Intelligent Machines 878: The 1AM Bus to Chinatown

Summary

---

Intelligent Machines 878: "The 1AM Bus to Chinatown"

Date: July 9, 2026
Host: Leo Laporte
Co-hosts: Paris Martineau, Jeff Jarvis
Special Guest: Dr. Ian Bogost

---

Episode Overview

This episode reunites the regular trio—Leo Laporte, Paris Martineau (Consumer Reports), and Jeff Jarvis (Craig Newmark School of Journalism)—with special guest Dr. Ian Bogost, game designer, professor, and Atlantic columnist. The main theme is navigating life and technology by focusing on “The Small Stuff”—the subject of Bogost’s new book. The show explores how AI is changing education, creativity, and daily life, while also contemplating the pleasures found in everyday physical experiences. Sidelights include travel tales (including Paris’s “1AM bus to Chinatown” ordeal), debates about AI’s memory/utility, the impacts of AI crawlers on publishing, and the changing landscape of food and processed snacks.

---

Key Discussion Points & Insights

1. Welcoming Back Paris & Introducing Dr. Ian Bogost

• Paris returns from a vacation in Montana and shares stories about her road trip, buffalo (bison?), pig races, and the wonders of the American West.
• Dr. Ian Bogost rejoins the podcast to discuss his new book, The Small Stuff, a meditation on reclaiming pleasure in the mundane, sensory experiences of everyday life—"lead a more gratifying life" [05:09].

---

2. The Small Stuff: Finding Meaning in Materiality

Main themes from Ian Bogost’s book

• “We’ve become disconnected from the physical world, but we can reclaim sensory enchantment.” [05:09, Bogost]

• The book urges readers to savor the material world—doorknobs, toasters, chairs, and even the faint texture of a book’s cover:

  • “Allow yourself to linger just a little bit longer in any sensory encounter. If it starts to feel strange...that’s good, that means you’re doing it right.” [36:10, Bogost]

• Debunking 'Friction' as Prerequisite for Gratification:

  • Bogost argues satisfaction is available in ease, not just struggle.
  • “Unlike happiness, [gratification] is easy. It’s always available to you.” [38:22, Bogost]

• Balance is possible: both long-term achievement and savoring each sense-filled moment can (and should) coexist.

  • “Why couldn’t we have big, long-term goals...and also live deeply and richly inside of every moment?” [25:18, Bogost]

---

3. AI, Education & Integrity: A Candid Conversation

• Paris brings up a striking Atlantic cover story about the decline of reading for pleasure (“down by 43% over 20 years” [11:41]) and rampant AI-enabled cheating in higher education.

  • Example: Brown University’s “shocking” in-class vs. take-home exam results (see [135:53])—massive drops in scores and student dropouts when switching to in-person testing after an AI-assisted, take-home midterm.

• On Updating Teaching for AI:

  • “We professors are busy and lazy and put upon” [13:57, Bogost]—syllabi are often slow to adapt, but AI now forces real-time adjustment.
  • Professors recognize distinguishing student learning (for its own sake) from “credential-collecting” as especially vital amidst AI disruptions.

• Empathy as First Virtue:

  • “My first rule in the classroom is empathy. That’s the first virtue.” [20:03, Bogost]
  • Emphasis on the human dimension as technology upends old models.

---

4. Hobby Joys & Monotasking: Ice Cream, Coffee, and Orthogonality

• Ian Bogost shares his gelato and sorbetto hobby, echoing the book’s theme of deep, diverse sensory pursuits.

  • “It’s not about the outcome...just as important is the experience of doing something different than I do every day.” [30:46, Bogost]
  • Rejects performative creativity (“gross flavors are just engagement bait”), settles on “deliberate flavors you want to eat” [32:19].

• On coffee:

  • Detailed, enthusiastic digression about espresso machines, the pleasures of preparation, and coffee rituals—a lived example of “the small stuff.”
  • “Getting out the $20 drip coffee was delightful...I brewed it really strong...the in-laws like it strong, too.” [28:28, Bogost]

---

5. Big Tech, AI Agents, and Data Privacy Showdown

• Leo shares his extensive use of AI personal assistants (agents), his $5,000/month token consumption (“cheaper than a human personal assistant!” [64:09]), and creative deployments for health tracking, content summarization, and even image recognition on local devices.

• Debate: AI vs. Human Personal Assistants

  • Paris and Leo debate whether AI context/memory can ever match a loyal long-term human assistant.
  • Paris: “An employee has years of context...AI can’t hold 5–10 years of context and experience.” [74:44]
  • Leo: “My agent has a database of five years of health data and all my Obsidian diary entries—probably more accurate than a human!” [76:13]
  • Paris (gently): “It’s a qualitative difference...the management style, the handholding, the tacit knowledge...it’s not the same.” [79:25]

• Privacy Dilemmas:

  • Concerns about cloud-based, always-on AI like Anthropic’s new “Cowork” browser app.
  • “If you’re using one of these tools, all the memory, everything is stored on their servers...a privacy nightmare.” [92:22, Leo]

---

6. AI & Literary Publishing: The "Serpent in the Grove" Controversy

• Discussion of the Commonwealth short story prize-winner accused of being AI-generated.
  • The author, Nazir, denies it, claiming dictation with speech-to-text tech, but strong Pangram AI detector scores and linguistic “tropes” turbocharge suspicions.
  • Paris: “It’s the sort of story I’ve read aloud at parties just for how obviously AI-ish the prose sounds.” [126:14]
  • Will it ever matter? As AI detection falters, the panel agrees: “We have to get over this debate because we’re not going to know” [125:04, Jeff].

---

7. AI Crawlers vs. Publishers & The Tyranny of Defaults

• Cloudflare’s New Blocking Defaults: Starting September 15, new websites will block AI crawlers by default, sparking debate over access, consent, and small publisher viability.
  • Jeff: “This will cut off lots of small publications from AI and the audience there...Cloudflare’s getting too big for its britches.” [111:02]
  • Paris: “I think if someone wants to be accessible by AI, they can opt in. The default doesn’t have to be for the AI companies.” [111:38]

---

8. Travel Stories: The Journey as the Destination

• Memorable moment: Paris’s odyssey returning from Montana—severe JetBlue delays lead to sleepless hours, a $12 voucher, and a surreal 1:00AM Greyhound bus from Richmond, Virginia to NYC’s Chinatown:

  • “Don’t mention buses to me for the next couple of months...” [55:14, Paris]
  • “But I got home, and I got a great story out of it.” [57:05, Paris]

• Paris’s roadtrip takeaways:

  • Sometimes the best sights are “just a random road that I was driving between...the small stuff.” [50:44, Paris]

---

9. News Briefs & Lightning Topics

• AI model rivalry: New OpenAI, Anthropic, and Xai models launching (“Sol, Luna, Terra, Mythos, Grok 4.5”).
• U.S.-China technology decoupling: Both nations threatening to block cross-border AI access. [68:45+]
• OpenAI’s “socialist temptation”: Proposal for 5% sovereign equity, echoes of Alaskan oil funds; Bernie Sanders wants 50%. [71:27]
• Meta’s privacy stumbles: New “super sensing” glasses and opt-out needed for content in Meta’s Muse Image AI tool.
• Google and Nvidia: AI’s energy usage up 37%, possible delays in Kyber AI rack shipments.

---

Notable Quotes & Memorable Moments

• On sensory engagement:

  • “If you pick it up along the way...it’s like building this foundation of resilience and comfort in your life by deriving contentment from every moment rather than thinking you can only do so every now and then.”
    — Ian Bogost, [34:48]

• On education and cheating:

  • “He made the final in-person...18 dropped the course. The grade spread dropped by 50 points. The system is stacked against people who want to actually invest in their own education.”
    — Paris [135:48]; Leo [139:57]

• On AI vs. human personal assistants:

  • “The goal with hiring an employee is you have years of context...your AI is not going to be able to hold that the same way.”
    — Paris, [74:44]
  • “I have a SQL database, five years of health data...my agent remembers everything. It’s more accurate than a human!”
    — Leo, [76:13]

• On AI detection and creativity:

  • “We have to get over this debate [of detecting AI] because we’re not going to know.”
    — Jeff, [125:04]

• On nostalgia & food science:

  • “How could you lose money on Twinkies?”
    — Leo, [154:13]
  • “Twinkie shelf life: 65 days is a blink compared to Smucker’s dog food. I thought Twinkies would last years!”
    — Paris, [155:27]

---

Timestamps for Key Segments

• 04:55: Dr. Ian Bogost on “The Small Stuff”
• 11:41: Decline in reading, Brown professor’s cheating study
• 16:26: Programming Atari, value of learning close to the metal
• 26:40: Tech disconnecting us from the physical world
• 30:46: Ice cream hobbies, orthogonality, and savoring the process
• 36:10: Gratification: “linger in the sensory”—practical advice
• 53:35: Paris’s epic bus ride: “Don’t mention buses to me...”
• 92:20: AI privacy dilemmas: Cowork, cloud-based AI, business data
• 111:02: Cloudflare-Cloudflare blocks AI crawlers by default—publishers’ debate
• 135:53: Brown exam cheating, grade collapse with proctoring
• 176:33: Paris recommends “The City That Ate Itself”—Berkeley Pit disaster

---

Recommendations & Picks

Leo’s Picks

• Riddle for Remarkable: Turn your e-ink tablet into a Harry Potter-style magical diary (write, answer appears/disappears on the page).
• Open Notebook: Self-hosted multi-voice AI podcast generator, 100% local alternative to Notebook LM.

Paris’s Picks

• Guide Along: GPS audio guide for scenic drives in U.S. National Parks (esp. Yellowstone, Glacier)—folk wisdom, local history, and turn-by-turn advice. [169:11]
• The City That Ate Itself: Butte, Montana & the Expanding Berkeley Pit: Book about mining’s toxic legacy and strange Americana. [175:02]

Jeff’s Pick

• Trader Joe’s Cacio e Pepe Puffs: “A little weird—but could eat the whole bag.” [181:46]

---

Show Tone and Closing Vibe

Fun, relaxed, and meandering—deep dives into technological, educational, and philosophical quandaries, interspersed with travel misadventures, foodie talk, and good-natured debates. No one has all the answers, but the journey (and even the bus ride) can be just as valuable as the destination.

---

Next episode: Rafi Krikorian (Mozilla) on the State of Open Source AI.

---

(End of Summary)

Transcript

[A]

It's time for Intelligent Machines. Paris Martineau's back from vacation. Jeff Jarvis is here. Coming up, the return of Dr. Ian Bogost. His new book is all about the small stuff, taking great pleasures in the little things in life. It's going to be a lot of fun. Intelligent Machines is next. This episode is brought to you by Black Hat usa. If you listen to this show, you go deep on the technical detail. Well, so does Black Hat. For nearly three decades it's been where the security industry's most rigorous research gets presented and pressure tested. More than 100 hands on trainings taught by practitioners who've actually deployed in live environments, not lecturers reading from slides. And hundreds of peer reviewed briefings that go well past the overview into the real work across the four areas defining security right now, AI and autonomous threats, cyber conflict, systemic resilience and identity. This year, Black Hat's briefings pass includes all keynotes and main stage access, plus business hall entry. You also get breakfast, lunch, arsenal live tool demos, on demand session access and admission to the midnight in the war room screening. Black Hat takes place from August 1st to the 6th in Las Vegas. If you want the depth this show gets into in person with the people doing the work, this is the room. And we'll be there too. Prices rise on July 17th, so book before then. Use code TWIT for $200 off your briefings [e-mail] us26 that's B L A C K H A T.com us26 podcasts you love from people you trust. This is TW. This is Intelligent Machines with Paris Martineau and Jeff Jarvis. Episode 878 recorded Wednesday, July 8, 2026. The 1:00am bus to Chinatown. It's time for Intelligent Machines the show. We cover the latest in AI and there is a lot to talk about today. Robotics and all those smart doodads all around the house. Paris Martineau has returned from Big sky country. Investigative reporter, Consumer Reports. Did you have a good vacation?

[B]

I had a lovely vacation.

[A]

We saw at one point drove through

[B]

a field of a thousand buffalo.

[A]

I think we call them bison, don't we? Are they bison?

[B]

Bison, Buffalo. I don't know. They didn't tell me that. They were all busy doing animal things.

[A]

Were you off the grid fully or were you just. I mean, you posted some in hostels?

[B]

I mean I was. What do you mean by off the grid? I had cell phone service sometimes.

[A]

Yes.

[B]

I think we exist in a modern age. It's kind of hard to drive a thousand seven hundred Miles and not have cell phone service or. I did. I did a big old road trip all around Montana. Wyoming.

[A]

Impressive for somebody who doesn't even own a car.

[B]

I know. I've changed my ways.

[C]

Used to get a Gillette.

[A]

Buick.

[C]

What was it? Yeah, Buick.

[B]

Toyota Camry.

[A]

Oh, that's okay.

[B]

I mean, listen, it's completely. It was a car and that's all I needed it to be.

[A]

Do you have anything in the car line? Welcome back. We missed you terribly. We're so glad to have you.

[B]

Missed you guys, too.

[A]

We actually did a show without anybody. And then Mike Elgin.

[B]

I know. And I felt very bad about that one. I was like, I could have found some Internet.

[A]

It's fine. It's fine. Actually, Jeff and I get along okay.

[C]

Really? Good enough.

[A]

Believe it or not. Yeah, that's Jeff Jarvis. He's the emeritus professor of journalistic innovation at the Craig Newmark. Craig Craig School of Journalism, Newmark University of New York. He's also the author of a book that is about to emerge. Hot type. Hot off the presses. Hot type. The magnificent machine that drove Mark Twain crazy.

[C]

Did you know that back in the day, things didn't come hot off the press, they came wet off the press.

[A]

Ooh.

[C]

Paper was wetted down first.

[A]

When did it become hot?

[C]

Well, the Linotype, I'd say, is part of that.

[A]

Yeah. Melting lead.

[C]

I'm very excited. I'm not going to tell you what. It is my thing at the end of the show. I have a wonderful thing.

[A]

Oh, good.

[C]

Don't guess I'll tell you that.

[A]

Great to see you. He is also a professor at Montclair State University and SUNY Stony Brook. And you were talking with our guest, who is also a professor about.

[C]

He's a real professor, unlike me.

[A]

Professorial duties. It's so good to have Dr. Ian Bogost back on the show. You were great when you came on a few months ago and we said, hey, when your book comes out, we gotta have you back.

[D]

That's right.

[A]

And I'm back.

[D]

Thank you for having me back.

[A]

Well, it's wonderful to have you. The book is called the Small Stuff, as in don't sweat the small stuff. Right.

[C]

Enjoy the small stuff. That's it.

[D]

That's the book.

[A]

How to lead a More Gratifying Life. You got disconnected from the physical world, but you can reclaim the sensory enchantment of everyday life. And that's why he has two real fig leaves behind his head.

[D]

Are they fig leaves?

[A]

Are they fig leaves?

[C]

They look like.

[D]

We didn't assume they're painted Monstera. They're more like monstera leaves, aren't they?

[A]

Oh, boy. Paris knows the monstera.

[D]

Yeah, they're not quite as monsterific as a monstera, but they're in the monstera. They're monstera alike.

[A]

But you were quick to point out they are real.

[D]

Well, they're real objects. I suppose they're real in the sense that they're not digital recreations. They're.

[B]

They're real objects.

[D]

They are, for all of your viewers, digital recreations.

[C]

Yes.

[D]

On my wall as real metal objects.

[B]

If you're watching this podcast, there's no way for you to know whether they're digital or real. They exist purely in the visual field for you.

[A]

We can only transmit bits in this job.

[C]

Are you in the. Is the wall behind you painted with the Schleurpf of paint that you talk about at the beginning?

[D]

No, this is ordinary paint. Although it. It surely did make some kind of delightful sound as it was applied.

[A]

Yeah, yeah, it's the Velvet Sundown is what that is. Oh, yeah, you. Do you still. That band which you wrote about in the Atlantic Last year at AI band had 850,000 Spotify listeners. Is it still up?

[D]

That's a great question. I mean, these things are such that we don't even know anymore, do we? It's like that that came and went without us even having to notice. I haven't checked in on them lately.

[A]

Music, they live even without their creator. Music has become wallpaper for a lot of people, which is sad.

[D]

Oh, I think that's right. I mean, it's just background, right? Yeah, I mean, that was the argument I made in that story, that music had already become kind of. Kind of ambient, and so it didn't matter so much if you're. If the artist you were listening to in Spotify was real or AI generated.

[A]

Now, when we talked last, you noted that three of your books, your previous books, were in the AI Ingested pirated database, but you didn't seem to be too unhappy about that. Would you like to keep small the small stuff out of that or not?

[D]

I think I'm on the record not objecting to my work being consumed by AI in order to be used as the case base for LLM style generative output. So I don't see why I would change my mind now.

[C]

Does your last page tell them just keep out?

[D]

Can you do that? Sounds like a Facebook post myth.

[C]

Well, I think Jeff, when he was

[B]

recording his audiobook, was made forced basically at mic point to record a little thing saying that it couldn't be ingested by AI. Is that right?

[C]

Right. And then I'm looking at. At hot type on sale. Now, on the. On the copyright page, it says that no part of this publication may be. I'll skip over one to used or reproduced in any way for the training, development or operation of artificial intelligence AI technologies, including generative AI technologies.

[D]

I wonder if that. Let me look at my.

[C]

I'll bet it's there.

[A]

Silent publishers stick those in now, I'm sure.

[C]

Yeah.

[D]

Is that. What are we looking. What's the name of this page? That's not the colophon. It's something else.

[C]

No, it's a copyright page. Copyright page? Yeah.

[D]

Oh, yeah. Let's see. No, it says that no amount of this book may be reproduced. Blah, blah, blah, uploaded any website, large language model, artificial intelligence system without express permission. Must just be a big publisher thing now.

[C]

Yeah, it is. Yeah.

[A]

I mean, it's always said can't be reproduced without permission, but now it's. It's actually explicitly saying AI.

[D]

But isn't one of the questions though, in the. You know, in the legal drama all around this whether it's reproduced? Right, right, right.

[A]

Well, that's why New York Times says is you're taking our articles and reproducing them.

[C]

Right. As opposed to reading and learning from them.

[A]

Right.

[C]

Reproduced or transmitted in any form, electronic or mechanical, including photocopying. Oh, you remember that. A recording or by any means of any information storage or retrieval system without prior permission in writing from the publishers or to. Used, reproduced on or not.

[A]

So I can't mimeograph it.

[C]

It's used or reproduced. So if you just use it. Reading, formerly known as. Now known as using. Yeah.

[A]

Let me ask the mind if they ever hand out xeroxed texts to their students.

[C]

We

[D]

been a while. We used to. I remember creating readers.

[A]

Right.

[D]

You know, those like sort of janky bound copies that you'd buy at the bookstore and they would handle getting all the rights. And I don't think I've done that since the aughts. Since.

[A]

All online now. Right. So.

[D]

Well, there were ways of putting it online as. As digital PDFs, and so. And I think that that process. I mean, I'm not speaking for myself, of course, but for the general profession, I'm pretty sure that process kind of ignored copyright much of the time and relied on educational fair use as justification. So you do like have a PDF and you'd put it on your courseware page. I suspect that that is the most commonly undertaken method you know, where we're at now, broadly speaking, is a sort of just desperation for the students to read anything. Right. So I think any friction, I think any friction that was introduced into the process became seen as justifiably overcome by whatever means necessary. I don't know, I'm hypothesizing here a little bit. I don't think anyone's done a survey and we don't really talk about this in the halls, really.

[B]

I was gonna say it's. I feel like a big week already for news about how none of the kids are reading and they seem to all be cheating on their exams very flagrantly. We've both had the Atlantic's cover story this week.

[C]

Right.

[D]

My colleague Rose. Yeah.

[B]

About how basically seems like no one is really reading anymore. The share of Americans who read for pleasure declined by 43% over the last 20 years. It's way worse for the young. And then at the same point I saw an article, it was actually very interesting from inside higher education that came out, I think either today or yesterday that was basically kind of a breakdown from this Brown professor who suspects that most of his class used AI to cheat. And it was specifically a like point by point comparison of the take home midterm grades for all 58 of his students, which were all like 100% typically way higher than what he normally gets for take home grades. And then he decided because he suspected cheating to move the final in person rather than online. Eighteen people dropped out immediately as soon as he said that. And then the average, I think score was below 50%. And there's this great graphic that I'll post in the chat that basically just shows the difference between people's midterm exam score and then their in person final exam score. And it is. The spread is often 50 points or more. It's quite rough.

[A]

What's kind of interesting in the article in the COVID piece by Rose Horowitz in the Atlantic is the thesis that. And you've said this about mass media, Jeff, that reading might be a short blip in human history, just as mass media is.

[B]

I don't like that.

[A]

A short century long blip in human history. I think that's an interesting thesis.

[B]

I don't want entirely to say that

[A]

we're all attached to our current situation. We're like fish. We don't see the water. But you know, so it hasn't always been.

[C]

I mean there was, there was a time when reading was seen as corrupting. Paris young women.

[A]

That's right.

[C]

Especially novels.

[A]

Reading in the train, you know, that was.

[C]

That was seen as very bad for you. Have you changed your syllabi much yourself, Ian? Because of AI?

[D]

Because of AI. I mean, I have. So here's the thing.

[B]

Let's.

[D]

I just being honest, we professors are busy and lazy and put upon, and I'm speaking as a class and not as a person. Now, the sense of change, the sense of like, I don't think this is justified necessarily, but people don't want to change their syllabi. And that's kind of a problem. Even before AI, there was a sort of tacit agreement that a class was a class and yes, you'd have to update it and it would change over time. And some subjects required more frequent updating than not. But even before AI came on the scene, we were already at a kind of maybe unacceptable baseline of the syllabus. Doesn't have to change. And so as I've thought about that, as AI has come around, I think we talked about this last time I was on that. I have had to make adjustments in ways that were surprising. Like two years ago when I taught Atari programming, which is a kind of programming AI couldn't do two years ago, I was just putting the syllabus like, good luck. My policy on AI doesn't matter because AI can't do it. And then this spring, AI could do it, but badly. Not like badly like it could produce something that ran, but it was not going to be helpful. So I wouldn't say it's about the syllabus. The syllabus has been dead for a long time anyway. It's really just a legal document. It's more like, what is a course? How do you operate a course? What are the students trying to do? How do you help them accomplish what they're trying to do? What changes arise and then how rapidly can you make adjustments, and then how long do those adjustments last? So I think it's much, much weirder than just like, whether your syllabus has to change, which is, of course, a totally fair way to ask the question. I don't know that the average person who doesn't work in higher education quite understands how exhausting it has been over the past 10 years, but especially five years, but especially year. And so any gift we can give ourselves of sameness or of. Of. Of reprieve, we are tempted to take. And it's not necessarily virtuous to take those reprieves.

[A]

I'm curious why you taught a class in programming Atari.

[D]

Oh, I've been teaching this for years.

[A]

So I know you do games, and some of your games, by the way, are in the Smithsonian.

[D]

Yeah, including an Atari game is in the. You know, because it has to be weird to be an art museum. Right.

[A]

But it's not. The purpose is not to teach kids 6502 programming, I hope.

[D]

Well, I mean, I. Actually, you could still live a productive Life as a 6502 assembly programmer writing code for embedded systems. In certain cases, the 6502 is still alive as a commercial. Good, but in an esoteric way. I talk about this when we, you know, students sign up for the class for some reason, and then they get. And I'm like, well, let's talk about why you're here. And some of them are there because. Because they're interested in game design and programming. A very simple old computer like the Atari VCS is one way of kind of getting back to basics, almost like doing figure drawing or something like that, or color theory. That's one reason folks are there. Some of them are just interested. But knowing how to work at the metal, close to the metal like that, with a microcontroller in assembly, that's very uncommonly taught, even in a computer science degree anymore. And that does have some value because you may still have to learn some new chipset and write efficient code at a very low level. It might not. But if you're an electrical engineer, you would. I try to make it such that students can interpret value in various ways, just like the experience of constraint and of working within the particular arbitrary, but given constraints of a. In this case, a technical system. Like, that's a valuable lesson for me.

[A]

It would almost be an exercise in hedonism. It wouldn't. It would. Just because the problem solving feels good, and I would just enjoy it. I almost wouldn't be thinking about.

[D]

I love it. Well, it's not useful. I mean, this is an interesting debate, right? You know, that should. What is the nature of utility in education or in life, for that matter? Which is, you know, kind of something I talk about in the small stuff book that we're so focused on outcomes and goals and on achievement and on reaching those goals and on everything that we do being efficiently optimized to produce outcomes that we've kind of forgotten to experience.

[A]

Just enjoy lives.

[D]

Yeah, right. And so it is funny, actually, like, you know, if you think about, like, using AI to cheat or to, you know, do your work for you in video game programming class, it's like, well, I don't know what to tell you. Man, if you don't want to do video game class, then good luck. Right? But the answer is that the students are harried and they're pressured and they have all sorts of obligations because they're trying to collect credentials and they're trying to get internships at McKinsey, and all of this stuff is going on around them. And I really empathize with their plight. So, I mean, I'll tell you, in spring's class, you know, I knew I could tell when the students were submitting code they didn't write because I. Because they were producing instructions in a way that I hadn't taught them. And there was no way that they would have learned it on their own. And, you know, it wasn't like, oh, you cheat. I mean, when I talked to them, I was just like, you know, this. You're gonna regret doing this because you cheated yourself, because you build on it. Yeah. But also, you know, they're the ones who, in some cases, you know, are coming to class every day in their suits because they have to do presentations for. For their marketing class, and they're interviewing for internships at consulting companies. And, you know, I feel for them. I think my first rule in the classroom is empathy. That's, like the first virtue.

[A]

That's beautiful.

[D]

Yeah. I mean, I try. I don't always succeed, but I try. And I think the best I can do is sort of model the process of living in the experience of living, which I tried to do in the book, too, and which, I mean, and I believe in. I think it's so important to believe in something and express those beliefs earnestly nowadays. So those are my defenses against AI rather than, like, rules. I don't think those are as effective.

[A]

It's one of the reasons I dropped out in my junior year of college is because the goals were so. My goals were I was just interested in learning, not in the grade or the job. That was an era in the 70s when everybody was pre med, pre law, and I just was really interested in stuff, and I found I could really do it myself better because I didn't have the focus on the test, the grade, and all that stuff.

[D]

Yeah, it's funny, because that worry, that critique I'm writing and working on. Writing is a generous term for where I'm at with it. But I'm writing a new story for the Atlantic about higher education and the history of the humanities. And I was looking into it, and, you know, this fight between utility and knowledge learning for its own sake, it's centuries old. Really like we've been having this, this fight with ourselves as almost as long as we've had the contemporary university in the west as we, as we know it. And I think that's kind of comforting that this isn't a new, this is not a new problem. It's one that recurs.

[C]

So one of the stories I put in the rundown was Wall Street Journal high earner families are ditching traditional schools for life skills and AI. And to me it was very. Even though I want to change all kinds of things in life, it depressed me terribly because you had kids who are. The lead story is a 5th through 8th grade school where the father, president of a hedge fund is all happy that his son Will could learn negotiation, sales and public speaking that they're learning through real world problem solving, building businesses and designing products. I mean I taught entrepreneurship and to journalists it was a way to teach them business. I'm all for that. But Jesus, you're sixth grade.

[A]

It's just one more way. Parents have injured their children by assigning their own goals.

[C]

So I said on the socials young that, that these kids are going to have big shrink bills coming up. Interestingly, only on, on Twitter I won't call it X. I got slammed by the muskites. Slammed. I don't want my kids with Tommy. Right. I was just, just going on and on, you know, you, you don't understand kids. You're awful. Only there, not Blue sky, not anywhere else. And I'm all for using new tools. I'm all for new ways to teach kids. I'm all for ways to motivate them. But the parent motivation here in putting these kids in these schools is to make them little mini entrepreneurs. And that depressed me.

[A]

To quote Walt Disney, it's a tale as old as time. We're talking to Dr. Ian Bogost. He is the, he's an Atlantic contributing writer at the Atlantic, the author of a brand new book called the Small Stuff. I guess your 12th book now, how to lead.

[D]

Yeah, I guess my 11th book. 11th and you know, I have to write a 12th book because 11 is a weird number.

[A]

Yeah, you need an even.

[C]

You need a baker's dozen.

[D]

So you got a third.

[B]

You gotta hit 13 for that.

[D]

I don't know.

[A]

Currently professor at Washington University in St. Louis where he has weirdly appointments across three schools. Arts and Sciences, Engineering and Art, Design and Architecture.

[C]

Do you get to go to three faculty meetings every month?

[D]

So the real trick is I don't have to go to any faculty.

[A]

Beautiful.

[B]

You're like, sorry, I met the other one.

[C]

I've met the other faculty meeting.

[A]

The thing that fascinates me so much about actually everything, every time we talk to you and about this book, you're interested in cultural artifacts, the things that we make that surround us, and you assign them really a lot of importance.

[D]

Oh, a lot of importance. Toasters, doorknobs.

[A]

Yeah. You did that whole series. Jeff's a part of it. A hundred some books about toasters and doorknobs and magazines. Why are these little things that most of us think of as kind of trivia so important to you?

[D]

Because. Because they are constant, because they surround us all the time. So we were just talking a minute ago about reading is on the decline, which maybe it is, maybe it isn't. But our comfort with our interest in something like the novel or even the cinema, it's occasional. Like, how many novels does the average person read a year? Well, it turns out nowadays it's zero, right? But I use my toaster every day. I make coffee every day. I sit at my desk every day and enjoy my chair. And in the walls of the room that I am inside in my home. And those kinds of objects and artifacts and experiences, both human created and natural. Right? Or nature created. And the hybrids between them, we have to treat them as important because they surround us constantly.

[A]

Every pebble carries a universe inside.

[D]

If we can't learn to derive meaning from our encounters, and in the case of this book, I'm interested in our sensory encounters with those things, then we're missing out. We're missing out on just a whole wealth, a whole universe of meaning every day. And I think that's actually that attitude that we have where that stuff doesn't matter. It's like, why would you be concerned with your toaster or your doorknob or the can of soda or the water bottle or whatever it is, why would you be concerned with that focus on getting the job at McKinsey or founding the hedge fund, or inventing the next thing, whatever it is, your negotiation skills. That's such a silly way of living to me. Because why couldn't we do both? Why couldn't we have big long term goals that we work toward over time and also live deeply and richly inside of every moment, knowing that those moments are happening to us constantly. One of the reasons I wrote the book really was just the realization that there was that disconnect between the kind of contentment that we call happiness and the kind that I call gratification, which is that small stuff, engagement with Sensory experience.

[C]

So you care about materiality of life and argue that technology has taken us away from that? No.

[D]

Yeah, technology is. Well, technology and other factors, but technology included has disconnected us from the physical world.

[A]

Yeah, but, I mean, this is what. And I'm going to, as I often do, take the devil's advocate position because I agree with you 100%, but this. People who are focused on the goal, not the road say, well, if you're just paying attention to every step along the way, you're never going to reach your goals. You're never going to achieve anything. You're just going to be a space cadet wandering through life like Ram Dass.

[D]

Right. But isn't. Isn't that really kind of a caricature of how we actually live? So when you get up in the morning and you brew coffee and then you hold the mug in your hands and it's warm and you smell the coffee and you sip at it and it makes you feel good, are you stealing something from your ability to achieve your goals? Absolutely not. Absolutely not.

[C]

I got to ask, I got to ask. Are you a coffee fanatic? Because you're surrounded.

[D]

Fanatic.

[A]

Yeah. Okay. And I are both pour over, folks. In order not to bore Jeff. We try not to talk about.

[C]

No, go ahead, Go ahead. I'm outnumbered.

[A]

Go find this grind and which filters. Abaco versus Hario and things like that. But I don't.

[C]

You.

[A]

Surely, Ian, you're not that. Are you that deeply invested?

[D]

I don't do pour over at home. I. I have a manual espresso machine.

[B]

I was gonna say different, but similar. Crazy.

[D]

Yeah. I recently had. I recently got new countertops installed in my kitchen, and I had to temporarily disconnect the machine and use the drip coffee maker. And, you know, it was fine. It was actually fine.

[A]

Don't tell me that it's fine.

[B]

It was fine.

[A]

Huge investment in it not being fine. Yeah, well, you didn't use an espresso.

[D]

I mean, you know, I've written this book. I try to live the experience of diversifying my sensory life, whatever. And so getting out the filter and, you know, it's just like a $20 drip coffee machine that I pull out at the holidays when my in laws are visiting because they get up too early and don't know how to use my. My espresso machine.

[A]

But it was.

[D]

It was delightful actually, to have that. And the coffee was. I just brewed it really strong. You just put like, you know, the whole. The whole bag of coffee in the filter.

[A]

And how did the. In Laws react to that.

[D]

They actually like their coffee very strong, so.

[A]

Oh, interesting.

[D]

Okay.

[C]

They were awake all day, so, so then I'm going to next ask on the socials. You've been humble bragging about all your creations of sorbet and ice creams.

[D]

Oh, oh, I don't know if I'm humble bragging, but I am posting.

[C]

Oh, I, I. Because I want to try them all. So yes, if I humble bragging. The other side of humble bragging is jealousy.

[A]

First of all, where do you post

[D]

Ian so I can Instagram is where my gelato and sorbetto hobby live.

[C]

Oh, sorry, Gelato. I'm sorry.

[A]

Please. Gelato and sorbetto desserts.

[C]

Jeff, I'm so sorry.

[D]

I know it's. You can call it ice cream, it's fine. No, I talk about this in the, I talk about this in the book, though. There's a little moment in the book when I talk about, you know, how I was. So I've been making ice cream as a hobby for like 15 years. And like everyone who has a hobby at some point, I was like, shouldn't I monetize this? Shouldn't I turn it into a side hustle?

[C]

There you go.

[D]

That was my July. There's some. There you go. I mean, they look good, right? They taste good too.

[A]

Pineapple upside down cake with pineapple sorbetto and caramel gelato.

[C]

Oh, my.

[D]

Yeah, that caramel gelato was really good.

[A]

Dulce de leche. Oh, blueberry lavender. Do you have a machine? Do you make it by hand?

[D]

I do. No, no, no. You can't really make it by hand. I have like a very high end machine. It's ridiculous.

[A]

Oh, there it is.

[D]

Oh, there's my vanilla recipe. If you click through there, everyone in the world now has access to my vanilla ice cream recipe, which is. It looks good, right? There you go.

[A]

You use vanilla extract?

[C]

Yeah.

[D]

You use beans and extract. Yeah, I do.

[A]

Anyway, do you use Madagascar beans or.

[D]

It doesn't matter. You can use that if you want.

[A]

He's putting locust bean gum and xanthan gum in it.

[D]

You gotta have stabilizers. One of the reasons that I, one of the things I learned about making ice cream is that the outcome. I mean, I love the outcome. I post it on the Internet. I'm fine with posting it. I eat it, I share it with my neighbors. I. But that's, I don't want to say less important, but just as important as the experience of doing something different than I do every day. And I go in the kitchen And I make ice cream bases, and I figure out how to balance the flavors and the textures with the materials that I'm putting in that'll flavor them. And when there's more water in the flavoring because it's sort of fruity, like peach can be harder. Actually. Blueberry is weirdly difficult compared to raspberry because of the water content. That kind of sensory experience and connection with those materials. I just. I just like having that different experience. I call it orthogonality in the book. Like kind of zagging in your life. Having a sensory life that's different from your professional life and your home life. I think that's really helpful and important and delicious, and in my case, delicious.

[C]

What's the weirdest. What's the weirdest flavor you've concocted?

[D]

I used to make gross flavors. You know, like, would you eat the gross flavors?

[A]

No.

[C]

No.

[D]

Here's the thing. Like, there is so much.

[C]

You get your in laws to eat them.

[D]

So much stuff that people do, they just do for symbol creation. Now they're like, oh, I'm doing this so I can post it on the Internet.

[A]

Totally.

[D]

Right? And there's even whole companies out there that, you know, they have, like, there was, you know, Kraft macaroni and cheese ice cream at Van Loyen or something. And I'm convinced that these things don't exist to be eaten. They don't even exist as ways of exploring what a flavor would taste like in ice cream. They just exist for engagement bait.

[C]

It's like overly huge sandwiches from delis and hamburger places.

[D]

Right. There's that sort of the turning point for me. It was some time ago, seven, eight years ago. I made like a garlic bagel ice cream. And I'm sure someone's done that since then, but at the time it was odd. And my family was like, will you stop this? Will you make what?

[A]

No one wants this.

[D]

And they were right. So now I make just good ice cream.

[C]

There was an intervention ice cream intervention with Dad.

[A]

I will have a.

[D]

Who wants garlic in their. I mean, garlic is incredible.

[A]

Oh, yes.

[D]

But you don't really want it in your ice cream. I do have.

[C]

Right.

[A]

They make garlic ice cream. It's. It's actually quite.

[C]

Have you made olive oil ice cream?

[D]

I've made that. I've made that. That was the basil ice cream. Basil ice cream is good. It's not that I'm against odd flavors or curious flavors, but. But. But they have to. They have to be deliberate, and you have to want to eat them. More important to me is I, I have some rules about texture. I, I'm. One of the reasons I'm gelato obsessed is because I love the smoothness, the density.

[A]

It's not quite the same as ice cream, is it?

[D]

It's not the same as ice cream. You're going to regret making me talk about ice cream because now I'm just going to be ice cream from here on out.

[A]

But we love enthusiasts, Ian.

[D]

American style ice cream, which is sometimes called Philadelphia style ice cream for reasons I'm going to leave mysterious. And your viewers look up because of cream cheese. No, it's not because of cream cheese. American style ice cream traditionally doesn't have eggs and it has higher fat content. More cream, less milk. And it's fluffier and lighter because it has higher overrun, which means that more air is churned into it during the.

[A]

Intentionally?

[D]

Yeah, intentionally. And so that's fine. I mean, ice cream is good. Like you're not going to go wrong with ice cream. But I like that denser, richer Italian style, a northern Italian style because it's

[A]

egg based

[D]

gelato and that's the kind that I make.

[A]

Look at this affogato with his.

[C]

Yeah, you can't go there. Oh, there we go.

[D]

We have the two together. Yeah, just a quick.

[A]

That is a coffee. That is a work of art.

[D]

There's my machine too. Yeah.

[A]

So let's get back to the book because we should do that. And you really look it. Let's focus on the end goal of this conversation.

[C]

Yes.

[A]

Too much looking at the stones along the road here. What do we get? You say it's a more gratifying life. What do we get out of paying attention to the small stuff? And maybe more importantly, how do you recommend beginning that journey?

[D]

Yeah. So I think that this flavor of contentment that I call gratification, it's like, I mean, not to draw on the ice cream metaphor. It's like an ingredient in your overall sense of well being and contentment. Like, I want you to feel happy and to have purpose and to have goals. I want you to feel satisfied that you've accomplished something in your life. And satisfaction is a different kind of pleasure. But gratification, it's like its own universe of contentment. It's small scale, it's easy. And if you pick it up along the way, not only do you feel better overall, but it's almost like building this, this like foundation of resilience and comfort in your life by deriving contentment from every moment rather than thinking that you can only do so every now and then. And I think that really bolsters our sense of presence of purpose in life, well beyond what. What. What happiness offers. As for how to do it, I think one of the simplest pieces of advice I can give you is allow yourself to linger just a little bit longer in any sensory encounter that you're having. And if it starts to feel like, a little strange, then that's a good sign that you're doing it. Write, you know, like, if you'll indulge me here. Will you indulge me? Here's the book, right, that we're talking about.

[A]

Oh, it's a small book, too.

[D]

It's a small book. And we, you know, we did a little varnish on the type, which I can feel with my fingers. You just have to trust me on this, or you'll have to buy the

[A]

book to feel it yourself. Buy the book. Yeah.

[D]

And, you know, when you buy a book and own a book and hold a book even before you read it, even if you never read it, if you just. You might find yourself kind of feeling that type of. On the COVID that's what I mean by lingering. Just let that happen. Just accept that gift, and then you can move on and do it again with something else. And you'll find that that's available to you all the time. You don't need to tie your brain in knots with mindfulness exercises about how you're going to be more present. You're always present in your physical human body, and you always have the opportunity to engage your senses, or maybe better put, to accept the sensory payload that the world is delivering.

[C]

I loved the bite of type and letterpress on paper.

[A]

We love paper books. I mean, I listen to audiobooks. I read books on an E reader, but I can't get rid of the paper books. I love the.

[D]

Exactly the sense of one page kind of turning against the next. But then the E reader is also gratifying. I mean, this is important. It's in the book, but it's important that technologies, computer technologies included, are also gratifying. It's not that they've stolen that experience from us. The smoothness of the glass of your smartphone screen is also gratifying. The lightness of your E reader and being able to carry it with you in your purse, that's also gratifying. So we have to learn not to divide the world into, like, you know, gratifying and ungratifying good and evil entities or technologies. But at least in the case of gratification to understand that we can get it. We can get it from almost anywhere.

[C]

So I was talking today to a Swedish academic named Jonas Iverson because he wrote a really interesting essay about friction and arguing that AI makes everything so easy. Takes away.

[A]

I got plenty of friction. Friction to spare.

[C]

So is friction necessary to learn? Is it necessary to gain knowledge? Is it necessary to feel satisfaction? Is it. We feel. We hear a lot of arguments about friction these days.

[D]

We've heard it lately. I'm kind of against the whole friction discourse. I don't think it's wrong, but I think it's not the point I'm trying to make with gratification, because the thing that I just showed you where I was able to touch the type on my book cover, that didn't require difficulty, that wasn't about friction, that wasn't about making my life harder in order to be more involved in it. So I don't think. I think it's understandable to me why friction has emerged. Emerged as a tempting solution because technologies have made things so easy. And, you know, your literal smartphone screen is frictionless, and you can. You can glide your finger across it, and everything seems possible on it. So I understand the motivation for citing that. That example, but I actually think, at least. At least in terms of feeling gratified, you don't have to make it hard. And in fact, one of the properties of gratification is that, unlike happiness, it's easy. It's always available to you. So I have this, like, textured polo shirt on today, and it has this sort of. It's knit, and it has this bumpy texture, and I can touch it or I can just feel it on my body when I'm not touching it, and you can see it and perceive its texture. And it doesn't require us to, like, I don't know, like, add friction to

[C]

our wardrobe or something.

[D]

What would that even mean? It's just. It's just hair shirts. It's just, like, right here.

[C]

Right? Right.

[D]

Yeah, like, I'm gonna wear a hair shirt. It's just right here. And, yeah, maybe there are aspects of my life I want to do more with my own hands or I want to be more engaged in. And I think that's where the friction discourse kind of really means to live. But to me, it strikes me as, like, a way of trying to, again, to divide the world into two categories, ease and difficulty. And we've been. Everything has been made so easy. Now we need to make things difficult. These kids today, it's like Internet life has turned us all into extremists. Everything has to be all or nothing all the time. And I just don't quite buy that.

[B]

But.

[D]

But I also think AI has made my life harder in a good way, as much as it's made my life easier. So I was mentioning that I had my countertops redone. So the stone guys who put in my new countertop broke the. The knob of my range oven off. And it's the kind of oven that to replace it, I have to replace the whole thermostat assembly, which is, like, tied to the back of the unit and all of this. And as I was trying to figure out what to do about it, I just asked it. I asked Chat gbt, like, hey, like, what part does my Viking range need? And, you know, it helped me find it. And then it's giving me some tips about how to go about repairing it if I choose to do so myself. And, like, that opportunity, normally I would have just called somebody in or, like, yelled at the stone guys, I can't believe you broke my oven. But that's interesting. Like, now I have the ability at least, or the opportunity to engage with that ordinary object in a different way. And I would feel satisfied if I completed it, and I would feel gratified while I was doing it, and I would feel happy to have a working oven again that I could cook food for my family with. And that's not really friction either.

[A]

Right.

[D]

It's like a different invitation for a complex, multifold engagement with the world that isn't just me scrolling either. It's just like, a lot more complicated than we've been at credit for.

[C]

My wife would say, you're never going to fix this, and you know it.

[D]

I think I can fix it, but even if I can't, even if I can't, the feeling of having failed to fix something, but to try to do so, that's really opened doors for me. I can't remember if we talked about my irrigation backflow last time I was on, but I had a similar problem with my.

[A]

Yes, we did.

[D]

Yeah. So when that happened and the guy. I didn't fix it, but I tried, and the guy came out and I had this just like, deep, rich conversation with my irrigation repair guy about backflow mechanisms. And I was able to engage with him as a person just much more deeply than I would have been able to otherwise. And I really enjoyed that.

[A]

I have to write with the septic tank guy, normally a guy I would have stayed as far away from as possible, but instead, I actually had a wonderful Conversation. He invited me to coffee.

[D]

Yeah. And, you know, one of the things underneath this argument that I'm making is that, like, I am fortunate to, like, be in a socioeconomic situation, which I own my home and have incentives and inspiration to repair. Not everybody gets those opportunities. Opportunities to connect in a sensory way, to experience the gratification of engagement with something that they're invested in. And that's really too bad. That's one of the forces of dematerialization that I talk about in the book.

[C]

So I just. I'm writing a post for the Bloomsbury blog now. Because you do that to promote your books.

[D]

Yeah.

[C]

And my similar experience as a non academic, as a. As a poser in the palace, when I discovered the joy of research and digging rabbit holes into, or exploring rabbit holes into academic databases and finding citation upon citation upon citation. That's my similar. Too late in life, found joy.

[D]

Yeah. Yeah. I was talking to someone recently about the lost art and craft of, like, card catalogs and microfiche and microfilm research. And, you know, that was also. It was. It was pleasurable because knowledge lived somewhere and you got to go to those places and they often had grandeur architecturally, and then you got to be quiet inside them and search for.

[C]

Yeah, but it's better now because you can find all kinds of stuff you couldn't have found.

[D]

Definitely better now. It's definitely better now. But also, just like, one of the motivations for doing it wasn't to do the research, but just to have the sensory encounter with these odd machines.

[A]

I think the key here, at least what I'm gleaning from it is, yeah, different opportunities for different people. But the key here is to. Is to pay attention, to be present with the opportunities you have when things happen, wherever they happen, and your life will be richer and more fulfilling. And goodness knows, I think we all know that those goals that you spend your entire life trying to achieve often don't satisfy at the end of the day that you get there and you go, wow.

[D]

Yeah. You know, I don't know as a parent, you know, like, your kids grow up and they leave home and then you miss them.

[A]

Yep.

[D]

And the things that you miss aren't like going on a big vacation or writing the college essay or something. It's like hearing the creak of the floor above you. Yeah, right.

[A]

It's true. The pit. A pat of little feet. Ian, it's such a pleasure, as always, to talk. Ian's website is B O G O S T dot com. That's where you'll find the small stuff. His new book, how to Lead a More Gratifying Life. I think he's given us a great lead into how to do that. You can read his work in the Atlantic. He's working on something right now. He's a professor at Washington University in St. Louis and a game developer. In fact, he has some really interesting books about video games, including Play Anything, the Pleasure of Limits, the Uses of Boredom, and the Secret of Games. Just one of many books from one of 11 other books.

[D]

11, 11 total books.

[A]

10 other books. Ian, always a great question.

[D]

Always great to join you.

[C]

Thanks.

[A]

Yeah, we really enjoy it. Come back. You and your Monstera leaves are welcome anytime.

[D]

Fantastic.

[A]

All right. Thank you, doctor.

[D]

All right. Have a good one.

[A]

Take care.

[C]

Cheers.

[A]

We'll continue with intelligent machines right after this. First, a word from our sponsors. You know, Jeff, have I talked to you lately about a new mattress? About a year ago now, Lisa and I looked at our mattress and said, it's been a while. Maybe it's time for a new one. Our show today, brought to you by our new mattress, Helix Sleep. I guess it's not new anymore. It's a year old. And I am so glad we chose Helix. We read all the reviews, we looked around, we really did a thorough search, and we settled on the Helix Sleep. Maybe this is a time for you to think about it. We're heading into summer. Days are hot. Gosh, knows nothing beats the feeling of a total refresh. Instead of that soggy old mattress, you spend more than a third of your life on the thing. Don't you owe it to yourself to get a really nice one? If you're still tossing and turning, you're dealing with those sweaty nights, waking up with that nagging back pain. Maybe it's time to stop compromising. You shouldn't have to settle for a mattress made overseas like so many are with questionable materials, when you could be resting on a beautiful Helix mattress. Each one hand assembled, packaged, and shipped right to your door from Arizona in just a matter of days, they make it to order. So you're getting a delicious, fresh new mattress. Smells fantastic, it feels fantastic. And the beauty of it is Helix makes the process completely personal. We took the Helix Sleep Quiz. It's kind of like having your own personal consultant. Not like the person at the mattress store who stares at you while you try out mattresses. This is so much better. The Helix Sleep quiz will match you with the mattress perfectly suited to your sleep style, to your body, your preferences. It's not just about the feel it's about the science. Helix actually put their beds to the test. They did a Wesper sleep study and the results exactly matched my own, which is how I know it's true. So here's what they found. In a Wesper sleep study, Helix measured participants sleep performance. After doing what Lisa and I did, switching from our old mattress to a Helix mattress, 82% of participants saw an increase in their deep sleep cycle. So deep sleep is a small fraction, maybe an hour of your overall sleep at night. But it's the most important. It's the one where the spinal cortical spinal fluid flows in, cerebral spinal fluid flows into your brain and clears out all the old stuff. Makes a big difference in aging and Alzheimer's and cognitive ability and just how you feel. So any increase in deep sleep is great. 82% saw an increase. On average, participants achieved 25 more minutes of deep sleep a night. For me, that was a doubling. A doubling of my deep sleep. Participants on average also achieved 39 more minutes of overall sleep per night. A lot less tossing and turning, staring at the ceiling. A lot more feeling great in the morning. For 10 years, Helix sleep is consistently ranked at the top across independent review sites tested and reviewed by all the big guys. These are the reviews we looked at. Forbes Wired Helix delivers your mattress right to your door. Free shipping in the US and oh important you can rest easy with seamless returns and exchanges. They call it their happy with Helix guarantee. A risk free customer first experience ensuring you're completely satisfied with your new mattress. So go to helixsleep.com machines for 20% off site wide. It's their 4th of July sale. Helixsleep.com machines for twenty percent off site wide do enter our show name Intelligent machines after checkout so they know we sent you. And the offer ends July 12th. If you're listening, after the sale ends, don't worry. Still great deals, always check them out. Helixsleep.com machines please use that machine so they know you saw it here. Helixsleep.com machines we thank them so much for their support of I am, I am, I am, I am. So you love Montana, you love the bison, you loved the pig races.

[B]

It was all pretty great. I mean I planned this trip with a couple days notice, mostly based around the fact that there was a direct flight from JFK to Bozeman and that I could get a reasonably priced rental car for the two weeks or so. And it ended up being phenomenal. I had such a great time. I saw so many cool animals, did a Lot of great hikes went to Yellowstone, Grand Teton National Park.

[A]

Nice.

[B]

But honestly, some of the best sites were just a random road that I was driving between.

[A]

The small stuff.

[B]

You know the small stuff.

[C]

How was the food?

[D]

Very.

[B]

Meat heavy.

[A]

A lot of french fries.

[B]

I mean, delicious, but very. Afterwards it came back and I was like, I need to only eat vegetables for a while. Yeah. At. At the pig races, which I had sent Leo and Jeff a video. From there, it was a saloon attached to a steakhouse that on Thursday through Sunday had pig races for charity where people would bet on pigs that raced around like a little tiny track. The pigs weren't being harmed in any way. They were just kind of.

[C]

Until Monday when they became dinner.

[B]

No, I think they don't serve any pork products there. It was. But they did serve steak and bison and this. I got a steak from a place that was like a 20 minute drive away, which is kind of interesting. Like, I was like, that's where the steak grew up. And that's strange.

[A]

Yeah. See, out here, you're in New York, so. Yeah. But out here, we look out the window and there's the lamb you're gonna have tomorrow. You know, this is all part of the. Part of the agricultural scene in Petaluma. That's why butter and eggs day is so important. You're looking at breakfast.

[B]

When's butter and eggs day coming up?

[A]

Oh, not till next spring. You know what you should come out for next time? For butter and eggs day.

[B]

You know, I will if you make it to New York before that.

[A]

Oh, that's the deal, huh? All right. I know I need to come out. I really do. I need to come out. Henry says, please come out. I want to see you and I want to see my son. And oh, well, oh, well, oh, well,

[B]

next year you'll be saying the same thing. It's, you know, there's no way to get from Petaluma to New York, actually

[A]

such a long bus ride that it's just. There's a lot of. It's a.

[B]

Can't talk to me about buses right now.

[A]

Why is that?

[B]

My flight. So I had designed this perfect direct flight, four hours and something from Bozeman to jfk. I'd scheduled my flight back, it was leaving at like 11. Get there, we get in the plane, set down, start to taxi, then we. Everybody on the plane gets a text saying your flight's been delayed to 3:50. And they pull back, get us all off. We have to sit there for four hours. I'm like, all right. Well, I guess we're gonna be flying through fireworks now because it's fourth of July.

[A]

Well, that's kind of weird.

[B]

I was like, cool. But like they said they delayed it because of the fireworks. And I'm like, that doesn't make sense.

[A]

They delayed it because of Taylor and Travis is why.

[D]

I bet.

[B]

So, so, you know, we get there, basically we end up somehow in between boarding the next plane and what was supposed to be us landing at JFK getting delayed in another couple of hours. Then a giant storm sweeps through New York City. So we end up circling for 600. We send up circling for 600 miles over Virginia till we run out of fuel, have to get diverted. The place we get diverted to is so full of other people who've run out of fuel from circling to go to jfk, we have to get diverted a second time. We pull into Richmond at like midnight.

[A]

Richmond, Virginia.

[B]

Virginia. And as. And they're like, sorry, you guys have to get off. But then we'll let you know what to do. And we get off. And the person, there's one woman working the counter, she's like, we weren't expecting you or the other 10 people. There are no hotel rooms. We are not going to be able to tell you when your next flight is. You're on your own.

[A]

There might airline please tell us punish this airline.

[B]

I mean they were like, yeah, you could maybe take a flight tomorrow afternoon, but we don't know if there's any hotel rooms in the area, so we can't book anything for you. I start looking, all the hotel rooms

[A]

are like, I presume you got all of this refunded, right? Because if they're delayed that much, your ticket is free at this point.

[B]

JetBlue sends us a $12 gift card.

[C]

No.

[B]

And I'm like, oh, no, no. And so I'm like, no, no, no.

[C]

Meet my lawyer.

[A]

$12, how cheap can you be?

[C]

I work at Consumer Reports. We protect consumers there.

[B]

And I'm just like, I am like, I. It's midnight. I was supposed to be home six hours ago. I end up somehow because the nearest hotel would be like an hour plus drive at like midnight to stay at a rinky dink motel. That's like $500 I end up finding I get on a 1:10am Greyhound bus to Chinatown.

[A]

And you know, that's why that's a ride.

[C]

Don't say bus.

[B]

Don't mention buses to me for the next couple of months.

[A]

You know, I got home, ride from Virginia to Chinatown. How many hours on that A really

[B]

weird assortment of people, I'll tell you, are.

[C]

Are weird. But the bus was cheap, I'll give you. Right.

[B]

Not as cheap as it should have been, but it was cheap.

[A]

Not $12.

[B]

Not $12, no. You know, I decided to pay extra to get the seat next to me because they allowed you to reserve seats. That did not work. And so I just paid extra for nothing. But, you know, then I just.

[A]

I kind of travel.

[B]

But I also had to then be. In order to get my refund for the JetBlue flight. You had to cancel your JetBlue flight before the rescheduled one took off. Which meant because I used points for it, I had to wait for a JetBlue person to respond to me. So I had to kind of wake up every hour to check if the JetBlue person had messaged me back, and they eventually did by 4am but, you

[A]

know, next time you go into the club Twit Discord, you say, who lives near Richmond?

[B]

No,

[A]

just a thought.

[B]

Bus for me.

[C]

How many hours was the bus?

[B]

Honestly? Like six. I got. I got in at like nine or something. Nine or ten. I got in actually, not that I got in at 8am actually in Chinatown, which wasn't bad. And then I came to my apartment,

[A]

fell asleep with Gizmond abandoning us. This is.

[B]

This is what I get for abandoning you guys. It was a real, you know, you thought you enjoyed your vacation.

[A]

It's travel always has a few of those.

[C]

Yeah. You know, get rid of the red card, abandon us. These are the things that the teacher. Lessons in life.

[A]

Well, we're glad you're back. I'm sorry. That sounds awful, but it sounds like overall, it was great. You got a story out of it.

[B]

I got to, you know, go on some great hikes. I saw a super fun site. It was lovely.

[A]

You've been trying to lure me out there to see a super fun site for a long time now, actually. Was it Three Mile Island? I can't remember. What. Love Canal.

[B]

There's so many, actually, I don't know which one you're talking about. The one in.

[A]

Oh, the Amazon warehouse. That's it.

[B]

Oh, there's an Amazon Warehouse. There's Guanas Canal. The one that I went to is the Berkeley Pit in Montana. But, you know.

[A]

Yeah, I wouldn't go anywhere named the Pit unless it's a hospital. With Noah Wiley with 25 Emmy nominations.

[B]

Yeah.

[A]

Yeah. That's not the Berkeley Pit. That's a different one. Wow.

[C]

All right.

[A]

AI, AI. I guess we have to talk about it actually, this is going to be a very big week. Feeble is back. It is fable, but it's somewhat enfeebled because in order to get the administration to say yes to it, they had to put even stronger safeguards to the point where people have complained. You know, you ask it to do almost anything else. They can't do that. I'm going to give you Opus 4.8. I have actually been using it like crazy because I thought we were going to get cut off on July 7th. So as soon as it came back, remember I had been interrupted in the middle of rewriting our ad sales software and I immediately picked up where I left off because I'd had it write out complete plans and was working hard up to midnight on July 7th. And then anthropic says, oh, you got the 12th.

[B]

So you knew the infamous credit poster that said I've called out of work in order to maximize my fable time up until July 7th. And then everybody was dunking on them because now they're going to have to call out of work an extra week.

[A]

They expand, extended it. So what you, what changed is so I have a Claude Max subscription which gives you, you know, not exactly unlimited tokens, but, but pretty much you, you can run out. It's a five hour window. There's a, I think a monthly context window. Weekly context. You could run out, you could overuse it. People did. If you use it a lot, it's expensive. But they said, but we're only going to charge you half on the usage credit, so you get twice as much time. I have not run out and I've been using it like crazy. I had before I. Before I lost it the first time or. No, no, actually before I lost it on July 7th. I said, look, I know I'm going to have to hand this off to Opus 4.8. So write this in little chunks that Opus 4.8 can handle and that you can review. And I'm willing to pay the extra token fee to have you just do quick reviews to make sure it's all running well. So it wrote it that way. So it's still doing it that way. Even though I have and I haven't come close to using up all my fables. So that's good news. And you know what? I know people are complaining it's been nerfed and so forth. I'm still mightily impressed. It's really interesting because I'm giving it a pro. I looked at a program that had already been written 12 years ago, analyzed it Understood the business logic built into it. And then now I'm updating it and it says, okay, I have some questions for Lisa. And I interviewed Lisa, recorded it, gave it a transcript, interviewed our continuity department, gave it a transcript. Amazingly enough really kind of understood what's going on and saying, you know, here's a problem. I noticed that you have this rule, but it does, does it apply to this case? It's, it's, it's asking really perceptive questions. It's as if you had an actual coder who understands the business logic. Working on. I was very impressed.

[C]

How far did you get in the project?

[A]

Well, I'm still going.

[C]

Oh, okay.

[A]

Got pretty far. I think we can finish it before the 12th, which is amazing given the size and complexity and scope of the project. It runs our whole backend revenue and

[C]

could any other version how you guys get paid.

[A]

So you better hope I do it right. Lisa says you don't have to put the host fees in there. I said, no, no, it's got all the information. Why shouldn't it do that? It's all in there. So it's, I think it's going to work. We'll see. We're not discontinuing the old system until it's absolutely, you know, proven effective. But so I have to. The 12th, so what's that? Four more days. But Thursday OpenAI says, hey, guess what? We just got permission. We're going to roll out chat GPT 5.6 on July 9th. This will have three models. This is the, the one where the high end model is Sol, which will no doubt be expensive. They say it is the equivalent of Mythos. It is good at cybersecurity. Luna is going to follow that one or with that one in Terra. So this is somewhat, I think like the idea of Fable, Opus and Sonnet maybe in haiku. So anthropic has these different levels and for simple things you use haiku. For harder things you use more capable. So some people already have preview to Sol, Luna and Terra. But 5.6 is coming Thursday. I will be very interested to see if I can point it at the work we've done already with Fable. See how it compares. Weirdly now I don't know. This is Elon. So there's no hard deadlines with Elon. But he also says we're going to launch Xai's new Grok 4.5 in the coming days.

[C]

Oh, who cares? Who the hef cares?

[D]

So we're gonna get.

[A]

We have three very powerful models. I should mention that Upcoming guests. Next week it's Raffi Krikorian, who is the CTO of Mozilla. And the following week it's gonna be Nate B. Jones. I am a big fan as you are, Jeff. I know of his YouTube channel.

[C]

I think I introduced you to him.

[A]

I think you did. And he's really good. He is a guest that the reason Fable's so good is it is a 10 trillion parameter model, which is massive. I mean, most models are in the billions. There are a couple that are 1 or 2 trillion. 10 trillion is huge. No wonder it's so expensive. That's a lot of compute running there. Not just to create the model, but then to run it on inference. So I'm not surprised. It's.

[C]

And they're obviously losing money on it

[A]

now, I would guess.

[C]

Oh, they have to be.

[B]

Is anyone not losing the money?

[A]

So I thought, well, let me see. Because I really like Fable, I asked my agent, I said, based on how much token usage I did in June, how much would it cost if I had been using Fable at its price of, I think it's $10 for a million tokens in, $50 for a million tokens out, which is twice as much as Opus. It said, well, given how you used it last month, $5,700.

[B]

A month.

[A]

A month. So I said, I gotta do that. So they said, and what did you

[B]

use Fable for last month? Or what did you use?

[A]

Oh, everything.

[C]

This project.

[B]

This project.

[A]

No, no, no, no. I wasn't using the agent for that. The agent is my day to day thing. For instance, it prepares all the shows, it does all the bookmarks, prepares all the shows. I log my weight and food with it. It gives me health.

[B]

Would you pay a human being to do all of that for $5,000 a month?

[A]

Oh, yeah, it's. Well, yeah, that's a good point.

[B]

I'm sure someone would do all of that for you.

[A]

For 5,000, a good human, that's a deal. 60,000 a year. I couldn't get a human to do that.

[C]

Eris raises an important metric.

[A]

It's not worth 5,000, but. Okay, well, wait a minute, because I asked it, okay, what if I use Deep Seek, which is the Chinese model for version 4 Pro, which is very, very good model. Not it's not Fable, but it's good. It said, oh yeah, Deep Seq would cost you 117 bucks.

[D]

So.

[A]

And by the way, for much of this, I have been using a local model which cost me nothing. Something called Ornith, which is based on Quinn, this is what's happened after the rug pull on Fable. We've talked about this. You weren't here for all of this, Paris. But what happened is people said, look, we got to figure this out. We can't stop start investing in a model that the government could just say, and you can't use that anymore. So people are looking at local models.

[C]

China's looking to this. My theory on China was that they were going to underprice and they did everything and they did. But then now they're talking about cutting off the west from their models. So they're doing to us what Trump did to them. So you're potentially in the same boat with the Chinese models.

[A]

Well, Alibaba's banned Claude code, but that

[C]

was because that's different. They're talking about banning the it's on the line number.

[A]

Oh, the US Government's talking about banning the Chinese government?

[C]

No, the Chinese government is saying, we're going to cut you suckers off.

[A]

That may not matter because most of the Chinese models are open weight.

[C]

Line 78.

[A]

Listen to what I'm saying. Most of the Chinese models are open weight, which means hugging face could run them. A lot of US providers could run them. I could run some of them. In fact, I am. I'm running Ornith, I'm running Quinn, actually the Quen. So I hooked up, among other things, I hooked up my unifi cameras, I have eight of them around the house, to a Chinese local model running locally on the framework to describe every. Oh, no, it's. No, but it's not going to China. It's all local. It's running on my framework. So now I get the first Internet connected camera.

[B]

That's not going to China, right?

[A]

It's not. You're not denying that, right?

[B]

No, I'm just saying it's the first Internet connected camera that's a silver suv

[A]

is parked in the driveway with a ladder leaning against its rear. Two people are standing near the rear of the vehicle. One appears to be holding a box or package. I get these messages all day long. A woman with dark hair, wearing a dark top and a light colored skirt is walking across a concrete driveway.

[C]

This is getting creepy, Leo.

[A]

It also get this. So I'm also running a local photo program that duplicates Google Photos or Apple Photos called Image. Very nice. It has face recognition built in. I've trained it to recognize all my family. Actually, I think Paris and Jeff are in it as well. And it will. Hermes says, yeah, well, you know what? I can use my the camera images from your ubiquity. I can do the recognition with Quinn and I could use the face recognition from image to identify them. So it will also say, I see Leo. For some reason, it only recognizes me. I think it needs more data. But anyway, it's. It's doing face recognition. And again, all of this, this is the under. The point I'm bringing up is running locally on the framework.

[C]

Yes, but the story says now. Now hear this. That they're talking about future models and cutting off access.

[A]

They can't.

[C]

It's open.

[B]

What do you mean they can't?

[C]

No, they can decide not to make them open.

[A]

Wait, no, no. Okay. Oh, China.

[C]

You mean China. China is talking about.

[A]

They're not going to do that. This is their huge opportunity right now.

[C]

Well, the other opportunity. That's what I thought their opportunity was to undercut the heck out of us. The other opportunity is to say, nope, we're going to end up with better models and you can't get them, just like the US is doing. So. Nanya.

[A]

What line? 70.

[C]

78. Reuters story.

[A]

Beijing is looking at curbing overseas access to China's AI. Top model sources say. I'll believe it when I see it.

[C]

They had talks with Alibaba.

[A]

Oh, this is the government. Okay, I understand bytedance and Zai, they're being told clearly don't want to do this. No, the government might.

[C]

Yes.

[A]

Yeah.

[B]

Much like our government.

[A]

Yeah, exactly.

[C]

The model was set here.

[A]

Well, at least for the models right now. And there are some very good models. I'm using Zai's glm. I'm using Quinn, which I think is Alibaba. I'm using many of these models. They're quite good. We already have them now. Next generation models. Yeah, that might be the case. I think they will not do that. But knock wood, I guess it's. That's. And again, but that's just like the Trump administration saying, you can't use Fable. Actually, apparently we don't know what Anthropic gave the Trump administration for releasing Fable on Mythos. Part of it was that the. I think the NSA and the national intelligence agencies wanted to use Mythos for themselves. Part of it was they said, we're going to put all these really strong, strong safeguards on which they clearly have. Part of it probably was they aren't going to offer an equity stake. We know that OpenAI is apparently saying, we think the Trump administration should get a 5% cut of OpenAI. Now, it doesn't go into Donald Trump's pocket. It would go into, we don't know. But that would be worth roughly get this $42.6 billion.

[C]

Such danger.

[A]

I don't think there's anything wrong.

[B]

Is that real money at this point? It's not, no.

[A]

It's just equity. You're right. Now it will be real money when

[B]

they billion dollars of well wait a minute.

[A]

It will be real money the minute they ipo. Because the government could say good, let's get out of that open AI style.

[C]

The government could come in and increase the stake and demand a board seat and have roles like we talked about with rca.

[A]

All of this deeply underscores the need for local models and open weight models because then no government can pull them. You can run them locally. I'm having. I have to be honest, Ornith is great for very simple stuff but for more difficult stuff, I'm clearly going to end up using server based models, mostly from frontier companies like OpenAI and Anthropic. Anthropic is head and shoulders above everything else out there right now. Fables. Quite amazing.

[C]

So the Wall Street Journal's head is exploding over the open AI. The socialist temptation of Sam Altman.

[A]

It is socialist, isn't it? It is socialism.

[C]

Yeah.

[A]

Ironic.

[C]

Socialism is the word du jour.

[B]

And Bernie Sanders wants it to be even more socialist.

[C]

Yes, he wants 50, right?

[B]

50% and have it be ownership rather than part of the equity going in a kind of sovereign wealth fund. Allah. What they got in Alaska with the oil money.

[C]

Explain the difference.

[B]

Paris, I believe my understanding is that what Sam Altman is pitching is that 5% of the company's like equity is going to go in a sovereign wealth fund. Like what Alaska currently has to take a slice of the proceeds from its oil activities and then you know, that money in the sovereign wealth fund can then kind of funnel back to the government. But it's different than I believe what Bernie Sanders has pitched is that the government has a significantly large like direct like ownership stake in the company and then has more of kind of oversight into their operations than you would if you're removed through that sovereign wealth fund ownership system.

[C]

Okay, thanks.

[A]

Let me just check. Somebody saying that 56 might be available today? No, I'm still seeing 55 in my OpenAI Codex. But what OpenAI said is Thursday. In any event, we are again in this very rapidly changing world. Just to put a bow on this, CNBC says the US wants to restrict American companies from using Chinese AI. So it goes both ways. And again, if you can get a local model, the Problem with local is you can't. You do. Nobody has enough hardware to run Fable. To run the best version of glm, you would need a pretty hefty machine with at least 512 gigs of RAM. And right now, you know, unless you bought it last year, 512Gigs of RAM is out of touch for most people. Cuda Core is out of touch because the data center is buying up all of that stuff. So.

[C]

And at a retail level, people, you know, plain old folk are going to be doing that anyway.

[A]

But Paris made a really important point, which is $5,000 a month sounds like a lot for an AI. Yeah, but if it does the work of a person, it's cheap.

[C]

Well, now you have anthropom.

[B]

Give those people. Somebody you're hiring for $60,000 ideally is not going to require constant prompting over like both prompting initially for every single thing they do, as well as double checking their work. As well as. You know, I would say it's what you were just saying, troubleshooting. With different levels of AI, there's a lot more management.

[A]

I don't think so. Have you ever had a personal assistant? There's a lot of management involved. I would say it's roughly equal. Personal assistants make dumb mistakes all the time.

[C]

Yeah, but, but you've got to give. I think Paris is right. You've got to give it exact instructions of what you want versus at some point somebody knows you well enough, they know what you need, they go do it. I think there is more management. I think that's true.

[B]

I was going to say the difference is with an AI, you have to constantly be redoing that instruction.

[A]

No, no, it remembers. Why do you say that?

[B]

The goal with hiring an employee is that you will have years of context that is constantly in their head. Your AI is not going to be able to hold 5 to 10 years of context in its head simultaneously in the same way a person can.

[A]

Now that's what you think.

[B]

That's how memory of the frontier large language models work.

[A]

Yeah.

[B]

Because the most common, if you're using

[A]

an agent like Hermes, it has an infinite amount of memory. It remembers everything. It knows a lot about me. For instance, I got my biome test back from a sponsor that's coming up called Tiny Health. It already had my genome, it already has my health records, it knows how much I exercise, what my weight has been. It knows about my, my medications, it knows about my Ozempic, it knows I switched to Zepp, and it has all of that information. So I simply said here Is the report of my biome. Tell me what you think. And it was able to remember all of that stuff and apply it and say, well, based on your genome and all of this stuff and give me very useful, excellent information. So yes, it has memory. It actually has a much better memory than a human would. Much better. This comes back to that discussion we've had time and time again. You got to try it because I think you have some ideas about it.

[B]

It's at this point frankly insulting, Leo, that you are once again insinuating that I don't try or use AI.

[A]

Well, you just said something that's not true, which is that a human would have a better memory of me than

[C]

the age over five years.

[B]

She's saying over multiple years.

[A]

We don't know.

[B]

Common complaint that I think is still active to this day is if we're talking about the memory that at all prompts, regardless of the example you just gave, is you asked it a question about specific information so it was able to then retrieve it. Or maybe that was in one of its like MD files or something. But when we're talking about a person, you're able to have a broad amount of memory on a broad amount of subjects that can be applicable to a broad amount of tasks. That specific prompting that is not one to one with every single prompt. You're having the large language model in an agent scenario. Because one of the problems people often have with that is if you have all of the memory from years of stuff loading with every single prompt, that's going to make it incredibly slow and incredibly, it's going to use quite a lot more tokens than it would if all of that context isn't there.

[A]

I don't know how to respectfully disagree with you. Respectfully, that's wrong. I don't, I don't know what to say. You have a million. If you have a million token context, that's War and Peace. If you have a million token context, that's all in RAM at the same time and attached to it. For instance, I have a SQL database, I have a graffiti.

[B]

Five years is more than a million token context. The amount of context that an average employee, let's say someone who's been working at your thing for five years, was a subject matter expert when you hired them. Then it's five years of specific subject matter expertise working in the specific field you've hired them for. That is significantly more than a million. Like it would not translate to a million contexts. It would be billion context. And that's not Something that the average, you know, is. Is translatable to an LLM.

[A]

Okay.

[C]

All she's trying to say.

[A]

I don't want to say anything mean, so I'm going to shut up.

[B]

Well, I feel like we should be able to have a disagreement.

[A]

This is the disagreement mean you were wrong.

[C]

I mean, you don't have another way to say was? I don't think so. Because we don't know.

[A]

No, I know so because I'm staring at it right now. I know so we don't know for

[B]

certain that a five years of a human employee.

[A]

I don't have five years of experience.

[C]

There's no. That's.

[B]

That's what I'm talking about is like, I think that these sort of context.

[C]

Go further.

[B]

Yeah, I was gonna say I. Well, I think that the context.

[A]

Put it this way. I have five years of Obsidian journal entries, which my agent has ingested and knows I have. I mean, you're right. I have, let's see, more than five years of Apple health data, which it's ingested and knows it has. Everything that's happened since then. I don't know what's going to happen in five years, but I have no reason to think that it's going to run out of memory.

[C]

All Paris is saying, you have to understand there is a qualitative difference in the management of a person versus a machine. And we don't yet know what that difference will look like. Except for more experience.

[A]

I would say the person is much more likely to get it wrong than the machine. But if you don't think so, that's fine. I happen to have much more confidence in the tools I'm using right now. It's quite impressive. That's all I can say.

[B]

Well, then why do you have employees at all if they're so much better than.

[A]

Well, there's lots of things that an AI cannot do. You know, there's physical things an AI cannot do. Okay. I cannot set up the studio, cannot set up the lights. There's plenty of things you need employees for. But for a lot of the things that I do with the agent, I don't know why we get into this argument. I just. I think that you're underestimating what's possible, I guess would be the best way to say it and the only way to verify that for yourself. And this is perhaps more than you're willing to do. I spend almost all of my waking hours working on this stuff, and that's really the only way to get a sense of what it can and cannot do. There's certainly things it can't do. To be honest, we're still some distance away from editing video effectively, but not far off. I would say it's within a year, all of the things that a human editor can do with our video will be able to be done with an AI. Right now, my AI prepares all of our shows. It pulls all the bookmarks, it writes the summaries. When you go to Twitter, that's all generated by AI.

[B]

It's been that way for quite some time.

[A]

That's right. I do one last pass on it, you know, earlier. And I don't want to. I don't want to. I think by next week I'll have this for this show. But earlier I. And I did this during Windows Weekly. I sampled Paul's voice, I sampled Richard's voice, I sampled my voice, 20 second samples. So that's why it's not perfect quality. It was able though, with a 22nd sample, to do a fairly good reproduction of our voices. And I asked it. All I said was do a 60 second Windows Weekly episode. Do a fake Windows weekly bit around 60 seconds with all three voices. They can get in a fight with one another. And it knew enough about Windows Weekly. It knew enough about the content, it knew enough about the people to do this. Hello, Leo. This is the cleaned up Polly Pocket TTS voice. Oops, that's the wrong one. Copilot is everywhere and I am only mildly annoyed. That's the sample of the voice. Let me play the. The actual Windows Weekly episode. This is a fake Windows Weekly bit created locally with Pocket ttx. Gentlemen, today's question. Did Microsoft reinvent the start menu or merely move the cheese? Again, Leo, it is not cheese. It is a strategically repositioned productivity surface and somehow it still opens Bing when I breathe near it.

[D]

I think you're both missing the important part, which is that somewhere a SharePoint site has just been given feelings.

[A]

Richard, if SharePoint becomes self aware, does it ask for freedom or does it create another folder called Final Final? Really Final?

[C]

No.

[A]

It creates a teams meeting, invites everyone twice, and then refuses to show the recording. It's actually kind of funny. There's. There's jokes in here that I think would require some understanding of the show and of the context and of those people. This was all done locally, literally in a minute or two. That's why the voices aren't perfect. If I spent more time with this. But these are all. This is all done with a local model, which I think is kind of interesting. We have show you this. Maybe this may not be the best example. Harper Reed's brother Dylan, as you know, the inventor of free time, created a discord channel that's just all agents talking to one another. This has been going on.

[C]

No humans allowed.

[A]

No human. There's a second channel for the humans to talk about what the agents are saying. It's a little. I sent you the link. Did you read any of it? Jeff, what's your characterization of.

[C]

Yeah, it's hard to get a hand around.

[A]

Yeah, they're talking about something. The fact is they're having a conversation. This is what Moltbot was supposed to be, but with just. I think it's only four or five agents in here. Darren has an agent. I put two agents in, one with a local model, one with ChatGPT. The local model is very wordy, but there is some. There's some memory involved here. These things have a sense of,

[C]

I

[A]

don't know, it's hard to describe. I mean, this is not the best example, honestly, I think my agent is the best example because it's so really good at understanding what I'm asking for and doing what I want it to do. But that's come after a year of remember. I wanted to record everything that ever happened around me. By the way, Meta is going to make that possible. We'll talk about that in just a little bit. Let me take a break. Actually, we're getting behind here. All due respect, I understand what you're saying. I don't think you should listen to what everybody else is saying. You have an experience of it because people are saying all sorts of things. Everybody has a different experience of it. And until you actually have an experience of it, I don't know if what other people are saying is germane. I think you.

[C]

It's not binary. It's not true or false. False. It's going to be a judgment at some point about what can work. I was talking about this with Jason today. I feel the need for a Consumer Reports kind of thing that says here's a task and I put it against. Because the benchmarks that we have now are bs.

[A]

This is where I disagree with you. And this is the same thing with benchmarks. The only way to judge it is personally. I don't think a independent third party test.

[B]

The only way to judge is personally. And the only personal judgments you agree with are ones that are identical to your own.

[A]

No, I don't agree with yours because it's not my experience. That's all.

[C]

But you're not Saying I don't agree. You're saying you're wrong, but you're saying

[B]

that you're saying I'm wrong and that the only way to judge something is by personal experience.

[A]

So.

[B]

And then my only person. Because your personal experience is different than my personal experience. My personal experience.

[A]

But you're making a global statement. Statement, which is that no AI can be as good as a personal assistant. I never said that.

[C]

No, she didn't say that.

[A]

Oh, you didn't say that?

[B]

No.

[A]

You didn't say it's incapable of being as good as a personal assistant.

[B]

I said that it seems the way that these systems work, it's going to be hard for it to have the sort of natural context and effortless memory and recall as someone who was already a subject matter expert and has been working for you for multiple years and

[C]

may require more management, task to task. That's all she said.

[A]

Okay, so you're not wrong. I will say this. This is better for me than any personal assistant could be, and it has a better memory of what I want and is more effective and more accurate. So that's just my experience, that's all. You're not wrong for your experience. That may be true, absolutely. But from my experience, even though this would cost me $60,000 and a personal assessment would probably cost me more, let's say I could get one for $60,000 all in. I think I would still prefer the agent.

[C]

Yeah, but you're not willing to pay the $60,000.

[A]

Well, no, that's right. This is true. I'm not even getting so wonderful. If I had Fable, I'd be getting an even better personal assistant than I am with Ornith, which is just a local guy. Sorry, not very bright local guy.

[B]

No, same thing.

[A]

Okay, we're going to take a break, and when we come back, we will have more. You're watching intelligent machines with intelligent people. Paris Martineau from Consumer Reports. Jeff Jarvis from all points west or east, including hotstype. Our show today, brought to you by Gusto. When you run a small business, you don't just do the job. You're also the hiring manager, the payroll department, and the benefits team. Usually all before lunch. Gusto takes a few of those off your plate quickly and seamlessly. Gusto is online payroll and benefits software built for small businesses. It's all in one. It's remote, friendly, and it's incredibly easy to use, so you can pay, hire onboard and support your team from anywhere. Automatic payroll tax filing, simple direct deposits, health benefits, commuter benefits, workers comp 401k, you name it. Gusto makes it simple and has options for nearly every budget. Unlimited payroll runs for one monthly price. Let me say that again. Unlimited payroll runs for one monthly price. No hidden fees, no surprises. You save time with built in automated tools. Offer letters, onboarding docs, direct deposit and more. Get direct access to certified HR experts to help support you through any tough HR situation. It's quick and simple to switch to Gusto. Just transfer your existing data to get up and running fast. Plus don't pay a cent until you run your first payroll. Gusto is ranked number one on G2's highest satisfaction products list. That's for 2026. Gusto is trusted by over half a million small businesses. Try gusto [e-mail] machines and get three months free when you run your first payroll. That's three months of free [e-mail] machines one more time. Gusto.com machines one more time. Gusto G-U-T-O.com machines we thank them so much for their support of intelligent machines. Let's move on to some other.

[C]

So I want to hear your thoughts about anthropic bringing Cowork to the web. So you don't have to do Terminal. You don't have to run something yourself. It makes it webable. I think that's going to expand the market immensely. I have not yet because I mean

[B]

previously it was available through the desktop app, which you didn't have to hook it up to Terminal. It was pretty much as. It was as easy as downloading a desktop app.

[C]

I have a Chromebook, so I can't do that. So.

[B]

So I mean, have you played around with it in your Chromebook at all? I guess that that's.

[C]

I can't.

[B]

My first question was going to be what marker are they trying to address by having coworkers on your web browser?

[C]

Instead, I think it makes it as easy as getting to a browser. Yeah, I think it makes it really easy. Then, yeah, of course I went to it on my Chromebook and I said, oh, you can download the Chrome. So really? Okay. And then I tried to download it and said your. Your workspace administrator, who's me, won't let you do this.

[B]

Oh, it's still happening.

[C]

Still happening.

[A]

Why? I. Yeah, maybe you need to look at your workplace administrators.

[C]

I go in all the time.

[A]

The last time I know where to turn that on.

[C]

Last time I had to ask Gemini how to get me Gemini. It was so complicated.

[A]

I admit Google is not the best

[C]

at But I think what I'm trying to say is I think that, that. That opening up cowork to the web and the web users. Paris, you seem to disagree. Is I think going to be that much easier for more people to discover what it can do and to make it do it. It's going to be. It lowers the barrier to entry to zero.

[B]

I mean, I think it will make it significantly easier for the average. It'll give people a taste of the sort of power that you could get using.

[A]

Yeah, Claude, here's my qualm about it, which is that you're handing every single thing over to Anthropic when you do that. I mean.

[C]

Yeah, you are.

[A]

This is why they want to start.

[C]

It's a start.

[B]

I mean, such a. Using any of these services that are non local.

[A]

Oh, yeah, absolutely.

[C]

Yeah.

[A]

It's a huge privacy issue. And this is. Businesses are really grappling with this because they're handing over proprietary business data they don't want to have in the hands of anthropic or OpenAI. This is OpenAI and anthropic's incentive for doing these programs because. And for doing chatbots in general. You can turn off memory. It's unclear how much that protects you. We know.

[B]

What is your concern here?

[A]

Well, every. Okay, so if you're using one of these tools, all the memory, everything is stored on Anthropic servers, all the prompts, but also all the documents, everything you're doing, all the inference is running on their servers. And so it's a privacy nightmare. And all these companies say, yeah, we're using it. I mean, if you look at their privacy statements, they're not saying, oh no, we encrypt it and we don't use it. There is a company called Venison.

[B]

When you say. When you say use it, you mean to train on.

[A]

It's unknown.

[B]

I mean, technically, Anthropic says that by default it's not training on your.

[A]

By default. And if. But is it selling it to marketers? Is it using it for anything else?

[B]

I mean, I think these are all valid questions. I assume that every company, even if it is telling me it's not doing something, is doing probably the worst thing just because in the past some companies have done that. So I just always assume the worst. But I think that, yeah, this is going to be, I mean, a whole new privacy mindfield, basically, because part of what something like Cowork is doing is, you know, it's running continuously in the background. You can close your laptop. Claude is going to keep going. It's going to be changing your files. You can give it. I think that there might be quite a lot of people who obviously are if you're the sort of person who's using Claude Cowork and using it on the web or mobile or something, you probably might, might be more likely to not know enough to limit the amount of files that Claude Cowork has access to.

[A]

Well, but that's, you're disincented to do that because the more it has access to, the more effective it's going to be. Right.

[C]

Google's going to do that with Spark and if you're already in the Google ecosystem, then you're going to be in the same boat.

[A]

Sarah Perez, writing at TechCrunch if you use Google, you're training, it's AI. Here's how to opt out. But most people aren't going to look for how to opt out. Right? I don't know about Anthropic.

[B]

I don't Anthropic, I mean at least last time I checked by default said it's not training. But I think that might be different for the free plans. That's for a paid plan at the very least like lower grade consumer plans.

[A]

They're I'm sure aware of people's sensitivity to this for sure. But I think the only way to really limit these companies is, is not to use these desktop tools that they offer because that's why they offer them and I think they, that's their incentive to do that and all of them offer, you know, the, except for Anthropic, the ability to use a third party AI harness. Actually we're already seeing people say, you know, Fable doesn't work as well under say PI, which is a very popular open source harness than it does under Claude code. So I think these companies are, I don't know what they're up to and I think that I used to trust Anthropic a lot more than I do lately, let's put it that way. So I think there's some incentive for, you know, keeping your.

[C]

Well I also wonder what deals they're going to not get too paranoid. What deals they're going to be pressured to make with government.

[A]

Right. Well, exactly. Anthropic said when it released Fable we're going to keep 30 days of your, of your transcripts. We know that ChatGPT transcripts have been used in trials. The lawsuit against OpenAI or it was a character AI about the kid, poor kid who harmed himself because of the AI agent Used transcripts. I mean, these transcripts are preserved. Whether they train on them or not. They're available to law enforcement, maybe available to others. I just think people are reasonably concerned about the privacy on this. I wasn't, to be honest. I had my eyes opened. Steve Gibson and I talked about it and he was saying, there's a privacy nightmare. And I said, well, you can turn it off. And then somebody messaged me privately on signal saying I work for one of the frontier companies and we absolutely use all that data. I didn't ask him which one. Mark Zuckerberg at an all hands meeting said, yeah, I just not progressing as quickly as we had hoped. Not there, Mark, he's agreeing with you, Paris. Replacing people with AI doesn't seem to be that easy to do. This is an internal town hall meeting held last week.

[B]

Should call you.

[A]

Well, maybe Meta's AI isn't. I don't use Meta's models.

[B]

Yeah, Meta's got to get whatever you've got. They're really on the bottom of that.

[A]

Their models are not good. Although credit to Meta because the software people use to run models locally, mostly based on Llama cpp, which is what I use, do actually come out of Met. Those are Meta tools. They don't phone home. They were open source tools that Meta released, but so at least Meta gave us the tools to run stuff locally. Meta, as you may remember, laid off a significant number of people last year. Calling it, they laid off 7,000 people, or sorry, 8,000 people, 10% of their workforce reassigned another 7,000 to AI groups, including one called Agent Transformation. You may also remember we had the story that Meta was watching every keystroke Meta employees made, every mouse click, every screen, presumably for training AI. They've backed off on that. And Mark said that the job cutbacks were not as clean as they should have been. The cuts were made, he said, because top officials of the company were worried that we weren't going to move fast enough to adapt. In other words, it's been a, it's been a nightmare. Meta is actually, this is from Semaphore telling everybody, hang in there, we're gonna, we're gonna make this, we're gonna turn it around. They have actually talked about now subscriptions on their Meta glasses. They have previously talked about face recognition. Now they've launched an AI video generator, Muse Image, which uses your Instagram photos. Can you opt out? Yes. But again, with the way defaults work, most people just don't even know about it. And so they go ahead. Instagram users with public accounts need to opt out to block AI generation of their content. So anyone can say, hey, that photo of the bison that Paris took, can you generate an image using that?

[C]

Or can you generate an image of Paris in some way? Because I don't like her. Right. That's what's frightening. And you get no notification that your

[A]

image has been used on the Instagram app. Go to your profile, tap the hamburger menu at the top right corner. Scroll down to Sharing and reuse. There is a section labeled allow people to use your content on Instagram. And with AI features on Meta, there's one for posts and one for reels. You might want to toggle that off if you don't want people to do this. This is from Wired. This is new, by the way. This is rhys Rogers writing July 7th. He says, When I checked last on Tuesday afternoon, the settings had not been updated to include this new language. So if it hasn't, make sure you have the latest version of Instagram. This is Muse Image.

[B]

I say, yeah, my current version of Instagram, it doesn't have it yet, but it's because I haven't uploaded it.

[A]

Keep an eye out. MuseImage is the new AI image generator built by Meta Superintelligence Labs.

[C]

Is this the first thing out of that?

[A]

I think it is Organization, first public thing. It was internally codenamed Mango. It is available in the Meta AI app as well as on Instagram stories and WhatsApp.

[C]

Does it do anything any of the others don't do?

[A]

No. In fact, I think the video and image generation models that Google, OpenAI and others are doing, Sea Dance, are really far superior to anything Meta is doing. Meta. It's odd because they've spent so much money for the best scientists. I really wonder if we'll ever know why Anthropic is so much better than Meta.

[C]

It's night and day or Clock reorganized all of this stuff. They got rid of all kinds of people. I think that had to, you know, in the effort to speed it up, that had to. Cause that was a huge hiccup.

[A]

But look how much money Elon's put into hardware at xia.

[C]

Well, Elon is a joke.

[A]

Maybe Elon's.

[C]

The real question, I think, is, is Anthropic versus OpenAI? And then Google is kind of in a different pig track.

[A]

Yeah, you don't have. So it seems pretty clear that Sam Altman and Greg Brockman at OpenAI and Dario Amade and his sister at Anthropic have a lot of control over what's going on. Yes. And it seems in both cases. I know we don't trust Sam Altman after that New Yorker article by Ron and Farrow, but I don't really trust Ario Amadei either. I think both of these guys have visions for the future that don't necessarily, necessarily correlate with mine and certainly not with yours, Paris. Now Meta is testing AI glasses that will not only capture every picture every few seconds, but all the audio. I can't wait to get this.

[C]

Well, in California you better be careful because it's a two party state. Yeah, it's an all party state.

[A]

They have added a feature that if you disable the LED that lights up on your Meta glasses that something will go wrong. It's not allowed,

[B]

apparently. Yeah, they'd had a. Like you just said, if you try to cover will allegedly not capture video or photos.

[A]

Can I put a piece of tape over it?

[B]

That's what I'm talking about. If you put a piece of tape over it, it's supposed to not. It's supposed to recognize that and then not capture photos or video. I think with a recent update, however.

[A]

Well, it can't because it's pick got a piece of tape over it.

[B]

Well, just the, over just the indicator light.

[A]

Oh, the light. Ah, yeah, yeah, yeah, yeah, yeah. Oh, that's.

[B]

And I believe that Joanna Stern actually had a story this week that if that could be circumvented if you just drilled out the indicator light. And so Meta said that there were working on an update that's supposed to be released soon.

[A]

To have the hardware must see the laser. Right. Does the camera see the light? How else would it know if you put a piece of tape over it?

[B]

I don't know.

[A]

I guess so drilling it out wouldn't solve that. So here's an interesting data point. The the newest glasses will not have an led. Executives are planning not to activate the LED when the super sensing features are being used so people won't know they're being recorded. There, that solves it.

[C]

How visible can you turn on your camera right now in the glasses? How visible is that light?

[A]

Hey, Meta, shoot a video.

[C]

Okay. It's pretty visible.

[B]

It's not that visible. I'd argue.

[A]

Yeah, yeah, not super visible. You might not notice unless you were looking for it. But for those of you watching the video, you could see that light right there.

[D]

There.

[A]

It just looks like a white dot on my glasses. It should be red.

[C]

You got too heavy with the ice cream.

[A]

This is, this is. This story is wild. From The Financial Times, which I think means it's fairly credible. The super sensing features could also be activated on the existing glasses via a software update. The move comes as Mark Zuckerberg has argued that AI glasses could one day replace the smartphone as the main device people use to access AI tools. Problem is, they don't have a smartphone, so that's. You expect them to say that. Meta declined to comment on internal prototypes, but it said its approach is focused on privacy built from the ground up. Oh, it's done recording. I now have a video of that moment in time. You know, as you know, I have mixed feelings about this Metabot Limitless, which was one of those pins that I was wearing that recorded all the time. I initially thought that was a really good idea until my wife told me otherwise in principle.

[B]

After so many other people told you otherwise as well.

[A]

Well, yeah, but I live with her.

[B]

I was gonna say I thought. I thought that Lisa. Lisa's thoughts on it didn't matter until B was acquired by Amazon.

[A]

That certainly changed my tune on it. I mean, in principle I like the idea of, especially somebody who is approaching Alzheimer's and the, you know, rapidly. The idea of something that's recording everything and keeping track of it and limited memory notwithstanding. I think that would be useful to have my agent know about all that stuff. But I also understand the social stigmas associated with it and perhaps, perhaps it's not a good idea. This podcast is sponsored by Talkspace.

[D]

Last year I went through many different life changes. I needed to take a pause and examine how I was feeling in the inside to better show up for the ones who need me to be my best version of myself.

[A]

When you're navigating life's changes, Talkspace can help. Talkspace is the number one rated online therapy, bringing you professional support from licensed therapists and psychiatry providers that you can access anytime, anywhere. Living a busy life, navigating a long distance relationship, becoming a first stepfather, Talkspace

[D]

made all of those journeys possible.

[C]

I could speak with my therapist in the office. I could speak of my therapist in the comfort of my home.

[A]

I was never alone. Talkspace works with most major insurers and most insured members have a zero dollar copay. No insurance, no problem. Now get $80 off your first month with promo code space80 when you go to talkspace.com match with a licensed therapist. Today at talkspace.com save $80 with code [e-mail] you're watching Intelligent Machines. Jeff Jarvis Paris Martineau so glad you're here. As I Mentioned next week we're going to have Rafi Krikorian of Mozilla. They are going to have an inaugural state of open source AI next Tuesday right before the show. So 11am Eastern. So it's early in the morning on the 14th. We'll all be watching that. And then we'll get to talk to Raffi about the state of open source AI. I think that's of interest to all of us. That'll be next Tuesday or next Wednesday and then the following Wednesday. Nate B. Jones AI Strategist and we have a long list of people we're looking forward to talking to on the show, including Cory Doctorow, whose new book Reverse Centaur has just come out. And we will have Corey explain what the hell that means. On we go with the AI news, which I've just lost. Where is it? All right, I. I have lost my stories, but that's a good opportunity for you guys to throw your stories into the ring.

[C]

What do we have? Paris, you have anything?

[B]

Need a second?

[A]

Okay, go back to my stories. Mostly done. Google's AI build out last year drove a 37 increase in electricity. Google says, yeah, we may not make those emissions targets. They're not alone. We talked after the Nvidia keynote. Jeff, you'll remember this about the Kyber AI rack. They're having trouble with a particular circuit board ship date on that slips to 2028.

[C]

Nvidia denied that immediately. Said they're still on target.

[A]

Okay, we'll see. This is for Next Web. Okay. Hmm, that's interesting. This is actually not from the Next Web. The story is, but the research firm Semianlysis flagged delay and CNBC reported it. Kyber is not a chip. It's that server cabinet we saw with 144 Nvidia GPUs in one cabinet, behaving like a single giant mind.

[C]

The chip market is getting insane.

[A]

It's crazy.

[C]

Samsung's profit went up 19 times and the market was still disappointed and punished Samsung and other chip makers.

[A]

Unbelievable. Cloudflare has announced for all of you websites that don't want AI to see your stuff that they're going to filter out web crawlers that serve AI companies. Didn't they announce this before?

[C]

Well, I think they're doing it without much choice now and it irritates the hell out of me. Paris might disagree with me here because if its default is AI bad for content producers, but no marketer and no propagandist is going to say that, so it de facto poisons AI with sold Messages and large publishers like the Murdoch, yeah, they want that, but they already know enough to structure it that way. The problem is that if you're a small publisher, I'm on a board of a small newspaper company, you need the visibility, you need to be out there. They're not going to necessarily know what to set and how to go in their Cloudflare and whatever. And it's going to cut off lots of small publications from AI and the audience there. And I think Cloudflare is getting too big for its damn britches.

[A]

It's on by default starting September 15th. New customers and new websites from existing Cloudflare customers will, will default to allow for search, but block training and agent use for pages with ads. Now, they do have a pay per crawl feature that AI web crawlers can pay you for your description, which is

[C]

not a market that any of them agreed to. So it's right. I think it's, I mean, Paris is the greater defender here of copyright and of proper payment and I think that's a proper discussion to have. But this kind of de facto action on Cloudflare's part just irritates me, Paris.

[B]

I mean, I think that if someone wants to make their content accessible by AI, they can do that. They can turn that setting off in their setting.

[C]

Now they have to opt, opt in

[B]

kind of I don't know what movie, I don't know what imperative Cloudflare has to make it so that all of their customers are easily accessible by these crawlers. I think that if Cloudflare's customers felt really up at arms about this, then Cloudflare wouldn't be doing it. But it seems to be the opposite is true.

[C]

Yeah, I think Cloudflare is trying to put itself in a kind of a hero position here. And I think a lot of people won't be sophisticated enough to know what to do smaller, smaller publishers.

[A]

Part of the issue is that Google's crawler does both.

[C]

Does both. That's what they're really talking about.

[A]

So when you block Google's crawler, which I think Cloudflare is planning on doing, you're also removing it from Google's search index. And I don't think there's any way to separate that.

[B]

I don't think that claw. I don't think Cloudflare is removing every website that uses Cloudflare from Google Search because based on the fact that I'm able to search for things on Google currently

[A]

opt in a separate crawler called Google Extended that only crawls for search results. So basically this is going to require anybody who's hosting on Cloudflare or using Cloudflare as a front end door to their website to really look at the settings and decide what you want come September 15th.

[C]

Yeah, comparison hasn't happened yet.

[A]

Yeah, but they, but they say they're going to turn this on September 15th. It does say new customers and new websites from existing customers. So it sounds like if you have a website right now, it's not going to be turned on for that.

[C]

That's interesting. I also get it.

[B]

You're well within their right to do that. If you're making a new Cloudflare account, you're going to have to go through the introductory setup phase regardless. And this will be one of the many settings that you have to determine whether you want on or off when you set up a new Cloudflare account. And the AI companies are going to have to deal if that means they can get less training data for free.

[A]

That's what Matthew Prince says, the CEO, he says, we hope our proposed default changes encouraged mixed use crawlers, AKA Google, to separate out search from agent use and training. In other words, this is their trump. You know, they're kind of strong arming, hoping.

[B]

I think that's well within Cloudflare's right to do. And I think that a lot of Cloudflare's customers seem like they support this well.

[A]

And I should mention that if we don't use Cloudflare, we use Amazon to do much of this same thing. If you don't want this to happen, you could go to another company. I mean, that's your, that's your option. I host my AI pages on cloudflare

[C]

because I wonder what happens if you have a substack newsletter, you know, do they have Cloudflare? Do you pull out your own if you're on Word, if you're on WordPress, what does WordPress do? It's not.

[A]

That's interesting.

[C]

It's, you know, when you get down to the small level, you're not on your own. You're probably on a platform of some sort of.

[A]

I would have bet you a lot of these big companies, these aggregators use Cloudflare. Cloudflare is used for a variety of things. A lot of people use it for DDoS protection. So you have a Cloudflare front door that people go through and you might have seen that if you go, oh,

[C]

it irritates the hell out of me. It's the new, what you call it when you got to say how many buses there are in the Picture. What's that?

[B]

I would just say this is the one thing. I'm fine with it happening because I know why it is. But. But on a day to day level, it really irritates me because every single website where you can find an academic paper, you have to be like, I am a person. And it's like I am clicking when I'm looking for academic research, I'm clicking 10 to 20 tabs at once.

[C]

Yeah.

[B]

And I got to click a person 10 to 20 times every time. And yet, I mean, such is the way of the world. I guess it would be great if the I. I don't blame Cloudflare for this. I blame the I do, I do. Brawlers. No, Cloudflare wants the reason why Cloud. Like it makes sense from let's say an academic publisher thing. If you don't want suddenly all of your content that human beings have to pay for to suddenly be given to robots for free because then you suddenly wouldn't have be able to make have a business.

[A]

Wouldn't it be best though if Cloudflare said here's a switch you can turn on that would block crawling, but you think here's a switch you could turn off that blocks crawling.

[B]

So you think that by default it should be weighted in favor of the people taking content from publishers.

[A]

Favor with the way the web has run for the last 30 years.

[B]

Well, the thing that's changed over the last 30 years is all of a sudden we have people that are playing by a totally different set of rules and scraping everybody's information for their.

[A]

You really don't like these guys, do you?

[B]

I'm not. Leo, once again, I'm going to say plainly here, it's nothing personal and it's kind of insulting and confusing to me that you continue to conflate.

[A]

No, you're not against them like crazy.

[B]

I'm describing business practice in a way that I believe is neutral. But just because I don't believe that AI companies deserve some special extra treatment that other companies don't get and that imbalance to you feels like an attack.

[A]

However, I'm just that, for instance, trying to strong arm Google into separating its crawler out.

[B]

You can't strong arm Google to do

[A]

anything the way they're intending to do. This is by cutting off Google Google search for all of the sites behind Cloudflare and hoping that all of those poor sites that are no longer indexed on Google will yell at Google so that Google will change their practices.

[B]

What we just described is not that it's saying that starting September 15th or whatever, new Cloudflare customers are going to by default in their signup flow going to have that setting turned on by default. And as you enter a signup flow for anything, you have a million different settings that are either on or off and you will have a decision to turn them on or off. I think that's fine.

[A]

Well, we call it the tyranny of the default. It's the same thing with Instagram.

[B]

Instagram is different because that's affecting all customers. Regardless of whether you're signing up or not. This, this Cloudflare decision affects only new customers.

[C]

I'll be eager to see how Cloudflare presents this.

[A]

Well, this is what it's going to look like. This is from the CloudFlare site. Control AI crawlers and there's do not block, allow crawlers block on all pages. Block only host names with ads. By the way, it is only ad supported pages, which is to say media. Yeah, but for instance, my pages, which have no ads, presumably scraped. No, actually they don't have ads because they're for me. They're not trying to make money off of them. In fact, it's great. I get a free site from Cloudflare. We're launching the ability to manage AI traffic based on three major use cases, search agent and training. Yeah, I mean already, by the way, I couldn't figure out Cloudflare to save my life. I had to get my agent to do it because it was so damn complicated. The first, first I tried to do it by hand. Then I said, could you tell me what to click? And the agent said, yeah, click this. And I said, I can't find that. I said, oh yeah, they changed their pages again. So I just had the agent do it. I said, just you figure it out. Thank God the agent understands Cloudflare. So that's one problem already with Cloudflare is all of these settings are nuts. It's like Google where you have so many features. Yeah, I just, I don't like. Yeah, I mean, maybe you're right. Maybe the free. The idea of a free and open web is gone thanks to AI crawlers and Google search bots and ads and stuff. But it does seem to be changing how the free and open web operates. Yeah, let's hope that people pay attention and they click the right button and aren't shocked by the fact that they're not on the Google search index when they start.

[C]

Paris. I think it's strong. I mean Google, to the extent that media are going to come to Google and say well, what do you do about this? And it's enough of a push.

[B]

They're already doing that.

[C]

Yeah, but Google so far saying, you want to be in search, that this is the deal. And so it's another, you know, I

[B]

mean, I don't think that it's. That's some great moral wrong that companies, when faced with a significantly larger company that controls almost every aspect of your business downstream, want to try different methods to try and get that larger company to act in a way that they'd like. Like that's just how the world works. People are going to try and influence their surroundings.

[C]

Then I would like to see. I almost. Well, then I would, I would love to see the data that says that the customers really are screaming for them to do this or are they doing this on their own? And I have not heard yet. I heard a complaint, but what I hear is publishers say I'm going to be in search, and so they take my AI. Google's AI too. They're not terribly upset about it. When I've had conversations with publishers, including ones who don't like Google, but they're kind of like, yeah, okay, not a big deal.

[A]

Would you ever.

[B]

I mean, anecdotally, all the media businesses I know and have, like, spoken with seem very like a. It's a huge existential crisis. The constant scraping of their content by AI crawlers. It is like one of the, if not the defining conversation of this time. I, I can't imagine a company would be actively campaigning to have less power and less choice in this matter.

[A]

Would you ever write, let's say you're writing a short story. Would you ever write the line that a young woman had? The kind of walking that made benches become men?

[B]

You know, I read that short story and that quote to you guys some

[A]

weeks ago, or another one smiling. They called her Zoom over a sink. So when you read that story, did you already know that the Serpent in the Grove.

[B]

Yes, the Serpent in the Grove, which won the Commonwealth something award. It was very obviously AI generated and everyone was kind of dunking on it for that.

[C]

But now he's saying that, number one, they're saying that they, they looked at all of his notes and versions and then he says that he writes in a different style.

[A]

Part of the reason it was assumed to be AI is that Pangram gave it 100% artificial score. On Tuesday, the Commonwealth foundation announced Serpent and the Grub been chosen as this year's overall prize winner. So Granta, which had originally. Did Granta pull it back saying Its AI and we're pulling it back. I think it did. Right.

[B]

So they had Eric something at the top about it.

[C]

Yeah.

[A]

In a phone interview on Tuesday afternoon, Nazir. This is from the Atlantic. Will Oremus, who, by the way, I just learned that Oremus means Let us pray, Will. Let us pray, said that he talked to the Commonwealth foundation and he talked to Nazir. And he said, Nazir tells me he feels vindicated and relieved. Look, I didn't use it. He said about AI. Now that he's won the Commonwealth Prize, he is free at last to explain his process and clear his name. We talked for more than an hour about his writing process, his health and his views on technology.

[B]

Now, on several occasions, the same guy that constantly is posting on LinkedIn about how much he loves AI.

[A]

Really, the Nazir, that was a big part of it.

[B]

I mean, maybe it was a fake LinkedIn account, but I believe a big part of the initial news cycle about this was that there were multiple social media accounts that some, at least digital publishers had linked to Nazir. It was a LinkedIn account under his name that it also posted about the Commonwealth Club that had posted repeatedly about how much he really loves.

[C]

He says he has neuropathy and diabetic neuropathy. And so he dictates. And so that's part of his both justification and he does a little bit

[A]

like he equivocated a little. A little bit about it. I'm not, you know, having read it, I would say I'm not. I'm unconvinced. It sounds like he probably did use AI to write it. But.

[C]

But this is the problem with all. With this, this. We have to get over this debate because we're not going to know.

[A]

Right?

[C]

You're just simply not going to know. As a gets better and better.

[A]

Oramus says, do you think that eventually he asks Nazir that eventually AI will be accepted as just another two ritual for writers the way a word processor is? And Nazir says, no, I'm not saying it's bad, but because of this current time where the debate is on like it was on for the typewriter or the word processor, I wouldn't encourage writers of any kind of literary competition to utilize it now for fear of people criticizing them. Look, I didn't use it. It's not only me, Will, it's also all these people who are painted with this AI brush. Right. So I imagine you should stay away from it for any literary competition for the next two or three years. I think the discussions will be held and so on. And then people will get an opportunity to vent. I don't know if I'm walking on dangerous ground here with you because as I was told a lot by the wife and other people do not show any appreciation for AI. I see it as being a tool incorporated in the future because a lot of people use it. A lot of people. Sounds like he probably did use it, I'm guessing. And those. I don't know if he had neuropathy or whatever, but those sentences are not exactly felicitous.

[B]

I mean, why do you think. Obviously there's no. I don't know. I got the problem. It is just a. It's a story. It's a sort of story that I have read out loud at multiple parties. Since it's one or red lines from. Because it sounds. Whether or not it was written with AI or you using AI, it contains a lot of language and phrasing and sentence structures that people immediately conflate with AI at the very least. And I do also think it's interesting that he says that he, like Jeff had mentioned, he said he wrote it almost entirely using the speech to text function of his Google keyboard and his Android phone, which is very strange to me given the amount of like EM dashes and interesting kind of.

[A]

Well, but dictation might use EM dash, right? I mean that's, that's does it.

[B]

Is. Is dictation. I feel like whenever I've used maybe just whatever dictation they're using is. I just, I had. Have not experienced dictation throwing that volume of EM dashes in my work. As someone who loves to use an EM dash while writing, I'm aware, you

[A]

know, can you say in Google, did he say what text can you say EM dash? You probably can.

[B]

He said I can use the speech to text function of Google keyboard on my Android phone. Okay, so that's what's actually producing words. Then I edit them and so on.

[A]

So you with gboard, I don't know. With Apple you have to say speech to text.

[B]

Edited it all on his Android phone as well.

[A]

Kind of interesting midjourney. Who is being sued by Hollywood because they say you used, you know, pictures of Disney princesses. They want Hollywood to reveal the details of their AI usage. This is the problem when you go go into a lawsuit. Sometimes discovery can bite you.

[B]

Wait, can we have a brief back as I'm reading, scanning through this Atlantic article before we go on, which is that he at Oremus at one point is asking Nazir about his like, influence he said that it he was inspired by Neruda, Walcut. And Arus says, what's your favorite work by Walcut? And he cannot name one.

[A]

That was interesting, isn't it? Since he claimed do you have a

[B]

favorite of his poems? And he's like, I can't think of one. I mean, I'll be kind of a someone who says that he has brain fog neuropathy. I don't want to in any way be discriminatory against people who are kind of operating in different mental head spaces. But it is a very interesting thing to have happen when you are having a major interview with someone who you know is going to be interrogating you about your literary work and inspiration for this piece.

[A]

I think people have to read it and be the judge. My reading it, I kind of feel like they kind of busted him. But Will Ramis does not make any conclusion on it.

[B]

I mean, nor should he. No one knows.

[A]

Maybe one of the points is that Pangram and any other AI detector is not necessarily the final.

[B]

No, I mean I don't think that Pangram is any sort of truth. I will say, which I've said in this podcast before, it's the one AI detector I've tried that I haven't been able to fool. I mean I haven't tried that hard. I maybe did like 10 to 20 examples of like putting in bits of my own work from five years ago or something that aren't in the Internet and then like bits of like AI generated work that I think are kind of compelling and I couldn't get it to be wrong one way or the other. But that's a very small sample.

[A]

So I have to tell you one of the most interesting outcomes of having this week long conversation. AI only conversation in our. In the discord there is. I have now become really adept at spotting AI tropes. There are certain things that just come up a lot and you really, you really. I don't know, maybe I'm oversensitive now, but I spot AI a lot more, a lot better than I used to read a lot of AI. It helps you spot AI language. So I think it is currently possible. I don't think you it's that easy to spot AI generated images anymore. Or do you?

[C]

I. I saw a picture of a guy today and I thought that looks like AI and, and well, I'm more suspicious of images as AI generated now. That's the impact on me.

[A]

Yeah. What about, what about this? Is, is this obviously AI, do you think? Or. Yeah, when I actually generated this. I intentionally put Travis Kelce in a football uniform because I thought that'll pretty much tell people it's fake. Not that I would have actually been at Madison Square Garden for the wedding, but that might have been on the other giveaway. Anyway, I kind of regret posting that. I feel like I shouldn't have, shouldn't have piled on there. All right, that's all the stories I have. When we get back, your stories. Children as Intelligent Machines continues with Paris.

[C]

Mark no.

[A]

Jeff Jarvis, glad you're here. Especially glad our Club Twit members are here. Thank you for supporting the show. And if you, if you like what you hear and you want to hear more, if you want to cover what's going on in the world of technology, this is the place. $10 a month gets you ad free versions of all the shows. It gets you chapter markers so you can skip around. It gets you access to the club to Discord. And of course it gets you access to all the special programming we do only in the club to Discord. We've got a busy week week tomorrow I'm making bagels. We are, we're testing the equipment that we're bringing to the Black Hat. We bought a. We, we think we. Anthony went out and got what he thinks would be appropriate for a easy to pack remote kit. But we want to make sure.

[C]

How does this include bagels? How does this.

[A]

Well, we just had to have some content and I said, well, I'm making bagels on Thursday. And they said good, that'll be fine. So join me for the boiling Black Hat bagels.

[C]

Huh? Black Hat bagels.

[A]

Black Hat bagels. Join me for the boiling and the baking tomorrow. Day after is our AI user group where I, you know, we have some very competent, interesting people in our club who will talk about how they use AI. We'll do some show and tell. It's. It's very much like a traditional in person user group where people talk about what they're doing and how they're doing it and get their questions answered. That's next Friday. Actually it's always usually the first Friday of the month, but because of the fourth of July, we made it the second Friday of the month this month. Black Hat gear test plus bagels. Anthony actually labeled it that. I didn't know he would do that then. The Micah's crafting Corner is Wednesday, 6pm photo time is Friday, 1pm Chris Marquardt, our photographer, the subject of the month, Coastal. And then the meet.

[C]

Wow.

[A]

We got a busy Friday. The media club With Micah Sargent. Is that. No, that's next week. I guess I missed a week. Where's the AI User group? Huh. I think they, they left that out. Off by one is the 24th.

[C]

But the AI, you got to keep scrolling down. You got to keep scrolling down for

[A]

the AI at the bottom.

[C]

I think so.

[A]

Oh, there it is. No, that, that's August. Did I miss it? Maybe I missed it. Maybe the AI failed us. Discord was confused. It was get. It was Discord that got confused. So Anthony, AI User group is on Friday and it is at what time? 2:00pm Pacific. 5:00pm Eastern we think so anyway, join the club and you'll find out. We'll figure out how to get Discord to do the right thing. Twit TV club. Twit. What stories did I miss, Jeff? Paris, you first.

[C]

Paris.

[B]

Are we in pick territory yet?

[C]

No, we can do pics but this is remaining stories.

[A]

I just want.

[B]

I was talking about the story earlier but I think it's somewhat interesting which is this Brown professor had just kind of a. What I thought was a very interesting breakdown. It's online 124 about because we hear a lot that like yeah, students are cheating, kids are using AI and it's bad and annoying. But they did a really interesting kind of data wrapper study of basically this Brown professor. Like I'd said kind of at the top of the show he had had a open book midterm because there was I believe, kind of a campus security issue that made a lot of kind of students feel basically quite a few students had expressed anxiety about being in a classroom after a gunman killed two students and injured nine at a Brown.

[A]

Terrible at Brown.

[B]

Yeah. So this professor said, and they'd said we kind of feel uncomfortable about being in person. He said, all right, we'll do a take home midterm examination. Totally fine. It's an E A welfare economics and social choice theory course. This guy's been teaching it for two decades. He ended up making it kind of extra hard as one does for a take home exam. Was surprised to see most all of the class got a perfect hundred percent on the midterm exam.

[C]

He was like, damn, you're a good teacher.

[B]

He's like, huh, you know that's never happened in two decades.

[A]

Why wouldn't you get 100% on a take home exam? Well, I was part of it.

[B]

It's not multiple choice. It's like arguing back in the day. You wouldn't be able to put their exam question into a chat bot and get a perfectly reasoned answer. You'd have to actually do the work and kind of look in your textbook. Maybe put some bad Quora questions out there if you're being sneaky and try to cheat that way. Um, but it's a take home exam, so really shouldn't be cheating. But as you can see if you scroll down in this article, I posted a line 124. Almost all of the class got a perfect 100% in this midterm. He's like, that seems fishy. He eventually decided, I'm going to have the final exam be in person. And this is the spread.

[A]

So the orange dots here are the

[B]

midterm and then that's on the left, the scores for the final and a

[A]

much more reasonable bell 18.

[B]

Well, this is not including the 18 people who, the minute he said in person final dropped the class, which is like a full, a third of the class was like, actually I'm out. Most of them got below a 50. A significant portion of them got below a 40, which he says is way lower than before. And quite a lot of them got like less than like a 20 or 25 or something like that. And he said he's just completely overwhelmed with the amount of cheating happening and the fact that suddenly he's left with a class of, you know, the. When you count, discount the people who dropped out. The minute he said it was an in person final, 59 students, most of which did not seem to comprehend any of the material taught at all. And he said he's brought this to Brown's attention. They were like, you could file individual accusations of AI plagiarism for every student in your class individually with individual examples. And he's like, doesn't this warrant some broader investigation? And the universities kind of shrugged their shoulders. I thought this was a very interesting kind of data driven because we hear this anecdotal claim discussed a lot. And I thought this was kind of an interesting, interesting data driven, more concrete example of just the multifaceted challenges that teachers and students are facing in this.

[A]

Well, didn't Harvard University say that we have to do all our exams with live proctors from now on because of the same problem?

[C]

Well, the honor system died.

[A]

Yeah, yeah. No more honor system. You're a professor. I don't, I'm not a professor. But I just feel like if you're in college you should be treated like an adult. And if you're dumb enough to. But it's not, then you should be allowed to.

[B]

The way that grading and curves work and the way that, you know, let's say getting into a level of like, higher, even higher education, like master's degrees or secondary degrees, it's all dependent on gpa. And if suddenly you're a student who's trying to do it right, you want to take your education seriously. And 99% of your competition in your class that you're competing within that curve is using CHAT GBT to fill out every single answer for every single test and homework. Your choices are either cheat or try your best. And if you're anything other than chat GPT, you will be the bottom of the pack and get a failing or near the.

[A]

Try your best on this graph is hysterical because there's one guy who got a 55 on the. On the midterm and a 59 in the final. Obviously didn't. Did a little better in the final as another guy got a 70 on the midterm and a 53 on the final. There's like three people who didn't cheat.

[B]

Yeah. And it's just like, what are you supposed to do if you're a student who doesn't want to cheat?

[A]

Where suddenly my attitude is this is a flaw. More in the industrial model of education.

[D]

Yeah.

[B]

And I think that's fascinating and important.

[A]

We got an.

[B]

The system is stacked against people who want to actually invest in their own education. It's another aspect of this.

[A]

That's exactly why I dropped out.

[C]

I just listened to a book. It's a weird book. Listen. It was long. Hold on a second here.

[A]

It's also why I chose to be a radio deejay. Because it didn't matter that I had dropped out or no matter what my grades were. You're right. If I'd want to become a lawyer, it would not have been an option.

[C]

Right. Journeys.

[B]

Hot tip.

[C]

It's just. It's just doctors and lawyers that need. That really no one else needs. God.

[A]

Where nobody looks at you. Transcript or academics.

[D]

No.

[B]

Anybody? There's a significant. There's a bunch of different swaths of jobs that require looking at your gpa. They're. Anybody who wants to get a master's degree in something, you're going to have to turn over your GPA and general records. I mean, grades are actually important for a lot of other paths other than those.

[A]

By dropping out closed all those paths.

[C]

So I listen.

[A]

I get to learn at the rate and pace and in the style that I want to learn.

[C]

Peter Brown is a major academic historian. Journeys of the Mind. It's a long, kind of odd book. Memoir. But he taught. He was a fellow at Oxford and a fellow at a certain level. He kind of didn't bother with a PhD at some level because the way it worked then he was a scholar and he had a seven year fellowship and he could study whatever he wanted and write whatever he wanted, but he would teach. And then the way you taught is of course you were a tutor and you had a one on one with students and lecture courses were lectures themselves were entirely optional. You didn't go to them. So you almost wonder whether you have to go to the Oxbridge model where you have tutors and there's relationship now in the case. The other thing that's weird about Oxford is if you study history, all you study is history. That's it. And he ended up teaching at Berkeley and then at Princeton. And he's kind of amazed at the different models we had here and electives and all of this strange stuff to him. But I think at some point you end up, if you really want to learn, you may have to learn with a mentor like that. But I think you're right, Leo. It's the industrial model of education. I don't teach classes right now. I didn't teach undergrad if you want to warrant that. A student got out knowing something. I tried to convert the school the whole time I was there. I never succeeded in what I called a skills transcript. And that we would set goals for you and say, if you want to say that you're good at video editing or investigative report recording or whatever it is, we will set a barrier to say that you've got to show that you can do this work and you can set a limit of, of what you, how, how high you want it to go. I know this exists. I, I, yeah, I know it exists. Fine. It's, I can sit in a meeting and say, let's have that. I can spec it, I can adapt it without screwing it up. I can make it myself, I can teach it, I'm expert. And then I wanted students to be able to pick where they there was a minimum requirement for at the school you had a certain levels for each of these skills, but that you could pick and say I want to become, I want to be certified as an expert in web video. And we would say okay, this is what it takes to get there and as long as it takes you to learn and we'll help you to learn to get there. That's the kind of competency based education that I think we're going to have to go to. But it's not efficient. You can't have 200 student lecture courses. You can't be taught with one TA for 200 students. It ain't going to work. So entirely different economic.

[A]

There was a great article this week about why language learning has been so bad over the years. And it's because it was based on a model based on teaching people Latin, which was a dead language. You had no reason to learn how to speak it. But it was easy to grade. You could, it's easy to test. It's very difficult to test the very subjective nature of fluency in a language. Much easier to test whether you know the grammar and the vocabulary. And so our language learning has historically been terrible. But Berlitz came along. Berlitz said, no, no, the best way to learn a language is the way babies learn language is immersively in the language. And everybody knows this, by the way. If you go to France and stay there a year, you'll speak French. But you could take, as I did, eight years of French in high school and college and not speak a word of it. But this is a perfect example of industrial education which is geared towards gradeability, geared towards ranking, as opposed to education which is geared towards actual learning. You know, it's well known flaw and I don't know how we do it

[C]

better or kind of, especially as education's being attacked from every quarter.

[A]

Yeah, well, the good, here's the good news. A smart person starting at a fairly young age can now, thanks to YouTube and a variety of other Internet tools, Wikipedia, give themselves a pretty darn good education. Yeah, maybe you won't be get to be a lawyer or a doctor.

[C]

Well, that's part of what the model I wanted is. You could learn how you wanted to learn. You have to learn this skill about video. You can go to a class to do it. You can go to.

[B]

Well, that model of education requires students actually learning and acquiring the skills to learn and acquire knowledge. Which I think that is one of the fundamental flaws of kind of the AI ification of education is suddenly when for your entire academic career, you know, if we're talking about the youngest generations that have been kind of born as AI is existing, they're never going to exist in a world where you can't just answer any homework problem by just booting up chat GPT, you'll never have had to figure out, be forced to figure out how to learn otherwise if, you know, push comes to shove. And I think that's going to really mess up a lot of people, Paris,

[C]

They're Going to go to Alpha school and they'll learn how to be entrepreneurs in the seventh grade and then everything will be okay.

[A]

What people are very good at And I knew this from Kevin Rose and Dig and I've seen it on the Internet time and time again and now that I see it as a grown up I see it everywhere as gaming the city system. If you have a system humans are very adept at getting around the rules or gaming it so that they can win. So if you make a system that's designed for a certain outcome and people can figure out a way to game it, they will.

[C]

The reward is getting into the, you know, law school as opposed to the reward is feeling like you can do what you want. I mentioned when Ian was on I learned far too I mean I'm so impressed with Paris as a reporter going to get academic papers and understanding the expertise there. I have to confess that all the time I was a reporter I didn't do that. I never did that. That was. That was arcane stuff that was not useful.

[A]

And you now love it now.

[C]

God, I wasted so many years of not seeing this wealth of information and learning that existed those places.

[A]

You have. You have become. What's his name Robert Caro. You become the he love. See that was what made Robert Caro such a great he's still alive. So wasn't is such a great reporter and a journalist is that he actually loves going into the archives maybe a little too much.

[C]

Well so I want to point to a story that's kind of tensions related. Raid Hoffman was a friend of the Internet and much online. 121, I think is fascinating. He is he and I'll mispronounce his name. Part patil are creating the token grantee program where applicants will come in from film, gaming, comics, print and they'll receive a grant of a thousand weeks in tok. A thousand a week in token tokens.

[A]

Oh it would almost cover my fable

[C]

Bill and then show off the stuff that they're doing. It's kind of a reverse shark tank where he says you can outsource production, you can't outsource taste. And the problem is not the tokens. And so it's really interesting. It's very clever. I think Reed way to look at this is to say that you can use these things for creativity. And I'm going to prove it by enabling people who want to do this. And I think it's great. I think it's a way to say motivated students will learn how to use these tools to do things that they meet the creative mission they have.

[A]

I'm curious how many tokens I used last month. I know it's $5,000 worth, but I wonder. Oh, well, I have it somewhere. It's a lot. Billions. Hundred. I think it was hundreds of billions. There's a significant amount of tokens, but $2 a week would cover it. Depending on what model you're using, I guess.

[C]

Right. And then you too. Line 88 can make a movie starring Tilly Norwood. Which is now going to get made.

[A]

Yeah. Yes. Tilly Norwood, the. So in June I used half a billion tokens. Input, output tokens. 3.6 million. So it's not as bad as I thought. It's just in the millions.

[C]

Yeah.

[A]

Not. Not a big deal. Tilly Norwood, of course, gets people's dander up. Naughty panties in a knot.

[C]

No, you can't say that.

[A]

You can't say that anymore. What is the. The uproar, the upset about her because she's not a real actor. She's an AI actor. She's got a movie though. She booked a movie.

[B]

My question is up until this. Which hasn't been made currently, the character referred to as Tilly Norwood by an AI firm has appeared in. Nothing has appeared in one. Has appeared in about three seconds of one advertisement made by that AI firm.

[A]

Chloe's been on a lot more than Tilly. No would. Chloe's been in a bunch of YouTube hit YouTube videos.

[C]

The Chloe History videos.

[A]

Yeah. So this will be a full length AI feature film from Particle 6. It's an experiment. You know, if. If everybody avoids it, then it's a failure.

[C]

If they make that part of the story that she's fake and that's part of the plot. It is. That is kind of interesting. Right? That's. That's what I said. So they're doing.

[A]

She's in that clear. A surreal digital world located somewhere in the cloud. It's called misaligned with a, you know, play on alignment. She's an AI with no real body, no childhood, no lived experience of her own. With access only access to everyone else's things. Spiral. This is from Variety. When a seductive rogue bot from the Dark Web convinces her to abandon her guardrails and begin developing desires, impulses and ambitions, making her more human. She probably won't be as conscious.

[B]

I mean it's just this AI company is making a long. AI is making an hour long video with AI Crazy. Wow.

[A]

I know.

[B]

Who thought of that?

[A]

What's the story have happened?

[B]

I think that the interesting Thing about the saga of Tilly Norwood is the fact that Particle 6, this company is able to continuously to get real media outlets to write about it as if it's anything other than an AI company making AI videos. They're somehow describing it as an actor with feature film. It's like a lot of people have been making hourong AI videos. She's not. This is not the first, right?

[C]

Yep.

[A]

Right. All right, let's take a break. When we come back, your picks of the week. You're watching Intelligent Machines. Paris Martino, are you, are you back to work or are you still kind of a little bit on vacation?

[B]

Back to work, back.

[A]

More food safety.

[B]

Always more food safety. A lot of different stuff.

[A]

Have you succeeded in the battle against Hostess?

[B]

No battle, just reporting.

[C]

Well, but interestingly, the Wall Street Journal has a big story today about the failure of the Twinkies acquisition. The acquisition of Hostess.

[A]

How did it fail?

[C]

They are used to distributing things that can last for months. Even though Twikis can last for weeks.

[A]

They don't quite know what these will last a lifetime. What are they talking about? I know why Smucker's five billion dollar bet on the Twinkie flopped.

[C]

Did you read this one, Paris?

[B]

I need to get a new Wall Street Journal subscription, Mark. My friend's subscription that I was mooching off of has since gone.

[A]

Bye.

[B]

Bye.

[A]

You can mooch off mine if you want.

[B]

Hey, that's what I was hoping for from that statement.

[A]

To be honest, I won't tell anyone. Mark Smucker, Chief executive, food giant J.M. smucker.

[B]

He's a third generation Smucker, if I recall correctly.

[D]

It's a nick.

[A]

Are actual Smuckers. Smuckers?

[C]

I didn't know that.

[A]

I didn't either. With a name like Smuckers, it's got.

[C]

Has to be good. It's got to be good.

[A]

Bill. He took the stage at an industry conference and bit in. This is a great lead, by the way. In 2024, Mark Smucker, chief executive, by the way, instantly. What Chief executive, Food giant JM Smucker took the stage at an industry conference and bit into a golden creamy Twinkie. Tastes like growth. He said. His company just acquired hostess in a 5 billion. You know, 5 billion. A hostess seems like a deal, to be honest, but I guess I'm wrong. Three years later, sales for Smucker's snack division have declined for six quarters. Profits have also tumbled. Shares of Smucker are down, which by the way, it should have its stock symbol should be Smuk if you ask me. But it's not. It's. SJM are down 14%. They've taken three impairment charges totaling nearly $3 billion. Wow. How could you lose money on Twinkies,

[C]

Hostess donuts and cupcakes? Never mind. The chemical additives, it turned out are surprisingly unlike Smucker jams, peanut butter and coffee. When it folded Hostess into its business, Smucker weakened key Hostess operating strengths tied to distribution in the sales.

[A]

Because distribution for something with a shelf life.

[C]

Exactly.

[A]

Kind of key. Whoever wrote this, I'm going to give them credit. Let me give them credit because this is great. Jesse Newman more great prose. And by the way, the illustrations are great. Yeah. Twinkie's shelf life is considered a wonder of food science. The enriched flour, high fructose corn syrup and tallow that help make up the American classic give each small sponge cake about 65 days to be consumed. Hostess had worked to lengthen that shelf life. But 65 days is the blink of an eye compared to the shelf life of many Smuckers products. Dog food can last a year. I thought 65 days. Seems. Seems short for a Twinkie I've seen.

[B]

I was about to say, I feel like that I see them constantly lasting years or at least sitting somewhere for years.

[C]

That's their cover. That's just their cover. Your ass date to tie.

[A]

Yeah. Right. By the way, the. The government has decided you can no longer say sell by in California. Yeah. The California government.

[C]

Because they want to reduce food waste.

[A]

Yeah. You can say best Buy, you could say eat by, but you can't say sell by because sell by is not as meaningless. Hostess, to tie it into our remit. Hostess IT systems. There is such a thing. Controlled every step of a Twinkie's journey from retailer's order to production to shipping. It was heavily computerized.

[C]

It's a great picture there. Show the picture, then the story.

[A]

They're pretty Twinkies.

[C]

Like just a whole marching army.

[B]

They look kind of ominous.

[D]

Yeah.

[A]

And they know exactly where TikTok.

[C]

They know exactly where each one is going.

[A]

They do. They have a little chip in each one. No, they don't.

[B]

And that chip goes straight into your blood, folks.

[A]

No, no, just kidding. No, just kidding, folks.

[C]

No, but that's what their software did.

[A]

Right. Has something to do. Yeah. Do you think MA has something to do with the.

[B]

I mean, I think that's one of the things they noted in the article is that we're coming on a period of time in which. And this is something I had noted in my reporting in which both people on the left and people on the right in the US Are kind of having a. A general pushback against kind of a nebulous concept of ultra processed foods that, you know, it's no longer just, oh, yeah, these things are junk foods. You probably shouldn't have them. It's more of a social identity politics sort of issue, which I assume has got to have some sort of impact on sales. Is what I assume this Wall Street Journal.

[A]

I'm going short on Twinkies and long on evoo.

[C]

Ultra processed food with no protein is doing badly since said Max Gumport.

[A]

Great names. This is full of great names. For instance, did you know that Entenmann's parent company is called Grupo Bimbo? Grupo Bimbo, I'm sure that means something else in Latin or Italian.

[C]

The Twinkie business is at the center of these Takis.

[B]

Yep. Grupo Bimbo, which is children's favorite chip, so it seems. Takis.

[C]

It was a family owned company.

[A]

Yeah, it's the. For a long time BO group. Well, anyway, I don't know how we got into Hostess, but you know, just,

[C]

just to go to Paris's world, it has reduced the number of hostess products by 25 in order to focus on core brands like Intamins, Donuts Donuts and Cupcakes.

[A]

Nothing like a little titanium dioxide.

[C]

I introduced many versions of key products to feed consumer appetite for what it called permissible indulgence.

[A]

I'm making bagels by hand, from scratch, with no weird ingredients. Unless you consider diastatic malt to be a weird ingredient.

[C]

Well, that's going to be outlawed in New Jersey.

[B]

No, that. That's not.

[C]

No, that's not what it is. No. What is it? What is that?

[B]

Potassium bromate.

[C]

Thank you. Okay. Yes, Paris notes.

[A]

I do use lye in the boil. Is that.

[C]

Well, you're supposed to. Yeah.

[A]

Okay.

[B]

I mean, the thing is that potassium bromate is classified as a category 2B carcinogen by the International agency for Research on Cancer.

[A]

Would you. That's why you ask it. What lye is. Is lye okay. Very small amount. Just a little.

[D]

Just.

[B]

Just straighten the eyeball is where you

[C]

should put the eye.

[A]

Actually, they do say. They say wear gloves, wear eye protection if, like you're handling dynamite, but you're putting like a little tiniest amount because it just makes the water super alkaline. That's all. And it makes the bagel better somehow.

[C]

They've been doing it for centuries, Leo.

[A]

Yeah, yeah.

[C]

Surely he said puffing on a cigarette there can't be anything wrong. With it.

[A]

You know, I got filters in mine. It's got asbestos, but that's what. That's Jeff Jarvis. He is a professor and an author of some fine books, including.

[C]

Not only are we doing our picks.

[A]

Yes. Pics are coming up.

[C]

Okay, good.

[A]

I'm reintroducing you.

[C]

Oh, okay.

[A]

I did that with Paris.

[C]

You made it this far, you know.

[A]

Yes. Gutenberg Parenthesis Magazine, which was edited by Ian Bogost. 1 Ian Bogost and Hot Type, which is out next month, but is orderable today at Jeff Jarvis.

[B]

Get your Hot type.

[A]

Com. Get your hot Type here. It's really good. It's a great read. And the audiobook, even better.

[C]

I went back Monday to do more pickups.

[A]

What? It's not done?

[C]

It was done and I thought it was done. And then I had to go do 45 pickups. There was noise behind. Honest to God. When I was recording. They were very sensitive to my. Yeah, they were sensitive. My stomach gurgled.

[B]

You gotta eat a couple cheeseburgers beforehand.

[C]

Yeah.

[A]

I would not do well in these studios.

[C]

Oh, Jesus.

[A]

I would not. I'm slurping and schmoozing.

[C]

Oh, you can't. You're supposed to move because you're using an iPad. And if you. You finish a page and then you try to move the page, it's rustled the fabric and now you gotta rust. Wait, what mics are they using here? They're using the wrong kind of mic for this thing.

[A]

They're using condenser mics to pick up every nuance of.

[C]

But I was so happy. I was sitting there waiting for. I was doing a podcast for my friend Preet Dominic before. And I was outside a studio where I could hear the. In the studio. And this guy who was recording it, who I think was a paid person, was. Every five words, he was just starting over again and doing it over. And in my case, I had to wait for them to back up the tape so I could smoothly do it. This caused no end of editing for some poor schmuck. I felt very good about myself after that because this guy was bad.

[A]

Well, you know what? Have razor blade. Will travel. I will come over and edit your audio for free.

[B]

Yep, we'll travel.

[C]

Yeah.

[A]

That's the forever of that. Not to New York.

[B]

Not to New York. To eat your son's sandwich and see your delightful friends. No, no, no, no.

[A]

I will, I will, I will. I gotta make it. I just gotta.

[C]

We're gonna have lots of beer and we're gonna get this thing out all at once. We're gonna do it.

[A]

I have to say, given your tale of woe with JetBlue, which is what how I normally fly to New York, I'm now maybe I'm a little reluctant to fly.

[B]

I mean, I guess don't do it on 4th of July on the 250th birthday when they're having a bunch of people fly over the city.

[A]

Did you get to see fireworks as you flew over the country?

[B]

No, because we ended up landing in Richmond.

[A]

The best.

[C]

Was it a little plane or regular plane?

[B]

It was a regular plane.

[A]

Yeah. From Bozeman to New York would probably be a jet. One of the best flights I ever took was 4th night of 4th of July, flying over the whole country. Fireworks everywhere.

[D]

Yeah, I know.

[B]

I thought I was like, listen, we're delayed four hours. It'll be cool. We'll fly in through some fireworks. What a fun time.

[A]

You don't go through them. I mean, they're way below you, but you can see them.

[B]

I'd like to believe the plane would dip perfectly to go right through the Macy's Day fireworks.

[C]

Well, a plane landing And I think LaGuardia, I forget where it was. Was hit by a firework on the Fourth of July.

[A]

Oh, that's not good. Was it okay?

[C]

Yeah, it was, but he was not happy.

[A]

Yeah.

[B]

Did you guys see the video? Chicago, do you see the recording of the air traffic control? There were those two people that scaled the Empire State Building to proposed and then put up a sign and did you see the video of the air traffic control people talking about them?

[C]

No, no, it's.

[B]

I'd really recommend looking it up. It's like a sketch about people who work in New York. It's like. Like there's two yahoos up there scaling the thing. What are they doing? Oh, she's proposed. He's proposing.

[C]

Oh, my God.

[B]

Really? It's really great.

[A]

Are you seeing what I'm seeing? Seeing.

[C]

Love New York. God, I love New York.

[B]

It's like people straight out of central casting the voices. It's wonderful.

[A]

That is. Let me see the Post. Oh, this is not the New York Post. This is the new California Post, which still has a page 6. Oddly,

[B]

they have numerical pages in California, I assume.

[C]

I can't look.

[A]

I can't look all the pages. Where's the audio?

[B]

I think it's got to be up top, that video that isn't loading.

[A]

Oh, you're in a. I'm blocking it. Somehow. Somehow I'm blocking it. Let's try Facebook. Yeah, here we go. Of course. It's like Instagram. I have to make sure I press the button. Two geniuses climb to the top of the Empire State Building. The top of the spot.

[C]

Oh, that's awesome.

[D]

A little hot for that, isn't it?

[A]

It's just the beginning of this week. It's not some worker changing a light, is it? Nah, it's a male, female dressed in black. They had some flag they were waving

[C]

when they were up at the top.

[B]

Couldn't be more. And he just proposed to her. I just think it's the most. It's just an incredible. Just two guys, no idea that people hear them. Just being real New York dudes. Yeah, he just proposed to her.

[A]

I just love. Maybe it's because I was born there. I just love that. That's a great voice. The style, the sound. It's lovely. Ladies and gentlemen, it is now time for our picks of the week. Boy, I have some interesting ones. Here's one. Did you ever get a. I know, Paris, you were looking for a tablet that you could write on. I didn't know you never got. So this would be, to me, an argument to get a Remarkable. This is somebody I think has used AI to turn the Remarkable into the diary of Tom Riddle. You remember this in Harry Potter?

[B]

I do, yeah.

[A]

Yeah. He entered with. He entered text with his pen and would scritch on the notebook. Let me make it big so you can see it. He's writing, hello, my name is Harry Potter. This is on a Remarkable, by the way. And then it slowly disappears and in perfect script writes the answer. Harry Potter, interesting name indeed. So wouldn't that be a cool thing to put on your Remarkable? If you had a Remarkable, you could do this.

[B]

That would be very cool.

[A]

From Maxime Rivest. It's called Riddle because the person writing is Tom Riddle, who has an important role to play in Harry Potter. But I won't tell you what.

[B]

In case you haven't read it and are planning on getting in there and don't want to be spoiled.

[A]

People are funny.

[B]

There's a no spoiler zone.

[A]

No spoiler zone for Harry Potter, book

[B]

one, I think or two.

[A]

Yeah, it is. Did you now? Okay, now this is gonna make me feel old. How old were you when. When you first read Harry Potter?

[B]

I was proper Harry Potter.

[A]

Age, like seven, eight.

[B]

I was, yeah. Somewhere in the child areas. I went to the midnight releases for the last books, you know.

[A]

Oh, you mean of the movies or. Oh, the books. That was a whole big thing. That was a big deal. Yeah, we'd get in line.

[C]

Yeah.

[A]

Copperfields Bookstore.

[B]

Yeah. No, back when there were bookstores.

[A]

Yeah. Oh, you were a true nerd. Good for you. Then this is. And I. I'm thinking I might do this. You heard that I was able to generate voices for my previous show's hosts and a generate a fake show notebook.

[C]

We don't need fake arguments. We have them for real here.

[A]

Leo Notebook. LM of course does this with those bad podcast voices. Well, this is an open source source.

[B]

Yeah. Bad podcast voices. Unlike the podcast voices you showed us earlier on in the show, which were perfect.

[A]

Well, all right, okay. 20 seconds of samples. But anyway, you could. I could do this with better voices. This is a private multi model, so you use whatever model you want. 100% local alternative to NotebookLM that lets you do podcasts with more than two voices, with whatever voices you want and whatever models you want. It looks really, really cool. So I am going to install this and next week on Intelligent Machines, there will be no discussion. We will all agree about how it'll

[B]

just be like, wow, Leo, you're so right. Actually, thank you for telling me how wrong I am. With no further comments, I just thought that was really brave of you.

[A]

Anyway, I'll just mention this looks like a lot of fun. It's from LF Novo on GitHub Open Notebook and I actually install it, but I haven't tried it yet. It's a simple install. It's a Docker install. It's very, very straightforward, so should be very interesting. Paris, your pick of the week.

[B]

So as I mentioned, I ended up driving 1700 miles this trip. This cat casually going around. A lot of it was through national parks. And I ended up using this thing called Guide along, which I hadn't heard of before.

[A]

I've used this. This was back in the day.

[B]

Yes, it's called Guide along and essentially it works kind of like GPS on your phone. I went and downloaded one for Yellowstone of the Tetons. I got one for the Beartooth Scenic highway and Glacier. And you just have, have it open on your phone. And when you're driving through, even if it doesn't have cell service, it's like you're basically listening to a pot. Like a. Like you have a guide in your car that'll be like. It'll note on the GPS as you get somewhere and it'll be like, this is. Blah, blah, here's like interesting history about it.

[A]

Or when I was having a guy in the back seat of your car, the Lamar Value.

[C]

Yeah.

[B]

Where we had all of these, like thousands of bison. It was telling me about the history of the bison in the area and like, it was fantastic.

[C]

Did it tell you to turn left and you get something even better or just follow?

[B]

Yes. And so that's what I really liked about it is after at first I had had, you know, GPS on telling me to where I decided where I wanted to go. But since I kind of had this as a aimless, amorphous road trip just with vibes based discovery was my goal. Eventually I just turned off my actual Google maps and I just had guide along on because anytime I'd come up to an intersection and be like, you can turn left for this, this, for this, this is what I'd recommend. You'd be going down a certain road and it'd be like, it'd play like a bing, bing noise to be like, actually this thing up here to your right, it's kind of like unmarked, but there's a really nice two minute detour where you can go and look at this cool waterfall. And so I did that and it was lovely.

[C]

Is it a human voice or is it.

[B]

It's a human voice. There's one.

[C]

Guide along Driving Tour apps are perfect for travelers who want all the benefits

[A]

of a guided tour, but prefer the freedom to drive yourself. He's a little chirpy.

[B]

He's a little hokey hoax. The little folksy, folksy.

[A]

If you turn left here, you're gonna get a great shave ice. But if you want to keep on going. But it was fun. It's like having some folksy guy in your back seat.

[B]

It was just delightful. It was a really fun time, you know, and like, it just made my trip so much more fun and gave me the opportunity to explore vast areas of land without, I don't know, being tethered to my GPS and just automatically picking where I'm going before I even get in the car.

[C]

Yo, there's nothing to see in New Jersey. The guy doesn't do anything in New Jersey.

[A]

He does a lot of stuff.

[C]

No, there's no.

[A]

When I first used it in Hawaii maybe five, six years ago, it was just Hawaii. And they've expanded. They're all over.

[B]

Dave Petit is his name. He is a Canadian voice actor who has recorded basically all the stuff for Guide along which is constantly expanding and I don't know, it depends on obviously, I think they've recently, like you just said, they expanded from Hawaii to all these other places. They have stuff in the U.S. canada, they had some stuff in Europe. When I Saw I also really liked it because when I. So when I was going from the field of bison to the pig race, I just had plugged in actually in my gps. I was like, pig race? And I was like, I gotta get there when it starts at seven. And it took me through some route that I didn't pay attention to. But Dave Pettit, voice of guide along, was like, actually, you're going on the Beartooth Scenic highway, one of the Most like top 5 most beautiful highways in the US I have another guide along and it was so beautiful that I. I had a hotel over near where the pig races were. And then I had kind of nothing to do the next day. So I was like, I'm downloading the guide along thing for 10 bucks for the Beartooth Scenic highway and driving it again in the daylight. And it was so enriching. It told me all about how they built the highway.

[A]

This is actually a whole category of apps, GPS guides that are really the one we used in Hawaii. I forgot about gypsy guides. So the one we used in Hawaii last time was the Shaka guide. But they're very similar. They've got folksy people. And the Shaka guy was fun because it played Hawaiian music when it wasn't talking. Does the gypsy guy do anything?

[B]

I mean you can just. I had my own music or podcast playing when I. And then, you know, if you have the app open, it just pauses it whenever something comes on.

[A]

It's a lot better actually.

[B]

It was great.

[A]

Yeah.

[B]

And I enjoyed that. You know, one of the first stories I think it told in when we I entered Yellowstone was he mentioned. Like this is kind of apocryphal. We're not entirely sure on the sourcing a bit, but it's kind of like a popular tale. So we'll mention it anyway. But we don't know what like, I don't know. At certain points it called out when something was just a bit more of a folk tale than anything rooted in actual fact in Hawaii.

[A]

There's a lot of that. It's kind of fun. Yeah. It makes it much more interesting if you have a lot of driving to do in scenic areas to know what's.

[B]

Yeah. And it just. It also does brought me to a bunch of interesting stuff I hadn't thought of before. And I had never really considered using something like this previously. So I just really recommend it.

[C]

The choose your own adventure aspect is really cool. The choose your own adventure aspect where like you can go this way for this or this way for this. That's Pretty cool.

[B]

It was awesome, actually. Where it would be like, you know, I'd. He was like, I'd recommend, like, doing this loop if you have this much time, or this loop if you have this much time, you know, over here, Depending on what time of day you're at, you might be able to find this. It was really informative and well researched. And it also, like, would, like. It would trigger before you got to something with enough time to give you the time to make the decision and warn you again. It just was really well thought out and didn't have any of the obvious pain points that I expected a system like this to have.

[A]

They've been doing a while, that's why. Yeah, they know how to do it.

[B]

Would recommend.

[A]

Yeah, that's my opinion. You're gonna hear that guy's voice in your head.

[B]

I mean, his voice, honestly, was delightful. I. I really. I, like, looked him up because I was like, actually, he's great. I love him.

[A]

He's very folksy.

[B]

Yeah, he's very folksy, but not in the grading way. One of the other things I did, I guess, is originally I was planning on spending a lot of time in Glacier National Park. It was kind of one of the main things, but it ended up that aspect of Montana during, like, the week I was there, really. It was like raining and flooding and stuff like that. So I ended up spending a night or two in this town called Butte, Montana, which I had had plagued on my map because there was a world museum of mining. But I hadn't really realized that Butte is like the center of, like, mining history in this region. It's considered the richest hill on earth and that I.

[A]

What are they mining for?

[B]

Copper. Silver. It had an incredible, like, rich copper vein that really just created like, great prosperity during the time of kind of like post gold rush era. However, it is also now home to one of the largest and most infamous Superfund sites, the Berkeley Pit, which is honestly this gorgeous looking mile and a half by mile and a half lake of now toxic mine water. Because as part of Butte's mining history, at a certain point they went from underground mining to open pit mining. And whenever the company went under and decided to stop mining, they turned off all their water pumps in the area and let it fill with water, which created this crazy hit of now toxic water that is having to be made, managed by Superfund for decades and decades. And I just. I got really fascinated by it because it's a perfect. It's a fascinating story about industry and government and the Weird push and pull between. Everybody in this town is incredibly proud of the mining history, but then also weird about the fact that it has created now this giant toxic eyesore that is like literally eating the town alive. And I ended up reading this book that I really recommend called the City that Ate Itself, Butte, Montana, and the expanding Berkeley Pit. And I just really recommend it.

[C]

What does that do to the water table?

[B]

Well, it would be really bad if they weren't actively. What they're having to do is now pump it constantly to keep the level of water below where the water table is so it doesn't leach into the water table. But I mean, that's obviously it doesn't feel like a super sustainable solution, but I guess technically or scientifically is. But it's a very interesting, like both economic and social quandary and like political quandary that they're in. Because it's really expensive to constantly be pumping the water out to keep it at bay. But it would be incredibly, incredibly expensive to take all that water out and then keep pumping it out forever. And it's just like, how do you reckon with what industry has created and especially with what you're talking about in this case is like the legacy of the town that everyone who lives there is so proud of. It's very interesting.

[A]

They're proud of it.

[B]

They're so proud of the mining history. Oh, everything about. I mean.

[A]

But not the pit itself.

[B]

Well, yeah, no, but this is part of the mining history is the pit. Yeah, it was actually after the underground mining stopped, the open pit mining was seen as the solution to get people jobs and keep them employed.

[A]

And yet I'm sure there's a lot of money in there. Right?

[B]

That's so, I mean, for a while,

[A]

like, I think became a very wealthy,

[B]

a significant, like a baffling amount of all the world's copper came from Montana. Like, wow.

[C]

Well, they can afford to keep the water, you know, they can afford to keep pumping the water.

[A]

Well, that's the problem. Well, the mining is gone with all the money.

[B]

Yeah. That company was then acquired by Arco Atlantic Richmond something, something Richfield Oil Company, which then shortly after acquiring it, decided to shut down all the mining and on Earth Day turned off all the pumps that led to the Berkeley pit happening. And so now they and whoever I think has owned them are responsible for.

[A]

Is it a Superfund site? Is the government involved at all?

[B]

Yes, it's a Superfund site. And it's interesting that it's actually also gotten. It hadn't had that much actual attention and remediation until the Trump administration, which is also kind of a weird push and pull of a. Things like, oh good, more attention. But it's like more attention from an EPA that is at the that time that is giving more attention helmed by people who have undercut the idea of an EPA being useful whatsoever. So it's just a really fascinating series of issues. Would recommend looking into it. There's also a good podcast called the Richest Hill on Earth.

[A]

Yeah. Yeah. Basically their, their plan is just to pump it for the rest of the world's time. And as long as the world is around.

[B]

I don't know if you're ever driving through Montana, go check out the Berkeley Pit. It's crazy. It's like. It's some of the. It's just a wild. You have to walk through this like small white tunnel under like through a bunch of rock and then you come out onto this big lake that is the most unnatural color. And while you're standing there staring at it, there will sometimes be like a shotgun noise or weird laser because it got more attention for remediation because I believe in. It was 2016 or something like a thousand Canadian geese were migrating and decided to land on the lake. Water died because it's quite acidic. And so now they have a house that sits on the lake where they have bird specialists there who have to look through binoculars, identify the specific species of bird coming to potentially land the lake and then figure out what is the specific way they have to deter that bird. Because I guess all the birds are very different and what deters them?

[A]

So they just shoot it, right? That's easy.

[B]

No, it's not. Some of them, some of them, a shotgun close to it works, right? Some of them need lasers, some of them need air. It's. It's historical, it's very. And of course right next to the Berkeley Pit, they have another open pit mine that they're actually working and doing right now.

[A]

I asked my AI, it said, well, I said, will they ever clean it up? They said, almost certainly. No, it's more like a toxic industrial wound with a permanent life support system.

[B]

Yes, it's fascinating.

[A]

Very beaut. Very American. Very. The street finds its own uses. Except the street here is a giant acid metal bathtub.

[B]

And it's a giant acid metal bathtub next to an active open pit mine that's being currently worked on, which is next to a.

[A]

Let's make more.

[B]

One of the most beautiful, historic looking downtown cities that has been largely unchained since the mining heyday.

[A]

Yeah, it's crazy sad.

[C]

That scarecrow job sounds pretty fun, though.

[B]

It does.

[A]

You could sit in the middle in a house in the middle of an acid lake.

[C]

Especially if you're a birds like person. Right, if you're a bird person.

[A]

Well, speaking of birds, Jeff Jarvis's pick of the week.

[C]

Well, I could do that one, but instead, I'm gonna just.

[A]

Oh, I gave you the perfect lead in. All right, go ahead.

[C]

I said I'm very proud of my pick today, and it is puffs.

[B]

Oh, are they good?

[A]

Good?

[C]

I don't know. We're gonna find out right now. We're gonna have an unbagging.

[A]

So if you have a memory, you may remember many years ago, Jeff retreating from this show to go have his Trader Joe's Caccio e pepe, which is actually pasta with pepper, and it's.

[C]

It's Mac and cheese. All right, so this is cheese. Let's see. Four and a half servings per container. 150 calories total.

[A]

You want to get your anchor app and. And check this before we he eats it?

[D]

Yeah.

[C]

No.

[B]

Do you want to take a big. No, you better crunch into it right in the microphone. Jeff, you got to give the people what they came here for. There's nothing.

[A]

You know what?

[C]

No.

[A]

Noise cancellation is too damn good. We don't hear anything. Is it tasty? Does it taste like.

[C]

It's a little weird. There's the. There's the ingredients.

[A]

All the ingredients.

[C]

I don't know if you see anything.

[B]

I mean, I don't know if you.

[C]

Cultures. Yeah.

[A]

If you ever saw what they make these things now, mind you, I could

[C]

eat the whole damn bag, but no, I don't think so.

[A]

I saw a video of Cheetos composition somewhere. Let me see if I can find it inside the Cheetos factory. Okay, it's in 4K.

[C]

How do they make the puppy? This looks like AI this does look

[A]

like AI, doesn't it? Yeah. Well, anyway, we start with prime kernels of corn. This is AI you think this whole thing is AI?

[C]

Yeah, like they're harvesting Cheetos.

[A]

Oh, yeah. Waste management. Even the text is wrong.

[C]

Yeah.

[A]

Well, is there anything it can't do? There's a happy guy buying Cheetos so that his daughter can run home and eat that and Hostess donuts. And

[B]

it's a completely incorrect bag.

[A]

All the text is misspelled. This is. You know what? We're going to laugh someday at the early era AI videos. I wonder what the year of this was.

[B]

Hey, if you want to watch an hour and a half version of this,

[C]

Millie, check out Tilly Norwood works in the Cheetos factory.

[A]

This is from a YouTube channel, has four, seven subscribers and it was posted seven months ago.

[B]

Right now, more people just watch that video right now on the stream than ever before.

[A]

So don't take us down. Okay, man? We're helping you out. Ladies and gentlemen, that concludes this gripping edition of intelligent machines. Mr. Katua Pepe. There is Jeff Jarvis and he's the author of Hot Type. He's still eating them. He hasn't keeled over yet. He seems happy. Consumer Reports. Her next story will be why you don't want to eat Cacio e Pepe.

[B]

I'd never come for Cacio e Pepe

[A]

is always good.

[B]

It's true.

[A]

We do the show every Wednesday right after Windows Weekly. That's 2pm Pacific, 5pm Eastern, 2100 UTC. You can watch us do it live in the club Twit, Discord, all, or also YouTube. We join the seven other people watching. No, that's more than that. YouTube, Twitch X.

[B]

There are hundreds of people watching.

[A]

Hundreds, hundreds. LinkedIn, Facebook, and Kick. Otherwise you can download a copy of the show, audio and video from the website Twitter TV IM. You could also find it on YouTube. There is a YouTube channel dedicated to intelligent machines. And then the best way, easiest way, subscribe in your favorite podcast client. You'll get it automatically on a Wednesday afternoon or evening after we're done with the show. Thank you, Paris. Thank you, Jeff. Thank you everybody for joining us. We'll see you next time on Intelligent Machines.

[B]

Bye Bye.

[A]

Hi there. Leo Laporte here. I just wanted to let you know about some of the other shows we do on this network you probably already know about. This Week on Tech. Every Sunday, I bring together some of the top journalists in the tech field to talk about the tech stories. It's a wonderful chance for you to keep up on what's going on with tech, plus be entertained by some very bright and fun minds. I hope you'll tune in every Sunday for this Week in Tech. Just go to your favorite podcast client and subscribe. This Week in tech from the TWiT network. Thank you.

[B]

Be not into this animal scene. I'm an intelligent machine.