Squawk Pod: AI agents with Airbnb CEO Brian Chesky & Walter Isaacson – 10/01/26 | Audio Only

Bring in show music, please. >> Hi, I'm CNBC producer Katie Kramer. Today on Squawk Pot, Walter Isacson, he wrote the book on Elon Musk, and now he's looking at the leaders of artificial or super intelligence. >> We've [music] never had a technology that's seemed this dangerous since maybe the atomic bomb. >> Big existential questions, but this student of history says it's all about the people. Elon Musk really doesn't like Sam Olman. So, a lot of these things are personal, but that's the way history is. >> Then, an extended interview with Airbnb CEO Brian Chesy, who says, "Silicon Valley has an AI rhetoric problem. >> We use words like super intelligence. Super intelligence sounds like an alien species. [music] The average person does not want an alien species smarter than them. I think the average person feels like AI is happening to them, not for them." and how AI leaders can get on the same page maybe outside of Washington. >> I think the real work is not happening in front of a bunch of photographers at the White House. >> Those big conversations plus the latest moves at Google's Gemini, the new co-CEO of the Paramount Warner Brothers combo and weight loss drugs and [music] earworms. >> I just hope they don't buy that song. >> They will. They will. Foundo. >> It's Thursday, October 1st, 2026. [music] Squawk Pod begins right now. >> Stand back by in three, two, one. Cure, please. >> Good morning everybody. Welcome to Squawkbox right here on CNBC. We are live from the NASDAQ market site in Times Square. I'm Becky Quick along with Joe Kernan and Andrew Ross Sorcin. And here's a look at where the major indices year-to- date stand with everything you've been watching right now. Year-to date pretty good year for the markets still. S&P is up by 11 and 3/4%. The NASDAQ is the huge winner. It's up by 15 and a.5%. But even the Dow is up by 6%. But if you were looking at what happened over the last month, not so great. Over the last quarter, not so great either. Treasuries, the 10-year and the 30-year yields hitting their highest levels since 2002. This morning, we saw a big pickup. Uh, in fact, the 10-year was all the way up to 534 at one point this morning. We'll keep an eye on that. Right now, it's sitting at 531. The 30-year is at 565. There have been people speculating that we could hit 6% before the end of the year, and that'll be something to watch. The 2-year right now sitting right at 4.9%, but that 30-year mortgage hitting 7.6%. That is its highest level since November of 2023. And that's the issue that people watch so closely for the housing market, too. >> For the quarter, the S&P was up. >> Yes. For the quarter the S&P was, the Dow was down. >> Was up 2%. But after it just doesn't feel like it was up. >> No, it doesn't. I mean, for the quarter, the Dow was off by about 2.7%. >> But the S&P actually >> was up for and I think the Nasdaq was too for the quarter. >> Meanwhile, take a look at energy prices because that's been the issue we thought uh that was weighing on the market. The White House now telling Germany and France to release their emergency diesel inventories to help ease global fuel prices or face a potential US diesel export ban. That's according to Reuters. a separate report saying that Chinese refiners have suspended exports of oil products until further notice. That follows comments from President Trump at the White House yesterday as his administration continues to consider an export ban. >> I think it's something that we think about and we talk about every day, but it just seems that it would have a negative impact on gasoline. So that that would go up a little bit and diesel would come down a little bit and we think we're in a very good place. It's going to start to come in. According to AAA, the national average price of diesel this morning is $6.39 a gallon. That compared to just $3.71 a year ago. All right, let's take a look at Google parent Alphabet unveiling its latest AI model yesterday. It's called Gemini for Argon. The company said that the new model sets a new record in real world software engineering. Ties for first in cyber security and leads other performance benchmarks and professional tasks including finance and legal. Google says that it's already using argon in its data centers to optimize memory, freeing up hundreds of terabytes without buying additional hardware. The company will begin releasing that model to trusted cyber security partners and work with the US government on pre-release safety evaluations. But Bloomberg is reporting that some company insiders say that the model struggles with day-to-day tasks, including coding. >> It's an interesting one. It's going to be very interesting because it does on on so many of the different baselines, it seems to be ahead. >> And yet then you're hearing this news that maybe on a practical basis, it doesn't work the same way. So, we're all going to have to see what happens. >> Whoever wins AI wins. I think that's very important. But I think the technology has very real risks and you know the mechanism how we address those risks is still under discussion. But we all need to work we all need to work together to make sure that we can win and we can win safely if we if we do this right. If we work with the president and everyone here we can win safely. >> Tech executives questioned anthropic CEO Dario Amade at the White House on Tuesday about why he was being so extreme and public about the risks of AI. That's according to a Wall Street Journal report that said Nvidia's Jensen Wong was one of the leaders who questioned Amday. It said that Amade told the executives that it's important to be honest with the public about model capabilities and not play down risks. When Amade brought up safety concerns during the White House lunch, met as Mark Zuckerberg responded the best way to address them was following through on the principles of the tech leaders signed document on Tuesday, >> which he had a lot to do with apparently. >> Mark. Yeah. >> Yes. Well, this goes back to why why I was trying to ask Brad about this yesterday. That document was not something that came up at the lunch. That document had >> been worked on for days prior as it was being passed around by all of these. >> So, this was kind of a pressure campaign. You will sign this. It's it's only 308 words. All right. That's >> I think it's I just think it's fascinating as you know the undercurrent is >> you know you got Jensen and others and you got Amday and others and and they're both sort of saying okay we're trying to judge who's more who's promoting their self-interest more at at at any both of them are is one more than the other is is is it is it really sincere safety concerns that Amade is talking about is it really uh that Jensen thinks it is safe and it's not about selling chips. >> It it could be both. I mean a little bit of both. This could be a situation where they both are speaking their own truth as they see it where you know one thinks we could be h hurtling down a road thinks if you slow us down you are going to keep us from reaching our full potential. >> But in life if you don't like at least consider the cynical side of of anything you're talking about. You're just Right. >> Follow the money so you're not a duke. Right. >> Right. But I think there's a consistency issue that's just worth noting over and over again >> with both of them or with uh >> not actually with both of them. First of all, I don't think that Jensen historically has talked about safety. Safety was not like part of that conversation. If you go back in time, >> he didn't think it was necessary. >> He may not have thought it was necessary. All I'm suggesting is Amod >> has always >> Sam Alman, Greg, Musk. >> Well, that's scary because they know more than we know then. >> And and when you when you say Musk, then again, my ears perk up. But I'm saying that's what they've always been saying. This is not like some decades. >> This is not like a new phenomenon. And so >> science fiction writers been saying it since 1950, >> right? But what I'm saying is this idea that they had some kind of financial um incentive in 2015 when open AI didn't exist to say these things, >> right? >> To me suggests that you have to take it a little bit more seriously and not constant sort of this move. There's this constant uh idea that it's every it's in everyone's best interest. They were saying it before it was in their interests ever. >> Well, I'm back to the doomer side of of if if it really is as great as everything is, then no one's going to have a job. So, we could all die, but we won't have a job before. >> They were saying it before any of these companies existed and yet they still pushed the boundaries and they're the ones who have unleashed these things that these agents that have now had, you know, more than 10,000 incursions into other places. But that's but the I I'm only saying it that the people who are doing this were always worried about this always knew that if you're going to go down this path, if you're going to build this type of technology that there was always these risks involved and at least >> but they're also the ones who have unleashed these risks. >> Sure. But I would also suggest that I'm so much happier that we actually have people who are working on this who are talking about it honestly. I appreciate the question >> and the truth is had they not had the individuals who are running these institutions not been doing that there are lots of other individuals by the way in the technology community some of whom signed that document who would not have made these issues public at all but >> but it's not just those CEOs who are saying it's guys like Demis and others who have raised these concerns from the get-go not running >> and it's not like the other people are plotting and know full well we're heading towards Armageddon and and they're doing it anyway for money. I mean, they probably genuinely disagree that it's such a, you know, there's panicans and there's non-panicans. And in in everything we ever talk about, markets, pandemics, whatever, there's there's a way of looking at things where you're you're always scared, excrementless, and then there's you don't need to. I don't know, but we don't know at this Most of the public would have been much happier had some of the internal studies at Meta Facebook over the years around social media where they knew >> the issues and they didn't bring those forward. They didn't they never spoke about them. And why didn't they speak about them? Because it was not in their financial interest at all. I'm suggesting you can go you can go look at the leaders who signed that document or what they're saying and ask whether you believe that they were publicly honest with people over the in in the past about what it's done to about what it's done to our children in terms okay that that that's connecting the dots to where you know you could see bullying and and shaming and fat and and all the things that have come from social media I guess we should have understood because it's human nature but this is the end of humanity they had internal studies >> that it was going to do that to children >> that it was doing this to children that it was doing this to children and they never made it public. The point is these people you can say what you can you can you can say what you want about these people >> um having some other incentive. All I'm suggesting is that they're they are having problems. They know they're having problems and they're telling you about them. >> Yeah. >> Right. >> That's very different than what's happened in the past. Some are and and some are are less and and and and we should have a diversity. >> Although the latest things that we've heard from open AAI all came about because there was a whistleblower that left Anthropic that had worked at OpenAI who had worked at Open AI before. All of the last three weeks that we've talked about it was because there was an internal whistleblower. >> But could a whistleblower? >> No, no, no. The hugging face >> the hugging face wasn't found was not released by OpenAI. That was found by Hugging Face. They're the ones who >> you're talking about Coxin or or >> Yeah, I'm talking about Coxin. The reason all of this came out, Dario's >> open letter that he wrote was in response to a whistleblower, but a whistleblower that that's that's blowing a whistle and saying that we could all be dead by 2030, I don't even take that seriously. So, I mean, whistleblowers that know factually that something's going on is one thing, but saying in his opinion, what he's looking at could be the end of humanity in a couple of years, I mean, you take that with a not even [laughter] sure what we're arguing about. >> Right. Right. Well, that that's not a whistleblower. That's a I don't know what that is. That's a someone who that's hyperbolic panic. >> But he reflect he repeated. >> No one else really thinks that. Don't they do they >> about temporary >> but you have defined humanity differently in terms of it could be >> it's not >> there could be many people that die from some horrible situation but not everyone right that's that's what you were saying shut down the grid shut out if you shut off the grid right people >> you know heating in a in a >> right >> in a in a freeze >> David Ellison has hired former Mattel CEO Enan cries to be his co-CEO The merger between Paramount Sky Dance and Warner Brothers Discovery closes. Ellison remained chairman of the combined company. Cries worked closely with Warner Brothers. This is part of that partnership with Mattel to produce the Barbie movie. It's a fascinating move and most interesting to me the idea in terms of just title that David Ellison is effectively going to share the title of co-CEO with Enon. Um, if I remember, Jeff Shell was, you know, when he was in that sort of operating role, I believe he was considered the president of the company. I don't know if David Ellison was planning to share the CEO role, but maybe >> in terms of being able to bring, >> right? So, >> over you you needed to offer offer that role, but Kz did a did a pretty good job at at Mattel doing some pretty interesting things. stock. I was going to say the stock never I mean it had it had its moments uh but it never moved in the end. >> But the transformation of the company in terms of becoming an IP company probably the right way to go eventually. The these these toy companies have just gotten slaughtered. Tariffs have played a big role on that. The demise of of toy stores like Toys R Us have played a role in that. And Cry was pretty good at some of the accounting stuff that was taking place behind the scenes. It's just weird that Okay, so he helped with the Barbie movie. Okay. >> Well, it's not just the Barbie movie. They have a whole slate of movies that they [clears throat] put through, but yes, it's probably an unexpected thing. People will say though about the co-CEO roles. They'll point to Netflix where you've got Ted Sarandos and Greg Peters as being there, but yeah, often co-CEO roles don't last permanently. >> It's tough. >> It's tough. >> Who were those two guys at Croup? That was the original co-consul. >> Well, has a long though history. I mean, people think of him as a toy guy, but really, you know, he was in the production business um for a long time. Uh sold a maker at one point, so he he's he's been in this space and and done pretty well. >> And check out the shares of Eli Lily. Nothing happening there, but check them out anyway, I guess. Uh results from a phase three trial show that its weight loss pill found deo uh helped patients lose weight and uh reduce the risk of a heart attack, stroke or hospitalization. And I just hope they don't buy that song because they will they will they will found they will >> they're going to because they all buy those things >> and they plan them in your mind. Do you watch at night? What's how many of these not as much >> every pharmaceutical company >> drug has a soul >> you know if if this hadn't occurred to them you have now given them the idea >> if it happens consulting >> yeah if it happens I want money >> will be next >> coming up on squawk pod doom and boom the AI regulation debate with [music] author and biographer Walter Isacson he's looking at the builders of the newest technology >> they all talk about pdoom meaning the probability we'll have a total doomsday and they say [music] maybe it's 10% 15%. I said, "Well, why are you doing these things then? Why don't you get together and regulate?" And of course, competition. >> But wait, [music] I want to be the first one to kill the planet. >> This is Squawk Pod. >> Hop and Becky Q. >> Welcome back, everybody. President Trump is calling on tech giants to self-regulate. Joining us right now with his thoughts on how the AI race compares to similar events in history is CNBC contributor Walter Isacson. He's of course an advisory partner at Pella Weineberg, a professor at Tlane and the author of Elon Musk among many others. Um Walter, it's great to see you today. >> Good to be back. Megan, >> you're an incredible student of history. What you must be watching this with an eye to what's happened in the past. >> Well, absolutely. And we've never had a technology that's seemed this dangerous since maybe the atomic bomb. But whenever we have these things, we have uh new groups that get together, including industry groups, like we saw earlier this week when Trump brought them together. But we also finally figure out regulatory regimes for them. I've got this series coming out soon called The Builders, which is a podcast series. Every one of these people is in it, and everyone thinks this could be dangerous. They all talk about P doom, meaning the chance that we'll probability we'll have a total doomsday. And they say maybe it's 10%, 15%. I said, well, why are you doing these things then? Why don't you get together and regulate? And of course, competition. Competition between themselves makes that impossible. Competition with China makes it >> Joe kind of said this tongue and cheek yesterday, but wait, I want to be the first one to kill the planet. >> Yes, exactly. I mean, what what is your goal here? One of the things that was interesting, I'm watching uh uh Mustafa Soleman who I'm interviewing for the builders. He he asked the basic questions which is what is the purpose of building this? Well, the obvious answer is to help humanity. Now, if you're doing something that seems so uh uh likely to cause harm in some ways to humanity, you probably ought to say, "Let's figure out how to regulate it." So I didn't say the plan the the planet. I think the planet will be fine. >> The planet. Yeah. >> Be like a be like a neutron bomb, right? >> Exactly. Well, yes. Exactly. >> That only killed people. Left the building. >> Left everything else standing. Um you you've talked to all of these people in the builders. I I I guess you went to Sam Alman, Elon Musk, U Deis obviously. And Dario and Danielle are together. a day >> and and the the things that they actually agree on. Every single one of them said that there is a less than zero chance that this does something to wipe out humanity or some portion of humanity. >> Oh yeah, they believe it's a non-trivial chance. They I mean there are two things to worry about. One that it just really unleashes something. I think hugging bear for me. I was on this show right when hugging bear happened. >> Whoa. This is a hugging face. Hugging face. That's a turning point because I was an AI optimist. So I think that became a problem. So it could do something like now that we're trying to create these autonomous agents that get more and more autonomous do things without our control then they're going to be out of our control. The big difference in history between the personal computer revolution which started in the 70s and this one is that with Steve Jobs, Doug Engel, Bart, Alen K, whatever the personal computer was supposed to connect you more closely to your computer. This is supposed to create agents that are disconnected and do things without you. That is not actually a great idea. >> What do you Oh, go ahead. >> I was just going to say what you think their motivations are. I've argued that these people have been relatively consistent since I've known them about this issue of safety. But then you have other people Jensen was sitting here earlier this week who said these safety concerns are are are crazy that that that effectively, you know, this is an engineering problem. Mark Zuckerberg would tell you there's liability laws in place already. So, you know, you should shouldn't be worried about these things. >> Yeah. I mean, what we saw at the White House summit is there's a group and David Saxs is a leader of it uh who are very influential and say, "Look, let's not worry too much about this. Let's not lose out to China. We have the FTC, the Department of Justice, liability laws that can regulate it." But to answer your question, yeah, for 15 years, I mean, Elon Musk got into this game, Sam Alman got into this game, Dario got into this game because they feared that a runaway uh artificial general intelligence could be very dangerous. So, this is not some jenned up let's have regulatory capture thing. This is a real worry. Let's not throw it away. Elon Musk wanted the type of agreement that we saw coming the idea of them kind of regulating each other and watching each other. Do you think self-regulation works here? >> I think that's a good idea and I think uh Musk has wanted that for quite a while. We talked about it on the builders which is embed people from the other companies in your company so that we can figure out if you're doing something really bad. It's like checking each other's homework or something. That's probably a good idea. Is it a totally sufficient idea? No. Because even when open AI was creating its own thing, it escaped its own sandbox. So I think we need uh probably more than that. Dario Amade coming out of that uh White House summit said this was good but it's not sufficient. And then you had the other camp saying, "Hey, this is good and this will probably help handle it." >> The idea of allowing each other in um just from the competitive kind of juices that would be flowing with that. >> It's not only competitive juices. I mean, let's get straight here. And one of the things we did in this builder podcast is this is a personality thing, too. I mean, Elon really doesn't like Sam Alman. By the way, Dario doesn't trust Sam Alman, but even Demis Hassabis and they've just released the new Gemini. You know, he had a falling out with his co-founder of Deep Mind, Mustafa Salem. And so, a lot of these things are personal, but that's the way history is. >> I think back to Oppenheimer and all the people around him. >> Yeah. You know, but these are all larger than-l life people playing with larger than-l life technologies, and they happen not to be very friendly people to each other these days. So the notion that they're going to go into each other's company, have employee level access and say, "Oh, this is what you're doing wrong. Trust me." Or I, >> you know, this is not really the long-term solution. >> Walter, as a genius in history, are you familiar with this guy Samuel Butler, uh, English author? >> Yeah. Well, I mean, >> you are a while. >> 1872, >> the book of machines, >> right? You know in 184 >> argued that machines were evolving far faster than the biological life and would eventually develop their own consciousness and humans would become entirely subservient and it was >> in the 1840s and I have this book called the innovators. >> I thought it was Isaac Azimov apparently not >> in the 1840s you had a love lace famous Adah Love Lace the first Lord Byron's daughter who created the first computer program. She was the one who understood how digital zeros and ones could uh convey all information. You had the lit smashing the looms that were using those punch cards in the 1840s. You have Samuel Butler. You had this whole same argument. The main argument on the singularity which is AI will happen right >> starts with Allan Turing you know and the imitation game but then his colleague J good comes up with the notion of the singularity or machines will keep improving themselves >> that's a ways off someone said earlier today that it I think it was Brian Chesy said it >> yeah in interview said that the the so-called uh AGI singularity is still how far do you think >> the AGI singularity may have already happened. I mean, in other words, we may be getting artificial intelligence that and general intelligence. >> I thought we weren't close on recursive self-improvement yet. That that cat's not out of the bag yet. It's just AI aided. >> No, I think we're seeing recursive self-improvement, machine learning, >> not to the point. >> And it's not yet at that hockey inflection point where it just totally takes off. >> That's a bad analogy to use. >> Yeah. Okay. >> The hockey stick. That's why Bill Gates no longer even talks. Now he's worried about something else. And you're going to ask him about climate change while >> I'll ask him about fears on AI because those are good. >> I know. >> Walter, let's talk about something we've spoken a lot off air about and that's just the idea that you know the rules in the in the internet, the rules in the digital world should be the same as the rules in the real world. >> Absolutely. And and maybe that's where we like how do we get to that point where >> well one way we get to that point which is uh the rules in the physical world are if you make something really harmful and it goes out and destroys things you're going to be sued. Now that's something that at the White House summit all the people at the White House said well we already have a product liability laws. Well, I'll tell you when there 10,000 product liability lawyers putting up billboards in your state of New Jersey saying if you've had some scam, you had your bankers talking about the scam and you've been hurt by AI, call 1800 sue AI company. I think they'll the AI companies will be begging for. >> So, it is going to go through the court system because right now the Department of Justice, the FBI could be the ones who are following up on that. We know there have been 10,000 different accounts where agents have broken in and done things they weren't supposed to be doing. Nobody's really gone after that yet. There was the first lawsuit, I think, filed early. >> Yeah, I think the lawsuits are starting to be filed. It's kind of hard to know, but there's no there was legal protection, say, for Facebook if somebody posted something about section 230. It is clear that uh protection doesn't apply. >> It's unclear what the look, the issue, and I think why it's going to be hard for some of these suits to win currently is it's unclear what the damages are, >> right, >> from these from these episodes. So whe when these episodes happen, you have to prove that there was damage to you in some way. >> I think in the c the case that was just filed this week by the nonprofit organization out in California, they're citing violation of a California rule. Um I mean that's a good point, but there's going to be damage. >> Oh no, I'm saying there eventually will be. I'm just saying that a number of the examples that have come out for example around open AI were you know doing something unexpected with you know the education department's website right >> it's unclear whether getting on that website unto itself >> created a damage all your data >> but if they hack all your d No once you get into a privacy situation it's going to be different >> is worried just about the hacking of data I'm worried about all the things that could be unleashed from you know boteterror weapons these >> and how close do you think we are on that? I mean, I'm I'm much more concerned about the grid, the electric grid, which is something you could touch today. >> Bioweapons, I think, are a little harder. >> Look, the grid air traffic control, bioweapons may be hard, but you can create I mean, let me go back to the good side of AI. What these companies should be focusing on is things like protein folding, which won the Nobel Prize for Demis, and saying, "How can that help cure cancer?" Right? >> Uh instead, they're creating these weird autonomous agents that go floating off on their own, uh go using your bank account numbers to make restaurant reservations for you. I mean, how important is that to society to have some autonomous agent who can look at your calendar and cancel a restaurant reservation? No, that's not important. We should be curing cancer and doing some real things, even going after your favorite topic and helping with climate. [laughter] >> Yeah, we'll be fine. Uh, that I'd rather do the restaurant stuff. All right. Uh, Walter, thank you. Um, are we done? >> Yeah, we are. >> Rabbit. Rabbit got >> coming up next on SquawkPod. Airbnb is leaning into artificial intelligence with search, property comparison, and social features. CEO Brian Chesy details just how AI native his company is becoming and how it's changing the nature of travel and of work. >> I am extremely bullish that ultimately AI is going to create a lot more jobs than it destroys. My concern is that there will be this valley [music] where maybe a lot of jobs will be destroyed before the new jobs are created. >> How his job as CEO has shifted and the long road to [music] reskilling, plus agents, accords and much more right after this. Welcome back to SquawkPod from CNBC. I'm producer Katie Kramer. This week, Airbnb unveiled new features both online and on the ground at bookable properties. Users will have AI search, [music] AI property comparison tools, dynamic pricing recommendations for hosts, and even grocery delivery and baby and sports gear rentals, all directly in the app. It's part of Airbnb's [music] overhaul toward becoming an AI native company. And it's inching towards CEO Brian Chesky's goal of being what he calls quote the Amazon for travel and living. [music] Our Andrew Ross Sorcin caught up with Chesky to discuss the news and also what it means in this whole 2026 world of AI [music] fear and excitement. Here's that interview. >> So, how do you think about this in the context of AI agents? Over the past couple weeks, we've now seen a whole number of new agents, whether it's Meta's Muse or Instinct or Dots out of Chat GPT emerge. And there's now a question and you and I have actually talked about this for a very long time about the the UI, how people are going to interact with all of these services in the future and whether there's going to be sort of one agent chatbot feature that everybody uses as their default that then goes and effectively crawls everybody else's sites and does stuff on their behalf. >> Um unclear if there will be one agent. If there is, it's probably not an app. it probably will exist at the operating system level. Um, but I also think that I do think agents are the future and chatbots are not. And I think right now the interface for an agent is a chatbot. And so it's kind of like taking a jet engine and putting on the back of a of a bicycle. Um, agents are not chat bots are not good for shopping. If [clears throat] you tried to book a hotel or something on Instinct or Muse, you'll notice it's not yet a good experience. Now, it is really good for like getting paper towels or something that's much more of a commodity, but the problem is you want rich features. You want lots of photos, you want comparison tools, you want filters. So, we still need an interface that's more than a chatbot that is an agent. So, imagine an agent that is a full branded experience. Um, brands are going to want to be able to feel like there's an software developer kit and your agent doesn't just find things and bring them to you, but they almost bring the store to you so you can see and be able to compare. Um, the big question, Andrew, is is there going to be some breakout consumer agent that becomes the front door to the internet? I don't know. Um, that [clears throat] agent would have to be the equivalent of a full operating system. They'd have to be the equivalent of a super app. They can't just be like a chatbot. And if somebody built that, maybe they could be the front door to the internet, but it's not going to be what we've seen yet. >> How critical though is it for you at Airbnb to be the front door for all of the various services that you're offering rather than somebody else's front door and potential agent going and crawling your site. >> I think it's really important. Um, you know, I have I I have a friend who worked at Netflix once and he said back in the day when Netflix was a streaming app before they had original content, they had a mantra. We have to become HBO before HBO becomes Netflix. In other words, Netflix had to make original content before HBO had a streaming service. So Airbnb, we have [clears throat] to build a really good agent before these one-sizefits-all agents develop enough user interface functionality that suddenly, you know, we don't really have a a purpose to be a destination. Um, so I think that's one point of the answer. Um, I don't know if we know there will be one agent to rule them all. I definitely don't want to say there won't be. I'm not convinced there will be. I think people are going to probably have a lot of different types of agents. How do you feel though about potentially agents searching your site, effectively potentially bringing people to your site and maybe even transacting with you, uh, but doing it potentially without a human in the loop depending on the situation. I mentioned this because Muse, for example, has now been blocked by Amazon and there's a discussion about whether there's going to have to be a negotiation between every agent out there and every storefront. >> It's a really good question. Um I think that we'd be totally I mean chatubt has really helped Airbnb. A lot of the traffic that comes to Airbnb um from chatubt converts better than traffic at Google. So I think there's a scenario where traffic that comes from agents could be really good for Airbnb. The problem is if the agent books on behalf of the user, you know the ho if I'm a host, I want to really be able to communicate with the guest. 90% of hosts communicate with the guest before booking. the guests need to have, you know, profiles in Airbnb. So, as long as I still feel like I'm talking to that person, whether they're between agents or between actual users, you really, you need to make sure that there's still a deeply human connection. So, that's really what we would focus on. Can you ever imagine a day and do you even consider it a risk that everybody has an agent including all of the hosts that have properties on your platform that effectively decide to host themselves and that the various agents are connecting with each other's agents and therefore are able to disintermediate anybody who has a marketplace. Yeah, I think that's similar to what if everyone in the old world had an app. And I still think if everyone had an app, then suddenly they still have a discovery issue. You still have a trust issue. Um, you know, everyone could have built an app just like before Airbnb, everyone could build a website, right? We we didn't invent the idea that your home was on a website, but we basically were able to collect and aggregate all of them. We were able to build a trust system that was uniform. We were able to provide $3 million of insurance like uh equivalent for protection. We're able to provide customer service. We're able to provide all the tools for all the host. So actually I think that is where it's going, but I think we're going to be the ones to build that host agent and that host agent will be having an intermediary Airbnb agent. But you're still going to want this trust system. I think >> um you're in the heart of the valley. You've made some provocative comments over the years with us and others about the culture in the valley and what's happening in terms of just the rapid progress. I'm so curious about where you land uh about some of the discussions and headlines we've seen over the past couple of weeks as it relates to AI, AI safety, uh the idea of, you know, humanity being uh at at risk in all of this and others saying that there's a sort of a regulatory capture effort underway by some of the big model makers. >> I um I I I know most the founders. I think I mostly take them at their word that this isn't primarily this is not about regulatory capture and by the way if this was there's a very simple solution which is you could regulate only the large companies and allow startups to get to a certain scale before regulation that would essentially be a pure antidote against regulatory capture. Um I do think there should be something um you know I think Bill Gates went on and said that you know pharmaceuticals have basic kind of regulation that you don't want to wait for people to sue you. Um, at the same time, I think the technology is really early. Um, I think that the timelines for the um for the doomsday scenario are probably overstated. I do think that cyber attacks are probably not overstated and we got to take this really really seriously. The most important thing as we as a valley, we need to get we need to get on the same page. Almost every other industry, the CEOs are all willing to get in a room together. and in our industry, unless they're meeting Trump at the White House, they're really rarely ever willing to get in a room together and even talking this out. I think that's like the first problem is that we need to like kind of get all on the same page about and share information about like our strategy. So, I do think that with great obviously power comes great responsibility. This is a cognitive revolution and I think that we need to basically take the language from doomsday down a notch. I think that the way we talk about AI in Silicon Valley is really bad. Um, we use words like super intelligence. Super intelligence sounds like an alien species. The average person does not want an alien species smarter than them. I think the average person feels like AI is happening to them, not for them. There's very few consumer applications in AI that are actually improving people's lives. Most AI is enterprise. People hear that as job automation. Um, you know, it is we the anti-AII is unfortunately becoming a bipartisan issue. You would imagine all these red states should probably want data centers who create a lot of jobs like electricians, but the public is afraid. And the reason why is not just the rhetoric, but there is a massive divide between Silicon Valley, which is a giant bubble, and we are in such a big bubble that we don't even realize it's a bubble, and all of society. And I think we in Silicon Valley, we talk a lot about humanity and like this abstract, but we don't talk about people like a regular person in Ohio has a child who cares about the affordability and like how is their life going to be better. So I think that we have a major problem of disconnection between Silicon Valley and society. And if this is not corrected, I think post midterms, you're going to have a whole bunch of presidential candidates come out. They're going to have to have policy positions. They're going to have to have positions on AI. And my worry is their positions are going to be very, very negative. And so I hope we in Silicon Valley, this is a wakeup call to us that we need to really bring society along on what we're doing. Otherwise, if people feel like something's happening to them, they're going to fight back. And I think that's what's starting to happen. >> I got to unpack a whole bunch of things you said. At one point, uh, you mentioned the idea of being in a bubble. I don't know if you were talking about that in the context of being in a bubble culturally in the valley or being in a bubble as in a market bubble >> or maybe both. >> I meant cultural bubble. Um um I meant that we are in a cultural bubble disconnected from society and a lot of us are on Twitter. Um but Twitter is still a very small microcosm of the public sentiment in real life. Um, now with regards to market bubbles, um, I think two things might be true. I think, um, I I mean, I have a kind of a radical point of view, I suppose, Andrew. I don't think we're in a like window or cycle like the internet. The internet was like a window. And in that original window, you had a few big companies emerge, basically Google and Amazon. In the original computing era, you had Microsoft and Google. And then in the kind of social networking era you had in mobile you had you know meta and I mean us and Uber were like smaller versions. I think now you see anthropic and open AI. My kind of radical view is this is unlike anything including the industrial revolution. Um if technology doubles every year for 10 years that's a 1000x we're like the beginning of a thousandear cycle. I think you could see 20 30 40 more anthropics. But that also means I think there's way too much money going to companies without products right now. I mean, you have researchers with like three employees with no product getting $5 billion valuations. Um, and they don't even have a product and their money is going to GPUs which are very expensive that are appreciating. And I I I worry like what is the business model? Are these going to become like very expensive aqua hires and are there enough buyers for all these companies? So, I think two things can be true where we might be underestimating the overall size of the AI opportunity, but we're kind of maybe plowing too much money right now into too many dreams that won't materialize. >> You referenced the president getting everybody in a room to try to get on the same page. A number of them signed a page pledging um how they would try to approach the safety issue. I'm curious whether you think that's enough. >> I think we need way more detail. Um, we need and and and I'm not I I don't know enough to like even have a thoughtful opinion about this, but the stakes are high enough that I just think everyone needs to come at the table and have meaningful conversations. And I think the real work is not happening in front of a bunch of photographers at the White House. the real work is going to be you know researchers working with the CEOs to really think about like what are the risk and I think that maybe the simple way to do this is to not try to solve the next 10 years like this notion of the singularity is still quite a ways away um the immediate risks are like cyber security like let's start there cyber security what should we do about cyber security because that is a clear and present near-term danger um you know using this for boweapons that's going to take a bit more time. So I would imagine not trying to have some grand theory but just one step at the uh you know every 3 to 6 months identify what are the really big practical risk. Um and that's one of them. I think the really really big topic we're not talking about is job displacement. I am extremely bullish that ultimately AI is going to create a lot more jobs than it destroys. My concern is that there will be this valley where maybe a lot of Joabs will be destroyed before the new jobs are created and do people want the new jobs or will they be prepared for the new jobs? So there's going to be some like massive societal education. Now we've gone through this with the industrial revolution that happened over a hundred years. Are we prepared to like retrain 300 million Americans over the next 10 years um if they don't even want data centers in their states? Um, I think that, you see what I'm saying? I think there needs to be a lot more education. We need to make simpler tools for people. I do not think we are preparing society for what's to come. And society is absolutely not, they're not prepared for what's about to happen. And we all just need to come together. I'm not that person. Um, I hope people with much more stature than me can be the glue that can bring tech and society together. I I really think that's important and it's really not happening. >> What's the latest in terms of how you meaning how Airbnb is operating as it relates to AI and is it changing um in terms of your jobs uh how many people you're hiring, firing, keeping on? What's that look like to you now? >> It is incredibly disruptive. Um and not necessarily bad disruptive, but like everyone's job is changing. My job as a CEO has changed. I mean from the time I wake up to the time I go to bed my job is totally different than it was 6 months ago. uh six months ago the typical job of CEO was to basically have meetings to figure out what was going on your company right like how's the cos going company you have meetings now suddenly you have an AI that can give you all this like information so suddenly like there's a whole bunch of people whose jobs it was to give me information their jobs are changing now suddenly I have all this information what do I do with this information how do I mobilize people you know almost nobody next year at Airbnb will be writing any lines of code. The idea of writing code is going to be probably gone completely next year. Essentially syntax coding. >> No, no code. >> No code. No one will rewriting code. Now they'll be prompting code, but they won't be handwriting code. That is almost gone. And I think it will be completely gone next year. Now engineering won't be gone, but suddenly what they did is totally different. So we now have to retrain thousands of people to do a different job. I don't know how much you can speak to it, but um I know you've been working on your own uh separate AI project. Can you give us an update? >> Um I would love to come back on the show and maybe do a dedicated session on that. Um what I will say is um I know a lot more about AI than I did 9 months ago. I I'm a designer, so I'm not going to masquerade as a researcher, but I've probably met with like a hundred researchers um in the valley in the last nine months. Um, I'm really really focused on consumer AI. Um, I've been kind of, you know, quiet, I've been loudly pronouncing chat bots are not the right interface for agents. So, I'm hoping to um try to maybe further that conversation so I have more than just rhetoric. And um you know when I talked a little bit more before about the bubble, the financial the investment bubble, you know, one of the things I'm seeing is a lot of researchers raising a lot of money to start companies. And I think my advice to investors if they're investing in researchers is make sure the researchers want to be entrepreneurs and you're not just giving them a bunch of money to do research for a couple years. It's a very kind of wild time in the valley. Um, but I'll give you updates when we have something. [music] >> Brian, thank you again. Appreciate it. >> All right, that's Squawk Pod for today. Thanks for listening. Squawkbox is hosted by Joe Kernin, Becky Quick, and Andrew Ross Sorcin. Tune [music] in weekday mornings on CNBC at 6 Eastern. To get the smartest takes and analysis from our TV show, Right into Into Your Ears, please follow SquawkPod wherever you get your podcasts. Have a great Thursday. Happy October 1st. We'll meet you right back here tomorrow. And we are clear. Thanks guys.