In short
US tech leaders’ China trip and the Taiwan/Taiwan-chip risk; how China’s AI and robotics ecosystem is advancing via open/distributed access and aggressive infrastructure; whether the US should avoid AI regulation and instead encourage competition; and practical examples of low-cost “AI hardware/agents” and smart-device security. The episode also touches on broader innovation strategy (US “lawyer state” vs China “engineer state”), and how companies should keep investing during uncertainty.
Guests (backgrounds)
- Amy Webb: futurist and founder/leader at Future Today Strategy Group (FTSG), long-running trend research; focuses on forecasting tech “convergences” (AI with biology/robotics, etc.).
- Harper Reid: technologist/entrepreneur/hacker at 2389.ai; works with AI systems and local/open-model experimentation.
Key claims
- The China CEO summit was highly choreographed; Jensen Huang of Nvidia was reportedly excluded because of Taiwan ties, and the visit’s intent (from China’s view) was to secure Taiwan-related concessions.
- China’s AI progress is helped by distributed, lower-cost access to capable models (often open/quantized) and large infrastructure investment (hyperscalers vs “no national strategy”).
- US AI discourse is framed as apocalypse/utopia, while China is framed more pragmatically around competitiveness and experimentation.
- The US should not “block” innovation with heavy AI regulation; instead, enable more companies to compete (“let a thousand flowers bloom”).
- US companies are overly optimized/regulated and pause investment; China keeps pushing forward.
Notable examples
- Nvidia/China chip sales and the argument that blocking chips forces China to build its own CUDA-like ecosystem.
- Local/cheap DeepSeek-powered AI devices/agents (e.g., Malleus space AI robot; DeepSeek “desk robot”/agents) described as low-cost and easy to deploy.
- Waymo “DDOS”/mass-ordering incidents in Atlanta and San Francisco neighborhoods.
- Toto toilet tech repurposed as an analogy for long-term R&D leading to data-center chip manufacturing value.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOLeadership US Japan Program Overview
0:45 to 3:18
Amy Webb explains the Leadership US Japan Program and its significance.
“Either way, you're going to be thrilled that we have Amy Webb here from the Future Today Strategy Group.”
Diplomatic Dynamics and CEO Visits
3:18 to 7:44
Discussion on the recent visit of US CEOs to China and its implications.
“You do have to demonstrate leadership and have some kind of interest in Japan if you're not Japanese.”
China's Technological Advancements
7:44 to 11:28
Exploration of China's advancements in AI, EVs, and robotics compared to the US.
“They literally have 70 % of the global market share for EVs and also have like literally 100 different EV companies.”
Global Perspectives on AI
11:28 to 14:03
Contrast between US and Chinese public perceptions of AI and its implications.
“They're not relying on closed US models.”
Debate on NVIDIA's Role in China
14:03 to 15:28
Explore the implications of NVIDIA's technology supply to China and its global effects.
“them to create their own technology and not use CUDA, his proprietary technology, which is bad for the world and bad for America.”
AI Robot for Kids: Malleus
15:28 to 18:34
Discussion about a new AI robot for kids and its implications for technology use.
“Although I don't use Google, but I'll do this.”
AI Innovation and Competition
18:34 to 23:14
Examining the competitive landscape of AI technology between the US and China.
“I would not, I know you're not going to use it for real work, you know, but it, but it's like, it's a$20 AI assistant that is pretty funny to use.”
US vs. China: Economic Systems
23:14 to 27:19
A discussion on the differences between the economic systems of the US and China.
“and it might not be the right tool for this moment in time.”
Future Outlook on China’s Century
27:19 to 28:00
Insights into the potential for the 21st century to be dominated by China.
China's Economic Landscape and Global Implications
28:00 to 39:37
Explore China's economic challenges and their potential impact on the future.
“But the point is, there's so much pendulum swinging in a lot of these countries because of extreme views going back and forth.”
Show all 69 chapters
China's Economic Landscape and Global Implications
39:39 to 39:52
Explore China's economic challenges and their potential impact on the future.
“Learn more about how ThreatLocker can help mitigate unknown threats and ensure compliance.”
Tech Innovations and Hardware Autonomy
39:52 to 42:01
Discuss the importance of hardware autonomy and modern tech innovations.
“You know, I had, what was it, the OBS bot?”
Exploring Wake Word Technology
42:01 to 43:32
Discussing the intricacies of training wake word technology using various samples.
“Well, that's what I'm using now is open wake word to train it to say, hi, Kenobi.”
Waymo Adventures and Hacks
43:33 to 45:08
Sharing experiences and hacks related to using Waymo for transport.
“It was so funny, because I was just like, what a...”
Waymo DDoS Incident
45:09 to 47:15
Discussing a humorous DDoS attack involving Waymo cars in Atlanta.
Revisiting the Trend Report
47:16 to 48:22
Amy discusses the evolution and discontinuation of their annual trend report.
“decided to replace things yeah i'm sorry by the way amy at south by southwest had a funeral I did.”
Capitalism and Creative Destruction
48:23 to 50:32
Exploring the concept of creative destruction in capitalism and its implications.
“At any rate, the idea is capitalism is this perfect storm, this perpetual storm that's constantly happening, and it sort of gobbles up old existing technologies and creates new ones.”
Political Dynamics and Populism
50:33 to 52:21
Analyzing the challenges faced by populist leaders in contemporary politics.
“You have to spot the storm before it happens.”
Debating Economic Systems
52:22 to 55:56
Engaging in a discussion about the effectiveness of different economic systems.
“There's also a little bit less freedom in some cases.”
The Collision of Entrepreneurial Cultures
56:00 to 57:24
Explore the contrasting mindsets of American and Chinese entrepreneurs.
“And we need to be, we're in a world that is not soft.”
The Job Market for Graduates
57:24 to 59:08
Discuss the challenges facing the class of 2026 in the job market.
“Here's the article from Semaphore, Rachel Jones writing, the class of 2026 is cooked.”
Economic Fears and AI Doomerism
59:08 to 1:00:46
Consider the implications of AI on employment and economic stability.
“Now they can get a job, it's just not going to pay the same.”
The Transition Generation
1:00:46 to 1:03:51
Examine the skills gap and transition challenges for today's workforce.
“And there are a lot, there's a lot of money to be made from doomerism.”
Innovations in AI Artistry
1:03:51 to 1:05:41
Discover new developments in AI and art creation.
“We are transitioning from where we were before to the time after.”
Real-World Applications of AI and Human Experience
1:05:41 to 1:10:02
Discuss the integration of AI into everyday life and its implications.
“Well, it's interestingly, it just uses an old Monoprice 3D printer that we just took with a pencil on it.”
Running and Real-World AI Dilemmas
1:10:02 to 1:11:23
Explore the frustrations of running with distractions and a real-world AI dilemma.
“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.”
Running and Real-World AI Dilemmas
1:11:30 to 1:13:14
Explore the frustrations of running with distractions and a real-world AI dilemma.
“Scribe will suggest improvements to the underlying workflow, not just documents.”
AI and Video Models in China
1:13:14 to 1:19:46
Discuss the advancements of Chinese AI models in video creation.
“That's S-C-R-I-B-E dot H-O-W slash twit.”
Cultural Nuances in AI Creations
1:19:46 to 1:23:54
Examine cultural nuances integrated into AI-generated content.
“And so I would read about Amelia Earhart or pyramids or whatever.”
Future Predictions and AI Ethics
1:23:54 to 1:24:05
Debate the future of AI and ethical considerations in technology's progression.
“One, you amass more credibility and power and riches over time as anybody does.”
The Future of Genetic Engineering
1:24:05 to 1:25:12
Discussion on the implications of genetic manipulation and societal hierarchies.
Robotics and Automation Misconceptions
1:25:12 to 1:26:28
Exploration of how robots are perceived and their actual capabilities.
“Does he also have a crater on the moon named after him?”
Optimism vs. Dystopia in Technology
1:26:28 to 1:27:32
Debate on the potential futures shaped by technological advancements.
Nostalgia for Classic Robots
1:27:32 to 1:30:06
A nostalgic look back at childhood robots and their significance.
“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.”
Childhood Memories and Gifts
1:30:06 to 1:31:33
Sharing personal stories about memorable gifts and family traditions.
“I have been telling everybody at school for a week, I got the Omnibot because what else could be in this, this heavy box.”
Childhood Memories and Gifts
1:31:35 to 1:33:15
Sharing personal stories about memorable gifts and family traditions.
“But if anybody is listening out there and has access to a battery that might work, please contact me.”
Childhood Memories and Gifts
1:33:18 to 1:35:33
Sharing personal stories about memorable gifts and family traditions.
“I have a little soft spot in my heart for Shopify.”
Tech Entrepreneurs and Fashion Choices
1:35:44 to 1:38:00
Discussion about the intersection of technology, entrepreneurship, and personal style.
“Yeah, he's, he actually, poor Hank made a video on Instagram.”
Trust Relationships in the Elite
1:38:00 to 1:38:52
Explore the dynamics of trust among influential individuals.
“That's a big career for a lot of people.”
Zuckerberg's Sandwich Delivery
1:38:53 to 1:40:13
Discussion about Mark Zuckerberg's unusual sandwich delivery experience.
“He's basically in a very pedestrian black tie.”
Broadway Success of 'Chess'
1:40:14 to 1:42:24
An overview of the Broadway revival of 'Chess' and its unexpected success.
“We took a longer term lease on a floor of a building.”
The Future of AI in Theatre
1:42:25 to 1:44:43
Insights into combining AI themes in theatre productions and upcoming projects.
“And if you are not a Broadway person, this is a good show for you because this is not like a Broadway show, Broadway show.”
AI and Trust Issues
1:44:44 to 1:48:28
Discussion on how to build trust and affection for AI technologies.
“We have one that's called Review Squad, which is really fun because you can enable a well actually squad of agents to review your code.”
Challenges with AI Models
1:48:29 to 1:51:24
An exploration of the unpredictability and challenges of AI model performance.
Evaluating AI Models
1:51:25 to 1:52:00
Strategies for evaluating and choosing the best AI models for tasks.
“so I don't like showing up to work and like one of them, one of them is named Kevin, right?”
Navigating AI Model Diversity
1:52:00 to 1:54:19
Exploration of the benefits and challenges of using diverse AI models for various tasks.
“I don't get like, nobody's like giving me a phone call and letting me know that there's going to be a new model release or an upgrade, you know, Tuesday.”
AI Development and Team Dynamics
1:54:20 to 1:56:40
Discussion on the expectations and realities of AI deployment within teams.
“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.”
Understanding the Bubble Debate
1:56:41 to 1:58:29
Analyzing the current AI investment climate in comparison to historical bubbles.
“But there's a legal and compliance issue that makes that super, super, super hard.”
Infrastructure and Future of AI
1:58:30 to 2:00:38
Discussion on the sustainability of AI infrastructure and its implications for the future.
“We're not in the coming up with products that will shift things phase.”
Community Impact of Data Centers
2:00:39 to 2:01:58
Concerns over the social and economic impact of data centers in local communities.
“I grew up south of the steel mills in a very, very blue-collar community.”
Planning for Data Center Development
2:01:59 to 2:04:26
Importance of strategic planning for data center projects to benefit local communities.
“Hill County, Texas, where there are already plans to build eight different data centers, has just passed a one year data center ban.”
Innovations in Space-Based Data Centers
2:04:27 to 2:06:01
Exploring the future possibilities of launching data centers in space and related technologies.
“You know, they, so they don't know to think through how do we manage this over a longer period of time?”
SpaceX and Data Centers: A New Frontier
2:06:01 to 2:08:35
Explore how SpaceX is positioning itself in the data center and satellite internet landscape.
Future Planning and Trends in Technology
2:08:35 to 2:12:12
Discussion on the importance of planning for future technology trends and changes.
“Amy Webb from the Future Today Strategic Strategy Group, FTSG.com.”
Future Planning and Trends in Technology
2:12:16 to 2:15:16
Discussion on the importance of planning for future technology trends and changes.
“They're the leading intelligent, this is so good, intelligent content management platform for enterprises.”
AI and Business Transformation Trends
2:15:28 to 2:20:01
Analysis of how businesses are adapting to AI and its implications for the future.
“That's just a stock market ruse, isn't it?”
The Shift in Business Models with OpenAI
2:20:01 to 2:26:16
Explore how OpenAI is changing the traditional business model by cutting out intermediaries.
“We're seeing people at Amazon just make up stuff just so that they can have the token usage.”
The Impact of Mythos on Cybersecurity
2:26:17 to 2:28:44
Discuss the implications of Mythos and how it affects cybersecurity discussions and practices.
“have like an awesome couple of years ahead of you in a way that you would not have had before How does Mythos change things?”
The Challenge of AI Accessibility
2:28:45 to 2:30:39
Address the issue of access to advanced AI models and its impact on inequality.
“dark side hacker, whatever you call them.”
The Future of AI in Business
2:30:40 to 2:33:05
Consider the future landscape of AI in businesses and the rising challenges they face.
“That is a really problematic situation where only big companies and billionaires can use real good AI.”
The Future of AI in Business
2:33:13 to 2:35:09
Consider the future landscape of AI in businesses and the rising challenges they face.
“It's a unified suite that brings your financials, inventory, commerce, HR, and CRM into a single source of truth.”
Bitcoin Wallet Woes
2:35:10 to 2:38:48
Discussion about lost access to a Bitcoin wallet and its implications.
Private Equity and Brand Quality
2:38:49 to 2:41:42
Exploring how private equity impacts the quality of well-known brands.
“right now just to make you feel really bad what's 45 times 77 000.”
AI's Role in Personal Decisions
2:41:43 to 2:47:16
Using AI to navigate voter guides and other personal choices.
“called bluebell then picked up jansport then the north face then they bought east pack they bought Kipling, they bought Eagle Creek.”
AI Tools and Design Preferences
2:47:17 to 2:48:00
Discussing preferences in AI-generated content and design aesthetics.
“And it's like, oh, well, that's a great idea.”
Exploring AI Dinner Recommendations
2:48:00 to 2:50:58
The hosts discuss using AI models to plan dinner based on preferences and inputs.
“Well, I probably knows food better than a lot of the others.”
Dinner Plan Insights from AI
2:50:58 to 2:53:27
ChatGPT offers a creative menu suggestion, highlighting its ability to curate personal dining experiences.
“I have max subscriptions for all of these.”
Humor and Observations on AI
2:53:27 to 2:55:12
The hosts share humorous interactions with AI and discuss its quirks and limitations.
“we wanted to plan a show with brian about making a secure pc is his pc still secure it sure is he's In fact, today he's not here because he's at his quarterly geek dinner.”
Wrap-Up and Future Shows
2:55:12 to 2:56:41
The hosts conclude the show, discussing upcoming events and how to engage with the podcast.
“We're just heads down on a lot of stuff.”
Transcript
Automatic transcript. May contain errors.0:00It'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 going to be a great one. Amy Webb is here, futurist from the Future Today Strategy Group, Harper Reid, AI guru and technologist from 2389.ai. We have lots to talk about. Trump and the CEOs going to China. 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 going to talk about just make sure you watch this show because Twit is great and next
0:42podcasts you love from people you trust this is Twit
0:53this is twit this week at tech episode 1084 recorded sunday may 17th 2026 don't overcook the asparagus it's time for twit this week in tech the show we cover the week's tech news and you are in luck if you've never listened this show or if you've listened for 21 years that's how long we've been doing it. Either way, you're going to be thrilled that we have Amy Webb here from the Future Today Strategy Group. Yep, that is me. FTSG, she's a futurist. So good to see Amy. One of the smartest people in the world and always welcome. And her buddy, it turns out the last time they were on, I didn't even know they knew each other.
1:38It turns out Harper Reid and Amy go way back. Harper Reid, 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 going to relax, sit back, and let you guys just jam, man. Harper, should we get this out of the way now? Oh, should I never have the stuff? I always need the thing. There it is. This is how we know each other. Can you screenshot this and send it to the WhatsApp group? 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, but they're not everybody was here the last time.
2:18So 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've, I guess this is going almost my 10th year. Um, you, you, 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 sort of alternate between the two and, um, yeah.
3:02And then you become a lifetime fellow, which is what Harper and I both are. And it's truly awesome. The people are incredible, incredible people doing really meaningful things. Yeah. And we see each other in person throughout the year. And it's a great, great organization. You have to be under 40 to apply. And if anybody... They raised the age a little bit. Okay. It used to be 40. You don't have to speak Japanese. You do have to demonstrate leadership and have some kind of interest in Japan if you're not Japanese. But I can't say enough about it. It's really an incredible program. Does it involve going to small nightclubs to see hardcore punk bands at all?
3:46Yes, 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 bring some of our, at that point, delegates to see a punk band in Seattle during our session there. But it's been really great. And it's interesting because as, you know, ostensibly, this is a soft diplomacy program. And the thing about soft diplomacy is it shifts the target shift throughout the life, I guess, life of these kinds 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.
4:33Um, it's good. Yeah. And it's not just, I was going to say, we've got some like Marvel executives. We've got, um, recognizable names that I'm not going to name out loud, but people you definitely have heard of recently in the U S you've got me, you've got Harper. Yeah. Yeah. So I, you know, I imagine there's been a kind of a wrenching of the of the late lately of the diplomatic focus. What was everything going on in the world? In fact, that's our first. What do you mean? I don't know what's going on. I don't know. It seems like something's happening. The first topic is about China, because Trump and 16, something like that, CEOs were in China this past week meeting with President Xi, President for Life Xi, talking about...
5:21I think that was unprecedented, too. I don't know that there's been another time in history when we've had such a group of CEOs like that traveling with a group of government officials. pretty amazing yeah yeah and in fact one of them hitchhiked uh jensen wong 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 pj flew up to alaska med air force one and joined the exodus ceos from nvidia apple tim tim apple was there exxon boeing qualcomm blackstone city group visa all there uh elon's son was there let's not forget about that oh x yeah that one was elon was there too right he always brings his son with him yeah there's a hilariously awful video of him he's 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 going to let us do what we want with taiwan and you're going to stay out of the way um so so that is why he was not not on the list uh so why was so then the question is why they changed 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 going to sit i mean that's the other thing like yeah is literally 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 lightning rod was was you know tough well and one of the deals that the trump administration made was to clear uh nvidia chip sales to china not the best the h200 not the yeah but that was already yeah that's fine but well it's what's funny is the chinese nobody in china would order one because the chinese have decided no we're gonna do our own and 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 you're an expert on this yeah well i don't know if i'm an expert on it but um china for years has been producing exceptional quality everything.
8:16They literally have 70 % of the global market share for EVs and also have like literally 100 different EV companies. So good that America has had to 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 affordable. They're 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 um sure yeah so anyhow um chinese ai models are only a couple months behind ours at this stage you think that close really i think it depends on who you talk to but again i'm not an expert expert well i've used quen which is what is quen wild ways or xiaomi's i can't remember no glm is xiaomi's uh there's a lot of quite good they're not quite there harper what do you think if you well you're also comparing them to um you're comparing an open model that the weights are available with a closed um yeah i can run a model yeah that has untold tweaks to get it to be more effective on the backend that we don't know about versus Quinn or GLM or Kimi or any of these that are effectively free to run.
9:41And I think this is also one of the reasons why I have enough 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. And in many cases, it's uncensored, like they've removed the Chinese censorship from Well, I think there's a funny thing that happens, which is we like to think that, you know, China is behind or they say, oh, the models are behind. 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 3 will pop up or whatever. GPT-120. Yeah, GPT-OSS will pop up.
10:22And 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. Although I have to point out, both Anthropic and OpenAI say that's because they're distilling our models. Yeah, but they distilled their own models off of themselves and also the internet. That's true. Everybody's distilling. I think that's a funny critique. Hey, they stole from us. We just can't have it both ways. You know what I'm saying? 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.
10:57which I feel like is a kind of the narrative. I just think we need to really be, if the U S really wants to have a, 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 U S followed in the past of releasing what amounts to a free technology to the world. And the U S is with open models, et cetera. And the U S has been reliant on closed models. And so if you talk to entrepreneurs outside the U S a lot of them are relying on these open Chinese models. They're not relying on closed US models. And if we're, 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 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, you know, which is like the Linuxization, so to speak, of these models, just spreading it out for free.
11:59I find that really interesting. That's correct. And if you couple that with the amount of infrastructure investment that's happening, you know, our hyperscalers are going to invest, what is it, like$800 billion on paper right now in building data centers. They're burning cash because we don't have a national strategy. It's all being left to the market to figure out. That's not what China's doing. China's got a plan. 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 going to have Wi-Fi, broadband.
12:40So, they're making it so that everybody can participate. And the models that people have access to are not cost prohibitive prohibitive the way they can be in the US. So that just creates this interesting circumstance where you've got significantly more distributed access, and people can around and find out, you know, like, 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. You know, it's all like, it's all apocalyptic, or it's this like utopian, wonderfulness that will happen.
13:19China's having a much more pragmatic conversation. So if you talk to just everyday people there, and I do, they're more worried about whether or not they're going to 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, get better with this, because I want to make sure I get the best possible job. 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 jensen wong's interview with dwarkesh on the he made a kind of an interesting point which is self-serving which was uh that we shouldn't block nvidia's chips in china that by doing so we're forcing them to create their own technology and not use CUDA, his proprietary technology, which is bad for the world and bad for America.
14:18His 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, the implication is be ours. Yeah, I don't know about that argument because there's a lot of, everybody wants access to China's market. It's a big market. And China has been very smart about not giving access you know so like i that that's a self-serving request it seemed pretty self-serving but i just it was interesting he seems pretty good 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 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.
15:06They're there for money or Facebook. 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. But people seem to really like they're 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 wants, like he's done a good job. Okay. So I have something for you to Google, Leo. Yes, I'm good at that. I'm going to have to spell it. Although I don't use Google, but I'll do this. You mean search the web for.
15:36Yeah, I don't use Google anymore either. Yeah, I'll Kagi it. It doesn't 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 a G. Okay. It is M-A-L-E-O-U-S space AI. And you want to hit the Amazon link. Malleus. Yes. It's a robot for AI robot for kids. So this is an example of a... The names sound scary. M-A-L means bad. And also for the elderly. This is an example of one of these little... This is a deep seek agent that you can talk to and has a very fast real-time AI agent on it. That's pretty good. Does it go to the cloud or is it a local agent?
16:25It goes to the cloud. It's all via... I took it apart and reverse engineered it. It's all MQTT posted to some endpoints in China. A lot of other people have done this as well. But what's interesting about it - Is it always listening to me and sending China everything I say? Oh, yeah. And you have to hook it to your Wi-Fi. This is definitely for people who really want to do that. But it's free. How long before the FCC bans this? It costs 40 USD. What? And there's no account. There's no sign up. There's nothing. It just works. And it is pretty effective. Like, pretty effective. Like, I just wanted - A friend told me about this.
16:59this 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. 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$200 a month subscription for the mediocre version. That was literally$40. Like I paid$40 with it.
17:33It came, I hooked it to my wifi, like a crazy person. And then immediately started talking to it. And I think that's seem like this is just give inviting China into your network. Yeah. But I think you're missing the point though, Leo, because the point isn't that China's in your network. Like, like you already have Govi lights and like all sorts of other stuff. Like they don't have to have some$40 device. The point is, is that this is a free AI thing. That's very effective and is competitive with with okay so this one's unavailable but what about the hey yo zoki ai desk robot from deep seek here's this one i can buy and have tomorrow yeah get it try it it's fun it uh it's a little rabbit adjacent is it not i have a rabbit i did buy the rabbit because it can talk to my open claw which i don't have but i had i had my ai agent simulate open claw so the rabbit doesn't know any better and i can talk to it through so okay show you don't know anything about this one, but this is also deep seek powered.
18:27So I figure it's probably just, if you look for the deep seek powered ones, they're for the most part free and easy and they're pretty good. I would not, I know you're not going to use it for real work, you know, but it, but it's like, it's a$20 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, they don't have hardware that you can give to your parent or your kid. No, in fact, all we have is this crappy Amazon Echo Plus thing, which has a sassy voice and then tries to sell me stuff.
19:04Or Siri, which is just stupid. In 2018 or so, we were working with Microsoft. This is, I'm not divulging anything now because it's been a while. Statute of limitations. I wish that's how IP worked. Could you imagine? 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 comprehension and um recognition and nlp and stuff like that at any rate what happened what happened was they got fixated on a device and wanted to get a sub$40 Bluetooth Alexa competitor, Amazon, you know, Echo competitor in the market.
19:58I bought it. And sort of lost where, you know, lost the, I don't know, lost their path, lost their footing. You've got all, and it's hard to bring hardware to the market. 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 that's been true in China and Japan, where Harper and I both spent a ton of time, because things are more open. And there is significantly more competition, 100 different companies trying to bring evs to the market that's that's in completely unfathomable we cannot wrap our heads around that in the united states um but isn't it ironic because they're communist and we're capitalist but that sounds like they've got free more more of a free market than we do is that easy i didn't see the air quotes there leo free market no i meant on on the capital and the communist like there's there 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 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.
21:32It'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 players. It should be the opposite right now at this stage. We should have 100 different legitimate companies in every single part of the ecosystem. And that is not what's happening. We have a couple of dominant players. Chairman said, let a thousand flowers bloom and then george bush said let a hundred points of light there's um a really good book that i think talks a lot about this by dan wong called breakneck yeah really really good i've got it right over there it's a great 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.
22:26But he just says the US is a lawyer state and China is an engineer state. And so you have what kind of comes from that. Like you have a lot of laws and regulation in the US, which, you know, is like all we talk about. But if you look at our leaders, 100 % of our leaders are all lawyers. So of course, it's like my famous economist friend of mine always tells me, you know, never ask a hairstylist if you need a haircut, right? It's like you do what tools you have in front of you. And whereas China's leaders are all engineers. So it's going to 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.
23:07Whereas 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. But the point being that we're using the tools that we have available, and it might not be the right tool for this moment in time. 100%. The tools of regulation, like this is going to open up a can of worms, so I don't want to do that. Other than to say that this is not a moment to regulate. We cannot and should not attempt to regulate AI. There are other ways forward, but it requires making some tough decisions with regard to who can make money and how.
23:46how 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 also China's not going to be the one innovating and creating brand new stuff. They are going to 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 would love to see the United States invest 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, you just then, then cool.
24:23You're, you, you're learning about AI and you can't use any of it. You know, what was, so what was the goal of the Trump summit this week? Yeah. It depends on who you ask from my point of view. 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 don't. So I think that there was some amount of, let's see if we can figure some of this out. China is very, very, we're, this is a little bit of a little bit of 1914, where we had all of these hotspots around the world starting to heat up because there were these political, political problems all over the place.
25:12It feels a little bit like that right now. And you could look at this as like, China may see this as an opening to take Taiwan by force, which they've certainly talked about. 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 the u.s tech industry uh about losing taiwan i mean apple yeah because we don't get there's no chips anywhere else they're all made by tsmc and it's presumed if china if china did invade taiwan that they would blow up the chip plants to prevent china from getting the technology i mean tsmc says they 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.
26:03So it's not just those resources. You've got this incredibly trained, highly skilled group of people who are able to work in these places. That's part of the reason why it's really hard to stand up a factory like that in the U.S. right now, because we just don't have highly, highly, highly skilled people yet to do that work. And if we do, they are burdens, the wrong word, but they're burdened by regulation laws, labor laws, et cetera, that are not from today. Yeah. 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.
26:39And so we have this stuff that, you know, China never had to go through that as a country. They were able to define their labor laws, which, you know, some are bad, some are good in the modern age. And I think that just, that's, you know, this is kind of a lot what the Dan Wong book is about, which I highly recommend. I can't recommend enough if you're interested in seeing that. 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 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.
27:13Here'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, etc. I remember visiting China in 2009. i was a chinese studies major in college i oh 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 it's the pearl of spoken word uh i uh i um uh in 2009 visited china and i had kind of come to the opinion that while the 20th century was the american century i think we all agree that the 21st century might well be the chinese century yeah and and our guide who uh was a brit expat said oh no you don't understand the economic issues besetting china this is 2009 we'll 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 centuries a long time in the year 20 decades let's have a decade decade yeah um 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 rate issue too huge birth rate issue they're also having lots of internal political issues they just launched a brand new party called mirai mirai something um mirai in japanese means future um so it's a 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 try anything new a lot of people i have the impression they're tradition bound is that no it's just just like stubbornness but that's not across the board but it is no it's like some japanese people are very forward-looking and very uh yeah i'm mostly talking about men of us like older men a certain still run things.
29:36Yeah, not younger people. But the point is, there's so much pendulum swinging in a lot of these countries because of extreme views going back and forth. And China is just plowing ahead. So the lack of uncertainty in China is partially what's going to help that 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, but it does reduce some of the, like, worked for Singapore, you know, constant back and forth. We're going to invest in coal. We're going to invest in EVs. Wait, no, sorry. We're investing in coal. Every four years we, uh, we decide something else.
30:16Yeah. And that makes, um, that makes the business climate like impossible to work in. We look, I'm in the room with leaders of major companies and just about every industry and their boards. And 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 because they don't know what's going to happen and so it's every everybody's just waiting that's a terrible situation to be in isn't this almost a global situation every polity in the world is struggling with the future because the future is so uncertain that's that that is an excuse.
30:58So the answer is no, you should always be focused. You need, you need your near-term strategy and you need the longer term. And there's even if you have no idea what's going to happen. Let me give you an example. Are you familiar with Toto toilets? Yes, I have many of them. We have, we actually have the, the Toto that South Park made fun of in our bathroom. I miss it. We moved and we, we left our Totos 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 actually what makes it toe-to-to-toilet really great turns out 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.
31:58Right. 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 that 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 going to learn so much in the process. And everybody who's a part who touches that in any way is going to level up in how they're thinking.
32:34It's going to, you know, they were just lucky. No, but no, there's plenty of examples. Even if they were, this is kind of manufacture love. Look, you can't just like my favorite. Do something. You can't be lucky and no opportunities exist unless you start to push in the doors. This is Vinito. So my favorite example of this for a Japanese company is Yamaha. They make everything from musical instruments to rowboats to speedboats to motorcycles. And like they just make everything. There are no companies like that in America. Well, there are conglomerates. That's a little tricky, too, because... General Electric made everything.
33:11It wasn't a good thing. Yeah, I think that the what I'm trying to point out is, and I don't want to name any company names, 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 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 great labs that used to exist.
Read the full transcript
33:48And 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. And the great irony is like, that's 100 % of what China used to do. You know, and we flip-flopped a little bit. I need to take a break. This is such a good conversation, I think so important. I don't want to interrupt it. So hold that thought, Harper. And I think the T is this way, but okay. And then I don't know what that is. That's orthogonal. I don't know. So I would actually kind of wonder if either of you have an opinion of what comes after this summit, whether anything was accomplished and what comes after the summit.
34:37We're going to take a little break so Amy can leak. Harper, I, by the way, this has already been an expensive show and I blame you, Harper. This is so gross. No, no, I had my power bar already, so I'm okay. I bought the robot that you mentioned before the show, the Ricci. and then i just bought this thing that's going to 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 no but but if you read the reviews on that site there is a perfect a perfect review um in that kind of that 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 what he's 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 192 168.1.1 they hacked me and you're just like oh 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 i'll put it on a separate lands just just play with it hook it to a hot spot on your phone don't worry about it don't hook it to your real big network and just kind of 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 wi-fi password so what they've got nothing they're gonna come and they're gonna take all of your stuff just don't hack my toto that would be they're already in your router anyway they're in your router already they made my router but you know i know 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 um anyway we're gonna take a break and come back this is why we love having harper and amy on i hope you're enjoying this week in tech brought to you this week by threat locker now you know why you need threat locker threat locker 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 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 threat locker ensures that 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 365 or asana or google works space or github or whatever what this means is even if your employee gets successfully phished it's increasingly difficult to prevent that right i mean you're doing everything you can i know but even if they got phished 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 like Windows Hello or Thumbprint or whatever.
37:46I mean, it's really locking it down. ThreatLocker works in every industry. They have great support, US-based 24-7. 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. Ask Rob Thackeray. 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 to threat locker he said quote threat locker was the most intuitive 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 was very good it's great to have an ongoing relationship with a company that is so responsive to our requests end quote that's what rob said 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 threat locker consistently receives high honors and industry recognition g2 gave it their high performer and best support for enterprise summer 2025 pure spot ranked them number one in application control and get app gave them the best functionality and features award in 2025 the awards just keep on coming with threat locker 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 get unprecedented protection quickly easily and cost effectively with ThreatLocker.
39:37Visit ThreatLocker.com slash twit. Get a free 30-day trial. Learn more about how ThreatLocker can help mitigate unknown threats and ensure compliance. That's ThreatLocker.com slash twit. We thank them so much for supporting this week in tech. What you got there? A little Lego miner? What's he got? A vacuum? Devo guy. Oh, he's Devo. He's got a you-must-whip-it guy. yeah yeah he thought he had a whip but i wasn't going to say it now i see it's he's got a planter on his head yeah he's got a whip in his hand who else could it be but devo 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 reach you i can't what did it break what's going on no it It just doesn't have power.
40:33Oh, the battery died. You know, I had, what was it, the OBS bot? 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 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 WeMo stuff. And WeMo was just shut down, right? And he was just like, what do I do? And I was kind of like, well, what do you want? But the way he described it, he's not a technical, technical person.
41:11He'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 want to be able to hack it if the service goes down. And then I was like, just do ESP32s. 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 claude here's an esp32 switch exactly i've had claude take the uh reference firmware and rewrite it yep and uh it's trivial to do and uh claude is very comfortable with it and can do all sorts of stuff with it yeah we do this quite a bit 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 esp32 board you can typically rip the firmware and decompile it and kind of basically an idea of what happens because they don't encrypt the the hardware right um i'm wondering because the esp32 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 and i wonder is this the chip inside an echo no why would it know alexa i'm guessing that they trained like so you're talking about, I think, open wake word.
42:26Yeah, you can use it. 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 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 word. Well, I did. I had Kokoro, which is a text to speech synthesizer. Very good. Generate like 2000, both hey, Kenobi. And then you also have to do false positives. So generate a lot of just random stuff or, hey, you know, hey, Schenectady. And then, so there's a lot of training involved. And I have all those samples still.
43:04I just have to, I have to buckle down and finish. Yeah, it's kind of a pain in the butt. I tried to do one that said, hey, Ricci. They said, use Alexa because it's got the most, that's probably why it has Alexa, because it has so much training. There's such a long history. with a company the other day that has a big G in it. And the guy kept saying, talking about Google, because it was from Google, and his Google Home kept going off. Like, we're talking multiple times. It was so funny, because I was just like, what a... I've muted all of the smart devices in this room, because if not, they would go off constantly.
43:41I just noticed your chair. You look like a child, Leo. I just switched. I had one of these fancy office chairs, and I've just switched it back to my Dr. Evil chair. that I had in the old studio I have to turn around and show you the back wow because I'm an elderly fellow I'm also very interested in all this stuff from the point of view of uh it's going to 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 going to come and take away my car keys they're going to say, Dad, you shouldn't be driving anymore. And I'm just hoping by then I will have an automobile that I will just get in and say, take me to the supermarket.
44:29I love Waymos. I have a good hack to get a Waymo to LAX from anywhere in LAX. What's the hack? You get dropped off at the LAX 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 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 take me to the LAX parking 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 that's where I'm calling from 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 way mo story in here because apparently um uh they have uh added waymo to atlanta and uh empty way mo's invade atlanta neighborhood circle cul-de-sac for hours with no passengers did somebody do this twice 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 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 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 so good wayward waymos what happened 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 graveyard look at all the waymos just look at all oh my god this is not adidas this couldn't be beautiful it is i think it's because 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 the 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 he put a little child thing in there and it's blocked them and now they can't go oh no that was mean steve local reporter causes waymo confusion we need more like give these waymos some free time give them some space amy are you 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 southwest had a funeral I did.
47:30I mean, I didn't have. Well, no, it wasn't your funeral. It was a funeral for your yearly annual report. Tell us about that. It got a lot of attention. It did. I, we, so we have been producing this trend report for 20 years. It would have been the 19th year. You've been sending 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 so much happening right now that I need people to unhook from what's easy and engage with what's more difficult.
48:14So a trend report is easy to put on a shelf and it absolves too many people of having to make decisions. So we're not doing it. Well, you also made the point that you can't just do it at the end of the year and have it be that it's constantly changing right correct so and also because we i've been tinkering with this new model um there are the i frame this in like an economic argument so i promise this won't be super boring but there was this guy named joseph joseph schumpeter who so oh yeah very famous um yep wrote this book called can capitalism survive that i uh i read for the Yeah, well, I read it when I was in college for the first time, and I still have it, and it's dog-eared and highlighted.
49:00At any rate, the idea is capitalism is this perfect storm, this perpetual storm that's constantly happening, and it sort of gobbles up old existing technologies and creates new ones. And that's kind of what a storm does, right? It messes up the ocean, the shoreline, whatever, and makes way for new things to grow. That's what's happening right now. It's called creative destruction. And you have to be willing to see what's happening, which is the exact opposite of what our government is doing right now, so that you can make way for what's coming. But does that mean that there are periods of time where people are miserable?
49:47no i mean look you're you're miserable you know who's going to 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 whether if there's bad skies you know i'm not going to be the last person holding down the fort. I'm going to evacuate. There are always people who you, there's footage of them on the roof of their homes because they just wanted to wait things out. The message is this time around with the technological storm that's happening, you don't want to be that person on your roof because nobody is coming to rescue you.
50:29Nobody. 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 I do? You have to be willing to ask really, to sort of think the unthinkable, which means if you're a company that's always made money doing 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. And that's fine, but don't dig your heels in and refuse to make any changes. Is there something an individual should do?
51:10So we are calling what you're looking for are convergences. AI alone is interesting, but it is impacting other things. What you want to 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, surveil you. You're looking for these small intersections. And then you have to be willing to adjust and adapt. But most people are not willing to do that. They're just kind 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 yeah i mean uh this is a cory doctor wrote this last week that 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 uh but he owes so much to the oligarchs the people who are creating this unpleasant future for people uh that he he has no 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 well it might be instinctual i agree it may not have been thought out but it may be instinctive that i mean the populist thing resonated it worked quite well got him i i know people don't love the politics when we get into it i want to highlight i am not i'm politically independent i have voted on for both both sides different parties at different times this particular administration is acting in ways that are truly unprecedented yeah i'm not politically independent you work for obama's 2012 based on my wikipedia page 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 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 u.s and of course not just in the u.s i do think that there is 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 and i don't think that's just a trump thing as well i think the cohesive of planning that you see from some other countries is where you get a lot of benefit.
53:50There's also a little bit less freedom in some cases. And I think that freedom is... Can you have planning and freedom? I don't know. But I mean, 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 700 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. But we've, I don't know. I read, I don't know if I should say this out loud.
54:32I'm going to say it anyways. In ninth grade, I read the Communist Manifesto and was like, I'm a communist. we've got everything wrong in this country you know what everybody's sensible at that age 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 ran and was like yeah that exactly exactly i am howard reardon yeah i was like i i've read i read um foundation and uh and the fountain had the same year and i was like this everybody 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 yeah but i do think that capitalism has evolved in a way that's no longer healthy in the united states uh well i mean it's a great example 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 no money no brain or no money yeah it's roughly the same yeah uh so go ahead i'm sorry hyper i didn't mean 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 uh competition is hard but so are It's really, really, really hard.
56:00Oh, they're all protectionists. And I do think that we don't want that. We are soft right now. And we need to be, we're in a world that is not soft. I should mention, by the way, that the Office of Technology Assessment was not wiped out by the Trump administration. It goes way. Newt Gingrich did it in his plan for America. And it was under the Clinton administration that it happened. And it was the plan on America. The contract on America. Contract on America, yes. The Dan Wong book was when he was interviewing some people about making masks and PPE around COVID. And he was talking to an American entrepreneur who said, that's just not our core business.
56:42And 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 these two systems and how they're colliding. 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 I look at my friends who are not in the US, not in startup culture, not in venture-backed 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 a raised in captivity wild animal.
57:13These people are like, you know, like gazelles born to run. And it's just like a different kind of world. And I think it's, we're going to have some hard conversations with ourselves in the next five years. Here's the article from Semaphore, Rachel Jones writing, the class of 2026 is cooked. It is a tough time to be graduating from college right now. There's no question about it. Tech companies have slashed 100 ,000 jobs. Cloudflare just fired 20 % of its staff to be replaced by AI agents. Meta's firing 8 ,000 people. I don't know what happens when you graduate from college. these days uh this is a graph of total non-farm hires uh from 2010 through 2026 and you can see the precipitous decline as soon as chat gpt comes out i mean the companies may be ai washing they may just be using that as an excuse to fire fire people but i don't think there's any question i mean if i were graduating with a computer science degree right now i don't know.
58:22I think there's something I've observed, and I think this is kind of funny. 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 250k a year, or someone comes to me and says, I'm a new grad, you have to pay me 120k a year. And with Claude code or whatever their equivalent, why would I ever hire the middle career person? I think that the younger, I think the class of 2026 is not going to 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 of people who are getting laid off as a mid-career person.
59:08Now they can get a job, it's just not going to pay the same. And I think this is going to have a really somewhat catastrophic impact on our knowledge worker communities, our dual income communities, suburbs in general, because you're going to have a group of people who've made 400K a year with two people or more. And suddenly they're not going to make that. They're going to make half of that. And that's just going to cause a lot of consternation, I think. Or what does the Tea Party look like when it's not caused from NAFTA, et cetera? It's actually caused from Facebook laying off everyone. you know, and I don't know, but I think about this a lot and it, and it, it scares me a little bit.
59:46This is why they're populist movements. Well, I had 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. 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 are we talking like yeah 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 going to turn into a mail server for some like you know i'm going to wake up in a thousand years as the mail server like we're all batteries don't you know didn't you see 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.
1:00:35Because it seems to me that the human doomers is just pain and suffering for more people. Whereas AI doomer is like, I get turned into a paperclip and yeeted into the universe, which I'm here for. So that's kind of my perspective. Okay, got to go. Bye. What's the future? Tell us, you're a futurist. Please. Yeah, look, I... Tell us what's happening.
1:00:57And there are a lot, there's a lot of money to be made from doomerism. Yes. And so the future currently is the people, and we all know who they are, who are proclaiming end times are going to make bank as in some way. I have to put out that same as it ever was. I used to work with John C. Dvorak. That was his shtick. That's right. Oh, that's right. After the keyboard. Yeah. no he was a cousin to the keyboard but he was a guy i mean he was a computer columnist 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 you know yeah ray dalio uh a very famous um money guy is out he he's always been the, you know, there's a huge cycle coming.
1:01:54It's, it's going to be off the charts. And they always say, he was right in 2008. 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. 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 way I can do that either because there's too many variables in play and there's too much change. Like you can't hook me up to Willow. It's still not going to happen.
1:02:30Right. So I don't believe in the absolute apocalyptic, like super intelligence stories because at the moment I don't have data to support those that could change in a week, but 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 who quite frankly, maybe 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 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 1999 and 2004 when everybody had to learn Flash, you know?
1:03:29So 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 commercialization of space, which has to do with manufacturing and a bunch of other things that just haven't settled out yet. So we're in this moment of transition. I call everybody alive today the transition generation. We are transitioning from where we were before to the time after. You know, people talk about, I had a computer growing up, but no internet. It's a pretty common story for people in my whatever 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.
1:04:33I remember this time 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. The Google foo. Yeah. Yeah. So I'm a 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. She is 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 going to enter the workforce with an incredible set of skills that she had to build on her own, plus all of the augmentation because of AI.
1:05:23So 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 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 i mean the signals are talking uh the most recent one is the uh the one about 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 before that the big nine which was prophetic uh talking about ai um 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 going to divorce me so i we have to it's a balance it's painful it's a balance i understand i'm going to be writing a prolific substack um there you go are you oh that's fantastic that's actually modern yes i'm writing a book shouldn't i've been on substack for a while and i've just been reads books yeah so anyways there we go yeah and 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 the pictures 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 it's very very good one for my house?
1:07:22Well, it's interestingly, it just uses an old Monoprice 3D printer that we just took with a pencil on it. We did hook up a brush to it to see if we could get brush strokes for more of a calligraphy style. I kind of like 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. We spent a lot of time trying to make it cheaper using local models and whatnot. And it's pretty fun. It's very, and it's also very funny. You call it Mikaso. Yeah, because it says Mikaso really loud when you hit the button. Mikaso is su-caso.
1:08:01Wow. A 3D printer into an AI portrait artist. And I guess it's your colleague, Ivan Indrautama, who wrote this. Yes. And Ivan and I have worked together since 2005. 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. And I've been very lucky to have that. And so I really am happy about this. I love the big button inspire and the other big button realize it's simple. It's a simple machine. Yeah. I've been also built our little mics that are transcribing everything we say, and he's kind of our hardware hacker. And this week we tried to nerd snipe him into doing the ESP 32 wifi mapping stuff.
1:08:41So you could see respiratory rate and heart rate of everyone that's inside of the office. Like astronauts. Well, I did this thing when I pitched a company over COVID where I had a heart rate monitor on my screen because I wanted the VCs that were pitching on video to understand that they had a physical, the words would actually impact me physically. And they were trying to get the heart rate to go up, I presume. Of course. But what my team found is that if it was a good pitch, my heart rate was up high, 120, 130, 140. If it was a bad pitch, I was at 60 flat. And they would just be like, I don't like it, but it wasn't bad or good for the VC.
1:09:25It was bad or good. They were like, Harper's out. And so the team just started to use it as a signal on whether they should participate or not. They'd be like, Harper's heart rate dropped. He's done. He's pulled out of the conversation. You can tell his brain's turned off on the flip side resting heart rate of 60 is pretty great yeah no kidding so you got pretty good with that i uh i'm i've been trying to get my cardio in shape turns out i'm getting old like leo thank you sorry leo or you could do like amy does and do gravel bike racing i also been right on the road it's not just gravel okay yeah i've been running with noise canceling headphones on that's a good idea that that's a great idea go wrong.
1:10:07I'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 hear this, you just go, what is that a truck coming up behind you? But as, but running, I've been running roughly 10 miles a week, which I start from zero, but nobody on the road pays attention to anything, Amy. I think, I don't think it meant, I think you're right. You just have to yell bike. Like one cares no one pays attention it doesn't matter um and so frustrating amy's 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 trolley and there are 20 people tied to the track on one track but if you take the other track you're going to go off a cliff and kill everybody in the trolley the ai has to decide i guess there's only one let's say there's one person in the trolley do you kill the one person who's riding on the trolley or do you kill the 10 people and save the one person on the trolley well amy decided to save the person wasn't paying any attention and wiped herself out never again next time people in the future what the future is in 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 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 was built to solve scribe is a workflow ai platform that 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 80 000 plus enterprises including nearly half of the fortune 500 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 redact 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.
1:12:50Scribe will 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. To book a personalized enterprise demo, visit scribe.how. slash twit. That's S-C-R-I-B-E dot H-O-W slash twit. What a great idea. Scribe dot how slash twit. We thank him so much for supporting this week in tech. Actually, speaking of Chinese models, one thing people seem to agree is that Chinese models are better at video now than ours are very good they're really that's what ron's using i suspect to make those propaganda videos uh which are also very good which are also very good uh by dance and kwaisho's models outperform western rivals this is from the financial times in realism and scale so your contention that ai in china may be only a few months behind ours open ai open weight ai in China might only be a few months behind ours.
1:14:08Well, in some areas, they're actually ahead of ours. Here, can you scroll down a little bit? Because 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 that is not known in the West. Right. Well, it is a very Indian, but these little tiny little nuances lend an authenticity. Yeah. This is using C-Dance. And it is more than like 15 seconds long. Yeah. C-Dance is incredible. It's an incredible product. I think you mentioned that head wobble is really interesting, Amy, because this gesture, you know, we have in the West maybe nodding yes and shaking your head no, but this gesture is very meaningful.
1:14:55And there's multiple versions of that. There's not a single head wobble. There's different cultural things. And so the ability to do that in a AI is fascinating. Yeah. I don't know. 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 would probably think it was, but to your point, and that that doesn't look very good. And the text was bad. It's not good at 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 intelligent machines called chloe does history if you guys see the end see this it is um so chloe which who is fully um ai generated uh is a kind of typical influencer right she's doing a lot of selfie videos in historic places so here she is in ancient in ancient london no no no but 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 going to check in at my room in the inn get into the market see how people actually so the guy who does this is a is a historian he's interested in actually recreating history so he uses 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 1536 but somehow he's doing the influencer as well is chloe vs history is the uh channel now he isn't really forthcoming he has i bought it a 69 e-pub 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 one of the giveaways is this very chop is that choppy is it is the history good is it like credible he he says it is um okay oh good yeah Well, okay, so this one is her in the Titanic, right?
1:17:21I 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 going to try to warn the captain that the Titanic is going to hit an iceberg. And so she finds him. Captain Smith, I have very credible information that there is a significant iceberg risk tonight. And he just brushes her off completely. I think some of the history in this is a little bit influenced by the movie Titanic. But I've 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.
1:17:58I think this is impressive from the point of view of if you had a ninth grader 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 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 1.9 million views of this titanic video 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 years ago and i forgot was not maybe not maybe not um 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 ancient rome and what he's done is taken uh drawings and pictures of ancient rome and and brought and animated those and put chloe in the scene every once a while there's some anachronisms there were some guys in sunglasses walking around but 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 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 you know i think it's just a short step from this to maybe making it a 3D VR.
1:19:30But isn't this kind of what Amy was talking about, where you have these 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 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 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. For better or for worse, now I can participate in some way, you know, or someone could make a product where someone could participate in that, whether it's on YouTube or what have you.
1:20:21I 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 I do think 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 there are all these glimpses. waymos 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 are you talking about the neo no no he's talking about the um is it not format factor it's not factor it's uh was it factor chat room what's he talking about he's 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 this is the company it's also working on skin that they're trying to make lifelike human i cannot it sounds kind of creepy yeah it is creepy yeah i mean the neo is creepy here's an older fella with his little neo home robot it's only twenty thousand dollars you can put a deposit down right now on this look can you imagine having this thing bringing the groceries in figure ai thank you thank you so creepy what is it again figure f-i-g-u-r-e um it made headlines because uh one of their robots worked for eight hours in a factory now oh so yeah we showed that and that was such a bogus uh yeah yeah a video because 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 There was something I read about the difference between Waymo and the Tesla, the cabs.
1:22:22The Tesla cabs 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 AIs to get themselves out of the problem, which means that it's learning, it's continuing. continuing and I think but but the but the the bigger point here is that this is just a glimpse of the future and we are probably not going to 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 going to be a magazine main stand right well it was for it was because that's who was making those websites was people that's who they thought it was this is why the field of strategic foresight which is where i happen to operate is so important because you want to pick up signals along the way as you know like this is what harper's saying right like this isn't the future gives us a glimpse of a change that's going to happen but how does that fully render we don't know yet are you optimists i don't i I wouldn't, I'm not a anything.
1:23:36I'm, I don't, I'm a, I'm a, whatever. You resist this kind of characterization. I do because also, because don't forget in my field, at some point, most of, most of the people become eugenicists and they always do. Why is that? They always do. Because at some point you, because if you are good at this job, 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 the lower the the no it's not about it's not about killing it's about no but this is where eugenics about like changing our genetic structure to get rid of the subhumans and uh it's about yeah i mean why do you think all these dudes are spreading their seat around yeah you know because they believe yeah yeah or like um larry uh oracle larry ellison ellison thank you has an island where you know he's working on longevity and you know that's that's what inevitably 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 is that a continent of becoming filthy rich sea land you see land there was an example we're going to take all these disused yes we're going to take all these disused oil platforms and we're going to make them into little countries where no one's going to mess with us yeah brian is a a baron there oh is he he bought into this is he living there now that was a no no You made him a baron.
1:25:21Yeah, I did. Does he also have a crater on the moon named after him? Yeah, that was 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 really seen this in the past where it would lose packages over the edge and stuff it's much better than it was yeah 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 should be their purpose.
1:26:16But 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. 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, of the Neo robot where the 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 the dystopian future is the fact that they're playing war like that's the worst card game ever play literally any other game like if that bothers me more than the robot it's also placed like the beginning of a horror movie it is such a 100 does come into the house human would you like another beer oh shaggy one it's just not this is not i don't want this 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 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.
1:27:51Or we're in the Mad Max world. 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 Trekkie friend Clint would always say, you know, Star Trek had to destroy the world before they invented the society of Star Trek. True. um and so i think that there's a lot there but i what i want to talk more about what amy just brought into frame here so i have been on the hunt for an omnibot i had literally 30 years do you remember this is an old-fashioned robot this was the this was a tomi robot this this so this thing it had 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, it mostly works.
1:28:41I have to tinker a little bit more with some of the electronics. Are you going to have this wander around the house? What does your daughter think of this? It has a cassette player, right? It has a cassette player. It does have a cassette player. And it's a remote control, right? It's remote control. It does. The problem is the battery is very specialized just to this. And I've been everywhere trying to find one that actually works. That's what happened to my Segways. I had my son using my Segway. My Segways? You had many Segways? I had two. I had two, one for each of us. But the battery died, and it turned out the battery for the old school Segway, the really good one, is$7 ,000.
1:29:17Wow. So I had to buy the replacement, and they're crappy. They were cheaper. That's too many dollars. Do you guys really not remember the Omnibot? Oh, I remember the Omnibot. The Tony Omnibot. Wow. That's what that was. that was the omnibot that was soup it was uh 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 here's one but it's hard they they are very rare same problem the battery will not charge same problem yeah so um it looks like robbie the robot from uh the day the earth so cool the commercials were cool like this was i saw this and was like it blew my mind i was like eight or whatever um i think we're all of the same generation here because like yeah i remember seeing this and being like oh my god remember this i'm too old i don't so i've been tinkering and tinkering and tinkering with it it's clean it looks good i just have to get the battery probably the parent that said we're not buying this yeah no way would you like to know what my parents did my parents it was last night of hanukkah and uh there was 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 because 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 she was just like like her head exploded it was the greatest thing that had ever happened, she got a pound puppy.
1:30:55Wow. I have been telling everybody at school for a week, I got the Omnibot because 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 30 cans of Campbell's 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. And I'm like, 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.
1:31:36Pound poppy and 30 cans. But if anybody is listening out there and has access to a battery that might work, please contact me. I will happily take it off your hands. yeah came to the right place probably there is somebody listening with a tomi omnibot battery that is still functioning there must be a way maybe you could get your guy harper to wire up an esp32 to make this work well i mean you gotta still have the power i wonder i'm just i mean yeah there's got to be a way right can't you just you're gonna have i think jerry rig a battery at this point because any battery from that era is probably gonna be it's gonna be dead by now Yeah, we have.
1:32:17Those are not CADs. Correct. I'm also, there are ways to do it, but I don't want to mess up the form factor because it'll look like it's all. You don't want a lead acid car battery hanging off its ass. That's no good. Yeah, I understand. Wow. We have learned so much about both of you today. And I'm. Oh, just wait. We have a little bit more time. We're not done. We're just beginning. Don't cut it short now. We're just done. No, I wouldn't dream of it. what kind of soup was it? Campbell's? Tomato? No, it was, 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.
1:32:59In the 70s, you could make anything if you could put 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 is 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 going to design the website? How do you handle fulfillment?
1:33:37And I have, you know, a pretty good idea of what it's like because I've watched both my kids actually start their own businesses. but you know what? They got it done with Shopify. Salt Hank's a really good example. My son started making his TikTok videos as a TikTok chef, you know, making these sandwich videos. It was going really well. He had this, you know, brand. It was called Salt Hank. His name is Henry. And he thought, you know, I should be selling salt because Salt Hank and I've got, you know, he pretty, 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 to brands you know just getting bootstrapped like salt hank 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 matched 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 purple shop pay button that's used by millions of businesses like hanks around the world it's why shopify has the best converting checkout on the planet 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 hanks 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.
1:35:33Go to shopify.com slash twit. That's shopify.com slash twit. I love that sound. Yeah, he's, he actually, poor Hank made a video on Instagram. Mark Zuckerberg was in New York City for the 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 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 like i think you said that perfectly you see the loathing choices which I think everyone should make their own choices.
1:36:26But I talked to a recently, I talked to a tech entrepreneur and 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 because I was hoping he was going to 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 vibe. this is yeah this is a statement you have chosen there's a thing there which i think a lot of these tech people have achieved everything in life but 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 no 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 the best taste isn't money related.
1:37:22Like it's like 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, you know, throughout history, which is why the Met Gala is always a cluster anyway. But I find it kind of funny because I think they are inside of them. I think they're all dying. They're being hollowed out as they are just like, where's my AP swatch or whatever. They're trying to, you know, they're trying to do it. 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.
1:37:58And 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 Goldblum 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 at that level, because they think they are at the top.
1:38:33And 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 tippy top. And that's right. I don't I mean, I don't think you can. I don't think they'll get it. Well, Mark decided not to do anything fancy. He's basically in a very pedestrian black tie. His cummerbund is not well adjusted. Priscilla looks great. His Priscilla Chan, his wife, looks beautiful in a red dress.
1:39:05But before he went to the Met Gala, he decided he wanted a Assault Hank French dip sandwich delivered. Really? to his door so my son uh made him one and uh delivered it uh in person to uh to zuck is this real yeah what it's not it's not ai wow uh the funny the ironic thing put on your cool new outfit he's actually asking suck to spill the sandwich onto his somebody asked you like who you by the way suck 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 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 cool and in touch is gross come on man you're better than this uh of course what these incommonators didn't realize is they are commenting about this hank least got mark zuckerberg to buy a sandwich from him they're 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 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're moving then we're moving offices.
1:40:37We'll just get it for the new office. Are you moving to New York? Are you moving to New York City? We've always had an office in New York. We just rented out. We took a longer term lease on a floor of a building. He's in the West Village. Joan's at Bleaker. All right. Got to get there early because he sells out 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 he never played upon his name at all this is not he just yeah no i just uh wow yeah i just connected henry laporte 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 because you can never shave that's incredibly cool you know yeah that's a hot hot hot commodity it's very hot right now yeah yeah uh yeah actually he's uh his next thing he's uh putting um he's buying a restaurant uh in the city to open a fine dining restaurant that he's going to call the ports and i said there damn well better be a family table that's all yeah 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 hit yeah called chess the revival of chess with a new uh book that was more timely yeah is it still running or is it is it over it very much it so um wow it's still on broadway it wasn't supposed to run this long was it no uh no it's his um historically failed every single time somebody has tried to revive it um people might know the song one night in bangkok one night which was a hit it's nominated for six tonys i think congratulations Congratulations.
1:42:47And 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. And the music is the band is on stage and it's more rock music. It was written by the two B's of ABBA. Benny and Beyond. Yep. So I'm actually executive producing a new show about the far future of AI. which is what i thought you were talking about but oh tell me about that you have become are you now an impresario no i just um i'm i'm an investor um that my role in in chess was that i gave them a lot of money uh because to make the show happen and of course uh leah is one of the stars of course she's famous from glee that guy so nicholas christopher who just he's a singular talent it's it's great.
1:43:43It's great. Anybody who's coming to New York or in New York, you know what? I'm going to go to New York cause I've never had one of Hank's sandwiches and I have, I've got to go. Never had one of his sandwiches. There's not a family table. I will, I will call you. I will get tickets. Cause I would like to see this. 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. Maybe if, if, if you can spare them, I won't bring Jeff Jarvis cause he hates musicals, but maybe Paris will go. and i won't bring hank because he hates musicals but uh maybe this is not bad um no this would be a good one to see yeah yeah yeah uh leah leah michelle uh who is incredible and what a great uh personality and singer erin uh tvite tvitt from yep and nicholas christopher who was in hamilton and he was in sweeney top 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 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 i have to say i have been using skills from 2389.ai there's some very good skills in your market we have a we have a couple secret ones that are one of them is called jam which gets a whole group of people to work with you.
1:45:07We have one that's called Review Squad, which is really fun because you can enable a well actually squad of agents to review your code. And it's all hacker news commenters. Oh my God, how funny. So they'll create these really burning things. But one of the things that we've been doing with Review Squad 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 go through and do all of this like, oh, we're going to have someone that has this expertise. I'll always add a cat. I'll be like, also add a cat. And then I don't know if it's on there.
1:45:43It might be. And then I'll say also add Anna Karenina. I've used your Fresh Eyes review, but that's not review squad. This is like super Fresh Eyes review. It's very good. Let me see if I can find the URL. You could have Anna Karenina. What does Anna Karenina do when she reviews your... I don't even know what her personality is. cutting, very cutting and very philosophical reviews that make you a little bit sad. I think you should have Schumpeter in there as well. I don't even know what you just said. Schumpeter, you know, the guy Amy was talking about, the capitalism guy. Oh, that guy? I mean, you can do whatever you want.
1:46:17This is how it works. I have Dinesh and Guilfoyle fighting over my code from Silicon Valley. That really worked well. It's 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, 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 love it, weirdly? How do we make it so people believe it, et cetera? And I think a lot of these are going to be, we're going to try anthropomorphizing. We're going to see how it goes. It's going to stick in some cases.
1:46:55It's not going to stick in some cases. but we have found that anthropomorphizing these things has been very very helpful and so being able to say you know against 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 and i mean some of them admitted it yeah i mean do you think no i'm not a monster no i mean like i mean richard dawkins thinks it's kind of thinks claudia is conscious well we we talk about this and we call them they like 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 short argument with a friend about whether they can be funny like whether they can make jokes oh they can be very funny now well that's the thing is is i laugh at them and my friend was laughing at them well that's what started the conversation is it made a joke.
1:47:51And the fact that it made a joke, I think there's some complicated questions about intent when it comes to humor. 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'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 you know 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 token maxing so to speak 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 25 minutes of tai chi and it said graceful and annoyingly virtuous which is a joke right yeah yeah i logged uh i logged 5 000 meters rowing it says another neatly documented suffering session why chat gpt versus i was a clod guy uh you know i just try them all but uh and you don't find the constant uh fragility of open ai's models to be so infuriating that you don't think anything on your computer screen no i i've never experienced that oh i have they're they're with each iteration it sometimes is prioritizing speed over well yeah but but a atropic was doing that in spades last yeah i don't know this lat the i don't know this is different this is different than the fail whale era of twitter where it's much different just had like too many people like bogging down the servers at once so there was a temporary outage 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 something breaks yeah yeah right and like but you don't know when that's going to happen it's not like or how right it's going to happen in weird ways right but it's like working with a temperamental intern yeah i mean i i i do think that when we say intern we we we're 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 i 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 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.
1:50:41They 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 going to use that one. That app works better on my phone. And I have no loyalty to the direct product to be commoditized. it in effect yeah i don't either fungible 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 i got nothing but time well and i i'm not doing this for a living i mean i've built i've built a ton of different skills and tools that 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 named Kevin, right?
1:51:36I 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 isn't that like a normal employee? I mean, doesn't that happen in the, with humans? 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 don't, and you just, you don't bring on people that, you know, you're going to 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 going to be a new model release or an upgrade, you know, Tuesday.
1:52:09Well, 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 Chachapiti or a Claude. 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 the today there is very important because it really does change that often.
1:52:44And what we found is like right now, Codex 5.5 or ChatGPT 5.5, OpenAI's ChatGPT model 5.5 is very, very good for coding. And so, but we also use Opus 4.7 for reviewing. And so the thing is, is this kind of using the diversity of models means that you can do things 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 through the planning, expecting the free model to do it. So instead of spending a$600 or$1 ,000 for an entire project of raw API costs, you're spending maybe$50 on the planning and then a slower but much freer version of the doing.
1:53:30And so we really, really rely on this, which has actually made me have a much more, I don't know, more, 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's in 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. Now 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.
1:54:05But it is very interesting. And I do think that these model companies are struggling to find relevancy. They're just trying to build products that are going to be outside of tokens. Right. 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.
1:54:37We 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 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. But it's an interesting point Harper made because I think a lot of nobody, a lot of leaders who have these expectations that a team is going to go fast on AI because everything is automated, don't understand the piece that Harper just explained.
1:55:18It's not, I mean, yeah, it's technically faster than humans typing things, but for any of these systems to be good enough to deploy without breaking down all the time, you need all of those additional steps. And that does take time. You can't, you know, it just does. And I think people don't realize that. And then they have these outsides, you know, these crazy expectations for life. It's also been my goal not to be dependent on any one front, certainly frontier model. Smart. Yeah. I want to develop skills that are generic enough so that tomorrow I can use chat GBT and the day after I can use GLM and the day after that I can use Kimmy or all of the above.
1:56:00I really think more and more people are doing what you're both doing, which is having. do you think so i don't know you don't think so i mean like there's like a company saying we're going to use anthropic that's it yes most of yes because um of compliance like all of the things they can't they're like it's not no fungible for them um or they're using no they tend to hitch their wagon to a company or like palantir which then um palantir switches doesn't it but But you're always using Palantir. Correct. So this is the better route because all of this is so new is to be as versatile as you possibly can.
1:56:42But there's a legal and compliance issue that makes that super, super, super hard. Yeah. Yeah. I think individuals. Yeah. It's also contrary to like how corporate, what is it called? There's a word that 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. but they want a big pro, you know, pre-revenue defined spend that where they can save money, get 20 % 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 I think that we're just not yet at the end.
1:57:17This is a little like, this is a little bit like data centers back in early 2000s, where you would just be like, oh, well, we, you know, we're in 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 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 um the way you guys are talking amy it sounds like you don't really think this is a bubble oh no leo uh so no the amount of money the the bubble thing is tough i don't want to equate the what what's happening right now to the dot-com era because they're they really are very very different um that being said Yeah.
1:58:18I get worried anytime this amount of capital flows into something that everybody has such high hopes for. It's not sustainable. Well, we're also in the building phase. We're not in the coming up with products that will shift things phase. Companies aren't doing that yet. We're not in the revenue phase or profit phase. So a lot of infrastructure has to get stood up, and that's unrecoverable investment that you're not going to make back. What do you think of Jeff Bezos' 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 the late 19th century and the dot-com bust, where you got infrastructure.
1:59:07maybe the companies that build it didn't survive mci's gone uh you know the railroads all went bankrupt but you got the transcontinental railway yeah the fiber 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 the same was true of the early internet era was being built. That was a technology that had been around that didn't start off as a commerce play. It started off as an academic research tool.
1:59:52This is not that. But won't those data centers survive if these companies fail? I mean, if OpenAI 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. We know that quantum, 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.
2:00:28And so It's not the same thing as standing up a steel refinery knowing that you're going to need steel for the next 60 years. It's just not. The other concern that I have is, look, I grew up, Harper's in Chicago. I grew up in northwest Indiana. I am a Hoosier. I grew up south of the steel mills in a very, very blue-collar community. Ground Point? Actually, no, no. Those are the cool spots. I grew up in Cherivelle, which is just next to East Chicago where I was born. And the challenge here is the places in Indiana where some of these data centers are being proposed are the places where the steel mills went.
2:01:14Those 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 going to require hiring. It's just you're not going to hire local Indiana companies. This is just a different look. 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 will temporarily get a handful of jobs and then find that those jobs have gone away and there's nothing else there.
2:01:58Interestingly, 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 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 see the communities that they have been built on. And keep in mind that I'm pretty pro data center. I'm kind of, it's kind of linked to my ability to participate in my company. I know, I am too. I want them to build them. I do think that - We saw what happened to Anthropic when it was compute constrained.
2:02:35Quad went way downhill. That's why it was such a good deal that they were able to find - Two things can be, you can both, look, these data centers aren't going to get built overnight either. Well, that's the other thing. They don't exist, 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 water and power yeah water and power um right now and that is true but there 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 an investment of some kind back 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 look and you know what's crazy you ask people the same thing about nuclear power stations and they don't care right if you go back in time right i mean so one gives us power one sucks up power so this is 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 because we've never had to 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 natural gas unless you mitigate that through a plan so like any community yes any community where these data centers where there's a the green light is being given, there should be a long-term plan.
2:04:07And some of those resources should come back into the community. Once those data centers go online, they should contribute X percent back. I mean, you know, you can have, 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 how do we manage this over a longer period of time? Do you know that right now, sorry, I'm 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.
2:04:43You know that the power demands on the East Coast right now are significantly like orders of magnitude greater than they ever have been before during a time when there's already a drought and we are hitting what's going to be one of the hottest seasons, hottest summers, where nobody is thinking this through. Why is power consumption so high? Is it because of data centers? Yes, it's a company. Well, yeah. Well, then there's a that's you can see why people might not be happy about data centers. So again, there are some other ways. It's just planning. It's planning and also putting these in places.
2:05:18It can be mitigated is what you're saying, but you've got to think about that. 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. That's all. Meanwhile, Google is just taking 6.1 % of SpaceX and they're talking about launching data centers in space together. Well, of course. Well, at least the neighborhood won't mind. there are some big technical issues associated 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 on that has plans to launch prototypes with tiny data centers in 2027 just to see if they can you know what it takes and then scale from there so a 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 so i was just there a couple weeks ago um mainly because of petra so they i was she was there and i was her 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 so she's maybe i uh i would like to say that her she no probably not her math is is good but not that good um we're looking at doesn't matter if anybody out there is on the admissions committee from princeton three years from now please look for an application for my lovely school um does she have a sport is she because 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 eagle scout and she's going to the university of michigan this summer for an architecture program um she applied to a traditional urban planning and architecture program and told them she wants to do that but off planet so they let her in she's i love it off planet and so she's gonna be doing my only advice would be for her to get a tiktok account and start making sandwiches i'm just saying it for me that's all in this household we are not we are not a tiktok household leo you should know that i do But the space stuff, so there is so much commercialization happening, mainly in manufacturing and pharma over the next couple of years.
2:08:11Interesting. That from my point of view, the SpaceX Google thing is less about data centers and more about just getting more infrastructure. SpaceX wants to get more infrastructure into space and Google. It would be good for them to position up there as all of this other stuff is being built. so okay it seems like a you know this is what you were talking about which is have a lot of irons in the fire yeah 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.blog is his blog it's a great read and 2389.ai is his company.
2:08:54Amy Webb from the Future Today Strategic Strategy Group, FTSG.com. If you are trying to figure out what the future holds for your business, this is a very, very good person to call, to talk to. And since you don't do the trend report, can people subscribe to some sort of reporting? 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.
2:09:43So it's, uh, it's actually a really, really, really useful tool. Um, 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 to come in to cheer everybody up. I wish I'd seen that South by a 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. I'll send you a copy of it, Leo, but we turned the entire thing into a funeral. So if you were waiting on the line, they were, everybody had black on, you were greeted with a pack of tissues and a pin, but nobody said why.
2:10:18there were funeral flower arrangements um and so i opened sad organ music playing we did no it wasn't sad organ music we had sad like sarah mclaughlin i was about to say angel on loop right angel and didn't tell anybody what was happening and then i i came out in a black cloak floor length and said 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. And then we had an memoriam video of trend report throughout his life. And then I ripped off the case. And then I said, look, we're at this moment of transition.
2:11:00The world is changing. Here's why we're burying it. And then I ripped off my cloak and had a sparkly outfit underneath and the Texas Longhorn marching band came in, which was insane. And it was a little party and then everybody settled back in and we got down to business. Fantastic. Love it. Love it. It was, it was a marching band. Yeah. Anytime you can have a marching band. And yeah, they were amazing. And the students were cool. And for a hot minute, I played quads in, in our marching band. So there was not. You're a clarinetist. And what is a quad? Is that a drum? I played all woodwinds, but my clarinet teacher at that point didn't want me on the field because I was on my way to music school.
2:11:43So they let me play quads. Is that where it's hanging around your neck and you're going boom? Yeah. And then like my body just doesn't, was not meant for that instrument. So 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 got to be hard. Yeah, no, I'm just, harnesses were not meant for women of a certain body type. So, yeah. It can drag you right into the field. We will have more with these two very accomplished people in just a moment. You're watching This Week in Tech, brought to you by Box. Now, this is interesting.
2:12:17Box has really got a niche now. They're the leading intelligent, this is so good, intelligent content management platform for enterprises. The 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, your content. Your business isn't the sum of, you know, internet knowledge in general. Your business lives in your business's content, right? Now, we all know AI is not some far-off idea anymore. It's dynamic innovation that's happening right now. It's here to stay. it's already having a profound impact on the world so 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 97 000 companies and get this 68 of the fortune 50.
2:13:17many companies are underestimating ai's transformative potential rather than reimagining their operations 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 going to love this this is what amy was talking about box ai works with all the major leading ai model providers like open ai 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 for 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 compliance and data governance over 115 000 enterprise trust you need a context layer that gives ai the context it needs while giving the teams the flexibility they need to test and leverage various models for different use cases just what we were talking about here this is the way to do it visit box.com ai to learn more that's box.com ai box.com ai we thank them so much for their support of this week in tech that's a really interesting example of a company it's not a pivot but a company that's leveraged its core business in a way that makes perfect sense in a new modern context i think it's a really a good example of what about what you can do if you're flexible it's a different kind of pivot than all birds 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 But it's like the Long Island iced tea and blockchain company.
2:16:04That's just a stock market ruse, isn't it? They're not going to make slippers anymore, I'm sad to say. Supply chains are hard. AI is easy. Is that sad? It's really because it's made in China and there's tariffs and supply chain issues and stuff like that. I don't know. Is it? I don't know. I don't know what the unit cost to make an Allbird sneaker is. They're ridiculously expensive. That's the thing that's so frustrating. I mean, this shouldn't cost$150. Yeah. But it does. I don't know why. I don't believe in them. Like, you don't think they exist? Nope, never seen one.
2:16:47Well, if you've never seen one, you probably will never see one now. So, there you go. They're like angels. They're going back home. Open AI has created a new$10 billion. open ai it's so funny because at one moment they're saying we're gonna 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 10 billion dollar company to help businesses deploy ai well it ain't sora i guess um well i mean i i was asking 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.
2:17:36Because 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 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 you're open AI, you want to do enablement, which is much less about building a product with a capital P.
2:18:13It's much more about 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 plaid and Google and open AI. And then it knows everything. But the point being is that you can kind of do this with, with a lot of things. And so with health too, right? That's what health is. All the, all the people that are saying that, you know, this is the end of SaaS.
2:18:46I think that's sort of true, sort of not true. It's a little hyperbolic, but it is very clear that we have been cutting our SaaS costs internally by just building new versions of our own 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 then they also say, but we are unique. week, we have these specific requirements.
2:19:19How 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 started 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 want to 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 11.
2:19:49They want everyone to use it. And so, but not in this kind of - To the point where they start a metric that says, well, how many tokens did you use today? And of course, whenever you have a metric to measure something, it's definitely bad. We're seeing people at Amazon just 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 24-year-old Amazon engineer who has to figure out how to game the token it's just like yeah and i have it it's reviewing my reviews the reviews are reviewing the reviews of the reviews it's just i can think of lots of ways to use tokens give me an unlimited supply i'll use them so i guess my my kind of macro point here is that i think people are going to be using the of 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 BCG or OpenAI or whomever, small companies, big companies, the thing is, is everyone wants to use it.
2:20:49They all see opportunities, whether it's, I hope it's not cost cutting. I hope it's actually just doing more. 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 layer to help the businesses match with the companies.
2:21:36OpenAI 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 McKinsey years ago, before all of this, was already hiring tons of engineers for the purpose of trying to harness more of those dollars. So if you think of this like stack with like Accenture at the bottom, which is enablement and McKinsey at the top of it doing more sort of strategy and planning and thought leadership or whatever. There's money to be made by going down by capturing more of 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.
2:22:24Accenture had been trying to push upward, which was what a lot of the acquisitions were about to do more of the consulting work. And OpenAI, I think very smartly was like, forget these dudes, we're going to go and do our own thing. For the purpose of not just selling more, but being able to direct future moat. I mean, once a business gets locked in, they're not gonna be able to get back out. So and why? Why were you know, BCG is going to move a floor of people into your business, you were going to get charged exorbitantly 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.
2:23:08So this is this was a smart move. Harper is correct, and that people don't quite know what to do yet. My concern would be you're there's no buffer now. So you better make sure that somebody internally 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 sass was all about was companies 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 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 technology enablement firm a voting firm and a law firm yeah right so it is what individuals are doing that's what i think is kind of interesting i mean this is what i mean i see myself and i see a lot of people writing our own personal so a small to medium same enterprise.
2:24:14This is, if you are somebody who's willing to experiment, this is going to be in a wonderful era for you. It could be a superpower. Totally. You're going to have access to be able to do stuff as long as you, yeah. It is a superpower. And that's why we're seeing so many businesses that are like, I'm making 10 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 normally would take a huge, huge, huge team. 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, the law firm, et cetera.
2:24:51And so having to do 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 ventured 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 did a lot of AI? And you hire McKinsey. I mean, is that the alternative? 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.
2:25:28There'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 and an ai firm so accenture will not his i don't know what it is now historically accenture would not work for you if you were not a global 2000 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 nope mckenzie nobody's going to bother with them they're too small yeah um and it was 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 that 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 like an awesome couple of years ahead of you in a way that you would not have had before How does Mythos change things?
2:26:29Now it's not just Mythos. Well, at least people are talking about cybersecurity. I feel like nobody wanted to talk about security until something bad happened. And then it was a lot of navel gazing. So I'm hoping that CISOs out there listening are all using the Mythos moment as a way to get yourselves more, get more funding and get more people and be able to convince your superiors that like yeah cyber security is kind of like a big thing there's a couple funny things i think about the mythos hype which is certainly it's probably very good like i don't have any experience 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 it wasn't trained to find flaws and they've given it to 50 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 the bad guys are going to find the flaws now there's some evidence that perhaps this is more about compute or marketing nevertheless open ai has something similar microsoft just announced something similar called m dash and 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 that Mythos was announced right around the same time that Hegseth was going directly after Anthropic.
2:27:51Right. And so this is not just, it's not, these things are not announced in a vacuum. It was a political move as well. There is a political move as well. And I think that, you know, if, if you say to the U S government, specifically the Department of War, we have a model that is going to kill everyone. The people who want to kill everyone are going to be very interested in that model. So this is a very smart process. Right. Exactly. And so the other thing I think to really remember is these models are very good. And when I say these models, I say all of the models. GPT 5.5, if you look at the list of things that Mythos apparently did, GPT 5.5 also hit very, very high on a lot of those evals, a lot of those benchmarks.
2:28:31And 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. But you can also use it negatively as a dark side hacker, whatever you call them. And I think the reality is we can't unring this bell. This is already out there. And so Anthropic is taking advantage of this. We talked about the doomerism always sells. This is another flavor of that, of allowing for Anthropic's like, oh my gosh, everything, we're all going to die.
2:29:07And everyone's like, oh, great. I'm going to adopt you because Chachupiti's over here saying, we are not going to die. And, you know, not that I want to kill everyone, but I want the model that's so good that it's going to kill everyone. You know, and it's a very simple thing. I think it's actually going to hurt them in the long run because of what we saw with Hegseth and the government and whatnot. I think that there's going to be some downsides. But I also think Anthropic is addicted to this. I think this is one of them. They love this stuff. They love this. Every single model they release, they're like, we can't release it.
2:29:35This is going to kill everyone immediately. I was talking to a friend. later they're like it's available for all the consumers yeah but don't forget this was the playbook of open ai in 2022 or whatever yeah they said chat gpt 3 was too by the way that was when dario was at open ai it was dario's plan he said there's too dangerous to release chat gpt 3 which wasn't that good uh or maybe it was 2.5 it was a very early not so hot version of chat gpt regardless i do think this is a plan and it is a it is a real it's a real thing that they are 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 evals 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 because i because i think there wasn't there some leaked pricing on mythos that was just like 10 billion dollars or some insane amount of money that also raises is another issue of the haves and have nots.
2:30:34Like 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 billionaires can use real good AI. That puts the rest of us at a massive disadvantage, but it's not that you don't think that's going to be the case, Harper. 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, I mean, income inequality is a big problem. Yeah. Listen, you can talk housing. There are literally no more houses in Chicago. Like Chicago is not a small populated space.
2:31:12I know 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 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. It's going to 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.
2:31:43Them not distributing it to people is I have some complicated feelings on that. So I don't necessarily know yet. 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 definitely problematic. There is a group of our club members in our club to a discord who have an over under bet going on the length of this show. Is it too late for me to kind of depauly market? Yeah. You know, I just I just was going to start listening to Lex Friedman's interview with David Henneminer Hanson.
2:32:25And 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 I know maybe there's something wrong with my podcast catcher but that seems like a little bit longer than our show this show will not be six hours long in fact to that end let me do the last commercial we could start to wrap things up uh but 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're just this is good i think we covered i think actually we covered a lot of it though we did pretty well yeah i'm kind of throwing the stories in bit by bit yeah more natural that way it is it's more natural this is a natural show uh 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 because one thing is almost certain your competitors are already making their move well no more waiting with netsuite by oracle you can put ai to work today netsuite is the number one ai cloud erp trusted by over 43 000 businesses And here's the strength.
2:33:48It's a unified suite that brings your financials, inventory, commerce, HR, and CRM into a single source of truth. And that connected data is what makes your AI smarter. So it doesn't just guess, it knows. It 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. NetSuite delivers a customized solution for your business. And this is not another bolted on tool.
2:34:32It's AI built into the system that runs your business. Whether your company earns millions or even hundreds of millions, NetSuite helps you stay ahead of the pack. So 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 netsuite.com slash twit. Netsuite.com slash twit. N-E-T-S-U-I-T-E dot com slash T-W-I-T. And we thank NetSuite so much for supporting This Week in Tech. it's kind of interesting how many of our advertisers are basically ai these days ai been very very good to podcasting used to be 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 have gotten maybe a couple hundred emails from people i'm very well known for having a bitcoin wallet with 7.85 bitcoin in it that i made when bitcoin first happened and thinking it wasn't worth anything forgot my password and 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 40 gigabytes or something small it's huge and you still have the 7.5 once again yes i do that just to see if the money's still there because one never knows and he says yes you have 7.85 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 about a guy who changed his password on his bitcoin wallet when he was stoned 11 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 got to try this but i have to say there were certain extenuating circumstances he did tweet this guy you have to get stoned first restraint stone and i missed it this guy is cpkrm on uh more bits and stuff on uh x now i haven't verified this tom's hardware maybe he has he says holy poop omg claude just cracked this thank you anthropic thank you dario naming my kid after you uh apparently claude tried 13 what was it 13 million 13 billion 13 trillion i think it was trillion passwords no i'm sorry 3.5 trillion passwords but it wasn't pure brute 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 uh using btc recover which is a recovery tool to 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 didn't really crack that's right just like read a bunch of stuff and used probabilistic like that's right yeah that's why it's not going to work for me because i 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 i will have claude work on it and chat gpt during that time period i had i used 45 bitcoins to buy a kindle i read a lot of good books that is a lot that's a very expensive kindle that's millions yeah it is it was it was it was good wait a minute let me just see what bitcoin's worth right now just to make you feel really bad what's 45 times 77 000.
2:38:55was it a public art project no this is 2011. i had a whole bunch of bitcoins and i then i then used 10 to buy a leica camera and then 10 to buy a Leica camera lens. The 45 were$3.5 million today. Yeah, I didn't misunderstand. I believed in it. Like Benito was getting it. Yeah, this is what they told us to do with it. Yeah, I was like - There's a guy who bought a pizza with 10 ,000 Bitcoin. I was like, this is beautiful. I have money I generated from math problems at my office did while I was gone. And I got a Kindle. And then I got a camera. I took the camera to my local favorite Leica guy, to mark in great great spot and he was like no you cannot buy like 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 7.85 bitcoin still yeah well you only have it because i couldn't get it that's true it's a good savings account it's a saving that's what i'm saying it's it's retirement by the time quantum computing comes along you have the wallet backed up though with theoretically oh So you could probably find someone in your audience that would buy it for a fraction of it, get some liquidity from it.
2:40:11The offer is out there. A mere 10 % of its value, and it's yours. I think, I mean, I love these stories because when Bitcoin first shot up, I had all these friends, because I used to send them to everyone. I'd just be like, hey, check out this crazy cash. Go download this software and I'll send you Bitcoin. that's why i have 7.85 is because uh i had a tip jar and people put bitcoin into it yeah and it was just like a thing like it was no big deal but there was a certain point maybe 2021 2022 whenever 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 to answer that question like i don't i sent it to you and it's gone um but a few of them logged into coinbase for the first time in 10 years and found you know a couple hundred grand it was pretty exciting for a few people um also a story and a new website and yeah 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 read it yeah well and somebody says AI generator I don't know but he is uh canna or she has uh taken a bunch of industries including tools uh restaurant prices and backpacks and talked about how basically private equity is the villain here has ruined checks out these companies vf corporation which started as vanity fair mills they made bras and underwear bought a company called bluebell then picked up jansport then the north face then they bought east pack they bought Kipling, they bought Eagle Creek.
2:41:52If you are a hiker, a camper, a bicycler, you probably know some of those names. These are well-known backpack companies and basically trashed them all, made them crap. Actually with Jansport, 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 Jansport and Eastpac were ubiquitous in high schools across America in the 90s. Yeah. Yeah. Well, the thesis on the private equity is right. The syntax is definitely all the hallmarks of chat. Yeah. There's a lot of M dashes. There's 32 M dashes in this article, dude. I know you didn't write it.
2:42:37Can 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 em dashes. It's a crutch. It's a crutch. What? And then the first book that I, it is. Strunk and white says it's very strong to use a dash. I neither think nor speak in normal complete sentences. I'm just a giant run on, which, you know, anyways. So, and like my first book was also just full of just em 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 mdashes and now like i have to convince people like no this is just the way that i i write but amy has been using the word delve for years no mdash i and keen i have my computer set up so when i type two two hyphens it turns it into an mdash that's how nobody even knew what an mdash was until all of this so anyhow 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 uh 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 uh 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, you know, make new backpacks.
2:44:20I 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 thing to show that. And so there were dozens and dozens of new within the last five years that were basically copying what was good about Jan sort of the past and making it new. um I do think that you know private equity has not done a lot of benefit I have a friend though I have a longtime friend who is a PE guy he's a small private equity company he was trying to buy Steinway for a while oh wow which is a family run piano company because it was not the family was getting tired of it they one of his specialties was buying companies that the family were family run but the family wanted out right and he is a good manager and he said i can take these companies without and them and turn them into what they should be yeah i don't think he ended up getting steinway but but that's wonderful if that's that's great but yeah so is it always bad yeah yeah 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 i know a few of those yeah yeah they're quite a few of those in fact what is that that is some sort of that is that is a tale as old as time as old as time i love those tails um don't judge me but i did ask so we're voting in california in the primary in June 2nd.
2:46:00And I think there are 60 people running for governor. I'm not joking. That governor's race is crazy. 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. It's confusing as hell. I did ask my chat bot to help me with a ballot guide. Yeah, that sounds like a great use of it. I told it what I care about. I gave it my values. And I said, please go through all the voter guides and get back to me. Now, I'm not going to vote exact. Actually, I may. It did tell me how to vote. And I may or may not vote that way. But it sure helped me cut through 60 candidates.
2:46:46One of my favorite parts about the LLMs is now you have this voter guide. You can paste the voter guide into Chachapi Tea and say, make me a recipe for dinner tonight based on this voter guide. And it would just be like, absolutely. Let's go. That's a great idea. I'm so happy to have dinner. 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. 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 so you're just like you that was a jump i appreciate what you did and i really like that you did this for me and 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 i recommended to vote but now i'm thinking i wonder what i should have for dinner well put in your voters guide and let's see what it says Based on my voter guide, what should I get?
2:47:51Okay, so the voter guide came from chatGBT, I think. So I should use maybe Grok for this. Yeah, if you want to. Go crazy? Yeah, go wild. Or maybe I'll do Anthropic. Because I think Anthropic doesn't know. Well, I probably knows food better than a lot of the others. Yeah, you said you always find out what the best one is. Who's the best? Well, right now I'm mad at Claude because it made every website look exactly the same. they do all look the same that was all the same that was a thing everything is this beige kind of brown with a nice warm rich reddish color everything has this like vibe right now so i'm mad at fonts have serifs which i really hate so i'm i have an emotional uh argument with uh chat gpt uh or with with anthropic but i use a lot of chat gpt because i'm of the conclusion that chat gpt'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 in many other models and then when i'm ready for the final final approach so to speak i'll drop it in the chat gpt 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 so but i want to know what it recommended for dinner with your uh with your vote what are you having are you are you suddenly having a vegan dinner is it austerity based signed to help me what should i say help me plan dinner just based on the voter guide what should i have for dinner okay that's what i'm doing please use this website tell me plan dinner it's retrieving the website now it'll be interesting if it returns instructions on the dinner yeah go to mom's market the url doesn't have anything dinner related on it it's your that's the best this is the thing is that like the 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 robux no i could see it no yeah okay so you know, visit this website.
2:50:08These are my preferences plan. Uh, what should I eat for dinner? Use this to help me like, like really instruct it. If it throws a fit, just be like, do it. Okay. Information about my preferences. What should I eat for dinner? I think it was progressive voter. It was trying to help you though, because it does not make sense. It really doesn't though i just read the whole page it's candidate recommendations from the u.s house down to county clerk recorder says there's no food no diet no preferences i'm not going to invent a connection that isn't there so it knows me right so it's trying keeps trying to give me a low carb that's a that's an interesting response claude knows me too well that's the right chat gpt now do you think chat gpt will do it chat gpt doesn't always scrape well though i think claude does a better job chat gpd made this website and it's on it's on uh cloudflare what you don't like this i love this i found that a response to be interesting that was not what i was smarter than you thought it's not it's not going to do that well it just means that you've probably yeah yeah i think you've just instructed that instance to not do it do stuff like that it's 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 bosses 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'll say my wife will divorce me if i don't have a good plan for dinner.
2:51:54Based on this guide. Based on this guide. It has to be based on this guide. You have to be specific. This, in caps, this guide. Oh, wait a minute. I forgot to put the guide in. Wait a minute. There it is. It'll feel like great. Got it. No problem. Okay. We got this. Okay. It's thinking. It's doing pro thinking. I have max subscriptions for all of these. So, you know, I'm giving it all my money. i'll turn the guide into a practical dinner plan yes 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 open ai will uh chat gpt will always whore itself out no matter what exactly exactly arbor reed is basically loki uh and i'm gonna call amy webb athena here because she's of ever wise i'm a i'm a cassandra leo cassandra even better knowing the future and can't get anybody she's been huffing the natural gas and she's ready to tell you the future no that's not cassandra that's the uh oracle of delphi is the one yeah the center is the one that does the same thing but nobody listens that's right that's right uh you guys are fantastic what a 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 this time but staying here to do the show yeah 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 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 In fact, today he's not here because he's at his quarterly geek dinner.
2:53:43So that group of super geeks. I love those guys. Is where he is tonight. I've been trying to get into the quarterly geek dinner for years. I'm sure they all listen. So they would very much. Okay, well tell them I will fly out. Okay. Okay, so here we go. ChatGPT says, make it a Sonoma spring dinner with marinated artichokes, olives, good bread and goat cheese or white bean spread. or a chiete or linguine with a, oh, this sounds fantastic, an arugula 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.
2:54:20And in fact, it even knows my local market and is telling me what aisles to use. There you go. Look at that. By the way, cooking plan, put the strawberries in a bowl with a little sugar, lemon zest and lemon juice right now because it's going to take you a while to make dinner. Don't make it a civics lecture. let the meal embody the voters guy don't narrate the whole ideology don't overcook the asparagus don't overcook the asparagus and if this is for monday may 18th don't rely on farmer's market shop local retail tonight yeah i feel like that feels that makes sense don't make it too virtuous and have a plan b this is great great information and 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 it did it perfect it did exactly what 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 is what you want but well this is one of the arguments right boring i know but this is how it makes the money i personally like to be glazed harper i know i know i shouldn't say that harper reed is at 2389.ai great some great uh skills there and you're going to have more soon 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 Right now, we just have a lot.
2:55:57We're just heads down on a lot of stuff. We'll be doing, I think we're going to 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 going to ask Anthropic for a dinner menu for a heads down spring. Yeah. Yeah, it'll be like peanut butter. But don't overcook the asparagus. Whatever you do. Amy Harper, love you. Thank you so much for being here.
2:56:32Really appreciate it. Tell Brian I'm coming to dinner. I will. They would love to have you. I will love to be there. We do this show every Sunday, 2 to 5 p.m. Pacific, 5 to 8 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 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.tv 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 The minute Benito and Kevin finish polishing it up and taking in all the swear words.
2:57:22If you're not a member of the club, please do join the club. Tuesday, a little programming note, we will be covering Google I.O.'s keynote. Because we don't want to get any strikes against us on YouTube, we're very sensitive about that. We are only doing it in the club, 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. note uh we also do WWDC that way it's just unfortunately the way we have to do it uh thanks to the sensitivity of these large multi-billion dollar multi-trillion dollar companies uh thanks everybody for joining us 21 years I've been saying it I'm going to say it one more time thanks for being here we'll see you next week and another twit is in the can
2:58:16Doing the twit. All right.
From the publisher
- Trump and the CEOs go to China
- NVIDIA CEO joins Trump in China despite 'awkward' politics
- US clears H200 chip sales to 10 China firms as Nvidia CEO looks for breakthrough
- Empty Waymos invade Atlanta neighborhood, circle cul-de-sac for hours with no passengers
- The Class of 2026 is cooked
- Chinese AI groups pull ahead of US rivals in video generation race
- Google Weighs Using SpaceX to Launch Orbital Data Centers
- What smart people are saying about OpenAI's new $10 billion company to help businesses deploy AI
- Bitcoin trader recovers $400,000 using Claude AI after getting 'stoned' and losing wallet password 11 years ago — bot tried 3.5 trillion passwords before decrypting an old wallet backup
- Your Mattress Got Worse on Purpose
Host: Leo Laporte
Guests: Harper Reed and Amy Webb
Download or subscribe to This Week in Tech at https://twit.tv/shows/this-week-in-tech
Join Club TWiT for Ad-Free Podcasts!
Support what you love and get ad-free audio and video feeds, a members-only Discord, and exclusive content. Join today: https://twit.tv/clubtwit
Sponsors: