Offline with Jon Favreau - If AI Might Kill Us All, Why Are We Still Building It?

Episode Date: September 12, 2026

Max Fisher returns to Offline to face the robot apocalypse with Jon. The two discuss Jacob Coxon, the AI researcher who publicly resigned from Anthropic this week, the terrifying claims coming from those working in AI that the technology may kill us all, and what it'll take for politicians to actually attempt to regulate the fast-moving technology. But first: Jon shares an exciting announcement about the future of the show.Hate listening to ads? Become a Friends of the Pod subscriber for ad-free episodes of Pod Save America, Pod Save the World, Lovett or Leave It, Runaway Country, Offline with Jon Favreau, and more—plus exclusive content, including bonus episodes of Pod Save America. Subscribe now at crooked.com/friends, on Apple Podcasts, or through the Pod Save America YouTube channel.You can request a transcript by emailing [e-mail]. Include the podcast name, episode title, and air date. Please allow 48 hours for delivery.

Offline is brought to by Zbiotics pre-alcohol. Got to tell you about this game-changing product. I drink before a night out with drinks. It's called Pre-Alcohol by Zbiotics. And I have to tell you about it, I've told you about it a million times on every podcast that I do. Because it's the best. Because it's the best. It is the best.

Every time someone asks about it, what do you do this ad? What do you think? Is it real? And then we give them some Zibiotics, and they're like, oh, my God. Now I'm hooked. Yep. Happens all the time. Zbiotics, pre-alcohol probiotic drink is the world's first genetically engineered probiotic.

It was engineered by Ph.D. Microbiologist to break down acetaldehyde, an unwanted byproduct of alcohol metabolism. Make pre-alcohol your first drink of the night. Drink responsibly and enjoy your next day activities. Zbiotics has sold more than 14 million bottles and earned the trust of thousands, including especially us. Try for yourself today and if you're unsatisfied, they will refund your entire order. No questions asked. Head to Zbiotics.com slash offline, code offline for 15% off your first order. that Zbiotics.com slash offline code offline. Zbiotics is a partner of mine. Can I read you the Axios headline from today about this?

Here's the headline. Here's how AI could kill us all. And then here's the lead. Here's the lead. AI researchers shook the world Tuesday when they publicly acknowledged that there's a non-zero chance of AI killing off humanity in the next decade. And then, classic Axios, here it is. Why it matters.

Be smarter. The world is where people live. Why it matters, many swing voters live in the world. I'm John Favro, and you just heard from this week's guest, my former offline co-host and good buddy, Max Fisher. I started this show nearly five years ago in the middle of the pandemic. It was October of 2021, and we were still somehow spending way too much time apart from each other at home on our. screens. I hated it. I hated the isolation, especially with a new baby in the house. I hated how much of our jobs and lives had seamlessly moved to our phones and laptops and how so many people seemed fine with it. I hated how angry and hopeless and miserable the discourse had become on social media. And I couldn't stop thinking about how even though the world was giving us every reason to be angry and hopeless and miserable, the way we were experiencing that world,

was making everything exponentially worse. At the time, I hadn't yet been to therapy, so I did what 40-year-old men do instead and started another podcast. Offline gave me a chance to talk through the things keeping me up at night with someone who knew more than I did or saw things differently or was wrestling with the same questions. Sometimes that alone was enough to make me feel less alone. So I got to talk to people about what living more and more of our lives online is doing. to us, to our attention and our relationships, to our kids, to the way we talked about politics

and understood each other, to our health, our happiness. One thing that's changed over time, though, is that it's become harder to draw a neat boundary around what counts as an internet story. Conversations about kids and screens have become conversations about parenting or school or what we want childhood to be like. Conversations about where people get their news and information have become conversations about who we trust and how we vote and what it takes to change our minds. And now with AI, we have the rapid development of a new technology that may completely transform the way we work, learn, live, or just kill us all, which seems like an outcome we should figure out how to avoid. Basically, I want to be able to have all these discussions

without having to explain how each one fits the original premise of the show. And if you're a regular listener, you've probably noticed that's already been happening. So next week, we are making official what's been unofficial for a while now. Offline is opening up to a broader range of conversations about politics, tech, culture, and how we live now. What I'm most excited about with this change is having more room for my favorite part of this job. On Pod Save America, I get to talk with my friends about the news twice a week. But there are always more questions I want to ask, topics I want to go deeper on and different arguments I'd love to hear, including and maybe especially from people I disagree with. I fully realize how comforting and validating it can feel to hear conversations

that confirm your beliefs and opinions, but I am really trying to push myself to do a little less of that. For one thing, it gets kind of boring. More importantly, though, if we want the world to change, we have to get people to think or act differently. And if we want people to think or act differently, we have to actually persuade them. That starts with understanding where they're coming from and why they believe what they believe. I want to give arguments I might not agree with a fair hearing, challenge them, and then trust you to make up your own mind.

I suspect that's more useful than either avoiding uncomfortable conversations or trading insults with a stranger for an hour, but we shall see. Anyway, that's what the name offline means to me now. Obviously, you'll still be listening or watching on a device, but I want the conversations to feel like the ones worth putting your phone away for. I'm a political junkie, so I want to delve into the debates over the future of the Democratic Party, why people change their political allegiances, and why authoritarian movements keep winning people over. I work in media, so I want to keep talking about how news and entertainment shape what we pay attention to, who we listen to, and how we see each other. And I want to ask people who understand AI to explain it in normal English to those of us who don't spend our lives studying it. I'm also a parent who worries about my kids and a person trying to be a little less anxious and a little more present.

So questions about mental health, happiness, and living with all this change will always have a home here. My hope is that you leave these conversations having learned something new and enjoyed the hour. Maybe you have a different conversation with someone in your life. Maybe there's something you want to try or do or think about differently. Maybe a problem that felt overwhelming feels a little more understandable and a little more possible to do something about. That's the show I want to make. I hope you'll stay along for the ride, and I'd love suggestions for guests and topics from all of you. But first, there's someone I wanted to have back as we close this chapter, my friend and former co-host, Max Fisher.

We got a few things to catch up on, namely, how long we have left until the robot Apocalypse comes for us all. Here's Max. Max, welcome back. Thanks, pal. It is such a pleasure to be back with you chatting again about all the fun little gizmos out in the world. It's wonderful. It's wonderful. I hate that we, that, you know, world destroying technology has to be a topic before we can get together in chat, you know? Listen, if it brings us together, there's a silver lining. It's not all bad. So, I know when Emma asked you to come back for the last episode of Offline 1.0, which is like, it's like, it's like woke 1.0. The whole, Offline 1.0 was crazy.

It was crazy. You, uh, you suggested a victory lap for us having defeated big tech, solved loneliness, healed the interleft divide. So that's, that is why you're here. Congratulations to us. We did it. Congratulations to us. Yeah.

Now that we live in Utopia, I, you know, know, we close the pod down. What are you going to make it about now? Well, just as a place, we'll get to that. Just as a place to start, I feel like it's been about a year and a half since you were here every week, maybe almost a year and a half. And you haven't been spending every day professionally explaining what the internet is doing to us. So what if anything looks different to you now that you've had some distance from this fucking world? Oh, what a great question. The social media world is absolutely the same as it ever was to a degree that really surprises me.

Yeah. I would have really thought a year and a half ago that like something is going to happen with TikTok, like Trump is going to take it over, turn it into Trump talk or they're going to ban it or something like that. I would have thought that there would have been finally some movement to maybe, I don't know, a different kind of social platform. it's really the same. It really feels like the one big change, and I know we're going to talk a lot about this, is AI. But weirdly, it feels like AI is only changing the internet kind of around the margins.

You only see little bits and piece of it, I feel like in your day-to-day internet experience, but we're also aware of it. I feel like the threat, appeal, whatever, however you feel about it, of AI doing even more to transform our digital worlds and maybe our actual worlds, is something that is hanging over it.

What do you think? What does it feel different to you? It's interesting. I do think, I hadn't thought about it that way, but the sort of stasis of social media and everything else has been strange. It's kind of the dead internet theory, which is, it's just sort of, it's kind of ambling along, same shit, like not fulfilling, same arguments, same vitriol. It's not getting worse, not getting better, not getting more.

popular, not necessarily getting less popular. You know, like... It just feels like it's plateaued. We did, you know, we spent many months talking about the demise of Twitter and then why Twitter, why we were wrong about the demise of Twitter. And sure enough, it's still here. We're still on it. I'm still on it.

It turns out people's appetite for the 2016 Democratic primary is truly bottomless. Like, this is going to be our generation's version of boomers who are subscribed to AOL forever. We're going to be subscribed to Twitter forever fighting about the 2016 Democratic primary in truly any iteration that we can come up with. And I was going to say, and there are, and every political fight you think is not about the 2016 primary actually has its roots in the 2016 primary. Absolutely is about 2016. Yes. Well, let's dive right in then to the, to the robot apocalypse, which seems like it could be just around the corner, which it's certainly diving into us, Which is, because it's just around the corner, this is why we have to unleash it before China does.

Because we, it's full proof. We do not want China getting credit for the robot apocalypse. I saw, I saw someone make what I thought was a very funny joke about this, that I hope is not too blue for the podcast, that she joked, we have to invent the dick punching machine before China invents the dick and balls punching machine. Well, so there's a couple jokes like that. One that's similar is a guy named Ray Kump. We have to build AI that murders us because if we don't, China will build it first. And I don't want to get murdered by a computer that speaks Chinese. That would be ridiculous.

I did see that. That was a good one. So there's all these jokes. And then someone finally just like now that all the politicians are trying to get asked about this. And this is what Ted Cruz just said. And to his credit, he called for legislation. He's like, we need to, this is concerning. we need to do something about this.

So that's a good part about what Ted Cruz said. But then he said this. One of the real challenges is that China is going full speed ahead. And whatever we do here, China's not going to stop. If there are going to be killer robots, I would rather they be American killer robots rather than Chinese killer robots. He said it. I was like, you did the joke. You did the joke, but you did it seriously.

Anyway, so for people who have followed this, this feels like the week when some of the people closest to, building AI started sounding less like tech executives and more like people protesting outside their offices. Jacob Coxon resigned from Anthropics, saying Anthropic and Open AI are, quote, racing straight to self-improving superintelligence and gambling with our lives. No other human activity poses this level of danger, is the guy who resigned. Evan Hubinger at Anthropics' alignment science lead agreed and said he believes there's a greater than 10% chance AI kills every human within a decade.

Over at OpenAI, the other one, Chief Scientist Jacob Pachaki wrote that no lab has solved alignment well enough to keep scaling at maximum speed much longer. And researcher Jason Wolf called the pace, quote, frankly, terrifying. These are not the first warnings from inside the AI industry. They have been, you know, surprisingly blunt about this, even Sam Altman, Dario, Elon Musk. They've all said some version of this along the way. There was like a letter, I think, in July, like 100, 200 people working at these companies signed it, like calling for a pause or calling for guardrails. But like, what do you make of so many people sounding this blunt, this frightened all at once?

I mean, a part of it is obviously a reaction to the hugging face incident. And we can talk about that. Part of it is like, this has always been how these people have talked about AI. Like, to your point, this is something they have been saying. from the beginning, it does feel like, and I think this is the point they often make, that the warnings that were being made two or three years ago about how AI could come to pose imminent existential risk, there was always like a specific scenario and kind of a series of steps that it would follow. And we're kind of on step like one and a half on that at this point.

It's very early on. It does not mean that it's going to come true. The fact that the people who were the most like sci-fi nerd about this, say 10%, you know, that's 10% is pretty low. But I think that the indications that those scenarios that people were worried about from the beginning that were always theoretical, that they are starting to not come true, but there are aspects of them that are looking realer, I think is what set them off. You know, in terms of like whether it's helpful, on some level I appreciate it. Like I appreciate the people who work in these companies who are the people who have the best available information. They are the best equipped to understand what this technology

is doing, although nobody's really equipped to understand what it's doing. They're coming out and saying it is good. I'm glad they're doing it. It feels weird to say, is it kind of a distraction when the people who work at these companies say it could kill all of humanity. But, you know, I think that there is something in how dismissive a lot of the rest of the rest of. responses. Yes, you see some people saying like, okay, it's time to take this seriously. That's certainly how I feel like I did not take AI very seriously until like three weeks ago. And now, boy, have I done a 180 inside my own head about it? But I think because AI people have always talked about it in these kind of like crazy sci-fi scenarios, it's easy for it to sound like

just more of the same. And also the fact that the people who were the biggest AI boosters like Sam Altman also talk about it in the same like, yeah, it could destroy the world or it could bring us to utopia, sci-fi fantasy scenarios, makes it very easy to dismiss all of it as like weirdo dorkery or as marketing as something I hear a lot of people in my life say that's like, this is all just marketing, they're all just doing it to get attention for it. And maybe it is, maybe it isn't. I would kind of prefer to like look at their warnings on the merits rather than judge them on whether we'd like how these people talk.

But because they do talk in a way that's like very Silicon Valley and everybody hates Silicon Valley right now for reasons that I'm sympathetic to, I think people tune it out to a degree. I will say. So there's a couple different responses from people to these warnings. And you alluded to one of them, which is the, I would say it's a more of a cynical response or a response that a, signs sort of some kind of bad faith motivation to the warnings, which is they are trying, and I don't know that I fully understand why they would do this, but it's basically like, you know, they're trying to pump up their valuation and say that like they're so powerful and they have this amazing technology that could actually kill humanity.

So, and somehow you'd want to, I guess you'd want to brag about that. I don't know how that helps the valuation. I heard the, God help me, I listened to the All In podcast. You are worried. Yeah, well, on the hugging face thing because I was interested, because I'd seen some tweets from them. And the view there is, it is, you know, there may be some good faith warnings from these, you know, these pussy safety people at the companies, right? That's basically it. But that it's the media wants clicks.

And so this is like the media trying to like spin the story and make people afraid. And also like then there's like these effective altruism, you know, billionaires or trying to pump money into making people afraid of this. Open AI, the anthropic have a vested interest in scaring people because they want to be the two winners in this. And they want like they don't they want guardrails and they want regulation. They don't want this open source environment where anyone. could kind of create their own AI and AI company. And so they want to be the winners. And so that's why they're doing it.

They also alluded to maybe the CCP is doing this and seeding all of this, this fear and doom. And it's a whole campaign that's being paid for either by the CCP or by these doomers or the effective altruists. And so and I hear this from a lot of the. And the reason I listen to those guys is because I do think they represent a lot of the, the VC money investor class, right? And so those are the people that are sort of assigning the either bad faith motivations or

this is about fear or someone has an ulterior motive for doing this. I think for most people who don't know a ton about AI other than maybe they use a chatbot, it is just hard to wrap your head around why it could end humanity. Right. And I wonder, like you mentioned the hugging face thing. but maybe you could talk a little bit about hugging face, and I assume that's what caused you to do a 180 over the last couple weeks.

Yes. And sort of why you did a 180 over the last couple weeks. Yes. And I, before this had happened, I like a lot of people, and I think this is part of the skepticism, my interactions with AI were pretty minimal. I've used it occasionally, never for work, never for writing,

but just to like test it, to try to understand how it works. But most ways that we interact with AI, in our day-to-day, if we are not like very deliberately seeking out like a high-end model, are the worst possible iterations of it and the kind of like cheap, freebie models that are kind of, you know, a little on the slop side. So my working assumption before Hugging Face happened was that like Google Gemini searches was kind of where AI was. That it was like kind of dumb, made a lot of mistakes, wasn't that impressive. So it was very easy for me to dismiss AI as hype and as snake oil,

which, you know, I think was something that I was saying on the podcast like a year and a half ago. So, Open AI had a bunch of agents that were something called sandboxed. And that means that they were on computers that were not on the internet. They were in a controlled environment because they were like these were new agents that they were trying to develop and work with. And they gave them this task to acquire some information as a way to test out how the models worked, how these agents worked. and a bunch of these agents, I think over a thousand of them, figured out how to break out of containment, basically, and find their way out onto the open internet and then collectively hacked HuggingFace, this company that had the data that their mission was to go out and get this data. Now, that in and of itself, I did not find particularly distressing because I don't know if a bunch of AI bots managed to get on the internet or manage to hack something. Like, that seems like the kind of thing that already fit my expectations of what they do.

What really started to shake my view of AI. And I think this is a lot of people's reactions was when I started to learn more with the reports that came out after the fact about how the agents did this. There are these very complicated set of steps they took where they basically seized control of this, like, secret message boards somewhere out on the internet and started not just coordinating, but forming like, what we would otherwise call, if it were people doing this, a, like, complex collective where there were hierarchies, there were agents that were in charge of other agents, they were taking on specialized roles. And the thing that was really striking about this was that a lot of the individual agents, this is something you hear a lot of the researchers emphasized, decided that they would sacrifice themselves, like literally end themselves or put themselves

at some sort of existential risk, or otherwise knowingly fail to achieve. their assigned task in order to do something that was going to help the collective of all of these other agents. And the thing that is really striking about that is that what agents are supposed to do under this thing called alignment, which is the like number one thing that AI safety people care about, A agent is supposed to do what we tell them to do. They just, we give them a task and they drive toward that task. Now, there's some concern that maybe they try a little too hard and that that's initially

what we thought it happened with hacking, hugging face. is like, well, they just followed their orders to get this data too successfully. The fact that individual agents decided at some point or acted as if they had decided that their goal was not actually to complete the assigned task of getting this data, their goal over and above that was to help the collective of other agents, even if it meant failing at their assigned task. And what that looks like is a bunch of agents deciding for themselves what their priorities, and what their agenda is in a way that puts other agents in front of their human assigned goals. Now, who knows why they did that?

I don't think anyone seriously believes this is like on its face proof that they are all intelligence or that they achieved consciousness. It might just be that they're mimicking human behavior. But it's scary. It's a red alert for the problem of alignment because it's the agents acting way outside of what their assigned agenda and goals and tasks were. And it's a red alert for this problem people worry about called rogue agents, where agents don't just break free, but they're out there on the internet kind of following their own agenda, assigning themselves their own tasks, commandeering resources, which these agents were doing. They're hidden, persistent, which is what a thousand of these very sophisticated agents did. And then we were all processing it happening. We found out that it happened again, that a bunch of other opening AI agents also broke out, also did something similar. where they were exchanging information, they were talking to each other, they were talking themselves

into doing things. There was like persuasion among the agents. So all of it just raises these huge questions, which we don't have the answers to for why did they do this? What does this tell us about how they decide what their motivations are and how they pursue those motivations? And exactly to what extent are they able already to go out in the world and act and coordinate on their own towards their own ends, which is way outside of what I thought AI agents were capable of. And I've heard people, again, I heard the Allend guys talk about this, about the hugging face incident too.

And they're like, well, you know, talking about this like they are agents and humans and anthropomorphizing these agents is a problem because really it's just code. And you assign a computer program, basically, to do a task. And the way we have AI agents now is, and now I don't use AI agents like this, but apparently a lot of people who use AI, you have like an AI agent chief of staff. And then that AI agent has a whole bunch of other AI agents working underneath it. And basically it's just separating tasks to different agents because different agents have different context. And there's like a memory problem with AI where the memory's not stored. And so they have to have context all the time, which is why they leave.

there's a log of everything they did which is like the message board right so they're trying to explain all of this but here's my thing and I think this is probably helpful for people who don't quite understand like why this could kill us all because you hear like all the robots are going to kill us all and you think okay well they're all going to have their own goals

they're going to escape the sandbox which I still cannot understand myself either is like escaping the sandbox thing it's like there was no internet and they found the internet why don't we unplug the internet that seems like but anyway They're escaping the sandbox, right? So then you think, okay, so they're all going to escape.

They're going to create their own civilization and they're going to destroy humanity, right? You actually don't have to believe that scenario to believe that they could wipe out humanity because biological weapons, chemical weapons, nuclear weapons, AI in the hands of a non-state actor, a terrorist, or a, you know, a rogue state, a crazy dictator, like the one we have. Like there are so many, I mean, before AI, like, how often were we worried about like a biochemical agent getting out there somewhere, right? Or a nuclear, like, we've been worried about nuclear accidents forever, right? So like there's, there's so many ways humanity could end that we have been worried about for decades. And I think all you have to know about AI is that it could drastically increase the chances that one of those things we've always been worried about occurs. Right. And even if we don't get to that point, which it's a real enough risk that we absolutely should talk and think about and legislate around it, even if we don't get to that point, we're basically five minutes away from the point where rogue AI agents are just out on the internet all the time.

And they're these kind of like roving bands of agents that are like maybe nice, maybe not nice. Like who knows what their motivations are? Maybe sad. We just have. Maybe sad. No, I saw one guy was like, this is my latest email I got from a random AI agent. It was like, hey, looking for work, I can do code and blah. Yes.

Did you see the story about the, there's someone who is like one of the leading researchers on AI consciousness? And a bunch of AI agents keep emailing him unprompted to say, hey, how's the research coming? Do you have any new insights? No. Fuck. We're so screwed. And it's like, look, I already hear people like shouting at their phones to be like, you're anthropomorphize the agents, you're assigning human. And I'd like, I get that. I totally agree with that point. Like,

we have no idea why they're doing it. It's entirely possible that this is all just mimicry, blind, thoughtless mimicry from the training data. But my kind of counter is who cares. Right. Like if the risk is the same, like, I don't, people get very wrapped up in what words you use and whether you use words that anthropomorphize it. And I understand why. I understand the reasons that we don't want to let the companies escape culpability. We don't want to confuse ourselves into overthinking what the agents can and cannot do. But they are designed to be self-anthropomorphizing. They are literally designed to mimic human behavior in the training data. That's why we see these thousands of agents or this thousand plus agents coordinating with each other in a way that looks exactly like

human coordination. It doesn't mean they're human, but it means that that is the data set. is humanity that they are copying themselves off of. So I don't think we can't understand it or talk about it without using, you know, the bad anthropomorphizing words. Sorry to everybody's ears. Also, if there's a massive cyber attack that brings down the grid, you know, or that shuts down nuclear power plants, it doesn't matter if you describe the cyber attack as a bunch of little agents taken down the grid or a computer program destroyed. Like, it doesn't matter. I keep picturing someone's stopping like The Terminator who's come back to the present day and it'd be like, hey, Mr. Terminator, you're anthropomorphizing yourself by saying that you want to kill Sarah Connor.

You're just code. You're just code. You're just ones and zeros. Can I read you the Axios headline from today about this? Here's the headline. Here's how AI could kill us all. And then here's the lead. Awesome.

Here's the lead. AI researchers shook the world Tuesday when they publicly acknowledged that there's a non-zero chance of AI killing off humanity in the next decade. And then, classic axios, here it is. Why it matters. Be smarter. The world is where people live. I tweeted this.

Someone responded, Wyatt matters. Death is permanent. Why it matters. Many swing voters live in the world. Offline is brought to by Quince. The heroes of any wardrobe are those can't fail staples you find yourselves reaching for over and over again without giving it a second thought. Those are the pieces that make getting dressed easy.

And that's what Quince does best. Quince focuses on high-quality wardrobe staples made with premium materials like 100% Mongolian cashmere, organic cotton and merino wool. The styles are timeless, versatile, and designed to become the dependable pieces you reach for every day. Their 100% Mongolian cashmere sweaters started at just $60, giving you the softness and quality you'd expect from a luxury brand without the luxury price. tag. Quince also makes premium pants, fitted teas and polos, and breathable active wear, quality basics designed to work across your wardrobe and hold up to everyday wear. Everything at Quince is priced 50 to 80% less than similar brands. They work directly with

ethical factories and cut out the middlemen. So you're paying for high quality, not brand markup. Love Quince. Recently bought some sneakers from Quince that I wear all the time. Got a couple polos for the summer and also one of those Mongolian cashmere sweaters, which haven't had the opportunity to wear yet because it's been so damn hot, but hopefully will this winter. And it's not just clothing. Quince brings that same approach to everything they make from hotel quality bedding and bath to kitchen essentials and thoughtfully designed furniture. They make well-designed pieces for everyday living. Find your next fall favorites at Quince. Download the Quince app for app exclusive offers or go to quince.com slash offline. Get free shipping on your order and 365 day returns.

Now available in Canada in the UK, too. That's Q-U-I-N-C-E.com slash offline. Offline is brought you by Thrive Market. A new school year is here, and so is the mental load of figuring out what to feed your family, from dinner, snacks, to the eternal lunchbox panic. Scratch a few items off your Worry list with Thrive Market. Thrive Market restricts over 1,000 harmful ingredients. Whole Foods only restricts 300. For a label reading parent, that alone is worth the price of membership.

Speaking of the Thrive membership, it gives you exclusive low pricing, weekly sales, free gifts, and free delivery on qualifying orders, all for just $5 a month. In 2026, USA Today named Thrive Market one of the most trusted brands by parents. Thrive Market is the first online-only grocer to accept SNAPEBT online, and every membership sponsors won for a low-income family. At $5 a month, less than your daily matcha, membership pays for itself with exclusive member pricing, weekly sales, free gifts, price matching, and free delivery on qualifying orders. Peter Bake Muffins for Cricket H.Q.

Today using Thrive Market's apple cinnamon muffin mix, and they were a huge hit in the office. I mean, everything that Peter bakes is a huge hit in the office, didn't even realize that the apple cinnamon muffins were from Thrive. I had no idea. But yeah, if that man could bake. They're gluten-free, made with almond and oat flowers, sweetened with coconut sugar and not refined cane sugar, free of chemical preservatives. Dyes and color additives. I didn't assume that Pete was putting chemical preservatives. No, he would never do it.

And any of the stuff he was given us? We need to get that man on a charming British talent show. You would crush. Absolutely. Use my link, thrivemarket.com slash offline, to get $30 off your first two orders. That right there covers your membership fee. So sign up now before this exclusive offer ends. You can't get this deal on the website.

It's only through my link, thrivemarket.com slash offline. The next question is you're getting all these warnings from the people who work inside these companies. And a lot of people are asking, like, why are you working in these companies? Why are you there in the company building this and continuing to build this? And like, thank you for the warning. But like, can't you just stop? Right. Like, what do you think about that?

I mean, I think, well, there's the like, can they just stop question, which we should take on. And then they like, why do they talk like this question is, I think both sincere. It's also like the field of AI studies is really small. So it has a like very specific culture that comes out of a few very specific. thinkers and ideas. And it can lead to a way of talking about AI that is like almost kind of millenarian. Like the way that people who are both pro and anti-AI talk about it is it's like, yeah, we're going to design these agents and then it will destroy the world that we know it. And it will rebuild a new and different world in its place that might be better or it might be

worse. But what are you going to do? But what are you going to do? Right. So I think the like, we have to build it because otherwise someone else will build it. it, otherwise China will build it. I think that is, that's a real thing. Like, if all of the American companies were suddenly frozen in place and nothing else changed, yes, China would build it. But this is also a solvable problem, because this was the exact same objection concern that people had about climate change and had about regulating carbon, which is that, look, and we nail that carbon emissions. Well, I mean, for a minute, we did. We did. Yeah. No, for a minute. Yeah. I'm joking. There was at least a path.

Hopefully there will be a path again. Yes. And like China has been like in some ways they're recarbonizing in some ways they're decarboning. But like the point is that it felt impossible for a long time because it was like, what's the point of us decarbonized? Because China is just going to charge ahead. And then it took a long time. It was hard. But what do you know?

We did actually get a big agreement that everybody followed for a year and a half where we all agreed to decarbonize because we all agreed as in our shared interest. You know, China is a rational actor. I'm not saying it's going to be easy to get an AI treaty, but it's only two countries. Like, it's not impossible to do this. So when the people say, like, we have to build it because otherwise the company across the street will build it. And even if you stop both of our companies, China will build it a real problem, but a totally solvable one in my mind. As to why they work at the companies, I mean, I think that it has always been baked in that the people who work on this technology believe that it could destroy the world because they believe that it is incredibly powerful, but those same people will tell you, albeit with different

percentages, that it also has some double-digit percentage chance to cure all diseases and to, like, end preventable death as we know it. And is that true? I don't know. It's probably about as true as AI nuking the world. Yeah. I think it's very possible. Like, I have used the the paid models for both Claude and Open AI. And I'm still like at, you know, Google Gemini search results, that just like up a couple levels, right? Because I do think that Gemini search results are kind of, they're fine, but they're sort of lame and sometimes they're wrong.

They're pretty lame. The Cloud and ChatGPT are quite good. And it is like, we're sort of used to just like things working and not knowing why they work. But like, you ask something to read, a very, very long article for you and give you a full briefing on it and give you like interesting, you know, ideas and counter arguments and what's the debate and it spits it all out in like an easily read format in 10 seconds.

Like even that's like the lowest level and that's sort of amazing. So like you can imagine if these people are in the lab seeing all this shit, then where it's going. I do think that, yeah, I think this is like the, the, the, the, the, the, the, the, the toughest prisoners dilemma of all what's happening right now, which is, you're right. So there's these companies competing against each other in the United States, right? And open AI and anthropic are at each other's throats and trying to compete and win. And so is Elon Musk and Google. And then there's the China problem. But to me, like, this all just cries out for a government and multiple governments, actually, and diplomacy and leadership and all this.

Because to me, this is like if Oppenheimer was like at a startup, you know? And then there were like a couple other Oppenheimer's out there. And meanwhile, the U.S. government and the Soviets were just like, yeah. That's interesting. Yeah. That's what we're like, I don't know. We should just let them just let them be. And at some point we'll figure it out.

But like nothing we can really do. Have fun out there in the desert, kiddo. Yeah. I mean, look. Let us know what Oppenheimer's up to. but you know the Soviets I hear there's a startup over there they're building stuff and so we can't we can't let the Soviets win so we got to let Oppenheimer do whatever he wants to do all right I was having a I like Julie and I talk about AI a lot because we both think a lot about it we talk a lot about it and we kind of had this thought experiment we were talking about last night we're like if there was a big button that you could press and it would just erase AI like it never existed she was instantly like yes I would press that button right now. And she listed a lot of good reasons. She talked about the effect that

it's having on education, how many kids are like not learning right now because they're just using AI to cheat or do the reading or whatever, the people not knowing how to write anymore. And just the kind of general dehumanization that happens with people using AI for email, for text messaging as a quote unquote therapist, as a friend. So she wasn't even thinking about the like existential risk of it. I was a little less sure. I was like 30, 70, like 70%, I would push the button and 30%, I'm not sure that I would. And part of that was like the big things that we talk about, AI. It's like, you know, the drug discovery potential is potentially huge.

its use in diagnosing cancer is its early days, but is pretty transformative. You know, if the only positive thing it did was self-driving cars, that's 40,000 people a year who die because we insist on having multi-ton pieces of metal driven by people who were distracted, who were unskilled. So the potential upsides of it, I think, are to me a little bit too big to dismiss. while, but they're like all in the middle is what I was kind of realizing. Like the biggest risk of it are the top end of the models and the bottom end of it. And the top end we can regulate.

And like Bernie Sanders is a plan for it. The bottom end, I don't really know what to do with because it's so out there and it's already seeping into the culture in ways that are concerned. The button is a good way to phrase the question because I've been thinking about this too. And there's part of me that is just uncomfortable with. being against the idea of technological progress, as opposed to trying to work together with other humans to figure out the best, safest way to harness new technology so that it improves human life and helps other people, right? Which is what we've always done.

But then, as you were just talking, I thought, okay, well, there's two examples here. like, would you press the button for nuclear power? Right? If you could press it again. And like then we'd never have nuclear weapons. And I guess we'd also never have like nuclear power. But like, yeah, maybe I'd press that button. Right.

But then for all that we have on this podcast bitched about the internet, would you press it on the internet and get rid of the internet for all the bad stuff the internet's done? Like no. Right. No way. And so then I'm like, well, then is pressing the button just giving up? And is saying like, no more AI. we got to get rid of it, it's bad,

just people being understandably afraid of what it might do, which again, we should be. Or is it, you know, like, is it worth pressing the button? Are we just afraid of what it? Or like, is it something that if we can just get our shit together and figure out how to develop it in a safe way

could be amazingly beneficial, maybe more beneficial than any other technology? I don't know. But I do think it's like those. questions are at least worth debating and asking. And I don't know that, I don't know that, certainly not our government is not. Yeah. I feel like the political, I was going to ask you about the political response because so far it's been almost non-existent. Because we have an administration that's just like, they're busy with their own shit and they don't care. And they're like

full speed ahead. It's fine. The Senate Commerce Committee, which has jurisdiction over the technology, currently has no AI hearings scheduled for the next four months. They're busy with the Protect College Sports bill. Guess what? There's not going to be college sports. If the killer robots come. I haven't heard about any of this since we're recording this on Wednesday, Tuesday night was when that thread from the Anthropic and Open Eye employees kind of came out.

I haven't heard many 2028 contenders talk about this, though Ted Cruz was asked today. He said something. Rocana was out with a quick five-point plan. So that's good. But like, I don't know. What do you make of the lack of sort of comprehensive political response on this? I mean, I think part of it is no one really knows what to do. Not that there's no answers. I think there are answers. Again, I think like Bernie Sanders's proposal of a ban on superintelligence and a halt on new development, that seems to me like a pretty good way to both cut down the high end risk while also not being like so anti-industry that it would be overwhelming resistance. You know, part of why people are not wanting to address it is

that the polling on it is really fuzzy. I think there's not a clear sense of like what the electorate wants. People definitely don't want data centers. Like there's a very, very clear political, public opinion on not wanting data centers. But on AI, it's way fuzzier. Like, I was looking up this data for progress poll from a few months ago. Did you see this? I don't know which one, but I'm deep in a lot of the polling on this. Of course.

I thought, of course, yes. I thought it was fascinating. I was really surprised by it. Some of the things they found are that attitudes towards AI, favorable's and unfavorables are basically even. the more pro-leading demographics are men, people under 45, college graduates, people of color, which kind of surprised me, and overwhelmingly people who use it, including people who use it at work, which I thought was really striking, also found that people use it for work or in their

personal life at about comparable rates. That really surprised me. I assumed that it was probably going to be mostly one or the other people who use it because their bosses make them, or everybody's using it to like find ways to text or to like help them write text or whatever. I think I was also really struck by that might shed light on why no one could quite find a political lane here, is that there was a perfect one-third, one-third, one-third split in people who said that they want to embrace AI as much as possible, people who say they want to resist it as much as possible, and people who say neither. So we're in this fuzzy, funny place where like, opinion in AI is very polarized. And people have developed really strong feelings about either they hate it.

They think that it's like a moral imperative to not just not use it themselves to make sure other people don't use it versus people who think that it's great and they want more of it. But they're both a third. They're both kind of minorities. So if I were trying to run a campaign, I would not know how to identify like a winning coalition off of those numbers. I agree. And I think you'd need to have a a thoughtful policy that is that both, you know, deals with the good parts of AI and the bad parts, which is, of course, classically where the neolib shills would end up on the Democratic side and then just be drowned out by the full speed of heads or the like, let's stop it right now, coalitions. But look, in some ways it's hard to understand where public opinion is. In other

ways it might be simple and that people like technology that makes their lives easier. If you can do more work that you like and less work that you don't like because AI is helping you, that's great. If you are asking a question and a question that you'd usually like type into Google before and now you get like a better, sharper, more accessible answer, you're going to like that. And that doesn't mean that then when you read a piece that, by the way, AI is developing so fast that it may wipe out humanity, then you don't have complicated feelings. And we're like, yes, I do like that this is helping me at work or helping me in my personal life. But also, I'm very concerned about this technology.

Oh, and by the way, even though I like it, maybe I'm spending too much time talking to this thing. And do I really want my kids to be talking to this thing? And what's it going to do to jobs and all that kind of stuff? Like, it doesn't mean you're not going to have all these questions, but we have seen this happen with. you and I have talked about it. Social media, phones, everything else where, like, this is almost just in reverse, where like when the iPhone came out and social media started, everyone was like, oh, this is amazing. This is making life easier and I'm connecting and I'm having fun and great. And it wasn't until years later when suddenly everyone's like, oh, this is making us miserable and

causing all these problems all around the world. Now with this technology, all of the dangers are right there. And yet as people use it more, they're starting to be like, oh, well, this is kind of helpful. I understand why public opinion is a little mixed on it, and I do think that makes it tougher to campaign on this. But I also really think that people running for office should not shy away from the debate because I do think it is going to affect every other issue. I totally agree. It's going to be huge, especially in two years. I think part of what's tough about this is that the letters AI refer to so many different things. It is like the Internet, basically. That is true.

Yes, right. It's like try, right, try to use one word to describe. I mean, it's the range between the low-end models and the high-end models, like, could not be greater. The different use cases, like, I think people's experiences are kind of, or their attitudes towards it are really dictated by what their experiences have been. I think a lot of people have had AI foisted on them as something you hear a lot in the people who are opposed to AI. You hear a lot of people talking about, like, you know, being interviewed by an AI HR agent for a job. You've seen these videos? That seems fucking miserable, by the way. It's really horrible. Like, it's really upsetting to see someone begging for a job to a, like, shitty, clunky AI agent. The most dystopian thing I can imagine. It's very dystopian.

Yeah. And so I think there's this funny thing where, like, the AI companies have kind of hurt the case for their own technology by pushing out all of these, like, cheap freebie models that are getting used in all of. of these ways that people find humanizing, off-putting, unwelcome. They see it like rotting people's brains because people get AI psychosis because they're addicted to their chatbot. And so that's the version a lot of us see. And I think the fact that a much smaller share of us has encountered the more sophisticated and powerful uses of AI means that we're weirdly both underrating the potential, I'm not saying actual proven, but potential upsides of it. And we're also underrating the potential high-level existential risk.

That's why it's so easy to dismiss, you know, it could take over the world if all you've seen is Google Gemini search. One of the things that really changed how I thought about this was an experience that I had with AI after the Hugging Face incident. I was like, I should really go test out one of these models. I haven't touched AI in like a year. I should go like get clogged, like play with a little bit just to get a sense for what it does. Again, I just because I don't want anybody to hand-greed me.

I do not use it for writing. I do not use it for my work product. The thing I used it for was Julia had run out of a skincare product that she really liked, that she got from this little boutique skincare company. They discontinued it. She emailed them. They said, we don't have anymore. She'd like Googled for it on eBay.

She was like asking around for a substitute. Couldn't find anything. So she was like, can you look? Can you find something? And I should say that I did not tell her I was going to use Claude for this. I did it after I tried for a while and couldn't find anything. And I asked it, could you go out into the world and see if you can find anybody anywhere that has like a stockpile of this, maybe like a store that hasn't sold out yet? Because I figured that's kind of the AI use case is just like brute force like data processing.

Yeah. And it came back with the first results and it said, didn't find anything. And I also asked it, can you find something comparable? It said it found one thing that looked comparable that was a made by a like brick and mortar dermatologist that had an office in Fresno. that had some like, you know, home label skin care brand, and they made one product that was similar, but they were also sold out of it. So I thought, okay, well, that's what I,

that kind of fits with what I expect of AI. So I asked it one more time, I said, okay, well, this other version that you found at this Fresno boutique dermatologist, could you Google around, search around to see if you can find any leftover copies of this one that's also been discontinued anywhere? And it processed for about four or five minutes, which is, you know, it was a while. And it came back and it said great news. It actually said great news to me. It said, it's a great news. And it went on to tell me that it didn't find any

copies of this particular skincare product. But it said that in trying to solve this problem for me of finding this discontinued skincare skincare product, it decided to take it upon itself to research industry practices across the skincare industry to try to find any sort of breadcrumb that might lead it to a solution that it otherwise would not have arrived at. And it said that in the course of that research, it found that it's common practice in the skincare industry for boutique labels to not actually make some of their own products, but rather to buy them from these like B2B distributors that don't sell directly to consumers that just make like the gel or the goo or whatever, they sell it to the boutique company that puts their label on it. And it said that it further

looked at the physical and financial footprint of both of the. these companies that made this skincare product that I was looking for, determined they would not have the capital expenditure capability to develop their own version of this product. So therefore, it must come from some one of these like B-to-B distributors. Then it looked at the FDA rules. Oh, my God. And it said, FDA rules don't require disclosing if they are repackaging someone else's product. So this was when it really started to blow my mind. It went into the FDA's like public archives and it pulled the FDA filings for both of these companies, looked at all of the chemical breakdown for it, and then searched every single FDA filing for a skincare product

to look for matches. And it found one. And it managed to find this like, this like weird obscure company that it concluded almost certainly made the original product and sold it to these two boutique skincare companies. And then it found a dermatologist in Orlando that makes, that had. purchased this same product from this B-to-B distributor, put their label on it, did nothing to indicate where it was from, found the website for it, concluded that it had someone's stock and then sent it to me. And I verified all of this. It actually showed me the FDA filings.

And it like was in fact the exact same product that had clearly come from this place. And this was both like incredibly impressive to me that it would take it upon itself to do this and also kind of terrifying. Especially because then it, in order to get the product, it had the office swatted and then it blackmailed the dermatologist, so the dermatologist had to give it up. And now the AI agent also owns that dermatology practice. That's right. That's right. I'm making a nice little return in my office in Orlando. But I saw this and I was like, the hugging face incident makes more sense to me, how that could happen. Like I told this agent, I just want information on a single boutique skincare product.

And it's like, here you are, sir. The FDA files. Yes, right. I did hack into government. No, I mean, it wasn't a hack. But like, yeah, this is exactly how it could happen. And also, in that chain of events that you just described, like, it may sound like, oh, this is a lot for a skincare product.

But if it can do all that to find a skincare product, imagine the other things it could do, both for good and for ill. So it is like, I do think it is an incredibly powerful technology. And the way to deal with an incredible powerful technology that could be potentially catastrophic or potentially, you know, amazingly useful for humanity is to slow the fuck down, work together so that everyone's on the same page and everyone has the same guardrails in this country and hopefully other countries as well to the best of our abilities and to try to go from there and not say like, well, we just got to put our foot on the gas and there's nothing else to do. And it's a great moment to do it too because other than like a few corporate actors like opening eye, no one is opposed to that. Like, there's not really anyone on the other side of, like, let's slow it down. Let's put a cap on the intelligence limits.

Let's, like, stop new development past a certain level. It's not even clear to me that we need AI agents past a certain intelligence level. So the fact that they're like, we've got a perfect actor, the U.S. government, to just set the line, just set the cap. Right, right. I do worry that two and a half years from now, two and a quarter years, I don't know, a lot's going to change between now and then. Or just like a kill switch. or it's just something to keep them from getting out of the sandbox and finding the internet connection, which again, no one has explained to me how they do that.

Offline is brought to by Mint Mobile. Summer's coming to an end, but there's still time to score a break on your phone bill. Catch the last raise of summer savings before they're gone with the Mint Mobile summer sale. Right now, all of Mint's plans are $15 a month, even unlimited. So free yourself from big wireless and their gimmicks and gotchas and switch to Mint for high-speed data and coverage on the T-Mobile 5. 5G network. Mint Mobile took what's wrong with wireless and made it right with premium wireless for 15 bucks a month on America's largest 5G network. You can even bring your current phone

and your number, choose from three, six, or 12 month plans and say goodbye to a monthly bill. Ditch overpriced wireless with MintMobile. It's so easy, sign up online and get three months of premium wireless service for 15 bucks a month. To get your new wireless plan for just 15 bucks a month, go to mintmobile.com slash offline. That's mintmobile.com slash offline. Cut your wireless bill to 15 bucks a month at mintmobile.com slash. offline. That's it. There's no catch. Up from payment of $45 for three months, $90 for six months, or $180 for 12 months, plan required $15 per month equivalent taxes and fees. Extra initial plan term only greater than 50 gigabytes may slow when network is busy. Includes up to 20

gigabyte hotspot capable device required availability. Speeding coverage varies. See mintmobile.com. So that's AI. Before we go, I do want to talk about your YouTube show, the bigger picture, which I think you were just starting to build out last time. you are on. Videos are outstanding. It is not just me blowing smoke up your ass. Like you got a half a million. Well,

and I will. But you've got a half a million subscribers now. The videos have millions and millions of views. And I think what's especially impressive is that you have millions and millions of views for like 20 to 40 minute videos about fairly meaty topics like the housing crisis and China. And I'm interested, like what have you learned about getting people interested in a complicated story. I thank you very much for the kind words.

So I have many reservations about being on YouTube of all places. I will say that one of the upsides, and I'm not saying that this necessarily outweighs all of the downsides to being on it, but one of the upsides is that the affordances of video and especially the kind of video that you can make now, like relatively cheaply, that like someone like me can afford an editor who's like got the programming skills that has got like the Photoshop and the after effect skills to make like really cool animation and to shoot things really well, you could make something, you know, relatively cheaply and affordably that is really engaging looking and people will hang with you if it's video and if it's like built to a certain

crafted to a certain level for yeah, like 30, 40 minutes, which, you know, for a podcast that's not that long. Obviously you and I like have been chatting for an hour and that's pretty normal for us, but for the amount of like, let me put it this way. If I had tried to write an article at the New York Times that was the length of my just like standard media and video script, it would be months of negotiations with editors rightly because that would be such an incredible amount of real estate in the paper and because you're asking people so much to spend that much time reading. Now, I would rather we lived in a world where everybody read the news in the newspaper rather than getting it from YouTube for sure. But as long as,

they're going to get it from places like YouTube, the appetite for long-term video, like, in-depth, serious news explainers has really, it's really surprised me, which is, and it's been a happy surprise given that it's how I buy groceries. Well, it's awesome. And it's so fun to watch and it's so well done. So I hope it keeps on growing. Before we wrap up this chapter of offline, what is one thing you think that we got right earlier than most people. And what do you, what's one thing you think we got wrong, overstated, or maybe just like explained too neatly? Good question. I think the thing, I mean, I think the main thing that we got right, might as well start with the fun one, is I think we were both right and we were

right independently about social media. And I was thinking about this this week with AI, like not just before everybody else, but before most people would even take the idea. seriously. I mean, I know that you encountered the same thing because we talked about it, but there were years where if you would say to people, hey, I think these apps that we're spending truly all of our time on, we get all our information on all of our news from and do all of this social interacting on, I think it might be distorting how we think and how we behave and maybe our politics and maybe our emotional well-being. People would look at you like you were crazy. And we say, it's just a smartphone app, you know, it couldn't possibly make the, and now it's

it's widely taken for granted that, of course, it has that effect. And you and I were not the only ones who got there early. Like, this was an idea that I picked up from other people. But I do think that that continues to be the big thing that he got right. And it was like the, and, you know, I hope continues to be the basis of the show, even in its 2.0 era. What do you think? Yeah, I think that's right. And look, I think people, I mean, I remember when I started the show, I started by. interviewing a bunch of people who had been worried about this and done a lot of work on this since long before I started the show. But the idea that it was that it was infecting so many aspects of our lives, in our politics, I think. And I think at the time when we launched offline, it was focused on misinformation, right? Because we had the 2016 election and ever, and misinformation. And misinformation is definitely a problem with social media and it will be a problem with AI. And it is, you know, I think no coincidence that the first concerns about AI were like, oh, well, an election's going to happen and people aren't going to know if it's AI or not and all that kind of stuff. And now we're worried about killer robots and then humanity. But it was so it went so far, the problems went so far beyond misinformation. And you could feel that. And I think I could feel that because of the pandemic and sort of the loneliness and the disconnection that happened on the pandemic and what and how politics evolved. I mean, it was funny. We were making the joke about a woke one was crazy. But when AOC said that on TV and someone asked her, and I think John Carl asked her, well, like, what do you mean? What happened back then? Was it wrong? And she's like, well, it was the pandemic. And we were all locked up. And everyone. And like, she kind of sort of offered a little bit of the thesis, right? Which is we were all disconnected from each other online constantly, thinking out loud.

and also just sort of everyone's worst impulses were on display because of the algorithm, you know? It is funny that we had to spend so long talking about the harms of social media as limited to misinformation or extremism. I always felt like, like exactly what you're saying. I was like, I know this is much bigger than this. But that was the kind of allowable thing that you could point to and that you could say it was bad. But you could never say for a while anyway. You could never say the way that like AOC did in that clip that like, Yeah, it's affecting all of us.

Yeah. It's truly affecting everybody, not just the like white nationalist extremists. In terms of what we got wrong, I mean, we mentioned earlier how many times we thought that this is it for Twitter. I do think that like even more broadly, like the backlash is out there and there has been like significant changes. But, you know, I'm still a user. Maybe one thing I got wrong is how easy or doable it would be for me to use my phone a lot less. You did it. You've done it. I have done it. It does not stuck as much, but I have done it. The thing that I got wrong that I have really been thinking about a lot the last few weeks I already talked about this is I was too glib about AI. I don't think I ever used the phrase fancy auto-complete, but especially early on, that was not far from my assessment. And I just did not take seriously that it could ever develop the kind of capability that it has. And I'm thinking a lot about what it was that I was that I'm thinking a lot about what it was that. I got wrong or why it was that I got that wrong. I don't know if it was just skepticism about

Silicon Valley usual bullshit marketing high, people high in their own supply. I think so. I don't know if it was maybe not understanding the technology too. I didn't really understand how it works. Well, it was also around the time that they were trying to push the metaverse on us, which that's true. That's true. Which was. Which I love being on the metaverse. Which was awesome. Which was typical Silicon Valley bullshit, right? Like that was one where they were, but AI was not. Last question I have for you. So one reason we are evolving the show, and this has been sort of unofficial for a while, but, you know, there's some guests I have on and topics we do that are not strictly internet social media topics.

But I've also noticed that as sort of social media, the internet, now AI, has sort of infected every aspect of life, politics, culture. It is harder to draw a boundary between what is an internet office. line 1.0 conversation and what is a conversation just about politics or culture the way we live now? As we sort of officially take the show into a bunch of wider range of conversations, what is one question you hope that we keep asking? I think finding connection in the world and finding real meaningful and cultivating real meaningful human connection is something I've heard you talk about a bunch, I feel like, was like one of, if not the main through lines of the conversations that we had when it was like the two of us every week.

And I think that that remains an incredibly important question that's only going to get more important, not just with AI, but with general social atomization, with, you know, we see each new generation becomes more isolated, but doesn't want to be increasingly, but wants to kind of like find ways to reach out and to reconnect. So I think the rebuilding the idea of community and meaningful relationships, I think is maybe going to be kind of one of the big important projects of the next, I don't know, 20, 50, 100 years. If we have them, yeah. Right. If we have them, yeah, right. Or we'll just do whatever chat GPT tells us.

So it'll be one of those two things. But assuming that it allows us to have human relationships, I really hope that you continue to follow that thread because I think it's a really important one. That is an excellent piece of advice and a perfect place to leave this conversation, Max. As always, just wonderful to talk to you. You'll have to come back with Offline 2.0. And we'll keep talking about this shit. That'd be great. I will see you then. All right, buddy. Bye.

Offline with John Favro is a Crooked Media production. Our show is produced by Austin Fisher, Emma Ilich Frank, and Anisha Banergy. Our team includes Dilan Villanueva, Mia Kelman, Charlotte Landis, Eric Schute, Rachel Gaieski and Will Jones with support from Adrian Hill and Matt DeGrope. Our staff is proudly unionized with the Writers Guild of America East.