Artemis II Makes History, Nutella in Space, The US-China AI Race | Diet TBPN

Wild, wild weekend. Some really great whitepelling stuff, some very disappointing news. We'll go through it all. I read a bunch of different pieces this weekend, tried to sort of tie them together into the newsletter today, but thought we could kind of go all over the place starting with what we talked about a little bit on Friday was the Artemis II mission. It was scheduled to land at 5:07 PM Pacific time, and it landed exactly at 5:07 PM Pacific time, like within the exact minute. Everyone was joking thinking like, whoever's in charge of this should be in charge of Uber Eats delivery times or something like that, or DoorDash delivery times, because it was remarkably accurate. And I think they predicted it like days or maybe since the beginning of the mission, like everything was timed out perfectly. Did you have a feedback?

Yeah, I mean, you can like predict these things, right?

Yeah, it is physics.

Yeah.

But still, I mean—

We know, you know, when the next solar eclipse will be for the next 10,000 years.

Yeah, yeah, 10,000 years. But I don't know, it still feels remarkable that there is no, that there's like no flexibility.

But was that predicted pre-takeoff?

Yeah. Right? I don't know. We should dig into it.

Or was that like updated after they had exited?

Yeah, because you'd think there'd be something about like, oh, like this engine fired a little bit too much or a little bit. So we had to make a small adjustment. I don't know. We'll have to figure it out. Anyway, the reactions were really, really positive. Elon Musk said, welcome home to the NASA astronauts. Welcome home, Reid, Victor, Christina, and Jeremy. The Artemis II astronauts have splashed down at 8:07 PM ET, bringing their historic 10-day mission around the moon to an end. I watched it live and it was, yeah, it was a remarkable moment. I mean, we haven't done this in my lifetime. We haven't done this in a very long time. So Reed Wiseman says, "Thank you, Elon Musk. The four of us glimpsed the red hues of Mars far in the distance as the sun slipped behind the moon and there was zero doubt in our minds that the creative genius of our greatest minds will have us there very soon. Let's go." And so I really like this. This is great. No, no, no, it is remarkable. And this was inspiring for a few different reasons because I felt like, you know, People were not voicing skepticism publicly beforehand. Like, you don't want to jinx it, and also you don't want to be negative about anything, and it makes sense.

But the space people we talked to off air ahead of time were extremely nervous.

Yeah. Not even just the space, there were like people in every single person had a different take on like, oh, this seems risky. This is aggressive. This has moved very quickly. The government hasn't done something like this in a long time. And so, you know, can 'Can America pull this off?' Like, America, there's been a lot of worry about the government being able to do things effectively. And like all government, like many government projects, there had been delays and cost overruns. The country has been extremely divided. Everyone knows this. And this mission in particular required Americans from all different backgrounds and political persuasions to come together to work on a common goal. And we saw some of this, we can talk about later, but even NASA Administrator Jared Isaacman had been through his own back and forth on the way to getting confirmation. And so he was like sort of new on the job, even relative to this mission, which of course has been in the works for years. And so there were a lot of different things. There's also the pressure from the private space industry. You know, can the SLS work in this case? Well, everything did, and it was very, very good. There were lots of things that could go wrong. Even the Apple executives seemed to be a little bit sort of nervous about this. There's a post in here that we could pull up.

I would love to know how they test that parachute system.

I think they launch it off of a plane or something. I don't know. How do they do it?

Yeah, yeah. I know. I'm sure there's a good answer, but—

Yeah, but you have to imagine that it's 3 parachutes because it can probably survive with just 2. Yeah. And there's actually 2 stages of parachutes. So there's one set and then these break away and then there's a new set of parachutes once the atmosphere gets thicker, I believe. But look at that. It opens up perfectly. And what an inspiring image. Look at that. JAWS over at Apple said, "Welcome home to the Artemis II crew. Honored that NASA astronauts brought iPhone to space with them." Not the iPhone, not a couple of iPhones, iPhone. This is in the official Apple brand. You don't say the iPhone, you say iPhone. But they brought iPhone to space with them. One small step for iPhone, one giant leap for space selfies. And so NASA posted this on April 4th, said, "This view just hits different." They took a moment to look back at Earth as they continued deep into space toward the moon and they showed photos, basically selfies taken with the iPhone or with iPhone, I guess, of the Earth. And then Tim Cook waited until they landed safely. Congratulations to Artemis II on a successful mission. You captured the wonders of space and our planet beautifully, taking iPhone photography to new heights. And we're grateful you shared it with the world. Your work continues to inspire us all to think different. Welcome home. And so Aaron pointed out to the tune of 3 million views, notice that Apple didn't comment on the iPhone pictures from Artemis II until the crew safely landed. So, you know, everyone was like, you know, on the edge of their seats hoping for the good outcome. And that's exactly what happens. It was very, very high stakes, but it was also in many ways America at its best. Even the never-ending culture war took a backseat to this. There was this interesting back and forth between Jared Isaacman and someone who is not a fan of, I guess they deleted the post, but the Artemis II crew was listening to Pink Pony Club by Chapel Roan, and that didn't align with someone's politics. And so they said like, this is ridiculous. Ridiculous. How can they possibly listen?

Because the last 5 bands that have come up on the show, it sounds like a made-up band.

Chapel Road? She's big. Jared Isaacman was like, hey, let's cool it with the political rhetoric. It's not my choice in music, but the astronauts rode a controlled explosion into space on a journey farther away from Earth than any human before with everything around them trying to kill them. That's a crazy way to put it, but it's true. They can listen to whatever song they want. And I thought that that was a really, really important moment when everyone is so divided. And so the job is very much not finished. Artemis III, which aims to land on the moon in 2028, will be a much bigger challenge. And there's some extra context today in the Wall Street Journal. And we can talk about the difference between Artemis II and Artemis III and sort of where this is going. So the journal writes, Micah Maidenberg says, "Artemis II is a blockbuster. Landing on the moon will be a lot harder." And so flying around the moon may end up being the easy part for NASA's Artemis program. This month's Artemis II flight captivated people around the world as the agency pulled off the deepest human spaceflight ever recorded and the first crewed mission to the moon since the 1970s. NASA and its contractors must now get through a series of sprints that would culminate in astronauts landing on the lunar surface in 2028. President Trump outlined that expectation in an executive order he signed last year. The path to the lunar surface is open, but the work ahead is greater than the work behind us, said NASA Associate Administrator Amit Kastriya. Artemis III, planned for next year, will focus on docking the Orion spacecraft with lunar landers in low Earth orbit, a precursor to a planned landing on the moon. Some current and former NASA spaceflight officials are skeptical that a 2028 landing will be possible given the technical and operational milestones the agency and companies involved need to overcome. Among the challenges: showing one or both of the moon landers that SpaceX and Blue Origin have been developing can safely transport astronauts, and preparing new spacesuits made by Axiom Space. ULA needs to develop upper stages for NASA's SLS rocket. Space missions often take years to put together.

They gotta do logos all over the spacesuit.

They should.

Private companies should be able to help fund the mission by—

They really should. And there's another story in the journal here about that viral video of the jar of Nutella that ended up floating on Artemis II. I was convinced this was like VFX or AI when we pulled it up. Apparently it's real. We can dig into a little bit of like how this actually happened. Ben Cohen has the story in the Wall Street Journal. As millions of people all over the world watched Artemis, the Artemis II lunar flyby this week, they were minutes from seeing astronauts travel the furthest distance ever from Earth when they were suddenly captivated by another majestic sight. It floated through the spacecraft, tumbled right past an astronaut's head, and drifted across NASA's livestream, leaving the roughly 252,000 miles away with the same question. Wait, was that a jar of Nutella? Back on this planet, in a Persephone, New Jersey conference room, executives at the brand's parent company were taking their seats on Monday for their 2 PM operations committee meeting, oblivious to the flying object that had appeared far, far away at 1:52 PM. Their meeting was quickly interrupted by a message in the Microsoft Teams chat flagging that Nutella was in outer space. As it turns out, the people who spread Nutella to every corner of the Earth were more surprised than anyone to see it near the moon. They only found out about the most famous jar of gooey stuff in the galaxy when they followed a link in the chat to a social media post. Dang, how much did Nutella pay for this product placement? And we saw that post and we had the same question. So Nutella says zero. They did not pay for this. This is not product placement, but it is remarkable. They didn't know their chocolate hazelnut concoction was aboard Orion. They didn't even know that the astronauts took it with them. And it's still weird to me that astronauts can just bring random stuff with them, but I guess 'cause it's just a bus at the end of the day. You can put whatever you want on it.

Do you remember your first time trying Nutella?

Maybe not really. I'm not that big of a Nutella guy.

For me, it felt like the first day of the rest of my life.

Really? You're a big Nutella fan?

Not really anymore, but as a kid discovering that, that it was like a peanut butter-like thing that was just on an entirely different level.

It was magic. What is it, a condiment? Technically, what is it? I don't know.

A spread?

A spread? Is that a thing?

I don't know. It can be a lot of things.

But it's always sort of bothered me that it sort of LARPs as chocolate. Like, it looks like chocolate, but it's like hazelnut technically, which I think is like sort of a betrayal. I don't know. Sound off in the chat if you have strong opinions about Nutella. David says Nutella is like crack for kids, and Jordi is being paid by the Nutella Corporation. No, we are not sponsored by Nutella. I wish.

We'd have a big jar of it right here.

So it's a chocolate hazelnut concoction.

Yeah, it has hazelnut in it, but it has chocolate in it as well.

So Nutella, the corporation, did not know that Nutella, the hazelnut concoction, was aboard Orion. They still don't know which astronaut brought it.

They won't know for sure. You just know that for the next mission, Red Bull will pay any price to have cans of Red Bull floating around the cabin.

Somehow I feel like the NASA astronauts' tax records will be deeply inspected to see that they're not selling ad slots out the back.

We need some iPads floating by with like B2B SaaS.

That'd be good. Yeah.

That'd be sick.

They still don't know which astronaut brought it. And like us, they weren't even sure the video was real when they watched the jar hurtle across their screens at exactly the right angle for the label to spin into focus. It all looked too perfect. I couldn't have filmed it any better if I tried, said Chad Stubbs. Stubs, who is their chief marketing officer. What a great name for a CMO.

Stubs.

Of Ferrero North America, who owns Nutella. But once he reviewed the NASA footage and saw a levitating tub of Nutella, he knew that a marketing opportunity had landed in his lap and that he was no longer sitting in his most boring meeting of the week. It was a lot more interesting than talking about shipping details. And so from the conference room, they started a Teams group to discuss the logistics of their improbable operation. They called it Nutella Mission Control. Before most Americans had never seen the original video, they posted a slow-motion clip set to the iconic theme of 2001: A Space Odyssey. The tagline: Nutella is out of this world. And I wonder if they could just rip that on Instagram using like the integrated music functionality or if they had to quickly license that because getting an official theme from a Hollywood film like 2001: A Space Odyssey It's definitely within budget for something like this, but it's usually a back and forth with some emails. But maybe as a large marketing team, they had everything like wired up already. For as long as humans have been leaving this world, they have been taking products along for the ride. But in this age when every inch of the planet is sponsored, space has become the most prestigious real estate in marketing because it's the only place where marketing is banned. NASA has a strict policy against promoting or endorsing commercial products. And Tyler's booing and enforces it so aggressively that not naming brands might as well be part of basic training for astronauts. Unlike college athletes, they can't get paid for their name, image, and likeness. As long as they're employed by NASA, they won't be shilling for Nutella.

So yeah, I'm not saying the astronauts should be able to do it independently. I'm just saying that NASA should try to, try to build out a multibillion-dollar ad.

You got to go straight to the top. You got to go straight to the government. Yeah. And say, hey, you know, Lockheed's showing up with some stuff. SpaceX is contributing. Axiom Space is doing the spacesuits. Blue Origin's doing a moon lander. Why not Nutella chipping in as well? At least paying for part of it. I don't know.

It would just be extremely American.

It would be extremely American to make the Orion capsule look like—

That's what I was saying. Pay-per-view.

A NASCAR.

And you gotta have a pay-per-view.

Yeah, pay-per-view for sure.

Like you can watch the stream when they're just kind of hanging out traveling.

Yeah, yeah.

Anything like a landing, splashdown, takeoff, it switches to pay-per-view mode.

Yeah. I think they got to sell the windshield. They got to sell the windshield. When you're taking photos of Earth, you got to see tide. You got to see tide across the windshield. It's like, oh, seeing the blue marble from this distance is amazing.

Reminds me I have some laundry to do.

And I kind of have to put the camera in between the I and the D.

Podcast ads over during the stream too.

There is a lot of dead air.

Just letting you guys know in in T-minus 30 minutes will be coming around the moon. And this segment, this moon passing, is brought to you by Athletic Greens.

Great. Okay, so as long as they're employed by NASA, they won't be shilling for Nutella. When one Artemis II crew member let slip in a press conference that he was bringing an iPhone to get mesmerizing photos of Earth, he caught himself. He said, I don't, I think I can actually say that as a government employee, Reid Wiseman said. We have small, highly powerful computing devices that we'll take with us with outstanding cameras. And so, yeah, what is an iPhone if not just a small, highly powerful computing device with outstanding camera? While in the cosmos, they also found other purposes for those powerful computing devices. One picture shared by NASA showed Jeremy Hansen with an electric shaver in one hand and his iPhone in the other because he was using it as a mirror. And that wasn't even the most amazing part of the shot. Anyone who looked closely would have spotted another product in the corner, a container of Jif peanut butter. Now, what's interesting is that the government does have, at least with peanut butter— are you familiar with NIST, the National Institute of Standards and Techniques or Technology or something, NIST? So NIST is like our official weights and measures, like they keep like the canonical, like what is the 1 pound? What is 1 gram? They, and they have a whole bunch of standards for all sorts of different things. And then different companies can, can like agree on, okay, well we are both, you know, saving, reinventing and defining the gram. Yeah. But all sorts of things. And one of the things that NIST like has is peanut butter, reference peanut butter. So if you are doing some sort of lab experiment and you need to say that you are testing this product when it comes into contact with peanut butter, you can go to the government and get the most standard, the official peanut butter.

The official peanut butter.

Because otherwise somebody might say, well, did you use Jif or did you use Skippy or did you use something else? And this way you can just say one thing. And so there is a world where the government would say, okay, don't bring Jif peanut butter, bring NIST peanut butter. But I don't know. To find out more about space oddities, this journalist at the Wall Street Journal, said he called Robert Pearlman, who obsessively tracks them, as the editor of Collect Space. He told me something curious about outer space. The deeply ordinary parts of NASA missions resonate back home as much as the extraordinary. We remember the astronauts who flew around the moon and the flying Nutella. It makes us feel closer to the humans who have never been further away. After all, most of us will never see the dark side of the moon. Disagree. And I think we're going to have Peter Diamandis.

Looks like a Coachella It does look like they're twisted knobs or something.

What are they doing there?

Just—

okay. Oh, they're checking the watches. Okay.

Wrist check.

Wrist check. One product was already synonymous with the wonder of spaceflight. Do you know what we're talking about? Can you take a guess? What is the main product that was famously designed for astronauts in space?

Is it the ice cream?

No, no, that's close. The freeze-dried ice cream is up there. Whipped cream? No, what? Whipped cream? No. Tang? And no one knows Tang? You don't know the story of Tang?

I think that's before our time, John.

There's Tang. And then there's also, you know, those super bouncy balls? I don't know how apocryphal that is, but there's always been this story, at least when I was a kid, it was introduced to me that those super bouncy balls were designed as rocket fuel or something like that. And that the experiment went wrong and like, this was all they could come up with. Fake news. Probably fake news. As an 8-year-old, Like, I was like, this is lore. This is peak lore. This is peak lore. The Artemis 3 mission next year is supposed to help set up NASA and its contractor to attempt one or more visits to the moon in 2028. So that's Artemis 4 and 5. NASA's inspector general said in a recent report that both SpaceX and Blue Origin have run into delays developing spacecraft for Artemis missions. Each company has been working on in-space transfers of super cold propellants to power lunar flights, fueling operations that are still largely unproven. That is a very, very complex and new technology that we are— that the entire space community is clearly working on. The in-space refueling is sort of critical to actually getting to the moon in a meaningful way. A NASA safety panel separately raised questions about how quickly SpaceX's human lander based on its Starship vehicle would be ready. A landing operation of astronauts with the Starship lander within the next few years appears daunting and to the panel, probably not achievable. SpaceX next month plans to launch an upgraded version of its Starship rocket while Blue Origin is working towards launching a cargo lander to the moon with its New Glenn rocket. So all of this has like the backdrop of the SpaceX IPO. And you have to imagine that even if it was incredibly cost intensive, we're in this weird dynamic with SpaceX where the CapEx requirements of something like this and, you know, sending a rocket to the moon are probably less than Colossus V or some crazy data center. And so you could be in this interesting situation where Elon, is incentivized to move a lot faster, probably not with humans on board, but get even just a basic Optimus robot up there, get a lunar lander up there, just continue to deliver payloads because it just shows so many more milestones. And as you go public, I think it becomes more difficult to stay focused on this like 30-year mission.

Yeah, part of like the value of actually sending humans to space is entirely like kind of marketing and just to prove that it's possible. Like you would think that you would hope that a lot of this is—

Yes.

Like it's just very cool. It's very inspiring.

Yeah.

I feel like it's important. But at the same time, you would think that NASA should just be optimizing for how do we get as much mass as possible up to the, you know, whether it's space or the moon and just basically leaning a lot more into drones.

Yeah.

And going more for volume. Versus these sort of like high-risk, high-cost.

Yeah. Over the weekend in San Francisco, which I'm sure you all saw, there were a variety of attacks. Sam Altman posted a blog post covering a Molotov cocktail that was thrown at his house. Then there was a shooting outside Friday morning, Friday morning, early hours. He said 3:45 AM in the morning. He said thankfully it bounced off the house and no one got hurt. I saw another article that said that the suspect is in custody. And he says, and Sam goes on to sort of restate what he believes. He says, working towards prosperity for everyone, empowering all people, and advancing science and technology are moral obligations for him. AI will be the most powerful tool for expanding human capability and potential that anyone has ever seen. Demand for this tool will be essentially uncapped, and people will do incredible things with it. The world deserves huge amounts of AI, and we must figure out how to make it happen. And then he also says, it will not all go well. The fear and anxiety about AI is justified. We are in the process of witnessing the largest largest change to society in a long time, perhaps ever, maybe bigger than the Industrial Revolution. And so you would expect the, the life and the existence of the American populace to change over that period of time, and we have a duty to make it as smooth as possible. And he says we have to get safety right, which is not just about aligning a model. We urgently need a society-wide response to be resilient to new threats. This includes new policy to help navigate through a difficult economic transition in order to get to a much better future. AI has to be democratized. Power cannot be too concentrated. George Hotz actually had an interesting rebuttal to Sam Altman's post, sort of rearguing for open source, which is something that a lot of people have not been arguing for lately. But it was sort of interesting to see him continue to push back.

Was it just about open source or was it about sharing research?

It was, it was, he was saying that you don't have a, you don't have an obligation to open source the weights of a model that cost $1 billion to train. He's not arguing for that. But he is saying that you should open source the tricks, the research ideas, basically publish the research papers again and empower a broader community. Of course, there's a lot of, there's a lot of competitive dynamics there, but that is something that could potentially happen via regulation or something or happen just just due to just a competitive dynamic. Like there are other labs out there that don't have as much compute and might realize that they have great researchers and maybe they want to open source more. There are a lot of different ways that this could play out. So he says, I do not think it's right that a few labs, few AI labs would make the most consequential decisions about the shape of our future. And so there's been a bunch of, a bunch of back and forth about the attacks and what's driving them and how risky the rhetoric has been. I think in general, it's a very tough situation because you don't want to just spark more controversy and more discussion around this stuff. You mostly want to move towards more security and more positive—

Yeah, and of course, the notable, you know, it's been shared widely at this point, but the notable, one thing that was notable about Friday, the attacker from Friday is just they were sharing all of the, you know, if anyone, all the AI doom material. They were clearly consuming it.

Yeah.

Sort of caught up in it.

Andrej Mitrov, formerly of Andreessen Horowitz, out there with a new fund, Amp Public, said time is running out for technology leaders to show they care about public benefit above all else. Slow down your layoffs, reinvest in reeducation, mentor the next generation. We are all on Team Humanity. Humanity. And I think that's a good message. As the AI race continues to heat up in America, geopolitical dynamics have consistently acted as a binding constraint, limiting the viability of proposals set forth by AI lab leaders. And so we've seen this at Davos with various lab leaders saying, well, we would agree to a slowdown if we could all agree. And then in the Bernie Sanders, the data center ban, and all of this feels very intractable with the backdrop of geopolitical competition. If you don't have buy-in from all the different countries, you wind up just falling behind another country and you have the same dynamic again. And so that piece of the discussion has sort of fallen by the wayside because it's so difficult to argue if you're running a private corporation in America and you're like, I wanna make foreign policy now. That's a really tall order. Fortunately, I think people are starting to, at least investigate what the path towards some coalition between different countries might look like. And Sebastian Malaby, author of "The Infinity Machine" and former guest of the show, published an op-ed in the New York Times outlining one possible solution to the US-China dynamic. He says in, in 2022, the Biden administration tried to arrest China's development of artificial intelligence by denying it cutting-edge semiconductors. This was the CHIPS Act, which at the time I was very in favor of, but of course the policies have all evolved and there are much more complex situations with the entire semiconductor supply chain and how fast the technology is advancing. So President Trump has relaxed that policy a bit without a clear plan to replace it, but the chip export controls have failed. China's tech sector is too sophisticated to be stopped from building powerful AI. In pursuing an impossible objective, the United States is missing an opportunity to, to try for one that sounds fanciful, but which, after a recent reporting trip to China, I believe is more realistic. America should negotiate with China on a global pact on AI safety, which would impose universal limits on a technology that can do much good, but in the wrong hands would do much harm. The premise of the export restrictions was that the United States would be able to successfully block China's access to powerful AI chips. The premium chipsets used in AI data centers are the size of skateboards and can't be smuggled in a simple suitcase. And it's hard to put them to use without hands-on support from the chipmaker's engineering teams. But Chinese developers circumvented controls by training their AI models in chips located in other countries. This is always a question of even if you stop the flow of chips into the country, can you set up a holding company that allocates some—

Yeah, this is a question last year of like, wait, how is Singapore placing that many billions of dollars?

And Malaysia was another one that was— and even in the Middle East, there were always questions about, okay, well, if the Middle East gets chips, are they going to be able to have Chinese companies as clients remotely? And there were discussions of folks basically putting training data on hard drives and or model weights on hard drives and just flying them from one country to another. It's very, very hard to actually contain the movement of the critical pieces of the AI value chain. So its model builders also take full advantage of a process known as distillation. Every time a US lab produces a cutting-edge model, Chinese rivals quickly reverse engineer its capabilities and build a copycat version. The follower has the advantage, he says. American AI scientists used to say that competitors being able to fast follow would not matter. An intelligent explosion was approaching, the argument went, as AI systems would soon become capable enough to write upgrades to their own code. AI would create better AI, better AI would create even better AI. Recursive self-improvement would drive performance skyward. The nation that had just reached this so-called singularity first would be the winner of the AI race. Even if the fast follower were just a few months behind the leader, 3.5 years after the Biden administration shipped controls, AI is generating code to upgrade itself. The promised feedback loop has started, but the accelerating power of the leading models won't determine who wins the AI race. It's AI deployment that will matter to transform economies and armies. AI must be embedded into the business processes and weapons systems. The raw power of the cutting-edge models must be turned into applications. The upshot is that China and the United States are roughly level in the AI contest. Top Chinese models may be a few months behind American ones, and the relative position on military applications is difficult to ascertain as so much is classified. But on industrial applications, China seems to be leading. US-sanctioned companies such as Huawei and Hikvision are rolling out AI systems that perform maintenance checks on high-speed trains, managing mining operations, scanning water samples to assess pollution, and more. At Huawei's campus near Shenzhen, He recently took a ride in an autonomous car. A device in the passenger seat massaged my back and the steering was immaculate. Tyler, do you generally agree with most of these takes here? No. Okay.

What do you think is wrong? To start, I think Chinese labs are further behind.

Okay.

The only reason that they're close behind is because distillation.

Distillation.

I also think—

Yeah, and he's saying that the chip export ban is not working. Yeah, also, but I feel like the, the, them being behind is proof that it is working. And he's also saying, you know, the labs want to get to recursive self-improvement. We're starting to see early signs of that. And so I feel like there's some kind of inconsistency.

Yeah, I mean, also just on the chip stuff, like, we're clearly not policing it as hard as we could. We could be losing it way harder, right? I remember it was like Supermicro, we saw those videos where they had smuggled in the boxes and it was like extremely—

Yeah, the hair dryer.

Yeah, it's like, if we really wanted to solve this issue, like, we for sure could. Among like the leading labs in the US, you are seeing some kind of like takeoff, but like among them, right? It's like very much recursive thing.

Yep.

The models improve the harness, which improves the next model. Like, I think this thing is actually true. Like, you're not seeing open source labs kind of keep up, I think.

Yeah, I mean, it was very, very remarkable watching, like, DeepSeq not accelerate in the same way that I think a lot of people expected.

Yeah, I mean, like, Qwen is kind of what DeepSeq— what people thought DeepSeq would be. And Qwen models are good. But again, like, I think most of the reason that they're good is just because of distillation.

Yeah. Yeah. It's chaotic time. Anyway, thank you for watching TBPn today. Leave us 5 stars on Apple Podcasts and Spotify, and we will be off tomorrow, but we will be back Wednesday. At 11 AM Pacific. We'll see you then. Goodbye.

We love you.

Rolling flashback. Goodbye.