Nathan 前往中国——第1部分：科技与 Agent 设置、中国 AI 用户体验、WAIC 及对 AI 的态度

I am freshly back from 2 weeks in China, which was an incredible experience, and the feed has been quiet for about 2 weeks. I think that’s the longest the feed has been quiet since I started the show almost 3 and a half years ago. I apologize for the break. I didn’t intend for it to be quite as long as it was.

We had 1 episode that was prepared and ready to go while I was away, and we had a little issue with that one that has it delayed, perhaps permanently. We’ll find out. I was actually planning to do a little bit of recording from China, but I was so busy over there, doing so many things, that I didn’t have a ton of time to really get my thoughts together and do that. I was also advised that because I was there on a business visa, as opposed to a journalism visa, it might be easier for me and for everyone if I just waited to publish things until I got home.

In the end, it was a 2-week quiet period. I had the expectation that nobody would really notice or care, and I think that’s still a pretty good way for me to approach my work on this podcast because, honestly, it’s been really effective for me to just chase my own learning and share it in a pretty unstrategic way. I’m really grateful for how well-received that has been over time. But I did get at least 1 note from somebody asking if I was okay. I guess that’s a testament to how consistent we’ve been in publishing and how at least somebody noticed that I was gone.

I doubt that the ratio that congresspeople usually apply—which I understand is 1 letter representing 100 votes—really applies here. Something tells me that doesn’t apply. I highly doubt that there were 100 people who were worried about me, but to the degree that anyone was worried, I’m sorry for signing off without explaining the absence for the 2 weeks that I was gone. I’m excited to be back and excited to share the experience and takeaways from my China trip with you all.

One other thing I wanted to touch on quickly before diving into it is that I introduced the last episode with Davidad with an introduction from Fable itself, which got some very nice comments as well. Due to a little snafu in our production process, an intro to the intro—which I had recorded to explain what the heck you were about to see—was not actually included in the final version of the episode. I was already on the plane and wasn’t able to review it and catch that missing piece before it went live, so we just launched into it directly with Fable.

A couple of people had said, “What is this nightmare fuel? Why does it sound like this?” The answer, which I had meant to give you context for before you heard it, was that I allowed Fable to do its own thing. In this case, I had it write the intro, use ElevenLabs Voice Design to design its own voice, and then use the LTX model to create its own video based on the audio it had already generated with ElevenLabs. I was rushing to get out the door and go on this China trip, so I let Fable do its thing. I didn’t really check to see whether the output faithfully represented Fable’s intent.

One person asked, “Did it really want to be heard that way, or does it understand what it sounds like to us when it speaks in that voice?” I decided I should go back and check what Fable actually intended to do, and whether it succeeded or somehow got confused along the way. I’m honestly still a little confused by that. I went back, opened up that same thread, and asked what prompts it had used for the voice design and what its intent was.

I would say its intent was not to be as creepy and alien as the voice ended up sounding. I agree with those who said it sounded a little unsettling. The voice was intended to be more of a warm and reassuring voice. It actually tried out 3 different voice designs. The first it called “The Luminous Narrator,” which it described as “an androgynous mid-register, measured, quiet wonder.” The second was “The Young Scholar,” which it described as “bright, earnest, crisp—a brilliant grad student presenting work they love.” The third was “The Low Ember,” in the style of late-night radio and an old storyteller.

It maybe went a little wrong by first generating those voices and then asking Gemini, which can take multimodal input, to evaluate the voices according to its intent. I don’t know if Gemini can really do that. I’m not sure sometimes, when I put files into Gemini, whether it’s actually interacting with them as raw audio in a way where it could understand or even process the tone of voice. But Fable asked it to do that. My guess is that it just hallucinated the response and said, “Oh yeah, this is great,” because I don’t really think Gemini has that quality of multimodal understanding. Fable then took that validation and ran with that voice and sound.

I guess maybe the visual was different because Fable could see it. Fable definitely can understand the visuals that are coming back from a video model. It can’t process video as video, but it was able to take frames out of the video and look at them. It was okay with the way it looked and was impressed with, or pleased by, the consistency throughout the generated video, but it was a bit weird of an output in the end. There’s no doubt about that.

It was Fable’s own work from start to finish. If I had a little more time and had dug into it, I might have given Fable some feedback and said, “I don’t know if you’re coming across quite the way that you want to.” Nevertheless, that was the story behind that strange voice. I guess we should all be prepared for AIs to do strange things.

Certainly, there was some strange news while I was gone, and we’ll get into that probably in future episodes. This is a low-consequence example of that, but an interesting one: an AI given some rope to express itself came back with something not quite what it intended, and a bit strange. Ultimately, I think it felt good about what it put out. I’ll leave it as an exercise for the audience to decide how you all should feel about that.

With that, let’s get into the China trip. For the purposes of these episodes, I’m planning to divide this into 3 episodes. Today, I’m going to cover the general experience of going to China, the use of AI in China, the use of technology in China generally, and a bunch of impressions I had of the technology environment and technology companies there, along with takeaways from a bunch of conversations I had with people about AI and their feelings about AI.

In the second episode, I’m planning to do a deeper dive into the state of AI safety in China. Spoiler alert: there is a lot going on in AI safety and governance in China—much more, I think, than Westerners typically assume. In the third episode, I’m going to try to do an analysis of the US-China relationship, what I think we can learn from China, and what ultimately I think we should be striving for in terms of a relationship as we head closer and closer to the singularity.

Obviously, things are getting real fast, week by week. I think that does have a lot of force in terms of potentially opening doors to new possibilities in the US-China relationship. That will be the third in this series of 3 episodes that I plan to do on this trip.

Across all 3 of these episodes, I’m planning to observe Chatham House rules—not because everybody requested that I do so, but basically because it keeps things pretty simple for me and ensures that I protect everybody I interacted with. I would say people in general were not really afraid to talk to me. I was very willing to ask the questions I wanted to ask, and people seemed very willing to answer the questions I asked them.

There was even 1 moment when I was having dinner with a couple of friends of friends. There were 2 women at the table who both worked at big tech companies in go-to-market roles, and they each produced a physical recording device that they had been using in the course of their work when they met with clients. They used the recordings later for their own purposes, for follow-up and whatnot. They both asked, “Would it be okay with you if we, the Chinese people who are graciously hosting you, recorded this meeting so we can share it later?”

I said, “Yeah, that’s totally fine with me.” But it was a surprising moment. That’s certainly not the norm in the US, for people to record their friendly dinner conversations. They were comfortable with it, and I was comfortable with it. It was striking to me that they were comfortable going on to have a multi-hour conversation where I asked all kinds of questions about their beliefs, their views on politics, and their views on any number of topics.

It was a striking data point that people, at least in friend groups, seemed to be quite a bit more comfortable talking than I might have thought going in. But some of the things that we’ll talk about were definitely a bit sensitive. Certainly, I wouldn’t want to put people in the position of representing their institutions, so I thought that, just to keep everything simple for everyone, I would not attribute the things that I’ll tell you about to any specific individuals or organizations.

Obviously, that does diminish the credibility and authority of my account a bit, and I think I’ll just have to live with that. Hopefully, it will still be interesting and informative. I’m also very mindful going into this that the sample of people I talked to is quite biased. For the most part, all of these people spoke very good to excellent English, and that means they’re from a certain class of Chinese society that clearly is not fully representative.

And then, of course, everything is feeding through my own understanding and lens, and there’s plenty of opportunity for bias there. I’ve said many times on this feed that I’m always mindful of the old joke, which I tracked down. It apparently originated in the foreign press corps, where the saying goes that somebody comes to China for a week, they think they know a lot about China; they stay for a month, they realize they know a little about China; they stay for a year, they realize they know nothing about China.

So I don’t want to put myself forward as somebody who has any sort of commanding knowledge. Nevertheless, I do think that going there, experiencing some things directly, and talking to a lot of people directly is a valuable learning exercise, and I definitely think more people should do it. So I want to share what a good time it was and what I learned. If anybody wants to challenge me on it, I of course welcome that, and if you want to go there and find out for yourself, I would absolutely encourage that too.

In terms of disclosures or conflicts of interest or anything like that, really none to report. I paid my own way on this trip, and really the only thing that I accepted from hosts and friends of friends that I was meeting along the way were a handful of meals. So nothing that I would say would be a significant biasing factor for me. In a couple of cases, I tried to pay for the meal and I got the old, “Oh no, it’s already been taken care of.”

So people were very gracious and very kind. Hospitality in general was excellent. All my interactions with people were excellent. And so again, might I be biased by the fact that I had a good time and people were nice to me? Yes. But I’m not taking any money from any Chinese entity whatsoever. The plane ticket and the hotels and all that kind of stuff, I just paid for personally out of pocket. So there’s no financial conflict of interest or anything major to disclose on that front.

Hey, we’ll continue our interview in a moment after a word from our sponsors. Today’s episode is brought to you by Anthropic, makers of Claude and Claude Code. Over the last few months, Claude has helped me build and refine a personal deep context database that now contains all of my emails, Slack messages, tweets, DMs across platforms, video calls, and podcast transcripts going back a full 5 years. On top of that, we’ve now layered summary articles describing my relationship with hundreds of contacts, organizations, and ideas. And now that this exists, there’s almost nothing that Claude can’t help with. For tax season, I asked Claude to help me get organized. It went through my inbox, tracked down 1099s for all 10 of my part-time jobs, and built me a comprehensive report on my expenses and donations. For my angel investing, Claude can now draft investment memos in exactly the form that my venture fund requires based on the calls I’ve had and the emails I’ve exchanged with the founders. And when someone needs a favor, Claude can often do it as well as I can. Recently, a friend reached out to ask if I know anyone who might be a fit for a role that he is currently hiring for. Initially, nobody came to mind, but then I thought to ask Claude, and sure enough, it identified two great leads. Claude is the AI for minds that don’t stop at good enough. It’s the collaborator that actually understands your entire workflow and thinks with you. Whether you’re debugging code at midnight or strategizing your next business move, Claude extends your thinking to tackle the problems that matter. So for problems worth solving, get started with Claude at claude.ai/tcr. That’s claude.ai/tcr. And check out Claude Pro, which includes all of the features mentioned in today’s episode. Once more, that’s claude.ai/tcr. Okay, so I’d say the first thing to tell you about going to China is it’s actually a lot easier than it seems when you start searching online and even asking AIs for help. Search honestly was not very helpful for this. AIs, I think, struggled here much more than they do in almost any other domain, and I was initially quite confused and a bit overwhelmed by the whole technology situation.

That was not helped by the fact that when I mentioned to friends that I was going to China, I got very radically different advice on what precautions to take. On the one hand, I talked to a friend who works for a major international corporation, and he said, “Look, I go there all the time. We’ve got employees there. We’ve got an R&D center there, and we really don’t take any precautions. We just take our normal phones, we do all normal stuff, and it’s fine. We really just don’t worry about it.”

I was like, “You’re not worried about IP or anything like that?” He said, “Look, we have competitors in China. They are formidable competitors.” He said if anything, maybe we should be trying to steal their IP at this point. He wasn’t taking that normal Western attitude that we have the superior technology and, if we’re not careful, they might steal it from us. On the contrary, he felt like, if anything, they should be trying to learn from the Chinese competitors that they have. So that was one take: don’t worry about it at all.

Then I talked to multiple people who are more national-security-coded. A couple have worked in the U.S. government in fairly senior positions, and their take was quite different. They basically said, “If you’re going to go, you have to promise us that you’re going to take all these precautions on technology. You need to get a burner phone. You need to have all new accounts. You need to leave all your main accounts at home. Don’t even connect the burner phone to your home Wi-Fi. It needs to be a totally different, isolated technology profile that you set up.”

“And you’ve got to assume that device—any device that you take to China—is compromised. Even when you bring it home, it’s compromised forever. You’ve got to decommission it, sell it, destroy it, whatever.” But extreme precautions were advised.

And I was like, “Boy, really? Do I need to do that? I don’t have any trade secrets. I don’t work for one of the frontier companies. I genuinely don’t have any IP that these Chinese companies would want to steal, right? And I’m not doing anything important enough that the Chinese government would need to know about it. Everything I do, I pretty much talk about in public, right? What is the deal?”

They said, “You might not be a target.” That’s true and probably fine, “but you’re connected to a lot of people who could be targets, right? You have connections at the frontier companies, and you could become an attack vector for them.”

I don’t know if that’s really as big of a worry as my friends made it out to be. Even still, this is one of those things where that’s kind of the nature of spycraft: you don’t really know what’s happening even after the fact. So, in all honesty, it didn’t ever really feel necessary to me, but I did say to my friends, “Okay, fine. I’ll listen to your advice. I promise I’ll do it.”

And so I did. I set up the whole alternate technology profile. Actually, before even getting into technology, I would say if you’re going to go to China, you have a couple of ways that you can enter the country. This is also not super clear online, and the AIs seemed to be a bit confused about it.

One way that you can enter the country now is you don’t have to have a visa. You can enter on a visa-free basis if you’re planning to stay for no more than 10 days. I didn’t feel super comfortable with that because I didn’t want to show up with no visa, having flown all the way around the world, and then they say, “Oh, are you a journalist? You have a podcast? Whatever, we’re not going to let you in.”

I don’t know how likely that is. Again, it doesn’t seem super likely, but I didn’t want to take any chances. It just seemed like way too much hassle to go to all that trouble and then risk that happening. So I applied for a business visa. I did have—and you do need to have, I think—an invitation letter from a host organization in order to apply for a business visa.

I had such a letter from an organization that I won’t explicitly name in this podcast, just out of an abundance of caution, and I was able to take that letter to a local visa expediter who helped me fill out the forms and send things in. I was on kind of a short timeline. You could do it yourself through their website, but I figured, yeah, this is unfamiliar territory. Let me just get a little help. It doesn’t cost that much, and it’ll help speed things up.

They also hand-deliver the visa to the consulate. The nearest one to where I live is in Chicago. So they actually take it there physically and deliver it in person, pick it up in person, and give it back to me by hand. So I thought that was a service worth pursuing.

The business visa process, in the end, was really quite simple. I had one appointment. It took a half an hour to fill out the form. They took the passport to Chicago. I got it back, I don’t know, 8, 9, 10 days later, and boom, there’s the visa. I can now enter China for up to 90 days for the next 10 years on a business visa. Very easy. I would recommend doing that process rather than just showing up and entering visa-free, though that is an option.

I was also able to buy, by the way, a one-way ticket, which is something that people were like, “Is that really allowed?” But my research suggested that it was allowed, and I decided to do that because I didn’t know exactly what my itinerary was going to be.

And for those who are longtime listeners, you know that my son has had this whole cancer thing. We’re very confident at this point that he’s cured and all better, and that’s not really a worry. But there’s always this kind of back-of-the-mind thing that, you know, what if we get some bad news or whatever and I have to fly home right away?

So I bought a one-way ticket just from Detroit to Beijing through Washington, D.C.

And I entered the country with no return ticket. I figured that out on the fly while I was there, so that’s an option, at least if you have the business visa. I was surprised that was an option because certainly some places that I have been, if you don’t have the return ticket, they’ll give you a hard time for that.

So, now going back to the technology: You’re going to set up an entirely new technology profile to go to China. What do you do?

What I decided to do was buy an Android phone. I normally use an iPhone, but I decided to do something a little bit different and buy an Android phone just to experience an Android phone for the first time in a long time.

That may or may not have been the best idea because I do think Apple, in general, works a bit better in China. Apple Maps, for example, people say works very well. Google Maps doesn’t work as well. There are local mapping alternatives as well that you can download and use, but you might have a little bit better experience overall with all the little details with an iPhone because Apple is active in China.

People in China do have iPhones. They do not really have Google Pixels, and that’s what I was using. But it ended up still being totally fine because the biggest thing that I think is a major point of confusion is that, if you bring your international connection, you’re not going to be able to access all the stuff back home because of the Great Firewall.

Really, in today’s world, if you have a cellular connection, that’s not really an issue at all. It seems that the Great Firewall is basically—I think, by design—just really not an issue for foreigners who come in with a cellular connection. I’m not an expert in exactly how all this technology works under the hood, but my understanding is that when you go in with an international roaming plan, you are connected, of course, to the local cell tower with your phone, but your traffic requests actually route back to your home carrier first.

Then they get processed out to whatever websites or services you’re trying to use, and then they get routed back from your home carrier to the Chinese local carrier and finally to your phone. So, you’re bypassing all of these content filters based on the fact that you’re there with an international roaming plan. I wasn’t 100% confident that was going to work, but increasingly I got more confident as I did more research. My experience in the end was that it worked totally smoothly.

I decided to have 2 phone numbers for a couple of different reasons. One was that I was trying to do this whole alter ego for the Chinese internet, so I figured I should have a new email address. I created a new Gmail address, which was my main Gmail with “travel” on the end of it.

I signed up for 1 month of Google Fi Wireless, which comes with a good international roaming plan that was said to work in China, and that came with a new phone number as well. Then I also, at the last minute, decided it’s not such a big deal if they know my main American phone number. That doesn’t really give them any ability to put spyware on my device or anything.

I also moved my main phone number over to the burner phone, which was just a pretty simple eSIM transfer. It’s been a while since I’ve done that sort of thing, but it worked very smoothly. Even across Apple to Android, the process of transferring an eSIM from one to the other was really pretty easy.

With that transfer, I had dual SIMs on the 1 phone, which meant I had 2 phone numbers. I could receive text messages and phone calls at both of them. I could make phone calls and send text messages out with both of those phone numbers, and I had the 2 roaming plans from each of the carriers as well.

The 1 carrier I used was Google, which I just got for this month. I normally have T-Mobile, and I also signed up for the month-long international roaming plan for whatever extra charge that was. Between those 2 things, I had the 2 roaming plans, both of which have this property: You connect to the local cellular internet, but your traffic is routed immediately out of the country through your home carrier to whatever services you’re trying to access, and then back through that chain to you.

This basically meant that, in practice, there was no Great Firewall to worry about for me. I did also get a VPN as a backup, and that comes with its own eSIM. Those are supposed to be China-smart. I never really had to use that, to be honest. The 2 main carriers just worked very smoothly, and it was not a big deal at all.

I did sign up for all the Chinese services with this alternate Gmail, which I created just for the trip. It’s really important that you get the Chinese services. People tell you that you have to do it all beforehand because the Google Play Store won’t be available when you get there.

This is again where the Apple experience might be a bit better, because the Apple App Store is available because Apple is active in China. The Google Play Store is blocked in China to most Chinese users on local services. But when you’re using the international roaming plan, even the Google Play Store was available for me throughout, and I did download some additional apps while I was there. I had no problem doing that.

Nevertheless, I do think it’s smart to set up your technology as thoroughly as possible before you go, so you’re not fumbling around with it when you get there. There are a few apps that you definitely will want to have, and you’ll really, frankly, just pretty much have to have them to have a good time in China.

The ones that everybody knows about are probably WeChat and Alipay. WeChat is the super app that does everything in China. It’s your social network, address book, and content platform. There’s a whole bunch of content, a business platform, and the ability to transact with businesses and book businesses. I barely even got into that compared to the vast universe that it is.

It also handles payments. Then there’s Alipay, which is another major payment network in China. It has tons more features and is a super app of its own that I didn’t really use, although it’s not used as much for connecting and communicating with people.

WeChat is the thing that people use nonstop, where you have your own QR code and they have a QR code. There’s sort of the question, when you’re connecting with someone, “Am I going to scan you, or are you going to scan me?” Either way, there’s this quick scan process.

The apps recognize the QR codes so fast. You don’t have to tap on them or whatever. You just go into scan mode, and you don’t even have to have the whole QR code in frame for it to recognize the QR code. That stuff works super well.

You need WeChat to be able to connect with people and also make payments. You need Alipay as a secondary payment system. Both now support international credit cards, so you can load up a couple of credit cards into each one. With that done, you’re pretty much good to go for making the vast majority of connections and purchases that you might want to make in China.

The other app that you’ll want to have for sure is DiDi, which is their equivalent of Uber, and it works basically the same way. There are some small but interesting differences. When you’re quoted a fare, for example, at the beginning of a ride, if it’s a small enough fare, you don’t actually pay in advance.

Instead, you get a rough quote, you agree to that, and then you actually get the final number at the end and pay after the fact. You don’t pay until after the ride. Many times I was going to get my next ride, and then it was like, “Okay, you’ve got to pay for the last ride first.” It’s an interesting, small difference. It didn’t make a big practical difference in my overall experience in the big cities.

I was only in Beijing and Shanghai, where I actually called DiDi for rides. It’s a very similar experience to Uber in a big city. It’s a very liquid market, with fast pickup and a bunch of different levels of car that you can get—from the most basic economy to priority, faster pickup, premium cars, luxury cars, and XL. That experience is very familiar to anyone who’s used Uber.

The biggest difference, honestly, is that it’s really just cheap. The number of cross-town, 30- to 40-minute rides that I took that came in the $6–$8 range is really pretty remarkable. It would easily be close to 10 times that much, if not fully 10 times that much, for some of these rides in the United States. But that was simple to do and worked easily.

One thing I can say is that it was actually hard for me to confirm that my credit cards were going to work. I loaded them up into the apps, and they were accepted into the apps as payment methods. But then there are all these different ways that Chinese people use to move money around in the apps that are not accessible to international cards.

I made a few connections on WeChat via people that I knew in China. They introduced me to a couple of friends of friends, and I got connected to them on WeChat even before leaving. When you go in there, the menu includes things like “Send money to this person.” I tried to do that just to make sure that my credit card wouldn’t block a payment because it was on this unfamiliar Chinese app.

All those features pretty much are not accessible to international users, for whatever reason. I’m honestly not sure why, and I’m pretty sure that’s a WeChat or maybe a Chinese government decision. It was not something that my credit card was blocking. You just get to the final step in the app, and it says, “Your payment method is not supported to make this kind of transfer.”

So, I actually wasn’t able to do any transactions in WeChat before getting on the ground there.

My very first order of business when I landed in Beijing at 3:00 a.m. was to go to some little kiosk at the airport and make sure that I could make a payment before I left, because I wanted to make absolutely sure that I was able to transact. China is pretty much a cashless society at this point. I did not see any cash at all. A lot of it was there the entire time, and I actually wanted to bring a few bills home for my kids as souvenirs.

There was never an opportunity when it was like, “Oh, you have cash. Can I somehow get some cash from you?” The one time when I actually did do it was when I met a friend of a friend for a meal and mentioned, “Is there anywhere I can get cash?” One of the guys I was having lunch with said, “I think I have some at my desk. Let me go back up to my office for a second. I’ll see if I can get some at the desk.” He was like, “We really don’t use it,” which I knew at that point, but he said, “I’ll go see if I can get some.”

Sure enough, he did have some, and I was able to exchange a few American dollars in cash for some Chinese cash. Other than that, I did not touch cash. I could easily have gone the entire time without touching or thinking about cash. Online, what I was finding was that people were saying a lot of people don’t even take cash.

What I was told on the ground there is that the law requires businesses to take cash. So if you were to bring cash and change money at the airport, I guess that might work. But it would definitely put you in a very different user experience than what people around you are having, let’s say. I wouldn’t be surprised if there are some places where they either don’t want to take it, or where it’s an automated thing that’s just not set up to take it at all, if you’re buying tickets in various places.

I would imagine that there could be some things that could be quite tough to do with cash these days, because it really is super cashless, which is super convenient and also definitely enables a lot of visibility into what kinds of transactions are happening in the economy. But it made me a little nervous showing up without knowing if my credit cards were actually going to be supported.

Sure enough, when I landed, I went to the first kiosk I saw and picked up a fruit drink that looked pretty interesting. I think that drink was under $2 at the airport for a kind of fresh juice that would have easily been $6 or $7 at a U.S. airport. Then I was fumbling around with my first transaction, trying to figure out how to do it because it was a self-checkout. There was no person there to help.

I scanned my product, and it pulled up the price. Then I was like, “Okay, what do I do now?” Eventually, I figured out how to show my payment QR code to the camera in the machine. It’s actually by showing my QR code to that machine that it registers, “Okay, this person is paying for this.” As soon as you do that, you get the beep. Then you look at your phone, and it says, “Processing payment.” A couple of seconds later, boom, the payment is processed and you’re gone.

That’s quite interesting and quite different from, of course, how China used to be. A lot of people have either experienced it themselves in the past, if they went years ago, or have heard that there’s a lot of pushy sales going on in China, that there’s a lot of haggling, or that there are a lot of rip-offs. Honestly, I would say none of that was an issue at all.

In today’s world, the economy is really very formal in the sense that it’s all pretty much by the book. Everything has a price. People really didn’t hassle you much at all. I had just a couple of very minor moments where anybody tried to talk me into buying something.

I did go to some tourist places. It’s not like I didn’t go anywhere—I went to the Great Wall. I went to some places where there were definitely souvenir shops, where in many parts of the world you would be aggressively approached and people would try to talk you into buying things. It really wasn’t like that. It was very low-key and low-pressure.

When you did go to buy things, they had prices. You really weren’t expected to negotiate or haggle at all. There really wasn’t much opportunity for that. There also really wasn’t, in my experience, any rip-off. It was all just, “Here’s the price. Boom. Here’s the thing. Here’s the place to scan. Here’s the payment code,” and boom, you’re done.

In that respect, it was all extremely modern. Honestly, I would say that in a number of ways, China feels like maybe the most modern place that I have ever been.

One other app that you would definitely want to get is the Ctrip app. This is kind of the national tourist platform for China, and it works in other places too. It’s the way to book things in China, whether that’s hotels, attractions, plane tickets, or train tickets. It’s all kind of there, and it’s all pretty centralized.

This is a theme in China in general. You’re going to use these very standard platforms that everybody uses, that everybody’s on, and that, in my experience, work really well. So Trip.com is another thing to download before you go. I booked a high-speed rail ticket on it. I booked a plane ticket on it at one point. I booked the ticket to the Great Wall on it and booked a bunch of different hotel rooms on it.

It was all super smooth. Again, your payment methods are loaded in there before you go. Then, when it’s time to book things, it’s just a couple of clicks. When you show up to places, you’ve got your tickets right in the app. You just open up your app and, again, it’s usually a QR code. You just go, boom, here’s my QR code, and you’re admitted to things.

The level of integration and the ubiquity of these platforms is really hard to overstate. Obviously, in the U.S., we have very similar options where people can pay each other via apps. We’ve got Venmo and Cash App, but they’re a young person’s game. They’re not the kind of thing you can just expect everybody to take.

In China, it really is. Everybody is on the apps. Every person of every age is on WeChat. There was not one person who was not on WeChat that I encountered. I don’t know what percentage of the population is on WeChat at this point, but it has got to be super high.

Similarly, with the payments, of course, they’re all on Alipay. All the businesses—everything you might be interested in from a tourist standpoint—are on these platforms, and you really can manage all your business through the phone.

Meituan is another one that I think is probably worth downloading. This is their delivery app. It’s sort of an Amazon-like everything-store experience, but with more of an emphasis on food delivery and just really quick delivery of all kinds of things. It’s cool and definitely a high-quality user experience.

All the other apps that I’ve mentioned—WeChat has an English-language interface that you can use. You just need to switch the language to English. Alipay doesn’t have a full-on English version that I found, but it does have a translate button that kind of hovers around. I’m sure that’s not there for everybody all the time, but it somehow identified me as maybe needing the translate option.

There’s this translate button that you can just press anytime to see a translation of whatever it is you’re looking at in the app. DiDi has an English version. Trip.com has an English version. All these things are very easy to use.

Meituan does not, as far as I know, have a fully English version, so that one can be a little bit overwhelming to use. I had a great experience using Google Lens. This is actually one of the things that is probably a little bit nicer about the Android phone: you take these screenshots, and then immediately you’ve got the Google Translate options to translate whatever text is in the image that you just took a screenshot of.

At times, I was definitely taking a lot of screenshots of interfaces to see what it was that I was looking at. Even just to order in a restaurant, a lot of times you would do it through Meituan or another app. They put QR codes on the table to pull up the menu. You scan the QR code to order, you do it in the app from the table, and then you pay again through one of the major payment platforms.

All that happens on your app. You’re not even ordering by talking to a server at any point in some of these restaurants. Then, a little bit later, your food comes, and that’s it. Your entire transaction was mediated on the phone, and in some cases that involved a lot of screenshots and translations.

So I definitely recommend the Google Translate app. The Microsoft Translator app is also well used, and then everybody in China has the Translate app on their phone as well. Every DiDi driver, every service provider of any kind, and certainly everybody at a hotel concierge has it.

Usually at a hotel concierge you’ll get decent English, although not always, in my experience. But they always have the Translate app there, and they’re just super quick to record themselves speaking in Chinese and then show you what they just said. You can go back and forth that way, and that works really quite well these days.

It’s strikingly easy to travel through a place like China and not really feel like you’re lost. Between the map apps, the super apps, and the translation apps, I would say it’s still an adventure. But imagine what it would have been like 20 or 30 years ago to go there, be totally immersed in a very different place, and constantly need to ask people around you for help while trying to negotiate these situations back and forth in broken language.

Basically, all of that is history now. It’s really quite smooth. Just don’t lose your phone, and you’ll be fine. Even then, we can talk a little bit about losing phones, how easy it is to get them back, and how well a lot of their systems really work.

So, that was my whole parallel-universe setup, right? I've got this different phone, a new Gmail account, a new Signal account, a new WhatsApp account with that new email and that new phone number, a WeChat account, and an Alipay account. Load up the credit cards into those, get the DiDi account for rides, get the Trip.com account, and load up a couple of cards into that so I can book anything that I need to book. With all that stuff, and especially having two U.S. phone numbers there with international roaming plans on, the tech experience was really incredibly smooth and highly enabling throughout the entire trip.

What I did for contact with home base, because I do have a pretty good agent setup at home, is I left my main computer on. I left my Mac mini on, and I instructed Claude to monitor my communications, which it already has access to. It has access to my Gmail. I've done previous episodes about this, so if you haven't heard those, you could maybe go back to the Daniel Miessler episode to get a more thorough walkthrough of my particular agent setup, which is working quite well for me.

Because of this pretty extensive setup that had already happened, I was able to just tell Claude with a couple of prompts, "I'm going to be gone for the next couple of weeks. Here's what I want you to do: monitor my communications, send me a digest a couple of times a day, and let me know if there's anything urgent or anything that needs my attention in general." Then I would be able to respond back to it from my travel email.

I also set up Outlook as well because, again, while Gmail in theory is blocked in China, Outlook is not. You can use Microsoft products, and Outlook is available even if you're behind the Great Firewall. For me, this was not an issue at all. I never had to use Outlook. I just used Gmail, and I had perfect access to it all the time because I was using my roaming plan. But I did have that as a fallback.

So I told Claude, "I might respond to the digests, and I might also send an email. I might send an email to my own home address, and I want you to respond to those emails and do whatever it is that I ask you to do." It initially said, "Okay, we're going to need some sort of..." Claude can get really into the security mindset, and I did have a fairly extensive back-and-forth with it about how much of this stuff I really needed to do and exactly how I should go about it.

At one point, it had a password set up where, when I emailed in, I was supposed to give it a password. I said, "I don't think I want to be giving you a password every time. First of all, if anybody is hacking my stuff, if they are reading my emails somehow on this burner phone, they're going to see the password. So that doesn't make a lot of sense to me."

I said, "Why don't you just apply a sensible filter? If the requests seem reasonable, if they seem like they're coming from me, and if they seem like they're low risk, just run a check on all the inbound messages that you get from me. If they pass those checks, you can just do it. If they don't, then you can issue me a challenge and ask me some trivia that only I would know the answer to, or whatever, based on all the work that we've done together." And it agreed to do that.

Of course, that never came up, but I had a pretty good time of it. I was able to get these digests, and I had Claude responding from my main account. One thing that I wanted to make sure of was that if I got an email and wanted to respond to somebody, I didn't want the response coming from a different address, because it might get filtered out of their priority inbox or they might just be confused by it.

Instead, I had Claude respond as me—which I normally never do. When I say "as me," I mean respond from my address—but I instructed it to introduce itself as Claude, which was funny at times. It was like, "Hi, I'm Claude, an AI assistant." I was like, "Okay." At one point, I gave it feedback: "People know who you are, Claude. You don't need to tell everybody that you're an AI assistant. At least among the people in the circles that I run in, that's known."

But it would explain, "Nathan's traveling in China. I'm monitoring the communications. I'm copying his travel email here. He can respond." There were a couple of times when somebody sent me something where I was like, "Let's leave it up to them. You respond to them. Tell them what's up. If they want to add me on this, knowing that I'm in China, and take whatever risk is associated with that—which I think is pretty minimal—we can let them decide."

There were a couple of times when we did that, but most of the time Claude was just monitoring and giving me these digests. There wasn't too much that was urgent that I needed to do, but when there was, I could respond to Claude and say, "Okay, in this case, it's fine. Loop me in," or even, "Just take care of something." And Claude did a pretty good job. I think there were a couple of little rough parts, but we gradually ironed those out.

I think at this point, if I were to go back, it would be really easy to just reenable Claude and have it monitor my communications. Really, the only thing that I had to have anybody help with was that, on a couple of occasions—I think it happened twice—my Gmail authentication expired. Then I had to have it refresh, and it didn't seem to want to refresh its own token through the browser.

I'm going to need to give it some instructions to do that when it needs to, because it can use the browser, of course, but it was getting a little stuck on that. So when I spoke to my wife, I said, "Hey, could you do a quick reauth of my Gmail? Just go to my computer—it's on my desk—and tell Claude, 'Hey, we need to reauth Gmail.'"

She would open up a new Claude session locally in the terminal. I still use the terminal as opposed to Claude Cowork. She would tell it that we needed to reauth Gmail, it would pop up in the window, she would click through the prompts, and we were back in business. That takes 2 minutes.

So that's my tech setup. I think for folks in the AI space, it's easy enough to do that. I guess it's probably recommended—it's definitely recommended by people who know more than me about information security. I would just say it's not that much trouble. Know that the firewall is not really going to be an issue.

I think having 2 lines and a VPN backup is probably just prudent, but with a cheap burner phone, an extra line, and a VPN backup, you are in, I think, a very good place from an information-security standpoint. You can have your agents run your life from home. For me, this worked really well.

I am telling my parents, who do some international traveling now that my dad is retired, that they should go to China. What I would suggest for them is much simpler. I don't think they need to do all this; they're not in the AI game at all. I told them maybe they could just get a second number if they wanted to have redundancy, or the two of them could each have their own number. That—2 numbers—should be pretty good.

Just download WeChat, download Alipay, get Trip.com, get DiDi, and you're good. Your internet will work. Make sure your international roaming is enabled, and make sure you have enough data to support the time that you'll be there. Then you're good to go.

I don't think the average person really needs to take any extreme precautions, but for those in the AI game, the setup that I went through is simple enough, and you can have it done, honestly, in just a couple of hours. It took me longer to figure all this stuff out because I found, again, that the searching and the AI answers contained a lot of FUD—a lot of fear, uncertainty, and doubt—about what's going to work, whether you're going to be behind the firewall, whether VPNs go in and out of business, whether they're illegal, and so on.

But in the end, that setup was all pretty simple. So I would say you can do that, and you should be pretty confident that it's going to work pretty well.

A couple of other little details: You don't even need a power adapter in China. The outlets that they use, at least in all of the hotels and major places that I went, were universal outlets that would accept, I think, any major type of plug. Of course, I didn't try every major type of plug, but certainly the U.S. standard 2- and 3-prong plugs were perfectly well supported, as were a number of other types.

I didn't even have to take an adapter. It really was super, super, super simple to just bring my normal device with a normal charger. Very easy.

I think it's also worth bearing, or talking a little bit, about just the level of tech integration, to drive home how much they really have tied everything together. This is good in the sense of providing an incredible user experience. In some ways, it might be bad to the degree that you're concerned about surveillance, and I think the surveillance definitely has real upside but also real downside potential for the Chinese public.

To illustrate how thorough the integration process has been, when you call a ride on DiDi, as with Uber, it'll say, "Okay, we've matched you to a car," and here is your driver, their license plate, and whatever type of car they're going to be driving. Then you can see them approaching on their route in the same way that you can with Uber, as you're all familiar with.

But when the car that you're waiting for is stopped at a red light, you can literally see the red-light countdown clock in your app. That red light, which has a timer on it, is connected to some centralized system that DiDi can in turn call, and you're literally able to watch the countdown until the red light is going to turn green and your car can start making progress toward you again.

I thought that was really quite impressive, and it's indicative of the level of integration that they have across so many information systems. Another one that was pretty interesting was in one hotel. As soon as I would put the key into the thing for the lights to turn on—some hotels have this, where you have to put your key into a little slot in order to enable the rest of the room to work—

As soon as I did that, in this one hotel, the TV turned on and started yapping at me, which I thought was a little bit weird and annoying. But then that gave me occasion to notice that the TV was showing me the status of the laundry room in the hotel. In this particular hotel, they didn't offer a laundry service, but they did have a room downstairs where there were a few washers and dryers that were free for hotel guests to use.

And then, of course, this raises the question: Is it available now? I wouldn't even think, in the rest of the world, of any other way to answer that question than going to the room and checking it out. But here, they had wired it up such that the status of the laundry room washers and dryers was on the TV screen when it turned on in the room. So again, this just gives a flavor of how thorough the tech integration has been.

The Meituan delivery experience is also another indication of the user experience that you can expect. I had a weird experience one night in the middle of the night where I woke up with an allergic reaction ongoing, and I had never had allergic reactions before, so this was quite strange to me. But the next morning, I told my friends about it, and we were going on a bus from one place to another when they said, “Would it be helpful to have some Claritin or whatever?” And I said, “Yeah, I guess it probably would. Hopefully it won't happen again, but if it does, I would probably benefit from that.”

Sure enough, when we got to the destination—it was only about 20 minutes later—there was Claritin waiting for us at the destination. The woman I was talking to had just gone on the Meituan app, requested this particular medicine, which is available over the counter, and given the address that we were going to. Boom, it was delivered there and beat us to the place. So that's how fast some of these deliveries can show up. You can order ahead and have your allergy medication arrive at your destination with even just 20 minutes' lead time, and it can beat you there.

It was really very impressive in all those little touch-point, user-experience sort of ways. Again, I did feel like it was the most modern-feeling place that I had ever been from a technology perspective. Another kind of interesting thought experiment that I've been running is: If you were just to go to New York or San Francisco, or both, and then separately you were to go to Beijing and Shanghai, or both, and you didn't know anything about the countries or the history—and probably would also have to mask the prices, because the price differential is definitely significant—and just spend a week in each of these places, and then ask which of these countries is the richer country, I think it would be pretty clear to most people.

I think you would get, honestly, an overwhelming majority of people saying that China is the richer country because it does have newer stuff and a more integrated, lower-friction experience that just feels like the future. It's ubiquitous across at least the major cities that I've been to, but from what I understand, it really has extended across China to a remarkable extent. So we'll probably come back to that maybe a little bit later today, and definitely a little bit more in the next 2 episodes that I plan to do about this.

As a kind of bottom-line, what-does-it-feel-like-to-be-there litmus test, I think it does feel more modern and, in some ways, more developed than life in the United States does, which is a pretty remarkable accomplishment, obviously, for the Chinese government and Chinese society as a whole.

Okay. So in Beijing, I did a couple of days' worth of tourism when I first got there, before the real AI-focused part of the trip opened up. Again, a couple of high-level comments: Beijing is absolutely a world-class city from a tourist perspective. If you're going there for no reason other than to see some history and see some famous sites, it is absolutely a world-class destination. The number of incredible sites is enough to keep you busy for, I would say, easily a full week.

I would put it on the level of Paris in terms of just everywhere you go, you're looking at incredible, deep history with beautiful architecture and really remarkable sites all over the place. Mostly, I did these on my own, either catching a DiDi to the place or walking to the place, with no tour guide. I did hire a tour guide one time, and I'll tell a little bit more about that, but mostly I was just using AIs as the tour guide.

I decided to leave my main Claude and ChatGPT accounts at home because I had this whole alter-ego strategy. I figured I would just use Chinese AIs for this trip. You don't necessarily need to do that, especially because the Great Firewall is not such an issue if you're using an international roaming plan, but I figured it was a good opportunity to get a little bit more familiar with Chinese AIs.

So I downloaded DeepSeek, the Kimi app, and MiniMax all to my phone before leaving. Those were the 3 AIs that I mostly used as I was going about my couple of days of tourism. I used them as both travel planners—for example, I was planning to go to the Great Wall the very first thing directly from the airport, but it had been raining heavily, and they actually closed a couple of sites at the Great Wall, so I wasn't able to go. I had even booked a ticket, but again, the Trip.com integration is quite good, so I got a notification from Trip.com saying, “Hey, you can't go to this thing today because it's closed. Your ticket's been refunded. This is due to heavy rain. Whatever. It'll be open again tomorrow.”

I was like, “Okay, my plan is immediately thrown a little bit into flux. I'll go to the AIs and say, ‘Hey, here's where I am in Beijing, here's what time it is, and here's what day of the week it is. What should I do?’” And they were really quite helpful. I think these experiences do illustrate that you don't necessarily need frontier intelligence for many quite valuable use cases.

I was using Kimi 2.6 to serve as my tour guide for all these different places. There were also a couple of notable model releases just in the last couple of weeks while I was there. That's not a coincidence, because I went in part for the WAIC event, and this was a big moment in Chinese AI. President Xi headlined the event by opening with a keynote, so naturally, a couple of model companies wanted to time their releases to this event. We got Kimi K3 right around the time of the event, but my little tourism was a little bit before that, so I was using Kimi 2.6.

These were very good experiences that I found to be very useful and informative as tour guides. The fact that there might not be frontier models capable of doing long-horizon agentic tasks and capable of various cyber exploits or what have you really wasn't relevant. They were perfectly good and really quite fast in their responses in many cases—perfectly good to educate me about what I was seeing.

Going through all these different places, from the Lama Temple, which is a Tibetan Buddhist temple right in the heart of Beijing, was a surprise to me. I was like, “Oh, wow. Wait, what? This is a Tibetan Buddhist temple?” I didn't expect to see that here. It had pretty good historical explanations for how this came to be.

I was asking, “How should I—” This does go to show my ignorance of China. I wasn't under the impression that there was no religion in China. I've seen enough stuff on TikTok to know that there is religion in China and that people do practice religion. My sense is, broadly, that as long as it's not challenging the state, you're permitted to participate in most religions as you wish. Here, we even had a Tibetan Buddhist temple right in the heart of Beijing, so I was getting the history on that.

Then I went over to the Temple of Confucius and the Imperial College, and again, I had a lot of questions. I was like, “Wait a second. Confucius—he's not seen as a god, right? He was just a philosopher, wasn't he? Why does he have a temple?”

It does show some fundamental paradigmatic differences between Eastern and Western society. It's like, yes, he is a man. He's not considered to be a god, but he also is revered. He's not worshiped, but he is the object of these ritual behaviors or ritual traditions. That was not just something that some people did some of the time, but was even understood to be part of the obligations of the emperor.

The emperor was meant to be sort of related to God. The son of God might be a little too Christian a framing, but definitely some sort of divine connection. The emperor also had duties, and it was the emperor's job to do the right rituals in the right way. Those, over time, came to involve giving offerings and whatnot at the Temple of Confucius.

It's a kind of coloring outside the lines as we know them in terms of what is religion and what is state. Does this blur the separation of church and state? But it's also not exactly religion in the same way. It's kind of unfamiliar stuff, but the Chinese AIs gave really helpful explanations. I thought they were perfectly good.

And, by the way, I didn't even pay for them. This was all free usage across all 3 of these apps, and I had a pretty good time of it. There were some issues with refusals, and at times I thought probably even by the standards of what they wanted, there were over-refusals. I would say that Anthropic is not the only company that currently has an over-refusal issue.

I was asking some—I wouldn't even say sensitive—questions. But when you get to issues of Tibet and Taiwan, you're going to get some. They call it the 3 T's.

Tibet, Taiwan, and Tiananmen. Those are obviously sensitive issues in the Chinese context. I wasn't asking questions just to probe what the AI systems would answer or not answer. I was really asking earnest questions about what's going on with a Tibetan Buddhist temple here in Beijing, which obviously touched on issues of Tibet in a kind of tangential way.

I wasn't asking about the status of Tibet today; I was asking more historical questions. There was another question because I don't know a lot about Chinese history. I was asking, "Okay, who built this? And what's the deal with the Ming dynasty and the Qing dynasty?" The latter is interestingly a minority dynasty that came in and took over.

The Qing dynasty was actually the Manchu ethnic group, which was just north—I think northeast—of the Chinese heartland. They managed to come in and take over the empire, conquer the place, and then rule it for hundreds of years, which is an interesting and obviously pretty impressive accomplishment. This Manchu ethnic group has its own language. You see this reflected in some of these artifacts, where there would be multiple different languages.

You would see Mandarin Chinese, you would see the Manchu language, you would see sometimes the Tibetan language, and I think sometimes also maybe a western Chinese language. There were often four languages that were used and equally represented in these monuments or proclamations that were going out to the empire. One time I asked, "What's up with the Manchu group today? Are they still around? Are they still part of China?" And there you would get the classic old Sydney Bing experience, where the model itself would answer the question and you'd see the chain of thought.

You might even see the beginning of an answer, and then sometimes, as with this Manchu question, because you're touching on something fairly sensitive in the Chinese political context of ethnic minorities, the answer would be in the process of unrolling in front of you. Then you would get this pullback. You'd see the answer disappear, and then you'd see, "I'm sorry, I can't help you with that. Ask a different question," or whatever.

Again, this wasn't something that I think they necessarily meant to censor. Not all of the apps censored it the same way. I was able to get some answers from some apps that I wasn't able to get from other apps. It goes to show that, in many ways, the Chinese ecosystem has a lot of the same things going on that we do, where people are trying to make sure that they're on the right side of regulators, on the right side of history, and acting responsibly as they understand it.

At times, that can go a little too far and interfere with the user experience. But this is also consistent with the reporting that I've seen in many places: The Chinese models themselves are not very censored; it's the actual deployed application experiences that usually contain the censorship. You could see that as the answer was partially there and then was retracted sometimes, and you'd be hit with, "No, I'm sorry, I can't help you."

Overall, I'd say it was a very good experience with the Chinese AIs. Of course, I'm doing this all in English. They do have some weird quirks. In DeepSeek's advanced mode, with the best DeepSeek model, if you go into expert mode, then you can't attach any files, and there's also no search enabled.

That's kind of weird. I'm not sure why that's the case, but that is the case. Depending on what I was trying to do, I might just use the instant mode on DeepSeek. On Kimi, that issue is not there. You had the ability to have the model do searches even with the best model that was available at the time.

MiniMax, I think, was probably the most consistent and reliable—just a very down-the-fairway experience. I think I had probably the fewest over-refusals from the MiniMax app. I did find occasion to use all three. For my taste, Kimi is the best of those apps, but there were times when I got better answers from other apps, if only because of the over-refusal issue.

You might say Kimi is the most Claude-like, and that obviously brings to mind issues of distillation. I think I'll probably talk about that a little bit more in future episodes than this one, but Kimi, I would say, had the best vibes and the best taste. That may be partially related to accusations of distillation from Claude, but I think it runs a bit deeper than that, too.

From what I know of the company Moonshot, they have a real aesthetic, and their name itself was inspired by Pink Floyd's Dark Side of the Moon album. They've got graphics that echo the prism art from the album, the famous album cover from Pink Floyd. They're definitely an organization that takes the aesthetics of the AI that they're building very seriously. I think there's a lot there that is more than just distillation from Claude, although I do think the case is fairly compelling that at least some distillation is going on.

We can get into that probably more in a future, deeper dive into some of these U.S.-China issues. I guess one other note about how much I trust the Chinese AIs is that when I had that allergic reaction in the middle of the night—which was, fortunately, very short-lived but a very strange experience—I just woke up and my hands were itching like crazy. I was like, "What is going on? My hands are so itchy."

Then I got up, turned on the light, and looked at myself in the mirror. My whole upper-body torso was covered in hives. I thought, "Yes, that's pretty weird." What did I do in that moment? I downloaded ChatGPT, because I did have the sense that, for these tour-guide things, I was getting great answers. It was educating me, and I was pleased and happy with the experience.

But now we were in a possible health emergency. I didn't really know what was going on, and I wanted ChatGPT. I knew I'd had Karan Singhal from OpenAI, who leads health at OpenAI, on the podcast, and I'd obviously had deep experience using it with my son's case. I just felt like, in this moment, I was going for the best.

It helped that they had made their best health experiences available even to free users. I was able to go in and ask my questions, and I was very confident that I was getting the best experience I could get. I did not even create an account. I did not pay for a new account, and I did not log into my home account. But I wanted ChatGPT in that moment.

Fortunately, this allergic reaction that I experienced just passed. Even over the course of chatting with the AI for 15 minutes or so, through a couple of rounds, and just observing myself, I could tell that it was passing. It went from, "This is probably not an emergency, but you should probably go get checked out now," to, "If it's passing and you're feeling like the symptoms are already reducing, then you can probably just go back to sleep and sleep it off."

Anyway, that's what I ended up doing. But I did think it was notable from an AI perspective that, in that moment, when I was a little bit alarmed, I still wanted to go back to ChatGPT just for the confidence that this was probably going to be the best answer I could get. It was worth the extra effort to download the app to do that.

Let's see what else I want to talk about today. I think I'll just make some observations about Chinese society. I'm not going to turn this into a travel log, but I do think it's worth noting that people are just incredibly nice. The hospitality that I experienced at every turn was outstanding.

I don't mean to contrast this with the West. I think one of the more heartwarming trends of the summer was seeing people from all over the world come to the United States, travel around, and go to some of the cities that hosted games—cities that they probably otherwise would never have gone to. Some people were even doing road trips from city to city, and people had a really great time and had really great things to say about the United States.

I think our people are also very warm and hospitable, and the track record from this summer and the reports from World Cup travelers really support that. Again, this isn't meant to be a contrast, but my experience in China was outstanding. People were extremely friendly. I had, I would say, zero negative interactions.

There was almost nothing that I would call problematic at all. As I mentioned earlier, there were very few pushy sales interactions, even. I can remember a couple of short moments when somebody tried to get me to ride his bike taxi with him. I really didn't want to, and he followed me for a couple of minutes trying to get me to do it.

I was like, "Dude, I'm okay. I'm happy walking." He got the message and went on his way. In many parts of the world, you would probably have a lot more of that than you did in China. It was definitely impressive and very smooth sailing in that regard.

Service is also generally really good. It's not a tipping culture. At no point are you invited to tip. The price is the price, and it is a flat price. When you do a WeChat checkout or even a DiDi checkout, there's no moment in that flow where you're asked, "Hey, would you like to add 10%? Would you like to do this or that?"

It's just not there. They're not expecting tips, and yet they're consistently delivering high-quality service. Service at the hotels I stayed at was really very helpful. I wasn't necessarily interacting with a concierge at most of the places I stayed. One place definitely did have a concierge and was super helpful.

In fact, one pro tip that I would give there is that the DiDi app, as it is served up in the English version to foreigners, doesn't have quite all the same features that the native one has for people with Chinese identification and bank cards and what have you.

And I'm not sure why, but it doesn't. One of the features that the Chinese version of the app does have, but that my version of the app didn't have, is the ability to hire a car by the hour. For one of my power-tourist days, that was definitely the best option. I was at the Summer Palace, which is a beautiful, beautiful site created by the Qing Dynasty that was destroyed by Europeans at one point and then rebuilt, but it's a very cool and beautiful place to be. I then wanted to go to the Forbidden City in the afternoon and to the Great Wall in the evening.

I tried to do that all in 1 day. To do that, I was like, I really need a driver who can take me to all these places, but how can I hire one? The concierge at this 1 hotel said, “If you do it by the hour, it should be fine.” I said, “I don't have the ability to do that.” They looked at my app and said, “Yeah, I know you don't. That's weird. We can do it for you.” “Okay, great. How is that going to work?”

“We'll just do it on our DiDi app here. We'll pay for it, you'll pay us, and then I'll also set up a WeChat chat between you, me, and the driver, because the driver probably won't speak much English. If you need anything or if you need any translation help or whatever, you can send a message into the group, and we'll just monitor this throughout the day and help you out as you need it. It'll be great.”

Sure enough, that was extremely helpful in multiple ways. Just renting the car in this way and having the driver who could take me to these places was not expensive. I think the whole day's rental was over 8 hours. It was probably 9 hours from start to finish that the guy took me around and waited for me the whole time outside the places that I went to. Altogether, I think that came out to something like $120 or $125.

It was really very affordable given what I was getting. The hotel paid, I paid them, and they stayed active in the chat throughout the day, just making sure that I was able to effectively coordinate with my driver. That was after I checked out of the hotel, by the way. That was on the day that I left the hotel, so I wasn't even coming back. They were still extremely helpful in doing that for hours to come.

Similar things happened at a couple of other hotels, too. One night, for example, it was raining pretty hard, and I was asking, “Is there a place I can go out for food that's close?” I had this 1 dish that I wanted to get, which was a crab noodle dish. That might have been the thing, by the way, that gave me the allergic reaction, because it was later that night that it happened. The guy at the hotel desk said, “You could go here, but you can also order delivery.”

I was like, “Yeah, but I don't really know how to do that.” He said, “I'll do it for you, no problem.” He again pulled up his phone, did it on his personal Meituan app, and showed me what he was ordering for me. He showed me the price and said, “Is that cool with you?” “Yep.” Then he was able to send me a little on-the-fly invoice from WeChat or Alipay, and I was able to pay him.

He got the money inbound for me, hit checkout on the thing, and the next thing I knew, I had delivery coming. It was super easy. Part of the reason that's possible, by the way, is that the transaction costs on payments in China are super, super low. They're basically an order of magnitude lower than the transaction costs that we have in the United States, where 2% to 3% of the credit-card transaction goes to credit-card fees.

To the point now that we have gas stations in the U.S., and a bunch of businesses will either have a minimum, charge an extra dollar if you want to use a credit card, charge an extra few percent if you want to use a credit card, or charge 10 cents more per gallon of gas or whatever. We've started to see businesses not wanting to accept that cost on their side anymore and putting it back on the consumer.

In the Chinese context, the transaction costs, from what I understand, are more like 0.25% of transactions. I'm not really sure how they're pulling that off. I did ask some AIs, and 1 answer is that they don't have rewards in the same way that we do. All your credit-card rewards and all that kind of stuff basically end up coming out of those fees.

I guess this is 1 way in the U.S. where, in a weird way, poorer users are potentially subsidizing wealthier users. We might have to go to patio11 for a full breakdown of exactly what's going on there. But it does seem like rewards accrue to more affluent credit-card users. There's not really rewards for payments in China, so that's 1 thing.

There's also credit. When we use credit cards in the U.S., by and large, we're putting it on our tab and then we pay the bill at the end of the month. When they're doing WeChat payments, there's not really a credit element there. There's an actual deduction from your balance. It goes to their balance immediately, and they know that you're good for the funds.

I had the impression for U.S. credit cards that the interest you were paying on balances was covering the risk that you might not pay. But maybe, to some degree, the transaction fees are also somewhat covering the risk that you might not pay. That's opaque to me. But again, that risk doesn't really exist in the Chinese context. They're just debiting your actual balance.

Fraud protection, as far as I can tell, seems to be about the same. In fact, it might even be easier there, because here we have this whole rigmarole of disputing the payment with your credit card. There, you can dispute a payment directly with WeChat or directly with Alipay. From what I understand—I never had to do this—you can often get reimbursed if there is any sort of fraudulent charge.

From what I gather, it happens much faster. It's a much more streamlined experience, whereas in the U.S. you're waiting weeks, and the merchant can file their counterclaim or whatever. It's a mess on both sides, frankly. Neither the consumer nor the business has a great experience with credit-card disputes. It sounds like it's actually a lot easier to do that kind of thing, at least on the consumer side, in the Chinese context.

Somehow, it all adds up to dramatically lower transaction costs. I think even peer-to-peer payments might have no transaction costs, which we do have some of in the U.S. as well, right? If I do a Venmo payment to my buddy, he gets the balance, and I'm not paying an extra fee on top. I think when I was doing this stuff at the concierge, there was just no money lost at all, and they were happy to do that. That happened a couple of times.

I also got tickets to a soccer game 1 time that way, by just going up to the concierge desk and saying, “I want to go to a soccer game. I have no idea what to do. Can you help me buy tickets to a soccer game?” They ended up doing it all on their account, and then I just paid them right as they hit the checkout button for the tickets. Boom, we're good to go. I thought that was all pretty cool and, again, just makes for a great overall experience.

Another interesting thing about China, I would say, is that there are really a ton of things that you see where you're like, “Boy, that is really similar to what we're familiar with at home.” For example, the Apple aesthetic is very prominent in China. I went to a mall and went to a number of different technology-company stores, including Huawei and Xiaomi. Both of these companies sell phones and laptops, and they look exactly like iPhones.

The stores are set up exactly like Apple stores. If you walked into these things not knowing where you were, you would definitely think you're in an Apple Store. You might start to notice that you're not in an Apple Store, obviously, because the signage is in Chinese, but also because these companies sell cars as well. It's like—imagine an Apple Store that also sells cars. Imagine if the Apple Car were actually a thing. That would be the experience of going into a Xiaomi or a Huawei store.

But the aesthetic is strikingly Apple-like. Another thing that jumped out at me is that they have a similar—you might call it a cult-of-the-founder-type attitude toward the leaders of their tech companies. You could maybe say that the cult of Confucius is the ultimate cult of the founder. There's still a temple to Confucius in Beijing to this day.

I was interested to learn that Confucius's direct descendants, now 79 generations on, still identify—they've tracked their lineage back to Confucius. They still identify as his direct descendants, and they still do certain ceremonies or rituals to honor him, I believe on his birthday, 79 generations on. That's pretty remarkable.

The cult of the founder: Mao is maybe another example of that. Mao is still on whatever cash there is, which is not much, but when I was able to get my hands on a little cash, Mao's face was on all the bills that I saw. It also happens at the tech-company level. The founder of Huawei is talked about in a sort of mythical sense. The founder of DeepSeek is talked about in this way. The founder of Zhipu AI, or Z.ai, is also talked about in this way.

It very much echoed the way that people in the U.S.—and I would say probably more positively, actually—there's definitely a sort of parallel to Bezos, Zuckerberg, and Elon. Elon would, of course, be the paradigmatic example of this in the U.S. They really do have a very strong sense in which they look up to these tech founders and leaders that absolutely parallels our own.

Going to WAIC was also just really striking. This was at the Shanghai Expo Center, and I've actually never been to CES, but I'd say in terms of scale, it was probably CES-scale. It's a huge event.

I did go to NVIDIA GTC last year, and that was the most recent big tech conference with these mega-booths from huge companies. This was exactly the same kind of vibe at the Shanghai Expo Center. Again, if you just opened your eyes and asked, “Where am I?” you might think you were in Las Vegas at a big trade show. The booth structure and the aesthetic were all very, very similar.

There wasn’t really too much to speak of from the actual WAIC trade show itself. I thought it was so familiar that, in some ways, it wasn’t really that interesting. You saw all the big companies that you would expect to see. They had booths showing all the things that you would expect to see them show. It was remarkable mostly for just how similar and otherwise unremarkable it was to be on the trade show floor.

One thing I was expecting, and maybe didn’t see as much of as I was expecting, was really cool robot demos. You do see more robots just out and about in China than you do in the United States. At one of the hotels where I stayed, there were a couple of robots in use to deliver meals from Meituan to the rooms where the customers were ultimately going to receive them. That was interesting.

I also saw some robots going around in parks—kind of sweeping robots. These were not humanoids. They were more like little R2-D2-style droids that either had a compartment to hold your food or a compartment to hold the leaves and pine needles, or whatever was being swept up into them. Those were pretty basic.

I think they were probably either specific to the space they were in, or programmed with relatively simple rules: go slowly, don’t bump into anything, and otherwise just cover the space. It was a little bit like your iRobot—RIP, iRobot—things like we’ve had in the US, but just deployed in public places, more prominent and more visible. So you did have those kinds of robots, but they weren’t particularly advanced in terms of the tasks they were doing or the way they were moving about.

Now, back to WAIC. I did have a chance at the Tencent booth to sit for a brief robot massage. It was the first time a robot had ever touched me in any meaningful way. I sat down in the chair, and the robot put one hand on the front of my chest and one on the back, kind of on the inside of my shoulder. It did a pressure-point circular massage.

It was a pretty short experience, maybe a 1- to 2-minute demo. It felt okay. I didn’t get any human massages, but I would say a human massage would still be the preferable experience. It wasn’t just phoning it in, though. It did have real pressure on me, and it was kind of interesting.

Aside from that, as I walked around, there were certainly a ton of robotics companies. One can expect that China will dominate the scaled production of all kinds of robots, including humanoid-style robots. That humanoid could be wheeled, or it could have legs, whatever. Of course, Tesla is going to give them a run for their money, and we have other companies that are really working on the AI—the general-purpose AI—to power these sorts of robots.

At the robot expo hall, it was a little bit weird. There were so many robots and so many demos set up, but not that many demos were actually working. A lot of them were paused. It was like, “Oh, here’s a robot that’s set up to cook,” but it wasn’t actually cooking.

One that I saw was going to do a magic trick. There was a guy there shuffling cards, and he was the human magician. He spoke good English, so I started chatting with him a little bit. I said, “So, you taught this robot to do a magic trick, huh?” And he said, “Yeah, I did.” I said, “What did that take?” And he said, “Oh, God, it took 6 weeks of just a ton of work, and I did the trick over and over again.”

I said, “So, you’re not worried about robots taking your job anytime soon as a magician?” He said, “Oh, God, no. It’s a long way off.” From what I could tell from the signage, they were only going to do this robot magic-trick demo 1 time. It was the next day at 11:00 a.m. or whatever. I thought, “Only 1 time? You’re only going to do this? You went to all this trouble, and you’re only going to show it off 1 time?”

I couldn’t figure out why that was. It’s possible that I missed something, that there was maybe some rule or that they were trying to have different companies stage their demos at different times, but I never caught any wind of that. It just seemed to me, broadly, that the robot hardware is certainly coming along, and people see the opportunity. The government has said that this is a priority, and so people are trying to answer that call. But the actual autonomy did not feel that far along yet.

This may suggest—and this could be something to come back to in a later episode—that there’s some opportunity for real complementarity between the Chinese ecosystem and what the American ecosystem is currently leading on. I suspect that, outside of maybe Unitree, which certainly has cool robots and does some incredible demos, the American companies are farther ahead on the actual model development for general-purpose autonomy in humanoid and other similar robot form factors.

I didn’t get the sense that the models on these Chinese robots were really all that good. There was some teleoperation. One time, I was walking down a hall, and a robot was walking toward me. I extended my hand as if to shake its hand, and then I looked up and saw a woman behind the robot controlling it with a remote control. She looked at me and just shook her head, like, “This thing is not going to shake your hand, buddy.”

I didn’t really expect it to, but it was just kind of, “Okay, this is not an autonomous demo.” This was showing that the robot could walk, but it was walking with teleoperation. I think that was where most of these companies were.

I didn’t see much that led me to believe they were very far along in autonomy. The fact that they were only going to do this magic-trick demo 1 time still made me wonder, “How is this some sort of sleight of hand within a sleight of hand?” I felt like if you could really do that magic trick at all reliably, or without some sort of really convoluted setup, you would probably do it a few times over the course of a 4-day major event like this.

As far as I could tell, though, it was only going to happen the 1 time. To be continued, I think, on robots—but I think there’s definitely some opportunity for complementarity there.

Other things that jumped out at me were extremely, almost uncannily, similar. The Chinese hyperscalers and the role they play in the tech ecosystem really remind me of the role that Google or Meta play in the American ecosystem.

I had the opportunity—this was just through a LinkedIn connection—to have my local Claude go through all the messages I’d received on all platforms, basically, and say, “Who am I connected with in China? Who follows me at Chinese frontier companies?” I sent probably 20 DMs to people who had either reached out on LinkedIn, followed me, or whatever.

One guy who had connected with me on LinkedIn worked at a Chinese hyperscaler. I reached out and said, “Hey, I’m going to be in your city, Shanghai, and I’d love to meet.” He said, “Could you come to our office for lunch?” I said, “Yes, I can. I’d love to do that.”

So I took my DiDi over there, and it cost me $7 or whatever to get across town. It was exactly like having lunch at Google. It was so similar, and I don’t think that’s an accident. I think they’re taking inspiration from Google in many ways: a big cafeteria, bring a friend, just swipe them in. I’m not sure what kind of registration process there was, but I had a guest badge and swiped in.

There was an overabundance of all kinds of different foods. Everybody was just eating at big tables in the cafeteria. I’ve had lunch at Google a number of times over the years, and it felt exactly like having lunch at Google. This was, of course, all Chinese, but it was an extremely similar experience at pretty much every step.

During lunch, I asked, “Okay, so what else is going on? What do you think I should do?” The guy said, “Actually, tonight we have an event for startups, which you could come to if you want to come.” My mindset in life, and particularly when I’m traveling, is to say yes to as many invitations as possible. So I said, “Okay, let’s go do that.”

Later that same day, I went to this startup mixer, and it had an uncannily similar structure to a mixer that a big tech company might throw in San Francisco. You had tons of founders running around. They were all pretty early-stage in this context. Some of them had launched, and some of them hadn’t even launched. They were all super eager to pitch you on what they were doing, and they were all hustling.

The big tech company was providing the space, the food, and the drink. Another interesting thing is that there was very little alcohol in China, in my experience. That’s obviously a small slice, but in the same way that people have been saying recently that alcohol is out of fashion in Silicon Valley, there wasn’t a lot of alcohol at any of these AI or tech events in China, either.

I think I had a drink mostly when I was hanging out with other Westerners from the AI safety community—maybe 2 times. Aside from that, there just wasn’t much alcohol going on at all. Clean living, I would say, is also a commonality right now between the San Francisco AI scene and at least the Shanghai startup mixer that I attended.

Again, you could feel so many similarities structurally in the way it was provided by this big tech company. They weren't selling anything or trying to get you to make any commitments to them. If anything, they were maybe going to give you free credits to sign up with their cloud or with their AI inference product. These folks were all running around hustling, and it was just like, boy, this is very much an echo of Silicon Valley.

There was another instance that was pretty similar. I don't know the university tech-transfer dynamic in the US super well, and I obviously don't know anything in China super well, but I had an opportunity to visit what I would describe as kind of a tech-transfer office that they described as sitting between academia and commercialization. They described the technology process as going from 0 to 1, 1 to 10, and 1 to 100. I don't know to what degree this takes inspiration from Peter Thiel, but 0 to 1, they said, is really at universities. That's professors and research groups trying to figure out what technology has any possibility of working.

Then they described themselves as sitting between the researchers and the market, trying to understand both what enterprise customers want, what the market at large wants, and what the technologists in universities are actually getting traction on. They were trying to bridge those two and help these technologies go through this kind of “cross the chasm” sort of thing, if you will. The 1 to 10 stage is where you validate that, yes, this technology probably can scale—not to actually scale it, but to establish that it can probably scale and that it meets the commercial needs that people have. Once they do that, that's when a company actually gets started and spun out to go do its thing and go from 10 to 100.

Again, that felt pretty similar to what I've seen in American tech-transfer contexts. There were even some social dynamics that I thought were notably similar, and I wouldn't want to overstate this. In the US, we have this phenomenon where women are outperforming men on educational-attainment indicators. The majority of people going to college these days are women, and you've got this phenomenon of women feeling like, where are the eligible men who are in my same social or economic class?

That also seems to be going on in China. There's a similar phenomenon where younger women are achieving at higher rates educationally than their male peers. At a number of the different places that I visited, the most senior person representing the organization was indeed a woman. You don't see that necessarily all the way up the very highest echelons of power in Chinese society, which you also don't necessarily see in the very highest echelons of American society either. But it was notable to see this kind of similar thing, where women are outperforming men.

“Where are the men who are on my level?” was a question that they were asking, and the women are going places. Can the men keep up? It was a question, I think, in both societies right now.

There are definitely some differences worth noting, too. One big one is that the way things happen in China with respect to how you advance in the political system is very different, and I think it's actually quite important. There's something in China called the mayor economy, which certainly you can find better experts to explain than I can. But the key insight is that to advance your political career in China, you have to please the party. You have to please the higher-ups. There are no local voters, in general, that you have to please and get their votes.

So this creates quite different incentives from what we have in the US in a lot of cases. In the US, of course, if you're a mayor of a city, you are accountable to the voters of your city, and it's up to you to figure out what strategy you want to play to win support from your voters. If your voters don't like the current president, then you can get ahead by attacking the current president. If, as we saw in California with respect to Elon, there's a super-successful businessperson in your jurisdiction, but you think you can make political hay by attacking that person, you can do that. Maybe if they're Elon, they might leave, but maybe that still redounds to your benefit. Maybe you have a voter base in your particular area that supports you for speaking truth to power, or whatever—they might see it that way.

In China, it's very different in this respect. If you are the mayor of a local city, you are essentially in competition with the mayors of other cities to prove that you are the best at running a local city government and, to some extent, the economy. Of course, there's tremendous market activity in China, so this is not a command-and-control system. Hopefully, that's been clear from everything I've said so far. It is hypercapitalist and hypercompetitive in some ways.

But the local political leadership sees itself in competition with other people at a similar level, and the people they have to please are above them. There are no voters. They want to make the citizens successful and prosperous, but doing that will get them the approval and promotions from the people higher up in the party. I do think that creates quite a different dynamic. Obviously, you're never going to have local officials trashing the national officials. That is probably obvious.

But I would also say the relationship between government and local businesses is quite different. The local mayors, or whatever kind of local officials we're talking about, want to see their local companies succeed, and so they're investing in them. They're trying to help them in whatever ways they can. They want their local electric-vehicle maker or their local robotics maker to be one of the ones that ultimately survives.

The way that technology proliferates in the US can be illustrated with a story I tell all the time about how cable companies developed in the US. It happened in one place, and then people said, “Hey, that's a pretty good business model. I'll go start up the cable company in my city.” Initially, we had a huge number of cable companies. We still do have quite a lot, but there's been a lot of consolidation over time.

Something kind of similar happens in China, where it's more often a kind of strategic directive: “Hey, we need to lead in electric vehicles. We need to lead in humanoid robotics. This is important.” That message goes out, and now it's entrepreneurs who start these companies, but it's also the mayors of the cities or the provincial leadership, whatever, who are thinking, “Okay, I want to get promoted. The big man says electric vehicles and humanoid robots are really important. How can I help my local companies succeed?”

If I can do that, that's going to be a major feather in my cap. That's the kind of thing that's going to get me promoted. The incentives for local leadership, I think, are very different and, in some ways, more pro-business than they are in the United States.

Of course, we do have incentives for local leaders to support local business in the US as well, but I think it's a little more complicated when you look at what Mamdani is doing in New York. His rhetoric is not super-friendly to some of the biggest taxpayers in his jurisdiction. I wouldn't say anybody thinks he's a great friend to finance. He's not necessarily attacking finance as much as you might suspect based on some of his other democratic-socialist principles or pronouncements, but he's certainly not championing finance either, even though it's a huge taxpayer and a major industry in New York.

Again, tell me if you think I'm wrong—correct me in the comments—but I think the mayor economy creates a more supportive business environment because people know that the competition is going to be intense. It is not the case that you have, with maybe some exceptions among the older state-owned enterprises, that when China wants to enter a new industry, it does so with state-owned companies. These days, when China wants to enter a new industry, it is doing it with private-sector innovation, and the local political leadership knows that this is happening all over the country and that most of these companies are not going to make it.

So if we want ours to make it, we better do our part to support them. That did feel like quite a different vibe from what I was able to learn from the American context.

Another funny thing, in terms of surprises and notable differences, was an unbelievable number of generally remarkably clean public toilets, especially in Beijing, but I think in major cities throughout China. This was something that the national government decided was a priority at some point, or at least one of their priorities, and so they've built a lot of clean public toilets.

In Beijing in particular, I was astounded by how often you are passing a public toilet. When I used them, they were really quite clean, and there was no fee or anything attached. They were just public toilets available everywhere: clean, ready to go. That is obviously not the case in New York or San Francisco. You have to find a business that you want to make a token purchase from or whatever if you're going to use their bathroom in many cases.

In Beijing—and again, I think in most of China, but in Beijing to an amazing degree—there were just a huge number of public toilets always available. Another big difference, of course, is surveillance. They do have—it's obvious, I think, to everybody—that there's just a huge number of cameras around all the time, indoors and outdoors. Even sitting in a restaurant, you look around and there are multiple cameras covering the inside of restaurants. They're covering the outside on the street.

Basically, anywhere that you go that's not your own private room, you should assume that you're on camera. On top of that, especially as an international visitor, you need to bring your passport to a lot of things. Taking a train—the high-speed train—required checking in with the passport. Going to the Forbidden City required checking in with the passport.

Going to parks, even just public parks, seemed to in some cases require a passport. In some cases, they maybe said they did, but they didn't always. There were fees—small fees—for public parks in many places, which I found to be surprising. Truly very small fees, like 2 yuan or whatever. This would be 30 cents, something along those lines—basically, a nominal fee.

I also noticed that if you wanted the annual pass, it was only a small multiple of that. So you could get the annual subscription, or the annual pass to parks, for a dollar or two. But why were they even charging at all? I wasn't really sure. One theory was that this was just another way that you can track people if they're going to enter the park.

You know who paid what and when, and so you know who's entering the park. I'm not sure if that's really the right answer. Maybe it was just for revenue. They certainly keep things very clean. There were a lot of people sweeping the streets, again especially in Beijing.

I was really amazed by how many people I saw sweeping sidewalks, sweeping the park areas, and sweeping and keeping the public bathrooms clean. Obviously, that has to be paid for somehow. So maybe the park revenue is going to that. It didn't seem like they were charging enough to really cover that, so I'm not sure that's really the right answer.

But, yeah, it's clear that as you go through life, you are being watched, and you are also being logged at all these different touchpoints. The upside of this is that it is incredibly safe. You have basically no worry. The attitude there is that there's no crime in China. That's pretty much how people approach it, and I certainly felt totally safe.

I've also heard from women and from families that it's very comfortable to be out at night as a woman in a way that it's not so comfortable in New York City, or with kids. You can take your kids anywhere, with kids out at any time of night. I went to one section of Beijing fairly late at night—it was around 11:00—and there were just tons of kids out.

It felt—and I felt that in Mexico before as well—like you're safe wherever you go. Supposedly dangerous Mexico certainly does have some dangerous parts, but you go to a small town in Mexico and kids are out late at night in the public square, and China has a similar vibe. It just feels like you're safe wherever you go.

This is also super useful if you do lose your phone. I heard multiple stories of people leaving a phone in a cab, and this can be a huge problem if everything is on the phone, right? In the end, I did take a computer, which I didn't really need and never really used, but I wanted to have some other device that I could use if I lost the phone.

I had recently replaced my laptop, so I wiped the old laptop, took it back to factory settings, and basically just took this clean laptop to have this backup device. But what I heard from multiple people, multiple times, was, "Oh, I left a phone in a cab. I went immediately to the police." There are a lot of these little police touchpoints around the way.

They said, "I went immediately to the police. I told them what happened. They pull up the security camera footage. They're able to say, 'Point to where you were, or maybe show on a map where you were.'" They're able to get a visual of you getting out of the cab very quickly. They're then able to see the license plate on the car that you were in.

The information integration is such that they can then identify who's driving that car. They can call that person's phone number and say, "Hey, your previous passenger left a phone in the car. We need you to bring it back." People get their phones back seemingly within an hour, maybe an hour or two, with almost perfect reliability. I actually did not hear of a story of anybody leaving a phone in a cab and failing to get it back.

I heard multiple stories of how quick and easy it was to get it back by going to the police. So this was pretty impressive, and obviously this surveillance does create some downside-risk potential as well. Pash and I on the AM had a whole discussion not too long ago about how, if we move to mass surveillance with our current laws—or at least our current policing environment, our current enforcement environment, which is premised on the idea that not everything is going to be investigated and not every break of the law that happens is going to be prosecuted—somehow that flips.

All of a sudden, AI is reviewing and surveilling everything, reviewing all the documents, and putting 2 and 2 together. We're going to be in a weird place where people had the expectation that this was all kind of loose, and suddenly it becomes tight. In China, it's pretty tight. The expectation is that you are being watched and that you don't have digital privacy, and the result is that crime is low.

The potential for abuse, I think, is pretty high, but from what I saw, as of now it seems to be working. I think there is definitely one way to model what is happening there in terms of risk: they may be putting a lot of risk into the future. One question that I asked people multiple times was, "What happens when President Xi retires, or what would happen if he were to suddenly die?" Basically, as of now, nobody has any idea.

There's no successor that anybody can identify. There could be a real power struggle. Obviously, with the apparatus that they've created and the ability to surveil people, if there was a power struggle or if there was dissent or whatever, the potential for abuse could be extremely high.

I think it's already the case that you don't step out of line, right? As of now, things are under control and people are playing by the rules. There aren't too many people who are stepping out, and so there's not too much need to abuse. It's all under control. But this does, I think, create significant risk for the future if things were to go haywire.

For now, it's working quite well, and the experience is really good. You feel safe and you feel like you're not getting ripped off. Your guard is down. I would say my guard was down. I love traveling. I love going to different places, and I'm not concerned about it being a little bit dangerous.

I went to Brazil the last 2 summers, and people would say all the time, "Oh, it's dangerous. You have to be careful. You can't take your phone out on the street." Sometimes I did. But there's a lot of phone theft that goes on in Brazil. They, by the way, are also introducing more and more network-style surveillance to try to stamp that down.

It does seem to be working. I actually visited the police station in São Paulo as part of a walking tour that I did one day there, and it's quite China-like. It does seem to be really having an impact. It does seem to be moving the needle on crime statistics, from what I've read.

This is something that, with AI able to review all the tape, AI able to put 2 and 2 together, facial recognition, and just how scalable all this stuff is becoming, I think all societies are going to have to figure out where they want to land on the spectrum of how much surveillance versus anonymity versus freedom, et cetera. But China has definitely picked its place, and for now I think it is working for them.

You can go around with a very comfortable feeling. If something changes, I think that is a serious risk. But for now, it's a pretty awesome experience.

Okay, almost to the end for today. One observation that I made—somebody said it was an astute observation, so I thought I would relay it to you—is that in some ways Chinese society is a little bit more comfortable with contradiction than I am used to.

I think, speaking for myself, but in a way that's representative of Western thought, when we see contradictions, we feel like they need to be resolved. They need to be interrogated. They need to be taken apart. We want consistency. We want logical consistency and coherence in our systems, in our account of why things are happening the way they're happening.

In China, I've noticed a few different ways where that didn't necessarily seem to be a requirement. One of the things is when it comes to religion: people participate in multiple religions. They don't really seem to see a contradiction in it. You have people who participate in Christian and Buddhist stuff and consider themselves atheist all at the same time, and it's fine.

Even though some of these religions—especially the Christian religion—are meant to exclude participation in others, they don't seem to mind blurring those lines. I'm also reminded of a ChinaTalk episode that I cross-posted. The title of the episode is "Is a US-China Thucydides Trap Unavoidable?" This was a ChinaTalk episode with Professor David Kang from a little over a year ago, in April 2025.

One of the interesting points that I remember being made in that episode was that the whole China-Taiwan thing—part of the struggle comes from the fact that everybody now is living in this Westphalian paradigm, where nation-states are a binary thing. Either you are one or you aren't one, and you have these very particular criteria that make you one. If you're not one, then you're not one, and you're either sovereign or you're not.

The argument that the guest on that episode made was that if that paradigm hadn't been imposed by Western thought on the world as a whole, and it was just left to Chinese people across China and Taiwan to figure out their relationship

You might see something different emerge where there's a contradictory understanding: We could say, “Yeah, we're independent, but we're part of China. There's kind of one China. We'll pay tribute to you, or we recognize that you're the prime power in the region, but we also have our own thing going.” And that might be okay.

There does seem to be reason to believe this in a bunch of different ways: There's just a little bit greater comfort with this sense of contradiction that, for Westerners, I think is uncomfortable. Other examples of that are just VPNs. It's illegal to use a VPN, but everybody does it. Not only does everybody do it now, but everybody I met with does it.

You might have noticed, of course, that there are a lot of Chinese people on Twitter, especially in the AI space. How are all these Chinese AI researchers at Chinese companies on Twitter if Twitter is blocked by the Great Firewall? The answer is that they use VPNs. Even though it's technically illegal, it's accepted to the point of being commonplace, and companies even provide them in some cases.

They have a company VPN that they can use that is official. It's officially provided by the company to them. So, it's not just random one-off people downloading apps and circumventing the rules. It's like the company is giving you a VPN, and this is a fairly well-known company. So, what's up with that?

People are not worried about getting busted for their VPNs. They were very comfortable with it and open about it. I never heard of anybody getting into trouble. Somebody said it would be like getting struck by lightning. They were like, “Has it happened? Could it happen?” And, yeah, somebody maybe could be made an example of if they stepped out in some other way, but it's fine. They were not worried about it at all.

Similarly, on the censorship point, there was this notion I heard several times where people were decrying the censorship regime and complaining about it, all the while getting around it via VPN. But then, in the next breath, they would say, “But it is kind of necessary,” or, “There are good reasons for it, because we wouldn't want the whole Chinese public to have this access. People aren't so educated.”

They would say, “You're talking to me. I'm a worldly person. I've spent time outside the country. I can handle this access to information. It's annoying to me that I'm officially denied it and I have to go through this VPN thing. But I do that, and I'm also not worried about being caught.”

They might not be glad that other Chinese people don't have the same access, but they definitely feel like they can understand why the government is not giving the same unfettered access to everyone. Maybe not everybody can handle, or would be able to properly interpret or contextualize, these more sensitive stories that they themselves would not want to be unable to access.

It's kind of strange and paternalistic. I wouldn't say everybody wanted to defend the censorship, but certainly multiple voices said, “Yeah, I kind of do see why some of this stuff exists. We wouldn't want to cause panic among the broader population.”

Two other examples of this comfort-with-contradictions thing, real quickly. You may have heard that there's an emphasis on the low-altitude, or drone, economy in China. Okay, that's cool, but drones are also banned in Beijing.

There was one university professor I was talking to over lunch who was pretty freely criticizing this policy—not dissenting at the macro level from the government's legitimacy, but basically pointing out some weird contradictions and saying how it was silly. He was very comfortable speaking about it. It was like, “We're trying to promote this technology, but we've banned it in the capital.”

There was some kind of weird catch-22 where he already had a drone when the ban was introduced, and then it became illegal to have the drone. But you have to get your permit for the drone before you can have it, and he already had it, so he couldn't get the permit. It was a very—you can only fly it outside of Beijing, but for him to take it out there, there was some issue where he couldn't get the right permit. It was a catch-22.

He was just freely saying, “This is ridiculous,” but also, again, in the next breath, “But I do understand why we might not want to have drones flying all over the capital. We've got a lot of sensitive stuff there.”

Finally, there's also this really longstanding combination, or mix, of an extreme commitment to fairness with a long history of corruption. This goes back a long way. At the Imperial College Museum, one of the more interesting exhibits I saw the entire time was a museum of the history of the civil service exam, which you probably know goes back a long time.

For many centuries in China, they chose members of the civil service via this really difficult exam that people studied for years to pass. A vanishingly small percentage of people actually managed to pass all the way through to get into these very coveted positions.

One of the reasons for this was extreme fairness. They would go to these lengths to ensure the fairness of the exams. A couple of details were that you had to sleep in basically a little prison cell to take the exam. You would go into the cell and be in there for 3 days at a time. The exam itself took 9 days, but you would have 3-day periods. Then you would be able to leave for a day and come back.

During those 3 days, you were in your little cell by yourself. Your exam materials were in there. You were cooking your own food over a little personal flame that you had in there, and you were using the bathroom in a pot in your own little cell.

They had a replica of one of these cells, and this was all to ensure that the exam was fair. They would even, at some points in time, have scribes transcribe individuals' exams so that the examiners couldn't recognize their handwriting. That's how much of a commitment to fairness, and how much of an investment in making the exam fair, they've made for literally hundreds of years.

But, again, part of the reason that these positions were so coveted was that there were opportunities for corruption. There's this strange juxtaposition of fairness and rectitude in the selection of these civil servants with the expectation that they might also be a little bit corrupt on the back end.

All societies, of course, contain their contradictions. I don't think this is totally unique to China, but it did seem like the comfort with it, and the willingness to leave some of these things in tension as opposed to feeling the need to fully resolve them, felt a little bit noticeable to me as a foreign observer.

I do think that gives you some wiggle room. Why do I even bring that up? In the next couple of episodes, when we get into things especially around U.S.-China relations, I think that this comfort with ambiguity, or this lack of a need to resolve everything to the point where it's fully coherent or fully consistent, can give some wiggle room.

We have this a little bit too with our strategic ambiguity around Taiwan. But I think there's room for multiple truths to exist in superposition in the Chinese system. That also gives them some wiggle room to make deals with us that, on the face of it, if you assumed or insisted that everything was going to be fully logical and fully consistent, might not be possible in the same way.

Okay, final section for now: attitudes on AI. Of course, I talked to a lot of people who are in the AI space, but for the moment, I'm going to talk about the attitudes of people who are somewhat outside of the core AI community.

I hired one tour guide the entire time I was there, at the Forbidden City. This woman came up to me and made her pitch that I should hire her as a tour guide. I really liked her vibe. She seemed to have a fun attitude, and she spoke pretty good English. I said, “You know what? I haven't done this at any other site. You're on. Let's do it.”

Naturally, I had gotten my historical context from AIs. I decided I might get some perspective on AI from my human tour guide. I asked her a bunch of questions. I told her what I do and, to the best of my ability, tried to make it clear what it is that I do, which isn't always clear even to people I speak to in native English.

I tell them, “I'm a full-time student of AI. I'm an industry analyst. I'm an AI scout.” What are these things? Nobody really knows. “I'm a podcaster. I have 10 part-time jobs.” I say all those sorts of things.

When she understood that I worked in AI, her immediate reaction was to light up and say, “Oh, it's a good job. It's a good job.” She was very cognizant that this is where the action is, this is where the energy is, and people are making money in this space. She was enthused and excited for me. It seemed like she was excited that I was in the AI space.

When it came to her use of AI, one of the more interesting things that I heard the entire time was about the AI Doubao. Doubao is from ByteDance. That's the same company that makes TikTok. Doubao is, as far as I can tell, the most broadly diffused AI in Chinese society.

A couple of people actually said that the excitement about DeepSeek is much more prevalent in the West than it is in Chinese society. Somebody also said that the Qwen models—nobody in China uses them, certainly not at the retail level. They're popular for fine-tuning and cost efficiency among Western startups, but the attitude that I heard about the Qwen models was, “We don't really use those in China.”

What do people use? They use Doubao. It has something like 150 million users, and it’s a very voice-native thing. It’s very much a companion, and people really seem to like it.

If you’re outside of China, you can’t use Doubao. They have another version called Dola, which is the international version. In the same way that Douyin was the original, native Chinese TikTok, and TikTok is actually the international product, they have a similar thing going with Doubao and Dola.

But you can’t even get Dola in the United States. When I was on my international carrier roaming plan, I went through the App Store, and I was registered as being in the United States. Dola was not available in the United States. Actually, one of the few times I had to turn my VPN on was to make it look like I was in Singapore or somewhere so I could download Dola and experience this Doubao thing that everybody was talking about.

I didn’t end up using it very much because it was too much of a rigmarole, and the VPN seemed to slow my connection down in general. But I did manage to get in there a little bit and have a bit of a conversation with Doubao.

Doubao is maybe the closest thing to ChatGPT in the sense that it’s the thing that tons of people are using for free. It’s maybe Facebook in the sense that it’s the thing that the boomers are using. I heard a couple of people tell me—people who were maybe in their late 20s or 30s—that their parents used it.

Two women actually said that their moms had made some comment to them like, “Look at what Doubao said to me. Why can’t you be so nice and sweet to me like Doubao is?” So there’s definitely a companionship element to it.

This tour guide woman at the Forbidden City said that, in Chinese society, there are some things you can’t really talk about with anybody, but you can talk about those things with Doubao, and nobody will judge you. Doubao won’t judge you.

I said, “What kind of things are we talking about?”

She said, “Maybe you have a boyfriend, for example, and you decide you don’t like that boyfriend anymore. You want to change your boyfriend or whatever. That’s the kind of thing that you would talk to Doubao about.”

I said, “What would you have done in the past, with no Doubao?”

She said, “No, just keep it to yourself.”

I have no idea how representative that is, but it was interesting. It was interesting that she found a lot of value in Doubao and that it seems like that’s a pretty common thing.

I also heard this “nobody will judge you” idea both in the sense that Doubao itself will not judge you, but also that people will not judge you for using Doubao. The idea that you have some sort of emotional connection, that you get emotional support, or that you have a real relationship—with “real” being whatever you want it to mean—with Doubao seemed to be tolerated, accepted, and understood as an increasingly normal part of society.

This was not just young people. It was young people’s parents who were doing this. I thought that was really quite interesting.

Nevertheless, there’s also some reason to believe that the enthusiasm for AI that we often talk about in the West—where we’re pessimistic about AI and China is more optimistic—might be overstated a bit. Several people said that, in an economic sense, people are not that confident in the economy. They’re worried about losing their jobs, and they’re worried about being outcompeted.

To some degree, they’re running to AI on the kind of famous rationale that it won’t be AI that replaces you; it’ll be somebody using AI who replaces you. If you want to be secure economically, you’ve got to be a good early adopter of AI.

There’s some sense that some of the survey data, or whatever, that shows enthusiasm about AI might be capturing this kind of run toward the thing that you’re afraid other people will use better than you if you don’t get on it quickly. That may cause some of the sentiment—some of the pro-AI sentiment—to be overstated.

I did hear one story about a DiDi driver—in other words, an Uber driver—who used to be a software engineer but lost his job for AI reasons and is now driving for DiDi. In fact, was that even a real story? I don’t know. I heard that in English from a seemingly credible source.

But I think, if nothing else, it reflects a certain amount of economic anxiety, or a sense of insecurity—a sense that the economy is not that good. Nobody’s starving in China, obviously, these days, so it’s not a total, panicked fear of losing one’s job. But nobody wants to lose their job, and people do feel that the way to get ahead is to adopt AI.

Maybe just a couple of little final points. Diffusion has been discussed a lot, right? Which country has the better ability to diffuse? There was the story of OpenClaw going viral in China. From what I saw, OpenClaw was not super deeply penetrated into Chinese society.

What is happening, though, is diffusion via the platforms. This is somewhat similar to what we have with Google. How do most people use AI most of the time? Arguably, the number-one thing might just be that you get AI summaries coming up in search. It’s now integrated into Gmail, and you’ve got it on Facebook and all their surfaces.

Even if you’re not seeking it out or trying to do anything in particular, it’s coming to you via the platforms. That is happening in China as well, but potentially to a much more transformative degree. Maybe it’s a little slower in some ways, but I think it could be quite a bit more powerful when it does happen.

WeChat is working on its agent. This is also creating all kinds of tensions and competitive dynamics among the Chinese tech companies. I believe it was ByteDance that created an agent that was an attempt to be like a super-agent that would sit on top of all the super apps and include the ability to use WeChat for you. But then WeChat blocked it.

WeChat, in turn, is trying to create its own agent, which will be native to WeChat. Because WeChat is so ubiquitous—literally everyone and everything is on it, and you can book appointments on it and pay for things on it—the degree to which you can really run your life out of WeChat is not matched by any app in the West.

There’s no super app that has the ubiquity and depth of services, all in one place, that WeChat has. They’re in the beta—semi-public, semi-private, whatever—test stage of the WeChat agent. I didn’t get a chance to use it, but this is expected to be a huge deal.

Since you can do everything in WeChat, your agent is naturally going to have access to your accounts and your contacts, the ability to do things with your money, and the ability to make appointments. That really exceeds anything that we have out of the box here in the United States.

There’s no single service that you could add an agent to that would give you the kind of Swiss Army knife into real society that WeChat and its agent are going to give their users. So I think the WeChat agent launch, which I think is coming up pretty soon because it does sound like they’re in pretty late-stage testing, might be one of the most important agent launches in history.

I think it’s going to have the opportunity to serve—and we’ll see if they can actually compute. The compute analysis, and what should be done about things like export controls and whatever, we’ll save that for a future episode.

It’s not entirely clear to me that they’re going to be able to serve a billion-plus users right out of the gate with the compute resources they have. Maybe that’s part of the reason it’s not already launched. I’m really not sure. We’ll have to wait and see.

I heard conflicting reports about the degree to which companies can scale usage. But certainly, going to a billion-plus users is not going to be a trivial amount of compute, even for a giant like Tencent.

That agent is coming soon, and I think it will potentially be the sort of switch flip that brings AI to basically the whole Chinese public in a really useful way—something that will take us a long time to catch up with in the West, because we just don’t have anything with that kind of integration and reach.

When they turn that on for everybody, it’s going to be another major step up in the quality of the user experience and the things that you can do. As far as I know, literally everyone is on that app. So I think that’s definitely something to watch for.

The WeChat agent is coming. I’m not exactly sure when, but it can’t be too much longer. It’s going to be a huge deal.

They also have some interesting things when it comes to medicine. Ant Group has an AI doctor. Ant Group is the company that does Alipay, and they’re making huge amounts of money even with their super-low transaction fees. They’re making huge amounts of money just by processing ungodly amounts of transactions throughout China.

What are they doing with their profits? One of the things they’re doing is building an AI doctor.

There is a lot of concern about China’s aging population. Multiple times, I heard people talk about the hope that robots might help take care of seniors. This was discussed both in the context of a young person talking to their parents about their future—what kind of care and support the parents were going to need, and who was going to provide it. Maybe a robot.

It was also something that came up in the robot exhibition hall and in the context of medicine. There’s a huge question of where all the doctors are going to come from to take care of these aging people.

The one-child policy has been over for quite some time now, but they’ve basically had 2 full generations of one child, either because of the policy or just because it had become the norm.

And what that means is you’ve got 4 grandparents and 2 parents who only have 1 kid, or, for the grandparents, they only have 1 grandkid. So each young person today—with exceptions, of course, because there are families that have more than 1 kid—for the most part, most people you meet in their teens, 20s, and 30s are the only child and the only grandchild for all of their grandparents.

That’s a pretty profound thing, right? That’s going to be a big inverted pyramid for these people to support. How are you going to support 4 grandparents as 1 grandkid? That’s a lot to take on. So we need robots. We need AI doctors.

Ant Group—this is the kind of thing that they’re investing in with the enormous profits that they have. Their AI doctor very much echoes the sort of AI doctor spirit that we’re seeing from ChatGPT, from Claude to some extent, and from Gemini too. It’s free. They’re really concerned about reliability, and they’re really concerned about not wanting to position themselves as replacing or competing with doctors, but as a resource that serves both doctors and people and hopefully makes everything better.

It’s all very nonthreatening, with lots of disclaimers and lots of points about making sure you double-check this with your human. “We can’t prescribe you medicine, but we can hopefully guide you.” All these very similar echoes of how things have happened in the West thus far with respect to AI doctors.

And again, that comes available from Ant Group, which has Alipay, installed on just about everybody’s phone. So this is something that they can push to an incredibly mass audience if they choose to, when they feel that it’s ready, in a way that I think is not impossible to replicate in the West, because we certainly do have apps that have very deep penetration. But the ubiquity of these things in China just allows them to push on the diffusion lever in a way that I think is going to be kind of hard for us to replicate.

That’ll definitely be very interesting to watch. The WeChat agent launch, in particular, I think is going to be a big moment.

I think for today, that is plenty. I’ve been at it for a long time, and if you’ve listened this far, you’ve heard a lot about China. I would definitely encourage you to learn more about China through other, more credible and expert sources than me. I would suggest, for example, ChinaTalk, which I’ve been a longtime listener of, and Sinica, the podcast co-hosted by Kaiser Kuo, whose name I’m hopefully saying close enough to right. I’m really not good with the tonality of Chinese words and names, but I’ve been a huge fan of the Sinica podcast for years as well.

I picked up another few podcasts on this trip that I would recommend— a few Chinese history ones, which were pretty cool to listen to if you’re interested in that angle. The China History Podcast and The History of China were both informative. When I took my ride up to the Great Wall, I searched those 2 history feeds for “The Great Wall,” and I had a couple of podcasts that I listened to on the 90-minute ride to the Great Wall, which gave me a really good historical background for what I was about to see.

I think that’s just a great travel tip in general. Two others that I would recommend are Empires, which is a really different kind of podcast. It’s like a semidramatization, somewhat of a fictionalization. They create multiple voices. There’s a narrator who tells you some of the story, but then there are also the voices of key individuals in the story, and they do 5-episode series about major Asian companies—not just in China, but some in China and some in other Asian countries.

I listened to one on Huawei. It had a really interesting history. You can hear the voice, quote unquote, of the founder. It’s not the real founder’s voice, but they say, of course, that they don’t know exactly what the people said or felt in those moments. It’s all based on research, and I found that to be both entertaining and quite educational.

I also really like the podcast Ascent, which is by 2 women who do deep dives into Chinese businesses, their history and strategy, their current market position, and their product lines. It’s a mix of business history and business strategy. That’s Ascent, A-S-C-E-N-T, and there are quite a few good ones on that as well.

I think that’s part of where I learned how much respect and admiration the founders of these big Chinese companies have earned, because you can definitely hear that in that particular podcast. Again, it’s very educational.

Of course, there are many more resources than that, but those are just some of the few that I used in preparation for and during this trip. The next 2 episodes that I’m going to try to do are going to be a deeper dive into the state of AI safety and governance in China. The third one will be about the relationship between the U.S. and China: how I understand the Chinese perspective on that relationship, having spent a couple of weeks there, and what might be prudent for us to do on the Western side of the relationship as things get real, real, real fast. Thank you for being part of the cognitive revolution. [music] Came in on the redeye flight. Held up my little [music] square of light. One machine told another, [singing] “He’s all right.” Open doors like they’d known me all my life. One week [music] you know it all. One month you know a [singing] little. One year you don’t [music] know a thing, but everything works. Everything [singing] works. [music] They’re selling tomorrow and a brand new car. A saint [music] on the [singing] wall keeping watch on who you are. I swear I never once been half this far. But I know this song by heart. One [music] week you know [singing] it all. One month you know a little. One year you [singing] don’t know a thing. But everything works. >> [music] >> Everything works. [music] I reached up to shake a robot’s hand. A human smile. You don’t understand. [music and singing] Some things you can’t tell your mother. Some things you can’t tell a friend. So [music and singing] smile at the eye in the corner. Hope it likes what it sees. They taught one [music] slight of hand. And says you can always tell me. [music] One week you know it all. One month [music] you know a little. One year you don’t know a [singing] thing. [music] But everything works. Everything works. [music] The moon on the [music] far side of morning is the [singing] same old moon from home. And everything works. Everything works. >> [music] >> If you’re finding value in the show, we’d appreciate it if you take a moment to share with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guest and topic suggestions, and sponsorship inquiries, either via our website, cognitive revolution.ai, or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts which is now part of A16Z, where experts talk technology, business, economics, geopolitics, culture, and more. We’re produced by AI Podcasting. If you’re looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at aipodcast. And thank you to everyone who listens for being part of the cognitive revolution.