the u.s now and i'm i kind of i like being here i can't imagine um you're sort of going back to europe anytime soon um so you're loving it bottom line is you're loving america i'm deep in the california game you know what they say about the brits when you get them to california they never go back i think santa monica being one of those largest concentration of of brits outside of yeah they I love the sunshine. Is that right? And to wrap up here, obviously the most important thing, your club in Premier League football?

59:15Arsenal. I grew up in high B. Oh, my God. We have to be friends. Bukaya Saka, did you see the game on Sunday? I did not, no. I've been in New York doing my book stuff, so I haven't seen anything. Oh, in extra minutes, two goals to go up 3-1 against Man U. Yeah. Nice. I just, I am, literally one of the reasons we moved to London is I use football as a means of kind of connecting with my sons. And I find there's just that Premier League football is just singular. And my oldest is Tottenham, my youngest is Chelsea, and I'm Arsenal. I just, have you gone to, have you been enjoying any sports while you're in California?

1:00:00Yeah, I've been to a few ballgames, actually, and I've been getting into it. It's fun. I end up like leaving five or 10 pounds heavier at the end of the day. But I guess that's part of the spirit. That's called America. Eating and drinking. Yeah, that's America. Mustafa Suleyman is the co-founder and CEO of Implection AI, a personal AI service. Previously, he worked at Google as a VP of AI products and AI policy. Before that, he co-founded DeepMind, which was bought by Google in 2014. His new book, The Coming Way of Technology Power and the 21st Century's Greatest Dilemma, is out now. He joins us from Palo Alto and most importantly is a gunner.

1:00:37Is that what they call us, a gunner? That's right. Oh, a gunner. Yeah, there you go. I wish you the best of luck. And not only because you're an impressive guy, but I really hope that more uber-successful people like yourself continue to promote and evangelize what is, in my view, the most noble organization in history. and that is Western and democratic governments. And I think we need more people like yourself and your backers evangelizing our connective tissue and that is our respective governments. Anyways, best of luck to you, Mustafa. Absolutely. Thanks a lot. Cheers for having me, Scott.

1:01:20Algebra of Happiness. It's been a wonderful and yet a tragic summer. It was so wonderful in the sense I got to spend a lot of time with my boys. I've been exceptionally blessed and have the flexibility and resources now to spend a lot of time with my boys and do wonderful things over the summer. It's also been a summer of tragedy. I unexpectedly have lost two friends. My first friend, Scott Sabah, who was kind of, for lack of a better term, my wingman in New York. Whenever I'm in New York, I'd call Scott and we'd go. He always knew kind of the cool restaurant. We both loved to go out, eat, drink.

1:01:58We traveled a lot together. Anyways, about 18 months ago, we came back from Tulum, a group of guys, and Scott had a bump on his head and went and got it checked out and ended up he had leukemia. and when, but no problem, we can handle this. I was told we'd need chemotherapy. They can handle it with drugs. The drugs aren't working. We need to do chemo. The chemo's not working. We need to do more toxic or stronger chemo to the point where we needed to be hospitalized when he was doing the chemo. Oh no, the chemo's not working. We have to do a stem cell transplant. Son is a donor, does a transplant, successful, incredible technology, literally resets your blood type.

1:02:43We're in the clear. Oh, no, we're not in the clear. It's back. And it's converted to something, I believe it's called Richter's. And Scott died in the hospital. And this was, you know, 54, good shape, successful, survived by his parents and his three kids, you know, a real tragedy. And then about a month later, my freshman roommate from college, Craig Marcus, I was diagnosed with, I think it's called Guillain-Barre syndrome. It's sort of a bacterial nerve disorder. Craig was the kid in our fraternity who was the creative one who was designing the shirts, went to, I think it was called California School of Art and Design, and shot out professionally more quickly than any of us.

1:03:28And by the time he was in his early 30s, was like a creative director at a big ad agency. and, you know, one of those guys, no one had anything ever bad to say about this incredibly, you know, just nice, sweet guy and, um, get developed this disease, thought it was getting better. And good news about the disease is you can make a full recovery, uh, discharge from the hospital into a rehab clinic to kind of get his strength back, uh, took a turn for the worse back to the hospital on his way to take a test and his heart stopped and never started again, again, survived by his parents and two boys.

1:04:03And this isn't, there's no real kind of deep inside here. I'm still trying to wrap my head around what happened, though I think there's one of those things or two of these things where you just don't ever wrap your head around it. The observation though, is the thing that really struck me, especially in the case of Craig, is that immediately, immediately when he was in the hospital, there was like this picture of eight of us from the fraternity who we kind of all lived together our freshman and sophomore year. Three of those guys, David Kingsdale, Jeff Browdy, and Gary Leshgold, you know, these are guys with very busy lives, very successful families.

1:04:39All three of them immediately got on planes in Los Angeles to come spend time with Craig in the hospital several times while he was in the hospital. And then when Craig unexpectedly passed, immediately got on planes and came back and did did things like empty out his apartment, find the life insurance policy, organize the service. We actually ended up doing the service at my place a couple of weeks ago. 150 people showed up. Anyways, but these were three guys who immediately just wanted to do what they could for Craig and his loved ones. And it got me thinking about friendship and not only the benefits of the importance of friendship, but how bad a lot of men are at it and what a tragedy that is.

1:05:28One in four men say they don't have a best friend. One in seven men say they don't have a single friend. And it not only diminishes or erodes your life while you're alive, obviously, you're more likely to live longer when you have good friends. But it also friendship is something that endures. You know, these guys gave Craig, and it's a tragedy, but they gave him a good death. And I think that that is such a nice testament, not only to Craig, but it's a testament to how important friendship is. that you want. You want to push the limits of your comfort zone. You want to express friendship to other people because when something happens and you can't be here, you want people who, by virtue of caring for you and loving you, care for and love your offspring.

1:06:19So friendship endures. This episode was produced by Caroline Shagrin. Jennifer Sanchez is our associate producer and Drew Burrows is our technical director. Thank you for listening to the Prop G Pod from the Vox Media Podcast Network. We will catch you on Saturday for No Mercy, No Malice, as read by George Hahn, and on Monday with our weekly market show.

1:06:50Why are we genetically or anthropologically predisposed to being allergic to geese? Are we not supposed to be around geese? Right? Is that like geese and humans are not supposed to mix? Anyway, don't know. Adobe Acrobat Studio, so brand new. Show me all the things PDFs can do. Do your work with ease and speed. PDF spaces is all you need. Do hours of research in an instant. With key insights from an AI assistant. Take a template with a click. Now your Prezzo looks super slick. Close that deal, yeah you won. Do that, doing that, did that, done. Now you can do that, do that, with Acrobat. Now you can do that, do that, with the all new Acrobat.

1:07:27It's time to do your best work with the all-new Adobe Acrobat Studio. Mercury knows that to an entrepreneur, every financial move means more. An international wire means working with the best contractors on any continent. A credit card on day one means creating an ad campaign on day two. And a business loan means loading up on inventory for Black Friday. That's why Mercury offers banking that does more, all in one place, so that doing just about anything with your money feels effortless. Visit mercury.com to learn more. Mercury is a financial technology company, not a bank. Banking services provided through Choice Financial Group, Column N.A., and Evolve Bank & Trust members FDIC.