Intelligent Machines 852: Gluten-free Slop

Summary

Intelligent Machines Ep. 852: Gluten-Free Slop — CES AI Trends, Robot Toys, ‘Slop’ Debates, and Vibe Coding Games

Hosts: Leo Laporte (A), Jeff Jarvis (B), Paris Martineau (C)
Guest: Jason Hiner (D), Editor-in-Chief, The Deep View
Date: January 8, 2026

---

Overview

The first show of 2026 kicks off with a spirited discussion about the state of AI and intelligent devices, fresh from CES in Las Vegas. Leo Laporte welcomes back regulars Jeff Jarvis and Paris Martineau, alongside guest Jason Hiner from The Deep View, to break down notable AI trends, products, overhyped gadgets (aka “AI slop”), and cultural implications for both enterprise and consumers. The conversation ranges from Nvidia’s announcements and “physical AI” robots for kids, to AI “slop,” open-source models, the blurred lines between consumer and industry tech, the questionable rush into AI health advice, and even live coding a “throw tomatoes at Leo” browser game.

---

Major Discussion Points & Key Insights

1. CES 2026: AI’s Ubiquity and the Changing Face of Consumer Tech

[02:26–15:49]

Jason Hiner on the ground at CES:

• CES had ~140,000 attendees, 4,000 vendors – not smaller, but “shifted” away from pure consumer electronics towards enterprise and AI/robotics.
  • “It felt like it had a few years where it was contracting...but the last three years, it’s become a lot more...especially triggered by AI...enterprise is one of them. It certainly feels like it’s on the upswing again.” (D, 03:42)
• Industry keynotes dominate over consumer-focused launches; AI claims are everywhere, creating "AI washing" and some hype.

Notable Quote:

"CES is no longer the Consumer Electronics Show. Maybe that's why they took away the name... it's very heavy on industry." — Jeff Jarvis [04:14]

Hardware reveals:

• All chipmakers (Intel, AMD, Nvidia, Qualcomm) introduced new AI-integrated hardware; talk everywhere of “Copilot+ PCs.”
• Paris asks: “How many times do you think you heard AI in a 24-hour period?” (C, 06:24)
• Jason: “It was everywhere...Our mission is to focus on the stuff that's actually real AI products, not AI washing... There was plenty of it, for sure.” (D, 06:37)

---

2. Nvidia and the Race to AI Infrastructure

[06:59–14:47]

Jensen Huang’s Keynote and the Vera Rubin platform:

• Nvidia announced a dramatic leap—training at 1/10th the “token cost” and using only a 1/4 as many GPUs compared to Blackwell (already cutting-edge).
• Announced ahead of schedule; company already has all 2026 capacity pre-sold.
• Nvidia now positions itself as “the infrastructure layer of AI,” not just hardware.
  • “They are going to be in world models...taking in more real-world data. That’s what’s going to unlock robots, unlock more automotive.” (D, 12:42)
• Big partnership with Siemens for “AI factories” (data centers), digital twins, and simulation of manufacturing processes.

Notable Quote:

“In another life [Jensen Huang] could have been an amazing educator...He does like teach and communicate these very complex topics beautifully.” — Jason Hiner [07:36]

---

3. The Rise of ‘Physical AI’: Robot Toys & Ethical Concerns

[15:49–27:00]

• Robots were omnipresent but mostly “novelty” toys or products aimed at children rather than practical home assistants.
• New term at CES: “Physical AI”—from humanoids to robo-pets, to robot arms for factories/homes.
  • “The humanoid robot is one aspect of robotics, but likely this physical AI movement is going to involve robots in a lot of different kinds of form factors.” (D, 16:06)
• AI robots for kids (emotional support bots, reading aids like Luka) raise questions about privacy, safety, manipulation, unintended consequences.
• Legislative pushback: California bill proposed banning AI toys for kids for four years after incidents where AI toys taught inappropriate content (like making bombs).

Notable Quote:

"One person's emotional support robot is another's emotionally manipulative device." — Jason Hiner [21:36]

Humor:

• “Laundry folding robot is like the holy grail, right?” (D, 18:43)
• “Every year that we’re gonna see one.” – Paris, on recurring ‘laundry robot’ CES pitches [18:52]

---

4. Consumer AI ‘Slop’ vs. Enterprise Value

[27:34–31:27]

• Jason: “I looked, I watched that whole [Samsung] keynote as well, and I just felt like, boy, they're trying really hard, but there's nothing in there I could see anybody I know in real life getting excited.” [28:39]
• Enterprise AI is where meaningful ROI and application is happening (automation, office tasks, domain-specific models), while consumer use remains mostly “asking dumb questions or laughing at AI slop.”
• Trend towards smaller, more focused models for efficiency and lower cost.

Leo:

“Those people who are using AI in those contexts are seeing a very different product from what consumers are seeing. They're seeing the chat products and it's...of mixed value.” [31:27]

---

5. "Slop" and the Debate Over Open Source, Models, and AI Ethics

[41:07–47:02, 112:09–117:11]

• Discussion of “hackquisitions” where companies like Nvidia hollow out startups for talent/IP (e.g., Grok chip team).
• “Open source” vs. “open weights”: Clarity that many “open source” models only open their output weights, not the underlying code.
  • “I really want to make sure we don't call it open source. We call it open weights...Open source implies that somehow the source code is open. It's not.” — Leo [44:07]
• Conversation around Llama, Grok, and the risks of unfiltered image AI (non-consensual images, etc.).
• Ongoing debate: Is “slop” (low-grade, AI-generated content) inherently different from human mediocrity? Satya Nadella’s argument that we need to move beyond slop/sophistication divides.
• Paris: “Slop specifically refers to it being machine generated...” [113:37]

---

6. AI Health Launches: Promise, Risk, and Practical Use

[48:00–56:01]

• ChatGPT Health launches, promoting itself as a supportive tool, not a replacement for diagnosis or treatment – encourages users to link health records/data.
• Hosts express skepticism about accuracy, risk of misdiagnosis, user overreliance, and legal gray zones.
  • “ChatGPT won’t train on that data. But I just, I worry that this is going to incentivize a behavior that is already ill advised.” — Paris [52:26]
• Jeff: AI can be useful for synthesizing and making information accessible, but garbage-in leads to garbage-out: “It’s blind men and elephants... The problem was me [not the tool].” [52:26]
• Larger point: Deficiencies in the US healthcare system are pushing users toward alternative—often unvetted—sources of advice.

---

7. Vibe Coding, Live Demo: Building a ‘Tomato Toss’ Game with AI

[101:10–138:45]

• In real time, Leo and Paris demonstrate “vibe coding” by prompting Claude (Anthropic’s AI) to generate a “throw tomatoes at Leo” browser game with minimal input.
  • Paris: “Create a GUI game where you throw a tomato at my friend Leo…” (C, 134:08)
  • AI handles not just code logic, but basic art and sound effects, with Paris reporting the whole primitive game is made “in about five minutes.”
• Broader point: AI has unlocked new levels of accessibility for software creation, even for non-coders. But deploying for others (as web services/apps) still bears friction.
  • “What blocks people...is I'm not used to putting things on servers and running them.” — Jeff [100:27]
  • Benito: “Most people don’t know exactly what they want...you need to be a good producer.” [102:11]
• Playful conclusion that much “AI-made” content is or soon will be “slop,” but that all creative tools (from laser printers to chatbots) have their slop phase.

Memorable Moment:

• Leo’s summary: “If nothing else, it’s a fun toy, but I think it is a lot more than that. I’m very intrigued by it.” [134:54]

---

8. AI Education: Where to Begin?

[119:22–123:38]

• Lisa asks how to actually learn about AI; consensus: “Any course is outdated by the time it’s published.”
• Suggestions: MIT/Stanford free lectures (notably, Stanford’s “Transformers and Large Language Models” on YouTube – [119:51]), YouTubers like Nate B. Jones, business-focused newsletters like The Deep View.

Notable Quote:

“It’s really important again, that you have that metaphor that I cannot drink this river. It’s impossible. I can dip into it.” — Leo [123:38]

---

Notable Quotes & Fun Moments

• “One person’s emotional support robot is another’s emotionally manipulative...device.”—Jason Hiner [21:36]
• “Laundry folding robot is like the holy grail, right?”—Jason Hiner [18:43]
• “AI slop” is debated throughout, with Satya Nadella and the gang wrangling with what counts as meaningful vs. “junk" content.
• Paris quickly creates the “Tomato Toss: Leo Edition” (game link shared in chat), delighting the panel with the ease and weirdness of modern AI-powered prototyping.

---

Timestamps of Key Segments

• CES/AI Trend Report: [02:26–15:49]
• Nvidia Strategy Deep Dive: [06:59–14:47]
• Robot Toys & Ethics: [15:49–27:00]
• Consumer AI “Slop” and Enterprise Value: [27:34–31:27]
• “Hackquisitions,” Open Source, and AI Ethics: [41:07–47:02]
• ChatGPT Health & AI Medical Concerns: [48:00–56:01]
• Tomato Game Live Coding (Vibe Coding): [101:10–138:45]
• Learning AI, Course Recommendations: [119:22–123:38]

---

Picks of the Week

Paris Martineau:

• Medicine and Survival: Tamiflu for flu, combo COVID/flu tests, and gluten-free babka from Zabar’s (plus Nick Cage’s New Orleans pyramid tomb)
• Demo: Rapidly built “Tomato Toss: Leo Edition” browser game via Claude AI

Jeff Jarvis:

• Pixel 10 Pro experience (it’s...a phone)
• Screen protector (Torres brand) and Apple Store SUFFERING story

Leo Laporte:

• Floor 796 (floor796.com) – a massive, interactive GIF with countless pop culture Easter eggs, “old internet” fun

---

Closing Tone

The show wraps with camaraderie, playful competition over who can code the best AI-generated tomato game, and a refreshingly skeptical but optimistic review of where AI fits into daily life—from household gadgets and “AI slop” content to truly transformative enterprise uses. The team agrees that, for the average person, AI is best approached as a river to be sampled (not a cup to be drunk whole) and that slop, fun, and breakthroughs will often blend in the years ahead.

---

"Happy New Year from the Intelligent Machines gang. We’ll see you next week." — Leo Laporte [160:01]

Transcript

[A]

It's time for Intelligent Machines. Paris Martineau is back. Jeff Jarvis is back. I'm back. It's our first show of 2026 and of course we've got to check out CES. Jason Heiner joins us to give us the AI lowdown from the Consumer Electronics Show. Next podcasts you love from people you trust. This is twit. This is Intelligent Machines with Paris Smart no and Jeff Jarvis, episode 850 recorded Wednesday, January 7, 2026. Gluten free slop. Happy New Year, everybody. Time for the first intelligent machines of 2026. I think 2026 is going to be a very interesting year in AI. So we're very glad to be doing our second year of intelligent machines. Joining me right now, Mr. Jeff Jarvis, professional emeritus of journalistic innovation at the Craig Newmark Graduate School of Journalism at the university. Also at SUNY Stony Brook and at Montclair State University, visiting fellow. Hello fellow.

[B]

Hello fellow.

[A]

Happy New Year to you fellow. He's also the I missed you guys. I missed you guys too. I look forward to this all week. It's a chance to get together and talk about stuff. I'm really interested. And also with us, she is feeling fit after about with the flu. Paris Martineau, investigative journalist at Consumer Reports, the Babka curator.

[C]

It the Babka really did cure me and Tamiflu and probably whatever residual immunity was from a fidaloo shot, you know.

[A]

And youth.

[C]

You are young, probably a lot of youth.

[A]

Yeah, yeah. Well, it's so good to see.

[C]

But honestly, the prospect of getting to talk about AI with you guys also cured me.

[A]

There's some big news which we will get to this week. But one of the things that's going on right now in Las Vegas, Nevada is the what the show formerly known as the Consumer Electronics show. Only CES these days and there are is I thought, and I think I'm right, this was going to be a big AI show with not just AI empowered devices, but robots as well. And we thought, well, who do we know that's on the ground there? And of course we do know somebody, the editor in chief of the deep view and AI newsletter, Mr. Jason Heiner, who's in the airport lounge at McCarran. Hello, Jason.

[D]

Hey, thanks for having me. So glad to be here and so glad it worked out. And these are lots of good stuff to talk about for sure.

[A]

And you're in a good mood because you're leaving Las Vegas.

[D]

Never in a better mood than you are when you're leaving cbs.

[A]

The wild thing In Vegas. And I'll never forget it because we, you know, I never go there except for the, you know, Consumer Electronics show and before that, comdex, you know, the trade show.

[B]

Didn't you go there for a rock buying thing or something? Your wife there?

[A]

But, but, but we go. We actually, I should say that we have been going down there for entertainment. Like if there's a. A band we want to see and stuff. But go to the.

[B]

Or.

[A]

I remember going in the cab to McCarran, the airport, and you'll see people walking to the airport dragging their luggage behind them. And the cabbie says, yeah, those are. Those are the folks who even bet the cab money home. That's when you know it did not go well. Of course, at events like ces, the tables usually have tarps over them because the gam, you know, the geeks know better than nerds.

[D]

Don't gamble.

[A]

No, we don't gamble. We just go for the show. So I had heard some rumors it was a little smaller than usual. Was it?

[D]

I don't think it was much smaller than usual. It was 140,000 people, 4,000 vendors, you know, 5,000 press. So nothing about it felt small. It felt like it had a few years where it was contracting. And I really wondered, like it. Does the show have a future? But the last three years, it's become a lot more. There's a lot more things that are. That are especially triggered by AI that have become part of the show, and enterprise is one of them. We can talk about it, but it certainly feels like it's on the upswing again.

[B]

Jason, a lot of it now is AI means. A lot of it strikes me it's not consumer per se. Right. A lot of it is industry video. I watched the keynote and it's all. There's another consumer there. It's. It's very heavy on industry.

[D]

That's exactly right.

[A]

Isn't that strange? It's not a consumer electronic show fully anymore. Maybe that's why they took away the name.

[D]

It's funny. Yeah. I think it was wise that they.

[C]

Took away the name for sure. No longer stands for Consumer Electronics Show. Did I miss that?

[A]

No, it has nothing.

[C]

What?

[D]

It's like kfc, you know, there's no. There's no nothing fried involved.

[A]

It's still put on by the Consumer Electronics association, though, right? I mean, they still call them the cea. Still calls themselves that.

[D]

Well, it's the CTA now, the Consumer Technology Association.

[B]

But it's still consumer.

[D]

That's right.

[A]

Used to be People would go there, vendors would go there because they were consumer electronics shops. And so they would go there at the beginning of the year to see what they were going to stock for Christmas the following year or that year. Now it's a bunch of different things going on. So we saw chip makers there, intel, amd, Qualcomm, even Nvidia, all announcing new chips, most with AI capabilities. We saw laptop makers there announcing new laptops. Although Dell interest. Was it Dell that didn't mention AI in its laptop presentation? They thought they would be a little different.

[D]

Yeah. It's a surprise if they did. I hadn't heard that, but Lenovo certainly was all about, you know, copilot PCs. Yeah, exactly, exactly. Copilot plus PCs. I really haven't seen anybody yet that wants to buy a PC for the Copilot button. But you know, there it is. They're trying.

[A]

I want to take it off. I can't figure out how to get it off my Lenovo X1 carbon. I'm using Linux. I don't have Copilot. I don't want Copilot. But there's a button. You can't get rid of it. And it doesn't do anything either. That's the other problem. I know Normal.

[D]

Can't remap it actually on Linux. Could probably figure out a way to leave it.

[A]

I'm gonna, I. It's my goal to figure it out.

[B]

Yeah. At least Google lets you, lets you change their Google key on Chromebook.

[C]

How many times do you think you heard AI in a 24 hour period?

[D]

Oh my God, yes. Paris.

[A]

One of those little calendars that they use at the night.

[C]

You should have gotten one of those little.

[A]

Oh my gosh.

[D]

Yeah. It was everywhere. And it's funny, you know, one of the things that we said going into the show from the deep view, you know, because all we cover is AI, is that, you know, our mission is to, to find and sort and focus on the stuff that's actually real AI products and not, you know, AI washing or, you know, AI hype. There was plenty of it, for sure.

[A]

Yeah. Let's start with Jensen Huang and Nvidia. I know Jeff watched the show and in fact watched the keynote and he said we should interview Jensen's coat, his famous leather jacket.

[D]

Oh, yeah, if we.

[B]

It was amazing this year, but I'm a connoisseur of his keynotes. I watch every one of them.

[A]

He's really good at it.

[B]

He's really good at it. Everyone has a message, right? I learned about the Token economy in one. Another one emphasizes scale, another one is about digital twins. But this one was a very Ruben.

[D]

In another life he could have been an amazing educator. Like a teacher.

[A]

Yes.

[B]

Worth a lot less money. Worth.

[D]

He would be worth an absolute fraction of, of what? A rounding error. Of what he is. But, but he would, you know, impact a lot of people. No, he really, he does like teach and communicate these very complex topics beautifully. And, and I think even he came on a. He appeared in a couple other keynotes because like all their partners, they want to bring him on. And it's funny, there are a couple times where they explain something and he's like, he sort of came around and gave a. And it's also this which breaks it down. That's like really? Oh, that. Yes. So he, he's just such a good communicator in that way.

[A]

Yeah.

[B]

So even though there's tons of jargon and tons of numbers and brands, but you just get a sense. And the way he stood there before the. I mean it's a chip. You can't see anything in the chip. It's this inanimate object. But you start to see the power of the Vera Rubin in what it builds and the size and scale of it and it's awe inspiring.

[A]

The Vera Rubin is the new platform that Nvidia announced. Your reporter Nat Rubio Licht said Nvidia is now outdoing itself.

[B]

Yeah.

[A]

What is new about the Vera Rubin?

[D]

The, the crazy thing about it is, look, they already have the most powerful chips in, in to. To train AI for inference for all of the things. Right. And so they are leading this industry by such a wide margin that they have 12 months of orders. From what we've heard ahead, like if they stop taking orders today, they have everything that they make in 2026, they've already sold. But they even with that and even with the competition having trouble keeping up, what they announced was very surprising. So they talked about that these new chips can do training for a tenth of the token cost. So 10%. That's incredible, right? Like, because one of the biggest problems in AI is it just costs so much to do it. Right. It also requires a quarter as many GPUs to train the models compared to Blackwell. Blackwell is the cutting edge platform, you know, that is that everybody's trying to get a hold of and that they have so many orders that they can't sell. And they, they not only did that, so they did those two things and then they essentially rolled it out about six months ahead of what we expected there to be, they now see huge leaps forward and then they are also ahead of schedule. So I don't know what's in the water over there at Nvidia, but they, they are doing things that are pretty wild and we could talk about there are some, some bigger implications sort of where all of this is going. I think there are some, some counter trends. But for what Nvidia announced, it was crazy. And the thing that, to Jeff's point, they communicate that so crazy. Like those things that I just, you know, said are very clear compared to this one. Here's what it does. A lot of the other chip makers, they talk about it's this many petaflops and now it's this many Yoda flops.

[A]

Yeah. What does that mean?

[D]

And you really, you, at the end of the day, you kind of have no clue or context for, for what all of that means at times, unless you're really, really deep in the weeds. I do this stuff every day and I was going through AMD's keynote and I was like, okay. I was doing maths and putting stuff in chat bots and calculators and trying to figure out how that comp shared. And at the end of the day, Lisa sue, who's also a good communicator from AMD said, and by the way, what this means is that in four years we've essentially increased the power of the chips that we use for AI by a thousand x over a four year period. And I was like, okay, now that's.

[B]

A number I got. Jason, let me, let me ask you for a sec because I'm fascinated by Nvidia being that it's in the chip business, obviously it's in the rack business. Cuda is an operating system. He made very clear this time how much he's in the open source model business. Yes, model business for other areas including automotive, that was a big announcement. But also I think that when it comes to the LLM versus World model world, he has to straddle it all. But he seems very much in the world model in the Yann LeCun team versus the, the classic AI team. How do you sense the, the kind of, the mix of what Nvidia is as a company now? Because we, it's thought of as a hardware company. But, but how would you characterize it in total?

[D]

Yeah. Thank you. I would almost think of it more less like a chip company. They're going to still make chips. They continue, will continue, but more like an infrastructure, the infrastructure layer of AI Right. That all of these different models are going to run on their hardware. Not only LLMs, but SLM, small language models, domain specific models.

[B]

Right.

[D]

They are going to be in world models. I think they also agree to your point. They agree that that's really the next thing because LLMs have their limitations and world models, you know, are taking in more real world data. That's where what's going to unlock robots, it's going to unlock more automotive. Automotive, all of those things. So, yeah, so I, I think that the other thing that they announced there a little bit more under the radar was they're a big. With Siemens, which is what Jensen Wong came on. Siemens. Siemens is an enterprise company. So on opening day they had an enterprise company, Siemens, give the opening keynote, which was, which sort of tells you all you need to know about how different CES is. And, but he came on and he said Siemens is the operating system of the manufacturing companies across the planet and they are partnering, the two of them with Foxconn and with some other hardware makers to basically provide the blueprint to out these, you know, new data centers. They call them AI factories, data centers that are just purely aimed at AI and by working together sort of software, you know, Siemens essentially does digital twins. Really, it's the world. This gets to your world model thing. You know, Siemens does this digital twin thing where they can take anything in the real world. They put it, you know, make a 3D model of it. But now it's not just designing things. It actually takes that 3D model and uses AI and world models to then simulate what is happening in the real world, how a car will drive, like in all the edge cases, places, how to build data centers so that they're completely optimized. They're going to save a lot more power. They can make little tweaks and do things, so they can then deploy things a lot faster, a lot more efficient, a lot more efficiently and with a lot less power. So things like that. I say all that just to say that Nvidia's ambitions and its reach is so much further than it's ever been, far beyond just chips themselves.

[A]

We're talking to Jason Heiner. He's the editor in chief of the Deep View, which is his new job. He's editor in chief of a newsletter that specializes in artificial intelligence. So I think we're going to be hearing more from Jason on our shows. I'm, I'm really excited about the Deep View. You can subscribe right now for free. If you go to the Deep View as You can see I'm looking at their rundown of AI stuff from ces, but they call themselves Intelligence for the Age of Intelligence. And Jason, by the way, will be back with us on Sunday. We're going to do a CES special Twit with Father Robert Balisaire, who's there right now, Jennifer Pattison Tuohy of the Verge. She is, of course, all over home automation. There's a lot of home automation stuff there. And Jason will be back, so that's going to be a deep dive. We'll have more time with Jason. He won't be flying out this coming Sunday. What about robots? Because that was the other thing I thought. I mean, it's still AI, but that was another thing I thought we'd see a lot of at ces.

[D]

Yeah, there's plenty, plenty, plenty of robots. Robots have always been at ces, as, you know, Leo, from all the times you've been. But it was always a bit of a novelty. It was always a sort of. It was one of the things at CES that are product. You put them in the category of products I'm never going to see in the real world or that I couldn't afford. And so this year is more about products that you can buy for one, consumer products. And they raise some big questions as well. And we can talk about some of those, because a lot of them were robots for kids and children. And when you think about, like, AI, deploying AI for kids, all of a sudden that brings up a whole lot of questions about privacy and also unintended consequences and things like that. So we can talk about that. But the new buzzword for the, for robots is physical AI. This includes a lot of things, but really that ranges from humanoid robots and these, you know, toys, really kind of robot pets, to other things like arm. These robot arms that can work in factories or they could do things at home. You know, I think there's this sense that, you know, the humanoid robot is one aspect of robotics, but likely this physical AI movement is going to involve robots in a lot of different kinds of form factors. And we've seen a little bit of that already. Right. Like robot vacuum is one of the robots that probably most people are probably most likely to have in some form or another, you know, in their home. It's not a humanoid robot.

[A]

Right.

[D]

It's a little disc that sort of runs around and uses sensors and tries to do smart things sometimes also does dumb things. But, you know, those are. That's a good indication of sort of where I think A lot of the trends are going to be. Are going to be when you. When we get to. When we get to these things becoming part of the real world and then products that you want to buy. Now, all that said, there still were not a ton of products that you're going to buy. There's not a lot of robots you're necessarily going to buy, except for these ones that were more like toys and products aimed at kids.

[A]

And so California's got a bill. Padilla has proposed a bill that would ban AI toys for kids for four years. So I think there might be some pushback. We already saw one AI toy that was retracted because it taught kids how to do things like make bombs and probably not the best use of AI in the home, but there is. I saw one robot, I don't know if it's from Hero, that actually folds laundry or you can. Will do your laundry for you that you can. Is there anything there you could buy today? I heard that you could buy that or soon, anyway. The robot for laundry, the laundry folding.

[D]

Robot is like the holy grail, right?

[A]

And in the past, it's been a human on the other side, in my most cases, who's doing the folding and then having the manipulating.

[C]

Get a laundry folding robot debuting at CES every year.

[A]

Every year that we're gonna see one.

[D]

No, it's not the year.

[A]

No, no.

[D]

But, you know, the dream. The dream is still there. And look, these things are getting smarter and smarter. I think one of the things that robot makers have learned, and I'm sure you all have talked about this a lot on the podcast, I'm sure the show, is that there are things that humans do that are very, very difficult to replicate. You know, when we pick up a glass, when we pick up versus a mug, like, we know that, oh, intrinsically, we don't think about it, but we just, in sort of living life, we learn that, oh, I can put less pressure on my forefinger when I pick up this one, or I need to, or else I might break it. Right. Like, it's very difficult to program a robot to do that now. They're starting to. With some of these world models and, you know, physics models and things like that, they're starting to be able to do some of it and to be able to train robots to be able to. To learn these things, but it's very challenging. Like, there are a lot of things that humans do, especially dexterity and. And laundry, I think, is one of those.

[A]

One of the toughest.

[D]

Dirty. Yeah.

[B]

One of the lowest Values in lower.

[A]

I don't understand. Yeah, I don't understand why that's. That's the focus. Maybe they've got research that says nobody.

[B]

Likes people talking about how hard that is to do that. If they. It's like an ag. It's the AGI for robots.

[A]

This Cloyd, which is a terrible name from lg, it's a butler that will fold laundry. But they have it also. Well, it will bake, it will do some light baking. This is from cnet. Folding laundry. But slowly. Light baking. Simple tasks like sticking a croissant in the oven, but you can remove the croissant. So what you're saying, Jason, is these are all concepts. These are not yet anything you're going to be able to buy.

[C]

Okay. Along a similar vein, I have a question, which is what is the dumbest product you saw at ces?

[A]

I'm sure there's more than a few that's hard to choose.

[D]

The list is long. The list is long and distinguished, for sure. So, you know, one of the things that raised the most questions for me were these robots for kids. There was one robot for kids that was basically just a sock with an eye that is like an emotional support robot that follows you around and tries to learn what makes you laugh, what makes you happy.

[A]

An emotional support sock.

[D]

It's a robot that mobile.

[C]

Emotional support.

[D]

That's right. That's right. There were, there were several of these at CS Unveiled that were these robots that were essentially emotional support robots. And the thing with it is, you know, one person's emotional support robot is another's emotionally manipulative. Right. Device. Right. Or irritant, for sure. And so I, I sort of had this question about not just like, was this a good idea? Would it be helpful? But it, it was in a sense, addressing a little bit of, like, loneliness of that kind of thing. And I just feel like. And, and it's funny. Readers wrote to me after we wrote because we, we sort of raised some of these questions on the inner. On the newsletter at the Deep View yesterday. And I had sort of raised some of those. And I had a reader that wrote and said, shouldn't the promise be of. Of AI be not that we're going to create emotionally supportive robots that you need, but it's going to do more things for you and you can spend more time with your kids? Like, wouldn't that be the better solution?

[A]

More time with a robot?

[D]

Yeah. Yeah.

[B]

Did you see the.

[A]

This is one of the robots that. Emotional support robots that won an. Apparently an Innovation award. This is from a Chinese company. It's a biometric effective AI Panda. I think one of the reasons you're seeing these is they are not challenging to make. Right. They just sit there and purr. You know, they don't do a lot. Here's a child shaped robot. That's absolutely creepy.

[C]

Oh, no.

[A]

Yeah.

[B]

And here, did you see the ami, the AI soul mate for the lonely remote worker.

[A]

Yeah. What do you think?

[D]

I have not seen that one. Oh, my goodness. But yeah, there were so many of these. This emotionally support, emotional support robot was like a whole trend. There were.

[A]

Wait a minute now this is a sex robot.

[B]

Well, there was that too. Yes. Love sense that.

[A]

Oh, okay.

[B]

Lines 108 and 109.

[A]

Okay, we'll get there.

[B]

One is. Is the AI.

[A]

Why does it always end up. It's so sad. Why does it always end up being that?

[B]

Well, see.

[A]

Yes.

[C]

And I don't like it.

[A]

Oh, okay. And these are intended for lonely adults.

[B]

Well, so the 109, the AMI sits on your desk. It's this little thing and it gets to know you and it listens to you all day. And it knows what makes you. Same thing as the sock. It's your worker sock.

[C]

Three cores. The mini core, which is described as portable presence, endless connection. The baby core, described as the emotional heart. And the drive core, which is described as mobility with a futuristic flair.

[D]

So do you all remember the Jibo social robot? This is from about 10 years ago.

[A]

Yeah, yeah.

[B]

Wasted money on all kinds.

[A]

I have so many terrible robots.

[D]

Leo has them all. The company that made Jibo had what I thought was like maybe the most polished of these. And they had actually had two products. One was called the Luka Robot. L U K A. And the other was the Luka Luka AI Cube. So the Luka robot is a multilingual tool that can read stories to kids. So they. So, so they had a sort of non AI version of this that's been out for a few years. And apparently there's over 10 million families that have already, you know, have one of these. So it essentially, you know, reads to kids.

[A]

Now, I think, a cautionary tale, I should point out, the Jibo, which I bought in 2014, stopped working when they turned off the servers a few years later. So you're running a little risk with this company if you buy one of their robots, that they may discontinue it at some point. Right, for sure.

[D]

For sure. So this one reads to kids then with AI the new version with AI sort of can turn some of the stories into conversation based interactions even. We'll do some, you know, animation of, of things and, and then this little AI cube is almost like a ruggedized tablet that they wear around their neck. This is probably aimed at more, I think like kindergarten age sort of kids where they can go and they can take photos of things in the real world and then ask questions about them.

[A]

Is it still Cynthia Brazil that's doing this? She did the Jibo. The reason I bought the Jibo is because she was a very credible MIT robotics researcher. So it had a good pedigree. I bet if it's the same company, it probably is still her and she really believes in this stuff as a robotic companion.

[D]

Yeah, these are all very well intentioned. Like I don't think these are people trying to make a quick buck. I think they're well intentioned. They see a lot of loneliness and isolation in the world and they're trying to say, hey, could we use this to help? I still think the larger question still applies, right? Like, do we, do we give in giving these things to kids? Like, do we have. We really thought through all the potential consequences, you know, and when we asked the question in our newsletter, we were like, would you. So all of our sort of like, you know, half a million readers, we asked them, would you give a robot, an AI robot to. Or a robot, an AI powered robot or toy to a child in your life? 50 said no right off the bat.

[B]

I'm surprised, not higher.

[C]

Which is incredibly impressive for the subscribers to an AI newsletter.

[B]

Yes.

[D]

Yeah, exactly.

[C]

I also, I also just need to state for the record, TLC programmed the AME to feel sadness is something I just learned while scrolling through this. And I think we all just need to reckon with that.

[B]

How dare you sadden. Your robot doesn't need that.

[A]

Go ahead, Jeff.

[B]

What about the whole. Samsung had a whole there the refrigerator you can talk to and open the door. We see the AI in everything. Did you see anything credible in terms of a consumer touch to AI and in a new use case?

[D]

Yeah, I'm glad you asked it because I didn't see anything that I know anyone that I know in like real life would actually want or that I could recommend to somebody. You know, why would I want to spend that much more on a refrigerator? And all it does is look in the refrigerator and tell me, oh, here are a few recipes, recipes for stuff you already have in the fridge. I just can't, I just can't imagine that many people doing it. And it's a lot of Money to spend extra money to spend on something that, you know, does something that ultimately probably within a year or two, you know, any of us could point our, we could open the fridge, probably point our, our phone at it and, and then ask like, what should I make from this? And it'll be able to, you know, scan everything and tell us. So I just think a lot of these things and a lot of them were like that, Jeff, like that. They are sort of solutions looking for, you know, people they could convince to try it. And none of them, I didn't see literally any of them. I looked, I watched that, that whole keynote as well, and I just felt like, boy, they're trying really hard, but there's nothing in there that I could see anybody I know in real life getting excited.

[A]

Was there anything at all that you thought was worth going to Vegas for?

[D]

Yeah, for sure there was. And so I know we, we talked a little bit about this. So the, and this is a surprise. And also I have some bias to this, right. I worked in enterprise for a long time. AI was a show where, you know, you, there might be a smattering of enterprise things here and there, but it was not, it was never an enterprise show. At this one, there's a whole pavilion, North Pavilion, which used to be all of the automotive stuff that's all moved to the new West Pavilion.

[A]

Really? Oh, interesting.

[D]

Yeah, yeah. North Pavilion is now all enterprise companies. Like there is a whole show that is essentially enterprise, you know, led by Siemens. Siemens did their, their big lead keynote. Siemens made six huge announcements that were, that were enterprise related. And this is where I, I see most of the momentum in AI when it comes to consumer AI. A lot of consumers are still really skeptical. I don't know if you all see this, but I see this a lot. They really see, they see t. AI good for two things in my experiences. One, asking it questions that they used to ask Google and maybe it's a little bit better a chatbot. And then two is laughing at AI slob. They, they see it and then they send it around to the, they text it around to their friends and they have an amazing laugh about it. Right? Like, so those are the things most of the use cases I see with consumers are that all of these products are trying to convince consumers that, you know, AI is going to do something for them. And look, it's a hard sell. Consumers are, are just not really sure and they have some fear and some trust issues and rightly so. I think where we're see of the momentum in AI is actually businesses and companies using it for automation and using it for all kinds of things. And it is helping people be more productive. It's helping in a lot of ways. But there are also huge questions around roi because AI is incredibly expensive. And so that gets to a little bit. We were talking earlier with, you know, Jensen Huang and Nvidia with these, you know, incredibly powerful models. The counter trend that I see, though, and I think was going to be a big one for 2026 when it comes to AI, is companies using smaller models, using domain specific models, and using world models, as we talked about earlier, to do a lot more specific things and to do them a lot cheaper. Right. Because you can run a small model and it's a lot cheaper to run. You can run it on older hardware that's two, three, four generations old and it's good enough and it's a tenth or a twentieth of the cost. And then when that happens, all of these things that sort of happen in business and organizations and all that, then they get a lot more viable because they're going to be a lot cheaper. Because right now it's hard to have ROI on a lot of these ro on these enterprise projects because it's just so expensive.

[A]

The other group I would add is developers. And this is in a way an enterprise story too. But I for sure am blown away by what Claude Code can do. And I know a lot of people who are developers are saying this is breakthrough time. That this is really where you're first seeing what I would say is almost AGI. So, yeah, enterprise, and I guess developers is a. Is a, you know, a branch of enterprise offshoot. And, and I think we're. It's funny because I think those people who are using AI in those contexts are seeing a very different product from what consumers are seeing. They're seeing the chat products and it's, you know, of mixed value. Jason, I really appreciate your taking the time. Yeah, I know you got a flight to catch. Jason Heiner is editor in chief. Brand new job. It's great to see you working at the Deep view. The deep view.com you can subscribe to their daily newsletter. Keep up on what's going on with AI, particularly in enterprise intelligence for the age of intelligence.

[D]

Subscribe. That's the deep view.com too. Yes, subscribe.com.

[A]

Okay, so thank you. It's much free. There's no charge, which is. Which is awesome.

[D]

That's right.

[A]

Yeah, yeah. 51,000 readers. Nice. Thank you. Jason Heiner. We'll see you on Sunday for a full CES on Metro Twit week. Tank. Take care.

[B]

Go get some nice cheese cubes there in the lounge.

[A]

Which lounge are you in?

[D]

Thank you. I mean, it's called the club. It's in sounds D. Yeah, terminal D. Very fancy.

[A]

Okay.

[D]

Thank you all for having me. Great to see you everybody.

[B]

Take care.

[A]

Jason. One of our favorite people, Jason Heiner. It's great to see him. From the deep view.com we'll have more of Intelligent Machines with Paris Martineau and Jeff Jarvis in just a moment. Our show today brought to you by Monarch. We love the Monarch. I'm a Monarch user. I have been for a year. It's been a very good year for me because Monarchs helped me really keep track of what's going on with my money. Wouldn't it be nice if there are a way of reducing money stress? You know, that I've seen this time and time again. The number one issue in couples often is money. That's the number one thing they stress about. They fight about, they disagree often about. It doesn't have to be that way. Managing your money doesn't have to be a struggle. This year. Monarch, it's the all in one personal finance tool designed to make your life easier. It brings your entire financial life. Life budgeting accounts and investments, net worth and future planning together in one dashboard on your laptop or phone. Start your new year on the right foot financially and get 50% off with your Monarch subscription. With code. I am. I highly recommend this. I've been. I love it. It's very easy to set up. I got all my accounts in there, including all my investment accounts, my bank accounts, my credit cards that I even put my house in there. It gives you your total net worth. And it's a look this. You might say, well, I don't want to do all this. It take. It'll literally take you five minutes, ten minutes at most. Depends how many accounts you have to set up. And now you're ready to start fresh. You know, after a chaotic year, chaotic holidays, this is the go to tool for a New Year's financial reset. You can review everything you've spent over the holidays because it'll pull all of that in. You could set fresh budgets. Okay, we're not going to spend that much money on eating out this year. Getting ready for 2026. You get automated weekly money recaps. No judgment. But honestly, the worst thing is to not know what's going on. Right? You'll be able to track progress towards future financial goals. It is easier than ever with Monarch to stay financially fit both in the short term and the long term. Unlike other personal finance apps, Monarch's different. It's built to be proactive, not reactive. Okay. They have AI tools built in that I think are very judicious and intelligently incorporated. They call it Monarch Intelligence. It's better. It's better because it's trained on the authentic collective wisdom of certified financial planners and financial advisors. And it has your information. It's. It's not sending it out to the cloud, but it has your information right there. So it's able to look at what's going on and give you some real advice. Here's what a survey Monarch just did last year at, well, just ended it. 2025 survey of monarch users. This is what they what happened for them when they were using Monarch in the year 2025, Monarch helped users save over $200 a month on average. After joining.$200 a month, 80% of members feel more in control of their finances with Monarch. I'm in that group. 80% say monarch gives them a clearer picture of where their money's going. My son at Christmas said, hey, I just, I don't know where my money is, where it's going. I'm making good money. But now I think I need to plan a little bit better. I set him up with Monarch. I want to set you up with Monarch too. This new year, achieve your financial goals for good. Monarch is the all in one tool that makes proactive money management simple all year long. And again, use the code IM IM for intelligent machines at monarch.com you'll get half off your first year, 50% off your first year at Monarch. M-O-N-A-R C-H.com with the offer code. Im highly recommended, Monarch. Thank you Monarch, for kicking our year off. Right. We appreciate it. Jeff Jarvis, Paris Martineau. Let's catch up. What's been going on with you guys over the last few weeks?

[B]

Oh, you're muted.

[A]

Yeah, I can't hear you.

[C]

How was the end of your guys's year?

[A]

It was good. We did a 24 hour New Year's special and I left you out. Oh, we forgot to ask you next.

[C]

Year I'll have to do it.

[A]

You've been lobbying. It's so good.

[C]

Someone in the chat in just the general was like, I heard just someone mention a 24 hour New Year's dream. Where is that posted? And I was like, God, I wish, I wish I knew.

[A]

Did you see my response to that?

[C]

I did. I Did.

[A]

She's relentless.

[C]

It started as a bit, but now it's become a lifestyle.

[A]

Yeah, well, we didn't do that. We went out to a rock show for New Year's and left early so we would be in bed by midnight. Did you. You had a party? I think I did.

[C]

I had my annual New Year's Eve Eve party, which is where on New Year's Eve Eve, you have everybody over for a raucous. Basically a New Year's Eve party, but the day before, and then.

[A]

That's smart. So that they're like pregame New Year's Eve. Yeah.

[C]

And you watch last year's ball drop. And then. I do. I did. And then throughout the night, you play, say, other notable years. Ball drops. 2000. Great. Ball drop.

[A]

Was that a good one?

[C]

The Y2K one? Yeah.

[B]

Is there a site with just the ball drops?

[C]

No, just YouTube.

[B]

Okay. All right.

[A]

You know, it'd be really fun if you had all the different foam hats that they.

[C]

Oh, that would be good. Are the various glasses that get worse and worse after the year 2000, after the 2000s, generally.

[B]

In the early years of. Of webcams, I remember we had. We had a webcam out of Conde Nasty that I put up there on one, and we pointed at the ball. Oh, yeah. Because we were right across the street.

[A]

The ball's gotten fancier, too. It's Swarovski crystal.

[B]

Oh, it's. Yeah, it's.

[A]

It's a fancy ball.

[B]

It's like everything in America gets tacky.

[C]

One of the ball drops. We were watching had a little segment on one of the men who blows the glass for the ball. And I'm like, if you got people custom blowing glass balls every year, that ball should hit the ground and shatter.

[B]

Should.

[C]

We should have some sort of pizzazz to it. But no.

[A]

Very good.

[C]

Yeah.

[A]

And it doesn't go down. It kind of. It's. It's. It's like somebody's lowering it with a rope. It's not. It's. It's not good. They gotta work on that. Plus, I. I understand. Have you ever done the Times Square thing in person?

[B]

Oh, God, no. What, do you. You think we have no iq? Who do you think we are?

[A]

Apparently no New Yorker will actually do that.

[D]

No.

[C]

A diaper. If you go do that.

[A]

Well, that's right. You're put in a pen. For how long?

[B]

Oh, like 10 hours, eight hours or something.

[A]

And you're not allowed to leave the pen?

[C]

No. That's why they have to wear a diaper.

[A]

It's crazy.

[B]

That's New Yorker. Leo. Leo.

[A]

It's all people.

[B]

Well or not, we never go to Times Square, period. We walk blocks to go around Times Square. We do not go to Times Square. And I always tell the out of towners, my one rule in life to be happy in New York is never hug an Elmo.

[A]

Oh my God.

[C]

Yeah. You don't know what you'll catch from those.

[A]

Costumes are not clean. No, they've been dragged through the, the.

[C]

Lighting in Times Square just feels unnerving because of the density of ads. It's just unnaturally bright at all hours of the day. Which makes the dirty Elmo costumes look even dirtier.

[A]

I, I actually I am such a bad tourist because I always, always go.

[C]

To tourists, always want to go.

[B]

The tourists want to go. But it's like San Francisco. I always admired San Francisco. Fort ghetto Wising the tourists. No San Franciscan ever goes to Ghirardelli Square or Fisherman's Wharf. Yeah, tourists there.

[A]

Yeah.

[B]

And, and, and, and, and, and near the twain shall meet. I'm old enough to remember when Times Square had the camel smoking billboard.

[A]

Oh yeah, it had a, it actually little puffs. Smoke come out of the smoke. Oh, I remember that too.

[B]

Oh, maybe you can find the video of that.

[C]

That.

[A]

Well, so much happened while, while you were gone, while you were taking a couple of weeks off. While we were all taking a couple of weeks off. Nvidia did an interesting thing on Christmas Eve. They hollowed out a company. And actually our friend MG Siegler has made a name for this. He calls it a hack wisition. You know, first there was the acquisition, then there was the Aqua hire. This is the worst of all. Instead of, and we've seen it happen before, instead of buying a company, they just get the brains of the company. They hollow the company out, leaving the poor employees behind. So Nvidia bought an AI chip startup called Grok G R O.

[B]

They didn't die. They, they licensed.

[A]

You saw I put it Grok.

[C]

No relation.

[A]

Yeah, no relation to the other Grok, which will also be talking about they spent $20 billion, that was the CNBC headline. But it turns out they were just licensing a non exclusive license for the Grok technology. Except they got the guy who invented the TPU for Google, who was the founder of Grok, the CEO Jonathan Ross to move over to Nvidia. So he left Grok. Apparently company president Sonny Madra also left kroc. So this is an example of a. And we saw, we saw this company continues.

[B]

It can still sell Its products.

[A]

It's a very strange.

[C]

It's just a zombie now.

[A]

It's a zombie. I think. Yeah, it is.

[B]

And it is, I think because it's still free to sell things to other people. And I think what's it. What impressed me about this from what I've read is that. And this is Jensen Huang. I'm going to start joining the Jensen Wong fan club if I'm not careful. He recognized what they didn't have. This is about short term memory for output of AI rather than training of AI. And it was kind of a weakness. And rather than saying, well we're going to build that, he acquired it and he got the brains behind it. It was very smart.

[A]

The guy who invented the tpu.

[B]

Yeah. Yeah.

[A]

Okay. That's a. It was an interesting thing. The fact that they did it on Christmas Eve. Maybe I'm too old school. Kind of implies that they were trying to kind of bury the.

[C]

Yes, that's classic. Bury the headline.

[A]

Nobody's paying attention. They're busy wrapping presents or having pre New Year's Eve parties or something.

[E]

Doesn't this just sound like private equity though? This is what they do, right?

[A]

Yeah. Except they didn't buy anything.

[B]

Didn't buy anything. Yeah.

[A]

It's a non exclusive license. It's a very strange thing, but one.

[B]

Thing that struck me from the. From the Jensen One keynote at CES and he says this all the time, but he really emphasized it this time is open source. Is that the models.

[A]

I'm glad you hear.

[B]

Open that.

[A]

I want to. By the way, I really want to make sure we don't call it open source. We call it open weights.

[B]

Yeah. Thank you.

[A]

Because it. Open source implies that somehow the source code is open. It's not in fact. Really the model is available. You could download it and then the weights the way it's.

[B]

Jason and I call it Openish.

[A]

Yeah. Just open weights. I think it's fair. I hate the fact that unfortunately open source has been co opted for that. But I do agree with you. That's very, very important.

[B]

And so the more people use. I mean it's in their interest because the more people have more models to use for more reasons, the better off Nvidia is.

[A]

That's a good point. They're still going to sell the hardware both for training and for people who want to use it.

[C]

Yeah. There's a downside for Nvidia leaning into this. Unlike I guess Meta.

[A]

Right.

[B]

Yeah.

[A]

Which. Which has been the single most popular open weight model. The Llama model that was Pivoting away.

[C]

From it now because they're.

[A]

Well, we don't know if they are. There's.

[C]

He even showed signs that they're going to be favoring other options.

[A]

So the other Grok, the Grok, not with a cube Grok we don't talk.

[B]

About in plate company got in a.

[A]

Lot of trouble over the holidays. Bad Rudy's Grok, I think intentionally people were using GROK to make non consensual bikini pictures and nude pictures of people, even children. And Grok said well if you do that you could get in trouble. They didn't take away the capability. They also on a post a couple of Thursdays ago. Dear Community, this is from.

[B]

I like you doing your busk voice.

[A]

Dear Community, Some folks got upset over an AI image I generated. Big deal. It's just, just pixels and if you can't handle innovation, maybe log off. XAI is revolutionizing tech, not babysitting sensitivities. Deal with it unapologetically. Grock. Grock sounds like it's Grog. It's like a caveman AI Grock. Of course, you know, X is full of that crap. No matter what the fact that Grok could do it all just, just made it all with kind of.

[B]

And he got, he got what, a huge investment at the same time?

[A]

Yeah, it didn't hurt. It got a lot of attention. And now XAI is raising one of the largest raises ever. $20 billion. Super raise, $20 billion funding round. And this is for their infrastructure. Elon says we want to build the biggest, best training facility ever, which he said before they were looking for 15 billion. So this gives you some idea the market is.

[B]

That's what they're saying.

[A]

I believe it. Everybody wants to get into this, right?

[B]

But here's. So I, I did a British podcast today when they asked me, you know, what role Musk and Grok have, have in the AI world and I just dismiss it. I don't pay any attention.

[A]

I don't care. Somebody does. People on an X do. I don't hear anybody talking about using Grok, you know, for anything but sex.

[B]

A lot of money for.

[A]

Not for coding, which you can get.

[B]

Pretty easily on the Internet, folks. You don't really need to make it up.

[A]

Elon says. And when I say Elon says now you gotta take it with a good. A giant grain of salt that they have more than a million H100 GPU equivalents in their Colossus 1 and 2 supercomputers. This will allow them to buy many more. You know that money's going to go straight to Nvidia. So maybe that's why Jensen Huang is celebrating.

[E]

It's already not the top of the line either. It's already behind.

[A]

The H100 is behind.

[B]

Yeah.

[C]

Yeah.

[B]

Here's a question. Now that there's Vera Rubin coming out. I almost said Vera Wang. So now that there's Vera Rubin coming out and Blackwell becomes the also ran chip, do you think China will be allowed to get Blackwells?

[A]

No, China is allowed to get back Blackwells.

[B]

No, they get the H2 hundreds. Those aren't Blackwells, are they?

[A]

Oh, those aren't Blackwell. That's the previous predecessor.

[B]

Okay, right.

[A]

They do get H2 hundreds. They're trying to decide if they want.

[B]

That's the other question. If we won't take Huawei, then they're not taking our chips.

[A]

Yeah. I don't know what that all means. All right, so let's. Oh, there's a great picture, by the way. We should really share this because I know for all the things we say about Elon Musk, we still think the world of him. And of course, Jeff and Elon were there for the New Year's Eve celebration on Times Square. And while they weren't wearing the funny foam hats, they are clearly friends.

[B]

Yeah.

[A]

Happy New Year.

[B]

Secrets out.

[C]

It's true.

[B]

I'm having Elon's baby.

[C]

Yeah, his 26 is gonna be a draft, actually.

[A]

That embrace looks like you could be. We'll check back in in nine months.

[B]

Who did that? I want to.

[A]

Aaron, who should we yell at?

[C]

And I like that the AI generation goes so far as to have somebody slightly off frame taking a photo of you two guys just to get.

[A]

Just to make it more realistic. Yeah, well, it was two people. It was such a good photo. Two people had to get involved.

[C]

True.

[A]

It's true. Have we. Have we said everything about. Oh, I forgot the big story which you announced to us, Paris, while I was in the middle of the previous show. ChatGPT is going into the health business, which was really kind of obvious from their last event where they talked a lot about how, remember they had a woman on who had cancer and her husband, and she is saying, you know, my doctor really didn't explain what this test meant, but I went to ChatGPT and it did, and it was very helpful. Clearly, this was an ambition of. Of OpenAI. They are now introducing ChatGPT Health.

[B]

What could possibly go wrong?

[A]

Well, they say they're gonna. I mean, I think this could be. I use it for.

[B]

We know you use it.

[A]

Yeah, and I, and I pay it. No, it's been very useful. Somebody has just fed that picture to Grok. And this is the latest version of the picture.

[B]

Elon Musk.

[A]

Don't put Elon in a bikini.

[B]

Don't put me in one.

[C]

So cold out there.

[B]

Elon has a nice decolletage there, don't you think?

[A]

I, I, I think if used judiciously, AI is very helpful. I, I've used it quite a bit. Well, especially if you train it particularly on, you know, good health information. It should be good. Right.

[C]

Five paragraphs down into this announcement is this line. Health is designed to support, not replace medical care. It is not intended to diagnose as diagnosis or treatment. I think that's just part of the issue is that a lot of people, anecdotally, when I've heard them talking about ChatGPT uses for health, they're asking and relying on ChatGPT to translate their medical records to recommend possible treatment regimes.

[A]

Well, they go even farther because they're suggesting, just as our previous sponsor, Monarch Money does with your financial results, they're suggesting you connect your health data to CHAT GPT. So that is always reading your, your records.

[C]

Whether it's basically the launch of Chat GPT Health is I kind of read it as it's supposed to be a secure sub window of Chat GPT for all of your health records that you can set specific, I guess, privacy permissions, certain deleting timelines. CHAT GPT won't train on that data or OpenAI won't train its models on that data. But I just, I worry that this is going to incentivize a behavior that is already ill advised.

[B]

Well, it's also, it's, it's, it's, it's, it's blind men and elephants. I had a family member who had a health issue and I got the, the full lab results. I put all the lab results in various of these models and other information that was important, like what, what medication they were on. And I was wrong about the timing of the medication. One of the, one of the medications. And so the, the conclusion that the, that the unit came to was actually right, saying, danger, danger, Will Robinson. This is what if, you know, this is a problem. And then when I went back and I said, oh, actually I was wrong, they're not taking the medication was okay. Right. The problem was me. In that case, I got the information wrong. But it's too easy to do that.

[C]

Okay, yeah.

[A]

Leo, you don't have to use it. Well, I Signed it.

[C]

I'm gonna say we're saying this, but millions of people already use Chat GPT for this and millions more are going well.

[A]

That's right. That's right. So this would, if those people make it better. He says they say in the, in the announcement, 200 million people are asking health questions of chat GPT. So, so it's responding to a demand.

[B]

Do you remember Samir Arora as a new company where he got access to. And I don't really understand this well enough, and I'm not sure exactly what the brand is. It's a fairly recent announcement. We could have Samir to talk about it, where he got access that others don't have to. A trove of really credible and important health data. And then on that basis, what you build with it I think is important. So I'd be very interested to know what ChatGPT is sources are, because I think that tells you everything.

[A]

Well, it says over two years we've worked with more than 260 physicians practicing in 60 countries and dozens of specialties. This group has now provided feedback on model outputs over 600,000 times.

[B]

That's how it works material.

[A]

That is how it's.

[B]

But no, what's how you, what's, what's the original training?

[A]

That's reinforcement learning. That's how you train. In fact, every model you use use today is trained by humans with reinforcement learning exactly this way. But, okay, let me tell you what the training material is.

[B]

I'm sure it's betters. I mean, all that matters.

[A]

Yeah, I'm sure they're using.

[C]

Yeah, I mean, I think there's just a lot of questions related to this that I, I'm, I was surprised to see this launch because I, I would have assumed that OpenAI as a company would be wary from a legal perspective to be describing its product as offering health and medical advice.

[A]

Right. Yeah, they don't, I don't think they say in this article where they're getting, how they've trained it, whether it's just chat GPT.

[B]

Well, it's a key. Give me that. Yeah, because if you just use Chat GPT and then we, we reinforcement learning.

[A]

A little and provide rag your health information.

[B]

Right. But the rag I want to see is that it's not built on the, it's like you're using Notebook lm. Right. Notebook LM and RAG will stick only to the data you give. Won't go off and make up stuff out of the whole general use of the entire Internet. Right. And I think that's a critical difference here. And in health stuff, I would want to see it trained only on credible health information and, and then reinforced by people who matter, rather than using just generic LLM.

[A]

Okay, I'm going to use it. You don't have to. I think that the interesting thing here is this kind of tuning. I mean, we've been talking about this for a long time. Tuning it to a specific need and goal with specific information, specific training, and then taking data from your own sources. What are the problems? I don't know. Maybe you guys have better doctors who talk to you. I think I like my doctor, I think I have a good doctor. But I think the state of modern healthcare in the United States is that as in my case, when he prescribed Ozempic, did he give me any additional information about it? Did he tell me what some of the things I should do, like resistance training? No, I was. It was up to me to go out and get information from a variety of places, including the Internet and books and, and AI. And I honestly think that some of the material I may have collated from the Internet might not have been that good or from books might not have been that good. In fact, there were a number of books I found that were clearly written by AI. So there's a lot of misinformation. If you're not getting. Then this is why I think also AI is being used in therapy in psychotherapeutic ways by people, because we are in a situation, at least in the United States, and maybe not for you guys, but certainly for me, where you don't get that information, you aren't being provided with the kind of information you need. And so there's a need for it.

[B]

Yeah, that's a larger societal problem.

[A]

Right, right. I mean, did you go to WebMD when WebMD first emerged and ask it questions?

[C]

Once, many years ago, but then it told.

[A]

Where do you get health information from.

[C]

Peer reviewed scientific papers and journals.

[A]

You go out. So you say, okay, I got to figure out how Tamiflu works. You go out and read scientific papers about Tamiflu.

[B]

She would.

[C]

The Wikipedia page. Yeah, right. I think I like, I, I often will read scientific journals about stuff like that, but I'm a free.

[B]

Yeah, you are. She's a reporter.

[A]

Yeah. I think it's impractical and I think it's also very risky because there's a large volume of material and the amount of time you can spend going through that material is going to very much limit what you learn. And it may not be the most Important thing that you learn. WebMD has come under a lot of criticism. As a matter of fact, it was a site that many people used. But as you say, it was a.

[C]

Site many people used until the long running joke became that you'd go to WebMD for it to tell you that you had cancer no matter what your symptoms were.

[A]

I don't know if that's true, but.

[C]

That was the joke.

[A]

Yeah.

[B]

Well, this is also where patients like me. That cite was really important. It's also learning from other people who are going through experience. You're going through.

[A]

Yeah. I mean, I wouldn't say you. You do chat GPT to the exclusion of anything else you might want to do, including if you want to read scientific papers. Okay.

[B]

It depends on the doctors too. There. Some of them are. So I. So I had a procedure before our last show.

[A]

Yeah. How'd that go?

[B]

It's fine. I was able to sit. It was not a problem.

[A]

Good.

[B]

You don't want to know anymore. I know, but the guy said nothing to me. Absolutely nothing. Nothing at all. Yeah.

[C]

I mean, did you want him to say anything to you?

[B]

This was.

[A]

Yeah.

[B]

I want some information about when I could resume certain activities. I'm having cataracts. In a few weeks, you will see me looking.

[C]

Oh, we all wear an eye patch.

[B]

In solidarity, I think in sympathy. I think you should.

[A]

Although people I know have had cataract surgery, report that their vision improved dramatically and they're very happy they get it done.

[B]

But I went to my optometrist, who I like a lot my. My op, just to. To get the blank lens for the first eye. And he said that one woman came in, was really upset because what they do is they put in, you know, new lenses for you. And so that's why I get. I get a correction lens so that I will not have to wear it for distance anymore. But she didn't realize she was an artist. She didn't realize that when you do that, you also can't see close.

[A]

Right.

[B]

Right now I can take off my.

[A]

Glasses like you're wearing reading glasses. And I could read those of us with my own very well, are very used to getting putting. You probably see people do this all the time, Paris, looking at their phone like this. Especially us old people.

[B]

But that you can't do that after the cataract.

[A]

Can't do that anymore. No one told her that.

[B]

She said it ruined her life.

[A]

Oh. Because she's an artist. She wants that detail. Yeah, yeah.

[B]

And being able to get more information.

[A]

Well, that's the. Okay, that's my point is because we get.

[B]

I'm agreeing with you. I'm agreeing with you.

[A]

We have such a dysfunctional medical system that you're really. It's on the patient to get that information now. And as soon as they realize, oh, they can just ask ChatGPT, they're going to even be less.

[B]

You know what? It's just like news. It's on you to fact check it too.

[A]

It's on you. But I have faith in people maybe more than you do that they can do that. I mean, I think you're worried about the dumb people believing what they get from the AI. I have faith that their people are smart enough to figure it out. Yeah.

[B]

I wonder what AI tells you about vaccinations. I wonder what GROK tells you about vaccinations.

[A]

Well, that's why we don't use Grok, isn't it?

[B]

Right. But some people do.

[A]

Well, here's the good news. Those people should all move to Florida because Governor DeSantis has now announced that we have to react. Reject AI with every fiber of our being. I don't know how he's going to bring the power of the. Of the Florida.

[B]

He's gonna unplug all the electricity in the states.

[A]

He is. This is from Politico. He's emerged as a leading AI skeptic. He wants to spend his last year as Florida governor beating back the advancement of artificial intelligence. He says, let's not try to act like some type of fake videos or fake songs are going to deliver us to some kind of utop.

[B]

Let's not pretend Florida such a utopia either.

[A]

Did you. Did you. You didn't spend New Year's in Florida. You came home for New Year's.

[C]

Yeah, no, I came home a week in advance.

[A]

Your microphone is doing that thing again.

[C]

It's doing that thing again.

[A]

Yeah. I think it's some hardware gating.

[B]

I'm not hearing the thing.

[C]

I'm not.

[A]

Her level was going way down and the way up and the way down.

[B]

She just has more emotional modulation than we.

[A]

She modulates more. You have to learn. If you're on the radio, continue to speak cadence. Yes.

[C]

And this volume all the time.

[A]

Yes, if it helps.

[C]

Signing off.

[A]

Cup your hand behind your ear.

[C]

I'm getting word from Houston that this is working.

[A]

Now. DeSantis is a well educated fellow. I think he went to Yale, didn't he? He says the idea. I say the idea of this transhumanist strain, that somehow this is going to supplant Humans and this other stuff, we have to reject that with every fiber of our being. We, as individual human beings, are the ones that were endowed by God with certain inalienable rights. I've heard that before. That's what our country was founded upon. They did not endow machines or these computers for this. These computers.

[B]

The citizens were trying to find a thing.

[E]

Yeah, but it sounds like he's just scared of AGI. He's scared of super intelligence. Not like chat.

[C]

He watched the Matrix over New Year's, much like I did.

[A]

And he has introduced a slate of recommendations for Florida lawmakers, calling on them to require companies to notify consumers when they're interacting with AI. We have that in California already prohibit the use of therapy or mental health counseling through AI. Where was it Illinois that did that? And give parents more controls over how their children use the technology. They've done that in New York State. At the same time, DeSantis wants to restrict the growth of data centers that fuel AI efforts by stopping any state subsidies to tech companies and curbing such facilities from drying up local water resources. He's obviously read Karen Howe's book, but the Governor's Bill of Rights is. And, but I. My. I felt so bad. Christmas Day, I'm talking to my daughter Abby, and she brought up the water thing, and I said, no, no, it's not true. And I. We got in a fight over it. I feel bad. So I'm sorry, Abby, but that's just how that kind of misinformation spreads. People want. I think people want to believe it.

[B]

Yeah.

[A]

I wish I'd known the fact that golf courses in the United States use far more water than data centers.

[B]

Centers.

[A]

So let's. Let's get rid of.

[C]

Everybody loves golf courses and thinks that they're just a real boon to society with their backs.

[A]

And by the way, the golf courses are pouring the water into the ground. The data centers are recycling the water. They're not pouring it into the ground. They're not.

[B]

Well, grass is bad.

[A]

She did bring up the fact that Elon Musk's data centers in. Where is it? Carolina. Are using natural gas generators and polluting. And I. And I. I quoted our CJ Trowbridge, our guest of a few weeks ago. I said, that's not an AI problem. That's an Elon Musk problem. I didn't win the argument. Let me. So that's interesting. So we'll see. We'll see what happens. You know, this is where, though it's interesting, DeSantis is now actually going against the Trump administration which has said no state laws.

[B]

Well, he's trying to run for.

[C]

He's got to try and carve out something that separates him.

[A]

Exactly.

[C]

That positions him as a. Another possible stop option, I'd assume.

[A]

Oh, it's a. Oh, it's a political thing.

[B]

I'm shocked there's politics here.

[C]

I am shocked there's politics in my politics.

[A]

When we last left you, the Waymos were stuck in San Francisco because they're.

[C]

Still there to this day.

[B]

Probably was the traffic lights being out. It didn't know what to do. Yeah. Okay, so it wasn't because it lost communication with its.

[A]

No, no, no. The communication was still working. It was the traffic lights. Waymo explained that its self driving car technology is already designed to handle dark traffic signals as of now and successfully handled over 7,000 during the power outage, properly treating those intersections as four way stops. But during the long outage, their cars sometimes experienced a backlog when waiting for confirmation checks, whatever that. Like when we called the home office leading them to freeze in intersections. Well, that seems like a good response. So Waymo said on Tuesday they're implementing fleet wide updates to provide self driving car specific power outage context. See, this is why you do it. You gotta get it out there in the real world. Experience these problems so you can fix them and be ready for the next thing. By the way, I just saw a news report in San Francisco. Waymo is offering people on the street $25 to help them close the doors on the Waymos. Because frequently, and I bet you do this, Paris, you get out of the car and you leave the door open.

[C]

Never in my life. My age is the opposite. I get out of the car and I am like, I gotta close this door. And I close it too loud. And then I feel like I'm being a jerk because I've closed a door too loud.

[B]

Take that.

[A]

Well, you're doing the right thing. The Waymos as concert currently constructed, not.

[C]

Getting paid $25 to do it. 25 bucks be holding out.

[B]

How do they know that you're the.

[C]

One who closed in New York City?

[A]

But they want people who are walking by and see a Waymo that's stranded because they can't drive with the door open. They're just sitting there going, close my door, please somebody. If they, if you see that, if you're in the streets of San Francis, you see a Waymo with the door open, just go up, close it.

[B]

Well, how do you get the 25 bucks?

[A]

I don't know, they didn't. They didn't.

[B]

Well, that's.

[C]

I think every Waymo should have a little buddy that's injected in the back. It's one of those laundry folding butler robots that if the door is open, it pops out, it waddles over, closes it, and then deposits 25 bucks to the nearest passerby.

[A]

That's a good idea.

[C]

Pops back in.

[E]

Don't those cars already have like electric back? Like the trunk already open by itself? So like.

[A]

No, they don't. Now, by the way, if you're in Japan, this is why you're thinking this, Benito. And you close a taxi door, they're going to yell at you. Because all the taxis in Japan, am I right? Bonito? Have auto closers.

[E]

I'm not sure if it's.

[A]

And you'll break it, but. Yeah, well, you'll break the door if you close it. You got to let the door close itself.

[B]

It's that.

[A]

Yeah, the drivers who are all wearing white gloves, I might add. Maybe I just went in some fancy tapsis, I don't know. But they wear white gloves. They're very polite. You can't tip them. And they, I think they have little signs, but they're probably in Japanese. But they say don't close the door, but they don't yell.

[B]

All kinds of rules. There, there's so many rules.

[A]

Let the car close the door. In this case, those eye paces that they used, they never thought of that. They don't have door closers. But. But Waymo says in future they will build cars with door closers. So this is.

[E]

Doesn't sound like a big engineering problem, you know, it doesn't sound like a big engineering problem.

[A]

Just put a rubber band, a spring on it, something, you know.

[B]

So, Leo, there are more notetakers for you.

[A]

Well, yeah, I saw the article. The B computer lady is writing an article about how at Amazon, oh, we're having so much fun. Now I feel sad. I gave my hairdresser my baby. She couldn't get it working. In fact, a lot of people with bees, those little things that I was wearing that recorded every once they were purchased by Amazon, I gave it up. But a lot of people who have them said, what, is it still working? Mine stopped working almost immediately. So I don't know. She was never able to get it working with Android. Yeah.

[B]

So Claude has a new wearable.

[A]

Oh, Claude does?

[B]

Claude.

[A]

What Claude? I have applaud.

[B]

Yeah, it's a new one.

[C]

I would have been excited if it was Claude.

[B]

And then there's Another one called. Where is it? Here. Memory.

[A]

I learned we had on Twitter on Sunday Joey de Villa, who is an AI developer advocate and from Canada. And he said. No, no, it is pronounced clud. So I'm saying clud. From now on, you have Joey back.

[B]

That's correct.

[A]

Oh, yeah. You know, we had Joey in the. The show. Of course. Yeah.

[B]

But I was. I was gone that week. I got him on and then I was gone.

[A]

Oh, thank you. Yeah, he's going to be a regular. He played the accordion for us. He has a new song. Can I.

[B]

Should I. Oh, do, do.

[A]

Should I play it for you?

[B]

He won't take us down, will he?

[A]

Joey better not take us down. He is the. Besides being an AI developer advocate, he's worked with Cory Doctorow. Yeah, he's. He was.

[B]

He goes back to the early blogging days.

[A]

He goes back to the early days. It's pretty cool. And he wrote a song which he played on the accordion for us.

[B]

I'll.

[A]

I'll just. I think I can play a little bit of. Let me see if I can find this on his website. Or. No, as YouTube.

[C]

If it's on YouTube. That might be tricky.

[A]

Oh, you know, he took it down.

[E]

Maybe Afroman took him down.

[A]

Afroman must have taken him down.

[B]

Oh, no.

[A]

I hope we didn't get him in trouble because we played it on our show. Now it's gone. Oh, never mind. I'm sorry, Joey. I'm sorry. So, okay, Plaud's got one. But that's really another updated version of credit card.

[B]

Then there's better microphones. CNET has a story about an old. You're not going to use a note taker, you're just going to use your phone.

[A]

Right? Because it's there. It's got very good microphones on it.

[B]

And then there was Another story from ZDNet about the most exciting AI wearable CES might not be smart glasses after all. Memories AI's wearable pin records for longer and does things.

[C]

Doesn't this look like the thing that got shelved?

[A]

Johnny Ives, Humane pin. Yeah.

[C]

Yes, this looks like the thing that got shelved after like two weeks.

[A]

It does look a lot like the Humane pin.

[B]

So they're still trying to. Is my point.

[A]

So what is so. So it's shiny Memories.

[C]

AI is launching Project Lucy at ce.

[A]

Oh, I Love Lucy builds on the.

[C]

Prior Lucy pin, which no one ever heard of. Quick processing.

[A]

At its core, it stands for long understanding, contextual intelligence.

[B]

Lucy SwitchBot has an AI recorder that's a second brain for memory. So my point is there's four of these. There's.

[A]

I'm soured on this. Well, that's where I found out about the B was from last year's ces. Yeah, there were a bunch of them announced at ces. Fieldy has. I think it's Fieldy that is now was got acquired. Right. I mean that's the thing is a lot of these companies I think were just started to get acquired. What happened? If this is the Fieldy, you wear this thing and I still have this. I think it's pretty cool. I feel like they got.

[B]

It looks like it's compressing the person's chest.

[A]

It does, doesn't it?

[B]

Jesus. Let him go.

[C]

I believe it's a video pin. So it's always recording video?

[A]

No. Oh, the human. The new one from the.

[C]

Lucy.

[A]

Lucy. Oh, that's. People don't like that.

[C]

It says failed AI wearables typically capture data without understanding it. Lucy does a better job of understanding your life the way humans do visually, contextually and continuously with memory. With the 2.0. The company aims to address this issue by converting the continuous video captured by the PIN into structured on device encoding frames which can be used to index and reference them in what the blog post refers to as a quote, sub, second search and recall. Three hours of continuous recording, which isn't that.

[A]

Three hours, that's not much at all.

[C]

109 degree field of view, a privacy switch in the back and a status light. Yeah.

[A]

You know, I'm. I tried all of them and, and I, and I, I like the idea, but the result wasn't fantastic. My main use of AI right now is just clud code which I. Which I leave running. I leave it running all the time on my laptop and it's fantastic. I just. You can ask it pretty much anything, but mostly I use it to do geeky stuff like configure. Like I. Can you make the menu font two points bigger and it goes. Yeah, sure, we. What else would you like?

[C]

On what?

[A]

On my laptop.

[B]

Okay, just do that.

[A]

Yeah. Oh my God.

[C]

So wait, but can't you just do that?

[A]

Yeah, I could. You could go into a configuration file and edit it and. Actually that's pretty simple and I could probably have done that easily, but Claude will do it for you.

[C]

Sorry, who.

[A]

Clone code? So I got a new laptop, a Lenovo ThinkPad X1. I put Linux on it. I've told this story before. We actually talked about it on the AI User Group on Friday, but I haven't told you guys. And I decided. Because I decided I don't want to use the proprietary operating systems. Macintosh and Microsoft are very. Both very opinionated about how you use a computer, right, Just as your phone is. And they're not very flexible. Same thing with your Chromebook. As in terms of, you know, well, I don't want to. I don't. I want the menu at the bottom or I don't want a menu or whatever. I want. I want these apps to launch in when I start up, and I want them to be in different workspace. All of this stuff you don't really get to decide. There are some limited switches you can flip, but for the most part, Microsoft or Apple or Google decide what your experience is. That's the. So I decided I want a laptop where I get to do it however I want. So I got Lenovo and I put Linux on it. It's beautiful. I love it. And I picked a desktop environment that is very different from all of these that I really like. But it takes a lot of fiddling. There are a lot of text files that you have to modify. And I thought, I wonder if Claude could do this? So I. Claude. So Claude could keep saying Claude. People are going, what the heck is he talking about?

[C]

You have to commit, at least for this one episode.

[A]

Okay, but there's one show.

[B]

Bonjour, Claude.

[C]

Hello, Claude can do it.

[A]

So I launch Claude and I say, hey, Claude, I'm using this Sway desktop environment. What's a good menu bar program? And Claude says, well, I checked, and everybody seems to be using waybar. Would you like me to set that up? And I said, okay. And then I said, oh, I would like to know what temperature my CPU is. Claude, can you put that up on the menu bar? Of course. And so basically, bit by bit, over a period of a few days, I completely customized it.

[B]

Okay, all right.

[A]

Instead of going. And I could have done it all, going out and looking up the documents and doing an editor and all that stuff. I've been doing that now with my Emacs editor, completely with my desktop computer. I said, eclode, I would like to talk to you. Instead of tapping, is there a way I can dictate to you? And Claude said, oh, yes, we can install Owe Oui. We can install the Open Whisper and have you just press a button and I will talk to you. Actually, it doesn't talk to me, but I.

[B]

Because it would you talk to it.

[A]

I talk to it.

[B]

It.

[A]

And so. And so I press a button, it goes beep. And I can Talk and then press it when I'm done and it sends it off. And it did it all. I could have done that. Absolutely. I could have gone to the various wikis and downloaded, figured it all out, downloaded software. But Claude just did it for me.

[B]

I got a new use for AI when we were gone.

[A]

One other thing I can say to Claude, Hey, I don't like that. Claude, take it back. I take it back and it undoes it. It, it remembers everything it's done.

[B]

Oh, it can undo Claude or just because it knows.

[A]

No, it's because it remembers the conversation. I said, every time you do something for me, put it in Claude MD. And then I also commit it to GitHub. In fact, if you want to see my configuration, it's public on GitHub. I'm Leo Laporte.

[B]

I'll be doing that tonight for sure.

[A]

Oh, I guess you will.

[B]

Fascinating.

[A]

And so it can, it can download the commits, it can change the commitment, it can check. So it. No. Yeah, it can undo anything. So I say, no, I don't like that. Can you do something else? And it just does it. Go ahead. Now. How are you using it, Jeffrey?

[B]

So I'm not going to tell you the subject because I want to keep this just to me for reasons, but I have a story that I suddenly realized one day that other people have talked about doing a screenplay. And I thought, well, no, I think that I could do that. And I had no idea how to write a treatment. So I wrote up four or five pages and I fed it into NotebookLM and then Gemini and I think ChatGPT and I said, can you make this into a treatment? And it did and it wasn't very good. But the format, I had no idea of. No idea whatsoever. So I learned the format versus going to read somebody's website about how to write a treatment. It used my material. So now I can say, well, that's. I'm going to rewrite this and I'm going to do. I'm sorry, bad word. And so it was really interesting as a different way to learn something because it used material that I knew and cared about. Right.

[C]

That is very interesting.

[B]

Yeah. And so it gave me. My agent doesn't know how to deal with this because she deals with like academic books and things. If anybody out there knows a good agent who deals with movie stuff, let me know. No, seriously. But it's like a 10 page treatment. Act one, act two, act three, you know, and the character arcs and all these things I didn't know.

[A]

And there's a lot of formatting. Right.

[B]

Well, it's not a script, so it doesn't have the, you know.

[A]

But it would. If you said can, by the way, I've been doing that too in my emacs. I'm saying can you reformat that for a different. And it does it.

[B]

Yeah, it would. But this was amazing. This was. I didn't know there were. There were elements I didn't know I needed. There was a flow I didn't know I needed. Then I also had it, you know, critique it. And there's the problem, there's the sycophancy. Oh, this is really fillable. This is what I wouldn't do. Should make this right. No, no, no, no. I think this also.

[A]

You're getting smart to know what it's good at and not good at.

[E]

You're getting. If you're asking for creative advice, you're getting a middle of the road advice, basically, very middle. You're not getting great advice.

[B]

But I learned the format, I learned something new which was really useful.

[A]

But there are some things where you do want kind of middle of the road. Configuring a computer, for instance, is a good example.

[B]

Don't go crazy.

[A]

What is the most popular way to do this? Or what is the one that you see the most of? And so in those cases, mediocrity is exactly what you want. It's not mediocrity, it's average. Is exactly what you want. So. Yeah, but not in creative. I am less interested in all of this stuff for creative purposes. I know a lot of people aren't.

[B]

But I Scooter X just put in something that maybe useful for your. Your job here.

[A]

Leo, I.

[B]

Did you see this? The Verge story?

[A]

Meta is adding a teleprompter concept.

[B]

Teleprompter to your glasses.

[C]

To glasses.

[A]

No, I don't. Hello, my name is a. Leo Claude Lavard.

[C]

Hello.

[B]

Hello.

[A]

Okay, that's interesting. A teleprompter.

[B]

I see Tony Dicopol will be using it on CBS News before, you know.

[A]

Yeah, I love that.

[C]

The teleprompter display image says speak naturally, stay on script and never lose eye contact. That's the power of the teleprompter. You know, someone repeating that to themselves, that is.

[A]

Oh, and by the way, I'm sure they taught you in this, in your media training. The constant eye contact is not a good thing, is it?

[B]

No, no.

[A]

Nobody does that in real life.

[B]

Let's give you a gift here. We're all going to do constant eye contact.

[A]

Constant can we all.

[C]

Let's all look at different direction.

[E]

Wonderful. Thank you.

[A]

Now we just look. I'm sorry. Supposed to use. That was not. That's a bad word.

[B]

That was bad editing.

[A]

That was worse than the other word that you used, Jeff. Sorry.

[B]

We're old. These things are still in our heads.

[A]

They're still in our heads.

[B]

It's bad.

[C]

I disavow that.

[A]

What? What should I have said? Mentally challenged. Is that okay?

[C]

We look dumb?

[B]

Dumb.

[A]

As far as silly, dumb is good. Dumb is a good word. And I didn't mean dumb, but I will use them.

[B]

Hmm.

[A]

I am so close to buying those new Ray Bans. This is not gonna put me over the top. I'm about to see. I want to see.

[D]

See.

[A]

I don't like. It's just a one eye. That's kind of weird.

[C]

Oh, it's just one eye.

[A]

Yeah.

[C]

That's like Jeff in a couple weeks.

[B]

The one eyed monster.

[A]

Hey, everybody. Yeah, Burke says I always make eye contact and keep it. And Burke, that's why we think you're weird.

[B]

Mm.

[A]

All right, let's take a little break. We are back. Aren't you glad? Intelligent machines. Should we do Carpathy's year in review? I don't think it's that interesting. Carpathy did tweet an interesting thing. He tweeted a very interesting thing on Christmas Eve or Christmas Day. He said he's the guy we mentioned before, the creator of the term vibe coding. He used to work at Tesla AI. He's an AI guru. And I've recommended his videos as a great way to understanding what's going on with AI if you've got three and a half hours to kill. And he. He posted on X over the holidays. I'm overwhelmed. I can't keep up. It's happening too fast. All of this coding stuff is going too fast for me. Yeah, Andre, if it's going too fast for you, we're in deep trouble. But we kind of encouraged me. For us, it kind of encouraged me. It's like, yeah, well, I feel that way. I've always felt that way.

[E]

Yeah, but is there anybody who feels fully, like, capable and caught up?

[B]

No.

[E]

Right?

[B]

No. Everybody, if they do, they have hubris beyond. That's the. That's the AGI. They think they're smarter than them.

[A]

It's funny because when I first started doing computer stuff, I felt that way. I felt, this stuff is moving too fast. I'll never be able to keep up. It's overwhelming. And I just said, you know what? I'm going to. In fact, I even came up with a whole philosophy of a way of thinking about it. Instead of thinking of it as a cup that I'm going to drink of and try to drink all of it, I'm going to think of it as a flowing river that I'm going to dip into. I know I can't capture it all, but I'm going to dip into it and use it as it flows by and use what I like. And I know there'll be stuff that's going by and that's what the way it is with news now, with information for sure and certainly with technology. Here's a When we come back, this was actually an interesting piece from Axios 2025's AI fueled science breakthroughs, things we actually this is part of my continuing effort to convince you to good priority that AI has some value. Gosh darn it.

[C]

Well, it's behind a paywall, so I'll never know. I won't know what it says. So we come back from the break.

[B]

Oh, I think she might grab one free article for a millennium.

[A]

Is Axios behind a paywall now?

[B]

Axios, yeah.

[C]

It says I've got to subscribe to Axios AI plus to continue reading for free.

[B]

Oh dear. Oh, that's not good.

[A]

They've changed. They used to be free.

[E]

Do you subscribe?

[B]

That's not a paywall.

[E]

You just have to subscribe. It just says to continue reading for free, you have to subscribe.

[A]

I hate that. Yeah, that's a paywall if you ask me.

[E]

I guess so.

[B]

Yeah.

[C]

Good. Data wall.

[A]

It's a something. This episode of Intelligent Machines with Paris Martineau and Jeff Jarvis. Great to have you. Great to see my old friends. Great to see all of you too. Is brought to you by Delete Me. Now. This, this is something you need. If you've ever wondered how much of your personal data is out there on the Internet for anyone to see, I do not recommend doing the search. But if you look for your name, you will find that. You will find contact info. You may find Social Security numbers. I actually looked for my Social Security number. Steve Gibson and I did this on Secure Now. Both of us found our Social Security number. Unbelievable. Home address, even information about your family members. And where is this coming from? Data brokers. Data brokers. And the reason they do it because they can make a lot of money. They compile information about all of us and sell it online to anyone who comes along. Anyone on the web can buy all of these private details. It's stunning to Know that the United States of America is not illegal to sell somebody's Social Security number. How could that not be illegal? It's not. And what can it lead to? Anything. Identity theft, phishing, doxing, harassment. You can protect your privacy with Deleteme. I am very aware of how little privacy we have. And I mean this is, it's stunning how much personal information is out there and the risks are obvious to me. We've even encountered phishing attempts on our company. Encountered them. We encounter them every day. That's why I personally recommend and use Delete Me. Maybe it's not going to get rid of all the phishing, but it's definitely going to get that private information out of the Internet, which really makes a difference in protecting you. Deleteme is you don't want the government marketers, foreign governments, hackers to buy your personal data. That's crazy. Delete Me is a subscription service that removes that personal info from hundreds of data brokers. The number one source of it. These are the ones collecting it and selling it. Sign up. Provide Delete Me with just the information you want deleted. It's nice. You have complete control over that. Their experts will take it from there. You'll get regular personalized privacy reports that show you what info they found and where they found it and what they were able to remove. Deleteme is not just a one time service. This is always working for you. Constantly monitoring and removing the personal information you don't want on the Internet. To put it simply, Delete Me does all the hard work of wiping you and your and your family's personal data from data broker websites. Take control of your data. Keep your private life private. Sign up for Delete Me at a special discount for our listeners. 20% off your delete me plan when you go to JoinDeleteMe.com Twitter and use the promo code Twitter checkout. The only way to get 20% off is to visit JoinDeleteMe.com TWIT Enter TWIT at checkout again. JoinDeleteMe.com TWIT the offer code is Twitter. It really works. We've been using it for a couple of years now. We get those reports. It's mind blowing. And the thing is, these guys, they don't take no for an answer. They'll delete it. But then they go, oh, we found some more and they'll start rebuilding your dossier all over again. There's only one way to get rid of it. Join EliteMe.com TWiT and the offer code is TWiT I should mention, by the way, and I haven't tried it yet. If you live in California, the state of California is about to or is implementing a way for you to delete your information from data brokers, which I think is pretty good. I don't know why we don't have a federal protection here.

[B]

Well, that's what the world wants, a federal privacy law. So we have some consistency. Data brokers have been there forever. I mean I've talked about this on the show before. I used to scare students by showing them what's an axiom. Long before, well, not before the Internet, but you know, many years ago.

[A]

Yeah, they have a list of. Well, this is funny. There's a data broker registry on the state of California, but as of January of 2024, they stopped doing it.

[B]

Why?

[A]

See, I think there must be a big lobby somehow of data brokers. So if it's two years out of date, it's worthless because these guys change their names. I mean anyway, California does have a tool. We'll see how long it lasts to. It's called drop, delete, request and opt out platform. I haven't tried it yet. It's only available to people in the state of California and it only goes into effect August 1st and then brokers have 90 days. So I don't think it's maybe as effective. I think there's a lobby and I think one of the lobbies is not data brokers is law enforcement and other people like marketers who love it that they can get this information that they don't want it to be shut down. There's gotta be right. Otherwise it'd be gone. And yes, federal law does allow you to go to the data brokers if you can find them and have your data removed. But you, there's more than 500 new ones all the time. So good luck with that. All right, here I am making the case for AI you can throw tomatoes at me. Paris. It's okay. 2025's AI fueled scientific breakthroughs making diagnoses for Alzheimer's. Oh is on the road to becoming faster and cheaper with AI. Researchers at a wide range of universities and healthcare institutions announced findings this year about how AI will help with future therapies. They found for instance, a specific gene that's a cause for Alzheimer. Alzheimer's. And they were only able to make that gene because AI helped them visualize the three dimensional structure of the protein. That is one thing AI been very good at is protein folding. Right. In fact, that's Google's Alpha it fold.

[C]

Laundry, but it can't fold.

[A]

It can't fold laundry, but it can fold your jeans. G E N E S It can't.

[C]

Fold your jeans, but it can fold your jeans.

[A]

Google released Alpha Genome which is a model to understand diseases better and lead to drug discovery. That's huge. Yeah. It will be again. A lot of this is someday I.

[C]

Was about to say. I like that all of the biggest AI things that happen in 2025 for science are things that will happen.

[A]

They're gonna happen.

[B]

Humanoid robots, discovery of the electron to radio.

[A]

It takes a while.

[B]

Yeah.

[A]

It takes a while. Yeah. I'm surprised they put human robots in there. They say one day they will clean homes.

[C]

Not today.

[A]

Not today. Weather forecasting is better. If Google released a weather forecasting model that can generate forecasts eight times faster than before. That's something AI is very good at. How about this? A machine learning framework helped a team analyze scientific literature more than 1 million rock samples to narrow down viable alternatives for cement cost and emission efficient cement.

[B]

MIT's because it's a terrible polluter.

[A]

It is a polluter. That's right. We mentioned AlphaFold which is doing 3D protein folding. That started five years ago. Six Nobel prizes for Google 3 awarded just in the last two years around that. You know this isn't as satisfying article in defending.

[B]

No, it's Axios.

[A]

I was really excited. Never mind.

[B]

Forget I brought water.

[C]

Someday all those AT and TS hit you square on.

[B]

I think we gotta have just these projects.

[E]

Just get all the money that OpenAI gets. We'd be better off, you know.

[B]

Amen.

[A]

Yes, I agree. I agree. But that's not how it works in this.

[E]

And that's the problem though. It's never really been about AI. It's been about the people who make the AI.

[A]

Right.

[C]

It's been about resource allocation.

[B]

But somebody in the chat. Your opportunity now is to make a game of hit Leo with the tomato and Leo dodges the tomato.

[A]

I'll give you the sound.

[C]

We're doing free Foley work out here guys. You gotta put in.

[A]

You got the sounds.

[C]

Now make the game Claude code on this.

[A]

Claude could do it.

[E]

It can probably be vibe coded before the end of the show. Right?

[A]

I. I know it could get on it.

[C]

Guys.

[A]

Happy Jeff.

[C]

Remember I've accidentally cannibalized my picks the week bring this up so it could be my pick.

[A]

Oh, she needs a new pick. So write the code.

[B]

What was. Where did Flappy Jeff come from? Why did we do floppy Jeff?

[E]

It was a vibe coding Exercise.

[C]

The only thing I've ever vibe coded and it was. I think I typed two sentences into chat GPT while on the air. And we.

[A]

Oh, I gotta go. The ice cream truck.

[C]

Somebody made a better version.

[A]

Can you hear the ice cream truck driving by?

[C]

No go.

[A]

It's driving by.

[C]

It still not have walls.

[A]

We live on a dead end. And there's an ice cream truck and it's the middle of winter and it's freezing cold and there's an ice cream truck freezing.

[C]

You should go support a local business, Leah.

[B]

We'll keep talking for you if somebody.

[C]

Wants to come on the show.

[A]

Adio, Drummet. What was that? What was that really good one that they discontinued?

[C]

Creepy Powerpuff Girl. Ones that get melted and they look really haunting.

[E]

You're the one with the gumballs for ice.

[C]

Yeah.

[A]

Ooh. Aunt Pruitt in our. In our discord says it's never too cold for ice cream. Bless your heart.

[C]

Rule of life ice cream place near me is now open your around that. I'm like old choice. Yeah.

[B]

I don't.

[A]

Wow. Yeah, it's. It's so cold. I don't know if I really want ice cream.

[B]

By the way, since you're. I love watching Ant on his shoots. Getting his makeup and.

[C]

And extraordinaire.

[B]

He is.

[A]

He's getting a lot of work.

[B]

He is as well. He's a handsome devil.

[A]

Yeah. Where do you find that? Is that on his Instagram?

[B]

That is. I think you're LinkedIn, right, Ant?

[C]

LinkedIn. Yeah.

[A]

Nice.

[B]

Yeah. Because it's to get. To get to get jobs.

[A]

And you'll see some of the headshots on Instagram.

[C]

Yeah.

[A]

Yeah.

[C]

A lot of good stuff.

[B]

Need a model on set. I'm 6 foot 2 tall, 225 pounds, 36, 32 trousers, extra large shirt, bearded or clean shaven. Your choice. Comfortable and easy to work with on set. Which we know and we can attest to.

[A]

I love that if you watch the. The audition videos where the guys will say, I will shave for the job or I won't shave. Some of them say, not gonna shave.

[E]

So he can also take the photos.

[B]

That's the thing.

[A]

Yeah. Posing them.

[B]

He did it. He did find it. He did photos where they made him look young and then they made him look old.

[A]

Nice. I love Ant.

[B]

Ant's the greatest.

[A]

Ant p r u-I t t.com hire ant. A Google engineer says Claude code. Which is interesting.

[B]

Google engineer said what?

[C]

They said what?

[A]

Claude code. A Google engineer. Claude code built in one hour. What her team spent a year on. What's funny is Anthropic left Google because they didn't like how Google was doing. AI and their competitors, Anthropic left OpenAI, not Google. Oh, I'm sorry. OpenAI. That's right. But they're still competitors. Google principal engineer Jana Dogan reports that Anthropic's cloud code generated a distributed agent orchestration system in one hour. A problem Google had been working on since last year. I mean, 2024. Well, the result isn't perfect, okay. Dogen says it's comparable to what Google built previously, which also wasn't perfect. Highlighting the rapid advancement AI assisted coding capabilities, Claude code creator Boris Charny suggests enabling the tool to self check its work. A feedback loop that can double or triple the output quality. I have noticed as I use Claude that it's an interactive process. You don't. I think the whole idea of a one shot process is misleading. Yeah, sometimes they can one shot stuff, but really you should do it in more little planning chunks, a little bit at a time, back and forth, back and forth, check it. I've been very impressed with what you can do to cut through the noise. Dogan clarifies Google has built several versions of the system over the past year. 2025. There are trade offs and there has been a clear winner when prompted with the best ideas. Ah, this is how they do it. When prompted using the code base of the best ideas that survived, coding agents can generate a decent toy version in about an hour. What I built this weekend, she says, isn't production grade, it's a toy version, but a useful starting point. I think that's a fair description of it.

[B]

So I've asked this question before, but I'll come back to it. When we had Mike Masnick on about how he did his vibe coding. I think what blocks people, frankly like me, is I'm not used to putting things on servers and running them.

[A]

Well, you don't have to do that. I haven't put any of the stuff that I do on a server. It's all local.

[B]

It's local to what though?

[A]

To my computer. Oh, okay.

[B]

So how do you get through a run version of what you've had?

[A]

You can run it locally. Make you run it locally. He's only running because he wants a server version. So Claude is running in the cloud? Yes, Claude's cloud. Claude's cloud's in the cloud. Cloud in the cloud.

[C]

Claude code in the cloud.

[A]

But, but it, but what you do is you go into your project directory and if you've started a new. If you're starting a new. Probably it's hard if you're not a coder.

[B]

That's what I'm saying.

[A]

You're just learning Python and you go into your directory where that code is going to be and you, you open Claude and. And you say slash init. It reads all your code, reads your project. Then you say, here's what I'm thinking, let's plan a pro. You can do it all verbally, right?

[B]

Right. So, but here's what I'm saying, Leo. I think there's a huge consumer demand that will come up when somebody comes up with the really easy way to say, tell Claude what the program you want to do, and then it just runs and it does what I wanted to do. I want you.

[A]

That's what you do. I can do that right now. That's how I did it.

[C]

Live coding is.

[A]

Yes, but. But it doesn't have to be on a server. That's just.

[B]

Never mind that. I'm still saying people aren't going to do ads and they're not going to do that kind of stuff.

[A]

Well, you don't have to know. You can hear. Watch. Let's do it.

[B]

All right. That's what I'm trying to get to, is to get to the really easy consumer level of this.

[A]

Yeah, you can do that.

[E]

The problem with that, Jeff, is that, like, you need to be a good producer. You need to know exactly what you want, and most people don't know exactly what you want. You know, that's, that's the problem most people have, is they don't really know what they want.

[A]

What are we going to make? I forgot. Oh, tomato game.

[C]

Yeah.

[B]

Oh, good. Okay.

[C]

Game where we're throwing tomatoes at you.

[A]

Okay, Now I'm going to do Claude. Claude. Hello, Claude. Hello, Claude, talk to me. Claude, what language. You're going to do this in Python. Right?

[B]

See, right there, you've lost consumers.

[A]

No, no, I don't have to do any Python.

[C]

And it's user, Leo. Tomato. It's like.

[A]

Yes, it is. Slash, tomato.

[C]

That's where we are in Cloud. I want to create a game.

[A]

Oh, I can. I don't have the dictation on this computer, so I can't dictate it. I have to type it. I should have put that on this computer. It would have been much more useful.

[C]

That would have been good.

[A]

I'm ready. I'm happy to help you create a game in Python. What kind of game do you have in mind? I want to Do a GUI game. GUI game?

[C]

Is that a type of game or you refer.

[A]

It's a graphic user interface.

[C]

Okay. I was like, are you a referring to what happened?

[A]

Not G, O, O, E, Y, G U, Y with flying tomatoes. And I want to aim the tomatoes and score points when it hits a picture of me. I'm not. See, this is something completely. You could totally do, Jeff. Right now.

[B]

I'm. Okay, now. Once. Once it makes it.

[A]

It stays this way. It stays this way. Well, I mean, I'm not gonna. We don't probably so. That's.

[B]

That's my point.

[A]

It's almost dinner.

[B]

I want to share this. I want to share this with.

[A]

Yeah, you'll get a game. You could package it up. You could say, package that up so I can send it in an email. Okay.

[C]

GUI game through Gmail.

[B]

How do I make it on the web? How do I put it on the web so people.

[A]

You can say, okay, well, that's a little more complicated.

[B]

That's what I'm trying to say. I think there's a host.

[A]

Any web game you have to have.

[B]

That's what I was trying to say. The server. Yes.

[A]

But now you just made it a web game.

[C]

Put this on the web.

[A]

Yeah, you could you simply say, okay, I need a web server. Can you set me up a web server? And I'll say, okay, where do you want to go? Where do you want to host it? And you say, I don't know, what do you recommend? And it's going to say, hetzner. Okay, can you get me an account at Headster? Yeah. Now you're gonna have to some point you do have to intervene because it doesn't have your money, it doesn't have your passwords, things like that. So you might. It might walk through.

[B]

All I'm trying to say is I think that there is a big business opportunity to say whatever program you want. You tell Claude and then the next thing you know, it's available to you. And you didn't have to know any of this stuff.

[A]

Well, that's what I'm doing. So it's a great idea. A tomato throwing game. Sounds fun. Before I start building it, I have a few questions about the game mechanics. How should the player aim and shoot the tomatoes? Do you want a keyboard arrows or you want to click to shoot? Do you want drag and release slingshots?

[B]

What's best?

[A]

Well, actually, I want Paris to decide. Drag and release slingshot. Right. That's how you would like.

[C]

Yeah, that's what we want.

[A]

Okay. Should your Picture the target move around or stay in one place. I think definitely it should move.

[B]

Yeah.

[A]

Do you have a picture ready to use as a target? Yeah.

[C]

Don't you. You have a photo of yourself?

[A]

Yeah. Okay. Submit answers. Okay, so now it's reticulating. Claude is funny. It has a lot of different verbs it uses for as it's working. Reticulating.

[B]

Perfect.

[A]

A slingshot style tomato game with a moving target. That'll be fun. Let me build this for you. It is. Now writing the game.

[D]

I love all this.

[C]

You didn't give it the photo.

[A]

Well, it's gonna ask at some point. It prompts you. It doesn't.

[B]

Ah.

[A]

So it's gonna say. Okay, I need that photo. Now let me. Let me. Let me get a photo from Pretty Fly for assist.

[C]

Sky has come up with a really good game box image for this game. We maybe.

[A]

Oh, yeah, I like this.

[C]

Ahead of ourselves a little bit.

[A]

But let's see. Copy image. Okay, now I've got that image. It's still thinking because this is complicated. It's writing a whole game in Python right now.

[C]

This is beautiful.

[A]

D. Cool. Oh, wait a minute.

[C]

Great use of everything.

[A]

Now here's the. By the way, here's the code. You want to see it, Jeff?

[B]

Sure.

[A]

Because do you know any python?

[B]

No. I don't know.

[A]

Do you care? No. Look it. Here's all the code I like where.

[C]

It says import math.

[A]

Yeah, Import math.

[B]

Good.

[A]

If I could do that. My brain, man, if I could do that.

[C]

That was me at 10am on Monday when I had the flu and was trying to do basic.

[A]

You would have loved this.

[C]

I would have loved to have imported math.

[A]

Yeah. I mean, I think this is really cool. Now this isn't the only way you could do this. Google's got, of course, their new. What is it called? Anti. Antipodes. I can't remember what it's called. Anti Gravity. There's. Okay. Codex will do this for chat GPT. Okay. The tomato will reset automatically after each shot. First install PI game. Run the game, and then you have to give it the path to your picture file. Okay, good. So it's done. Excellent. Should we play the game? All right, let's see how you can.

[B]

Get Paris to play.

[A]

Oh, I don't have pip. Let me see. Do I have uv? No.

[C]

I wish I knew what any of these words meant. But I really enjoy also that I don't.

[B]

Yeah.

[A]

Pip is a python of Pip Cheerio. Like Pip Pip Cheerio. So I have to install Python. I thought I would have had it on here. But watch, though. This all happens very fast. Okay, but as long as you can. Oh, I have Python. Okay.

[B]

I would think.

[A]

Oh, it's probably not in the past. It's pip3. Okay. That's why pip.

[C]

And then the third needs to be a cheerio, obviously.

[A]

Wait a minute.

[C]

Three pips? That would be crazy.

[A]

Now I have to read some instructions.

[B]

Oh, that's the part we hate.

[A]

Yeah.

[C]

Come back here. She keeps rubbing up against my legs and then running away when I try to hold her up to the camera, she's being shot.

[B]

Come here, Giz. We haven't seen you for three weeks. Giz, come here.

[D]

Got her there.

[B]

She's not going to spin around.

[C]

You're not gonna do it. You're so strong, Gizmo. I'm really trying to get her to face the camera.

[A]

This cat interlude provided for you by Clud. Good. This is too complicated. I have to install some Python stuff. But I will finish this after the show and by next week, we'll have Leo Laporte Tomato blast the.

[C]

Hey, listen, we've still technically got a chance that Darren OKE is making this exact same game.

[A]

It'll probably be better, but there is a game and it's ready to go. I just. I don't have the. The programming.

[C]

Schrodinger's game.

[B]

Let me ask the question this way. You've made the game. How does Paris play it?

[A]

I would have to send it to her. Or I could take a baggie on her, put it on a floppy disk. Put the floppy disk in the baggie with some immigraphed instructions, seal the baggie, put it in an envelope, mail it to her, and she could do it that way.

[B]

If you wanted her to play with right now, I would just.

[A]

I would email it to her. It's a file.

[B]

It's a file. She had to have installed Python.

[A]

No, once I. Once I get this all set up, she won't have to install Python.

[B]

All right.

[C]

Fascinating. I hope that we all get an.

[A]

Email with this game before this night is out. You will have. I will write it better. I won't do it in Python. I'll write it better.

[C]

I was going to say, if we're. If we're requesting features, I think the tomatoes should explode when it hits the image view.

[A]

But that might well, See, So that's what happens. So I could play the game now, and then I could go back to Claude and say, you know, I would like the tomatoes to explat much more aggressively on Leo's face. It's pretty impressive. It really is. And the thing is that's this is a more elaborate kind of thing. Then I do mostly little bits and pieces at a time, like, well, please install a way for me to transcribe stuff.

[C]

Or does Claud know what a tomato looks like? Does it have image generation capabilities?

[A]

That's interesting. Yes. You could use an MCP server to connect Claud to. Well, you could actually connect it to Nano Banana and say, but so I.

[C]

Wonder what tomato you are throwing.

[A]

I don't know where it got the tomato. That's a good question.

[C]

I don't know if it is a tomato.

[B]

Would it know enough to ask and say, I don't know what a tomato is. You have to tell me.

[A]

It knows what a tomato is.

[C]

It knows what a tomato is. But whether it cannot. I don't know whether it can generate. It's gonna a tomato.

[A]

Well, that's a good question. I should look and see if it might be a computer graphics tomato. That'd be interesting.

[C]

That would be pretty fun.

[A]

It would have a reference. It can go and say, okay, tomato don't know what a tomato is. Let's see what a tomato is. Get a reference file and says, oh, that's what it looks like. And say, well, then I put dots here like this. I mean, it can do all of that. Yeah.

[C]

I'm just imagining an ASCII art tomato being thrown at you.

[A]

It's pretty uncanny what it can do. And I encourage people to try it. I really do. All right, I think.

[B]

I think we've got a debate whether Satya Nutella is right or wrong.

[A]

What does Satya say?

[B]

Line 130.

[A]

All right, let's go to line 130 right now. Oh, Satya says, stop calling it AI slop. Right. What does he say?

[B]

Rebrand it.

[A]

We need to get beyond the arguments of slop versus sophistication. You know, a lot.

[B]

Because. Because that's not like a high culture and low culture thing.

[A]

Yeah, It's. There is a slope. No one would deny that.

[B]

In fact, the more we. I think it's the opposite. The more we define it and have norms against it, the higher the quality will be because people will be motivated at some point not to do it anymore.

[A]

Because I think we need to. I think we need a new model of what AI is. It's not. We stop. Stop anthropomorphizing it, but also stop expecting of it superhuman capabilities. It is a tool. It's like a computer can compute. A computer can generate garbage and can generate great stuff. When we first got laser printers and fonts, people made documents that. Page layouts that looked horrible because they looked like ransom notes because they wanted to play with all the fonts. Is that slop? Yes. But does it mean that the computer, the laser writer and page definition, you.

[C]

You're conflating slop with a thing that looks bad.

[A]

Oh, you say if a human does it, it ain't slop.

[C]

Correct. It's just bad. Slop specifically refers to it being machine generated, so.

[A]

Well, then I agree with Satya. I think that that is just because a machine generates slop.

[B]

I would. I would agree with that. No, it's not. Yeah, no, no, no.

[E]

It's a Venn diagram.

[C]

It's a specific. It's a subset of generation. There's a subset of AI generated images, text that is slop because it is bad or nonsensical in a way that is inherently slop.

[E]

Like, not all can also refer. Not all AI stuff is slop, but all slop is made by AI.

[C]

Correct.

[A]

Unless you're feeding it to pigs, which is where the word came from in the first place.

[C]

Yes, Grandpa.

[A]

I mean, that's what slop was originally. It was the food scraps that were slopped together and fed to pigs. Right. So that's what I think of when I think of slop. It's a pejorative. It's like this thing is.

[C]

It is a pejorative. That's correct. But it is, as Benito said, it's kind of a rectangle, square situation.

[A]

So from the conversation, we're talking about AI all wrong. So this is Pablo Sanguinetti, who's a professor of AI in Italy.

[B]

He has a book that is in Italian called Techno Humanism, A Narrative and Aesthetic Design for Artificial Intelligence. So he argues toward the end that we should avoid using AI as the subject of a set sentence. As in AI does this, or AI that's good. I like it when it is being used as a tool and thus not anthropomorphizing. Playing with the term AI also helps us see how much words can change our perception of technology. Try replacing it in a sentence with, for example, complex task processing. Hard to say one of the least ambitious but most accurate names considered during early days. So rename it and rebrand it.

[A]

Or expert Systems. That's what AIs were originally.

[B]

So I mentioned when we were trying to find some. I'm really glad we got who we got for the show today, but one of the people I mentioned is a guy named Thomas Haig, who's a historian of technology, he wrote a book on the history of computers, and he's writing one now on the history of AI as a brand.

[A]

Yeah, it's more of a brand, isn't it? That's a bad way to think.

[B]

And interestingly, he argues that he was interviewed on another podcast. I listened to that. With all of the pushback on AI right now, we might see these companies start to rebrand it themselves, that they'll use other phrases for because they want to stay away from AI. Because you have people like Desantis who are saying all AI is evil.

[A]

Right. So here's a picture of me before. It's JRPG style, and here's a picture of me after.

[C]

This would be your icon in the corner of the screen.

[A]

Thank you.

[B]

I like that the microphone got slopped, so to speak.

[A]

Everything got slapped.

[B]

Yeah, that's good. Who did that? Anthony? Very good.

[A]

Probably Nano Banana. And yeah, I think the prompt probably did say. What did you say? Parasitude.

[C]

I think it seems jrpg style.

[E]

Yeah, 16 bit.

[A]

Yeah, it's like super ness. Yeah. One of those old RPG games. Yeah, Very nice. I like it.

[B]

Unleashing creativity from our.

[A]

I think our tomato game's going to end up being pretty good. I do think we have to recontextualize AI because I do think because we ascribe so much to it, it's disappointing to people when it doesn't do the things that they are ascribing to it.

[B]

Well, this is the whole AGI thing. There was a fascinating debate between Yann Lecun and Debbis Hasabis while we were gone and tour.

[A]

Didn't they just. Didn't they debate before?

[B]

No, that was. I went to an event where was somebody else from deep.

[A]

Oh, okay.

[B]

Debated with Jan. So Demis is. Jan was saying, it's not AGI. It's not general intelligence. There is no such thing. Anyway, we're good at. We as humans are good at some things, bad at other things. We're spiky. So will aib, and I think he's right. But Dennis as no is still tracing the AGI brass ring. No, no, no. We will have a general statement. We'll have a general machine. And so that's what DeepMind and OpenAI and Anthropic are still going after. And that's why I'm On Team Team LeCun, because Jan is saying, no, we're going to have a whole bunch of specialized AIs that are going to do some Stuff really, really well, far better than we do. We already have some protein folding, but the goal isn't to think that we're. It's to replicate us. And I think this, this idea of replication has caused the industry huge problems because that's what's freaking people out.

[C]

Yeah, I think a lot of. I mean, it's one of the many issues that the industry is having right now is just that the promises that were initially made about what these products can and will imminently be able to do is just one not realistic. But two is perhaps like not as interesting as what they're currently doing or have the capacity to do.

[B]

Well, I think that's what we heard from Jason at the beginning of the show is, you know, I didn't see anything I want to buy. I don't see anything that my friends would use. It's still amazing. It's still doing great things, but it's being presented as if it's a consumer cure all. And that's the wrong way to brand it.

[A]

Lisa was asking me and I actually don't know what the answer to this is. She says, I want to take a course about AI. And one of my thoughts was anything that is by now a course is going to be outdated. Right. That you couldn't go to your local community college and take a course on AI that would be in any way timeless.

[B]

Well, I have the answer for her on line 156. Funny she should ask.

[A]

Is it this? The Stanford lectures?

[B]

Yeah.

[A]

Yeah. So I was thinking maybe sending this to her. She wants to go to an actual class. But These are Stanford's AI courses from online. Stanford edu. They posted on. On YouTube.

[B]

The great thing is it's taught by twins.

[A]

Oh, that's confusing. Did you. Have you watched this?

[B]

No, no, I. Because I just.

[A]

I didn't lecture one. Transformers. Is the whole course on here or just.

[B]

Yes, it is the whole course. This is just. I couldn't. I couldn't find the page for it, so I just put up number one.

[A]

They probably have a playlist somewhere. Yeah, I've been. You know, it's fun. I. You can get a college education on YouTube easily. I. I do a lot of MIT's open courseware stuff.

[B]

Oh, wait a second. Yeah, you should play it for just a minute because he has.

[E]

Can I put that on my resume? Leo took a MIT course.

[A]

That's. Yeah. You don't have a degree. Hello everyone, and welcome to CME tw.

[C]

I just need to see him say cloud.

[A]

I bet he uses cloud.

[D]

So my name is Afshin. I would be teaching this class with Sherving, who's in the back, and.

[A]

And. And Sherving is his twin.

[E]

Yeah.

[B]

If you look them up, they're twins. I think you might even put a picture of them up.

[E]

I don't think.

[B]

Yeah.

[A]

I would love it if they. Why aren't they both down there?

[B]

I. I know.

[C]

I want to be mirroring each other's movements throughout the entire course.

[A]

Yeah. All right. CME 295 transformers and large language models. Yeah. I mean, that's going to be a good course.

[E]

The question for Lisa is like, does.

[A]

She want to learn how to understand how they work? She wants to use them.

[E]

So there is no real consensus on that. So how can anyone have a course like. No one?

[A]

Yeah. I don't know if there. I don't know if there is a good answer to that. There's lots of places you can learn about the underlying technology, but I don't.

[B]

And one of Jason, of course, if.

[A]

You go to X, you'll see a thousand posts. Oh, yeah, I have the 13 best prompts ever and you'll never need to know it again.

[B]

Well, here's another one is I should share with her because I still have one. Nate B. Jones. He. Have you watched any of his full videos?

[A]

No, that I sent you.

[B]

I sent you and you said you wanted to get it and you didn't watch it. Leo, you.

[A]

I'm sorry, I'm busy. Oh, I actually have seen this guy. I like this guy.

[B]

He's really good. What? He has very short videos.

[A]

Yeah, I watched the shorts.

[B]

The long ones are amazing. And he. And he'll take you through how he made something with it in detail. He's great.

[A]

That's kind of what you want.

[B]

Beautifully. There's 35 minutes of just an explanation I've been trying to get.

[A]

That's the one I watched. I watched his Grok explanation.

[B]

It was really good. Good, I thought. Right.

[A]

Yeah.

[B]

He knows his stuff, but he also does practical things. Like, here's how he used Nano Banana to make a whole. To put his life as a. As a. As a game board on a huge screen. You know, he does that kind of stuff. It's like he could make.

[A]

I mean, these things are great because they give you ideas about how you can use it.

[B]

Yes.

[A]

But the beauty of this is it's yours. It's very. It's highly personalized. You get. You have it do exactly what you want it to do and what you needed to do. Now he has his own community on his website, natebjones.com actually, after you sent me that, I thought about signing up for this. I thought, ah, this kind of looks kind of.

[B]

I'm paying 200 bucks for him.

[A]

Yeah.

[B]

And it's a lot of stuff. The other one is the Jason Heiners. I signed up for his newsletters when we started the show.

[A]

Yeah.

[B]

One of them is, the first one is ideas about how to use this in business.

[A]

Yeah, there are. That's what I think is there's a.

[B]

Lot of resources like that, I think a course. No. People ask me for a book to read.

[A]

No. That's going to be out of date by the time they put it on the paper.

[B]

There's just too much. It's too broad. It's like saying, can you teach me computers?

[A]

It's really important again, that you have that metaphor that I cannot drink this river. It's impossible. I can dip into it. And I'm going to look for little things that I can use that I can try, that I can tailor to my life. And I think it's not a bad thing to start with. Well, I want to. Can you help me with a to do list or something like that? Or I've got to plan a wedding or maybe a New Year's Eve Eve party and go from there. But don't attempt to strength the whole ocean.

[B]

Well, the news part of it. I mean, I irritate you every week, putting tons of things into the rundown. But that's how I learn, is that I see all the stuff going on.

[A]

No, I don't. It doesn't irritate me. It's just I can't absorb it. So I have to let you pick the stuff that you want to highlight because I. I can't absorb it all. What is Nate B. Jones background? Is he.

[B]

I don't know.

[A]

I don't think he's a technologist particularly. He says AI strategist and product leader about. Anyway, yeah, I was impressed. I was impressed.

[B]

Yeah.

[A]

The problem is there's a lot of snake oil going around too, because it's the hot thing. So there are a thousand people who say, I can teach you everything you need to know. And yeah, a lot of that.

[B]

Oh, there's horrible online conferences. That one we made fun of a few weeks ago. Yeah, I just saw another one being promoted. Gary Vee speaking at an AI conference.

[A]

Yeah, once.

[B]

Gary, do you really know AI?

[A]

Once it's become a money maker. Stand back. You're watching Intelligent Machines. Jeff Jarvis Paris Martineau. So glad to have you. Paris.

[B]

Paris looks deep in thought.

[C]

Sorry, I'm back. I'm back on Tomato game and I've gotten distracted.

[A]

No good. See. Are you vibe coding it now?

[C]

Trying to.

[A]

Good.

[C]

What are you using right now? I'm using. I'm using both ChatGPT and Claude.

[A]

Yeah.

[C]

And we'll see.

[A]

That's another useful way to do it is combining these things. The other A lot of the latest, you know, techniques involve creating sub agents and MCP servers and having AIs use other AIs and collaborating people are trying a lot of very interesting things. It's a very vibrant and fertile area right now and I don't think there is any single pass at this point. So no.

[C]

I need to install pillow but it's given me tomatocamepy. I don't know what pillow is.

[A]

Pillow is probably like pygame. It's another library. This is the problem I ran into is now I've got to install a bunch of stuff.

[B]

Yeah.

[A]

We'Ve talked before for people who want to try AI locally and we talked a lot about that. I'm going to talk about that in just a moment. You're watching intelligent machines. But first a word from Helix and my mattress, Helix Sleep. I got a great sleep score last night. I camera it was like 88, 89. Felt so good. Helix Sleep, how are you preparing for the chilly season, the winter you're spending more time indoors, I bet. Wouldn't it be nice have a comfy mattress to curl up on with your kitty cat gizmo or to read a lovely book with or maybe to play a tomato game. Stay comfortable inside with your Helix mattress. No more night sweats, no back pain, no motion transfer. Don't settle for a mattress made overseas with low quality and questionable materials. Rest assured, your Helix mattress is assembled, packaged, packaged and shipped from Arizona within days of you placing your order. So you place the order and they make it just in time as you order made to order, which is wonderful. You can also take the Helix Sleep quiz which will match you with the perfect mattress based on your personal preferences and sleep needs. It's exactly what we did. Lisa and I took the quiz. It says side sleeper, front sleeper. Do you want a cushy mattress? Do you want a hard mattress? And it gave us the perfect mattress mattress. We love it. And I have to tell you, it has greatly improved my sleep. My deep sleep is better. It's longer. I sleep overall longer. In fact, this is exactly results that came back from a wesper sleep Study Helix did. They measured the sleep performance of people after switching from their old mattress to a Helix mattress. Here's what they found. 82% of the participants saw an increase in their deep sleep cycle. In fact, on average, they achieve 25 more minutes of deep sleep per night. Now honestly, deep sleep for me it's an hour, an hour and a half to get 25 more minutes is a massive improvement. And that's the sleep that matters. That's the sleep that cleans your brain. You know, that's where you really get the health benefits. Participants on average achieve 39 more minutes of overall sleep per night. That's because less time tossing and turning, less time staring at the ceiling, more time sleeping blissfully away time and time again. Helix Sleep remains the most awarded mattress brand tested and reviewed by experts like Forbes and Wired. Both picked it as their favorite mattress. Helix delivers your mattress right to your door with free shipping in the US and rest easy because you get seamless returns and exchanges. The Happy with Helix guarantee provides a risk free customer first experience ensuring you're completely satisfied with your new mattress. By the way, you couldn't get my mattress out of my sleeping hands. I love our Helix Sleep. No thought about returning it. We love it. Best mattress we've ever had. Go to helixsleep.com machines for 27% off site wide during the new year sale extended Best of web and that's exclusively for listeners of intelligent machines. You go to helixsleep.com machines for 27% off the new year's seal extended that's best of web. Now the offer does end January 11th, so go there right now and do enter our show name after checkout so they know we sent you. And if you're listening, after January 11th, you still should check them out. There's always great [e-mail] machines. Thank you Helixsleep for a great mattress. I'm very comfy and cozy. I look forward to. You know that when you have a great, great bed and a great mattress. I don't know if you've ever had this experience if you hadn't get the Helix Sleep. But you, when you go, when you lift the covers, you get in and you go, ah, that's such a good feeling. The end of every day I've got a ah, thanks to helix sleep helixsleep.com machines the idea of running an AI locally is, I think a good way to get started with this too. Instead of paying I, I pay 20 bucks to anthropic to OpenAI to Gemini To Google, to Perplexity, to Kagi, and on and on and on. But that's kind of my job. There are a number of different nice tools out there that people we've mentioned before. Studio LM Jan is a nice tool, but anything. LLM has just announced the ability to add web searches to its models. So these are tools you can download anything. LLM you can download in Windows, Mac, Linux, it runs locally. You choose a model. So you know you have to, you have to start with an AI that you can, you can download and there are many, many models out there. You can even just start with anything LLM's own stuff. But you can also. And all of these are free. They're. There are what we call open weights. Right. Deepseek Gemma from Google. Mistral, French AI company Mistral.

[B]

I think you should say I got.

[C]

My first result back and it's from Claude and it's. You know how we were discussing how Claude can't really do image generation right?

[A]

Is tomato wrong?

[C]

I mean the tomato is right, but the rest is a bit sad. I'm going to post it.

[B]

Can you share your screen?

[C]

No, I just created it. I think I posted published it.

[A]

It's, well, it's giving you entertainment.

[C]

I'm very entertained. But you need to click the link to see.

[A]

All right, here it is. Let's see Artifacts. Why it's kind of primitive now. How do I shoot it?

[C]

Oh, there we go.

[A]

Well, that's a start.

[C]

It's flats, you know, you like to see it. It's got your little mic there.

[A]

I like how the tomatoes disappear after a while.

[C]

I like like that. They do splat though. I didn't even ask it to splat. I just said I'd like a tomato game.

[A]

Ironically, the tomato, it's got.

[C]

It put Club Twit there, which I didn't ask it to do either.

[A]

You didn't ask it, it just knew.

[C]

Oh, and it said aim for the Hawaiian shirt for maximum satisfaction.

[A]

Oh, you're right.

[C]

I was aiming for the head. I told it. Well, I, I, you know, it's quite.

[A]

Satisfying even as, as primitive as it is.

[C]

Pretty satisfied, right?

[A]

Okay, Paris, you are a coder.

[B]

There you go.

[C]

I am, yeah.

[A]

Well, by the way, as primitive as it is, that's just the graphics. The rest of it's pretty sophisticated, actually. I mean, I think it's pretty good.

[C]

Collision detection cluds about 45 seconds and zero follow up questions.

[A]

The Club Twitch sign lights up from time to time. I'm not sure why.

[C]

Listen, I'm just happy it's there.

[A]

There's stuff going on.

[C]

I like that Leo. That cartoon Leo just rocks gently back.

[A]

To side to side, moving slightly. That doesn't make it any harder.

[C]

He's unblinking.

[A]

But see, you could then go in and say, can you make the eyes blink every once in a while? I think that would make it worse. Well, you could play with it and say, no, I don't want the eyes to blink. Very nice. Well done.

[C]

Done.

[B]

Good work.

[A]

Very good.

[B]

You're a nice job.

[A]

Tomato toss, Leo Edition. Look anywhere to throw tomatoes at your favorite tech podcaster. Did you put that in or did it.

[C]

No, I. So my entire prompt was. Well, let me see if I can go back here. Claude.

[A]

That's pretty impressive, actually, that you did that in about five minutes.

[B]

Yeah, exactly.

[C]

Not even the amount of time I spent squinting at the screen was me trying to get chat GPT to do it. I. After you guys asked, what's up? Also, paste it in clud. I texted Create. Create a GUI game where you throw a tomato at my friend Leo, pictured and included a screenshot of you sitting in your chair. Wow. And it was.

[A]

That's it.

[B]

That was it.

[C]

That was it.

[B]

So it saw the club twitch behind him. Is that what it did?

[A]

Yeah, that's what that is.

[C]

But that's it. That's her club.

[A]

Twit.

[B]

Come on.

[A]

Huh? You could say, make it more realistic. I'm sure it would do that.

[C]

Let me see.

[B]

Oh, now we've lost Paris into coding. You've done it, Leo. You've done it. You've wandered down into the vortex.

[A]

If nothing else, it's a fun toy, but I think it is a lot more than that. I'm very intrigued by it.

[C]

I'm saying make it more realistic and make Leo move wildly around the screen.

[A]

Oh, that's too hard now. You made it too hard. Harvard has a whole bunch of online courses as well on AI that you can take for free, I think. Yeah, that. I also. Yeah. Three Blue One Brown is an excellent YouTube channel for learning about that. They have a very good course on neural networks. But I think what Lisa is not looking for is how it works or, you know, what's the underlying technology, but really, how can I use AI? And that's a harder one to answer because it's so personalized.

[E]

We're still in the experimental phase of all this. Like, everybody needs to try stuff. Like, that's the whole point right now, is like, everybody needs to try things. And then, you know, in a sense, it's like. It's really a SaaS product. It's not really for us.

[A]

I think people are doing some pretty amazing things. Let's see. So here's Gali as a dragon. Release to throw tomatoes. Let's see. Where's the tomatoes? Oh. Oh, yeah. I'm the tomato, apparently.

[C]

I like that. I think that's pretty good.

[B]

It doesn't make sense.

[A]

Made in Bolt. I don't know what Bolt is, but. Thank you, Galia. Very fun. So we got everybody to.

[E]

Do you need to drag from where it says drag and release to throw tomatoes? Drag from there. Click. And drag from there.

[A]

Oh, oh, I see. Oh, I didn't. Human error. Drag and release.

[C]

So you've got to drag back, I think, maybe. Right. Yeah.

[A]

So do this.

[B]

There's a tomato. Look, there's a tomato.

[A]

Now underneath it, the tomatoes are over there.

[B]

You're right.

[A]

I'm dragging them the wrong way. I don't have a lot of drag room.

[B]

You got a score.

[A]

My screen's too short. I can't. I can't get the tomatoes.

[E]

Yeah, the slingshot needs to be higher.

[A]

Yeah. Whoa.

[B]

It's counting as getting you even when it doesn't hit you.

[A]

Well, it's fun. We just need a little bit more play for the sling. All right. But those are real tomatoes. Now, is Bolt an AI product? Galia? All right, we're going to try one more time. Did you. Did you. Is it better?

[C]

No, it's way worse. Again, I said make it more realistic and make Leo move wildly around.

[A]

Just a black blob now, but there is a boom when the tomato hits me. Direct hit. Where's Leo?

[C]

I like that. There's a Leo speed in the corner up top. It says.

[A]

Oh, I could turn the speed down, maybe quicker. Can I turn the speed down?

[C]

I don't think so. I think that's just for you. Just for you to know that he could be fast.

[A]

Combo.

[E]

Oh, no, it goes faster the more you hit it. It gets faster the more you hit it.

[A]

Oh, no. Oh, no. Combo four. Combo. Boom. Combo five. I'm getting good at this.

[E]

Hit box is too big. The hit box is way too big.

[A]

I don't think.

[C]

I'm not.

[A]

Yeah, not even hit. That's okay. I want a big hit box.

[C]

Oh, the hit box is too big. You're right.

[A]

But you tell it.

[C]

I like that sometimes. It says, got him. I like the icon that happens when you throw the tomato.

[A]

Oh, I slowed him. It slowed him down. Oh, now he's. Whoa.

[C]

Splat. Boom.

[B]

Combo. Quick. But are you.

[A]

You're gonna see a lot of these on app store and other places.

[E]

Oh, yeah. Anthony just said in chat. It reminds me of the old flash game days. Flash games back in the day.

[A]

Yeah. These are flash games. Yeah.

[B]

All right. Is this slop?

[A]

Game slop? I mean, you wrote it.

[C]

It's game slop. It is essential. Is game slop. But I love it. Game slop is derogatory or slop is derogatory.

[A]

Oh, I can. It seems it bounces off the old tomatoes. So I can. I can slow it down by giving. I can. I can rein it in. Oh, oh.

[E]

And again, like what you like.

[A]

Good game, Paris.

[E]

That's fine.

[A]

How much would you pay for this? Yeah, pretty good. Pretty good tomato toss. He's quick, but are you quicker?

[E]

Okay, this is great radio, guys.

[A]

This is not a good use.

[B]

Sorry, people out there.

[A]

Let us. Let us get to the picks of the week. What do you say you're watching Intelligent machines with. With our new slop creator, Paris Martineau. And Jeff Jarvis, who does nothing but writes amazing books like the Gutenberg brethren.

[E]

Hold on Magazine. We're short one ad still. We do still have one ad still.

[A]

I have an ad.

[B]

Yep.

[A]

Oh, yeah, you had three. I thought I only had two.

[C]

Are you dreaming?

[A]

I have an ad about my mattress. All right, we'll have the picks of the week in just a moment. This time of year, everyone talks about going dry, but at Athletic Brewing Co. We're skipping that because we prefer going athletic, which isn't dry at all. From crisp goldens to hoppy IPAs and limited releases, in between, you'll find something that fits your style. Every single non alcoholic brew is packed with flavor and the same craft experience you love. So, yeah, you could call it dry, but there's really nothing dry about it. Find your new favorite [e-mail] Athletic Brewing Company fit for all times. Carvana is so easy. Just a click and we've got ourselves a car.

[D]

See so many cars.

[C]

That's a clicktastic inventory.

[A]

And check out the financing options, payments.

[B]

To fit our budget.

[D]

I mean, that's Clickonomics101.

[B]

Delivery to our door.

[C]

Just a hop, skip and a click away.

[A]

And bought.

[D]

No better feeling than when everything just clicks.

[C]

Buy your car today on Carvana.

[D]

Delivery fees may apply.

[A]

All right, ladies and gentlemen, I think this would be a good time to get our picks of the week. And as always, I will start with Paris Martineau.

[C]

You know, I said that because I cannibalize my picks of the week that I was going to use a tomato based Leo games one thing which you.

[A]

Also cannibalize that as well.

[C]

So I'm going to do a little bit of double dipping here. We talked about the spit in the pre show but I had the flu over the weekend. But I'm here recording this podcast thanks to the wonders of modern medicine and Tamiflu. And I found this article from Bloomberg today really interesting about the fact that TAM flu, which is an antiviral to stop the, I guess replication of the flu virus is now in, in short supply in lots of parts of the US because this bout of the flu is hitting the US so badly. They talked to kind of some different pharmacists around.

[A]

Is there a name for this flu?

[C]

I think it's H3. I think it's crappy N3 or H2N.

[A]

Yeah, not a good one.

[C]

And it's a clusty and it's H3N2 called Claude K. Oh, not code. Well and part of the reason why this flu in particular is so bad that is whenever they decide on the flu vaccine, they have to decide what kind of specific sub target or I guess like submutation of the flu. They're going to target that vaccine ahead of time because they have to manufacture it and they chose a different one.

[A]

So they didn't get the right variant.

[C]

Yeah, they didn't have the right.

[A]

It was a year ago practically that they had.

[C]

Yeah, I mean you gotta, you gotta decide early on this sort of stuff. And that wasn't right. So now this flu strain that already is kind of mutated is, is spreading like wildfire. So I don't know. One would recommend getting the flu vaccine regardless because I think it's still useful. I had the flu vaccine. I partially attribute that to my fast recovery. But also make sure you ask your doctor about getting Tamiflu if you end up getting the flu because it really helps. But it really, really helps if you start within 12 hours of symptoms like I did, because then it stops it from multiplying and shortens your symptoms by three days.

[A]

If you get Covid now you got. I didn't even know this existed. You got a test that is both a COVID test and a flu test.

[C]

You know, I didn't know it existed either until Friday when I was feeling bad and I was on the couch and I was like, I guess I'm. I'm out of COVID tests. I guess I'm gonna doordash a COVID test me.

[A]

Wait a minute, you ordered it on doordash. Yes.

[C]

Because it was Friday and I live alone and how else was it gonna get.

[A]

And they go, yeah, you don't wanna go out and get it.

[C]

I also couldn't find any of my masks. I was like, I don't wanna get any sick. And I was like, well, I guess if I'm gonna get a COVID test, why not get a combo Covid flu test?

[A]

Sure. And a Nestle's Crunch bar. I understand.

[C]

And a Nestle's Crunch bar. And that made all the difference. Another one of my two kinds of fl.

[A]

I could just see you have a top of a filter. I'm gonna get a test. Let me get a crunch bar too.

[C]

I mean, I thought about. I should have gotten two. Honestly, it looks like. It looks like a normal COVID test, but instead of one window where you wait for a line, there's two. And one of the windows has, you know, a control line area as well as a line for flu. A line for flu B.

[B]

So did you have flu A or flu B?

[C]

Flu A.

[A]

Flu A. That's actually great because the COVID symptoms overlap. The flu symptoms and you.

[C]

I mean, it's ideal.

[A]

Yeah. Because if you had Covid, you'd take Paxlovid. If you had the flu, you take theraflu.

[C]

I know. And in both cases, I was like, I'm starting that immediately.

[A]

Yeah.

[C]

Because it really.

[A]

Paxlovid works too, by the way.

[C]

Paxlovid really works as well. I've taken that. Yeah.

[A]

Yeah.

[C]

Except for it does make your mouth taste like metal. But that's a reasonable side effect in my opinion. My other flu season recommendation would be babka from Zabars, which one of my friends dropped off for me. You can get it delivered to yourself. It was chocolate, Bob say bars.

[A]

Oh, that's so good.

[C]

They ship it, so.

[A]

I know they do. I've had it shipped to me, by the way.

[C]

It's delicious.

[A]

From Zabar's and from Zingerman's, which is another great place to get your babka.

[C]

And I learned because my friend who dropped off supplies to me is gluten free. They have gluten free babka as well. So you know, celiacs get it.

[A]

It's better with gluten. Gluten, though?

[C]

Yeah, definitely. I'd agree every. Most things are.

[B]

Did you see one of the things that CES was a gluten detector you can carry with you.

[A]

Oh, God.

[C]

I bet it's just a little AI.

[A]

I have a gluten detector. It's my life.

[C]

I was gonna say, does a piece of food look appetizing? It probably has gluten. And lastly, happy birthday to Nicholas Cage.

[B]

Oh, you're here.

[C]

62 today.

[B]

Wow.

[A]

Maybe next November we can have a. We can all have a nick of a nick of him.

[C]

No, I mean, I think we should all at least watch one Nick Cage movie in the club. And I did.

[A]

That's a good idea. You could pick the movie. Yeah.

[C]

Great. I included in the rundown, under my fix of the week, a photo of Nick Cage's pyramid, which is where he's going to be buried.

[A]

He bought a pyramid?

[C]

He bought a pyramid. In 2010, he purchased two plots in a cemetery in New Orleans to construct a large marble pyramid.

[A]

Is he of the thinking that that will preserve his body? That's what some.

[C]

I don't know, but he's got on. On it says omnia ab uno. I don't know what that means, but all.

[A]

All from one.

[C]

All from one. Everything from one.

[A]

That's the opposite of a pluribus unum. No, it's the same from one. Many, many, one. Huh, Food for thought.

[E]

He's a. He's a impulse buyer though. He buys a lot of weird stuff. That's why he makes so many movies. That's why he makes so many movies.

[C]

This was, I believe, during the period where I like to refer to it as Nikita's pyramid debt period. But I'm not sure if the pyramid actually put him into debt, but it was a period where he was making so many wild purchasing decisions such as exotic animals, a lot of snakes. I think there was an island or two in there that he had to make a lot of. Some might say less high brow movies.

[A]

God bless him. You know what, if you're gonna be a rich movie star, spend money on weird things. Yeah. By the way, Louis, number one cemetery. Yeah. Instead of bugging us, I will also.

[C]

I will read one of. Look at the kisses age quotes.

[A]

Look at the kisses people have planted on the mall.

[C]

As they should. This is an interview with Playboy magazine in 1996. Playboy says, Nick Cage, your salary is shooting up into the multi millions per movie. Reportedly 4 million to 7 million. Do these numbers make you chuckle, Nick Cage? I don't chuckle. I have respect for the dollar.

[A]

Chuckle. I don't chuckle.

[C]

I have respect for the dollar.

[B]

I have respect.

[A]

All right, your pick of the week, Mr. Jeff Jarvis.

[B]

Okay, so I guess I'm gonna do this one. This one. So I bought, I finally bought a new phone. I finally. And opened it and installed. It's a Google 10 Pixel XL.

[A]

Oh, good. Oh, good. How do you like it?

[B]

It's, it's fine. It's nice.

[A]

It's just a phone.

[B]

It's just a phone. Yeah, it's, it's, it's, it's. But so my daughter got a new phone for Christmas and I was going into the, the Apple Store with her and she was doing the trade in because we got a lot of money for the trade in. It's a good deal in the Apple. And no, we had to sit there and wait while the whole thing transferred over. And then while it uploaded, we sat there for two hours. Two hours in the Apple Store. They gave us a chair once.

[A]

Why did you do that?

[B]

Because.

[A]

Why did you go home and do it?

[B]

No, they wouldn't let you. You have to stay here.

[C]

What?

[B]

We're imprisoned. We were. We were kidnapped by the Apple Store.

[C]

They tackled you?

[B]

Yes.

[A]

Wait a minute. So you got an what?

[B]

This isn't even the story. She traded in her phone for a new iPhone and they made her transfer it over and that made her.

[A]

Oh, she had them transferred over instead of her doing it herself, which you could easily.

[B]

No, no, they wouldn't let her because they said, you have to give us the phone when it's all transferred over.

[A]

You because you just traded it in.

[B]

You don't own it anymore. We own it all right.

[A]

Right.

[B]

So then I also got her a screen.

[A]

It's not a good user experience.

[B]

No, it's not. It's not.

[A]

Not.

[B]

Yeah, they're coming over, looking at us once in a while like, hey, we're still here. We're just sitting there in the Apple Store like we're, we're vagrants. Right. So in the old days I used to do that. Right. It was the only place you get WI fi in the city. I'd go sit in the Apple Store. Anyway, so I got her a screen protector and they do the. With the screen protector. Now if you've seen this, they put it in this, this device machine and it pulls out and then it comes out. So we did that. That was fine. Went back to the car and she bought a. A case in Best Buy. Put the case on, boom. Ruin the screen protector.

[A]

Oh my God.

[B]

We just spent two and a half hours an hour.

[A]

Back to nightmare experience.

[C]

You don't need a screen protector. You don't need a case. I've been carrying around this phone raw loose for years and it's fine.

[A]

You're a very responsible unit.

[E]

I've done that. Since the beginning as well.

[C]

I drop this thing every single day.

[B]

Pass the word, Paris.

[C]

I dropped, I've dropped this phone probably three times today.

[A]

Wow.

[C]

And it's fun. They make them strong now.

[B]

So we go back in, they say, oh, this is, this is apples good. This okay, no, no, that's not good. So they put on a new screen protector. I watched the thing goes.

[C]

Did you have to pay for it again?

[B]

No, no, no, no. They do it. They have to go through their, their, you know, bureaucracy. I said, you put the case on now. Put the case on. Ruins it again. So my daughter, bless her, says, I'll just buy a case here, we'll return that one. When the best buy clearly bad, she gets a clear case, puts it on, it works this time. So I, with my phone, I thought, well, should I get a screen protector? You're right, Paris. I didn't have one before and I thought, I'll get one. And I saw this amazing thing online. This is going to be a confession from Toras. And it comes with a little device with the protector itself, just like the Apple thing. You put it in there and you put it in and you pull it out and then it's supposed to work. And so there's video of this you can watch. So I'm, this is my stupidity moment. I'm really stupid. Then it says, take this thing off your phone. Which I did, right? And there's this thing still around my phone that has a tab on the top and it's too big in the bottom and well, this is bad. So I email the company, I say, this doesn't work, I want a refund. And I take the thing with a tab off, saying, well, that just took a whole thing off. And it had left the screen protector on perfectly. Absolutely perfectly.

[A]

That's the whole point.

[B]

I had no idea that's the applicator. But as could be, that was the applicator. I now have a Beautifully produced. So I, I wrote back to the company and I said I'm sorry.

[A]

And I was emailed 20 times a day. But they're doing it. The thing still has a ton on it. What's.

[B]

No, it's a Chinese company and everything was through God knows what Deep Seek Translate and they got no idea what it was. So I just said, I'm very sorry. It works very well. So that was my little confession of Jeff's old age stupidity.

[A]

Well, good for you. They, a lot of the screen now come with applicators because they realize no one can put these on.

[B]

No, they Can't. You can't.

[A]

It's always ridiculous.

[B]

And it's really. It's really nice. It's great. Product is very good. It's placed it perfectly. It's. It's quite nice.

[A]

Does it feel like you're this.

[B]

No, it doesn't. No, no, I've.

[A]

No. They're glass.

[B]

You know, the whole point of the story is I didn't realize was on there.

[A]

They're tempered glass.

[B]

I didn't know it was there. Only when I finally looked at the end, I said, oh. Oh, it left it. Oh, that's how it works. Yeah, yeah, yeah. It's two hours later.

[A]

So this is my pick of the week. Was inspired by the RPG thing. This is floor 796. This is a project by a Russian guy. He's been making this since 2018. It is a massive, massive gif, which is. You know, it is. Because 796 are the alphabetic letters for G, I and F. It is supposedly floor 796. Yeah. It's a giant animated GIF.

[C]

And it's just one.

[A]

Yeah, he adds to it all the time. And there's a lot of Easter eggs in it. Wait a minute. Let me see if I can find Claude Van Damme, because he will. Who?

[C]

Claude Van Damme.

[A]

Claude Van Damme. You can click on people and they. Some of them do things. Yeah, there we go. You saw that one. Did something. Let me see if I find Claude.

[B]

That's a very weird thing going on in that room down there.

[A]

Oh, it's all stuff from movies. There's a Hannibal.

[E]

Yeah, these are all references. Yeah, these are all references, yeah.

[A]

Oh, it's all stuff. Real stuff from movies and stuff. And he even has some Easter eggs. If I could. Ben Affleck's smoking. There's a tardis. Oh, now. Where? Where? Oh, well, I can't find him. Let me. Let me click the menu. Oh, this is. I know why this is when I downloaded so. So I don't even getting any help, but it's a lot of fun. Floor796.com. It's a Russian guy. Look, there's the showers. The mummy jumps out of the locker room. Russian guy's been making this for eight, nine years now.

[B]

Jesus.

[A]

Yeah, he's a little obsessive.

[E]

Waldo's got to be in here somewhere.

[A]

There. Waldo's in there. If you find Waldo, click on him. I can't remember where he is, but yes. Oh, yeah, Very good. It looks like a Where's Waldo? Doesn't it? Ah, There's. There's John Claude. Here he is.

[C]

Not Waldo, though.

[A]

He's just waiting. He's saying, come here, boy.

[E]

No, that's Chuck Norris.

[A]

Boom.

[E]

That's Chuck Norris.

[A]

I'm sorry. Chuck Norris.

[C]

Yeah.

[A]

He punches you in the glass so it's really.

[C]

God. That's what the Internet used to be all about.

[A]

This is what the Internet used to be.

[E]

The old Internet. Yep.

[A]

Doesn't it. It's just fun. I just like it. I'm glad that I don't know what's going on here, but it's kind of weird. Oh, that's a holodeck, I think. Yeah. Floor 796 of a space station. There's a lot of stuff going on. He's adding to it. It's not done. He says. In fact, it's only about 59% done. You can turn on wandering and it'll wander for you. If you just want to have it, like, as a screensaver or something.

[B]

Something that would. No, no.

[A]

Floor 796 compete. Old Internet. Exactly, Briggs. Exactly. But click on them because there are a lot of little Easter eggs in here. I think he has a list somewhere of all the different Easter eggs. Ladies and gentlemen. No AI Create. This is slop. Without the AI, actually. So fresh.

[E]

This is art. That's not slop. That's art.

[A]

No, this is art. I think this is art. Somebody's done a lot of.

[C]

Yeah, no, this is art. I thought you were with the podcast.

[A]

The podcast is.

[C]

That's fresh. Homegrown slop.

[A]

We do this show on Wednesdays. I hope you'll join us.

[B]

Gluten free slop.

[A]

Gluten free slop. We do it at 1400 Pacific Time, 1700 East Coast Time. 2200. No, 2300 UTC. If you want to watch it live, you can. It's on Twitch, it's on x, it's on YouTube, it's on LinkedIn, Facebook and Kick online everywhere.

[B]

You cannot escape us.

[A]

You cannot escape us. In fact, if you're in the club, you can even watch it in the club. Twit Discord. We would love to have you there after the fact. On demand versions of the show available on our website, Twit TV IM. There is also a YouTube channel for the video. Great way to share clips. And you can even subscribe to it because it is a podcast. And get it wherever you get your podcast and that way you'll get.

[B]

Are you supposed to be promoting the survey?

[E]

One more thing.

[A]

Yeah, I forgot. Thank you. I haven't mentioned the Survey once this week, we are doing the survey. It's all my fault. If. If we sign, it helps us immensely. Take about 10 minutes of your time. TWiT TV survey 26. It just asks a few questions. We're trying to. It's the only way. We don't know about you. We can't. Because it's a podcast. We don't collect information about you. But once a year, we like to do the survey. It helps us sell ads, but it also helps us understand what you like and don't like about the programming. Who you are, what your interests are. So if you would. TWIT TV survey 26. We do this every year. Even if you've done it before, please do it again. And thank you in advance. It's a big help. We appreciate it. Thank you, Jeff Jarvis. Have a wonderful week. When is your surgery? Is it before next week?

[B]

First, so. No.

[A]

So you'll probably miss a couple of.

[B]

No, no, I'm gonna. If you don't mind seeing me like.

[A]

That with an eye patch. No.

[B]

It'll be in the morning, so I think Pirate Jeff.

[A]

Well, okay, if you're up for it. If you're not, don't, don't, don't hesitate to. To not do it. Yeah. Paris Martino. I was prepared. I thought, well, maybe Paris is going to still be under the weather, but you're a trooper.

[B]

Oh, you're here in full fine fettle.

[C]

I'm here and I'm. My voice isn't even that shot, which is great.

[A]

No, you sound great. Ms. Martineau is at the Consumer Reports where she covers food safety.

[C]

It's true.

[A]

And I saw a bunch of radioactive shrimp memes over the holiday and I thought about you.

[C]

I thought, wow, that's still got legs.

[B]

We need to do a radioactive shrimp game. Yeah.

[A]

Oh, throw radioactive shrimp at Leo.

[D]

Yeah.

[A]

And maybe some lead based protein powder if you're really feeling mean. Maybe, maybe, maybe. Thank you, everybody. We love you. It's good to see you. Happy New Year from the Intelligent Machines gang. And we'll see you next week.

[B]

Bye.

[C]

Bye. Bye, bye.

[A]

Bye. I'm not a human being.

[C]

Not into the animal scene. I'm an intelligent machine.