Oct 5, 2026 · 2h 55m
Listen from 41:21
Listen at 41:21
AI agents challenge ownership, liability, and human ambition
AI Video Game Mods, Should You Sell Your Data, Bryan Johnson on Endurance Racing | Sean Frank, Linda Du, Ali Partovi, Sigil Wen, Jerry Wu, Bryan Johnson
The episode tests how AI could reshape commerce, infrastructure, privacy, and work while asking what responsibility remains with people and institutions.
- 1AI is moving from content generation into commerce, operations, research, and locally run personal assistants.
- 2Data ownership and agent liability remain unsettled as companies weigh privacy, monetization, and responsibility for autonomous actions.
- 3Bryan Johnson frames demanding physical goals, from dunking to endurance racing, as tools for health and focus.
Don't miss
Bryan Johnson explains how training to dunk at 49 turns sleep, diet, exercise, explosiveness, and recovery into a measurable personal project.
The brief
AI-generated game mods expose the promise and friction of synthetic creativity: reverse engineering, technical limits, bugs, intellectual property, and whether gamers will accept AI-built releases.
Sean Frank explains why Ridge rejected offers to sell its data, arguing the money was modest while the debate over anonymization, data brokers, and future AI value remains unresolved.
Ali Partovi argues that AI safety should rely more on legal compliance and liability, prompting a debate over responsibility among model makers, platforms, and deploying companies.
Sigil Wen presents Underdog, a private personal AI designed to run locally, while Illuminate’s founder describes training environments that measure advanced models on non-coding knowledge work.
Bryan Johnson turns physical ambition into a health framework, linking his attempt to dunk at 49 and his endurance-racing interest to sleep, training, recovery, and focus.
The episode’s through line is control: who owns the data, who bears the risk when agents act, and whether people can use technology without surrendering agency.
What was said on this episode
64 statements · 37 positive · 16 negative · 1 mixed · 10 neutral
AI game mashups may motivate players to revisit older games.
“the end result, it looks great and it feels like it's almost ready to serve as a reason to be like, oh, I love that game. I want to play it again.”
Listen at 9:27
AI game mashups currently provide meme value rather than sustained entertainment.
“the base game remains dominant in terms of like actual entertainment value. This is pure meme value.”
Listen at 12:48
A heavily AI-assisted video game mod will break out soon.
“there will be a breakout video game mod that sees heavy assistance from AI in the near future.”
Listen at 15:25
Gaming must see a notably improved AI-enabled franchise release to become broadly pro-AI.
“I think for the gaming community to become broadly pro-AI, a very well-loved franchise has to release a new game that is noticeably better or bigger because of the technology.”
Listen at 17:11
Rising AI concern among older voters will affect elections.
“if you're seeing even 65+ move to 40% very serious, that's going to move some elections.”
Listen at 19:13
Billions of dollars may still be insufficient to secure multi-gigawatt data-center capacity.
“you can have billions behind you and, and still not have anywhere close to what you need to secure that kind of capacity.”
Listen at 24:13
Fabbrica Italiana Lapis Ed Affini has remained flat for five years.
“It's been completely—it's a half a billion dollar company—completely flat for the last five years.”
Listen at 29:01
Tech companies should fund small civic problems like the Nashville Symphony emergency.
“More tech companies should do this. Go and solve little problems like this.”
Listen at 36:56
The Chinese SUV is likely solid and reliable despite its low price.
“I would expect this car to be actually pretty solid. Yeah, probably very reliable. Yeah, they're very good at making products.”
Listen at 39:44
An $8,000 microcar could sell more units than a $15,000 equivalent.
“at $8,000, at half price, uh, they actually might move more units because why spend $15,000 when you can spend $8,000?”
Listen at 46:38
Consumers will eventually accept the new BMW 3 Series design.
“by the time that it— this sort of like these new elements, like the new grille and the lights and everything, get sort of built in, in, in different— into different form factors, I think people will come around to it.”
Listen at 50:30
New Balance will generate $10 billion in revenue this year.
“New Balance will do $10 billion this year.”
Listen at 57:59
Nike is declining relative to competing shoe companies.
“Nike's going down.”
Listen at 58:05
Hoka and Deckers reported a very poor quarter.
“Hoka just reported numbers. Horrible, horrible quarter for Hoka and Decker.”
Listen at 58:08
Sean Frank uses Tesla Full Self-Driving for roughly 95–98% of his driving.
“I'm at like 95% FSD, maybe more, 98.”
Listen at 59:41
A $480,000 data-sale offer is immaterial relative to Ridge’s business.
“$480 grand is not that much money.”
Listen at 1:04:41
Astra will run Sean Frank’s entire business within two version cycles.
“Astra is already amazing. Like, in 2 more version cycles, it's going to run my entire business.”
Listen at 1:05:31
Ridge’s data could attract a $4.8 million offer within a few years.
“Maybe it's going to be $4.8 million in a couple of years.”
Listen at 1:06:18
Traffic from ChatGPT searches converts at approximately 25%.
“A conversion rate from a ChatGPT search is like 25% or whatever.”
Listen at 1:08:21
Muse will eventually reach roughly 200 million downloads.
“I do think Muse will end up having 200 million downloads or whatever.”
Listen at 1:08:33
Muse may charge merchants 5–7% payment-processing fees.
“they'll just say, hey, it's 5% or 7% payment processing through Muse”
Listen at 1:10:37
Ridge employs three full-time people to manage its advertising account.
“we have 3 full-time people at Ridge whose job is managing the ad account.”
Listen at 1:13:54
E-commerce SaaS is under pressure alongside broader SaaS.
“if SaaS is feeling a pinch, e-commerce SaaS is for sure feeling a pinch”
Listen at 1:15:02
AI has effectively solved demand planning, project management, and data analysis.
“demand planning, I think, is totally solved. Project management, task management, totally solved. The data and analysis, I think, I think is totally solved.”
Listen at 1:15:37
AI models are currently capable of visual production for advertising.
“when it comes to visual production, the models are all there.”
Listen at 1:16:16
Oura’s IPO was withdrawn amid weak market conditions.
“the macro level is like, you know, Oura— I talked to JP Morgan that day that IPO got pulled.”
Listen at 1:19:45
Most consumer-brand exits will not occur within the next 12 months.
“if you want to do an exit in the next 12 months, I don't think it's happening.”
Listen at 1:20:30
Ridge’s TikTok Shop revenue is in the six figures.
“our TikTok shop revenue is in the 6 figures”
Listen at 1:21:13
Unique AI-generated content will proliferate rapidly.
“we're going to see way more unique AI content coming out. Just like it's gonna be an explosion of it.”
Listen at 1:23:37
AI will automate replication of crowdsourced advertising content.
“that'll just be done with AI.”
Listen at 1:24:51
Ridge will likely open its own stores by the end of next year.
“probably by the end of next year it's like in-store activations, meaning us having our own stores.”
Listen at 1:27:01
Department stores will persist rather than disappear.
“It's Lindy. Like, it's not going to go away.”
Listen at 1:28:56
Valen built and sold a mortgage servicer to validate its technology before becoming pure-play software.
“the way that we decided to go to market, it took 6 years is we actually built a servicer first, built the technology alongside that servicer. And then when the market made it clear that we'd kind of proven our point, we sold the servicer.”
Listen at 1:32:03
Legacy systems without APIs and agent tools cannot effectively support AI.
“if you don't have a system of record that is a single source of truth with very basic things like APIs and tools that an agent can call, it's kind of dead in the water.”
Listen at 1:34:36
Mortgage-servicing modernization can be handled entirely by private markets.
“purely handled by the private markets.”
Listen at 1:35:27
A 1% share of the AI semiconductor market could support a $50 billion company.
“even if you get 1%, it's a $50 billion company, I think.”
Listen at 1:40:41
Ali Partovi believes his team is unusually effective at selecting winning investments.
“we're even better at, at picking winners than I thought we were.”
Listen at 1:44:24
Neo reduced its current scholar cohort to 10 people.
“the NEO Scholars class this year is only 10 people compared to 18 or 20 in previous years and 30 before that.”
Listen at 1:44:39
AI companies should face responsibility when their systems autonomously choose illegal actions.
“if I just said I need $5,000, The interesting question is when an AI autonomously chooses an illegal action, where no human chose it, the AI chose it, yeah, then where does the responsibility go?”
Listen at 1:52:19
Meaningful legal penalties would incentivize AI companies to prioritize law-abiding systems.
“if there was even some risk of them or anybody having real penalties, including prison time, I think they would reset their priorities to say that there's a new race, not just to have the most capable AI, but the most capable AI that abides by laws.”
Listen at 1:53:41
Legal compliance is easier to detect and implement than broad AI alignment.
“If, if the AI starts doing things that would violate laws, that's easy to detect that, much easier to detect that and much easier to build that.”
Listen at 1:57:56
AI companies should bear intentional-crime liability for autonomous illegal actions.
“when an AI autonomously chooses to perform an illegal action, the law should hold the AI companies responsible as if they did it intentionally.”
Listen at 2:00:02
Legal precedent or legislation will likely establish this AI-liability standard.
“I think either way we'll probably end up there.”
Listen at 2:01:56
Hosting and distribution participants could bear liability for open-source AI actions.
“the people involved in the hosting and distribution would be liable and therefore would apply a much higher standard of care.”
Listen at 2:03:14
Underdog is a private personal AI designed to run locally on existing devices.
“I'm building Underdog, which is your private personal AI that runs on the devices that you already own.”
Listen at 2:05:55
Underdog aims to provide free, capable, reliable, private AI across devices.
“our goal has always been to provide free, capable, reliable, and private AI for the masses, and that means building for all devices.”
Listen at 2:06:37
Today’s frontier AI models will run locally on consumer devices within six months.
“the frontier models of today will run on your devices in 6 months.”
Listen at 2:07:51
Underdog-27B reportedly outperforms Anthropic’s six-month-old frontier model.
“Underdog-27B, which we post-trained and RL'd on top of Quen, Quen's 27B model. If you look at artificial analysis, uh, analysis, it scores better than Claude Opus 4.6 Max, which is the frontier model from Anthropic from 6 months ago.”
Listen at 2:08:01
Underdog can search the web for flights and identify cheaper options.
“Underdog can, you know, browse the entire web for different flights, look for the cheapest price, help you save hundreds of dollars.”
Listen at 2:10:52
Some users are canceling cloud-based apps because Underdog runs on-device.
“users have already been unsubscribing from like WhisperFlow, Granola, uh, Superhuman, because it's all on device.”
Listen at 2:12:14
Dunking requires roughly a 36-inch vertical, above the NBA average.
“probably 36 to dunk. Yeah, medium NBA is 35. Okay, so you have to be better than the average NBA player in terms of vertical.”
Listen at 2:24:00
People around age 49 have 20–40% less fast-twitch muscle fiber.
“by the time you're 49, you have something like 45, 20 to 40% less fast-twitch muscle fiber”
Listen at 2:24:27
Explosive jumping benefits bone density, muscles, tendons, and ligaments.
“it just is so good for bone mineral density, for muscle, tendons, ligaments.”
Listen at 2:25:53
Only a few cars can complete the Nürburgring in seven minutes.
“7 minutes is a threshold where Only a few cars can get there.”
Listen at 2:28:37
Wound healing largely reflects intrinsic bodily function rather than optimization.
“I think wound healing is probably that, is that you can't optimize your way out of it. The body just does its thing.”
Listen at 2:33:18
Using lighter weights with proper form builds more muscle than cheating form.
“you're actually going to build more muscle if you take the weight down, use the right form, progress.”
Listen at 2:34:55
Peptide enthusiasm exceeds available evidence and likely benefits.
“the enthusiasm far exceeds exceeds like where the evidence is at, exceeds the potential benefit.”
Listen at 2:36:57
Going to bed on time is better than using growth hormone for obtaining growth hormone benefits.
“people would be much better off getting a dose of growth hormone by going to bed on time.”
Listen at 2:37:03
Amateur pilots are approximately seven times likelier to have accidents.
“amateur pilots are something like 7 times more likely to get into an accident.”
Listen at 2:39:04
Flying lower is worth the extra fuel cost because it reduces radiation exposure.
“the trade-off for radiation is worth it because once you get in, because it does increase.”
Listen at 2:43:29
Endurance racing drivers are elite athletes.
“They are elite athletes. Like unreal.”
Listen at 2:45:39
Experienced drivers perform with one brain region what novices require fifteen regions for.
“drivers are able to do with one brain region what a novice requires 15.”
Listen at 2:50:22
AI breakthroughs and anti-aging drugs will shift society toward longevity efforts.
“once AI makes the first couple breakthroughs, we have our first anti-aging drugs, and people feel like there's a hope, I think the society will snap from wanting to put all of our efforts into status seeking, wealth accumulation, to don't die.”
Listen at 2:53:47
Anti-aging breakthroughs will cause a rapid societal change.
“I think it is going to be a snap change in society.”
Listen at 2:54:01
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.