Airbnb CEO Brian Chesky on early rejection, customer focus & AI’s future in hospitality | E1735

26:27And remember, faster inputs equal faster outcomes. And velocity is how startups win. so go sign up for Miro for free at miro.com slash startups that's miro.com slash startups to sign up for free it really is like this modern conundrum that google finds itself and not to single out google i mean any big company has this i think zuck had it as well which is it's just too many people working on too many different projects there's no cohesiveness and then quality goes down and then the consumers are like am i supposed to buy a pixel or some other android phone am i Am I supposed to use Chrome operating system or the Android operating system?

27:04Which one is going to win out here? And then I'm a consumer. I have a Nest and then I have Google Home. And just that one cognitive dissonance I have, like where I open up Google Home to see all of my cameras and I open up Nest to see the old cameras. And I like my Nest better. And it's just chaos, right? And Sundar's got to be in the middle of this trying to referee it. And if he did the same approach. And I don't want that life. I had that life before the pandemic. And I said, I don't want that life. I don't want the life where I have to ask permission to get involved. I'm adjudicating disagreements.

27:35I'm managing bureaucracy and dysfunction. I want to be in the product with people making the product. I want to be talking to customers. I want to be communicating with them. And I want as little bureaucracy as possible. And that requires us to do as little as possible. There's an old saying, have few details and perfect every detail. And that requires focus. It requires you saying no. So we probably say no now more than we say yes to things. And I think that's pretty important. Yeah. The second business line really was experiences. Am I correct? Yes. Stays and then experiences. Yes. And is there even a third business line?

28:11No, no. It's really just, you know, probably broadly, like there's stays. 80 % of stays are short term. 20 % are long term stays defined by longer than 30 days. So that's kind of like an in between. It's an extension of the core. And then experiences, which has been kind of like an in between zone, because it was, we thought it was going to break out before the pandemic. And then we had to actually shut it down or pause it. So it's, we're in a process of we're going to reboot that. And we do, and just to be clear, we have some big ideas coming. We have some really huge ideas that I think will expand Airbnb way beyond travel, way beyond our core.

28:47It's going to launch next year. But I'll tell one more story. You know, I feel like companies, you know, like there's certain companies you like root for them. like a lot of people love their Tesla car and they like want Tesla to come out with a truck. A lot of people love in 2006, how many people wanted Apple come out of the phone? A lot. And the reason why is we all loved our iPod. How many of us love wanted Gateway to come out with a phone? Probably not many. And so I like to tell people, we have to have permission by the customer to do something new. And we only have permission to do something new if they love what we currently do.

29:21And if they're complaining on Twitter about upfront pricing, about having new chores and to check out about customer service taking too long, then we don't have permission. They don't want to do something new because they're going to think we're going to bring problems to that category. So I told the team, I said, we have some big ideas. AI is going to unlock so many of them, but we don't have permission to do new things until people love our core service. So we're going to basically create a blueprint of every single thing people are complaining about. So we created this storyboard of the end-to-end experience.

29:50then we got a hot we i got basically blueprinted out with the team all 150 screens on the app every user policy all 72 user policies now each of these user policies are as many as 100 pages long right like who gets a refund when and imagine jason you're a call center worker airman just hired you now like two customers in different languages are in a dispute and you're adjudicating between 70 policies that are as many 100 pages long that's a perfect example of ai Anyway, this is an opportunity where we realized, okay, we created this blueprint. And then we mapped tens of thousands of social media posts, millions of customer calls, and all these interviews with customers.

30:31And we put them on the blueprint. We looked at what is everyone upset about? What do they want fixed? And based on this blueprint, we were able to go stage by stage in the end-to-end journey and prioritize the top issues. And we said, we're going to get through every one of these issues before we do something new. And that was the whole premise of today. It was basically fixing what people wanted. And then we had one more thing. We're like, we got it. We have some unfinished business. This idea of people staying with each other in their homes, the original idea of Airbnb, it's kind of like the Volkswagen's Beetle or Nike's running shoes or Apple's iMac or whatever the original car, original product.

31:07I said, we got to revive that. We can't, we got to get back on the highway. We got to like reinvest in this core idea before we do new stuff. And so that was kind of what today was about. And hopefully, if we're successful, people feel like we listen. Let's talk about the user complaints and prioritization of them. Because how do you know when you're running something at scale like this, that this is something that's annoying but necessary? Like, you can't throw a party in an Airbnb. Yes. Like, I had my old home, which I was thinking of keeping, and I had an Airbnb. And only one time in a year was there a party thrown there.

31:45You guys handled it great. like there was a little bit of damage insurance picked it up um but you know like over and over again it has in the documentation do not throw a party here do not throw a party right you can't throw parties so how do you know when something is like you know just this is a deal breaker for people who are hosts this is a deal breaker for people who are going to choose a hotel or an airbnb to run an airbnb or not run an airbnb as a host and these are like minor annoyances but they come up frequently so there's like a certain amount of suffering in each thing right yeah and i'm curious how you prioritize the suffering yeah to alleviate it because you're also mitigating and arbitrating between these two groups of people the hosts who can't be burdened so much that they don't want to be a host because it's arduous and then the people who are leaving the home who don't want to have to do an hour of chores when they leave like it's reasonable put the dishes in the sink but do you have to mop the floor or take out the garbage okay maybe take the garbage that was okay?

32:46How did you prioritize that? It's both an art and a science. I think that analyzing it as a science and then prioritization is probably more of an art I can explain. So let's start with analyzing. The first thing we do is we try to just take all the inputs. So what are the inputs we get? Inputs are customer service calls. Inputs are social media posts. Inputs are we do like tens of thousands of like we do listening sessions with guests and hosts. And then inputs are like basically user behavior, right? You can like see what people are clicking, you can see retention, you can see when somebody churns.

33:21And then if there's a party, we don't handle it, right, they churn, they don't come back. And then we see people in their network don't list and you can start to measure that. So those are your inputs. Those inputs, you basically now have like an organization of like maybe millions of issues. And you can bucket them to like, say 100 200 types of things. The next thing you do is you look at the frequency and the severity. So severity, we might be like a safety issue, very infrequent, but when it happens is really serious. A frequent issue is like, I'm upset with a refund, not as severe, but very frequent that happens all the time.

33:56And then you then you start to look at relationships like, you know, we notice a lot of hosts are complaining about pricing be confusing. A lot of guests are complaining that hosts are charging too high fees. And then we start to realize, wait a second, if we make it easier for hosts to price, and they understand what guests are actually paying, they might actually do a better job pricing and guests won't complain as much. So you start to find like connections. So this is where that's the science part. Now the art part. Now it's actually picking stuff. Because you have a list, you have some matrix like severity, frequency, this and that.

34:34But the real ability is the art form of a group of us, 10 or 20 of us deeply understand the issue so well. We live the product. We use the product. I host, I've read thousands of things. You just take thousands of inputs, you put them in your head, and then you just go through and you can't paint by numbers. You can't be purely algorithmic is what I would say. That's why that's the art. You could, but I think it's never as good as your intuition, but your intuitions form all these data points and the intuition might be like you know it also might involve like engineering capability like this is going to be a really heavy lift so we have like t-shirt sizes a small lift medium lift large lift extra large lift so we'll prioritize high severity high frequency like low engineering lifts for example be a great quick wins quick wins yeah that with high payoff there might be a big lift but we need to do it to launch this feature later so so basically, there's like hundreds of inputs, you see how like, you're like weighing all these in your head.

35:36And I think there was this, like, I think developing a product, the best products, I feel like are a group of people that are deeply involved, and they can hold 1000 contradictory things in their head. And they can make all those different trade offs. We call that intuition. But I don't, I don't know if I love that, because it makes it seem like it's arbitrary, and not systematic. It actually is very deep and it can potentially be very technical. And so that's what we do. And I audit all the prioritization. And I personally decide on the final prioritization of every single project in the company that a customer will ever see.

36:12I don't decide, I mean, the final decision, I make the final call. And I only make a call if I'm informed. And I only am informed if my team understands. And so I make sure that I know the details, My directs know the details and their directs directs know the details. I also have a rule that if I have a direct report in a meeting, they can never call their direct to answer a question. If that's the case, they don't know the details and their scope is too big. So, I mean, within reason. So, that's kind of how I do it. And then we update it twice a year. So, we have this thing called the roadmap review.

36:44We bring in like top 70 or so product people in the company or people weighing on the roadmap. We spend a couple days together. We debate everything on the roadmap. app, we adjust it, and then we roll it out. And then we have an extremely robust program management function. Most companies ask product management to basically be their program managers, we separate product management with program management. And we have a very small product management functions, very small. We have a product marketers, product marketers are senior product managers. So we have no outbound people, you have to know inbound and outbound, the junior product managers don't need to do outbound.

37:20But then most of what people call product management, we call program management. And they're the ones keeping the whole thing on trains because everything has to fit together. And they're doing like really rigorous reporting every week about like where we are at the project. So it's, it's all integrated. And the net result is Jason, is I can literally know the performance of an individual engineer I've never met on a week to week basis. Because imagine if we're all designing a car, I know how good the tire team is doing when I see the car assembled every week and I see, hey, there's something wrong with the tire.

37:50But when they're like 100 different products, I don't have time to evaluate 100 different things. Yeah. And this is where constraint comes in. If you want to make great art, as you learned as an artist at RISD, like if you give some artist, you know, a thousand canvases and buckets and buckets of paint and a million brushes, like where do you begin, right? And how do you prioritize? You just get lost. Creative people want constraints. And that's the other thing. And I think what happened, and maybe this is a burger lesson, Jason, in Silicon Valley is I think in the last 10 years, we as entrepreneurs in Silicon Valley probably didn't have enough constraints.

38:23And it's kind of like we didn't ask for constraints. We didn't want constraints, but we probably needed them. I think that Airbnb before the pandemic probably raised too much money and we didn't raise as much money as others. And I think that in hindsight, like not having as much money means you have constraints. Constraints means you have to say no. You have to make hard decisions. If somebody's not performing, you can't just hire more people, you got to deal with it. And if a product's not succeeding, you had to actually fix it or sunset it, you know, so I think the constraints are really critical.

38:53Listen, we work with super early stage companies at my investment firm launch, you know, pre series A, maybe got a couple of thousand dollars a month in revenue, you've raised a couple of hundred thousand dollars, maybe a million dollars, right? It's the early days, year one or two of a startup. And I'll be honest, a lot of times startups, they don't have their insurance. They haven't set that up yet. They haven't set up their accounting properly. They're getting things cleaned up. In fact, I was recently had a great startup, but they didn't have D &O insurance. That basically protects your directors and officers.

39:25That's the D, directors, people on the board, officers, the people who work at the company, right? Directors and officers insurance is super important. So what do we do? We send them right to InBroker. our friends over at and broker are a business insurance company that's built specifically for startups. You just fill out a simple application, right? And then startups get four quotes for four lines of coverage in 15 minutes, four quotes, four lines of coverage, 15 minutes, easy, breezy, lemon, squeezy. That's right. They connect you with one of their expert brokers for unmatched service that goes beyond your policy.

39:59And listen, you might think, oh, it's too early to have insurance. It's not that expensive. It's not that complicated, because in broker makes it easy. So here's what I want you to do. Try and broker today with the code twist. And you'll get 10 % off their startup package at in broker.com slash twist. That's in broker.com slash twist. We love in broker, I use in broker, they're an amazing team. They do a great job for startups, whether you're in year one or year five, go use in broker.com slash twist. The zero interest rate policy led to a lot of lack of discipline and maybe you avoid difficult decisions like hey this division's working out okay but it's not great yeah it's never going to be profitable and you watched uber you know you guys had your media stuff you were doing for a little while um which you were passionate about i remember but like at a certain point it's about the guests it's about their experience and how many different projects could uber do they were doing micro mobility and that was interesting but it was just such a minor unprofitable piece of the business they deprecated it right yeah if people really want to get on scooters like what percentage of people in what percentage of cities when you did this cycle what do you think are the problems you solved that were the most rewarding for you because when you solve a problem so when you say immediately no you say cycle you mean which time horizon well the one you just did where today the release oh today yeah this release there must be things that you personally as you did the prioritization said hey these three things these are the big wins And now you put that tweet storm out with the images in it showing each one, which I love, by the way.

41:35Oh, thank you. When you get it directly from the founder and it's like, here's the image, here's what I was thinking, let me know. How did the public's early response, your host's early response, your customer's early response to what you thought the top three things were versus what they think the three things are? Yeah, so the things I prioritized was probably number one, anything around pricing. The number one complaint Airbnb is that the prices have gone up. And we started as an affordable alternative to a hotel. And Airbnb's had become over time less affordable. And I think we just like, I mean, that in itself was a potentially zero interest rate kind of phenomenon.

42:13And so that was the number one. It was like pricing transparency. The second one I focused on because I just was so tired of hearing about it and I didn't want people to deal with it was checkout chores. There was this, you know, turn the lights off and you'll leave is totally reasonable. Strip the bed and put the sheets in the laundry is not reasonable. You're on vacation, like, and especially if you paid a cleaning fee. So that is not reasonable. So that was the second one that I really prioritized. The third, and this is more on the opportunity, is monthly stays. That's a growing part of our business.

42:41People think of us as a travel company, but 20 % of our business is now housing. In Airbnb, there's a lot of problems with monthly stays. For example, you can only pay by credit card. Paying your rent with credit card means you have to pay a credit card processing fee. So we thought, well, what if we allowed an integration where you can pay by bank? We did integration with Stripe, and now you can save a bunch of money paying by bank. You can now pay in installments. We now lower the fees. We now have this really cool little dial. We now have really cool pricing discount tool so hosts can be encouraged to charge less and will promote the very best listings.

43:15So these are maybe some of the ones that I was probably most passionate about. Anything around pricing, anything around chores, anything around really expanding the definition Airbnb. But yeah, there were 53 upgrades that we made. So there were a lot of things. Yeah. And running a remote company, you're so thoughtful about how you're running the company now. And you have constraints. It seems like you've got it really dialed in. And listen, the stock market is responding to it. The stock market is looking for discipline. I think you were the first of the cohort to actually show it. uber just showed it they started hitting free cash flow dar has done a great job there and we're starting to see this discipline and zuck actually got got uh religion he cut 20 000 people and you know got rid of layers of management how is um when you are a remote company and you try to institute this are these things in sync still or do you miss being in the same place with each other or do the you know these uh retreats that you're doing and these off sites whatever you call them now are they enough or do you think at some point you want to be in an hq again yeah actually yeah so probably actually even good to clarify um we're not a so-called returned office company or remote company i would say like shopify and coinbase are probably remote companies as far as i can tell at last i checked i go to an office every week and many of my top people go to an office every week and we have offices open and we welcome people back to the office but we don't require people.

44:40We say you work wherever the hell you want, because we're so disciplined, so organized that we don't think there's going to be a hit to productivity. I basically came to the conclusion that especially with the accelerating rate of technological progress to AI, increasingly, you're going to be able to have more and more of a global talent pool. And so then you have to, if you're going to require people to come into office, you have to believe that you get a greater productivity gain by having them physically there than the productivity gain you have being able to hire anyone anywhere in the world.

45:06and for some jobs, like your core team where it's really creative and you're making stuff, I think you want to be physically together. But I do not need our accounting systems to be managed together physically around me. I don't need like thousands of people physically around me. I can't even talk to them. After about 50, 100 people, I don't know, I can't remember the conversations. I can't keep track of those conversations. So my vision is that we have a small, tight team of the most, some of those senior people, many product people. we built this creative studio and we're in San Francisco a lot.

45:38And then the rest of the company, they can choose. Some groups, they do come to the office. Other groups, people just work totally remotely. And then once a week, a quarter, we try to bring people together. Sometimes it might be a week every six months. And I think that kind of suffices for a lot of people. In other words, we kind of give teams the option, but the most senior people, we're pretty much together. We also travel a lot together. So like I'm here in New York, we probably had like 20 people here with me. So I was around them. But I think we're just still learning. We're still moving.

46:09And I think, Jason, the future is flexibility. I think that this is the worst technology we'll ever be in our lifetime. As screens get better, as bandwidth gets stronger, I think people are going to want more flexibility. And my prediction is even proponents of working in the office, I bet you a lot of those people are going to be somewhere else over the summer in an Airbnb. Yeah. It's easy for people to be like, yeah, we will. All these rich guys in New York, let's just be honest, all these rich guys in New York who said they want to call back the office. They're all going to the Hamptons for the summer and they're going to Europe in August.

46:39So even the return to office, those aren't 12 months a year. Those are now nine months a year. Yeah. And I mean, Europeans have their have a better lifestyle and they enjoy it. So maybe what happened during this pandemic, when we look back on and we do the postmortem, it's going to be, hey, you know what? Maybe Americans, we were slaves to the offices a bit too much. Yeah. We over indexed on it. And maybe there's some balance because when you do take a four-week vacation and you work half days for two of the weeks you kind of come back refreshed and you stay with the company longer you don't resent your company and you're not depressed because you feel guilty about not spending time with your kids or not checking off bucket list items and i think you're so right about the speed to hire that talented person versus the annoyance of somebody quiet quitting or somebody's phoning it in or somebody's working two jobs like of course you're gonna have abuse in every system there's abuse of people who come to the office screwing around with their door closed.

47:32So, you know, you're going to deal with some level of, I don't know, bad actors, whatever. And I kind of feel like if you want people coming to office, it should be because of collaboration, it should be for trust building, it shouldn't be for accountability. We have enough tools digitally that actually it's frankly easier to track people when they're working digitally than they're in the office because you can actually track everything they're doing and you can just have really good mechanisms like what did you get done today yes actually we have a whole program manager operation whose job it is to basically tally what everyone's doing and if we don't know what you're doing then we're not doing our job that's i think why people are so um some of these managers are so aggressive about this is because they don't actually know how to manage yeah and so since they didn't know how to manage they were managing by the fear of being in the office and the punch clock as opposed to did you assign the right person to work and did you check the work yeah and checking the work and assigning the work is a lot of work and prioritizing as you just said your biggest challenge as the founder is just keeping all this in your head yeah and trying to make the right decision on behalf of the customers right and so many managers are phoning it in uh i came up with the simplest system possible for my investment company which is at the beginning of the day have a cup of coffee and share with the team what you're going to try to accomplish and when you check out for the end of the day like make it a good checkout so you can go spend time with your family and just say what you got accomplished and if there are any blockers and when people just started doing that in the slack room oh it's such a big difference it was such a big difference and then when somebody left the company you know we did we looked at their end of the week and what they said they got accomplished what they were working on and we said hey here's the last three end of the weeks we put it into one document who should do this job should this job be retired should it be spread across three people should it be outsourced you can actually look at the granularity like you're saying when they do digital work it's all right there so then you can say does this position need to be here or should it be outsourced or automated you keep bringing up ai and i'm curious when you start what are you seeing in ai right now that's getting you excited internally running the company and then externally in terms of opportunities because travel is such a rich textured experience it's also so customized so you know so much about your customers you know when i'm in japan you know my trip to france you know the different places i have airbnb'd you know who came with me um and you know whatever else you know about the hosts and everything what do you think the opportunity is going to be in travel specifically for ai yeah so let me let me um if if you may indulge jason i'll like zoom out and just give you like kind of my mental model and you can tell me if this maps to yours so you have like the base of the foundation right the base models which are basically the large language model.

50:17So of course, GPT-4 is probably the preeminent base model. Google has some base models. Anthropic has a base model. Microsoft Reuters has their own base model. So you have like three to five big base models. I think of those base models as if it's like a highway, they're the highway. They're like the infrastructure. And, you know, in another generation, by the way, that probably would have been dealt by the government, right? It's almost like the Manhattan Project. These large language models are going to be eventually about$100 billion dollar supercomputers running these models. So they're going to be like massive.

50:47These are giant infrastructures. And we're not going to do that. We don't do infrastructure. Airbnb, we think of ourselves as designing the cars on the highway. So on top of the base model, you have the tuning of the model. And the tuning of the model is going to be as good as your sensibility and your customer data. And if you and I both asked ChatGPT a question, we mostly get the same answer. And we mostly get the same answer because it doesn't know who you where I am. And that's great for some questions. Like what was the like, you know, like, how far is the moon from the sun or whatever, like there's one right answer.

51:20But if you ask, like, where should I travel, your answer and my answer are probably different. And so, some problems are search problems. Some problems are kind of matching and personalization preferences problems. And so, what we want to do is we want to be one of the best companies for AI personalization. So, we want to develop really good tuned models. To do that, we have to change our business. And actually, one of the questions that Johnny Ive told me when we brought him on the team is he said, you need to switch from beyond, you need to go beyond where and when. Right now, you ask, where are you going?

51:50And when are you going? And we need to shift to who and what? Who are you? And what do you want? And that's really the vision. And so, what we want to do is we want to build these robust profiles. I want to start to learn, Jason, who you are, build really good, rich customer information. And then I can understand and personalize, like, where do you want to go? And also, what do you want in your life? Like you looking for inspiration, just get out like, you know, what do you what are you looking for? Do you like want to get healthy? Do you, you know, and you start to learn about people. And then we're also pretty good interfaces, and the application layer.

52:22And I think that Airbnb, that's where we're really, really going to focus, we're going to focus on the tuning of the models, the most personalized AI interface, and then really good application interfaces. Now, I think as far as interface, I don't think they're all going to be just text-based. For example, we were working on the OpenAI plugin. We were one of the first partners that was working with OpenAI. And at the last second, I pulled the plug on the plugin because I just didn't like the interface. I didn't think that was the right way to interface a travel. I told Sam, I said, long text outputs are low bandwidth.

52:54And then you give me another text output with widgets at the bottom. I said, I want something much more multimodal, more visual, richer. And if you're going to have access to GPT-4, why don't we put it in our app? So we pulled the plug on the plugin. And we think eventually, I think our real vision is Airbnb, at the largest sense, isn't even a product or service. I hope it's more like a travel community. What we're really building is a travel community. And then the role we have, the app in the travel community, is eventually we're like the ultimate AI concierge. We're like the ultimate host. Like Charles Eames, one of the greatest in our 20th century said, the role of designer is that of a thoughtful host, anticipating needs of the guests.

53:34So, that's what we should do. And we understand who are you? What do you want? Where do you want to go? And maybe we could even go beyond travel if we get there, right? And the part of that means you have to trust us to give us your personal information. It means we have to be a marriage of art and science. It means we have to understand a lot about like human psychology and know what you want. It's not just a technical problem. We have to design unique AI interfaces that are probably richer than just text inputs because you want to see and feel things, right? And I think it's much more immersive.

54:07So that's where I think it goes. That's the long-term vision in the interim. And that's probably long-term as in three to five years. I mean, that's not even that long-term. In the next year, what we're going to do is three things. One, engineer productivity or productivity. I mean, RCT, I think engineers can be 30 % more productive in the next three to six months. That's exactly the number I'm with. I'm watching my team interview founders for investment. I'm looking at due diligence and I'm looking at my own behavior. And it seems like one out of three tasks I can offload. And so I'm like, this is like this year, 30 % more efficient, which means headcount stays the same.

54:45But we just added a third more people. Like, this is unbelievable. It's amazing. And by the way, it's actually, if everyone's 30 % more productive, it's actually more productive than adding 30 % more people because, you know, in the mythical man month, every time you add somebody, they bring a communication and like, like, they bring a tax with them. So this is like productivity without the tax of more people. And I think the productivity doubles, you know, in somewhere between one and two years. It's a little hard to know, but it will be double. So we'll have the equivalent of twice the engineers without the productivity tax.

55:18and so that's the first thing is just getting everyone on the tools give everyone a co-pilot and getting everyone like just immersed and i just bang the drum like use these tools as much as you can i don't want to be like the luddites or like afraid of a computer in the 1980s right like we want to be using these tools that's that is so key i told everybody to make chat gpt for their home page in their browser oh i love that every time they open their browser it just reminds them hey use this and i said searches is obvious but when i told you they do they're like hey here's what i'm going to accomplish today and here's what i'm going to accomplish at the end of the day i just told everybody try that in chat gpt and try a plug-in and by the way when you were talking about plugins being like not exactly the right modality i've been using open table expedia kayak the last couple days for travel trying to see if it works and it's not as good as using some of the other like native interfaces yet i mean i'm sure it'll get better but i agree that it needs to be visual and it yeah it's got a long ways to go open ai is a choice they can either invests a ton on the interface layer, or they can decide that they're the brain and other people decide the interface.

56:22Either way it can work. But either way, the interface is, here's the thing I'd say. Do you remember when like the iPhone launched and Steve said like, the problem with most smartphones in the bottom half of the screen, the interface doesn't change? I think that we should think of every task as wanting a distinct interface. Like you have a hammer to get a nail You have a screwdriver for a screw. You have a spatula to scoop something up. Every interface should be custom designed exactly for the task. Do you think it's voice for travel? Or do you think it's flipping through videos that are TikTok style and then understanding which ones people spend more time on?

56:57I don't think it's voice. I think voice is for commodities. So if you basically say, I need to go from here to there, and there's no inputs, there's no decision making or visual discretion and voice. I think it's more like, I don't want to say like minority report, but I have this image of like, it's an immersive interface where it's words and images. And it's a conversational bot, but it's not pure text. It's got lots of rich visual videos, photos. It's a little hard to describe. I have to build a project. Well, if you haven't seen Minority Report, he puts on a pair of gloves, which MIT was working on at the time.

57:31And he's just moving stuff around in a virtual desktop. That's conceptually what I'm talking about. Yeah, I think it's actually a pretty brilliant one, because if you think about the function of magazines, which is how people did this previously, you flip through a magazine, what would you see in a magazine? A subhead, a headline, a table, an infographic, a picture, a two-page spread, a table. and all of that was evocative and kind of pulled you in and i actually really think this vision of trying to understand me as a person psychologically and where i'm going next oh jason's you know doing more skiing oh he's got a bucket list of places he wants to go skiing ah he's really into food and that trust you built up is interesting as well because i would have no problem authenticating with instagram and giving all my photos and with ai you could just tell from my photos that I take pictures of a lot of food and I take a lot of pictures of skiing.

58:24Like you now know a lot about me and I don't have to tell you, you just know my Instagram account. And I think the problem is all of us, and Airbnb is part of this problem. You come to Airbnb and it looks the same as it did the last time you came. And we're like a marketplace and everything's a transaction. And we assume you're like, we don't know anything about you and you just walked into a store and we got a bunch of stuff on a shelf. and I think that's a very 1990s 2000s Amazon paradigm of commerce and I think the future of commerce is more like somebody's showing you around and they understand you deeply and you have so much more control and it's significantly more personalized and everyone has a unique experience and so that's where I think it goes and I think that we don't have a search problem in the future we have a matching problem so we are going to use AI to match you to whatever you want and I think that the other thing is I think we're going to be really good I hopefully identity authentication.

59:16You know, the problem with AI is, what's the first word in AI? Artificial. The biggest risk before machines come after us is they become us. And we can't discern who is a machine and who is not. And I think that we're already seeing that. I mean, when that Pope photo circulated, I thought that was a real photo at first. I did too. I was like, that's weird. The Pope is going for it. Yeah. I was like, pretty awesome. He's got Balenciaga, go for him. But that's today's technology. So, we're about to live in a world where someone can sound like me, they can behave like me. And as we spend more and more time online, it's going to be hard to discern what is real and what's not real.

59:50So, I think in the age of artificial intelligence, the other thing people want is authenticity. And authenticity is whatever's real and whatever's authenticated. So, I think our brand is kind of authenticity. We're not going to ever be the most digitally immersive company. That's going to be social media or entertainment. Our value is really the physical world. We get you online, offline with people different from you all over the world. And that comes with starting with knowing who you are, authenticating your identity. Eventually, we may use biometrics because right now we use like a government ID.

1:00:21But I think eventually, I want to do something with biometrics to be much more robust. So, you can't really get around it. And then we have your identity. You trust us. We build this incredibly robust personal profile. And then we can just match you. And then hopefully, we're really good at interfaces. and we use the latest, greatest language models. Yeah, it's pretty amazing how fast these are working and what's going to be possible because we don't know what it's going to match people with over time. Yeah. I don't know what my, I don't know necessarily what's going to delight me. Exactly. Every time I go somewhere in the world that somebody else introduces me to, like my guide on the side, a friend, if you told me, hey, you got to try this place in San Francisco, cafe okinawa that i've been going to to get tonkatsu sandwiches i told like three people about this and because i love these when i go to japan and they're just everybody goes there and they're like oh yeah you're right that was great and now they spread to the next person and that kind of word of mouth i think like this ai is going to find little things like that where it didn't know that this was something that you would be delighted by totally and this is such an opportunity i think it's amazing and by the way i think i heard another one of your podcasts you said something that really resonates that I think it's just worth calling out.

1:01:35The cost to develop software is about to go down. And maybe a good analogy, Jason, would be like the camera. 130 years ago, very few people could operate a camera. And so therefore, you had to be a specialist to take a photo and there weren't a lot of photos. Suddenly, computer programming is just basically telling a computer to do something in its language. AI means you can now tell a computer to do something in your language. The moment you don't have to learn a different language, you can use your natural language to English, suddenly kind of everyone in a way can be a programmer like everyone's a photographer you don't have to actually have a skill now there will be programming skills but suddenly anyone can do that now when it's possible there's going to be so much more software we're going to live in abundance of software i think in a podcast you reference like a ski instructor or a ski company like a niche business that couldn't have developed hired great software engineers but now they can the cool the best engineer you've ever had developing their app and so suddenly the skillsets are different we have to understand skiing human psychology taste design and how to operate and have a conversation with that tool but you don't have to necessarily build the infrastructure just like we never built servers because we had aws but if we launched five years earlier we would have been really good at like kind of building out servers and stacking servers i mean your first half a million would have went towards a co-location facility yeah and now it's like you're just going to talk to this thing and it's going to make your software for you so i think this is going to create i think this is going to create millions of startups i think that like entrepreneurship is going to be a boon i think there's going to be millions more i think anyone can essentially do the equivalent of what software engineering only allowed you to do only five years ago it's going to be awesome for many people it's going to be wildly disruptive for others albert einstein used to have a saying the best way to keep your balance in a bicycle is to keep moving and i think the best way to keep your balance in the world the ai is to keep moving forward and just adopt the tools don't fight it and see where it takes us but it's going to be pretty well you say that i was just looking the writers guild was going on strike and like on the second page the third thing from the bottom was their update on um ai and they're like we're seeking a ban of using ai in the writers room using ai to ingest previous scripts and using ai to generate future scripts and dialogue and it said um the most the studios will commit to is doing a yearly review with the union about technology.

1:03:56And it's like, don't you realize you could ingest the entirety of The Simpsons or every late night joke? And then you would start on second or third base with all these incredible ideas and brainstorms for jokes, and the jokes would get better. It would and the shows would be more compelling. You're totally right. I think trying to ban AI is like trying to ban electricity. I just, you're going to be on the wrong side of history. The genie is out of the bottle. You can't put it back in the bottle. And I think that here's the problem, Jason. I think it's easier to imagine what jobs will be displaced than what jobs will be created.

1:04:25Because of course, we can imagine everything that now AI can do. We can't imagine everything we hasn't yet done because that requires us to conceive of what doesn't exist. But if we remember that at all other periods in history, technology created jobs that didn't exist. This can do that too. It doesn't mean we should be blindly ignorant or not be concerned. There's a lot of concerns I have about AI. I'm concerned about how fast it's going and is society prepared for the speed? But I think from a creative standpoint, as a person who went to Rhode Island School of Design, I would tell the creative community, you only have to be worried if you don't want to be a part of it because this is a creative tool for you.

1:05:03You know, computers are tools. The reason they put a handle on the back of the iMac was they used to have a saying at Apple, never trust a computer you can't throw out the window. Computers are tools and AI is a tool. and i think if we think of it as a tool then it's a tool for creativity and you're right you start on third base like i'm already noticing a chat chibi t like it makes my writing better i don't use i actually don't use most of what chat chibi gives me back but it gives me some words and i'm like summarize this the cold start problem is solved because you're like hey i want to write about i want to make a marketing plan and it's like oh well here's what goes in a marketing plan you're like oh yeah points two and seven let me double down on those or i can like i can have a thousand word thing, say, summarizes in four sentences.

1:05:46And maybe I don't use the four sentences, but it gives me ideas. By the way, before we go, I want to say, one, I keep discovering new uses for ChatGPT. Here's the coolest use I've discovered. I asked ChatGPT a question, at the most fundamental level, what business are we in? And then it spits out a not super insightful answer, like Airbnb is the business of travel and experiences and blah, blah, blah. And then the cool thing that you should do this is ask a follow-up and keep asking follow-ups still can't do it anymore. And so I asked, what's a more fundamental business? And it goes, actually, a more fundamental business that Airbnb is in is a sharing.

1:06:20And I go, okay, but what's a more fundamental business? Then it goes, a more fundamental business is human connection. And I go, what's a more fundamental? And it just keeps going. But you start to learn. And it teaches you its first principles, right? Because - It reduces it down, yeah. And then I did the same thing with AI. I said, how can AI transform Airbnb? And it has these really small ideas. And then I go, how can do even bigger? And you keep doing it. And so if you keep going, go bigger, or go more fundamental, keep going. Yeah, take more risk. AI starts to teach, it helps you discover the first principles in really like interesting ideas, which I think the follow up questions with these chatbots are always more important than the first question.

1:07:02Yeah, it really gets interesting the third or fourth time. And as you said, I think one of the most profound things you said here is we don't know what jobs can be created because they haven't been created yet. And if you look at marketplaces, whether it's eBay, whether it's Etsy, whether it's Airbnb, whether it's Uber or DoorDash, to different extents, these allowed people to create new careers for themselves. And how many people do we know now who became real estate magnets and they own 20 properties because of how well their first two properties did on Airbnb or they have collections on Etsy or Kickstarter?

1:07:37starter and who knows what this is going to lead to but it's going to unless there's not enough problems and not enough human creativity and energy then this is going to be incredible for society i believe i'm an optimist about it and i'm using it every day for multiple hours a day i'm playing with the plugins the web tools and i just started on replit like um putting up bounties and i'm started a python course because i'm like i just want to know how to like stitch this stuff together because the auto gpts to me are where this gets really wild oh yes if you can start putting a couple of the ideas we talked about today on autopilot and then say hey and try to make a better result every day it could just be trying to make a better result for how airbnb understands my travel taste and you say hey just keep trying to get information from jason ask him some interesting probing questions or look at his social media and and try to give him better ideas and then I give the thumbs up, thumbs down, either implicitly or explicitly, and you just start understanding me.

1:08:36Hey, listen, I know you got to go. This has been amazing, dude. I'm so glad we got to finally do this. I know, thank you for having me. I'd like to be back again soon. We got to not wait so long. Well, you know, the thing is, like I had all these notes about like the blocking and tackling. Like all great conversations, I think, and great CEOs. You've got some things on your mind about how you're architecting the company and then AI that are really profound. So if you're an entrepreneur, or you're going to want to listen to this with your team and probably don't put it on 2x speed because Brian and I both talk pretty quick.

1:09:08But a great conversation. Book end it with the Joe interview. And we'll see you all next time on This Week in Starters. Bye-bye.