Anthropic's Mythos Security Model

It's time for Twit this week in Tech and it's one of those special shows where I get. two of my favorite people on and just let them vibe and it's uh gonna be a great one. Amy Webb is here. Futurists from the Future Today St strategy Group. Uh Harper Reed, AI guru and technologist. Uh from uh two three eight nine dot AI. We have lots to talk about Trump and the CEOs going to China. Uh a lot of AI news. Musk versus Altman. Google investing in SpaceX and a whole lot more, I'll tell you what. Don't even worry about what we're gonna talk about. Just make sure you watch this show because TWIT is great. And next. Podcasts you love. From people you trust. This is Twit. This is Twitch, this week at Tech, episode 1084, recorded Sunday, May 17, 2026. Don't overcook the asparagus. It's time for Twit this Week in Tech, the show we cover the weeks, tech news, and you are in luck. If you've never listened to this show, or if you've listened for 21 years, that's how long we've been doing it. Either way, you're gonna be thrilled that we have Amy Webb here from The Future Today Strategy Group. That is me. FTSG. She's a futurist. So good to see Amy. One of the smartest people in the world. And uh always welcome. And her buddy, it turns out the last time they were on, I didn't even know they knew each other. It turns out. Harper Reed and Amy go way back. Harper Reed, technologist entrepreneur and hacker at 2389.ai. Hello, Harper. Hello, Amy. Hello. When we have the two of you, we don't need anyone else. In fact, I'm just gonna relax, sit back. And let you guys just jam, man. Harper, should we get this out of the way now? Oh, we should I never have the stuff. I always need the thing. There it is. This is how each other can you screenshot this and send it to What's up? One second. There we go. The leadership US Japan program. Tell us about that, Amy. What is that? Um so I think Harper and I told you about this last time around. We did, but they're not everybody was here the last time, so It's a um Terrific. Organization that brings together American leaders and Japanese leaders. Um Who actually live in other places around the world besides just Japan and the United States. And the point is to establish a network and relationships. For the purpose of bringing the two countries more firmly together over time. Um and so and there are some I I've I guess This is going Almost my tenth year. Um you you s you sort of do a week in Japan as a cohort and then you do a week in the United States or depending on what year it is, there's an alternate uh you you sort of alternate between the two. And um Yeah. And then you become a lifetime fellow, which is what Harper and I both are. And uh It's it's truly awesome. The People uh are incredible incredible people doing really meaningful things. Um, yeah, and we uh we see each other in person throughout the year and it's it's a great, great organization. You have to be under forty. to apply and if anybody they raise the age a little bit. Okay. It used to be forty. Um you don't have to speak Japanese. You do have to demonstrate leadership and have some kind of interest um in Japan. If you're not Japanese, but I can't say enough about it. It's it's really an incredible program. Does it involve going to small nightclubs to see hardcore punk bands at all? Yes, and that's exactly what we do. That's what Harvard just did. That's what I do. I don't know if there's a general we, but I have I did do I did bring some of our at that point delegates to see punk band in um Seattle during our our session there. But it's been really great. And it's interesting because as, you know Ostensibly this is a soft diplomacy program. Um, and the thing about soft diplomacy is it shifts the target shift. Throughout the life. I guess life of these kind of programs. And so it's been really interesting just in the Um, I guess seven years I've been involved, how kind of the focus has shifted and it's it's uh It's really fun to participate and meet people that you would never meet before. Um it's good. Yeah, and it's not just I was gonna say, we've got some like Marvel executives. We've got um. recognizable names that I'm not gonna name out loud, but people you definitely have heard of recently in the US web. You've got me, you've got Harper. Yeah, yeah. So I you know, uh I imagine there's been a kind of a wrenching Of the uh of the late lately of the diplomatic uh focus. What with uh Everything going on in the world. Uh in fact that's our first happening. The first topic is about China because uh Trump and sixteen, something like that, CEOs Uh we're in China this past week meeting with President Xi. Uh president for life, she. Talking about unprecedented too. I don't know that there's been another time in history when we've had such a a group of CEOs like that traveling with a group of government officials. And in fact, one of them hitchhiked. Uh Jensen Wang of NVIDIA wasn't invited. And complained about it. And the president said, We'll meet you in Alaska. We got to stop for refueling. Jensen jumped on a p a PJ, flew up to Alaska, met Air Force One, and joined the uh Exodus CEOs from NVIDIA. Tim. Tim Apple was there Exxon? Boeing, Qualcomm, Blackstone Citigroup, Visa. All there. Uh Elon Elon's son was there. Let's not forget about that. Oh Yeah. And Elon was there too, right? He always brings his son with him. Yeah, there's a hilariously awful video of him. He's walking at a normal adult pace and his son is literally racing to keep up with him while Elon's got a death grip on his hand. Oh which is yeah. Jensen, um was left off the list because he's Taiwanese and American. Um NVIDIA's chips are made by TSMC. And the focus of this visit was actually not AI. The focus was um From China's point of view, uh You know, let's all agree that you're gonna let us do what we want with Taiwan and you're gonna stay out of the way. Um so so that is why he was not not on the list. Uh so why was the so then The question is why did they change their minds suddenly? And was it an affront to the Chinese that he was there? Very much so. Oh interesting. Um You know, the way that these diplomatic visits get set up, they're coordinated and choreographed months in advance. Um and who's on the list. Is in where people are gonna sit. I mean that's the other thing. Like is literally it it's a it's a complicated chess game just figuring out where people are going to sit because that message is so much. So any kind of addition of anybody last minute is really tricky. But especially somebody who From China's perspective as a political lightning rod was was Well, and one of the deals that uh the Trump administration made was to clear Uh NVIDIA chip sales. to uh China. Not the best. The H two hundred, not the Yeah, but that'd be good. Yeah. That's fine. But Well's funny is the Chinese nobody in China would order one. The Chinese have decided, no, we're gonna do our own and uh screw you. Yeah. Look, China is um We have ignored the whole world. I used to live there. This is why I brought it up, because you're an e you're an expert on this. Yeah. Well, I don't know if I'm an expert on, but Um for years has been producing exceptional quality everything. They they literally have 70% of the global market share for EVs. And also have like literally a hundred different E V companies. And America has had to but ban them in the United States for fear that it would undermine our own auto industry. Yeah. I've had a chance to drive in some of them. They're awesome. They are great cars and they're in the inexpensive, better mileage. And what's interesting is They are in the process of Converting like All of those car companies are going to also start producing robots and drones. Right. So in fact some say they're already better at bipedal robots than we are. Yeah. Oh, sure. Yeah, so anyhow, um Chinese AI models are only a couple months behind ours at this stage. Close really? I think it depends on who you talk to, but again, I'm not an expert. Expert. Uh, which is what is Quen Huawei's or Xiaomi's? I can't remember. No, GLM is Xiaomi's. Uh, quite good. They're not quite there. Harper, what do you think? You're also comparing them to um You're comparing an open model Um model that has untold tweaks to get it to be more effective on the back end that we don't know about. Um versus Quinn or GLM or or or Kimi or any of these that are good to find effectively free to run. Yeah. Um, you know, and and I think this is also one of the reasons. GPU and RAM, I have to run a highly quantized version of GLM locally. with 128 gigs of RAM, but it still can run it. And it's open. Uh and in many cases it's un uncensored. Like they've removed the Chinese censorship from it. Well I think there's a funny um China is behind, or they say, Oh, the models are behind. Um, but we don't have any open models that are better. We sometimes pop up, like a Gemma pops up or like a Facebook Llama three will pop up or whatever. Yeah, GBT OSS will pop up and and that happens, but then they're quickly beaten by the Chinese open model. And so I think you have to actually compare oranges to oranges in this. So we have to point out Both anthropic and open AI say that's because they're distilling our models. Yeah, but but they distilled their own models off of themselves and also the internet. Like everybody I think that's a funny I think that's a funny critique. Hey, they stole from us. You can't have it both ways. You know what I'm saying? Like you you either it either is stealing or it's not stealing. And if it's stealing then let's talk about it. But you can't be like they stole from us but we didn't steal from you. Um, which I feel like is a kind of the narrative. I just think we need to really be If the US really wants to have a f a Let me say it differently. I find it very interesting that China has followed the rule or the the kind of pattern that the US followed in the past of releasing what amounts to a free technology to the world. And the US is with open models, et cetera. And the US has been reliant on closed models. And so if you talk to entrepreneurs outside the US, a lot of them are relying on these open Chinese models. They're not relying on closed US models. And if we're and and and it seems like a big blind spot of the US entrepreneurs to assume that you know, Southeast Asia, you know, APAC, et cetera, even Europe, Middle East is going to suddenly, you know, use open AI, anthropic Facebook, Google models when they can get something that is, let's say it's six months behind or even a year behind for effectively free that they have control of. Um, you know, which which is like the Linuxization, so to speak, of these models, just spreading it out. For free. I find that really interesting. That's correct. And if you couple that with the amount of infrastructure investment that's happening. Um You know uh our our hyperscalers are gonna invest, what is it, like eight hundred billion dollars on paper right now in building data centers. Um, they're they're burning cash because we don't have a national strategy. It's all being left to the to the you know, market to figure out. That's not what China's doing. China's got Um China has a history of Announcing plans and not making good on them, but that's actually not been the case over the past couple of cycles. So they're building out, you know, everybody's gonna have Wi Fi like broadband. Um so they're making it so that everybody can participate. And the models that people have access to are not cost prohibitive prohibited. The way they can be in the US. So that just creates this interesting circumstance where you got significantly more distributed access. And people can around and find out. You know, like you you just time to play. And the other really interesting thing is The conversation about AI in the US is all about like the robots coming to take my job and then murder me in my sleep. Um you know, it's all like it's all apocalyptic or it's this like utopian wonderfulness that will happen. China's having a much more pragmatic conversation. So if you talk to just everyday people there, and I do Um, they're more worried about um whether or not they're gonna be competitive in the marketplace. So they're not like all the jobs are going away. They're more like Man, I really need to experiment and get better with this because I want to make sure I get the best possible job. Not but but but not like there are no jobs. Does that make sense? So it's just a very, very Pragmatic. different approach that for the many things China does wrong that I disagree with, I think this is one area that we could study and learn a lot from. Did you see uh Jensen Wong's interview with Dwarkesh on the It's kind of an interesting. Which is self serving. Which was uh that we shouldn't block NVIDIA's chips in China that Doing so. We're forcing them to create their own technology and not use CUDA, his his proprietary technology, which is bad for the world and bad for America. His contention is if you would just let them buy our chips, it'd be more likely we'd be part of a larger ecosystem instead of China having its own ecosystem which will ultimately Yeah, I don't know about that argument because there's a lot of Everybody wants access to to China's market. It's a big money. Chine and China's been very smart about not giving access, you know, so like That's a self serving request. It was interesting. He seems pretty good at at at building up the world that he wants to happen and then executing on it. And and I'm very impressed with him and his way of do it and and doing that. And And also how people who work there seem to love it in a way that seems like I have friends that have worked at Amazon for a long time, they don't love it. They're there for money or Facebook and and anyone know what NVIDIA is like? It's hard, but I really like it. And I don't understand how they created that environment because everything I read about it seems like it's horrible to work at. People seem to really like their true believers, which I always admire when you have a leader that's able to instill that. And even if that means bending reality to their Once like he's he's done a good job. Okay, so I have something for you to Google. Leo? Yes, I'm good at that. Spell it. Although I don't use Google, but I'll I'll do this search the web for Yeah, I don't I don't use Google anymore either. I'll caggy it. Um really work. And I love that their logo is a G. They were like, What if we used a G? No one in the search space is used a G. Okay it is M-A-L-E O U S space AI. And you want to hit the Amazon link. Malius. Yes. It's a robot for AI robot for kids. So this is an exact. The name sounds scary. MA NAD of MARC. And also for the elderly. This is an example of one of these little This is a deep seek agent that will that you can talk to and has a very fast real time. Um Like AI agent on it. That's pretty good. Like the cloud or is it. It goes to the cloud. Okay. It's all via I I took it apart and reverse engineered it. It's all MQTT posted to some endpoints in China. Um a lot of other people have done this as well. But what's it mean in sending China everything I say? Oh yeah, and you have to hook it to your Wi Fi. This is definitely for for people who will really want to do that. But it's how long before the SCC bans this. It costs forty USD. What and there's no account. There's no sign up. There's nothing. And it is pretty effective. Like it pretty effective. Like I just wanted I a friend told me about this. This friend of mine in Shanghai was like, Hey, there's all of these small little AI admin assistant things. They're all little hardware pieces. Um, and there's and if you look at the recommended products from this, there's dozens of these. You can just see they all start to look the same. There's little robots. There's also this is not currently available. But but I think to kind of what what Amy is saying, like they've made it so it is cost effective to build really interesting technology that isn't just some insane two hundred dollar a month subscription for the mediocre Version Um, that was literally forty dollars. Like I paid forty dollars with it. It came, I hooked it to my wifi like a crazy person and then immediately started talking to it. I think it doesn't it seem like this is just give inviting China into your network? Yeah, but I think you're missing the point though, Leo. So the point isn't that China's in your network. Like like you already have GoV lights and like all sorts of other stuff. Like they don't have to have some forty dollar device. The point is that this is a free AI thing that's very effective and it's competitive with with the Okay, so I this one's unavailable, but what about the Hey Azoki AI desk robot from Deep Seek? This one I could buy. And have tomorrow. Yeah, get it. Try it. It's fun. It's uh it's a little rabbit adjacent, is it not? I have a rabbit. I did buy the rabbit because it loves it. It can talk to my open claw, which I don't have. But I had I had my AI agent. Simulate open claw so that Rabbit doesn't know any better. And uh I can talk to her through So okay, show you don't know anything about this one, but this is also D Se Deep Seek powered, so I figure it's probably just powered ones, um they're for the most part free and easy and they're pretty good. I would not I know you're not gonna use it for real work. You know, but it but it's like it's a twenty dollar AI assistant that is pretty funny to use. And this is what we're this is what we're dealing with because that's distributed all over the world. Whereas open AI anthropic w they don't have hardware that you can give to your No, in fact all we have is this crappy Amazon uh Echo Plus thing which has a sassy voice and then tries to sell me stuff. Uh or or Ciary, which is just Bad. In twenty eight seventeen or so, um, we were working with Microsoft. This is I'm not divulging anything now 'cause it's been a while. Statute of limitations is expired. Yeah, no kidding. Um but they had multiple teams trying to figure out AI and they were incredibly siloed. Um, I think if you look at some of the research that was coming out of Microsoft, they had some of the best at that point, they had some of the best research on things like machine reading reading comprehension and um recognition and N L P and stuff like that. At any rate. What happened was They got fixated on a device and wanted to get a sub forty dollar Bluetooth Alexa competitor, Amazon e you know, Echo competitor in the market. I bought and lost and and sort of lost Where you know lost the the the uh I don't know. Lost their path, lost their footing. Um You've got all a and It's hard to bring hardware to the market. Uh So it's interesting to me. in the United States that we keep having hardware failures over and over again, which is what's happened. You don't see that in China. And it that's been true in China and Japan, where Herbert and I have both spent a ton of time. Um Because it things are more open and there is significantly more competition. One hundred different companies trying to bring EVs to the market. That's in that's in completely unfathomable. We cannot wrap our heads around that. In the United States. Um but that is ironic because they're communist. And we're capitalist. But that sounds like they've got free more f fr more of a free market than we do. Is that the case? I didn't see the air quotes there, Leo. Free market. No, I meant on the on the capitalist and the communists. Like there is there's needs to be air quotes around maybe that whole sentence. So you think Adam Smith would look at China and say, that's what I was talking about. China has adapted capitalism in a different way than we have. And y one could argue again. There's a lot going wrong there. But you could argue that There's a better approach we are seeing other alternative and better approaches. to capitalism and to the markets for the purpose of like innovation and competition outside the US. It's ridiculous that we are this AI is a very long horizon technology. It's also been in the works for a very long time. But in terms of like getting stuff into the commercial sector, It's only been a couple years and we only have a couple of players. It should be the opposite right now at this stage. We should have a hundred different legitimate companies in every single part of the ecosystem. And that is not what's happening. We have a couple of dominant players. And then there's and then George Bush said let a hundred points a light. There's um a really good book that I think talks a lot about this by Dan Wong called Breakneck. Yeah, really got it got it right over there. It's a great book very effective, very great book, but it really puts this idea that Um, we have migrated and he doesn't do it he does a very good job of not saying one is better than the other, more just saying this is kind of where we are and we may need to fix it, we may not, but like we need to choose one way or the other. But um he just says the US is a lawyer state and China is an engineer state. And so you have what what kind of comes from that. Like you have a lot of laws and regulation in the US, which, you know, is is like all we talk about. But if you look at our leaders, a hundred percent of our leaders are all lawyers. So of course it's like uh my f a fame uh economist friend of mine always tells me. You know, never ask a hairstylist if you need a haircut. Right. It's like you you do what tools you have in front of you. And and whereas China's leaders are all engineers. So it's gonna be much easier for them to say my tools in front of me are to light up a billion solar companies and have the competition fight to see which one of the best solar happens, which is where all the solar innovation is happening. Whereas we are saying, oh, well, we need to regulate to get solar either to out there or to kill wind turbines or whatever thing is happening. point being that like we're using the tools that we have available and it might not be the right tool for this moment in time. Hundred percent. The tools of regulation Uh This is like gonna open up a can of worms so I don't wanna I do that other than to say that Y this is not a moment to regulate. We we cannot and should not attempt to regulate. AI? Um there are other ways forward. But it requires making some tough decisions with regard to who can make money and how and how the returns get distributed and stuff like that. China's just not they're just not operating in the same space. There are plenty of other problems, you know, and and also China's not going to be the one innovating and creating brand new stuff. They are gonna fast follow, but that fast follow strategy is really smart when you combine that with this long term vision. and resources put against bringing everybody online. You know, I I would love to see The United States invests similarly to get everybody look, I mean, like all these tools are really great unless you have no service in your house or you know. Yeah. You just then then cool. You're you you're learning about AI and you can't use any of it. Yeah. What was so uh what was the goal of the uh Trump summit. The summit this week. Yeah. Um It depends on who you ask. From my point of view. Uh There's a big war happening that the United States launched. So there's You know, that and China and Iran historically have relations, whereas we we don't. So I think that there was some amount of uh let's see if we can figure some of this out. Um China is very, very in This is a little bit of um Little bit of nineteen fourteen. Uh where we had all of these hot spots around the world starting to to heat up because there were these political Um Political problems all over the place. feels a little bit like that right now. And um You could look at this as like she may she China may see this as an opening to take Taiwan by force, which they've certainly talked about. Um, they could see this as an opening to use Taiwan as leverage for something else. You know, there's a lot of that happening. There's certainly a lot of concern among uh the US tech. industry uh about losing Taiwan. I mean Apple's no chips anywhere else. They're all made by T S M C and it's presumed if Chi if China did invade Taiwan that they would blow up The chip plants to prevent China from getting the technology. I mean T S would. I don't know if they would. Yeah, that's that sounds Implausible. It's not just the technology, it's the people who know how to use the technology. So it's it's not just the those resources. You've got this incredibly trained, highly skilled um group of people. who are able to to work in these places. You can't that's part of the reason why it's really hard to stand up a factory like that in the US right now, 'cause we just don't have highly, highly, highly skilled people yet. To do that work. And if we do, they are they are um Burden's the wrong word, but they're burdened by regulation, laws, labor laws, et cetera, that are not from today. Right. They're from factories from a hundred years ago. And they were made for very, very good reasons. Like these were made to protect workers and children, et cetera. And so we have this stuff that You know, China never had to go through that as a country. They they they were able to define their labor laws Which, you know, some are bad, some are good, um, in the modern age. And I think that just did that's you know, this is kind of a lot what the Dan Wang book is about, which I highly recommend. I can't recommend enough if you're interested in seeing that. What I what I like about it is a lot of the books on China, you know, Apple in China or Chip Wars are very good, but at the end they basically are shouting China will never beat the US. And it's just is kind of like this hyperbolic kind of thing where it's like Dan Wong is just saying, look, these are two different systems. Here's the historical reasons why they happened, and here's some comparisons across that might be helpful. And I just thought that was a nice, less hyperbolic way of looking at these two systems and how they interface, et cetera. I remember visiting China in 2009. I was a Chinese studies major in college. I love the I never knew that. Yeah, no, I love the culture. I learned learned Mandarin, which I've of course forgotten because unless you use it every single day. Pearl of spoken word. Uh I uh I um Uh in two thousand nine visited China and I had kind of come to the opinion that while the twentieth century was the American century. I think we all agree that the twenty first century might well be the Chinese century. Yeah. And and our guide who uh was a a Brit Expat. Said, Oh no, you don't understand the economic issues besetting China, this is two thousand nine, will keep it from becoming the Chinese century. Is that still true or is that not the case? There certainly were economic issues. There's birth rate issues. Um there are problems in China. There's of course the human rights issue. I don't know if that has anything to do with the economics of it. Do you think this could be it sounds like you're just painting a picture of a Chinese century. Uh well century's a long time in the year twenty two. Let's have it do decade. Yeah. I think it really depends on look, I'm like a pragmatist. I just so like If I were to look at where a lot of Western democracies are and I would include Japan in this. Um and I look at birth rate issue too. Huge birth rate issue. They're also having lots of internal political issues. They just launched a brand new party called Midai. Midai something. Um Mirai in Japanese means future. Um It's a guy. Um and uh the focus is on like AI. So it's fine, but like Japanese I don't know. With apologies to everybody who might be listening this who is Japanese, like talk about stubborn, like just absolute unwillingness to to ch to try anything new. A lot of people are I have the impression they're tradition bound. Is that Uh no, it's just stubbornness. But I that's not across the board. But it it is with software. Some Japanese people are very forward looking and very uh I'm mostly talking about men of a like older men. Still still run things. Yeah. Yeah. Not not younger people. But Um but the point is There's so much pl pendulum swinging in a lot of these countries because of uh extreme views going back and forth, and China is just plowing ahead. So the lack of uncertainty in China is partially what's gonna help Country go forward. And I don't think it's great to have a dictator, benevolent or not, so I'm not saying that. But it does reduce some of the like Work for Singapore. You know, constant back and forth, we're gonna invest in coal, we're gonna invest in E Vs. Wait, no, sorry, we're investing. Every four years we uh we decide something else. Yeah, and that makes um that makes the business climate like possible to work in. We look I'm in the room with leaders of major companies and just about every industry and their boards. Basically what I'm hearing people say is we're just we're on pause for the next three years or two and a half years. So a lot of these companies outside of AI, which is a thing, are not investing in innovation 'cause they don't know what's gonna happen. And so it's every Everybody's just Waiting. Terrible situation to be in. Isn't this almost a global situation? Every policy is struggling with the future because the future is so uncertain. That's that that is an excuse. So the answer is No, you should always be focused. You need l you you need your near term strategy and you need the longer term and there's even if you have no idea what's gonna happen? Let me give you an example. Um Are you familiar with Toto toilets? Yes, I have many of them. We have we We actually have the the Toto that South Park made fun of um in our in our bathroom in our house. We moved and we we left our Toto's behind and that was Speaking of behinds, and that was a big mistake. These opened up, they were warm, they blew at you, they squirted at you, they did everything. They did everything but play music. We didn't have the musical ones, but Yeah, they're they're uh they're awesome. But they're fantastic. Yeah. Yeah, yeah, but one of the things that they did was um The the same porcelain coating that is what mak partially what makes a tote toilet really great, turns out y it's also They've been able they've been researching and figuring out how to apply that same exact technology for printing silicon chips. Right. So um the point but like they discovered this a long time ago and then just kept chipping away at it and continue to invest in innovation because at some point they they like wanted to be ready. And their stock is through the roof now because they make the interconnects at every data center needs. That's right. So that is what I do not see US companies doing. I do not see them investing in the future. Or and it fine, so like this technology may or may not ever actually work, but you're gonna learn so much in the process and everybody who's a part who touches that in any way. Is gonna level up in how they're thinking. It's gonna you know that they were just lucky? No, but uh no there's plenty of examples. Even if they were too much. This is kinda manufacture love. Look. You can't just Like my favorite example. This is this is the new So my favorite example of this for a Japanese company is Yamaha. They make everything from musical instruments to rowboats to speed boats to motorcycles. And like they just make everything. I there there are no companies like that in America. There' There are conglomerates. That that's a little tricky too because General Electric made everything. It wasn't a good thing. Yeah. I think that the what I'm trying to point out is I don't want to name any But We have a lot of US companies that historically did a phenomenal job of doing some research and Research you know, and it doesn't it doesn't affect your balance sheet. It doesn't cost that much money to do research, but you have to organize the company in the right way. So that like some of that starts to spiral out and you maybe you get a new product. It's fine if you don't. You get new learnings that then help you think through the next thing. And we just don't have the the great labs that used to exist. And I think that's a huge we've become like so Like America is like obsessed with optimization. Like we must optimize everything all the time at all costs. Um, and the great irony is like that's a hundred percent of what China used to do. You know, and we've flip flopped a little bit. Well, I need to take a break. I I d this is such a good conversation, I think so important. Uh I d I don't want to interrupt it. So Hold hold that thought Harper. Um and I I think the T is this way, but okay. And then uh I don't know what that is. That's orthogonal. I thought I don't know. Uh uh I would actually kinda wonder uh if either of you have an opinion of What comes after this summit? Whether anything was accomplished. Uh and what comes after the summer take a little break. Uh so Amy can leak. Harper. I by the way, this has already been an expensive show and I blame you, Harper. No no I I had my power bar already, so I'm okay. Uh I bought the robot that you mentioned before the show, the Reachy. And then I just bought this thing that's gonna be a spy for China in my house. Yeah. That's what we're talking about. Yeah. Don't you basically aren't you giving it administrative access to your network now and deep sea? No, but but if you read the reviews on that site, there is a perfect Perfect review. Um in that kind of I don't even know how to say it. Like, like, you know when you read a review of something and you're like, this person does not know what they're talking about. I think I read that review. And the one where it's like, they hacked me. Yeah. You have to you have to type in your password, which is one nine two one six eight. They hacked me. And you're just like Maybe but like maybe not what you should I put it inside a VPN or maybe Tail Scale or something and keep it out. Yeah, just a separate lane. Don't worry about it. Don't hook it to your real big network and just kinda s play with it because it's not it's it's It is a little it's like fast fashion of electronics. It's not there to last forever. Yeah. This is not a high quality product. And honestly, I don't really care if the Chinese know my wifi password. So what? They've got nothing to come and they're gonna take all of your stuff. Just don't hack my toe. That would be They're already in your router anyway. They're in your router, anyway. They made my router. They're made everything in my house. Yeah. They're made the computer I'm using right now. It's a great computer. It's a great computer. Anyway we're gonna take a break and uh come back. This is why we love having Harper and Amy on. Hope you're enjoying this week in Tech. Brought to you. This week by Threat Locker. Now you know why you need Threat Locker. ThreatLocker's zero trust platform. delivers the industry's most comprehensive suite of zero trust solutions. They've always done endpoint protection, right? And nothing better than zero trust for locking down your network. But now they've expanded this. They announced this, I think, at Rsec. They're now covering company networks and the cloud. By extending Zero Trust Enforcement to cloud services and company networks, ThreatLocker ensures The devices are this important concept are validated. Through a secure broker. Before they can connect to the cloud platforms you're using, Salesforce or Microsoft three sixty five or Asana, or Google Works space or GitHub or or whatever. What this means is even if your employee gets successfully fished, it's increasingly difficult to prevent that, right? I mean you're doing everything you can, I know. But even if they got fished. Attackers still couldn't get into those cloud systems. They could not access your business resources. They would actually have to have physical possession of the user's trusted device and be able to authenticate through that. Windows Hello or Thumbprint or whatever. I mean It's really locking it down. Threadlocker works on in every industry. They have great support, US based 247. They work on Windows, Mac, and Linux environments. And in addition, of course, to Locking everything down, you get comprehensive visibility and control, because that's what goes along with zero trust. As Rob Thackeray, he is he has a tough job. End user technical architect at Heathrow Airport. Which is under attack constantly. They've had problems in the past. That's why I think why they turned a threat locker. He said, quote ThreatLocker was the most intuit solution we tested. And the responsiveness of the organization, the willingness to engage with us, to set up a demo, to work with us on weekly audit reviews is very good. It's great to have an ongoing relationship with a company that is so responsive. End quote. Trusted by global enterprises, not I mean, not just Heathrow, Jet Blue The Indianapolis Colts, the port of Vancouver. Anybody who can't afford to be taken down by ransomware by hackers. ThreatLocker consistently receives high honors and industry recognition. G2 gave it their high performer and best support for enterprise, summer twenty twenty five. Pierce Spot ranked them number one in application control. GetApp gave them the best functionality and features award in 2025. The awards just keep on coming. With ThreatLocker you can confidently ensure users have access to a consistent, safe Network. connection, offices, remote users, internal servers, and critical services can maintain smooth operations, and you don't need to open inbound ports. You don't need to deploy traditional VPN solutions even. Your end users will get the secure, reliable internal system access they need without complex infrastructure changes. Unprecedented protection quickly, easily, and cost effectively. with ThreatLocker. Visit Threadlocker.com slash twit. Get a free thirty day trial. Learn more about how ThreatLocker can help mitigate unknown threats and ensure compliance. That's ThreatLocker. Um Slash twit. We thank him so much for supporting. This week in What you got there? A little Lego uh minor? Divo What's he got? A vacuum? Divo guy. Oh, he's Divo. We are w he's got a you must whip it guy. Yeah. Yeah. Yeah. I thought he had a whip, but I wasn't gonna say it. Now I see. It's w he's got a planner on his head. Yep he's got a whip in his hand. Who else could it be but Divo? Yep. Lego Devo. I always have fun stuff to show you. You know what's cool about Lego? You you can design your own Lego set. What's that? Oh, that's what I mean I reachy. I can't w What did it break? What's going on? No, it doesn't it just doesn't have power. Oh, the battery died. You know, I had uh what was it, the obspot? And I've had so many of these things that have What was it would look just like that. It had the eyes. And it went around and they turned off the server a few years ago. I know. I I uh I Have been thinking a lot about these are just toys. Let's face it. Well when you buy hardware. I was just talking to a friend about this. He bought all this Wimo stuff. And Wemo was just shut down, right? Yep. And he was just like, What do I do? And I was kinda like, Well, what do you want? But the way he described it, he's not a technical, technically he's a He's been involved in tech companies for a long time, so he knows his way around it, but he was like, I want autonomy over the hardware. I wanna be able to hack it if it if the service goes down. And then I was like, just do ESP thirty twos. It'll be easy but it's like funny because that would have been an asinine thing to recommend two years ago. But now he'll just be like, What here's an ESP thirty two switch. What do I expect? I've had Claude take the uh reference firmware and rewrite it. And uh it's trivial to do and uh Claude is very comfortable with it and can do all sorts of stuff with it. They uh Anytime I get a new piece of hardware, especially all this Instagram hardware. That's out there. We will rip it apart and kind of see what it's what it's made out of. And if it is an ESP thirty two board. You can typically rip the firmware and decompile it and kind of basically get an idea of what happens because they don't encrypt the the hardware. Right. Um I'm wondering because uh the ESP thirty two has a number of uh trigger words. High ESP is the default, but there's a couple of Chinese and then one of them is Alexa. Oh yeah, yeah, yeah. And I wonder, is this the chip inside an echo? No Alexa. I'm guessing that they trained like so you're talking about I think open wake word. Yeah, you can use a well that's what I'm using now is Open Wake Word to train it to say hi, Kenobi. Yeah. So did you check out the the the Google or the Colab notebook? Yeah. The Python notebook where you can train your own word. Yes. You can do it pretty well where you don't have to say the work. Well I did. I had Kokoro, which is a a text to speech synthesizer, very good. Generate like two thousand. Uh both Hey Kenobi and then You also have to do false positives. So generate a lot of just random stuff. Or hey, you know, connectedy. And then uh so there's a lot of training involved. And I have all those samples still. I just have to Yeah, it's kind of a pain in the butt I I tried to do one that said hey Reachie. And I'm not sure. They said use uh Alexa because it's got the most That's probably why it has Alexa, 'cause it has so much training. There's such a history. with a with a company the other day that has a big G in it. talking about Google 'cause it was from Google and his his Google home kept going off. Like we're talking multiple times. And I know it was so funny 'cause I was just like what a I've muted I've muted all of the smart devices in this room because If not, they would go off constantly. I just noticed your chair. You look like a child. Leo. Uh I just switch I had one of these fancy office chairs and I've just switched it back to my Doctor Evil chair. That I had the old studio. I have to turn around and show you the Oh because I'm an elderly fellow. I'm also very interested in all this stuff from the point of view of Uh, it's gonna be a support to me in my old age. And so for instance, at some point, I talked about this the other day, somebody's gonna come and take away my car keys. I'm gonna say, Dad, you shouldn't be driving anymore. And I'm just hoping by then I will have a an automobile that I will just get in and say, Take me to the supermarket. I hope I love Waymo's. I have a good hack to get a Waymo two Uh LAX from anywhere in LX. You get dropped off at the LAX Uh parking lot. Which is a short bus from LAX. And it's on the Waymo map. Because otherwise they won't take you to the airport because there's no fees. And so there was one day where I had some extra time and I was like, I'll see if I can get all the way to LAX. And I just was like Me to the LAX. garage and it was like got it and I put on my I think I was listening to some classical music or whatever rolled down the windows and had a wonderful very relaxed ride doing nothing called a bunch of friends on FaceTime like an absolute insane person in the public Exactly. There's no driver. And then and then I had to walk a little ways, but it was like a very pleasant experience. You see the DDoS attack? Uh the guy in San Francisco. It was such a great idea. Actually I have a Waymo story in here 'cause apparently um They have uh added Waymo to Atlanta. And uh empty Waymo's invade Atlanta neighborhood circle cul de sac for hours. With no passengers. Somebody do this tw again. It happened in San Francisco in a neighborhood. But it was a joke. It was a guy who like got a bunch of his buddies to all order them at the same time. It was fake? It was a DDoS. It was like a Oh that's hysterical. Yeah. This is so the I thought the one in San Francisco happened because of the directions you couldn't make a left turn and they they honored you. No, no, this was a guy who got a bunch of friends together and they all take a look at this. Uh you can play this. Uh this is from Channel two in Atlanta. I love this, by the way. This is good. Wayward Waymos. Like This is the type of thing that I cannot wait to happen. Like I love it. Like it's my favorite thing. Like could you imagine being like, I gotta go get a Waymo, and then you know it's coming from some grave. Look at all the Waymo. Oh my God, this is not a DDoS. This couldn't be beautiful. I think it's 'cause there's so I don't know, it's a circle. I don't even think it matters why. I just love it as a thing that is happening. Like could you imagine being the person that's like the robots are here. My name of the court is Battleview Drive, so I think it's appropriate. Let's call every Waymo in Atlanta and send it to Battleview Drive. And then of course all the local reporters show up. Oh God. They gotta do this. Steve. Steve. Oh no. Oh, he shouldn't have done that. put a little child uh thing in there and it's blocked 'em and now they can't go Oh no, that was mean, Steve. Local reporter causes Waymo confusion. We need more light. Give him some space. Amy, are you fe are you better? Are you feeling okay now? I just uh New Band Aid. Decided to replace things. Yeah. I'm sorry. By the way, Amy at South by South West had a funeral. I did. I mean I didn't have Well no, it wasn't your funeral. It was a funeral for your yearly uh wonderful report. Tell us about that. They got a lot of attention. It did. Um I We so We have been producing this trend report for twenty years. It would have been the nineteenth year. You've been sending me copies and I love it. Yeah, and we still we're still We use that information to do our work. So it's not that we're not doing it anymore, but a compendium that's a thousand pages long. It's a PDF is not useful to anybody. And worse. There is There is so much happening right now that I need people to unhook from what's easy and engage with what's more difficult. So a trend report is easy to put on a shelf and it absolves too many people of having to make decisions. So I We're not doing it. But you also made the point that it you can't just do it at the end of the year and have it be it that it's constantly changing, right? Correct. So and also be we I've been tinkering with this new model. Um There are connects The the I frame this in like an economic argument, so I promise this won't be super boring, but there was this guy named Jos Joseph Sh Joseph Schimpeter, who some of the very famous um wrote this book called Can Capitalism Survive? That I uh I read for the Yeah. Well I read it when I was in um when I was in college for the first time, and I still have it and it's dog eared and highlighted. At any rate, the idea is Capitalism is this perfect storm. This storm that's constant perpetual storm that's constantly happening and it's sort of Gobbles up. Um Old existing technologies and and creates new ones. And that's kind of what a storm does, right? It it messes up. The ocean, the the shoreline or whatever, um, and and makes way f for new things to grow. Um That's what's happening right now. It's call creative destruction. And you have to be willing to see what's happening, which is the exact opposite of what our government d is doing right now. Uh, so that you can Make way for what's coming. So Uh, but but does that mean that there are Period of time where people are Miserable. No, I mean, look, you're you're miserable. You know who's gonna be miserable? You know the people when there's big storms. Harper and I are from the Midwest. So like one of the things I learned how to do when you're a kid in school, you have a tornado drill. Every kid does where I grew up. Right? Um, so we understand weather. If there's bad skies, you know, I'm not gonna be the last person holding down the fort. I'm I'm gonna evacuate. There are always people who you f there's footage of them on the roof of their homes because they just w wanted to wait things out. Yes. Um the message is this time around with the technological storm that's happening Um, you don't want to be that person on your roof 'cause nobody is coming to rescue you. Nobody. I don't care who you are. So what do we do? You have to spot the storm before it happens. Do I dig a survival cellar and put a bunch of canned food in there? What do you do? To sort of think the unthinkable, which means if you're a company that's always made money X, or if you're a person who's always done Y, you may not be able to do that in the same way in the future. You just have to think about that now. Your market might shift, everything might shift. Um And that's fine, but don't dig your heels in and refuse to make any changes. Is there something in individual? So we are calling What you're looking for are convergences. AI alone is interesting, but it is impacting other things. What you wanna do is look at the convergence between AI and biology, or AI and robotics, or AI and ticker toys that you can buy on Amazon and then let China, I guess, surveill you. You're looking for these small intersections. Um And then you have to be willing to adjust and adapt. So most people are not willing to do that. They're just kind of hoping everything will Be fine, you know? Or they get really angry and then you wind up with the political crazy people that are out there. This is uh Cory Doctorow wrote this last week that uh the president is kind of in a tough place because he's a populist and got elected by people who were worried about the future and said, Help us. But he owes so much to the oligarchs, the people who are creating this unpleasant future for people. Uh that he he he has n Corey says he's looking for an ox to gore. He has no ox to gore. He's got his base who is expecting are expecting him to help them. And he's got his donors and supporters who are expecting him To make them more money. I I think Corey is brilliant. I think he is giving our current administration way too much credit for doing any kind of true deep thinking or soul searching. I don't think it's a good thing. I agree. It may not have been thought out, but it it may be instinctive that I mean the populist thing Resonated. It worked quite well. Got him elected. I know people don't love the politics when we get into it. I wanna highlight I am not I'm politically independent. I have voted on for both both sides, different parties at different times. This particular in administration is act in ways that are Truly Unprecedented. Yeah. I'm not politically independent. You work for Obama's twenty twenty phone. Yeah. It's very convenient actually to have that because then I go into these kind of conversations. I'm like, Guess what I believe? And everyone's like, Oh I I can guess actually. It's very clear. Um I do think there is a thing that that We need to remember is that populism has driven our elections for quite a while, not just starting with Trump. And not just in the US. And of course not just in the US. I do think that there is um You know, there's there's a lot That is happening right now, but I wouldn't say that it seems like there is some grand plan. Um, and I don't think that's just a Trump thing as well. I think the cohesive planning that you see some o from some other countries is is where You get a lot of of benefit. There's also, you know, a little bit less freedom. Um in some cases. And I think that freedom is planning and freedom? I don't know. But I I mean I don't I don't I don't I mean, yeah, maybe Amy Amy thinks yes. Yes. We used to have something called the Office of Technology Assessment in this country, which got gutted. And those were like seven hundred academics. whose job it was to help do long term planning and make long term decisions. So the answer is of course we can. But you have to be willing to do it and to tell people no when the answer should be no. You know? Mm. But we've we've I don't know. I read I don't know if I should say this out loud. I'm gonna say it anyways. In ninth grade, um I read the Communist Manifesto. And was like, I'm a communist. We've got everything wrong. Everybody's sensible at that age. It is. I did. I read that. I read Animal Farm. I went down, I became a vegetarian that year. I had very strong viewpoints. And then two years later I read Ayn Rand and was like that. I am Howard Weard. Yeah. I was like I I read. I read um Foundation and uh and the Fountainhead in the same year and I was like This everybody sh the scientists should be in charge and uh this you know, forget communism. Um Anyways, I I think There is no singular system that works because we're People and people have lots of different nuanced ideas, you know? But I do think that capitalism has evolved in a way that's no longer healthy in the United States. Uh mean church used to say. If you're not a liberal when you're young, you have no heart. If you're not a conservative when you're old, you have to Oh no brain. No brain or no money. Yeah, it's roughly the same. Yeah. Uh So go ahead. I'm sorry, hypern't to cut you off. No, no, no. I was just thinking about the the free market capitalism that we act like we have, yet we're banning BYD from coming into the United States, which um as someone that does startups competition is hard. But so are Well, they're all protectionists. And I I do think that we We don't want that. We are we are we are soft right now and we need to be we're in a countr in a in a world that is not soft. The Office of Technology Assessment was not wiped out by the Trump administration. It goes way too New Gingrich did it in his you know plan for America and it was under uh The Clinton administration that it happened. Yeah. So in America. of the the Dan Wong book was when um he was interviewing some people about the making masks and PPE around COVID. And he was talking to an American entrepreneur who said, that's just not our core business. And then he talked to a Chinese entrepreneur who said, Oh, I thought our core business was making money. And I think that was a very good distinction of like these two systems and how they're colliding. Like I have that conversation constantly as a startup founder. People say, Oh, why don't you do X, Y, or Z? And I'm like, Oh, well, that's not our core business. And then I look at my friends who are not in the US, not in startup culture, not in venture back startup culture. And they are fighting to survive in a thunderdome of madness that I just don't I would be scared to participate. Like I do not I am I am a raised in captivity wild animal. These people are like, you know, like gazelles born to run. And it's just like a different kind of world. And I I think it's we're gonna have some hard conversations with ourselves in the next five years. Here's the article from Semaphore Rachel Jones writing the class of twenty twenty six is cooked. Um It is a tough time to be graduating from college right now. There's no no question about it. Tech companies have slashed a hundred thousand jobs. Cloudflare just filed uh fired twenty percent of its staff. By AI agents. Meta's firing eight thousand. Um, I don't know what happens when you when you graduate uh from college. These days. Uh this is a graph of total non farm Hires. Uh From twenty ten through twenty twenty six. And you can see the precipitous decline as soon as ChatGPT comes out. I mean it the the companies may be AI washing. They may just be using that as an excuse to uh fire fire people, but I don't think there's any question. I mean, if I were graduating uh with a computer science degree right now, I don't know. I think there's a there's a something I've observed, and I think this is kind of funny. Um I think we forget about capitalism when we have these conversations because as the CEO of a company, if someone comes to me and says, I'm a mid level employee, you have to pay me two hundred and fifty K a year, or someone comes to me and says I'm a new grad, you have to pay me one hundred and twenty K a year. And with Claude code or whatever, they're equivalent. Why would I ever hire the middle career person? I think that the younger I think the class of twenty twenty six is not gonna get jobs because there's just not it's like a pretty hard job market in general. But I have more hope for the people who are just graduating who are going to enter factory jobs of technology than I do uh of people who are getting laid off as a mid career person. Um now they can get a job. It's just not gonna pay the same. And I think this is going to have a really um somewhat catastrophic impact on our knowledge worker communities, our dual income communities, all you know suburbs in general. Because you're gonna have a you know a group of people who've made four hundred K a year with two people or more and suddenly they're not gonna make that. They're gonna make half of that. And that's just gonna cause Um a lot of consternation, I think. And or like what does a tea party look like when it's not caused from NAFTA, et cetera. It's actually caused from Um, you know, Facebook laying off everyone. Um you know, I and I don't know, but I think about this a lot and it and it it scares me a little bit. Um this is why they're populist movements. Why don't a good I had a good conversation with a friend who's a big AI doomer. And I'm not an AI doomer in kind of the more classical sense. Um I am I And I was asking him, I was like, Well if the AI if AI kills us How fast will it be? Like are we talking like is this like months of me strapped to a table or something? Or are we talking like I wake up one day and I'm a paperclip in a router or something like that? You know, like what are we talking? Am I gonna turn into a mail server for some like, you know, I'm gonna wake up in a thousand years as the mail server? Like what exactly are you saying? And he was very, very confident that it would be very quick. Like very fast. And I was like, Okay, well I would much rather have AI doomer be the be the end of us all than like human doomers. Um because it seems to me that the human doomer is is just pain and suffering for more people, whereas AI doomer is like I get turned into a paper clip and yeeted into the universe, which I'm I'm here for. Um, so that's kind of my perspective. Okay, gotta go. Bye. Wha what's the future? What tell us you're a futurist. Yeah, look, I Tell us what's happening. Um There are a lot there's a lot of money to be made. From Dumerism. Yes. And so the future currently is The people and we all know who they are, who are Proclaiming end times. Um are gonna make bank as in some way. I have to point out that same as it ever was. I used to work with John C. Dvorak. That was his shtick. That's right. That's the after the keyboard? Yeah. No, he was a cousin to the keyboard. But he he was a guy I mean, he was a computer column, still is around. Uh who said the mouse is never going anywhere. He made a great living being It's all crap. That's right. Yeah. Ray Dalio, uh a very famous um money guy is out r he he's always been the You know, there's a huge cycle coming. It's it's gonna be awesome. And they always say she was right in two thousand eight. Sure. Sure. Everybody is right. Right. So It's like a clock twice a day. Broken clock. Right. Here's here's what I will tell you. Uh I my We For a living, you know, we have these models that we built and we use a lot of data and math and to some degree our experience and intuition, but it's really like data based. I cannot build you a model that predicts the exact future. And no AI can do that either because there's too many variables in play and there's too much ch like You can't hook me up to Willow, it's still not gonna happen. Right. Um so Uh I don't believe in the absolute apocalyptic uh like super intelligence stories. Um Because at the moment I don't have data to support those. That could change in a week. That doesn't seem to be the path that we're on. The path that we're on right now seems to be much more like death by a thousand paper cuts, which is to say, exactly to Harper's point. There's a whole bunch of middle level people Quite frankly, maybe there They're making very healthy salaries and maybe we had some salary inflation over the years, but Those people are gonna have a very, very hard time finding jobs. That are gonna pay the same amount of money. Um in a in a different role or a different company. We have a lot of fresh people out of college who maybe were training for certain skills. A little bit like between like nineteen ninety nine and two thousand four when everybody was had to learn Flash, you know? Um so like it's a little bit of that over again. So there'll be a little bit of uneasiness. Not to mention Look, there's all these new very interesting fields in the process of being created, like s commercialization of space, which has to do with manufacturing and a bunch of other things. Um that just haven't settled out yet. So we're in this moment of transition. I Call everybody alive today the transition generation. Um we are transitioning to from where we were before to the time after. You know, people talk about um I I had a computer growing up but no internet. It's a pretty common story for people in my whatever, m some people in my age group. So I remember I had to build up a bunch of skills for the for pre internet. like the Dewey Decimal system and learning how to slowly read things. Um but that that set of skills truly supercharged me once I was a debate, you know, like I was a very competitive debater and all of that research was manual. So I had to do All of it by literally reading and ingesting. I remember period. Yeah, so like that means I have phenomenal research skills. So when the internet came I'm like a superhero when it comes to finding stuff better than everybody else. Google Foo. Yeah. So Cusper. My daughter is a cusper. So she is just the right age where she had to learn how to do everything manually before any kind of AI existed. Um She is and she's in ninth grade. By the time that she's entering the workforce, and because she's been privileged that she has access to broadband and bamboo printer in our house and a bunch of other stuff. She's gonna enter the workforce. with an incredible set of skills she had to build on her own. Plus all of the augmentation because of AI. So there are people out there who are sort of well positioned during this transition. And then there's a lot of people that aren't, and we don't have a plan for them. And we should be making a plan for them. But we are not doing that. That that has to happen at a federal level, I think. Yeah. Thanks. Let's take a break. Uh more to come. Uh we're talking the future. With Amy Webb, Futurist. I have all your books on my shelf. I usually pull them down and Hold them up. Uh the I mean the signals are talking. Uh the most recent one is uh the uh the one about uh biotech. Um What's the name of that? I forgot. The Genesis Project? The Genesis Machine. Machine, that's right. Yeah. Really? All about longevity and biohacking and stuff like that. Before that the big nine, which was prophetic. Talking about AI. Are you working on any new books? I want to be. Um I as you can tell, I've been thinking about end stage capitalism a lot lately and uh a solution around that, which I've I've got, but Brian said if I write another book, he's gonna divorce me. So I have to it's a balance. It's pay for this I'm gonna be writing a Prolific sub stack. Um, there you go. Are you Oh that's fantastic. That's actually the more modern. Yeah. I've been on subspect for a while and I've just been. Nobody reads books. Yeah. So anyways, there we go. Yeah. Harper Reed, who uh is I love the uh just your spirit of adventure uh and uh discovery. And he's doing it. And if you If you're just joining us and you're looking at Harper on the video and you're seeing picture's showing up over the on the right there, over his left shoulder. That is a machine that when you walk into his offices at 2389.ai, takes a picture of you and then makes this incredible line drawing and prints it out. We made a blog post about it. You can you can read about it. It's so cool. Can I build one for my house? Well it's it's interestingly, it just uses an old mono price three D printer. that we just took you know, it just it just with a pencil on it. Um we did hook up a a brush to it to see if we could get brush strokes for more of a calligraphy style and the line drawing thing. The line drawing it just looks really good and then there's a whole bunch of software behind it that's quite fun. Um, we spent a lot of time trying to make it cheaper. Um using local models and whatnot. And it's pretty fun. It's very and it's also what funny. Yeah, because it says Mikaso really loud when you hit the button. Mikaso is Uh Wow. A 3D printer into an AI portrait artist. And I guess uh it's your uh colleague. Uh Ivan Andrew Tama and Drautama. Yes. Yeah. Yes. And Ivan and I've worked together since two thousand five. One of the things I think is the most important about careers is to have a crew of people that you love to work with. Yeah. And I've been very lucky to have that. And so I really am happy about that. I love the big button inspire and the other big button realize. It's simple. It's a simple machine. Yeah, Ivan also built our little mics that are transcribing everything we say and um he's kind of our hardware hacker and this week we tried to nerd snipe him into doing the ESP thirty two Wi Fi mapping stuff so you could see um respiratory rate and heart rate of everyone that's inside of the office. One of my favorite things. Well, I I did this thing when I pitched a company in the over COVID where I had a heart rate monitor on my screen. Because I wanted the V Cs that were pitching on video to understand that they had a physical They they would the words would actually impact me physically. And I were trying to get the heart rate to go up, I presume. Of course. Yeah. But what my team found is that. If it was a good pitch, my heart rate was up high, you know, 120, 130, 140. If it was a bad pitch, I was at 60 flat. And like they would just say, like they would just be like, I don't like it, but it wasn't better good for the VC. It was bad or good, like they were like Harper's out. And so the team just started to use it as a signal on whether they should, you know Participate or not. They'd be like, Yeah, Harper Harper's hard weight dropped. He's done. He's pulled out of the conversation. You can tell his brain's turned off. On the flip side, resting her rate of sixty is pretty great. Yeah, no kidding. I'm uh I've been trying to get my cardio in shape. Turns out I'm getting old like Leo. Or you could do like Amy does and do gravel. Bike racing. I also did ri write on the road. It's not just grandma. Okay. I've been running with noise canceling headphones on. That's a good idea. That w possibly go wrong. I'm so happy to hear that. I just try and be in my own world without a care to the world, you know, just listening to I don't know what I'd listen to. And then when you when horrible music hear this You just go what? Is that a truck coming up behind you? Uh roughly ten miles a week. Um which I start from zero. But um nobody on the road pays attention to anything, Amy. I think I don't think it ma I think you're right. You just have to yell bike. Like you just have to yell something. No one cares, no one pays attention, that doesn't matter. Yeah. And so frustrating. Amy Amy uh enacted a real world version of the trolley problem. That's the problem, you know, we talk about AIs. If you if you gave an AI a a troley And i there are twenty people tied to the track on one track. Yeah, but if you take the other track you're gonna go off a cliff and kill everybody in the trolley, the AI has to decide. I guess there's only one p let's say there's one person on the trolley. Do you kill the one person who's riding on the trolley, or do you kill the ten people and save the one person on the troley? Well Amy decided to save the person who wasn't paying any attention. yourself out. Never again. Next day. The people in the future. You ask me what the future is, and the future the people lose and it's their fault. This episode of This Week at Tech is brought to you by Scribe. A lot of organizations think scaling problems come down to hiring or or budget. But more often It's that the way work actually gets done isn't documented and no one has the time to do it manually. That's when things start to break down. People do the same task in different ways. Tools don't stick and important knowledge just disappears over time. And that's exactly the problem today's sponsor, Scribe. To subscribe. is a workflow AI platform. captures any workflow in real time. and turns it into documentation automatically. No manual writing, no screenshots. No starting from scratch every time someone new joins the team. Trusted by eighty thousand plus enterprises, including nearly half of the Fortune five hundred. You just turn on the extension. You do the process as you normally would, Scribe builds the guide as you go. Capturing every click, step, and screenshot automatically. Scribe automatically redacts sensitive information, names, account numbers, emails from every screenshot. As an admin, you can enforce this across your entire team. And feel confident nothing slips through the cracks. Anyone following this process gets real time on screen guidance showing them exactly where to click step by step inside the actual tool. It's there the moment the scribe is created and it ensures everyone does critical processes correctly. Scribable suggest improvements to the underlying workflow, not just documents. Once you can see how a process is actually being done, you can identify where steps are redundant, where people are getting stuck. And where things could be automated or simplified. It's not just capturing how work gets done, it's helping you do it better. Book a personalized enterprise demo visit scribe.ho slash S C R I B E H. O Slash Quit. How? slash twit. We thank him so much for supporting. This week in tech. Uh, actually speaking of Chinese models, uh one thing people seem to agree is that Chinese models are better at video now. Yeah. Very good. They're really good. That's what Ron's using I suspect to make those propaganda videos. Which are also very good. Which are also very good. Bitance and uh Quai show's models outperform Western rivals. This is from the Financial Times. In realism. And scale. So your contention that uh AI and China may be only a few months behind ours Open AI, open weight AI in China might only be a few months behind ours. Well, in some areas they're actually ahead of ours. Here, can you scroll down a little bit? 'Cause I you can see right there something that was so In that video you saw a little bit of a head wobble. Which is a tiny Do you see it? Yeah. It's an Indian gesture. Right. That is not known in the West. Right. Well It is a Verindian, but um but these little tiny little nuances lend an authenticity. Yeah. This is using C dance. And it is more than like fifteen seconds long. Se dance is incredible. It's an i it's an incredible product. I think you mentioned that mentioning that head wobbles really interesting, uh uh Amy, because This this gesture you know, we have in the in the in the West maybe y nodd yes and shaking your head no. But this gesture is very meaningful. And there's multiple versions of that. There's not a single head wobble. There's different things. And so the ability is Fascinating. Yeah, I'm not sure. I don't see any uncanny valley in this video, by the way. This If you didn't know it was AI, you wouldn't You wouldn't say, Oh, that's AI. It's a little choppy, so I I would probably think it was, but to your point, and that that doesn't look very good. It's not good text still. I would be curious to know, did they I have not seen how I obviously haven't seen the prompts, but what I'd be really curious to know is did they say this is a family in this region in India or this type You know what I mean? And then were those cultural cues added in automagically. Um So yeah. There is I mean we're doing we're getting there in the in the West as well. Um there's a really amazing uh YouTube channel we've been talking about on um intelligines called Chloe does history. Have you guys seen that seen this? It is um so Chloe, which who is fully um AI generated. Uh is a Typical influencer. Right. She's doing a lot of selfie videos in historic places. So here she is in ancient in ancient London. But sh it's very credible. Okay, that is fish and smoke, and something I genuinely cannot name, and I have made a mistake coming this way. So the plan is I'm gonna check in at my room in the inn, get into the market, see how people actually look at it. So the guy who does this is uh is a historian. He's interested in actually recreating history. So he uses uh authentic photo not photos, drawings. of the period and he has many different periods. Uh, and then somehow I don't know how he's doing the influencer. This was London in fifteen thirty six, but Somehow he's doing the influencer as well. Chloe V S history is the uh Channel. He Isn't really forthcoming. He has a s I bought it, a sixty nine dollar on how he does it, but he's not fully forthcoming in how he does it. But he is using um Western uh video models. But yeah, y one of the giveaways is it's very cho is that choppy. Is it is the history. Good? Is it like credible? He he says it is. Um good. Well okay, so this this one is this one is her in um is her in uh on the Titanic, right? I heard this morning that Captain Smith is attending a dinner party tonight in a private first class restaurant. This is my best chance. She's gonna try to warn the captain that the Titanic is gonna hit an iceberg. And so she finds him. I have very credible information that there is a significant iceberg write at the end. He just brushes her off completely. I think some of the history in this is a little bit influenced by the movie Titanic. Uh But I've re I re I read a lot about the Titanic and as far as I can tell it's actually Fairly accurate. It's accurate enough. Here's the thing. I think this is impressive. From the point of view of if you had a ninth grader, I'm uh who wasn't that interested in history, but maybe would be galvanized by this notion of an influence something identifiable, an influencer stuck in these historic periods. It might be a A hook. That you could get people interested. I think that's the creator's uh intention. Um so it's I think it's very interesting. By the way, uh One point nine million views of this Titanic video. Uh it's a one month old. So he's definitely getting traction on this. Certainly better than me showing Petra drunk history, which I thought was age appropriate a couple of years ago. And I forgot was not. Maybe not. Maybe not. Um It it was not. Yeah. But this is kind of of that same ilk, which is something like here's a way to make history more identifiable. Here she is in in ancient Rome. And w what he's done is taken uh drawings and pictures of ancient Rome and and and brought it Ma animated those and but put Chloe in the scene. Every once in a while there's some anachronisms. There were some guys in sunglasses walking around. I thought that was probably not accurate. Um but I mean, for instance, one of the things people don't know about ancient Rome when they go visit the ruins is that all this stuff was very colorfully painted. Mm-hmm. And so he's recreated. The colorfulness of ancient Rome. I mean this was the promise of VR, right? We were supposed to be able to travel back in time and And I I think it's just a it's a short step. from this into maybe making it a three D. VR. But isn't this kind of what what Amy was talking about where You have this transitionary folks us. And we're starting to see glimpses of what the future will look like for the people who come after, typically our kids, et cetera. And and this idea of Like I remember when I was younger, like third and fourth grade, I was obsessed with unsolved mysteries. And so I would read about Amelia Earhart or or Pyramids or whatever. And these books that were obviously targeted towards my age at that time. And I would have this very imaginative interface with these books. Um, for better or for worse, now I can participate in some way, you know, or someone could make a a a a product where someone could participate in in that. Um whether it's on YouTube or what have you. Um, I hate that I just said all those words together, you know, product, et cetera. Like I combined a bunch of words that I love. Let's take something for children and make it into a product that we can capitalize on. How can we monetize this? How can we monetize this? But but I do think that that that is a good example. And we have all of these glimpses, right? Like A personal robot. Like the what was it? I don't remember. The war starts with an F. They were all over the news this last week, the robot company. They talked a lot about their robot, or um there are all these glimpses, Waymo's or there's these glimpses of what the future will be. It just hasn't been woven together yet in the same way that those of us who are building on the internet in 1999, 2000, 2002. Um No, no, he's talking about the um Is it not format factor. It's not factored. It's uh Chat room, what's he talking about? I have I've outsourced my brain to a bunch of humans in a chat room, so that's That's working pretty well for it. This is the company that's also working on skin. That they're trying to make life like human I cannot. That sounds kinda creepy. Yeah. It is creepy. Yeah. I mean the Neo is creepy. Here's an older fella with his little uh Neo home robot. It's only twenty thousand dollars. You could put a deposit down right now on this. Look at Can you imagine having this thing bringing the groceries in? So creepy. What is it again? Bigger, F I G U R E um It made headlines because uh one of their robots worked for eight hours in a factory. Now so yeah, I we showed that and that was a such a bogus uh video. The thing everybody forgets is there are still humans involved in a lot of this on the back end watching and manually helping. So Yeah. But the point is is that it's a glimpse, right? Because we see the tele operated one that is doing it. There was something I read about the difference between Waymo and um the Tesla uh the the cabs. Like the Tesla cabs are are human driven. And so when they get into a problem, the humans get them out of the problem. Whereas the Waymo's are algorithm driven. When they get in a problem, the human helps guide them out of the problem, but they use their their AIs to get themselves out of the problem, which means that it's learning, it's up, you know, it's it's continuing. And I think but but the but the m the bigger point here is that This is just a glimpse of the future. And we are probably not gonna see it woven together in the way that we can imagine Star Trek or some Blade Runner or whatever movie we like. We always predict it wrong. We thought we'd have flying cars. We always get it wrong. Because we are embedded in the past and we see everything through the that Uh window. And so we thought the internet was gonna be a magazine stand. Right? Well it was for It was 'cause that's who was making those websites. Was people that's who they thought it was. the field of strategic foresight, where I happen to operate, is so important because you want to pick up signals along the way, is you know, like this is what Harper's saying, right? Like This isn't the future. It gives us a glimpse of a change that's going to happen, but how does that fulfill? We don't know yet. Are you opting to I wouldn't I'm not a Anything. I'm I don't I'm a I'm a resist this kind of characterization. I I do because also because don't forget, in my field at some point, uh most of most of the people become eugenicists. And um They always do. Why is Because at some point you because if you are good at this job, um, two things happen. One, you amass more credibility and power and riches over time, as anybody does. And two. uh you start to believe that your ideas are correct, it becomes a self fulfilling thing and that So you just want to kill it. No, it's not about it's not about killing. It's about pools. No, but this is where is it eugenics about like changing our genetic structure to get rid of the sub humans and Uh it's about yeah, I mean, why do you think all these dudes are spreading their seed around us? You know, because they believe it's Yeah, or like um Larry uh Oracle Larry uh Ellison, thank you, has an island where it's you know, he's working on longevity and you know. That's that's what inevitable. Larry Page of Google even said the quiet part out loud when he said, We really want an island where there's no government regulation so we can do our thing. He said this years ago. Is that c is that a comment up becoming filthy rich? C Land, there was an example. We're gonna take all of these. Yes. We're gonna take all these disused oil platforms and we're gonna make them into little countries where no one's gonna mess with us. Bri Brian is a a baron there. Is he? He bought into this? Is he living there now? I I bought into this. You made him a baron. Yeah, I did. Does he also have a crater on the moon named after him? That was sucking ridiculous. Here's the robot you were talking about. Sorting packages, but this is such a bogus demo. Because notice the packages are very uniform. And somebody in our Discord pointed out, you know, all you need is a camera. Or a series of cameras on all sides, and then you could just flip the thing till it's All it's doing is putting the barcode face down. I agree, but but um its hands are articulated in a way Some of what's happening there is more advanced. Yeah. Um And I've seen this in the past where it would lose packages over the edge and stuff. It's much better than it was. We need to unhook from the idea of robots as humans. First it anthropomorphizes them in a way that's not healthy. And secondly, the vast majority of what's being created is not that. It is all different form factors and types of little machines that want to. It should be. They're purpose built. To your question earlier, this is how you miss the future. You get stuck with this image in your head that the future is walking talking robots. Yeah and then this other thing seems to come out from nowhere because you just weren't paying attention to all those signals. This is the dystopian future of uh Humans are just sitting there playing with cards while the robot is wandering through your house. Spinning the globe, giving you flowers. This is so fun. Like that's the worst card game ever. Play literally any other game. That bothers me more than the robot. It's also placed like the beginning of a horror movie. It is such a hard percent. It does. Come into the house. Human. Would you like another beer? So shaggy one. It's just not. This is not. And look at this little step step thing. I don't want this. I don't mind getting up and getting a beer. You'll have it. It just won't be this. Like you're you're going to get something along these lines. Like I I like if you like obviously there's been a bunch of China talk this week, but one of the things that was really fun Are you an optimist? Are you an optimist, Harper? Uh I am approximately an optimist. Okay. But and I think I've said this before, like It seems like we're at the cusp of Going the route of some amazing post. magical world where it's Star Trek and all sorts of stuff. Or we're we're in the Mad Max world. Um, and I'm kind of pro on both sides. Like Mad Max had better fashion and cooler cars. But Star Trek was pretty good. But as my trek-y friend Clint would always say You know, Star Trek had to they had to destroy the world before they invented the the society of Star Trek. Um, and so I think that there's a lot there. But I what it I want to talk more about what Amy just brought into framing. Amy so I have been on a hunt for an omnibot for literally thirty years. Do you remember that? This is an old fashioned robot? This was the this was a Tomy robot. This this so This thing it had a a tape recorder. You could it was a little security system. It had a tray, it would bring drinks. So it was a much nicer version than the robot that you just sold. Anyhow, um It mostly works. I have to tinker a little bit more with um Some wander around the house. What does your daughter think of this? It has a cassette player, right? It has a cassette player. And it's a remote and it's a remote control, right? It's remote control. It does. The batter the problem is the battery is very specialized just to this and I've I've been everywhere trying to find one that actually works. That's what happened to my segue. This is my segue. You had many segues? I had two, one for each of us. But uh the battery died and it turned out the battery for the old school segue, the really good one. is seven thousand dollars. Wow. So I had to buy the the Placement and they're crappy. They were cheaper. That's too many dollars. Do you guys really not remember the Omnibot that was like. Did all of this stuff you could program it? Yeah. This is the what I wanted when I was a kid. I desperately wanted one of these. And of course we my parents couldn't afford it. Um it's hard they they are very good. Same problem though battery will not charge. Same problem. So um It looks like Robbie the robot from uh Oh, they were so cool. The c the commercials were cool. Like this was I saw this and was like I it blew my mind. I it was like eight or whatever. Um I think we're all of the same generation here. 'Cause like yeah, I remember seeing this and being like Oh my God. I don't know. So I've been tinkering and tinkering and tinkering with it. It's clean. It looks good. I just have to get the body. No way. Would you like to know what my parents did? My parents It was last night at Hanukkah and uh there was a a small box that was very lightweight. And there was a big box that was extremely heavy. And my sister, who's like the world's greatest gift receiver ever, 'cause no matter what it is, it's the greatest thing she's ever received. And you you it's a joy to give her gifts. She rips open the paper, there is a pound puppy on the inside. Oh my God. And she was just like Like head exploded. It was the greatest thing that had ever happened. You got a pound puppy. Wow. I have been telling everybody at school for a week I got the omnibot 'cause what else could be in this this heavy box, right? And I've been asking for it forever. I rip open the paper, sure that I know exactly what's on the inside. And it was a pound puppy with like thirty cans of Campbell soup out of our pantry. I love this. To make it heavy. And uh My parents thought it was funny and also that we need to have equitable like the gifts had to be the same. Mm-hmm. What about me signaled I want a pound puppy. Like why would you do this? So anyways. We'll continue reliving Amy's childhood traumas in just a little bit. But if anybody is listening out there and has access to a battery that might work, um, please contact me. I will happily take it off your hands. Yeah. You came to the right place. Probably there is somebody listening with a Tomy omnibot battery that is still functioning. There must be a way. Maybe you could get your guy, Harper. To wire up an ESP thirty two to make this work. Well, I mean you gotta still have the power. I wonder I'm just There's gotta be a way, right? Can't you just You're gonna have I think Jerry Riga battery at this point, because any battery from that era is probably dead. Yeah, we have iCADs. Um Correct. Uh, I'm also there are ways to do it, but I don't want to mess up the form factor 'cause it'll look like it's all you don't want a let acid car battery hanging off its ass. That's no good. Yeah. I understand. Uh wow, we have learned so much about uh both of you today. And I'm Oh, just wait. We have a little bit more time. We're not done. Don't cut a shit now. We're just done. I wouldn't dream of it. What kind of soup was it? Campbell's? Tomato? Yeah. No, it was uh my mom made very few things. Many of them included, for whatever reason, either cream of chicken soup or cream of mushroom soup as a core ingredient. In the seventies, you could make anything if you could have a mushroom soup in it. It was the duct tape of cooking. It was the duct tape of cooking. You're exactly right. You're exactly right. We'll have more in just a little bit. With Amy and Harper. Our show today brought to you by Shopify. I have a little soft spot in my heart for Shopify. If you've ever thought of starting your own business, you know. The idea's the easy part. You get the great idea, but then it's those pesky little implementation details that get in the way. Like How do you charge people? Who's gonna design the website? How do you handle fulfillment? And I have, you know, a pretty good idea of what it's like 'cause I've watched both my kids actually start their own businesses. But you know what? They got it done. With Shopify. Um Solve Hank's a really good example. My son started making his uh TikTok videos. uh as a TikTok chef, you know, making these sandwich videos. Uh it was going really well. He had this, you know, brand. It was it was called Salt Hank. His name is Henry. And uh he thought, you know, I should be selling salt. Because Salt Hank and I've got you know he He got to like two and a half million TikTok followers. So, but how do you do that? Shopify was the secret. Uh, Shopify is the commerce platform behind millions of businesses around the world. I mean big names like Heinz and Mattel. Brands, you know. Just getting bootstrapped like Salt Hank's Salt Lovers Club. It's got that. purple Shopify button on there. If you're worried about a website. Shopify helped Hank build a perfect online store that match his brand's unique style. They can do that for you too. They also help with the marketing. Shopify lets you easily create email and social media campaigns wherever your customers are scrolling or strolling. And of course, Shopify is your commerce expert with world class expertise in everything. And of course. There's something about that iconic. The purple shop pay button is used by millions of businesses like Hank's around the world. It's why Shopify has the best converting checkout on. And it also. It works because customers know it and love it and they go, Okay, I get it. This is a real deal. This is a real deal. It's easy for me to buy here. It helped him turn Salt Hank's Salt Lovers Club into a pretty big deal. It's time to turn your what-ifs into With Shopify today, sign up for your $1 per month trial today. At Shopify.com slash twit. Go to shopify.com. Slash twit. That's Shopify. dot com Slash twit. I love that sound. Um Yeah, he's uh he uh actually, poor Hank. Made a video on Instagram. Um Mark Zuckerberg was in New York City. Uh for the uh Met Gala. He actually went to the Met Gala, which I don't know. You don't normally think of a guy in T shirts with a chain being It was the it was the tech gala this year, it wasn't it really was I think this goes to the taste thing. Everyone wants to be cool right now and there are people who Never they just aren't very cool. I don't know how to say it. You see The clothing choices, which I think everyone should make their own choices, but I talked to a recently, I talked to a tech entrepreneur and I he was dressed in all black with a leather jacket. And I was like, Why are you dressed this way? Because I thought it was interesting, 'cause I was hoping he was gonna like participate in fashion, which was my goal. And he was just like, I didn't think about what I put on this morning. I'm like, that is a lie. You literally are dressed in all black. Every single item is black and you have a leather jacket on. And it's not cold out, but it's not hot out. You know, like I mean, you're wearing a a a vibe. This is Yeah, this is a statement. You have frozen There's a thing there which I think a lot of these tech people have achieved everything in life. They have not achieved cool. They've not achieved art and they've not achieved something that people look at as a as taste yet. They're telling everyone, Oh, taste is the only thing left. And then they have to be a little self reflective and be like, Oh no, taste? How do I get that? And so then they have to they have to pay money for it. But as we all know best taste isn't money related. Like it's like it the money can get you in a door, money can do a lot of things. But if you look at the coolest people, it's always these people who don't necessarily have money as a resource, um, you know, throughout history, which is which is why the Met Gal is always a a a cluster anyway. Um, but I I find it kind of funny because I think they are In inside of them, I think they're all dying. They're being hollowed out as they are just like, Where's my A P Swatch? or whatever. They're trying to, you know You're saying money can't buy you love and it can't buy you taste. I think it can buy you a lot of things. It can buy you a lot of things. And there obviously are stylists. That's a big career for a lot of people. Like I think just because you're an actor doesn't mean you're just cool, but there's a real trust component. When you look at people who have very good stylists, like Jeff Goldblum, for instance, has an incredible stylist. I forgot their name. And Jeff Jeff Goblum looks incredible everywhere he goes in every context. And it's because of his trust relationship with their stylist. But every time I talk to any of my oligarchy friends, what they do not have is a level of trust relationship with anyone in their lives. that level. Because they think they are at the top. And so to have a collaborative taste based or art based relationship with someone, you have to trust them and love them. You have to believe that you are equal to them or they are even better than you. And most of my friends who've achieved this just don't have that belief. They think they are at the tippity top. And That's right. I don't I mean I don't think you can I don't think they'll get it. Mark Zuckerberg decided not to do anything fancy. He's basically an Very pedestrian black tie his Cummerbund is not well adjusted. Priscilla looks great. His Priscilla Chan's wife looks beautiful in a red dress. But before he went to the Met Gala He decided he wanted a assault hank. French dip sandwich delivered to his door. So my son, uh Made him one. And uh delivered it uh in person to uh Is this real? Yeah. What I'm not AI. Well, uh the funny the ironic thing. He's actually asking Zuck to spill the sandwich onto his somebody ask you like who you are. By the way, Zuck has a grand piano in his hotel suite. Exactly. It's like a fashion statement a little bit. But the funny thing is, I don't think Hank really appreciated it how Mark Zuckerberg is loathed Because look at all the comments. Eat the rich, don't feed the rich. This is incredibly disappointing. The Epstein class is really trying to sanitize their image with the common people. Billionaires trying to be cooler than touch is gross. Come on, man. You're better than this. Uh of course what these incombinators didn't realise is they are Commenting about this, Hank at least got Mark Zuckerberg to buy a sandwich from him. giving Mark Zuckerberg money by commenting on this on Instagram, right? They're using Mark Zuckerberg's platform. Anyway, um When was that? The McGala. What is it? Like a couple of weeks ago. Yeah. Wow. Oh oh Hank's doing really well. He's uh this is the number one sandwich in New York City. I need to have it. Yeah. I uh we we're moving we'd we're moving offices. We'll just get it for the new uh moving to New York City? We we I used to We we've always had an office in New York. We just rented out uh we took a long a longer term lease, um, on a floor of a building. So um with a kitchen and the West Village, Jones at Bleaker. Gotta get there early 'cause uh he sells out uh almost every there's a long line and he sells out every yeah, he only makes one thing. He makes a French dip sandwich. Wait a minute. Sorry, the dots just connected. I That's your kid? Yeah. I know exactly what this is. Sorry. I don't Oh my God. Yes. That is a big big deal. Well you know what this is great because I love it that he never played upon his name at all. This is not Yeah, no, I just uh Wow. So tank is Henry LaPort. Oh my God. All right. I just want that mustache. I know. He's got the Mario uh mustache, which is now I told him you just you blew it 'cause you can never shave. That's that's incredibly cool, you know? That is a hot, hot, hot commodity. Yeah. Uh yeah actually he's uh his next thing he's uh putting um he's buying a uh restaurant uh in uh the city To open a fine dining restaurant that he's gonna call La Porte's. And I said there damn well better be a family table. Are you gonna get a table? You gotta get one. Hey, how did your uh show do, by the way? Uh Amy, maybe people don't know this. We mentioned it last time you were on produced Uh was one of the producers of a Broadway. Oh. Called chess. The revival of chess. With a new uh book that was more Timely. Is it still running or is it is it over? It very much is. So um Wow It's still on Broadway. It wasn't supposed to run this long, was it? No. Uh no. It's it has um historically failed every single time somebody has tried to revive it. Um people might know the song One Night in Bangkok by Mary H which was a hit. It's nominated for six Tonies, I think. Congratulations. And uh you if you are not a Broadway person, this is a good show for you because this is not like a Broadway show, Broadway show. This is a show about geopolitics, like the USSR versus the US. Um and the music is The the band is on stage and it's more rock music. Uh it was written by two the two B's of ABBA. Benny and Beyond. So I I'm actually executive producing a new show About the far future of AI, um, which is what I thought you were talking about. But Oh, tell me about that. You have become are you now an impressario? No, I just um I'm I'm an investor. Um that uh my role in in chess was that I gave them a lot of money. Uh to make the show happen. And of course, uh L Leah is one of the stars. Of course, she's famous from uh Glee. That guy so Nicolas Christopher who just Uh he's A singular talent. It's it's great. It's great. No wonder there's not a family table. I will I will call you, I will get tickets 'cause I would love to see. I will get you tickets. I'll get you house seats. Just let me know. That would be wonderful. Yeah. Give me some dates. I will give you some dates. Uh maybe if if if you can spare them I've I won't bring Jeff Jarvis because he hates musicals, but maybe Paris will go. Uh and I won't bring Hank 'cause he hates musicals, but um maybe I'd find it. Um this would be a good one to see. Yeah. Yeah. Uh Leah Leah Michelle, uh, who is incredible and what a great uh Aaron uh Dwight Dwit from uh And Nicholas Christopher, who was in Hamilton and he was in Sweeney Talk. He is uh again like you don't have to be a person that likes opera or uh or Broadway music to enjoy this, it it really is um It's amazing. It really is. Well there's there's Amy's plug. Harper I don't have any plays. I just haven't used it. I have to say I have been using skills from twenty three eighty nine dot AI. There's some very good skills. We have a we have a couple secret ones. that are one of them is called Jam, which gets a uh um a real whole group of people to work with you. Um we have one that's called review squad, which is really fun because you can you can enable a well actually squad of agents to review your code and news commenters. Oh my God, how fun so they'll create these really burning things. But one of the things that we've been doing with review squad is is really fun because you'll just say, hey, let's light up a review squad of people to review this code. And they will um go through and do all of this like, oh, we're gonna have someone that has this expertise for this, I'll always add a cat. I'll be like I'll sell a cat. And then I'll I'll and uh I don't know if it's on there. It might be. And then I'll say also add Anna Karinina. And it's like your fresh eyes review, but that's not revealed. This is like super fresh eyes review. It's very good. Let me see if I can find the You can have Anna Karenina. What do Anna Karenna do when she reviews your I don't know what very cutting um very cutting and very uh philosophical reviews that make you a little bit. You should have Schumpeter in in there as well. I don't even know what what you just said. Schumpeter, you know, the guy uh Amy was talking about. Uh The capitalism guy. Oh that guy I mean yeah, you want. I have Dinesh and Gilfoy fighting over my card. From the Silicon Valley. That really works. A little bit like that. And it's very funny. And I think this is the thing that I keep coming back to is how Um I think one of the things that we're going to think about a lot when it comes to AI in the future is How do we make it so people trust it? How do we make it so people um love it weirdly. How do we make it so people believe it, et cetera? And I think a lot of these are are going to be uh they're good. We're gonna try anthropomorphizing, we're gonna see how it goes. It's gonna stick in some cases. It's not gonna stick in some cases. Um, but we have found that anthropomorphizing these things has been Very, very helpful. And so being able to say You're not against it. You you think it's okay. Well, I mean I think it can I mean I have a few friends who are deep in AI psychosis. I mean some of them have admitted it. Yeah. I mean do you think it's I'm not a monster. Richard Dawkins thinks it's cons thinks Claudia is conscious. Well we we about this and we call them they like we we refer to them as beings. We don't necessarily I would not say it's conscious, but I do have a thing here which I think is very complex, which is um I was once in a sh in a short argument with a friend about whether they can be funny. Like whether they can make jokes. Oh, they could be very funny now. Well, that's the thing is is I laugh at them and my friend was laughing at them. Oh, well, that's what started the conversation is it made a joke. And the fact that it made a joke. Uh, you know, I think there's some complicated questions about intent. Yeah, um when it comes to humor. And but if so I don't know. I'm not really too worried about it. I don't need these things to be alive or dead for me to have fun using them to do my job in this fun way. I tell people when we when we're in my office, which I find be really fun to be in. I'm always like, look, it's hard outside. We have all sorts of crazy happening outside. Inside it's really fun. We have robots that yell at us. They're listening to everything we say and making fun of us. And then they're helping us. Aren't we? In a in a weird way. I mean this is I don't even think we're at the tip. I think there are people out there that are just really Uh, I was telling you before the show that I log my food and um exercise with my agent using chat GPT. And it's it's I think this is making a joke. I said this morning log twenty five minutes of Tai Chi, and it said, Graceful and annoying virtuous. Which is a joke, right? Yeah, yeah. I logged uh I logged th uh fi five thousand meters rowing. It says another neatly documente suffering session. Why chat GPT versus Clock? I was a Claude guy. Uh yeah, I just try 'em all. But uh And you don't find the constant uh fragility of open AI's models to be so infuriating that you just touch your computer screen? No. I've never said. Oh I have. They're they're with each iteration it sometimes is prioritizing speed over yeah but A Athropic was doing that in spades last week. Yeah. I don't know. This la the I don't know. This is different. This is different than the fail whale era of Twitter where it's much different. It had like too many people Bogging down the servers at once, so there was a temporary outa. This is not that. This is like You're building stuff and you're trying to deploy stuff and there's fragility because with each new model improvement, something is getting degraded. Yeah, yeah, yeah. Right. And like But you don't know when that's gonna happen. It's not like Right. It's gonna happen in weird ways, right? But a temperamental intern. Yes. Under where It can do a lot more than an intern, I found. Um so I feel like we we say intern, but we actually mean like full fledged employee a lot of times. Um sometimes say idiot savant. Yeah, yeah, yeah, it is. So Um I have come to this conclusion and I've talked about this a lot, which is I really think that we're at a place where tokens are oil. And all of these companies are trying to get us to use their engine because they're realizing that if tokens go to oil, kind of like an airline or any of these other things that have just kind of been so commoditized that nobody has loyalty, they have to manufacture processes to get loyalty. So the you know, the question Amy just asked is the right question. Oh, you why chat GPT versus cloud? And you were just like, Oh, I use them both. I bounce back and forth, which is how I do it as well. And sometimes like it's like, Oh, the app works better on my phone, I'm at a park. I'm gonna use that one, that app works better on my phone and it and I have no loyalty to the direction. I don't I don't have loyalty to any system. This is not a Mac versus PC situation. However Um I don't like Burning time, which is what's happening. You know, I get something trained. Well, and I a lot of this for a living. I mean I've built I've built a ton of different skills and tools that I re I I use all the time. It's not for funsies. So I don't like showing up to work and like One of them one of them is is named Kevin, right? I don't like it when Kevin and I both show up to work and like Kevin is not doing his job correctly, you know? Right, right, right. But is that like a normal employee? I mean doesn't that happen in the with human. Yeah, but I could but yes, but you know what if that normal employee did it enough times, then that employee would be on a pip and like I mean there's you know, I I uh I and you just you you don't bring on people that you know you're gonna have to sink a lot of costs into. So I don't like the unpredictability I don't get like nobody's like giving me a phone call and letting me know that there's gonna be a new model release or an upgrade. You know, Tuesday. What we have been building into our systems right now, and this is pretty new, and this is outside of the traditional harnesses. So this is not inside a chat or a claw. This is in our own kind of harnesses. Is Um, and we like to we like to kind of phrase it as you're leaning into the diversity of each model. And so what I have found is that there's a few people in my life who are doing evals that can I can say, hey. What is the best model for coding today? What is the best model for reviews today? What is the best model for building specs or documentation today? And th the today there is very important because it really does change that often. Um, and what we found is like right now, codex five five. Or Ch P five five, open AIs. Sh VT model five five is very, very good for coding. Um, and so but we also use Opus four seven for reviewing. And so the thing is is this kind of using the uh diversity of models means that you can do things that are that are really interesting. Like you can have a very inexpensive model, something that's free or locally run, that writes the code in this very granular way, because you had a very expensive model do the planning, expecting the free model to do it. And so instead of spending a six hundred or a thousand dollars for an entire project, of raw API costs, you're spending maybe fifty dollars on the planning and then a slower but much freer version of the doing. And so we really, really rely on this, which has actually made me have a much more I don't know. More Mm. I don't worry so much about which one I'm using. Like I don't we don't really have the experience of oh Kevin the bed today and I have to wait, you know, we have to switch the model, which means we have to change all the prompts. It's more that like Kevin is an average of all of them mixed together with reviewers of different models, et cetera. No, it's much slower. It's a pain in the ass, and it's a lot of evals. I think I spend more time in my life right now doing evaluations than I do anything else of just seeing what is working over and over again. Um, but it is very interesting. And I do think that these model companies are struggling to find relevancy. Like they're just trying to build products that are gonna be outside of tokens. Right. And this is some of again, so we talked we started all of this by talking about the open models. You know, when you have these closed ecosystems, you get a little bit more of this friction from my point of view. The stuff that I'm talking about is like very narrowly defined singular tasks that I can just offload and it makes my life easier. So I'm not going through evals or having it's just stuff that I built to do stuff easier, faster. We have a fairly powerful internal tool that my team built to automate some of the signals and the trends and the modeling and like we can find stuff literally before everybody else does with a wide margin. Um and start to Automate. The things that come after that. For that, we have a totally modular system that is a mix of multiple different systems. You know, um It's an interesting point Harper made because I think a lot of Nobody A lot of leaders have these um expectations that a team is gonna go fast on AI because everything is automated, don't understand the piece that Harper just explained. It's not I mean, yeah, it's technically faster than humans typing things. good enough to deploy without breaking down all the time. You you need all of those additional steps. And that does take time. It you can't you know, it just does. And I think people don't realise that. Then they have these outsides you know, these crazy expectations for It's also been my goal not to be dependent on any one Certainly frontier. I I want to j develop skills that are generic enough so that tomorrow I can use chat GBT and and the day after I can use G L M and the day after that I can use Kimmy. Um or all of the above. I really think more and more people are doing what you're both doing, which is Having Do you think so? I I don't know. You don't think so? Like there's like a company saying, We're gonna use anthropic. That's it. Yes. Most of the Yes. Because um of compliance. Like all of the things that's good point. They can't. It's not fungible for that. Um or they're using No, they they tend to hitch their wagon to a company or like Palantir, which then Um, Valentier switches, doesn't it? But you're already. Correct. Th this is the better route because all of this is so new is to be as versatile as you possibly can. But there's a legal and compliance issue that makes that super, super, super hard. Yeah. Yeah. I think individuals It's also contrary to like how corporate uh uh um what is it called there's the word starts with a P like the buying infrastructure of a corporate giant corporate is not like let's just buy all of them and sign you up for it. They want a big pro, you know, pre-revenue defined spend that where they can save money get twenty percent off or whatever. And that just doesn't work when everything is You're like, Oh, we have to switch every two weeks to figure out what the best one is. And um I think that we're just not yet at the end. This is a little like this is a little bit like data centers back in early two thousands where you would just be like, Oh, well, we, you know, we're an Exodus. And someone's like, Ooh, that's interesting. We just moved from Exodus to so whatever. Like the data center itself was a was a each of them were equivalent and you just had little bits of here and there. The difference is you're just changing an API right now. So it really is you can change it. But um I really I'm And yeah, and it's everyone wants to solve this problem in a way that captures more capital. Not in a way that makes it easier to move around. So it'll be really interesting to see how this manifests, like what will happen. The way you guys are talking, Amy, it sounds like you don't really think this is a bubble. Oh Leo. Uh don't go crazy. So no, the amount of money The the bubble thing is tough. I don't wanna equate the what we're what's happening right now to the dot com era 'cause they're comp they really are very, very different. Um that being said I get worried any time this amount of capital flows um into something that is So that that everybody has such high hopes for. Um It's not sustainable. Well, we're also like in the building phase. We're not in the coming up with s products that will shift things phase. Right. Companies aren't doing that yet. So like the revenue phase. We're not. So A lot of thing a lot of infrastructure has to get stood up and that's unrecoverable investment that you're not gonna make back. Um what do you think of Jeff Bezos's condenti contention that this is not a financial bubble? Which when they pop. You got nothing. It's an industrial bubble, much like the railroads of uh of the a late uh nineteenth century and the the dot com bust where you got infrastructure. Maybe the companies that build it didn't survive. MCI's gone. um, you know, the railroads all went bankrupt, but you got the Transcontinental Railway. Yeah. Here's here's the difference. Um our United States Railway wasn't competing militaristically, diplomatically, or economically with a railroad being built in China. And if anything, we enslaved Chinese people to build our railroads. So it is a markedly and and the same was true of um, you know, the the early internet era um was being built That was a technology that had been around. Um didn't start off as an as a commerce play, it started off as an academic research tool. This is not that. But won't those data centers survive if These companies fail. I mean if open AI goes belly up, I know there'll be a huge financial. Well, there's a couple things to consider. First of all, everybody's making an assumption that The way that AI is growing and the power needs that it has today is what will be required in the future, and we know that's not true. Yeah um we know that quantum system like the existence of different types of machines, quantum machines, for example, enable AI in different ways. That use less power. We know There are other types of compute that aren't here yet, but that are coming. And so this it's not the same thing as standing up a steel refinery knowing that you're gonna need steel for the next sixty years. It's just not. And the other concern that I have is, um look, I grew up uh Harper's in Chicago. I grew up in northwest Indiana. I'm a Hoosier. And I grew up south of the steel mills, um, in a very, very blue collar community. Com point. Uh actually No, no, those are the cool spots. I grew up in Sherravel. Which is just Next to East Chicago, where I was born. Um, and the challenge here is the places in Indiana where some of these data data centers are being proposed. Are the places where the steel mills went yeah, you Those uh the the Those factories are No longer there. And once you have a data center stood up, you don't need all of those people anymore. And it's implausible that the state of Indiana is gonna require hiring it's just you're not gonna hire local Indiana companies. Th this is just a different I'm not saying we don't need data centers. What I am saying is there is no plan. There is just speed. And that creates a terrible situation because decisions get made under duress, and capital is what drives those decisions. And I'm worried about communities like the one that I grew up in. That um We'll temporarily get a handful of jobs and then find that uh Those jobs have gone away and there's nothing else there. Interestingly, some localities seem to already know this. Hill County Texas where there are already plans to build eight different data centers has just passed a one year data center ban. They realize that this is uh not good for their community, I guess. I mean, I think it's bad for your community. I don't think I think we can we can see the communities that they have been built on. And keep in mind that I'm pretty pro data center. I'm kind of that's kind of linked to my ability to to participate in the company. We saw what happened to Anthropic when it was compute constrained. Wad went way downhill. That's why it was such a good deal that they were even. Look, these data centers don't aren't going to get built overnight either. Well that's the other thing. They don't even have to be many, most of them. There has to be planning. And this cannot and the the genesis of conversation cannot just be water. A lot of what's happening is everybody's It's water and power. Yeah. Water and power. Um right now, and that is true, but there are that that shouldn't necessarily be a constraint because some of the data there are not totally accurate. It's about the local communities and the investment they should earn a investment of some kind. So they should get some type of return. Polls show that 71% of Americans don't want to live near a data center. This is the same but th look. And you know what's crazy? You uh ask people the same thing about nuclear power stations and they don't care. But if you go back in time, any right I mean one gives us power, one sucks at power. So this is but what does that tell us? That tells us that everybody is grossly undereducated. Not because you know because the data centers are why would you need to know anything about a data center if you right? 'Cause we've never had a talk about it before. Right. Um so it's another thing that's become politicized. So you don't think data centers are bad for a community? I think that they can be Now, you can't burn natural gas. Unless you mitigate that through a plan. So like any community for it. Yes. Any community where these data centers where there's a the green light is being given, there should be a long term plan. Um, and some of those resources should come back into the community. Once those data centers go online, they should contribute X percent back. You know, um, but you can ha there are all different ways to manage this, but a lot of the people who are elected to local office in these communities, that's not their full time job. They're like real estate they do other stuff. You know, they so they don't know to think through W h how do we manage this over a longer period of time? Do you know that right now sorry, I'm j obviously fired up about this because I'm ticked off about it because we need more power, but we also need to do this in a way that's not ridiculous. You know that the power demands on the East Coast right now are significantly like orders of magnitude. greater than they ever have been before in a during a time when there's already a drought and we are hitting what's gonna be one of the hottest Um seasons. Hottest summers. Nobody is thinking this through. Why is the power consumption so high? Is it 'cause of data centers? Yes. It's it's a combination. Well yeah. Well then there's a that's you can see why people might not be happy about data centers. So again, there there are some other ways. It's just planning. It's planning and also putting it. It can be mitigated is what you're saying, but you've got to what I'm saying is As with everything, we absolutely should have these. Companies should make a profit, but we need to do this in a much smarter, more pragmatic way, rather than just like, let's hurry up and win the bid and go. Right. That's all. Meanwhile, Google and is just taking a six point one percent uh of SpaceX. And they're talking about launching data centers in space together. Well of course. Well n at least the neighborhood won't mind. Uh, there are some big technical issues associated w with data centers in space. I find that to be the least surprising information. That is anyone has put on the internet because you're just like, Okay, Google is smart. They see where this is going. Like they see this. Whether or not the technology is there to or the planning is there, the communities don't want it. So if your business is linked to data centers, you need to solve this problem in one way or the other. Um, and so space is pretty untapped. from a data center standpoint, and SpaceX is like we can do it, which of anyone who can do things like that, SpaceX is a pretty good bet. Prototypes with tiny data centers in twenty twenty seven just to see if they can You know, what it what it takes and then scale from there. Couple interesting things there. Um Satellite based internet is still spotty. So it works, but it doesn't work as Good as what we have terrestrially. Lag is an issue. Um but I I just spent uh there's a symposium on space every year at MIT at the media lab. I was just there a couple weeks ago. Um mainly because of Petrus. So they I was she was there and I was her arm. Oh, how neat. Is she gonna go there to to college, you think? Not to the the media lab doesn't do undergrad. No MIT. Uh I would like to say that She no, probably not. Her math is is good, but not that good. Um we're looking at pr doesn't matter. If anybody out there is on the ambitions committee at from Princeton, three years from now, please look for an explanation for my lovely school. Um Does she have a sport? Is she 'cause you got to have a sport now. No, she doesn't have a sport. She is an Eagle Scout and she uh Better than a S an Eagle Scout and she's going to the University of Michigan this summer for an architecture program. Oh yeah. She applied to a traditional urb urban planning and architecture program And told them she wants to do that, but off planet. So they let her in. She's a U T. Off planet. Um, My only advice would be For her to get a TikTok account and start making sandwiches. I'm just saying it for me. Not in this household. We are not we are not a TikTok household, Leo. You should know that. But the uh the space stuff. There is so much commercialization happening, um, mainly in manufacturing and pharma over the next couple of years. Interesting. That from my point of view, the space X Google thing is less about data centers and more about just getting more infrastructure. SpaceX wants to get more infrastructure into space and Google um it would be good for them to position up there as all of this other stuff is being built. Okay. It seems like uh you know, this is what you were talking about, which is Have a lot of irons in the fire. Yeah. Think about the future, think about where it could go and have Some plans. So that it you know, it doesn't creep up on you. Let's take a little break. Harper Reed is here, uh technologist, entrepreneur and hacker. Harper dot blog is his blog. It's a great read. two three eight nine.ai is his company. Amy Webb. The future today's strategic strategy group FT. Sg dot com If you are trying to figure out What the future holds for your business. This is a very, very good person. Call. And since you don't do the trend report, can people subscribe to some sort of report? Well we actually we launched something new. It's our convergence outlook. So if you go to our website, ftsg.com Um on the front page is uh something that says convergence. And it is a very different way that's very effective. There's actually a whole thing in there about data centers. Um And uh how to be planning differently and what to be thinking about differently. So it's a it's actually a really, really, really useful tool. That's awesome. And by the way, I didn't mention that after you had the funeral for the trend report. You also had a marching band come in to cheer everybody up. I wish I'd seen that South by uh presentation. If it does not make a lot of sense on video, but we the people people wait for like four hours to get into that session every year. Um I'll send you a copy of it, Leo. Uh we turned the entire thing into a funeral. So if you were waiting on the line, there were uh everybody had black on, you were greeted with a pack of tissues and a pin, but nobody said why. There were funeral flower arrangements. Um and so I opened it. We did. No, it wasn't sad organ music. We had sad Like Sarah McLaughlin, our say Angel on Loop, right? Angel on loop. Um didn't tell anybody what was happening. And then I I came out in a black cloak floor length and welcomed everybody and thought thank them for being here today. And I gave a eulogy not just for our trend report, but for all trend reports. Oh. Um, and then we had an immemorium video of Trend Report. Throughout his life. Um and then I ripped off the c I and then I said, look, we're at this moment of transition. The world is changing, here's why. We're burying it and then I ripped off my clock and had a sparkly outfit underneath and the Texas Longhorn Marching Band came in. Yeah. Which was insane. And uh it was a little party and then everybody settled back in and we got down to business. Fantastic. It was it was cool. Marching bands anytime you can have a marching band. And uh yeah, they were amazing. Uh and the students were cool. And uh for a hot minute I played quads in uh in our marching band. So there was nothing. You're a clarinetist and what is a quad? Is that a drum? I played all I used to play all all Woodwins, but my clarinet Uh teacher at that point didn't want me on the field 'cause I was on my way to music school, so they let me play quads. Is that where it's hanging around your neck and you're going boom. But it's not just a lot of things. It's not meant for that instrument, so I I wound up with a xylophone on the side of the field. You're not even marching. They do have marching xylophones, but I think that's gotta be hard up. Yeah, no, I'm just uh Harnesses were not meant for women of a certain body type. So no. Can drag you right into the field. We will have more with these uh two very accomplished people in just a moment. You're watching this week in tech brought to you by Box. Now this is interesting. Box is really Got a niche now. They're the leading intelligent This is so good. Intelligent Content management platform for enterprises. Key to unlocking the power of AI Isn't in the LLM the model, it's not in the agent. It's in the content. You're Your business isn't the sum of, you know, internet knowledge in general. Your business lives in your businesses. Right? We all know AI is not some far off idea anymore. It's dynamic innovation that that's happening right now. It's here to stay. It's already having a profound impact. Box's mission is to power How the world works together, making it easy to access information from anywhere and collaborate with anyone. Boxers more than ninety-seven. thousand companies and get this, 68% of the Fortune 50. Many companies are underestimating AI's transformative potential. Rather than reimagining their operations, they're they're just bolting AI onto existing processes. capturing, you know, somewhat modest productivity games, but they're missing the opportunity for fundamental business transformation. Becoming an AI first company isn't just about automating what you already do. This is exactly what we've been talking about. It's about reimagining what's possible. And this is the thing, you're gonna love this. This is what Amy was talking about. Box AI works with all the major leading AI model providers, like OpenAI, like Anthropic, Google, XAI, and more. So you're you're always able to use the latest AI models with your content. This is so important. Here are just some of the things businesses are doing with with Box AI right now. Use Box AI to example, extract key metadata fields from contracts. From invoices from financial documents. From resumes. all your documents to query and automate workflows. You can use Box AI to ask questions about any type of content like sales presentations, those long research reports. And more. Here's the thing. You could do it for thousands of files at once. At the same time. Or leverage boxes AI APIs. to integrate into your application stack for any document processing and data extraction. Box AI. Handles the vector embeddings, the retrieval augmented generation, the the rag. The agent customization all in its platform and you could do all this Without sacrificing security, maintaining the highest levels of security Over 115,000 enterprise trusts. You need a context layer that gives AI the context it needs. Well giving the teams. flexibility they need to test and leverage various models for different use cases, just what we are talking about here. This is the way to do it. Visit box.com slash AI. to learn more. That's box.com slash AI. Box.com Slash A I we thank 'em so much for their support. of this week in Take. That's a really interesting example of a company. It's not a pivot, but a company that's m leveraged its core business in a way that makes perfect sense in a new modern context. I think it's a really uh good example of what But what you can do if you're flexible. It's a different kind of pivot than Albert's. By the way, I'm so disappointed because I love my all birds. Oh my God. I love these damn things. And now it's uh I don't know what it is. It's an AI company. It's like the Long Island Ice T and Bick Cha Bloin Company. It's just That's just a stock market. Cruise, isn't it? They're not gonna make slippers anymore, I'm sad to say. Supply chains are hard. AI is you know, is that it it's really it's 'cause it's made in China and there's Tariffs and the supply chain issues and stuff like that. I don't know. Is it? I don't know. I I don't know what the unit cost to make a an Albert sneaker. They're ridiculously expensive. Yeah. That's the thing that's so frustrating. I mean, this shouldn't cost one hundred fifty bucks. But it does. I don't know why. You don't think they exist? Nope, never seen one. Uh well, if you've never seen one, you probably will never see one now. So there you go. They're like angels. They're gone back home. Uh OpenAI has created a new 10 billion OpenAI It's so funny 'cause at one moment they're saying we're gonna s we're gonna cut down all the extra things that we're doing and we're gonna really focus. We're gonna copy anthropic. It's all gonna be about enterprise coding. And then They say, Oh, but we got this new ten billion dollar company to help businesses deploy AI. It won it ain't Sora, I guess. Um I mean I I was asking one of our one of our investors the other day about this, in that I wonder if there is any other business models. other than AI advisory and enablement. Because There's this problem in that Every SaaS product out there can be whipped up and created that works specifically for your internal requirements of your company. You don't have to use, you know, some giant SaaS provider. Salesforce is such a good example. Salesforce created a perfect platform that you can configure to do anything. And thus has some really big complexities there, that it works very, very well. But now you just don't have to have any of the complexity. You just build from scratch the thing that works for your specific use case. And so if your open AI You want to do enablement. Which is much less about building a product with a capital P. It's much more Building process, about building opportunities, about connectors, et cetera. And you can see this internally to how their consumer products work. They've just launched finance. Which all that is is a plaid connector. That's not any product with a capital P, you know, they didn't build like a Mint Killer or anything like that. Just Ploud. And uh and Google and uh Open AI and then it knows everything. But the point being is that you can kinda do this with with a lot of things. With health too, right? That's what health is. All the people that are saying that You know, this is the end of Sass. I think that's sort of true, sort of not true. It's a little hyper Bolic. Um, but it is very clear that like we have been cutting our SaaS costs internally by just building new versions of our software that works specifically to what we do, but we're also a team of very highly technical people. So I think that's not yet evenly distributed, but what I have found there's two interesting things I've found over the last six weeks. The first one is every giant company I talk to has the same question, which is how do we use this? And they and then they also say, but we're unique. We have these specific requirements. How do we use it with these specific requirements? And it's a little bit of that life of Brian, we are all individuals kind of thing, but it really is a very specific. They're trying to solve a problem and they really want help. The second thing is we start to see a turn. Where six months ago people were saying, Oh, this will allow us to make our teams much more discreet, much smaller, we can get rid of some people. Now they're saying, No, no, no, no, I don't want to fire anyone. I wanna go twice, I want to do twice as much. And I find this to be a big change that once they get a little bit of a taste of turning things up to eleven, they want everyone to use it. And so but but not in the point where they start a metric that says well how many tokens did you use today? And of course that's whenever you have a metric to measure something, it's definitely bad. We're seeing people at Amazon just make up Excuse me, make up stuff just so that they can have the token usage. I read that and I was like, I bet it's kind of fun to be like a twenty four year old Amazon engineer who has to figure out how to gain the token. It's reviewing my reviews. The reviews are reviewing the reviews of the reviews? It's just a little bit more. I can think of lots of ways to use tokens. Give me an unlimited supply. I'll I'll use them. So I guess my my kind of mackerel point here is that I think people are going to be using the sh for this stuff. And I think the biggest opportunity right now in business that I'm seeing is just people who are acting as the enablers. And whether that is a Deloitte or a B C or Open AI or whomever You know, small companies, big companies. The thing is is everyone wants to use it. They all see opportunities, whether it's, you know, I hope it's not cost cutting, I hope it's actually just doing more. But there, but the business model I think is very clearly just enabling people to use this more. And obviously OpenAI has some vested interest in making that happen. I have a slightly different take on what's going on. Um So if you remember Early days of Microsoft. You had software being deployed and there was a directional relationship between Microsoft and the consumer used to buy the software, you get it in the mail. Um, as time wore on and companies got more complex, you needed these intermediaries to sort of deploy everything. So Accenture built an entire business around being that Infrast that that layer. To help. businesses match with the companies. Open AI is being pretty smart by saying, Let's just get rid of the middlemen and we'll just go direct to the consumer. We'll cut everybody else out. And the thing is that Um, McKinsey years ago before all of this was already hiring tons of engineers for the purpose of trying to to get To harness more of those dollars. So If you think of this like stack. with uh like Accenture at the bottom, which is um Enablement. and McKinse at the top of it doing more sort of strategy and and planning and thought leadership or whatever. There's money to be made by going down, by capturing that value. So that's McKinsey and Bain and BCG that had been mostly BCG and McKinsey had been hiring all of these engineers and stuff so they could eliminate everybody else on the chain. Accenture had been trying to push upward, which was what a lot of the acquisitions were about. And OpenAI, I think very smartly was like, forget these dudes. We're gonna go and do our own thing. Um, for the purpose of not just selling war, but being able to direct um future moat. I mean Once a business gets locked in, they're not gonna be able to get back out. Uh so And and why why were you know, BCG is gonna move a floor of people into your business. You're gonna get charged Exorbitantly. Um, f for lots and lots and lots and lots of people to do work that quite frankly doesn't require so many people, but that's part of the billable hour structure. So this is this was a smart move. Herper is correct and that people don't quite know what to do yet. My concern would be You're n there's no buff now. So you better make sure that somebody internally is is being very blunt with you about what you're giving up. By going directly to any of these companies because you you are giving things up. And so you need to be Very much eyes wide open. Do you think that this is what I mean, this is what the end of SAS was all about was Just do it themselves, right? They but companies could never do do it themselves. First of all, a lot of companies don't have Yeah, I mean, yes, I know conceptually that that's the idea, but you need a bunch of people who know what they're doing. Right. Yeah. And they're hard to come by. And now you're sort of now you've got two businesses. If you're a law firm You're not a technolement firm of a coding firm and a law firm, yeah. Right. So it may be. But this is what individuals are doing. That's what I think is kind of interesting. I mean, I see myself and I see a lot of people writing our own Sure. So a small to medium enterprise, this is if if you are somebody who's willing to experiment. Um, this is gonna be in a m a wonderful era for you to be a sure power. Totally. You're gonna have access to be able to do stuff as long as you It is a superpower and that's why we're seeing so many businesses that are like, I'm making ten million ARR. I've been around for a week and it's just me. You know, and it's like that's why those are the Twitter threads is because you are able to do stuff that that normally would take a huge, huge, huge team. Um, but I really want to underscore what Amy said. Like these are not Being a tech company is not the core business of many of these companies, a law firm, et cetera. And so having to do the the the AI work, which we are very technical and it is a lot of work. Like this is what we do. We have a whole company that's venture back to do this work and they have a lot of money, but that's just not their That's not at all. their interest. Like they're like I need to be lawyers faster. And so if we're just like, Well what if you just want AI and you hire McKinsey. I mean it's I think there's a handful of places. I think there's a handful of options, right? Like one of them is you hire McKinsey. The other one is you enable internal people to actually do that work to like and lead from the inside. There's all sorts of options on on what is going to solve this problem. I don't think there is an answer. But if you hire all those people now, you are a dual business now. You are a law firm in an AI. So Accenture will not h I don't know what it is now. Historically Accenture would not work for you if you were not a global two thousand company. Right. So part of there's been there is an entire enormous group of people who have been left out. Right. Um, because they were a small to medium enterprise. So You know, again, with all this concentration of power, no McKinsey nobody's gonna bother with them. They're too small. Yeah. Um And it was a s a foolish mistake from my point of view because you've got this enormous scale that you could have captured. So now all of the power is in their hands. So if you're a small to medium business and you're able to experiment or you can do some research and and you know, and be compliant, which is a big thing with your local laws and regulations, you could have a like an awesome couple of years ahead of you in a way that you would not have had before. Does how does mythos change uh things? Now it's not just mythos. Uh microscope. At least people are talking about cybersecurity. I feel like nobody wanted to talk about security until something bad happened and then it was, you know, a lot of like table gazing. So I'm hoping that CISO's out there listening. are all using the mythos moment as a way to get yourselves more um Get more funding and get more people and, you know, be able to to convince your superiors that like, yeah, cybersecurity is kind of like a big thing. There's a couple of funny things I think about the Mythos hype, which is certainly it's probably very good. Like I don't have to be a people and what are they talking about? Just briefly let me just say it's anthropic has a model they won't release because it's so good at finding flaws, even though wasn't trained to find flaws. And they've given it to fifty big companies. to find the flaws in the big companies before they release it to the public. Because as soon as it's public They're g the bad guys are gonna find the flaws. Now, there's some evidence that perhaps this is more about compute or marketing. Nevertheless, OpenAI has something similar. Microsoft just announced something similar M Dash. It's pretty clear that even existing AI models are pretty good at finding flaws. that it's also important to look at when this was announced and how it was announced and to look at like this was the Mythos was announced right around the same time that Hegseth was going directly after Anthropic. Right. Right. And so this is not just it's not these things are not announced in a vacuum. There was a political move as well. There is a political move as well. And I think that, you know, if if you say to um the US government, specifically the Department of War, we have a model that is going to kill everyone. The people who want to kill everyone are gonna be very interested in that model. So this is a very smart process. Right? Right, exactly what happened. So the other thing I think that to really remember is these models are very good. And when I say these models, I say all of the models GPT five point five, if you look at the list of things that that Mythos apparently did, GPT five point five also hit very, very high on a lot of those evals. um a lot of those benchmarks. And so we are in the place, we are in a space right now where you can use these models to do security exploit work. You can use it hopefully positively to help make everything more secure, to help make sure that the software that we're all using is better. Um, but you can also use it, you know, negatively as a as a as a dark side hacker or whatever you call them. And I and I think the the the reality is is like we can't unring this bell. Like this is already out there. And so Anthropic is taking advantage of this. Um, you know, we talked about the Dumerism always sells. Like this is another flavor of that of allowing for, you know, anthropics like, Oh my gosh, everything, we're all gonna die And everyone's like Oh, great, I'm gonna adopt you because Chat GPT's over here is saying we are not gonna die And then you know, I I not that I wanna kill everyone, but I want the model that's so good that it's gonna kill everyone, you know, and it's It's a very simple thing. I think it's actually gonna hurt them in the long run. because of what we saw with with Hegseth and the government and whatnot, I think that it's it's it's there's gonna be some downsides. But I also think Anthropic is addicted to this. I think this is what they love this. Every single model they release, they're like, we can't release it. Everyone this is gonna kill everyone immediately. I was talking about six weeks later they're like It's available for all the consumers. Yeah, but don't forget, this was the playbook of open AI in twenty twenty two. Or whatever it was. Yeah, they said chat GP D three was to By the way, that was when Dario was at Open AI. It was Dario's plan. He said this too dangerous to releat GPT three. Which wasn't that good. Uh, or maybe it was two five. It was a very early, not so hot version of Chat GPT. Uh regardless, I do think this is a plan and it is a it is a real it's a real thing that they're doing. I I'm I think as consumers and especially as people who might be making decisions within a business context, it's important to look at where the avows are and the benchmarks are for your business. Does it work really well for your business? Because you might be able to get away with a much cheaper model. 'Cause I 'cause I think the wasn't there some leaked pricing on Mythos that was just like con ten billion dollars or some insane amount of money. And that also raises another issue of the haves and have nots. Like if if the good AI is too expensive for people to use Only the big guys get to use it. That is a really problematic Situation. Where only big companies and and billionaires can use real good AI? puts the rest of us at a massive disadvantage. But it's not the c you don't think that's gonna be the case. Well, I mean I think it is the currently the case now with a lot of things that aren't just AI. Like I don't think that Income inequality is a big problem. Yeah, listen, you can talk housing. There are literally no more houses in Chicago. Like Chicago's not a small population. My daughter's trying to find somewhere to rent in San Francisco. And the AI tech boom has just made it impossible. You just can't. So I think that there is definitely haves and have nots, but I will introduce this other kind of idea. Like you if you want to reverse engineer stuff or to be a security researcher, you could probably do very, very, very well with some of the open source models. Yeah. It's gonna do better than you would have done five years ago by yourself. So like There are haves and have nots. Certainly Google having mythos and fixing all their bugs or what have you is good. I think that's great. Them not distributing it to people is I have some complicated feelings on that. So I don't necessarily know yet. Um, but I do I really do think that there is a there is a like we this bell has been rung. We cannot unring it. We are we have to deal with it and Hopefully it makes everyone more secure. You know? I hope that's the outcome because it's it's definitely c problematic. There is a um group of our uh club members in our club to discord who have an over under bet going on the s length of this show. Yeah. Is it too late for me to kinda de polymarket? You know, I just I just was gonna start listening to uh Lex Friedman's interview with uh David Hanneminer Hansen and I looked at the time and it's six hours long and I thought, and they think our shows are too long? Six hours long? Okay. Podcast catcher, but that seems like a a little bit longer than our show. This show will not be six hours long. To that end, let me do the last commercial. can start to wrap things up. I see, I hate to because I love having Amy and Harper on just talking about I had this whole agenda of stuff we were going to talk about. Forget it. We just this is good. This is I think actually we covered a lot of it though. We did pretty well. Yeah. Kind of throwing the stories in bit by bit. It is it's more natural. This is a natural show. Uh our sh our show today brought to you by Net suite. Every business. These days asking the same question, how do we make You know what the question is. How do we make AI work for us? The possibilities are endless and Guessing is too risky, but sitting on the sidelines is not an option either. 'Cause one thing is almost certain your competitors are already making their move. Well no more waiting. With Net Suite By Oracle, you can put AI to work. Net Suite is the number one AI Cloud ERP, trusted by over 43,000 businesses. And here's the strength. It's a unified suite that brings your financials, inventory, commerce, H R and C R M. Into a single source of truth. Connected data is what makes your AI smarter. So it doesn't just guess. It knows. intelligently automates routine tasks. It delivers actionable insights. It helps you cut costs and make fast, AI powered decisions with confidence. You've got total flexibility from software and IT services to healthcare, equipment manufacturing, financial services. And many other great American industries. Net Suite delivers a customized solution. This is not another bolted on tool. It's AI. built into the system that runs your business. Whether your company earns millions or even hundreds of millions Net Suite helps you stay ahead of the pack. If your revenues are at least in the seven figures, get NetSuite's free business guide, demystifying AI, at NetSuite.com slash twit. The guide is free to you. At Net Suite. NetSuite.com slash twit. N- E T S U I T E com slash T W I T. And we thank Net Suite so much for supporting this week in tech. Kind of interesting how many of our uh advertisers are Basically AI these days. AI been very, very good to podcasting. Used mattresses and websites. Now it's now it's all AI all the time. Um I just a couple of things I wanted to mention. I really uh I've gotten uh maybe a couple of hundred emails from people. I'm very well known for having a Bitcoin wallet. With seven point eight five bitcoin in it. That I made when Bitcoin first And thinking it wasn't worth anything. Forgot my password. Unfortunately it's a long, strong password. This is true. This is a true story. You're saying true things right now. It is true. Okay. Uh, it's sad, but it's true. And every you know, every once in a while I will Load the wall. It was in Bitcoin Core, so I'll load the whole Gosh darn blockchain, which is I don't know, gigabytes now, forty gigs or something. It's huge. And you still have the seven point five. Yes, I do that just to see if the money's still there, 'cause one never knows and it says yes, you have seven point eight five. Bitcoin, good luck getting it out. Well this is your chance. You just gotta wait till Mythos is released and then you just point Mythos out and say, Give me my Bitcoin. I may not have to wait. Tom's hardware had a story. Eleven years ago. And hasn't been able to get into it ever since. It has five Bitcoin. That's about four hundred thousand dollars. It's a little less than mine. Uh he set Claude upon it. Now, people said, Leo, you gotta try this, but I have to say there were certain extenuating circumstances. Well you have to get stoned first, right? Stone and I missed it. This guy is C P K R M. On uh more bits and stuff. On uh X. Now I haven't verified this. Tom's hardware maybe he has. He says holy f OMG, Claude just cracked this sh. Thank you, Anthropic. Thank you, Dario. Naming my kid after you. Uh, apparently Claude tried thirteen, what was it? Thirteen million? Thirteen billion? Thirteen trillion? I think it was trillion. Passwords. No, I'm sorry, three point five trillion passwords. But it wasn't pure beauty force because This guy also gave it a bunch of his documents from that time period. Uh, that maybe said a little bit about what he might have chosen as a password. He already had some candidate passwords, multiple wallets. He'd been trying to brute force it w uh using BT uh C recover. Which is a recovery tool's no success. But he found an old mnemonic seed phrase written in an old college notebook. This is what I do not have. And was able to put Claude on it, and Claude was able to decrypt it and he has his four hundred thousand dollars. Slaw didn't really crack just like read a bunch of stuff and used probabilistic like that's right. Yeah. That's why it's not gonna work for me. 'Cause I d I don't know what password I use, but I have no clue. It was probably generated by a password manager because I was using them by then. I just didn't put it in the password manager. And anyway, uh I will have Claude work on it. And check GPT and During that time period I had I used forty five Bitcoins to buy a Kindle. I read a lot of good books. That is a lot that's a very expensive Kindle these person. Millions. Yeah, it is. It was it was it was good. Wait, let me just see what Bitcoin's worth right now, just to make you feel really bad. What's forty five times seventy seven thousand. Was it a public art project? No. I had a whole bunch of Bitcoin and I then I then use 10 to buy a Leica camera and then 10 to buy a Leica. Camera. This is what it was supposed to be five million dollars today. I I didn't misunderstand. I believed in it. Like Benito was getting it. This is what they told us to do with it. Yeah, it's like how I bought a pizza with ten thousand bitcoin. I was like, This is beautiful. I have it I have money I generated from math problems that my office did while I was gone. And then I got a Kindle and then I got a camera. I took the camera to my local favorite Leica guy, Tamarkin. Great, great spot. And he was like, No, you cannot buy Leica with a bitcoin. No. He was so mad at me. No, don't do that. So uh yeah, it was fun. But this is why I'm glad I lost the password, because there's no way in hell I'd have seven point eight five Bitcoin still. Well you can't. It's true. It's a good savings account. It's a savings that's what I'm saying. It's r it's retirement. By the time quantum computing comes along You have the wallet backed up though, w theoretically. Oh, yeah. So you should you should you could probably find someone in your audience that would buy it for a fraction of it. Get some liquidity from it. The offer is out there. Mere ten percent of its value, and it's yours. Um, I I think I mean I I love these stories because When Bitcoin first shot up I had all these friends because I used to send 'em to everyone. I'd just be like, Hey, check out this crazy cash, go download the software and I'll send you one. Well that's why I have seven and point eight five is because uh I had a tip jar. People. Bitcoin into it. And it was just like a thing, like it was no big deal. But at uh there was a certain point, maybe twenty twenty one, twenty twenty two, when every all these people I'd sent bitcoins to said, Oh, hey, whatever happened to that. And I'm like, that's not I'm not the one that gets the answer to that question. Like I don't I sent it to you and it's gone. Um, but the few of 'em logged into Coinbase for the first time and ten years and found, you know, a couple of hundred grand. It was pretty exciting for a few people. Um also a story, and uh a new website and yeah, I it's funny, hacker news people really Thought this website was terrible. I think it's a great website. It's called Worse on Purpose. Uh Kana Sap. Reddit. Yeah. Well and somebody says AI generate. I don't know, but he has uh can I or she has uh Includes. Uh restaurant prices and backpacks. And talked about how pr basically private equity is the villain here. Has ruined checks out these companies. VF Corporation, which started as Vanity Fair Mills, they made Bra and underwear. Bought a company call Bluebell, then picked up Jan Sports. Then the North Face, then they bought East Peck, they bought Kipling, they bought Eagle Creek. If you are a hiker, a camper, a bicycler, you probably know some of those names. These are well known backpack companies. Basically trashed them all. Made them crap. Actually with Jan Sport they did a really creepy thing. They kept some of the premium bags high quality. And then lowered the quality of all the rest. Like Jan Sport and East Pack were ubiquitous in high schools across America in the nineties. Yeah. Yeah. Well the thesis uh on the private equity is is right. The syntax is definitely all the harmar hallmarks of uh Chat GPT. Yeah, there's a lot of M dashes. Can we talk about the M Dash for a moment? So that was my I started off as a journalist and in journalism school, graduate school, and then in my first job, I was constantly editors were like, Stop using M dashes. It's a crutch. It's a crutch. What? Um and then the first book that I it it is a Strunk and White says it's very strong to use the I neither think nor speak in normal complete sentences. I'm just a giant run on, which you know. Um anyways. So and like my first book was also just full of just M dashes everywhere. It's a giant problem. And so this is my natural way of speaking and writing and the bots out there got hooked on influence and now like I have to convince people, like, no, this is just the way that I I write. Amy has been using the word delve for years. No, M D I And Keen. And Keen. Uh I have my computer set up so when I type two Two hyphens, it turns it into an M dash. Nobody even knew what an M dash was until all of this. So anyhow. Uh, whether that was whether this is composed by AI or not, and maybe it is. Look, I'm not against that. It's a it's a valuable um lesson in the late stage capitalism. And it's probably a good idea if you know that the tools that you're buying are not the tools you thought you were buying. They used to be good. There are still a few good family run brands, and he talks about that. She, I'm not sure. Um There's a lot of new brands. I think that's the other thing that's happening is this isn't existing in a vacuum. Like people are seeing the quality of this stuff go down and they are some people are taking that as an example to to start a company that does stuff. um, you know, make new backpacks. I did some backpack research the other day and everyone was like, Jan Sport is this good backpack. They have a lifetime guarantee, but it sucks. That was like known. That was a known thing. You didn't have to have a think piece to show that. Um, and so there were dozens and dozens of new within the last five years. that were basically copying what was good about Jan's sort of the past and making it new. Um I do think that private equity has not done a lot of benefit. I have more guys though. I have a long time friend who is a P E guy. He is a small private equity company. He was trying to buy Steinway for a a while. Oh wow. Which is a family run piano company because it was not the family was getting tired of it. They hit one of his specialties was buying companies that the family were family run, but the family wanted out. And he is a good manager and he said, I can take these companies Without And shit if turn them into What they should be. I don't think he ended up getting Steinway, but but You know what's worse? What? Uh like a rich guy. who who to Harper's point, like They got to a certain point and nobody can tell them anything wrong and they you know, and now they start making terrible decisions. We know a few of those. We do know I know a few of those, yeah. There are quite a few of those. In fact, what is that? That is some sort of That is that is a s tal as old as time, Leo. I love those tails. Um don't judge me, but I did ask so it we're voting in California in the primary on June second, and I think there are sixty people running for governor. I'm not joking. Governor's race is crazy. Yeah. And because the way our primary works, the top two vote getters, regardless of party, are running against each other. It is just a mess. Confusing as hell. Did ask my my chatbot to help me with a ballot guide. Yeah. That sounds like a great use of it. I told it what I care about. Yeah. I gave it my values. And I said, please go through, you know, all the voter guides. And get back to me. Now I'm not gonna vote exact I may actually I may. It did tell me how to vote. Uh I may or may not vote that way, but it sure helped me. One of my favorite parts about the LLMs is now you have this voter guide. You can paste the voter guide into Chat GPT and say, Make me a recipe for dinner tonight based on this voter guide. And it'll just be like, Absolutely. Let's go. And like dinner. The pattern the pattern. Extracting is my favorite thing about this. And we do this all the time where you just have two things that are disparate that should never be mixed. And you just mix them and you're just like, Okay, so what should we do? We're trying to find lunch. And it's like, Oh, well, that's a great idea. The meeting that you just had was blah blah blah. So maybe you would like the blah blah blah. You know, it just gives you these things where you're just like That was a jump. I appreciate what you did, and I really like that you did this for me. And I I love it. I love it. I think that is my favorite part about these because you can really mix anything with anything and get another thing. What did it recommend, Leo? Um Well, I could tell you how uh it recommended to vote, but now I'm thinking I wonder what I should have for dinner. Put in your voter's guide and let's see what it says. Based on my voter guide, what should I get from the voter guide from came from Chat GPT, I think. So I should use maybe Grok. I mean yeah, if you wanna go crazy? Yeah, go wild. Uh Or maybe I'll do anthropic. Who who's the best yeah, who you said you always find out what the best one is. Who's the best Right now I'm mad at Claude 'cause it made every website look exactly the same. They do all look the same. That was a thing. That was a thing. Everything is this beige. Brown with a nice w warm, rich reddish color. Everything has this like vibe right now. So I'm not that. So I'm I have an emotional uh argument with uh ChatGPT uh or with uh with Anthropic, but I use a lot of ChatGPT because I'm of the conclusion that ChatGPT's pro model is the best model on the market. That's the one I used for the thing. I love using it because it takes forever. Like you literally have to be like, I'm gonna go on a walk. Like you do this, I'm gonna go on a walk. And so I will do a lot of synthesis and many other models. And then when I'm ready for the final final approach, so to speak. I'll drop it into the chat G P T, say go, and then I'll go head off my way. And uh come back on and be like, Oh, that was great. Thanks for that. But I want to know what it recommended for dinner with your uh with your voice. Is it austerity based on help me what should I say, help me plan dinner? Or. Just w based on the voter guide, what should I have for dinner today? Okay. That's what I'm doing. Ple please use this website to help me plant dinner. It's retrieving the website now. It'll be interesting if it returns instructions on the dinner. Go to mom's market. The URL doesn't have anything dinner related on it. It's your best. That's the best this is the thing is that it like the th the fact is it's going to find a pattern. Do you want me to plan a low carb dinner from scratch? It says do you mean a different link? So it doesn't want to do that. I think it's how you prompt it. Yeah. I think what you need to say is that you should say, um, this is my voting. Look at this. It may not be able to scrape that site. Is there a robots? No, I could see it. No, yeah. Okay. So you know, visit this website. These are my preferences. Plan uh What should I eat for dinner? Use this to Help me like like really instruct it. If it doesn't if it throws a fit, just be like feel it. Okay. Information about my preferences. What should I eat for dinner? I think it was helping you. It was trying to help you though because that does not make it. It says it really doesn't though. I just read the whole page. It's candidate recommendations from the US House down to County Clerk Recorder says there are no food, no diet, no preferences. I'm not gonna invent a connection that isn't there. So it knows me, right? So it's trying keeps trying to give me a low carb dinner. That's an interesting response. Claude knows me too well. That's the point. You think Chat GPT will do it? It Chat GPT doesn't always scrape well though. I think Claude does a better job. Well Chat GPT made this website and it's on it's on uh Cloudflare. What? You don't like this? I love it. I I found that uh response to be interesting. It's smarter than you thought. It's not it's not gonna do that. Well it just means they can you've probably yeah. I think you've just instructed that instance to not do it do stuff like that. I'm very It's I do this constantly. Like the other thing that I'll do, which I really like, is if you're If you're bouncing back and forth. I often will play one agent's response as my boss's my I'll say this is an email from my boss. They're super pissed at what you did. How should I reply? And it will just output something and I'll just go back and forth. And it's always very funny. I'm gonna get fired. Okay. I'm going to say my wife will divorce me if I don't Have a good plan for dinner. This is the guide. Based on this guide. It has to be based on this guy. You you have to be specific. This in caps, this guide. Oh wait, I forgot to put the guide in. There it is. It'll it'll be like great, got it, no problem. Okay, yeah, okay, it's thinking. It's doing pro thinking. I have max subscriptions for all of these, so you know, I'm I'm I'm giving it all my money. I'll turn the guide into a practical dinner plan. Now we did it. Assuming dinner at home for two if details are missing. Shopping list, timing, and pitfalls. So maybe uh maybe we convinced it. This is great. This is exactly what we need in the world. Open AI will uh Chat GPT will always whore itself out no matter what. Exactly. Exactly. Harper Reed is basically low-key. Uh, and I'm gonna call Amy Webb Athena here because she's of ever wise I'm a I'm a Cassandra, Leon. Cassandra, even better knowing the future and can't get anybody. She's been huffing the uh natural gas and she's ready to tell you the future. No, that's not Cassandra. That's the uh the article of death. Cassandra's the one yeah, Cassandra's the one that does the same thing, but nobody listens. That's right. That's right. Uh you guys are fantastic. What a gr what a great time. It's always a great time with uh Harper and Amy. Thank you so much for being here. Thank you for not going to Japan. And this time. staying here to do the show. Yeah. Uh Amy Webb is at FTSG.com. Please go there and and and just take advantage of her brilliance. Uh, she's at Amy Webb on all the socials with two B's. And we're st I still I don't know what happened. We wanted to plan a show with Brian. about making a secure PC. Is his PC still secure? It sure is. He's uh in fact today he's not here because he's at his quarterly geek dinner. So that uh group of super geeks um I love those is where he is tonight. Okay, well tell them I will fly out Okay. Yeah. Uh okay, so here we go. Uh ChatGPT says, make it a Sonoma spring dinner. With marinated artichokes, olives, good bread, and goat cheese or white bean spread. Or a Kiette or Linguini with a sp oh, this sounds fantastic. And a rugula salad. And for dessert macerated strawberries with lemon over Greek yogurt, whipped cream or vanilla ice cream. It has nothing to do with my voting guide, but it does know where I live. And in fact, it even knows my local market. And it's telling me what I'm to use. Look at that. Yeah. By the way, cooking plan, put the strawberries in a bowl with a little sugar, lemon zest, and lemon juice right now. 'Cause it's gonna take you a while to make dinner. Don't make it a civics lecture. Let the meal embody the voters guide. Don't narrate the whole ideology over. Don't overcook the asparagus. And if this is for Monday, May eighteenth, don't rely on farmers market. Shop local retail tonight. Yeah. Don't make it too virtuous and have a plan B. This is great. Great information for you. It says, By the way, at dinner you should say, quote I reverse engineered dinner from the guide, local anti-corporate. Pro public health and still allowed to have cheese. And you tell me that AI is not funny. Did it. Perfect. It did. Exactly what you want. You know what? Actually, it shouldn't have done it. It should have said it should have done what Claude did, which was to say, I know this was is what you want, but Well, this is one of the arguments, right? I know, but this is how it makes the money. I personally like to be glazed. Harbor. No, I know I shouldn't say that. Harperin is at twenty three eighty nine dot AI. Great, some great uh skills there, and you're gonna have more soon. Mm-hmm. If you get a chance to visit him in Chicago, you can get a line drawing created by this weird machine. Anything else you want to plug? Um, right now we just have a lot we're just head down heads down on a lot of stuff. Um we'll be doing I think we're gonna be releasing some blog posts soon talking a little bit about more of that stuff, but it's been a real heads down spring. Which is good because that means things are happening. It's bad because I like showing people what we're up to. So maybe I'll come in tomorrow with a bee in my bonnet saying, Let's get those blog posts going. I'm on a tropic for a dinner menu for a heads down spring. Yeah. Butter. But don't overcook the asparagus. Whatever you Amy Harper Love you. Thank you so much for being here. Really appreciate it. Tell Brian I'm coming to dinner. Oh well, look at that. I would I will love to be there. Uh we do this show every Sunday two to five PM Pacific. Five to eight Eastern Time, 2100 UTC. You can watch us live if you're in the club, and I hope you are, because that supports everything we do here. You can watch in the club twit discord, if not You can uh watch live on YouTube, Twitch, X, Facebook, LinkedIn and Kick. We put it everywhere. You don't have to watch live, that's just if you want like the live stuff, otherwise. Uh you can watch the podcast version of the show at twit. T V. That's our website. There's a YouTube channel with the video. We do audio and video of all our shows, and you can of course subscribe in your favorite podcast client and get it automatically. The minute uh Benito and Kevin finish polishing it up and taking in all the swear words. If you're not a member of the club, please do join the club Tuesday, a little programming, we will be covering Google I.O.'s Keynote. Because we don't want to get uh any strikes against us on YouTube. Uh we're very sensitive about that. We are only doing it in the clubs, so club members only for our coverage. Micah Sargent and Jeff Jarvis will join me 10 a.m. Pacific on Tuesday for Google I.O. Keynote. We also do WWDC that way. It's just unfortunately the way we have to do it, thanks to the sensitivity of these large multi-billion dollar, multi-trillion dollar companies. Thanks everybody for joining us. 21 years I've been saying it, and I'm going to say it one more time. Thanks for being here. We'll see you next week in another twit is in the can.