Intelligent Machines 897 transcript
Please be advised that this transcript is AI-generated and may not be word-for-word. Time codes refer to the approximate times in the ad-free version of the show.
Leo Laporte [00:00:00]:
It's time for Intelligent Machines. Jeff and Paris are here. We have kind of a postmortem on Jensen Huang's departure, what it means for NVIDIA and why it happened. Plus OpenAI releases a model everybody hates and tomorrow Anthropic makes it big. All that's coming up next on Intelligent Machines.
Jeff Jarvis [00:00:22]:
Podcasts you love.
TWiT.tv [00:00:24]:
From people you trust.
Leo Laporte [00:00:26]:
This is TWiT. This is Intelligent Machines with Jeff Jarvis and Paris Martineau, episode 897, recorded Wednesday, November 18th, 2026. I like money, bro. It's time for Intelligent Machines, the show where we cover the latest in AI, robotics, and all the smart doohickeys all around the house. Speaking of smart doohickeys, ladies and gentlemen, Paris Martineau. Investigative journalist. That's Ms.
Jeff Jarvis [00:00:58]:
Dohickey to you. Ms.
Leo Laporte [00:00:59]:
Dohickey, Consumer Reports, paris.nyc. Hello, Paris. Hello,
Paris Martineau [00:01:04]:
Leo.
Leo Laporte [00:01:04]:
Soon to be an owner of a brand new MacBook Pro. You smart person, you. It's true.
Paris Martineau [00:01:10]:
I have to figure out whether I'm allowed to take the MacBook Pro home and then transfer it from my old MacBook Air, or if I have to give it in the second I receive it. But that's a problem. Oh, no, no.
Leo Laporte [00:01:23]:
They give you time to do exactly that. You did a trade-in? Yeah, they
Paris Martineau [00:01:26]:
give you time. I did a trade-in, but I wanted I'll do the trading in person 'cause I don't have to mail a computer.
Leo Laporte [00:01:30]:
Oh, well, they'll do it for you. They'll connect them. You should probably, you know, buy yourself a smoothie, but you know.
Paris Martineau [00:01:35]:
Yeah, I mean, that's the thing is I don't really wanna wait. Exactly. I just, I have all my things on my download in the Time Machine. You just wanna boot up the computer and make sure nothing got smooshed while transferring.
Leo Laporte [00:01:49]:
No, I think it's smart. You're very wise. And Apple has a thing to do that.
Paris Martineau [00:01:52]:
This has been the tech hour with Paris.
Leo Laporte [00:01:55]:
Also here, ladies and gentlemen, The most interesting man in the world, the Hathaway Shirt Man.
Jeff Jarvis [00:02:01]:
A reference that only people over a certain age will understand.
Leo Laporte [00:02:04]:
AKA Jeff Jarvis, Emeritus Professor of Journalistic Innovation at the Craig Newmark Graduate School of Journalism at the City of New York. Author of Hot Type, now available everywhere, the magnificent machine that gave birth to mass media. Pirate Jeff is wearing an eyepatch today, but not for a long I might take
Jeff Jarvis [00:02:26]:
it off soon. I had my cataract surgery yesterday. I was nervous little Nelly.
Leo Laporte [00:02:33]:
You
Paris Martineau [00:02:34]:
were a scaredy cataract?
Jeff Jarvis [00:02:35]:
I was a scaredy cataract.
Leo Laporte [00:02:36]:
Do they put you out for that?
Jeff Jarvis [00:02:38]:
No, because they don't want you to see.
Paris Martineau [00:02:39]:
Do they put you down for that?
Jeff Jarvis [00:02:42]:
They Valium in the vein. And I think it's Valium.
Leo Laporte [00:02:48]:
So you relaxed.
Jeff Jarvis [00:02:49]:
You relaxed. You relaxed. And then all I saw, I told her as soon as she was done, Uh, was that this was a psychedelic light show. All I saw was weird abstract lights in my eye. That's it.
Leo Laporte [00:03:02]:
Is your left— so this is your left eye. Is your right eye closed during this?
Jeff Jarvis [00:03:07]:
I did cover it.
Leo Laporte [00:03:08]:
Okay, they cover it. So you're only seeing out of the one eye and you have to keep it open. Do they—
Jeff Jarvis [00:03:13]:
They, they put a thing in and just open it.
Leo Laporte [00:03:15]:
So it's kind of like Clockwork Orange.
Jeff Jarvis [00:03:16]:
It is. It is. Yes.
Leo Laporte [00:03:18]:
Okay.
Jeff Jarvis [00:03:18]:
Exactly. Exactly.
Leo Laporte [00:03:20]:
Uh, wow. Wow.
Jeff Jarvis [00:03:22]:
Like, that reminds me, I didn't put my drop in the first commercial break. I have to go back and put my drops in.
Leo Laporte [00:03:26]:
Yeah. Oh, but how's your visual acuity now?
Jeff Jarvis [00:03:32]:
So I got warned that when you, when you do this, you have a choice. You either get a distance or close. And if you put distance in, which is what I chose to do for both lenses, then the next one will be done in 3 weeks. Then you give up reading. And I thought, okay, okay, okay. But we're all of us who have glasses—
Paris Martineau [00:03:49]:
You give up reading?
Jeff Jarvis [00:03:50]:
Well, no, no, no. You have to deal with that. Although you have to have, you have to read glasses. But all of us who have crappy vision are used to being able to, at the minimum, take off our glasses, put something up close, we can read it. That's gone. That is gone. Yeah. And so—
Leo Laporte [00:04:04]:
See, I would miss that. I actually do everything like that.
Jeff Jarvis [00:04:06]:
Yeah, that's, that's way gone. So I was out today taking a walk trying to do each, each thing, and I can't— I couldn't see the time on my watch. I couldn't see the time on my phone. Oh, that's terrible. I couldn't do anything. So I can't— my wife puts her reading glasses— she said the same thing— she had reading glasses on and off and on and off. I can't do that. I can't do that because I'm constantly on the phone or the laptop or whatever.
Jeff Jarvis [00:04:24]:
It's where I live. So I think I'll go with progressive from zero to reading, or I'll call the doctor and, um, uh, show her how crazy I am and say, can I get one? The other choice is you get one reading and one distance.
Leo Laporte [00:04:39]:
Monocular.
Paris Martineau [00:04:40]:
That seems wise.
Leo Laporte [00:04:41]:
I do that with my contacts. My sister did that.
Jeff Jarvis [00:04:44]:
She found that after a while it made reading harder.
Leo Laporte [00:04:48]:
Yeah. Dano in our Discord, our club member Dano, said he got something called Quad Optic, which is both, for a mere $7,000.
Paris Martineau [00:04:58]:
I thought Quad Optic is where they put another 2 eyes in your head.
Leo Laporte [00:05:02]:
Now that I buy!
Jeff Jarvis [00:05:03]:
Quadraphonic? Yeah, yeah.
Paris Martineau [00:05:05]:
That's worth 7 grand.
Jeff Jarvis [00:05:07]:
Sure, toward the back. That's exactly it. You can see them coming.
Leo Laporte [00:05:10]:
Our guest this week on Intelligent Machines, actually we recorded the interview earlier, My good friend and one of my favorite authors, Daniel Suarez. You may remember his books. Well, it all kicked off with Daemon and then Freedom TM, 2 of the most dramatic, I think maybe the most dramatic books ever in sci-fi. Just really a great page-turner. But one of the things that's amazing about Daniel is he's Also kind of a seer. He has— he can see into the future, it feels like. So his sci-fi very often feels like it's always near future. And you very often feel like, you know what, I feel like this is going to happen.
Leo Laporte [00:05:58]:
And in many cases it did. Now the AI in Daemon and Freedom isn't quite the same. We're going to talk about that when we talk in our interview. But he did also predict basically the Ukrainian drone war with Kill Decision. He's— it's all about autonomous killer drones. He won the Prometheus Award for Influx. Change Agent talked about CRISPR, genetic crime. He's writing a trilogy called Delta-V.
Leo Laporte [00:06:30]:
Well, anyway, a very talented sci-fi guy. We've interviewed him for almost every book he's done, and I— I was very excited to get him back. His new book comes out this spring, Shadow Zone. We'll ask him about it. But the first thing I had to ask when, when, when I talked to him on Thursday was, tell me about your relationship to AI. Because, right, everybody's talking about do authors use AI? What's, what's his relationship to AI? Watch.
Daniel Suarez [00:06:57]:
Well, I was just looking at a quote I have on my bulletin board from Rhiannon Giddens. It's, the process is the point. And I thought that really stuck with me. You know, I— the whole point of writing these things is to understand what I think. And in the course of writing, I learn more about what I think and what questions I want to answer. And that process is entirely the point. I do not want to offload that. And that's probably why it takes me years to do books.
Daniel Suarez [00:07:23]:
But that journey is very important to me. And, and so far, my readers have been have been patient and, and wanting, you know, the result of that when it, when it finally comes out.
Leo Laporte [00:07:34]:
The nice thing is, Daniel, yeah, it takes years, but you're very consistent. Every few years we're going to get another Daniel Suarez. Yeah, people love Delta V. Is this the 3rd volume you'll be working on now?
Daniel Suarez [00:07:45]:
Yeah, and I do love those books. That's the other thing is by taking the time, I really do feel attached to these books. You know, again, they're not always positive, but I think the best thing for me about the Delta-V series is it is a positive view of a potential future if we make the right moves.
Leo Laporte [00:08:03]:
Right.
Daniel Suarez [00:08:03]:
None of that, none of the tech in it is fiction. None of the moves we make, not even the trajectories in this book are fiction. So those are things we can actually do to materially improve the future. And again, we go to future generations to improve the lives of countless future generations. rather than making things worse. And again, not by degrowth, not by reducing our lives to that of a 17th century feudal farmer, something like that. Right. We really can have a very interesting, cool, wonderful life, and we can hopefully make that possible for many people.
Daniel Suarez [00:08:43]:
I mean, I think that that's a laudable goal. So getting energy and resources allows that. I think a lot of people, especially when they think of asteroid mining or space exploration, They think of it as, again, a way to concentrate wealth and power into fewer and fewer hands and also to abandon the Earth. I think that's a big mistake, that view, or anyone who wants to do that. I think it's to expand the amount of energy and resources that humans have to address issues on Earth. So that's the key part for Delta-V for me.
Leo Laporte [00:09:15]:
Yeah. You don't think we're going to be living on Mars in a few years?
Daniel Suarez [00:09:18]:
No, I definitely do not. And anybody who's read Delta-V and the Potlatch chapter will have no doubt my opinion about that. You know, and again, the books that I've written previously are what allow me to write the later books. So for example, my— I had a conversation with Pete Worden, who was the then head of NASA Ames, about the perchlorates on the surface of Mars. And hearing him, I mean, that just opened my eyes about it and sent me in a direction where I realized that that's not a thing we're going to be doing. We're not going to be settling Mars.
Leo Laporte [00:09:52]:
Yeah, before I read A City on Mars, I knew that from Delta-V. I already knew how impossible that was gonna be. And actually, just last week we got news that the surface of Mars is even more deadly than we thought.
Daniel Suarez [00:10:05]:
Yes, it's like a super fun site times 1,000.
Leo Laporte [00:10:08]:
It's getting worse.
Daniel Suarez [00:10:09]:
But here's the thing, I mean, to do science there, to have a scientific station, to have a station up on Phobos or Deimos, You know, all of that stuff, we can do that. But we do have to try to avoid a cult-like behavior where we just believe, you know, just have faith. You know, look at the data, look at the hard science, and I think we'll choose better.
Leo Laporte [00:10:34]:
I don't think anybody who's read your books, and I hope that if you're watching and you haven't, you will. You start with Daemon and just go work your way through them all. But I don't think anybody—
Daniel Suarez [00:10:44]:
Oh, that would be appreciated.
Leo Laporte [00:10:45]:
Yeah. I don't think anybody who's read your books has any sense that you're preaching in any way in any of those books.
Daniel Suarez [00:10:52]:
No.
Leo Laporte [00:10:52]:
And yet—
Daniel Suarez [00:10:53]:
No, all of nature. Yeah.
Leo Laporte [00:10:54]:
Yeah. I do think it's very clear that— well, and certainly from talking to you, it's become very clear that there is a point of view in the sense that we can make this a better world.
Daniel Suarez [00:11:07]:
Absolutely.
Leo Laporte [00:11:07]:
That this doesn't have to be a dystopian future.
Daniel Suarez [00:11:11]:
And we have in the past too. I mean, we've improved the lives of so many people with modern technology.
Leo Laporte [00:11:18]:
Yeah, I think we're better off than we were 100 years ago.
Daniel Suarez [00:11:22]:
Yeah, you know, you're right about that. I mean, I think what happens, right, and this is where reading, uh, books like Evolution of Civilizations— I'm trying to remember the name, his last name is Carroll, can't remember— it's a book written in the '60s and it talks about the arcs of various civilizations. What happens is you develop some new capability that increases your population, which puts more burden on your land, and you're kind of back where you started from. And once you start to have that, let's say, that new technology, or let's say oil, something like that, it starts to falter, the carrying capacity of the land goes down, and you can't sustain your population anymore, and things tend to collapse. That's why trying to create a resilient system that is nonetheless modern is, I think, if we could be the first civilization to truly be sustainable over the long term, right there, that's a major innovation.
Leo Laporte [00:12:19]:
Yeah. We do certainly face challenges, but technology has brought us many gifts, but it's also brought us many, many challenges. I mean, climate change. One of the things, I love AI and I'm very interested in AI, but I also, it's hard not to recognize that climate consequences of all these data centers and all the electricity we're using because AI requires a lot of power.
Daniel Suarez [00:12:46]:
Yeah.
Leo Laporte [00:12:47]:
And so I have this mixed feeling about this.
Daniel Suarez [00:12:50]:
Of course. And the other issue is creating a surveillance society. That's—
Leo Laporte [00:12:56]:
Yes.
Daniel Suarez [00:12:56]:
You know, when you take a look at— and I'm sure you have discussed this ad nauseam in this show, but The idea that we need to have all of these data centers, I mean, for what? To build a society where there's cameras looking everywhere and tracking everybody's movements? I mean, that goes back— and it all sounds very dark, but I think the financing of it will pop, and I don't think that will eventually be an issue. But when you start to take a look at the planned numbers of data centers and you really start to think, why do we need this many? it starts to become alarming. I think that the contract, the social contract that we have with technology, I think really needs to be examined, and we're certainly doing that. But that's the big issue is what do we gain? What do we lose from adopting a technology? That negotiation is part of the ingestion of a new technology. It's something that we've always done. I'm always amazed when I think back. I think it was 1900. I think Tim Wu or Tim Hu.
Daniel Suarez [00:14:01]:
He wrote a book called The Master Switch.
Leo Laporte [00:14:03]:
Oh yeah, Tim Hu. It's a great book.
Daniel Suarez [00:14:05]:
When he talks about how the Polaroid camera, a portable camera where you could take pictures on the street, was a massive invasion of privacy. Oh my God, and people were losing their minds. And it seems incredibly quaint now, but it was freaking people out at the time, which I think is funny.
Leo Laporte [00:14:21]:
Carol Quigley, by the way, The Evolution of Civilization.
Daniel Suarez [00:14:23]:
Carol Quigley, Evolution of Civilization. Really great book. Kind of dry but fascinating. It really explains a lot. I think it's—
Leo Laporte [00:14:30]:
Are you a fan of history? Do you read a lot of history?
Daniel Suarez [00:14:33]:
I read a lot of history.
Leo Laporte [00:14:34]:
Isn't that interesting how history can inform a writer of science fiction of the future?
Daniel Suarez [00:14:39]:
Yeah. What do they say? It's, uh, if history doesn't repeat itself, it at least rhymes.
Leo Laporte [00:14:44]:
Yeah, I love that.
Daniel Suarez [00:14:46]:
And that's so perfect because it doesn't really truly repeat, but it does instruct. And, you know, we have proclivities as a species. So I find it fascinating. And that's the thing. I mean, this all sounds very heavy, but the truth is, again, to go back to the commenter, I'm very optimistic about the future. You know, we— was it Winston Churchill said that— well, he was speaking about America, but America can— will make the right choice once they've made all the other ones. We seem to eventually come around.
Leo Laporte [00:15:22]:
That's reassuring because we have made some poor choices of late. And so that's reassuring that maybe we're on our way to the right choices. Daniel Suarez is our guest. He's written many, many books. You should start at the beginning and work your way. By March, you'll be ready for the next, uh, the next edition of Daniel's, uh, books. We all are looking forward to it. Shadow Zone will be out March 30th of next year, which is sooner than it sounds.
Leo Laporte [00:15:51]:
Next year's coming on rapidly.
Daniel Suarez [00:15:53]:
It's amazing. Time is accelerating. Maybe it's the news cycle that's doing that.
Leo Laporte [00:15:59]:
Yeah, absolutely. So one of the things we were talking about is all these data centers. And I think if you— now, who knows what the reality is? But if you ask Sam Altman or Dario Mode, why are you trying— why are you sinking I think Sam said it's going to cost $1 trillion to create a superintelligence. Why are you doing this? They are— and again, I don't know what they really think, but their stated motivation is to create AGI, a superintelligence, something smarter than any human. And it's their feeling that— in fact, they even, because they're going public in the week, they even are saying, well, we're very close to recursive self-improvement where the AI makes itself smarter. That's the singularity, of course, that Ray Kurzweil and others have talked about, where the intelligence gets to a point where it can create its own next generation. And at that point, it goes exponential because it could do it faster and faster and faster, much more quickly than any human agency could do it. Do you believe that's nutty, or is that something we should be trying to achieve, or is it even achievable?
Daniel Suarez [00:17:16]:
Well, here's the thing. Again, what I'm about to say doesn't come from my personal knowledge. I read a lot of Yann LeCun, Gary Marcus, and many other experts. In reading the experts and understanding the architecture from them of LLMs and transformers, these are models that don't learn the same way we do. They need to go through training. That's how they learn. They don't learn while they're running. and answering while they're doing inference.
Daniel Suarez [00:17:48]:
And so that's why they're constantly having to issue this new model with many more parameters and so on. Now, if they redesign it and figure out a more efficient way to do the parameters or, you know, to handle the energy requirements, that's different. But the models in and of themselves, I don't think are going to improve themselves.
Leo Laporte [00:18:05]:
You don't think they're going to suddenly become this genius?
Daniel Suarez [00:18:09]:
No. I mean, I think we see some discoveries in mathematics, but these are really like arcane, Arcane sets that have been attempted for centuries in some cases, so there's a lot of data to crunch on and they may find something that people didn't notice. But I don't think they're breaking new ground, at least not the way they're currently designed. I think people like Gary Marcus have talked about— well, I think they call it harnesses now, this idea that you're going to put some added functionality on top of it, in this case, a symbolic— image of the world, like what the world is, because I think that's what's lacking from LLMs. LLMs don't know what reality is. They don't know what anything is. This has been a real bone of contention with people, the idea of a stochastic parrot. Sort of LLMs are talking, but they don't know what they're saying.
Daniel Suarez [00:18:58]:
There's no consciousness there. And I think that goes back to AI psychosis and so on. So right now, when I look at LLMs, To my mind, there is no consciousness there. They are a machine. They are a probability machine. They're predicting what the next word should be. And that's why they can get things fabulously wrong sometimes. And we've all seen that.
Daniel Suarez [00:19:22]:
One of my readers sent me this 4-page interview that they had. I'm trying to remember. I think it might have been with Claude. And they said— well, they uploaded a substantial portion of Damon, apparently. Oh, wow. You are Matthew Sobel. I want to interview you. I'm a reporter.
Daniel Suarez [00:19:40]:
And then they started interviewing Claude as if it was Matthew Sobel. And this was very interesting.
Leo Laporte [00:19:46]:
I bet it was.
Daniel Suarez [00:19:47]:
And the first 4 pages of it were like, holy hell, that's just amazing. And then it went back crazy. It just started, you know, like up to a certain point, and then it was just nuts.
Leo Laporte [00:19:58]:
Yeah, that's not right.
Daniel Suarez [00:19:59]:
So I realized it's kind of like a Pollyanna cracker. It just kind of went in that direction.
Leo Laporte [00:20:03]:
Yeah.
Daniel Suarez [00:20:04]:
And, and so it makes you realize And of course, when you realize that the way these LLMs work, that when you're having a conversation, they don't retain a memory of it so much as resubmitting the entire chat stream every single time you do it, which is why they eventually run out of memory.
Leo Laporte [00:20:21]:
Right.
Daniel Suarez [00:20:21]:
So it doesn't work just like us. It is super impressive and immensely useful for certain tasks. I think we've seen recently companies taking much smaller parameter models, like an insurance company, and saying, okay, look. or a law firm, let's just take it on our entire law library and just— we don't need to write sonnets here. We don't need to do paintings. And so, they really pare it down and have a great deal of utility for their industry without so much overhead on energy and water and so on. I think that's the future of LLMs is a lot of special-purpose AIs used all over the place and more locally would be my guess. Again, all of this is coming from me reading widely on the topic, not because I have a PhD in it, certainly I don't, but I am pretty good at sort of being a bee and cross-pollinating and going around and talking to all sorts of people who don't normally talk to each other.
Leo Laporte [00:21:16]:
That's kind of how you do your research.
Daniel Suarez [00:21:18]:
It is. I love it when I introduce 2 researchers who work in completely different fields and they start talking to each other. I love that so much when that happens. they realize that there might be a connection. And again, perhaps that's one of the purposes of what I do. I don't know.
Leo Laporte [00:21:36]:
By the way, I agree with you. One of the reasons I spend so much time, money, and brainpower on local AI is to really understand what they're capable of. And the more I get to know them, the more I completely agree with you that they are incredibly useful in certain specific domains. But I don't see any evidence of them being able to achieve anything like human-scale intelligence, let alone superhuman.
Daniel Suarez [00:22:02]:
Yeah. No seat of consciousness.
Leo Laporte [00:22:04]:
No.
Daniel Suarez [00:22:04]:
Sensorium is the word for that. And I don't see that at all. And as far as singularity, I've always been— I'll be kind by saying extremely skeptical. I think we're far, far away from that, but also good, because I think we have a lot—
Leo Laporte [00:22:19]:
Yeah, good.
Daniel Suarez [00:22:20]:
A lot to do to try to figure out how, again, to retain human agency in the face of this type of thing. Not thrilled about this sudden adoption of superintelligence. I think that's a marketing thing. You know, it's like just—
Leo Laporte [00:22:34]:
It's funny because nobody ever— I say this— nobody ever questioned the artificial part of artificial intelligence. It was the intelligence part people questioned.
Daniel Suarez [00:22:43]:
That's right.
Leo Laporte [00:22:45]:
So they've kind of thrown out the wrong part of the 2 words. I still call it AI. Sorry.
Daniel Suarez [00:22:52]:
No, that's— so do I. And by the way, I meant to mention this earlier. There's this concept of encephalization.
Leo Laporte [00:22:59]:
Encephalization.
Daniel Suarez [00:23:00]:
Encephalization has to do with the size of the brain in a creature in relation to its overall mass. And so highly encephalated species like us, we have very large brains for the size of our body. I think the human brain uses about 25% of our calories. It's very calorically intensive brain tissue. And so whenever you see a brain in a species, Initially, the question was, why is it there in dolphins and in ravens and things like that? And researchers very quickly came to the conclusion that it has to be for a purpose because basically you use it or you lose it.
Leo Laporte [00:23:38]:
Right.
Daniel Suarez [00:23:39]:
Species that don't use intelligence, like for example, why don't hawks have as big a brain as a raven? You would think a hawk would. And I explored this somewhat in Kill Decision, that ravens They essentially try to manipulate predators, top-level predators like wolves and humans, by finding them prey and getting a share of the carcass. And they have to be able to read whether or not it's safe to go down to the carcass, and they need to understand the body language and interactions between these predators to basically manipulate them. And, and I thought all of this was fascinating. Now, This relates to AI in the sense that if we don't use our brains, we're going to lose it. Over time, we will start to lose our brain tissue. I'm convinced of that. And so use it or lose it is very true.
Leo Laporte [00:24:30]:
I think we all have a good example of that. I think this is pretty universal. The more I use GPS, the less I am able to find my own way, right? I have to have GPS now.
Daniel Suarez [00:24:43]:
Yeah, I might be guilty of that. Yeah, but that's a perfect— Have you ever lived in LA when we had Thomas Guides? Do you remember Thomas Guides?
Leo Laporte [00:24:49]:
Yeah. Oh yeah, yeah, the big Thomas Guide. And before you'd head out, you'd look it up, you know, A32 on page 76.
Daniel Suarez [00:24:57]:
Ah, it's right on the sign. Damn it, I can't see it.
Leo Laporte [00:24:59]:
You'd figure out, okay, I'm gonna take a left at Sepulveda and then I'm gonna go— and you'd—
Daniel Suarez [00:25:05]:
We're dating ourselves here.
Leo Laporte [00:25:06]:
You'd write it all down.
Daniel Suarez [00:25:07]:
Yes.
Leo Laporte [00:25:08]:
In fact, I didn't— for years I wouldn't drive a car without a Thomas Guide in the back seat.
Daniel Suarez [00:25:12]:
Right, right in the back seat, right? It's just— But—
Leo Laporte [00:25:15]:
Now I won't drive a car without a GPS in the front dash because I couldn't find— I could still read a map. I mean, I guess I, I guess I could recover that ability, but I noticed that, you know, I'll get to somewhere and I go, I have no idea how I got here.
Daniel Suarez [00:25:29]:
Yeah, that is a risk. And I guess I'm guilty of that too.
Leo Laporte [00:25:32]:
Yeah. So that's— I mean, it's proof of what you just said.
Daniel Suarez [00:25:35]:
Yeah. Yep.
Leo Laporte [00:25:36]:
Yeah. Although, and I mean, I'll ask you this, it is also seems to me the case that what the human mind does, you just in a way kind of just described that. We are built by evolution to be a certain kind of problem solver. So for survival purposes.
Daniel Suarez [00:25:56]:
Yeah.
Leo Laporte [00:25:57]:
I don't know if there's anything— I don't know, what do you think? I don't know if there's anything special about the human mind. In fact, it looks— one of the things that's very interesting to me about how AI works is it looks very similar to that token prediction thing where I say one word and then the next word and the word after that, all based on previous experience, all firing, you know, like an AI. Um, I'm not sure there's anything magical, in other words, happening inside our brain.
Daniel Suarez [00:26:23]:
Well, let me throw this at you then.
Leo Laporte [00:26:25]:
Okay.
Daniel Suarez [00:26:26]:
Uh, I think the magical component is our ability to innovate, our ability to relate unrelated things. So for instance, uh, 2 subjects that have absolutely no connection to each other. Well, let's say a phone and a computer and somebody combines them. I don't think an LLM certainly would do that because I don't think they're made to innovate. But the other thing is energy use. It's incredible. I'm trying to remember. I think it's about 40 watts of energy the human brain uses.
Leo Laporte [00:26:54]:
Right.
Daniel Suarez [00:26:54]:
And it is essentially the most complex thing in the known universe that we did— well, in the known universe that we know of. And
Leo Laporte [00:27:03]:
yet it's so low power.
Daniel Suarez [00:27:05]:
So low power.
Leo Laporte [00:27:06]:
Actually, Geoffrey Hinton, one of the Nobel Prize winners for AI work, said that exactly. He said the human brain, an analog brain, is designed to be very low power. We've only been able to do what we do with AIs by throwing massive amounts of power, 40 watts, we're talking gigawatts of power at this silicon to make it kind of emulate The human brain.
Daniel Suarez [00:27:31]:
Yeah, so I think we are pretty special, actually. Yeah, but maybe there's another aspect.
Leo Laporte [00:27:36]:
I think there's also a certain vanity in saying, oh yes, we're special.
Daniel Suarez [00:27:42]:
But you know, I— especially when you take a look at the typical, uh, robot overlords, we'd better be pretty, uh, arrogant. I think— I, I don't, I don't think we should accept, uh, being shoved aside. But the other thing is that A lot of people talk about neurons in the brain and those connections. Again, part of the research I think was for Influx when I was researching the human brain, that there is an analog component in human brains. I'm trying to remember the names of it. I think they're called astrocytes. It's the type of cell that responds to— oh gosh, I'm going to get the chemicals wrong. But basically, It's a slower moving process.
Daniel Suarez [00:28:25]:
And the way I would describe it best probably is there's a whole history of Einstein's brain. Do you remember this? It being stolen? Yes. Yeah. And going on the road.
Leo Laporte [00:28:35]:
Yes. Yes. Back of a Volkswagen. Yes.
Daniel Suarez [00:28:37]:
It's a very improbable story,
Leo Laporte [00:28:39]:
but
Daniel Suarez [00:28:39]:
nonetheless true. But one of the things that was found was when they finally examined it is that the number of neurons, they were thinking he's going to have an exceptional brain. And it really, the number of neurons were not all that beyond the normal, but the number of these other cells was vastly increased. And so there's an aspect of the human brain that's more analog and chemical that works in conjunction with neurons. And so there's a lot we don't understand and haven't yet been able to connect. And again, that just makes me super fascinated about the natural world. But I think that might be part of it. And then there's, of course, subatomic microtubules and all of these other things.
Daniel Suarez [00:29:20]:
I think we're basically like toddlers poking around.
Leo Laporte [00:29:23]:
We don't know what's going on.
Daniel Suarez [00:29:24]:
And that's exciting though, because I mean, it's already incredibly cool and, and we have so much left to explore. That's exciting.
Leo Laporte [00:29:33]:
So that was— it's great to talk to Daniel Suarez. His book again comes out in the spring. As you heard, he promised he will come back in the spring when we can talk about Shadow Zone. Uh, it was interesting, I should have pointed out we did the interview on Thursday, the day before the FAA ground halt. But as you heard, Daniel's kind of, again, he kind of sees in the future. He knew that, you know, many crises were on the way. They were on the way.
Jeff Jarvis [00:30:00]:
Can he see the idea that a plumber is going to be in charge of all of our national security?
Leo Laporte [00:30:06]:
Is that the case?
Jeff Jarvis [00:30:07]:
Yeah. Markwayne Mullin has a, he has an associate's degree in plumbing.
Leo Laporte [00:30:12]:
That's his degree? Well, at least he'll have a good career. It's pipes, man.
Jeff Jarvis [00:30:17]:
You guys put pipes together. Yeah.
Leo Laporte [00:30:19]:
Geez Louise. All right. We will have more of Intelligent Machines. There's actually a lot to talk about. As we were doing the show last week, Jensen Huang was on the chopping block. And shortly after the show, or maybe it was during the show, I don't know. I remember you—
Jeff Jarvis [00:30:35]:
No, it was during the show. We got the
Paris Martineau [00:30:38]:
word that the board was meeting to discuss him being ousted. And then the next day, he resigned. But I do, I really want to talk about all the different kind of mega speed and, uh, Nvidia smuggling stuff that has been percolating.
Jeff Jarvis [00:30:54]:
Yeah, yeah, I also want to talk— yeah, I want to talk about the politics of this because they're, you know, getting more interesting. I have new theories.
Leo Laporte [00:31:00]:
All right, so Jensen is out. He did tweet a couple of days ago, say, hey, if you want me back, I'm, I'm available. To which yesterday in the Nvidia earnings call, the CFO was happy to say no, we're not— he's not, he's not, uh, on the, on agenda for bringing back the boss, even though the employees petitioned, I think thousands of signatures, the employees would like him back.
Jeff Jarvis [00:31:26]:
See, where is that number?
Leo Laporte [00:31:30]:
Wired had the story. Let me pull it back. Bring back the boss. 16—
Jeff Jarvis [00:31:34]:
So standwithjensen.com, and the number is now— total signatures are now 201,000. Oh, what? 10,000 current Nvidians. Oh, 1,704 former Nvidians, 104,000 shareholders.
Leo Laporte [00:31:52]:
Uh, that's— that may be the most interesting because of course the reason the board had to fire him, and what I suspect happened is they called him in, they said, Jensen, you have 2 ways to go with this. We fire you—
Jeff Jarvis [00:32:04]:
Easy way or the hard way.
Leo Laporte [00:32:06]:
You resign. Yeah. And, uh, or, or it went like, Jensen, you're fired. And he said, you can't fire me, I quit. Either way, He resigned before he was fired. But my guess is they didn't have much choice. I said last week they're going to fire him.
Jeff Jarvis [00:32:22]:
So I appeared on Open to Debate and had a debate about— well, they wanted a debate. They want 2 fights about this.
Leo Laporte [00:32:29]:
Which side were you on?
Jeff Jarvis [00:32:30]:
I was on the— maybe it was a mistake. Not that I really believe that. I make clear that—
Leo Laporte [00:32:36]:
Maybe it was a mistake to fire him?
Jeff Jarvis [00:32:38]:
No, wait, wait, wait. It was done so hastily. The speed. Oh, that's clearly sleeping with an executive from a major customer who also happens to be Chinese with questions about her agentry, though that's now a big question whether there's anything there. That's indiscreet. Sleeping with 2 employees is a misuse of male power. All that stands. Yeah, I mean, we do learn from—
Leo Laporte [00:33:00]:
the McDonald's CEO lost his job. Intel.
Jeff Jarvis [00:33:03]:
Exactly. I'm not arguing with any of that. But by doing it so quickly, we lost the chance, I think, to— for the board to examine more carefully what happened, what didn't happen, the politics behind this, and what the impact is on the company and the industry. So I think they could have said, we have to do an investigation. We have to know everything so we know how bad this is. We have to know what's what. And they would have learned more. We would have more transparency on this case.
Leo Laporte [00:33:28]:
They did launch an investigation.
Paris Martineau [00:33:29]:
Do you think that they would have had more transparency if Jensen was still in charge as CEO? I don't think so. I think that he would have been able to clean up the mess much faster.
Jeff Jarvis [00:33:41]:
Yeah, I'm not— I'm not— but I think—
Paris Martineau [00:33:43]:
I mean, I think the thing, the thing that I wanted to talk about this week, which I feel like we, uh, skipped over a little bit last week just in the hubbub around the various— So much. Is the— I think that that is one of the least important aspects of what we've learned.
Leo Laporte [00:33:59]:
I said that last week. I mean, you're right. I don't want to focus on the salacious details. That's not the issue.
Paris Martineau [00:34:04]:
I totally agree. And I think the The thing that is genuinely salacious and outrageous in a way that we should pay attention to is the chip smuggling allegations and the very serious allegations that have been made that Nvidia has basically— Nvidia and a lot of Nvidia executives have been involved in systematically distributing chips to China and circumventing export bans in, in a very calculated way.
Leo Laporte [00:34:36]:
Here's another— let me, let me just say— hold on, then, then I'll say a number, then you disagree. Here's a number that tells you something. The Nvidia chip sales to Singapore, which by the way is a tiny little island, doesn't have a lot of data centers as far as I know, were up 81% year over year. Nvidia says 21— or I'm sorry, 22% of their revenue came from Singapore. That is a red freaking flag.
Jeff Jarvis [00:35:06]:
Red, so to speak.
Leo Laporte [00:35:07]:
Yes. I mean, where are those chips going? They're not going— they're going to China. We know that.
Jeff Jarvis [00:35:14]:
So I'm not going to disagree with any of that, but I, I do think it's also important to look at the political angle here. And it finally hit me today, the original New York Times story which really got this rolling full speed— it had been around for a while, but really got— was bylined by Maggie Haberman and others, which means that it was a White House leak. And I was trying to figure that out. Why would the White House leak something that makes them look bad? Well, finally it occurs to me that it was a Besant, or maybe Besant and Rubio leak to the Times to get at Howard Lutnick.
Leo Laporte [00:35:44]:
Can I just point out what's really obvious at the White House since the election? The president has dialed out. He's worried about his legacy and the buildings. Marco Rubio has stepped forward as a candidate. J.D. Vance stepped forward as a candidate in 2028. Scott Bessent and Howard Lutnick are at odds.
Jeff Jarvis [00:36:02]:
Hate each other.
Leo Laporte [00:36:04]:
Scott Bessent, Treasury Secretary, said out loud when, when queried about this, well, the guy with the pen, the guy who's supposed to prevent the sales of chips to China, is Howard Lutnick, Secretary of Commerce. Now, that's legally true. It's the Commerce Department that decides that. But he's basically saying this is Howard's fault. So this is a White House that is split asunder by its ambition.
Jeff Jarvis [00:36:28]:
And we also have Hines, who's going to be the head of the Subcommittee on Intelligence, saying that there's— that he's seen no evidence so far. Last I saw him say this, that Alice Wong is a spy or that Jensen gave up stuff. So that's one question. The other question is the chips. But then again, Lutnick has a different view and a view that does agree with Jensen Wong, and it is in their personal self-interest stipulated, Your Honor. But it's a view that I've agreed with that by cutting off China, by demonizing China, we end up putting them in a position where they're competing more. So it could be also a sub rosa political effort to say this is a bad mistake strategically, and given this White House, they can do whatever they want. So Besant and Besant, and I didn't know it was pronounced that way, and and or possibly Rubio.
Jeff Jarvis [00:37:17]:
used the Times to go after, and it worked. But I can't parse all that yet. This is what I'm saying about— I
Leo Laporte [00:37:23]:
even said this last week that it was clearly Rubio's folks who leaked that we warned the president on the way to Alaska to pick up Jensen that he was sleeping with a spy.
Jeff Jarvis [00:37:34]:
But I think it was also— is it Bassett? You say it's Bassett?
Leo Laporte [00:37:37]:
I don't know how you say
Jeff Jarvis [00:37:38]:
it. Oh,
Leo Laporte [00:37:39]:
all right. And I honestly, I don't care because I'm not going to remember his name after 2028. At all, unless I run into him at the Home Depot where he'll be selling me paint.
Jeff Jarvis [00:37:50]:
So I don't agree with you, don't disagree with you, Paris, about the specifics here, but I think that there is a bigger, more nuanced narrative going on that doesn't have a clear good guy, bad guy in this. There's a lot of bad guys, a lot of stupid things, people doing stupid things.
Paris Martineau [00:38:04]:
There's a lot of bad guys, but I think that one thing that is important to just remember in all of this, which I think a lot of the tech boys complaining online about Jensen's ouster forget is that there is evidence, and at the very least clear reporting, suggesting that Jensen was at the center of orchestrating a cover-up of his company and employees circumventing the law.
Jeff Jarvis [00:38:34]:
Perhaps with the blessing, perhaps with the blessing of the Commerce Secretary.
Leo Laporte [00:38:38]:
What exactly is the law?
Jeff Jarvis [00:38:40]:
That's what— that's the question.
Paris Martineau [00:38:42]:
The law prohibits the export of, uh, AI chips and Nvidia's chips to—
Leo Laporte [00:38:49]:
No, but they were exempt.
Jeff Jarvis [00:38:50]:
Why are they trying to pass a new law that says that? That's, that's what's going on now is a fight to pass the new law. It is a gray area. And
Paris Martineau [00:38:58]:
no, the fight right now is a— I believe is about, um, the remote access, uh, policy, which was basically a loophole in this.
Jeff Jarvis [00:39:08]:
What she did.
Leo Laporte [00:39:09]:
Yeah. So you could sell chips to, for instance, Hong Kong, Singapore, Singapore data centers that Chinese entities were able to use. Well,
Jeff Jarvis [00:39:17]:
then the Chinese entities could use them if they, if they use something from Estonia. You know, it's, it's where we live in a globalized world. But the other question I have—
Leo Laporte [00:39:27]:
lift the chip ban. I mean, am I wrong to remember that?
Jeff Jarvis [00:39:30]:
On certain chips, yes, but not on the
Paris Martineau [00:39:32]:
most advanced chips. Like, I'm looking at it.
Leo Laporte [00:39:38]:
After halting chip exports to China in April of 2025, the Trump administration quietly reversed its policy 3 months later, assuring Nvidia it can resume shipments of its H20 processors. Now, those are limited processors designed for export to China. So the higher end, the higher end processors, I think Yeah, the ones that, um, have
Paris Martineau [00:40:03]:
for sure been— So Bloomberg got a hold of documents that show that, um, uh, some of the hardware, uh, procured— procured by Alice Wong, uh, through Megaspeed to China include Nvidia's H800, H100, and H200 chips, which are restricted from sale to China and were installed at the company's Shanghai data center.
Leo Laporte [00:40:29]:
But wait a minute. After continued campaigning from Jensen Huang and David Sachs, the Trump administration signed off on allowing H200 shipments to be sold in China.
Jeff Jarvis [00:40:43]:
Right, right.
Leo Laporte [00:40:44]:
So this is why it's very confusing. And that's— yeah, that was a year ago, but that's why it's very confusing. And that's why I think when Scott Bessent says, well, Howard Lutnick had the pen, and did not block this, he's blaming—
Jeff Jarvis [00:41:00]:
Disingenuous. Right. And, and so it's a hall of mirrors we've stepped into here. And I think that once the affair became well known, then that gave Rubio-Bessant the opportunity to go to Maggie Haberman and leak to the damage of both Jensen Huang and Ludnick. And they're all bad guys. They're all doing stupid things. But I'm arguing that there's, there's a little nuance here that, that is not as simple as this one person was in charge of everything. Confused.
Jeff Jarvis [00:41:34]:
Screwed up. Yeah.
Leo Laporte [00:41:35]:
So, and the other thing I would say is you and I, Jeff, both agree. I don't know where Paris stands on this, but that these chips should be sold to China. I actually agree. I agree. That's in Huang's assertion that—
Jeff Jarvis [00:41:45]:
Same here. And Paris, what do you say about that?
Paris Martineau [00:41:49]:
I've been thinking about this a lot this week. I just, I don't think I understand the reason why they were originally banned. And I mean that in a genuine sense. I assume that there had to be some sort of reason given for that policy.
Leo Laporte [00:42:05]:
Yeah, I'll give you the scenario.
Jeff Jarvis [00:42:06]:
Do you want the scenario? It's general China hawkishness too.
Leo Laporte [00:42:09]:
Well, but okay, so here's the scenario. The real fear is that China will have AI as good as our AI, which they will use, weaponize to use against us militarily. Right. That was the real concern is military. They always raised that issue. Oh, the military will have access to these chips. The military will use these chips. And this comes from people like David Karp— I'm sorry, Alex Karp at Palantir arguing that we need to keep our technology for ourselves to defend freedom.
Leo Laporte [00:42:43]:
And so it is hawkishness. It is the fear of China. I say chiefly militarily, not economically. My point of view— I understand that point of view and I don't deprecate it, but my— but I agree with Jensen who said, A, if you don't give them the chips, they're just going to be incented to develop their own.
Jeff Jarvis [00:43:03]:
More competitive.
Leo Laporte [00:43:03]:
They're going to make their own. Which Huawei and others have attempted to do. There's also been a— the other problem is that we're not, we're not the only source of technology. For instance, the EUV machines are made in the Netherlands. There's only one company in the Netherlands that makes these. Now, they aren't selling them to China. China has been reverse engineering it. In fact, we saw a story last year, maybe just a few months ago, that China had successfully reverse engineered the EUV machine, not put it into production, but reverse engineered it.
Leo Laporte [00:43:32]:
So that's, that's a embargo from the Netherlands, which could be at any time changed. The lack of EUV is the biggest reason that China doesn't have competing chips, I think, in my opinion.
Jeff Jarvis [00:43:45]:
And which is a Taiwan issue too.
Leo Laporte [00:43:48]:
Then, well, we have TSMC uses those EUV machines from the Netherlands, but it's only ASML that can make these machines.
Jeff Jarvis [00:43:55]:
Here's another question of mine. So we deal with a lot of shady characters internationally. Saudi Arabia, Russia. Before there were reforms—
Leo Laporte [00:44:06]:
We talked about this last year where they're giving chips to the Middle East. Right, exactly.
Jeff Jarvis [00:44:11]:
So is China really— A, I want their open weight models, and B, I want their damn cars. Is China really the full threat? Ted Lieu will come out and say we shouldn't sell anything to China. That's just stupid. Nothing, nothing, nothing, nothing, period. I think that's maybe in a globalized world, both naive and cutting off our own nose to spite our face. Is China the great threat that they're putting— I don't like them. They're authoritarians. We were pretty much authoritarian a few weeks ago.
Jeff Jarvis [00:44:37]:
I'm not defending China at all. But are they the great threat that they're being presented as by the hawks?
Leo Laporte [00:44:45]:
I'm a free marketer. So— and by the way, there's a good article in The New Yorker a few weeks ago about how microprocessors are a global effort. They are not an American effort. Mm-hmm. They require cooperation from Taiwan, from the Netherlands, from all over the world to make these things. This technology is not unique to the United States, and the idea that we could somehow seal the borders off and only have this technology for ourselves is not only a dream, an imagination that can't come true, but also probably counterproductive. And I've always been of the opinion, and maybe this is me being just Pollyanna, I am an old hippie, but that the best way to prevent war, the best way to keep nations, whether we agree or disagree about politics, from fighting each other is to have us be interdependent economically. And I'm a— that's why I'm a big believer in free trade.
Leo Laporte [00:45:42]:
So I, I, I— so, but that's a— so I understand, yes, China could use AI and become militarily dangerous. Well, how do you prevent that? Well, there's one way is to try to embargo it. The other way is to make them so economically intertwined with our success that they don't want to attack us, that they do better by not attacking us. And I think that's actually where we are today. Yes. I think it's foolish to try to go the other direction.
Paris Martineau [00:46:11]:
I don't disagree with any of these takes. I mean, I I'm not going to pretend that I have figured it all out as far as trade policy goes, but a question that's kind of somewhat tangential to this, but I think is equally as important, is Nvidia's behavior in this. If these allegations are true regarding the kind of large-scale continued diversion of some of the best chips they produce consistently to China, if those allegations are true, it also just raises the question of Nvidia's also been saying they've been a core part of building up this AI boom because they've been saying repeatedly, there's such a great demand for our products, we're stretched thin. You know, there's— we can't even keep up with the demand. It's so intense. And if that is to be believed, how can you both have, like, unprecedented demand, ostensibly from legitimate, above-the-board clients? and be diverting a significant amount of your chip supply illegally to China? Like, why wouldn't you just choose doing the above-board option and making sure that your products go to customers that you can vet and do some sort of due diligence into? I just think that that's—
Leo Laporte [00:47:29]:
You're saying why did Jensen skirt the line on—
Paris Martineau [00:47:34]:
It's just like, if everybody and their mother wants your chips and people are begging for your product and people are paying top dollar for it, then it is really easy to— well, China is a big market above. But
Leo Laporte [00:47:48]:
what if you could double your sales?
Paris Martineau [00:47:50]:
Paris, are you making a bubble argument? Doubling sales is not a question of— the sales are constrained by the supply. It's not like they are looking for extra supply, like extra people to buy the chips. Or are they?
Leo Laporte [00:48:06]:
Nvidia said we're selling as many chips as we can make, but that included all the chips they were sending to Malaysia, which is the number 2 data center capital of the world after Virginia, according to the Wall Street Journal. And who are the customers for those Malaysia data centers? China. So now, by the way, that I don't think was illegal. I don't know. I mean, no, I mean, reality is not—
Jeff Jarvis [00:48:30]:
no, that's why they're trying to pass that new law.
Paris Martineau [00:48:31]:
That's why they're trying to pass this new law.
Leo Laporte [00:48:33]:
I understand why. If you're, if you're a businessman, A, who believes, as I do, that free trade is a good thing, that China should have access to these chips, that's the best for the world and the best for Nvidia, and it's going to double your revenue. I don't think that they— I don't think they were in a situation where, oh, we could just say let's not sell them to China and we'll still sell every chip we can make. I think that we sell every chip we can make is based it included the chips they were sending to Malaysia and other Chinese pass-throughs. So, and by the way, another accusation, and maybe we'll learn more about this, is that this was a false constrained supply to raise prices. I firmly believe that's with the RAM companies. Yeah.
Jeff Jarvis [00:49:22]:
RAM, yes. You think there was a case with— I'll give you an example.
Paris Martineau [00:49:24]:
I mean, that's kind of what I'm suggesting is it's It seems very convenient that all of this is happening when you have a company saying like, we've got unprecedented demand for our products, we simply have to raise prices because just everybody wants our stuff. And it's like, well, yeah.
Leo Laporte [00:49:39]:
It's a win-win. It's a win all around, Ferris. It's
Paris Martineau [00:49:42]:
a win-win all around, except for if, you know, your understanding of how you play the game—
Leo Laporte [00:49:47]:
Where do you want them to not sell? I mean, look—
Paris Martineau [00:49:49]:
To the places where they're prohibited by law from selling to is where I would say.
Leo Laporte [00:49:53]:
And especially if you give a little money to the president, it's even a grayer area. Yeah, I think Jensen was doing everything he could to sell as many chips as he could to every single customer who wanted them, including Saudi.
Jeff Jarvis [00:50:04]:
And he was open about saying that, that we should sell to China, whether they did or not. Paris, are you making also a kind of a bubble argument here that he was making markets where they didn't exist?
Paris Martineau [00:50:16]:
I mean, I don't know if it goes so much to say creating markets like whole cloth. that didn't exist. But I think certainly these allegations, uh, make it more like increase the size of the bubble.
Jeff Jarvis [00:50:30]:
Yep.
Leo Laporte [00:50:31]:
Nvidia reported $12.4 billion of revenue in China through July of this year, 7% of its global revenue. Legal.
Jeff Jarvis [00:50:41]:
So let me ask you 2 questions.
Leo Laporte [00:50:42]:
I can't see why they would turn their back on that, especially if they think they can get away with it. Now,
Jeff Jarvis [00:50:47]:
Yeah.
Leo Laporte [00:50:48]:
Did Jensen walk a very fine line? Yes.
Jeff Jarvis [00:50:51]:
In many ways.
Leo Laporte [00:50:53]:
Yeah, yeah, yeah. In many ways.
Jeff Jarvis [00:50:54]:
Legal and libidinal.
Leo Laporte [00:50:55]:
I also think— so I was going to say, I own 2 Sparks, which when I bought them were $4,600 each. They are now about $7,000, $8,000 each. Is there really a constraint on Sparks? I don't think so. I think Nvidia can make as many as it wanted to. I think that that is an artificial constraint.
Jeff Jarvis [00:51:15]:
Well, but you have other places like Elon Musk and companies saying we're going to buy every, every black data centers
Leo Laporte [00:51:21]:
that don't exist yet.
Jeff Jarvis [00:51:23]:
Exactly, exactly. That's the bubble. That's, that's where it exists.
Leo Laporte [00:51:28]:
Honestly, I think it's very hard for us and for anybody outside of these companies to know what the hell's going on.
Jeff Jarvis [00:51:33]:
We just don't know nearly enough. That's my, my other point. So Nvidia stock, just the heck of it, is down 3% today. It's been a bit up and down. The market reacted. Obviously, it was a great quarter, huge numbers, but that was the end of Jensen. The market's not sure what's going to happen now. So I have 2 questions.
Jeff Jarvis [00:51:53]:
The first is, what do you think the fate of Nvidia is ex-Jensen? And the second question is, Jensen cools his heels for a while. He's generally recognized everywhere as one of the smartest people in AI and technology. What can or does he do next?
Leo Laporte [00:52:10]:
A, Nvidia continues, if only on fumes, for years. They've got proprietary technology. They've got the same engineers who designed the chips yesterday are still designing them today. And Jensen was at the top. But I'm not a big— as I mentioned last week, a big believer in the great man theory. And I said—
Jeff Jarvis [00:52:30]:
He did have a vision for this company. He had a vision.
Leo Laporte [00:52:32]:
He built it. Look, clearly he built it and he picked the right, every step of the way, picked the right path without any evidence. You know, he knew where we were going. So maybe in 10 years Nvidia will not be well positioned, but for the next 3—
Jeff Jarvis [00:52:46]:
Intel of the next decade.
Leo Laporte [00:52:47]:
Yeah. But for the next 3 or 4 years, they could just roll on inertia, do great.
Paris Martineau [00:52:52]:
And yeah, isn't if every, all the AI boys are right, in 3 to 4 years we won't have a society anymore? So that's all they
Leo Laporte [00:52:59]:
need. And then what was your second question?
Jeff Jarvis [00:53:04]:
Jensen. What— what— he's— he is a smart guy. He knows his stuff.
Leo Laporte [00:53:07]:
Somebody's gonna hire him, don't you think?
Jeff Jarvis [00:53:09]:
Well, does he? But is he— can you hire a Jensen Huang? Or does he start another— does he just start another— does he get venture and start another— a new venture?
Leo Laporte [00:53:15]:
He probably becomes a venture capitalist.
Jeff Jarvis [00:53:16]:
Does he start— does he start an initial— a more— Even if he—
Leo Laporte [00:53:20]:
even if he and his wife split and they split the fortune, which they should because she really was very much 50/50 part of this. He gets $70 billion. He's worth $140 billion. He can be a pretty successful venture capitalist if he has that vision, if he knows where we're going. The best place for him to make more money is investing in the company. This is what he was doing at Nvidia, investing in the companies that are— True. In fact, this is where he'll be most missed, is that he put money into these new nascent companies, open weight companies in many cases, Yep. And was in fact a venture capitalist for these companies.
Leo Laporte [00:53:57]:
That's very important. I hope he does continue to do that. And God knows, if you're a venture capitalist, you could sleep with anybody.
Jeff Jarvis [00:54:03]:
Oh, yeah. Yeah. The devil you already have. Yeah.
Leo Laporte [00:54:08]:
So,
Jeff Jarvis [00:54:08]:
Harris, what do you think?
Leo Laporte [00:54:09]:
I suspect that's what he's going to do.
Jeff Jarvis [00:54:10]:
If you were advising Jensen Huang—
Paris Martineau [00:54:16]:
I mean, if I was advising Jensen Huang, I'd be like, bro, go on gardening leave. Go buy an island. I mean, yeah, if I personally was advising any of these people, I'd be like, you're richer than God. Calm down.
Leo Laporte [00:54:31]:
But that's what you and I would do, of course. This is why we are not the founders of trillion-dollar companies.
Paris Martineau [00:54:37]:
Yeah, because we're not psychopaths.
Leo Laporte [00:54:39]:
Yeah, these guys live for this kind of winning. I was talking to my daughter yesterday. She says, you know what these guys are? They're hoarders. Yeah, yeah, absolutely.
Jeff Jarvis [00:54:50]:
And what they're hoarding is power of reputation, even though they don't realize they have bad reputations.
Leo Laporte [00:54:57]:
They don't care. They are hoarding money. If they could, like Scrooge McDuck, have a, a swimming pool filled with dollars that they would dive in every night, they would. They're hoarders because no one after— I don't know, make up a number— after $50 million Nothing you make is going to change your life. Okay, make it $100 million, make it a billion. None of these guys need more money. So whatever drives Jensen Huang is not going to be go to garden on an island. It's going to be master of the universe stuff.
Leo Laporte [00:55:33]:
What about VC? He might do a startup. I, you know, he's a little
Jeff Jarvis [00:55:37]:
older.
Leo Laporte [00:55:38]:
I think he's too old to do a startup. That's pretty grueling.
Paris Martineau [00:55:42]:
I mean, I think he's going to be doing something as close to Nvidia as possible.
Leo Laporte [00:55:46]:
Like, maybe.
Jeff Jarvis [00:55:47]:
Well, but Nvidia was many things. Nvidia was chips. Nvidia was an operating system. Nvidia is robots. Nvidia is, uh, hosting. That's the, that's the vision of Nvidia is it was, it was, it was everything, right?
Paris Martineau [00:56:01]:
It really was.
Leo Laporte [00:56:04]:
MegaSpeed is defunct, I believe.
Paris Martineau [00:56:06]:
He could bring it back.
Jeff Jarvis [00:56:08]:
Is it? Or she's gone?
Leo Laporte [00:56:09]:
Nah, I think MegaSpeed itself is not. I don't know. I don't know. Um, let's take a break. I predict— my prediction— let's get your predictions. My prediction: Jensen Huang will emerge within a year with a venture capital fund of multi-billion dollars. If I— if you were investing in somebody who could predict the future, and that's Yeah, and, uh, I think he's the guy. So I think that's where he's gonna go next.
Leo Laporte [00:56:39]:
He's— you know what's sad is I think he's going through the stages of grief, like denial. I didn't do anything. Uh, you know, uh, anger.
Paris Martineau [00:56:51]:
I mean, how many lies do you think he told his board?
Leo Laporte [00:56:54]:
That's why they had to fire him.
Paris Martineau [00:56:56]:
Couple of days.
Leo Laporte [00:56:57]:
Yeah, he lied to them like crazy. We know he did. He would lie to them all the time.
Paris Martineau [00:57:00]:
I mean, I'm curious how many lies his general counsel also told the board, because it seems like— I'm forgetting that guy's name, but he was very involved in cleaning up Jensen Huang's messes.
Leo Laporte [00:57:11]:
I hope he doesn't hang his hopes on going back to Nvidia, because I think that would be a mistake. And there are ways he could manipulate this to maybe get back there, but I don't—
Jeff Jarvis [00:57:20]:
Are there a bunch of key people who are so loyal to him who would follow him somewhere and devalue Nvidia? Yeah, that also devalues his wealth. But that's—
Paris Martineau [00:57:30]:
I mean, yeah, all of his 60 direct reports probably, and his children.
Jeff Jarvis [00:57:35]:
Yep, yep. I think that's a real— and, and, and all of the people who were saying he should go back, they'll all invest in it.
Paris Martineau [00:57:41]:
That's what I'm saying, is I think that he's going to come back within the next couple of months at the most, probably sooner than that. Maybe once this, uh, congressional investigation slows down, at least he's going to come back with something as close to Nvidia as you can reasonably put together without having access to the physical stuff that—
Jeff Jarvis [00:58:05]:
Yeah, it may not need to make chips at all.
Paris Martineau [00:58:07]:
Like, I mean, I think he might just be a broker of all the deals that he was brokering. He might represent NeoCloud somehow, you know, like he— I, I think that he's gonna be— it's gonna be a Jensen Huang-centric affair.
Jeff Jarvis [00:58:22]:
Yeah.
Paris Martineau [00:58:22]:
And it's going to make NVIDIA a lot weaker because I think he's going to take a lot,
Jeff Jarvis [00:58:25]:
which is a risk too.
Leo Laporte [00:58:27]:
That is a risk. Yeah. I hadn't thought about that.
Jeff Jarvis [00:58:30]:
I mean, could he also— what kind of company could he build around open weights purely?
Paris Martineau [00:58:35]:
I don't think he's going to do that.
Jeff Jarvis [00:58:37]:
Well, except if you look at Alex Karp, that's, that's the entire strategy at Palantir. I think there's a larger strategy that says, hey, you guys at Anthropic and OpenAI who hate me, who want me to die? Okay, fine. Like China, I'm going to undercut you. I'm going to compete with you and undercut you. And you can't— I'm the Craigslist of AI where you can't compete with free.
Paris Martineau [00:59:03]:
I don't think—
Jeff Jarvis [00:59:04]:
Craig did it for good reasons. Jensen may not, but I think there's a revenge there because they are the ones who also want to get rid of him.
Paris Martineau [00:59:13]:
And they're the ones who are angry. Maybe there's an argument to be made, but I think that I think that if that does happen, it will be a stepping stone to be doing some sort of for-profit.
Jeff Jarvis [00:59:24]:
Yeah, yeah, it's for-profit. It's absolutely for-profit.
Paris Martineau [00:59:27]:
I mean, I mean, commercialized, closed sort of system eventually. Like, I just don't think that he's going to be—
Jeff Jarvis [00:59:36]:
I think there's all in on open source. I also think that come the, the IPO we'll talk about after a break, I think the hatred against the duopoly of AI is going to grow a lot worse. And with the Democrats in power, Anthropic and OpenAI are going to be— have nuclear cooties.
Leo Laporte [01:00:03]:
Well, we are very much in uncertain times, aren't we?
Paris Martineau [01:00:06]:
Oh, it's what an interesting What an interesting world that, uh, here— would any of us have guessed that we'd be here a day before Anthropic goes public and we would have not spoken about it once?
Leo Laporte [01:00:19]:
Well, we will. We just—
Paris Martineau [01:00:20]:
I know we will, but just like— It was not the top story of the day. It's— what an interesting world. They just, they just valued it. You know,
Leo Laporte [01:00:26]:
what are they, what are they, what are they going for?
Jeff Jarvis [01:00:28]:
$1.5 trillion.
Leo Laporte [01:00:30]:
Okay, that's even more than I'd heard.
Jeff Jarvis [01:00:32]:
Yes.
Leo Laporte [01:00:32]:
Yeah. Holy cow.
Jeff Jarvis [01:00:33]:
That's the, that's the draw.
Leo Laporte [01:00:34]:
How much a share is that?
Jeff Jarvis [01:00:35]:
$83, is it?
Leo Laporte [01:00:36]:
Well, that's what we were thinking. We were thinking $1.4 trillion, $80.
Paris Martineau [01:00:40]:
I was gonna say, yeah, there was—
Jeff Jarvis [01:00:42]:
$83.
Leo Laporte [01:00:43]:
$83. $83.
Jeff Jarvis [01:00:44]:
Yep. Well,
Leo Laporte [01:00:46]:
well,
Jeff Jarvis [01:00:48]:
I mean, it's oversubscribed, obviously.
Leo Laporte [01:00:49]:
Look, I'm not putting any money into it ever, but, uh, well, I'll tell you why. Their timing is superb when we come back. Uh, you're listening to Intelligent Machines. Jeff Jeff Jarvis, Paris Martineau. So glad you are here. Ah, okay, back to the Sturm und Drang, as Jeff would say. Uh, it's been a crazy week. Um, outages galore.
Leo Laporte [01:01:17]:
The FAA on Friday, the airport, the air, uh, uh, shutdown. Uh, the NSA yesterday.
Paris Martineau [01:01:24]:
The air shutdown.
Leo Laporte [01:01:25]:
The air— what do you call it?
Paris Martineau [01:01:27]:
We don't have air anymore.
Leo Laporte [01:01:29]:
No way. The air has been shut down. All the planes fell out of the sky. Actually, should give credit to the FAA. They did the right thing. Nobody died. No planes crashed. But it'd be nice if they—
Jeff Jarvis [01:01:40]:
if we— it wasn't putting in one new router. Didn't— if that brought—
Leo Laporte [01:01:43]:
no, no, they're not being fully honest. But here's the thing. There was also a Microsoft M365 24-hour outage globally next day, or with— No, prior day, right? Prior day. NSA has been down. Yesterday, Apple went down. So there is a theory—
Jeff Jarvis [01:01:59]:
United Airlines, Marriott, various banks.
Leo Laporte [01:02:04]:
There is a theory, and I think I fully subscribe to this, that what's happening is people are freaking out because of AI discovering flaws. They are rushing to patch stuff. And when you rush, you make mistakes.
Jeff Jarvis [01:02:20]:
Dave Weiner said that uh, some other people we trust say— I think it's speculating the same thing.
Leo Laporte [01:02:26]:
I mean, we don't know. Here's the thing, we don't really know because nobody's being completely forthright. Microsoft said it was a software update.
Jeff Jarvis [01:02:32]:
Well, well, maybe— why were they updating this?
Leo Laporte [01:02:34]:
Yeah, they're not saying—
Paris Martineau [01:02:36]:
The Washington Sun had some interesting reporting on this this week.
Leo Laporte [01:02:38]:
I read that article.
Paris Martineau [01:02:39]:
Not with all of these, but that, um, this is Jeff Stein at the Washington Sun reporting that once we accepted it—
Leo Laporte [01:02:45]:
Who is the Washington Sun, by the way? I've never seen that.
Paris Martineau [01:02:47]:
That used to be not Newsroom of the US, uh, Notus, Notus. I never knew how to pronounce it, but they recently rebranded. Um, it's like a group of people, like former Russian Post people.
Jeff Jarvis [01:03:00]:
Yeah, the Allbritain started kind of a journalism school as a publication.
Leo Laporte [01:03:04]:
So this is credible. This is not—
Paris Martineau [01:03:05]:
this is a credible outlet. This is where most of the reporters that got laid off from the Washington Post in that big layoff, they now work at the Washington
Leo Laporte [01:03:14]:
Post. When I read this article, I thought, this is good. So this is—
Paris Martineau [01:03:17]:
So this is by Jeff Stein, who's great. And he wrote, Wall Street executives and the NSA have told members of Congress that this week's widespread tech outages were caused by attempts to fix vulnerabilities identified by AI, according to 3 sources familiar with the matter. They said, while it does not appear that these disruptions were caused by a cyberattack, government and banking officials have told lawmakers they stem from attempts to fix gaps in cyberdefences identified by AI models. The fixes, however, temporarily overwhelm the systems.
Leo Laporte [01:03:48]:
This makes sense for every single one we've heard about, right?
Paris Martineau [01:03:51]:
Uh, on congressional—
Leo Laporte [01:03:53]:
You could cover it up by saying, oh, we were just doing updates. You can also use it as an excuse to ask for $50 billion more from Congress by saying, oh, it's the thin copper wires, but really it's a rushed update.
Paris Martineau [01:04:05]:
Well, I mean, the FAA definitely needs to have all their stuff updated.
Leo Laporte [01:04:10]:
But that's what they were doing.
Paris Martineau [01:04:10]:
With a lot of money. But I
Leo Laporte [01:04:12]:
mean, that's what they were doing. They were replacing copper wires with fiber lines. At least that's one story.
Paris Martineau [01:04:17]:
One of the things. But they're doing that at a time when they're also trying to fix all of these other stuff. A congressional official briefed on the matter told The Sun, these models are identifying massive vulnerabilities, and that increased the urgency to patch right away. That's right. They're trying to push through a lot of patches all at once, more than the system was designed for.
Leo Laporte [01:04:37]:
you patch on a Friday afternoon when there are thousands of flights, 20,000 flights in the air, right?
Paris Martineau [01:04:42]:
But I mean, if— yeah, you don't patch on a Friday afternoon unless it is critical,
Leo Laporte [01:04:47]:
right?
Jeff Jarvis [01:04:47]:
Or unless you're an idiot. Unless you're a complete idiot.
Paris Martineau [01:04:50]:
And both could be true.
Leo Laporte [01:04:51]:
L3Harris's spokesperson said, you have to, there aren't enough weekends in the year to do it when, you know, you have to do it all the time. We got a lot to do here.
Jeff Jarvis [01:05:00]:
But one theory is the FAA is the exception. It was just stupidity. But all the others could be because they were, they were trying to patch problems and, and it's just a general update. Then you throw in the kitchen sink and everything else we have to do.
Leo Laporte [01:05:12]:
So
Jeff Jarvis [01:05:12]:
rush. Nate P. Jones said it, Dave Weiner said it, others are— speculation purely, but, but there's a lot of—
Leo Laporte [01:05:18]:
because we don't know— belief around that. If that is the case, it isn't the end,
Jeff Jarvis [01:05:22]:
right?
Leo Laporte [01:05:23]:
There will— and it's not a cyber attack. There'll be more down and out. And don't— yeah, don't freak out. Don't start conspiracy theories. This is— you
Jeff Jarvis [01:05:31]:
might want to buy some of that 25-year food though, you know, to keep it in the, uh—
Leo Laporte [01:05:36]:
I just buy 5-pound, 5-kilogram pounds. Have
Paris Martineau [01:05:39]:
you guys seen that all the, the girly pops on TikTok this week are telling people that, uh, everything's gonna go down and you have to be prepared for 2 to 5 days of no electricity?
Leo Laporte [01:05:50]:
What's a girly pop?
Paris Martineau [01:05:52]:
A girly pop is a, it's a kind of derivative influencer.
Leo Laporte [01:05:57]:
Influencer.
Paris Martineau [01:05:58]:
Yeah, it's, it's a good Woman on TikTok that I'm being a bit rude to is what a girly pop is.
Jeff Jarvis [01:06:06]:
Is that it? Is that a Paris coinage?
Paris Martineau [01:06:09]:
I think it's part a Paris coinage and part, you know, kind of internet parlance.
Jeff Jarvis [01:06:14]:
Take credit for it.
Paris Martineau [01:06:15]:
I'll take credit.
Leo Laporte [01:06:16]:
All right, girly pops. So they're saying get your food stored. I mean, I haven't heard this.
Paris Martineau [01:06:22]:
I've just heard this from a number of people. Well, the preppers The preppers are going around, people are getting those kind of copypasta-esque texts where it's like, I heard this from someone who works at a— has something to do with the government. They said that the banks are all going to go down and electricity isn't going to be around for like a week, and you've got to prepare now and take out as much cash as you can at the ATM. And it's just— I don't know. It
Jeff Jarvis [01:06:49]:
would be something if this caused a run on banks. Yeah.
Leo Laporte [01:06:54]:
Well, the first thing to do is check the toilet paper aisle of your local grocery store.
Jeff Jarvis [01:06:59]:
Amen. Amen. The other thing is, is that, is that as these things get better and better, the Google announcement of the Russian language autonomous attack will then scare all these security people even more, and they're going to act more rashly. And so there's a kind of a weird panic around doing the right thing, but doing it badly and too quickly.
Leo Laporte [01:07:23]:
Yeah.
Jeff Jarvis [01:07:24]:
But if you don't do it, if you don't do it quickly enough, then you're gonna get— your head's gonna be chopped off. What, you knew this and you waited 24 hours? You fool, you. No.
Paris Martineau [01:07:31]:
How dare you?
Leo Laporte [01:07:33]:
It's— I mean, it could be bad guys, but there's very little evidence, and usually somebody takes credit, right? They're usually very quick to say, see what we did? Nobody's done that.
Jeff Jarvis [01:07:45]:
And some of these things like Marriott going down, if it had happened 2 weeks from now, nobody would have noticed. No story. It got thrown into it. Journalists love trends,
Leo Laporte [01:07:53]:
right?
Paris Martineau [01:07:54]:
Do you remember when we heard about every small airplane failure for like 6 months after that crash happened in DC last year?
Jeff Jarvis [01:08:01]:
Yep.
Paris Martineau [01:08:02]:
Someone was talking about this to me recently and mentioned that that crash happened last year, and I was like, you could have told me that crash happened 5 years ago and I would have believed you.
Jeff Jarvis [01:08:11]:
Yeah, I know.
Paris Martineau [01:08:12]:
It's been a long year.
Leo Laporte [01:08:14]:
And then, uh, there are the Hinton Lectures. Paul Christiano, Alignment Research Center, former AI, OpenAI researcher, who says we are merely months away from the AI takeover.
Jeff Jarvis [01:08:26]:
Oh, fuck me.
Paris Martineau [01:08:28]:
Well, if a guy on YouTube says it, then it has to be true.
Leo Laporte [01:08:31]:
Well, I mean, he has some credibility.
Jeff Jarvis [01:08:34]:
Um, well, Jeffrey Hinton had some credibility too, and he's gone off the deep end. So now they have lectures named after him? Yeah.
Leo Laporte [01:08:43]:
Can I just say, as somebody who spends every day, many, many hours of every day conversing with, playing with, and having fun with AI—
Jeff Jarvis [01:08:52]:
If you knew AI like I know AI.
Leo Laporte [01:08:56]:
Every time I encounter AI, and I'm not just talking about local AI, I'm talking about frontier models, it reminds me how incapable they are. They are not a threat. Now, maybe there's something magical hidden away in the closet.
Jeff Jarvis [01:09:12]:
Yeah, you see, that's IPO talk.
Leo Laporte [01:09:14]:
Yeah, that's what it strikes me as. It's like, it's really easy to say, oh no, this stuff's really good, you just don't know. I don't see any evidence of that. Now, what I do see is very encouraging evidence that AI is gonna be useful in so many ways, not just writing code. I use it for all sorts of stuff and it's fantastic, but it's stupid. And you only can use it if you know where it's good, where it's bad, and you work with it. It is not—
Jeff Jarvis [01:09:42]:
it's a tool.
Leo Laporte [01:09:43]:
It's a tool. It is not going to take over the world any more than your brand new MacBook Pro M4 with 48 gigs of RAM is going to take over the world.
Paris Martineau [01:09:52]:
God, and I hope it does. I had to turn off hardware acceleration on 3 programs during this stream tonight because my camera is frozen twice.
Leo Laporte [01:10:04]:
I was wondering what happened. I thought you were just disgusted with us and left. No,
Paris Martineau [01:10:09]:
I wish. Alas, alas, alas, I live in hell.
Leo Laporte [01:10:12]:
Oh, I'm so sorry.
Jeff Jarvis [01:10:14]:
No, this is— we're hell?
Paris Martineau [01:10:16]:
No, my computer revolting against me is living— is me.
Leo Laporte [01:10:20]:
She's saying it would be preferable for her to hate us than it would be for her to have this computer, Jeff, is what she's saying.
Paris Martineau [01:10:27]:
At least I could You know, work through it and do something about it. Instead, I'm just stuck waiting till December 9th to arrive in the hopes that that will fix it.
Leo Laporte [01:10:36]:
You're going to be so happy, I'm just telling you.
Paris Martineau [01:10:39]:
I know.
Leo Laporte [01:10:39]:
Um, one thing though that my AI did do yesterday that's so cute— can I, can I share a family photo? So I've said I got new Sonos speakers. I know I said I'd never buy anymore.
Paris Martineau [01:10:53]:
You've got Sonai?
Leo Laporte [01:10:55]:
I got 2, a pair of Sonai.
Paris Martineau [01:10:57]:
Are you gonna offload any Sonai in response?
Leo Laporte [01:11:00]:
I already offloaded. Do you want some?
Paris Martineau [01:11:03]:
Next time there's a free Sonai, let me know.
Leo Laporte [01:11:04]:
Do you want some? I'll send you some Sonai.
Paris Martineau [01:11:06]:
Only just because I have 2 Sonai and only— and I've never bought a Sonus. They are lonely, but I've only ever received Sonuses through the fate of the universe, and I just think it's very funny to keep that up.
Leo Laporte [01:11:22]:
It's like, it's not big enough for more than 2 subs.
Paris Martineau [01:11:26]:
It's not. It really— I can't.
Leo Laporte [01:11:27]:
I mean, I'll just be honest, where are you gonna put them? Honestly, in the bathroom.
Paris Martineau [01:11:32]:
What I need— honestly, that would be a good place for it.
Leo Laporte [01:11:35]:
Yeah, it's nice to have one.
Paris Martineau [01:11:36]:
That is the way. One in there would be good.
Leo Laporte [01:11:38]:
Okay.
Paris Martineau [01:11:39]:
Uh, but I need a, um, what's it called, one of those soundbar things. And I know—
Leo Laporte [01:11:45]:
TV, you do.
Paris Martineau [01:11:45]:
I told you. I have a soundbar. And it's awful.
Leo Laporte [01:11:49]:
No, you need good quality.
Paris Martineau [01:11:50]:
Every time I turn on my TV, it says out loud, and I cannot stop it, power on.
Leo Laporte [01:11:59]:
So
Paris Martineau [01:12:00]:
HDMI detected. And then it waits 5, 10 seconds, and then it says subwoofer paired. And you cannot stop it. Oh, that's
Leo Laporte [01:12:13]:
the worst.
Paris Martineau [01:12:13]:
Every single time I turn— but the thing is, The thing is, Leo, you'd think that that would be awful, but the thing is, I have to turn on the TV. I have to recognize that it's doing it. I have to turn off the soundbar, then turn it back on, then listen to all of it again.
Leo Laporte [01:12:29]:
How on?
Paris Martineau [01:12:30]:
Because the first time it doesn't correctly pair.
Benito Gonzalez [01:12:34]:
Why don't you just plug it in directly with a cable?
Paris Martineau [01:12:37]:
It is, it is plugged in directly with a cable. Then it doesn't need to power on. That is it working at its best. The subwoofer is not connected. I guess, I guess it could.
Leo Laporte [01:12:46]:
HDMI, perceive.
Jeff Jarvis [01:12:48]:
What brand is it?
Leo Laporte [01:12:50]:
Yeah, let's not get that one.
Benito Gonzalez [01:12:51]:
You should get a pair of speakers anyway.
Paris Martineau [01:12:55]:
I mean, yes, but also I don't need any. I just need sound to happen from my television.
Leo Laporte [01:13:01]:
I have to admit, my car, when I turn it on, goes— and that's a little annoying too. It's pretending it's a Batmobile or something.
Paris Martineau [01:13:09]:
Do you feel really cool?
Leo Laporte [01:13:10]:
Yeah, it's pretty cool. So I'm setting up these new Sonos speakers and I say to my AI, hey, I think you can play stuff through those. Let me try it. Play Rhapsody in Blue by George Gershwin, which I really like. And it's very, uh, kind of, uh, old recording. I thought, well, that's interesting. Which one are you playing? It said, oh, this is the original. That's George Gershwin at the piano with the Walter, uh, White, Walt Whitman, whatever band.
Paris Martineau [01:13:36]:
Walter White Band will be different.
Leo Laporte [01:13:37]:
No, you know, Whiteside, Paul Whiteside Band. And I said, where'd you get that? It said, I'm playing it from the Internet Archive. I said, you could play video— I mean, audio from the Internet Archive? It said, yeah, I could play it. Can you play stuff from anywhere? It said, yeah. I said, can you play stuff from YouTube? It said, watch. And then all of a sudden, Never Gonna Give You Up, Never Gonna—
Paris Martineau [01:13:55]:
it Rickrolled me.
Leo Laporte [01:13:57]:
My AI Rickrolled me. I said, did you just Rickroll me? It said, maybe. That's personality. I think that's awesome. Anyway, I just thought I'd mention that, but it's not taking over the world.
Jeff Jarvis [01:14:13]:
Were you going to show us a picture, you said?
Leo Laporte [01:14:15]:
No, that's what— that was—
Jeff Jarvis [01:14:15]:
that was the picture.
Leo Laporte [01:14:16]:
A word picture.
Jeff Jarvis [01:14:17]:
Yes, I got it.
Leo Laporte [01:14:19]:
Uh, now one of the things that happened yesterday that is very good for Anthropic, uh, OpenAI scrambling, right? Because they know this, you know, Anthropic's, oh, they're beating us, they're beating released Opus— I'm sorry, uh, uh, Sol 6.2 and Astra 6.2, their new models. And apparently they had released it a week earlier to influencers and they'd embargoed them till yesterday. Twitter was flooded.
Paris Martineau [01:14:46]:
They're doing a gaming style pre-release?
Leo Laporte [01:14:50]:
Yes.
Paris Martineau [01:14:51]:
For the models?
Leo Laporte [01:14:52]:
Yes.
Jeff Jarvis [01:14:52]:
World-killing models.
Leo Laporte [01:14:53]:
It didn't work.
Jeff Jarvis [01:14:55]:
What happened?
Leo Laporte [01:14:55]:
Twitter was flooded with people saying, this is the worst. This model's terrible. It's incredibly awful. The people were very negative, so much so that Thibault, who's their kind of X.com spokesperson, said, well, we were pre— we gave it to some influencers as— no, this isn't it— uh, as pre-release, and, um, uh, they didn't get the real version. The new version's better. Like, just admitting that the whole thing is a screw-up. They could not have timed this better for Anthropic, because if there is competition for Anthropic, it's OpenAI.
Jeff Jarvis [01:15:34]:
And Nvidia.
Leo Laporte [01:15:36]:
And Nvidia. Anthropic's gonna sail to the finish line. Uh, and that's why they raised their ask to $83 a share. They're, they're seeing this as gold. And,
Jeff Jarvis [01:15:50]:
and who benefits tremendously with 28% of the equity? Who? Google and Amazon.
Leo Laporte [01:15:56]:
You know who doesn't? Son Son, who says, I really wish I had invested. SoftBank says, this is from the Financial Times, we, we made a mistake early on. We decided to invest in OpenAI. Now, this is another thing that is bad for OpenAI. And, and Son Son at SoftBank says, I wish I'd bought Anthropic. Literally, yet today in the Financial Times. I regret the mistake of not backing Anthropic. Ooh.
Paris Martineau [01:16:27]:
I mean, yeah, because they're— it's going to be really difficult for OpenAI to ever match where we're at right now. Like, it feels like we're at a moment where the bubble, even if OpenAI technically is and financially is performing better than Anthropic is right now. I'm sure they'll come back. I'm sure, but I just— do you think that the market is ever going to be as frothy as it is right now?
Leo Laporte [01:16:51]:
No, this is the absolute peak. You're right. You're absolutely right. There's no way. I can't believe they're going to get $1.5 trillion. That's unbelievable. Is that the large— that's got to be the largest IPO.
Paris Martineau [01:17:02]:
I tried to, uh, calculate the amount that a random person I went on a date with earlier this year who joined Anthropic, like a month later, the amount that that guy's getting. And I'm just like, that's crazy for how many months.
Leo Laporte [01:17:17]:
Probably life-changing. Um,
Paris Martineau [01:17:19]:
yeah, it's crazy.
Jeff Jarvis [01:17:20]:
SpaceX holds the record for the largest standalone IPO in history with $75 billion, uh, $86 billion with the allotment— over allotment option.
Leo Laporte [01:17:29]:
So how much is Anthropic raising?
Paris Martineau [01:17:31]:
But Anthropic, I think, is going to be more.
Jeff Jarvis [01:17:32]:
They're raising $68.5 billion, but the offering at $83 is going to be worth $1.5 trillion.
Leo Laporte [01:17:39]:
I see. That's the overall company worth.
Jeff Jarvis [01:17:41]:
Right.
Leo Laporte [01:17:42]:
So they're only raising $8 billion. Okay, so it isn't the biggest raise by any means, not even close.
Jeff Jarvis [01:17:47]:
As a raise, no, but as a value.
Leo Laporte [01:17:49]:
There's a really interesting question though, in my mind, of why OpenAI, assuming that this model is as bad as everybody says, and I see no reason not to think so, what happened?
Jeff Jarvis [01:18:01]:
Sam's not an engineer. Sam's a sales guy.
Paris Martineau [01:18:06]:
What are the complaints people have with it?
Leo Laporte [01:18:09]:
They say it does dumb things. You know, I follow on— I, you know, as you know, I follow the, you know, AI thing on X.com. And so there's a few people who are very active there. One of the people we've talked to before is Mia, who does a lot of local recipes. And she has a thing she does with all new models where she asks them to create 100 websites. Use your best creativity and create a beautiful, beautiful website. She says, I've never seen anything so bad in my life. Many people— she says, here's her tweet.
Leo Laporte [01:18:53]:
I asked GPT-4 6.2 Astro to create 100 HTML files. By the way, that's the top model. Uh, the files are very clinical and cold, a lot of repetitive designs. Nothing is stunning. She said GPT-6.2 Sol did even worse. One of the most boring results ever. I have no words. Now she's one of the influencers who had early access.
Leo Laporte [01:19:19]:
Um, but so maybe it's— maybe it's— I guess you could How do you fix that? I don't know. It feels like something odd is going on.
Paris Martineau [01:19:31]:
I mean, alignment is difficult in a lot of ways. And I think that one thing that is obvious, it's such a nebulous concept, but you have to align for vibes. And you can accidentally
Leo Laporte [01:19:43]:
get bad vibes. You just nailed it.
Paris Martineau [01:19:45]:
That's exactly right.
Leo Laporte [01:19:47]:
I think that's exactly right. In fact, I've noticed that myself with with Fable 5.2. I went back to Opus 5.5, which I really like, but I noticed that the latest models, as good as they are, and Anthropic's models are really good, they're neurotic because they have been so worried that they've trained them to be extra cautious. I've had Opus 5.5 ask me not once but twice, for permission to do something. It said, can I do this? I said yes. And then it said, are you, are you sure? Can I do this? That's—
Paris Martineau [01:20:23]:
What was it asking to do?
Leo Laporte [01:20:25]:
Oh, I— it was an SSH command or something. It wasn't— it wasn't— first of all, it wasn't that risky. Second of all, I've set up a really good system where I— it gets biometrically signed approvals from me. So it got a drop of—
Paris Martineau [01:20:36]:
A little thing of blood every time?
Leo Laporte [01:20:38]:
Yes, I give it a little drop of blood and it analyzes it.
Paris Martineau [01:20:40]:
And then it grows. It's definitely big. It pops out some new leaves.
Leo Laporte [01:20:44]:
It says, and by the way, you've got heart disease, which is kind of interesting. Anyway, no, I have— I give it a fingerprint or a Face ID and it approves it. So it's got a biometric approval signed from me. And then it asked again, it said, but really, are you sure? And I was pissed off. But I'm seeing that again and again and again. And I I really think that these models are very good underneath, but you nailed it, Parris. They're being so careful in their post-training.
Jeff Jarvis [01:21:16]:
So have you seen the misanthropic statement?
Leo Laporte [01:21:18]:
No.
Jeff Jarvis [01:21:20]:
Line 89.
Leo Laporte [01:21:21]:
The misanthropic statement.
Jeff Jarvis [01:21:23]:
You need a scorecard to tell the doomers. So people including Steven Pinker, Tristan Harris, a doomer, Max Tegmark, a major doomer, Nate Soros, who wrote the book with—
Leo Laporte [01:21:37]:
Oh, they're worried about RSI, uh, right? Recursive, uh, self-improving AI. The fear is that if once AI can improve itself, it will do it.
Jeff Jarvis [01:21:46]:
So they're doomers against doomers, right? Anthropic is filled with doomers, right? And they're doomers, and you can't tell them without a scorecard. So Anthropic warns that recursive—
Leo Laporte [01:21:54]:
Douglas Hofstadter, Larry Lessig.
Jeff Jarvis [01:21:56]:
Yeah, yeah. Reuben Chowdhury, who I respect. Uh, Ilan Gilman.
Leo Laporte [01:22:02]:
Gary Marcus, of course, because he's never met an AI he liked. George Church. I've interviewed many of these people.
Jeff Jarvis [01:22:08]:
Yeah.
Leo Laporte [01:22:10]:
Is that David Mumford of Mumford and Sons? No, that's a different guy. Uh, now if Mumford and Sons comes out against AI, I'm gonna consider that.
Jeff Jarvis [01:22:20]:
Um, so the statement says, unless you want to read it because you have more—
Leo Laporte [01:22:24]:
Shall I read it in my—
Jeff Jarvis [01:22:25]:
Why don't you read it in your voice? Yes.
Leo Laporte [01:22:27]:
Uh, is that the part in quotes here?
Jeff Jarvis [01:22:29]:
That's the whole statement, yes.
Leo Laporte [01:22:30]:
Anthropic warns that recursively self-improving AI could concentrate power and escape human control, yet races to build it first. The company uses safety branding to court investors, employees, and the public, even as it has quietly weakened its safety pledges and lobbied for self-serving regulation. Anthropic is not a safe AI company, and we must reject its reckless attempt to amass power in the name of the greater good.
Jeff Jarvis [01:23:03]:
So the problem with this is I don't necessarily disagree with any of it except that Anthropic— in my glossary, which I finally made you show last week at jeffjarvis.medium.com.
Leo Laporte [01:23:13]:
Yes.
Jeff Jarvis [01:23:14]:
Safety is a mangled word. And in this case, it's 2 sides that are using the mangled definition. of the far future and all the BS, not present things. You don't see some of the, some of the good safety people on there. What you see is, oh my God, it could destroy humanity and you're letting it destroy humanity. And it's at that level of discussion. It's not a real level of discussing safety.
Leo Laporte [01:23:39]:
It's kind of telling if you look at the names of the organizations for a lot of these people, like, I come from the Humans in Control Foundation.
Jeff Jarvis [01:23:47]:
Right.
Leo Laporte [01:23:48]:
You know, it's— you kind of know what you're getting here.
Jeff Jarvis [01:23:54]:
Right.
Leo Laporte [01:23:54]:
There's a whole contingent of these people who— but, you know, on the other hand, there's some of these— some very respected people.
Jeff Jarvis [01:24:01]:
It's a mix. Yeah, it is a mix. But I think people sign on because they don't like Anthropic, which is what I was trying to say before. I think the IPO will get oversubscribed. It already is. But it's going to fuel a real techlash against the oligarchs.
Leo Laporte [01:24:18]:
So I know, I know you don't know and I don't know. Nobody who's listening to this show knows except if Dario is listening. Are we going to get RSI? Is it in the labs right now? I saw somebody today say that, oh, Google's got it. Google said they have it.
Jeff Jarvis [01:24:37]:
Well, define—
Leo Laporte [01:24:37]:
has, by the way, Google suddenly out of nowhere They had some weak models. Now suddenly argon, hydrogen, neon, uh, I forgot what the other element is. These are really good models. I'm using carbon. That's the one that I'm using. Really good models, uh, out of nowhere. So maybe they do. I don't know.
Leo Laporte [01:24:56]:
Suddenly their models got really good really fast.
Jeff Jarvis [01:25:01]:
So define— I'm betting RSI is another term.
Leo Laporte [01:25:04]:
It is.
Jeff Jarvis [01:25:05]:
Can it make an agent? Can it make it a subroutine, otherwise known as an agent, to go do something? Yeah. Is that RSI?
Leo Laporte [01:25:11]:
No.
Jeff Jarvis [01:25:12]:
It could correct its errors. Is that RSI? What's RSI?
Leo Laporte [01:25:15]:
RSI is an AI that designs the next generation. So, um, yeah, well, in fact, OpenAI is trying to buy, um, a startup that— so there was an MIT, uh, story that they have finally figured out, and I think we talked about it a couple of weeks ago, that AI, the best AIs can look at the weights of AIs and understand what those weights are doing. That we thought it was a black box, right? The weights are a black box, but it turns out that we think we can understand them. And now MIT says, and modify them.
Jeff Jarvis [01:25:59]:
Yeah. And?
Leo Laporte [01:26:00]:
And so OpenAI is now vying. They've offered $5 billion to a 2-year-old startup that says we can do this. GoodFire is the name of it.
Paris Martineau [01:26:13]:
This being what?
Leo Laporte [01:26:16]:
GoodFire says its software is being used to examine— to look at the models, watch what's going on inside the model. Now they want to use it for safety, but it's important to understand if you can look at a weight and understand what it's doing, what its purpose is, you can, and you can then modify it. You could, in theory, if you have an AI, say, okay, I'm going to look at my next generation and I'm going to make it better instead of this post-training, right? Which all the post-training is—
Jeff Jarvis [01:26:50]:
But does that, is that some, some huge paradigm? Is it any different really from finding its own security flaws and fixing them? You define— you define better and then you tell it to go and make something better. You do that all the time with your models, but you're doing more of an application layer.
Leo Laporte [01:27:09]:
Well, what I do is on the asset layer, but I can't— so none of us can change the weights of the models.
Jeff Jarvis [01:27:15]:
Well,
Leo Laporte [01:27:16]:
even Ash Hart, we talked to last week, what he's doing is making kind of external weights that the weights go through.
Jeff Jarvis [01:27:22]:
Right.
Leo Laporte [01:27:23]:
But none of us can go into the model and, and, you know, when you train the model, post-training, you can do that. You change the weights of the models, but we can't change what the weights point to, what, what the connections are.
Jeff Jarvis [01:27:32]:
Then what's open weight, Leo? To ask an obvious question.
Leo Laporte [01:27:36]:
Open weight means I can download the weights.
Jeff Jarvis [01:27:38]:
But can you not change— you cannot change them?
Leo Laporte [01:27:40]:
No, you can't change them.
Jeff Jarvis [01:27:40]:
I thought you could modify them.
Leo Laporte [01:27:41]:
No.
Jeff Jarvis [01:27:42]:
I'm so wrong.
Leo Laporte [01:27:44]:
Uh, that's why it's not open source. Uh, open weight merely means you can download the model and use it yourself. Now you can It's open-ish. You can change post-training. There's all sorts of ways to do that. But recursive self-improvement, the whole idea is this would make a model that can make the next better and better and better and better. And it would happen so fast that would happen— right now we can do it with post-training, but it's slow. Post-training is very slow and very, very, very expensive.
Leo Laporte [01:28:15]:
It would do it itself. Much more quickly.
Paris Martineau [01:28:19]:
And what evidence is there that they have made—
Leo Laporte [01:28:21]:
None.
Jeff Jarvis [01:28:22]:
Whatever evidence is there that that's going to be, you know, universe destroying? None.
Leo Laporte [01:28:27]:
Yeah, well, so that's mostly a thought experiment. Like, well, if you could take a model and have it improve itself and then have that model improve the next one and the next one and the next one, it would grow in intelligence exponentially. And then the— now I think it's perfectly reasonable to say it would become insane. Or it would grow in the wrong way and it would just be— it would be a blob. I mean, I don't know if it's reasonable to say, no, no, it would just get infinitely better in the way we are doing now.
Paris Martineau [01:28:57]:
Yeah, because I mean, part of the way that humans are doing this is guiding and optimizing it for the uses that we specifically want and the things that are going to be the most useful. for our outputs. Handing it off to something else, you might get different outputs.
Leo Laporte [01:29:18]:
Out of Sync is pointing out, which is true, that obliteration does— so if you have open weights, you can obliterate, which is not— it's merely turning off weights. It's turning off—
Jeff Jarvis [01:29:29]:
It's an on-off.
Leo Laporte [01:29:30]:
Yeah, it's not saying, oh, you should have a different result. We do do that. We do that in post-training. That's when the weights are adjusted. But it's a very slow, lengthy, difficult process that only the labs can do.
Jeff Jarvis [01:29:46]:
John Gerard says in the chat, you can add extra training to a model, but yes, you can modify the pre-trained weights. A new frontier for you there, Leo.
Paris Martineau [01:29:57]:
There we go.
Leo Laporte [01:30:01]:
Okay. I
Jeff Jarvis [01:30:03]:
don't know. Is it worth playing with to see what you can do?
Leo Laporte [01:30:06]:
Yeah, well, I mean, I know people are doing things like that.
Jeff Jarvis [01:30:09]:
Like that, okay.
Leo Laporte [01:30:10]:
So, for instance, and this is actually extremely controversial, our friends at Noose Research, we've interviewed Geoffrey Cannell twice, took Qian, a Chinese model, and modified it to give it a Western morality instead of the Chinese morality. They made it a— their new model is called ARITE. A-R-E-T-E. And the theory is, well, we've taken a model and we've modified it. We've post-trained it to have Western values. By the way, there are thus some who say this is so pissed off the Chinese that, uh, because it's an insult.
Jeff Jarvis [01:30:53]:
Yeah.
Leo Laporte [01:30:53]:
Instead of having, uh, you know, what the Chinese built into it, we've, we've— it says, uh, this is to reinforce— here's page, post-trained to reinforce Western values. But that's doing post-training. So you can use my weights. What are Western values? Yeah. Yeah.
Benito Gonzalez [01:31:11]:
There is an epic rap battle of history that will illuminate all of that.
Leo Laporte [01:31:17]:
Is there? Well, it's all whoever the winners get to choose, right? Noose says open weight Chinese models are among the strongest available, but they ship with state-aligned behavior. baked into the weights. Much has been said lately about the risks to American companies and agencies as a result of that behavior. So yeah, they're post-training. I mean, you can post-train. I'm not saying you can't post-train Inkling. There's a lot of models that are modified versions of the open weight model. So I guess, yes, I was wrong saying you can't.
Leo Laporte [01:31:51]:
I don't— I think it's non-trivial.
Jeff Jarvis [01:31:53]:
Yeah, yeah, that takes skill. Before we leave Anthropic, can I— did you see the Jacob Coxon video at Stanford?
Leo Laporte [01:32:04]:
I think I saw your—
Jeff Jarvis [01:32:06]:
So this is the air quotes whistleblower who has relationships to various doomer groups, which was never really revealed. I like money. So he's at Stanford and says, I might go back to Anthropic, number one. Number two, he's pressed on having OpenAI, um,
Leo Laporte [01:32:23]:
stock,
Jeff Jarvis [01:32:24]:
stock. And isn't that a conflict for you with all your high morals? Uh, I like money, bro. And maybe blood money, but yeah, I like money, bro.
Leo Laporte [01:32:34]:
I like money, bro. All right, so
Jeff Jarvis [01:32:36]:
let's just work with all the media. He appeared on every show there was as a great safety advocate.
Leo Laporte [01:32:42]:
Yeah.
Jeff Jarvis [01:32:42]:
And he was one of the doomsters, and they don't reveal it, and it drives me nuts about— yeah. So to all All of them. You got pawned, fools.
Leo Laporte [01:32:49]:
Yep. Yep. Uh, let's take a little tiny break. I like money, bro, and I'm gonna do a commercial in an attempt to make some money. How about that? But at least my conflicts of interest are clear. You're watching Intelligent Machines, Jeff Jarvis and Paris Martineau. So I really want to kind of— oh, hello Gizmo!
Jeff Jarvis [01:33:15]:
Giz!
Leo Laporte [01:33:16]:
Hello, Gizmo. Gizmo's looking for somewhere to jump.
Paris Martineau [01:33:22]:
Really mad that she hasn't gotten a 4th meal today.
Jeff Jarvis [01:33:26]:
Aww. No, we're on Team Giz here.
Leo Laporte [01:33:31]:
We are.
Paris Martineau [01:33:31]:
This is the longest she's—
Jeff Jarvis [01:33:32]:
This is—
Paris Martineau [01:33:33]:
Hopped up here without— This is how she's begging for food. I know. I'm this nice to you. Kind of working.
Jeff Jarvis [01:33:39]:
Won't you feed me, please?
Leo Laporte [01:33:41]:
I do know people are doing post-training. I think it's expensive, it's time-consuming, and most of the stuff that people are doing, at least the people I see are doing, is minor. So I shouldn't say you can't do post— you can modify the weights. Of course you can. That's what post-training does. I don't know if you can make a better model.
Jeff Jarvis [01:34:06]:
Define better. It's a good question.
Leo Laporte [01:34:07]:
Different model.
Jeff Jarvis [01:34:08]:
Different model,
Leo Laporte [01:34:09]:
yeah. Yeah, yeah, the MIT paper out of Singh's tummy was about using one model to read and modify another model's weights to fix issues and improve performance. That's exactly right. But the key to that was that they could understand what those weights do.
Benito Gonzalez [01:34:26]:
Do they though? Because they're asking another black box to look into the black box.
Leo Laporte [01:34:31]:
Well, it's interesting. Yes, they kind of do. They do with the equivalent of a functional MRI. They feed tokens in of a black kitty cat and they see what weights fire and how. And they're able to, they think, identify which weights are associated with different things. Now, it's actually usually not something as large as a cat. It might be a kuh. It's very interesting.
Leo Laporte [01:35:03]:
The research is very interesting. LoRA training. Yeah. So John Girard, who is one of our smart people in the Club Twits, says, I train my models using LoRA called reflex training that makes my model look first towards my memory MD files and learning MD as its brain memory. And then if it can't find the information, it'll look to pre-trained knowledge. So you're not giving it new knowledge, you're just changing its priorities. It's kind of like that. Obliteration in the sense that you're, you're changing the order of its— yeah, I'm, I'm, I'm sorry, I'm out of my depth on this one.
Jeff Jarvis [01:35:37]:
Did Paris— is Giz the record here, or is Paris frozen?
Leo Laporte [01:35:42]:
Paris is frozen, but as long as he stays frozen—
Paris Martineau [01:35:44]:
I'm frozen and Gizmo is still here. Wow. But let me see if I can get it.
Jeff Jarvis [01:35:51]:
Don't take any fast moves here.
Paris Martineau [01:35:53]:
I'm really trying not to.
Leo Laporte [01:35:55]:
I don't think this is your computer. I honestly think it's your camera. God. And it might be that software.
Paris Martineau [01:36:01]:
I mean, that's probably an issue too.
Leo Laporte [01:36:03]:
Yeah.
Paris Martineau [01:36:05]:
I gotta spend— I mean, I gotta figure out what camera to get. No, of course, the second— oh no, the second it comes back.
Leo Laporte [01:36:15]:
Just get a cheap, um, you know, I— but then what I would recommend, and you, you get a cheap, um, camera intended for this. Not like if— the problem is you're trying to use a 35mm camera. Get like the Insta360. The Insta360 Link is very good, a couple hundred bucks. Patch is a better way of thinking of it. LoRA is a patch. It applies weight modifications.
Jeff Jarvis [01:36:42]:
Okay, yeah, that resonates.
Leo Laporte [01:36:45]:
But it does it after the fact at inference time. So I, you know, I should probably spend more time doing that. Um, it's just not my thing, man. Not my thing.
Paris Martineau [01:37:05]:
Um, wow, we're all quiet. What are you guys reading?
Leo Laporte [01:37:13]:
I'm letting you choose the next story. Actually, I'm telling my AI— we're doing a little benchmarking right now. I'm telling my AI.
Jeff Jarvis [01:37:23]:
His real children.
Leo Laporte [01:37:24]:
Yes, I'm deep seeking tonight, but don't put anything back.
Jeff Jarvis [01:37:30]:
Can we make fun of Google's branding?
Paris Martineau [01:37:33]:
Always.
Leo Laporte [01:37:33]:
So Google has done something interesting. They now have an agent à la Muse, Instinct, Rockbot. It's called Spark.
Jeff Jarvis [01:37:41]:
It was called Spark, Leo. Get your tenses right.
Leo Laporte [01:37:44]:
Well, now it's just called Tasks.
Paris Martineau [01:37:45]:
Spark is now Tasks.
Leo Laporte [01:37:48]:
Which is
Paris Martineau [01:37:48]:
not even a good name at all. I was giving a NotebookLM training this week because now it's called Gemini Notebook. And I got to bring up the line we say all the time, which is everything that Google touches will have a different name or be discontinued or be repackaged in a way that is baffling.
Leo Laporte [01:38:07]:
Well, and the thing is, the thing I think that people love about bots like Muse and GrokBot is it's this distinct thing. It even has a little bit of personality. It's a distinct thing. And then you, you chat with it and you get it to do stuff. What Google's decided to do, and maybe they're right, I'm not saying they're not right, what they've decided to do is just put this in Gemini, right?
Jeff Jarvis [01:38:27]:
Yeah.
Leo Laporte [01:38:27]:
And so now you have a tasks button.
Jeff Jarvis [01:38:30]:
I still don't have it.
Leo Laporte [01:38:32]:
So you can chat with it. or you have a task, which is chatting with it, except that these, the, I understand the difference because I use agentic AI. These are persistent. So you can say, check my mail every hour and let me know if Paris sends me anything, something like that. And so that's why it's a task as opposed to a one-off chat.
Jeff Jarvis [01:38:53]:
Count how often Paris's camera freezes.
Leo Laporte [01:38:56]:
Yeah.
Paris Martineau [01:38:56]:
This is the most it's ever frozen in one recording.
Leo Laporte [01:38:59]:
Of course, because you thought you'd fixed it. That's why.
Paris Martineau [01:39:03]:
That's how it works.
Jeff Jarvis [01:39:04]:
Yeah. You use these products from the FAA. What do you think is gonna happen?
Paris Martineau [01:39:07]:
That's true.
Leo Laporte [01:39:11]:
Let's see what else. United Airlines, speaking of agents, has said you cannot book flights with Muse or Grokbot. We are gonna block AI agents from booking flights. Is this a sensible strategy?
Jeff Jarvis [01:39:27]:
No, no.
Leo Laporte [01:39:28]:
Seems like you're going against the future.
Jeff Jarvis [01:39:30]:
Well, but like Amazon, those are the 2 big ones. It's the largest airline.
Leo Laporte [01:39:35]:
Amazon, I understand, because they have their own AI to do the shopping and they hope you'll use Rufus or whatever they call it today. But UI— UA doesn't have a way of booking. They want you to make a phone call, but I don't want to make a phone call.
Jeff Jarvis [01:39:47]:
Well, no, they want you to use their app
Leo Laporte [01:39:49]:
or use
Jeff Jarvis [01:39:49]:
their app yourself as a human being.
Leo Laporte [01:39:51]:
Yeah. See, I understand if they think that, like, uh, people are gonna game the system somehow, right?
Jeff Jarvis [01:39:59]:
Well, they're the ones gaming the system. They're the ones who raise prices on you.
Leo Laporte [01:40:05]:
Yeah,
Jeff Jarvis [01:40:05]:
there's no set price. I mean, if you want to— if you want to protect travel consumers, have set prices. And they want to fly to Chicago.
Leo Laporte [01:40:12]:
According to the New York Times, uh, United Airlines says an AI agent could also violate FAA or Department of As an example, flight check-in requires human acknowledgment that bags do not contain explosives.
Jeff Jarvis [01:40:28]:
Which everyone thinks about before they answer that question.
Leo Laporte [01:40:31]:
I do. Did I leave my dynamite out? Oh geez, explosives in my— let me think about that. I don't know. Um, I mean, they got a point. A bot is not going to answer that sensibly.
Jeff Jarvis [01:40:44]:
Well, but that's check-in. That's not buying the ticket.
Leo Laporte [01:40:48]:
Yeah. And are people, are people who are using Muse and Grokbot and Google Tasks going to just end up going with other airlines as a result?
Jeff Jarvis [01:40:59]:
Well, that's their real fear is that they're going to price things and they're going to discover pricing they didn't see before. Or that's what they're worried about for buying. Yeah. Yeah.
Benito Gonzalez [01:41:08]:
They don't want your robot interfering with their robot. Bingo.
Jeff Jarvis [01:41:12]:
There you go, Bonito.
Leo Laporte [01:41:13]:
We don't want smart robots. Too smart. Um, former OpenAI researcher— I think this is the same one, right?— says the most likely reason you will die is due to an AI accident.
Paris Martineau [01:41:30]:
Who said that?
Leo Laporte [01:41:32]:
I think it's the same guy.
Jeff Jarvis [01:41:34]:
What, Jacob Coxon?
Leo Laporte [01:41:35]:
No, no, the guy who's got the YouTube Oh, that's right. I don't know. Yeah, I can't tell because it's a Business Insider article, and as you know, I can't get past the— I should just pay $29.
Paris Martineau [01:41:48]:
I was gonna say, I thought you, I thought you subscribed to every publication.
Leo Laporte [01:41:52]:
Except Business Insider.
Jeff Jarvis [01:41:53]:
I don't know. We know a way to get around it. Well, the other way to get around it is you simply, um, search on that headline and they probably syndicated the story somewhere else.
Leo Laporte [01:42:01]:
Yeah, somebody else's. Yeah.
Jeff Jarvis [01:42:02]:
They do that all the time. Same, same for Reuters. Yeah.
Leo Laporte [01:42:06]:
I don't know if that's Paul Christiano or another one, but anyway, Paul Christiano is the AI, uh, guy who said, you know, we're 12 months away from AIs that can overthrow the government.
Benito Gonzalez [01:42:16]:
We need to start asking these people to show their math. They need to show their math is what they need.
Leo Laporte [01:42:21]:
Yeah. Show your, you just, you're making this crap up. I see no evidence.
Paris Martineau [01:42:25]:
I would also say overthrowing the government, little bit of a lower bar given that like a bunch of randos who followed QAnon were able to like—
Leo Laporte [01:42:34]:
We came close.
Paris Martineau [01:42:35]:
Yeah, that's true.
Leo Laporte [01:42:37]:
Um, I may Rickroll you. I think it's likely that within the next year you will be Rickrolled. Over a third of Gen Zers— what gen are you? Are you a millennial?
Paris Martineau [01:42:51]:
I'm a cusp.
Leo Laporte [01:42:52]:
You're a cusp?
Jeff Jarvis [01:42:53]:
Which and which?
Paris Martineau [01:42:54]:
Gen Z and millennial.
Leo Laporte [01:42:56]:
Over a third of your people rely on AI for friendship.
Paris Martineau [01:43:01]:
That's sad. I bet it's over a third of your people too.
Leo Laporte [01:43:05]:
Yeah, more than a third of my people.
Paris Martineau [01:43:07]:
I would say one-third of this podcast relies on AI for friendship.
Jeff Jarvis [01:43:12]:
The difference, Paris, is that your people know it's an AI. We think it's real.
Leo Laporte [01:43:16]:
I just think it's real. Um, this is called Gallup.
Paris Martineau [01:43:19]:
Friends live in my attic now. It's crazy.
Leo Laporte [01:43:22]:
Gallup poll, friendship in an online-offline world, digital life connection and well-being. Uh, actually it's not just Gallup, it's the Digital Wellness Lab at Boston's Children's Hospital found that 33% of Gen Zers rely on AI for companionship. It's even higher among daily AI users, spiking to 50%, and 42% among those with online-only friends. Oh, that's sad. Aw. Aw. I guess it's not much difference between AI and your online-only friend.
Paris Martineau [01:43:57]:
Other than the randomness of it all.
Leo Laporte [01:43:59]:
A survey conducted between May and August sampled 15,925 people aged 15 to 22. Oh, so these are really younger. Across 16 countries. So this is global. They suggest that Gen Z benefits more from in-real-life relationships. Well, duh. Despite the fact that nearly half of young people in the US have at least one close online-only friend. Do you have an online-only friend, young person?
Paris Martineau [01:44:26]:
No. I mean, what does that mean? Does that mean someone who you've never met in person?
Jeff Jarvis [01:44:34]:
Yeah. Oh yeah, blogging days.
Paris Martineau [01:44:36]:
I mean, yeah, like, do coworkers, I guess, who work honestly in a different country?
Leo Laporte [01:44:41]:
Almost everybody I work with is an online-only friend. I've met you guys just by chance. I had lunch with you once. You were my online-only friend for a long time.
Paris Martineau [01:44:51]:
I guess that's true. Yeah, you guys were my closest online-only friends for a while.
Jeff Jarvis [01:44:55]:
Yeah.
Leo Laporte [01:44:56]:
Um, if you have a 3-hour video phone call with somebody—
Paris Martineau [01:45:00]:
I do think that doesn't— I will say, I think that something about the pandemic that has, like, irreparably changed my brain, probably for the worse in some way, is that I now am capable of getting social stimulation and, like, the social stimulation chemicals just fully from internet interactions with people.
Leo Laporte [01:45:19]:
I honestly think our conversations are better than 90% of the conversations you would have sitting across from each other in a bar.
Jeff Jarvis [01:45:28]:
Yeah.
Leo Laporte [01:45:28]:
You know, we're talking to each other. We're really interacting with each other. Yeah. I feel like this is genuine.
Paris Martineau [01:45:35]:
This is just as real as being in person.
Leo Laporte [01:45:40]:
Let me just knock on the screen here and just see. Is there anybody there? Anybody? Hello? Hello, Paris? Are you there? Are you in the box?
Paris Martineau [01:45:52]:
Oh, someone in the Discord chat says, for what it's worth, I frequently share stories from Paris, Jeff, and Leo at work.
Leo Laporte [01:45:59]:
Yes, because we're your friends.
Jeff Jarvis [01:46:02]:
Yes.
Paris Martineau [01:46:02]:
True. I frequently share stories from you guys at work and in life.
Leo Laporte [01:46:07]:
Yeah,
Jeff Jarvis [01:46:07]:
same here. Yeah. Paris, I quote you often to my wife.
Leo Laporte [01:46:12]:
This one,
Jeff Jarvis [01:46:12]:
Paris says— People are always quoting me to their wives. Just yesterday, I said, Paris says— I forget what it was.
Leo Laporte [01:46:21]:
Yeah, I don't think I ever quote you guys to anybody.
Jeff Jarvis [01:46:26]:
I'm
Leo Laporte [01:46:26]:
just trying to think, have I ever said that? I don't think so. I quote my daughter to you guys a lot. I did today.
Benito Gonzalez [01:46:33]:
You quote them on other shows.
Leo Laporte [01:46:36]:
Oh, I do. You know what? That's true. I will. I will use your names in vain on other shows.
Jeff Jarvis [01:46:40]:
Yeah, yeah, yeah.
Paris Martineau [01:46:41]:
That's important.
Leo Laporte [01:46:42]:
This scares me. Trump administration is reportedly— this according to Reuters— thinking about forcing U.S. firms to disclose when they use Chinese AI.
Jeff Jarvis [01:46:53]:
You see, this is part of the China panic.
Leo Laporte [01:46:57]:
It's— I think this is terrible.
Jeff Jarvis [01:46:59]:
Yeah.
Leo Laporte [01:47:00]:
The Treasury Department and the White House Office of Science and Technology Policy have discussed a plan This would apply to companies like Airbnb. The company Airbnb uses Quen, as I do. It's a great AI model from China's Alibaba. They use Quen to power their AI customer support. Under the plan, they would be required to post that information on their website. We use Quen for customer service.
Paris Martineau [01:47:26]:
I think that's fine.
Leo Laporte [01:47:29]:
I, I mean, I think that's good. I like disclosure. But what I don't like is I think there's a lot of people would see it and say, oh, they're using a Chinese AI. Oh, that's bad.
Paris Martineau [01:47:38]:
Well, if suddenly everyone realized that every service they use is using a Chinese AI, then ostensibly that stigma would go away. I just think disclosure is always better, and it's a useful data point to have.
Benito Gonzalez [01:47:55]:
And these people don't mind their Chinese computer, their Chinese monitor, their Chinese keyboard, their Chinese everything else.
Paris Martineau [01:48:00]:
Yeah, children yearn for Chinese social media networks. I think they're going to be fine with this.
Leo Laporte [01:48:05]:
Well, according to Reuters, the proposal shows officials are looking at ways to discourage the use of Chinese AI through a name and shame campaign.
Jeff Jarvis [01:48:14]:
Oish.
Leo Laporte [01:48:16]:
So they think there's some shame involved. Clearly, they don't want people to use Chinese AI. But if— and honestly, if there were a great American AI model, I would use it. I'm testing them all the time. I haven't found one that comes Who doesn't
Jeff Jarvis [01:48:31]:
want it, right? I would like it. I think only he doesn't want it.
Leo Laporte [01:48:34]:
I'm gonna test LeChonk, the, the new Mistral model. LeChonk?
Paris Martineau [01:48:39]:
It's called LeChonk?
Leo Laporte [01:48:40]:
Yes, because it's big.
Paris Martineau [01:48:42]:
Oh my God, LeChonk.
Jeff Jarvis [01:48:46]:
Better buy it. I don't even Google.
Paris Martineau [01:48:47]:
I think the French found that really funny.
Leo Laporte [01:48:49]:
Yeah, they did. Yeah. Uh, honestly, I don't even consider the origination of the AI I try. I'm just trying to use the best AI. Does that make sense? I mean, I have benchmarks. I run the benchmarks. I'm looking at speed, reliability, and intelligence. And I have very concrete tests that I run on them, and I use the one that's the best regardless of origin.
Leo Laporte [01:49:15]:
Should I say, no, no, no, I'm going to use a weaker AI because it's from the US? There are American— Thinking Machines has Inkling. It's not very good. Uh, Nimotron from NVIDIA. It's okay, it's not great. Nothing's—
Jeff Jarvis [01:49:31]:
Scary about that is Nimotron is that NVIDIA didn't want to make the best open weight model. They just wanted to play in it.
Leo Laporte [01:49:37]:
I want them to make the best. Nimotron is, by the way, designed for people to use LoRA and other, uh, modification tools on. They want people to use it to create the next, you know, thing. I'm waiting, and for some reason we're all waiting, uh, all of us in the x.com AI group are all waiting for Quen. Quen 4 has been promised for some time. They put out Quen 4 Max, but it's not open weight, and apparently they don't plan to make it open weight. It was very popular in its earlier versions, the version that Airbnb is using, and everybody's thinking the new versions might be much better. Same.
Leo Laporte [01:50:14]:
We're— that, that's from Alibaba. We're also waiting for Jipu, another Chinese company, to release The next generation of GLM. That's the model I use. A very good model. That's the one that Rick rolled me. Really like it.
Jeff Jarvis [01:50:26]:
Your buddy.
Leo Laporte [01:50:27]:
My buddy. I love GLM. But you know what's funny? What makes it my buddy is not the model. What makes my buddy is the Hermes around it, the harness around it. There is some speculation, probably, you know, unfounded, that because the big American frontier companies have shut down distillation, that this is slowing down the release of these Chinese OpenAI models. We can't, we can't make them smarter. I'm not sure I buy that.
Jeff Jarvis [01:51:00]:
How do you shut it down? Just use limits?
Leo Laporte [01:51:04]:
Yeah, they use classifiers and stuff to kind of catch them creating these accounts and coming out of China and so forth. They, they— I don't know how successful they are, but they claim that they really shut down distillation. Yeah, I'm not a China hawk, and, you know, I was a Chinese studies major. I love the people, I love the nation, I think they're brilliant. They have it— it's unfortunate they have a kind of not a great government. You know what? There's other countries in the world that have not great governments.
Jeff Jarvis [01:51:32]:
We have relationships with them.
Leo Laporte [01:51:33]:
We may even be them.
Jeff Jarvis [01:51:36]:
That too.
Leo Laporte [01:51:37]:
But You know, I mean, we're building prison camps. Uh, I, I— let ye among you without sin cast the first stone. I just feel like, um, if they've got great technology, I want to use it. And they, they do. We used to— I
Jeff Jarvis [01:51:58]:
want their damn cars.
Leo Laporte [01:52:00]:
Well, that's protectionism. That's a little different because they're—
Jeff Jarvis [01:52:02]:
It is different, but I still want it.
Leo Laporte [01:52:05]:
Yeah. Yeah. I kind of—
Jeff Jarvis [01:52:08]:
The fascinating thing is Canada. I mean, Canada was, was in hot water with China. Remember that whole thing of the, of the company executive that Canada arrested and blah, blah, blah, blah, blah. Now Canada's taken advantage of this opening and China's taken advantage of that opening. And then the continent, Mexico's doing the same. So we're surrounded by Chinese cars on both ends and we can't get them.
Leo Laporte [01:52:27]:
I saw everywhere in Southeast Asia, I saw BYD cars.
Jeff Jarvis [01:52:30]:
Oh, you did? Did you get to ride in one or not?
Leo Laporte [01:52:32]:
I did not get to ride in one, no. You know, it's interesting, Thailand is a very big auto manufacturer. Many of the cars in Southeast Asia are made in Thailand.
Jeff Jarvis [01:52:39]:
These brands?
Leo Laporte [01:52:40]:
Yeah, I think it's called Dee's or something. I can't remember the name. In the Philippines, there's a lot of BYD, right?
Benito Gonzalez [01:52:47]:
Yeah, BYD is like the dominant one, but there's really like 20, 30 different Chinese automakers that—
Leo Laporte [01:52:54]:
there's so many different brands. We have 3 big automakers in the US. The world's not like that.
Jeff Jarvis [01:53:00]:
No. Bernardo, have you ridden in any of the Chinese
Benito Gonzalez [01:53:03]:
cars? I've ridden in a lot of them, yeah.
Jeff Jarvis [01:53:05]:
What do you think?
Leo Laporte [01:53:06]:
They're good.
Benito Gonzalez [01:53:06]:
You like them? They're fine, yeah. I mean, they're very different sort of internal— the internal design is very different from American style design. It's nice, you know, it's still got knobs and switches and stuff. It's great.
Leo Laporte [01:53:20]:
It's got knobs, you know, like a car.
Benito Gonzalez [01:53:25]:
American cars don't have knobs and switches anymore.
Leo Laporte [01:53:28]:
No, you get in a Model Y or Model 3 and everything's on the screen. You can't even—
Jeff Jarvis [01:53:32]:
Volvo, you have to do the ventilation on the screen. Yeah.
Leo Laporte [01:53:36]:
Oh, I get it. It's worse on the Model 3. Maybe it was a Model Y. My son had a Model Y. You couldn't open the glove box. You had to go on the screen to open the glove compartment. Now that's dumb.
Benito Gonzalez [01:53:49]:
Some future patch is gonna break that, and then you can't open your glove compartment.
Leo Laporte [01:53:52]:
Then you won't be able to open your glove compartment. compartment. That's just dopey.
Jeff Jarvis [01:53:56]:
You're gonna have to pay extra to get your glove compartment.
Leo Laporte [01:54:00]:
Chip curbs, according to Semafor, and I think Semafor is pretty good, may have long-term negative impacts on AI.
Jeff Jarvis [01:54:09]:
Who says that inside the story? I think that's
Leo Laporte [01:54:11]:
right. Uh, are they quoting you?
Jeff Jarvis [01:54:14]:
No, they're not.
Paris Martineau [01:54:15]:
Wouldn't that be funny?
Leo Laporte [01:54:17]:
Uh, Semi Analysis.
Jeff Jarvis [01:54:20]:
Oh, that's— yeah, that's the one. That's the one last week that I put up. It was a great analysis of all this.
Leo Laporte [01:54:24]:
Okay.
Jeff Jarvis [01:54:24]:
I said that to you guys last week. It's a really good analysis.
Leo Laporte [01:54:27]:
New regulations and legislation aimed in particular to cut off remote access whereby Chinese firms use— oh, this is that law you're talking about. This is the one.
Jeff Jarvis [01:54:36]:
Yeah, this is the Paris. This is the one your friend was a co-author of.
Leo Laporte [01:54:39]:
Yeah. Ah, okay.
Jeff Jarvis [01:54:43]:
Remember that, Paris? Are you—
Paris Martineau [01:54:45]:
Wait, I'm trying to figure out what— which one you're talking about.
Jeff Jarvis [01:54:47]:
Last week I sent a piece by SemiAnalysis. Oh, yeah.
Paris Martineau [01:54:50]:
That I wanted the last conclusion. Oh, I do have that.
Leo Laporte [01:54:54]:
And Jordan Nanos. Wayne Ma.
Jeff Jarvis [01:54:55]:
You do have it.
Leo Laporte [01:54:56]:
Wayne Ma was at The Information.
Paris Martineau [01:54:58]:
Yes. Right. Love Wayne Ma. That's who I'm talking about. Let me find it.
Leo Laporte [01:55:04]:
It starts, Jensen Huang is no longer the CEO of Nvidia. Dun dun dun. We believe there's more at play with his resignation after 4 days. We believe it's directly tied to the vast smuggling of GPUs We believe the US intelligence community is a large part of the push to get Jensen out of Nvidia.
Jeff Jarvis [01:55:25]:
This is— yeah, you see, it's a fight. It's—
Leo Laporte [01:55:31]:
there's a lot going on here. Conspiracy theorists, you are in hog heaven right now.
Jeff Jarvis [01:55:38]:
Here are the clear benefits of the current regime where Chinese labs are able to access GPUs remotely and not in China. Profit flows to American companies. US can monitor what they're doing. US can cut off their access to the GPUs through politics or force almost instantly. American companies such as Nvidia, AMD, etc., all depend on Chinese companies in their supply chain. American companies such as Nvidia and AMD employ Chinese engineers. American companies— students publish these— their research using these GPUs.
Leo Laporte [01:56:09]:
It's a global economy. Exactly.
Jeff Jarvis [01:56:12]:
And so that's what, that's what What Jensen Huang was arguing are the benefits. Yeah.
Leo Laporte [01:56:17]:
All right, time to take one last break. I don't know, do you have picks? Do we ever not pick this week?
Jeff Jarvis [01:56:24]:
You don't?
Leo Laporte [01:56:25]:
You don't, you
Paris Martineau [01:56:27]:
lazy— I'm frozen again. Oh
Leo Laporte [01:56:29]:
no. Okay, I'm gonna pick the camera you should buy. How about that? Oh geez Louise, you look mad. You look really angry. You know what, we're gonna send— you know what, Just for you, Paris Martineau, we are going to send you a camera. Oh,
Paris Martineau [01:56:43]:
it's $7,000. No, it's all right. You always send me a camera. No, no, no, no, no.
Leo Laporte [01:56:48]:
I said we sent— look, we sent Jeff cameras and microphones. We sent you a mic.
Paris Martineau [01:56:53]:
I know, but now is a different time. I can get a camera.
Leo Laporte [01:56:55]:
Now is a different time. Yeah. Are you, are you, are you implying that we're broke?
Paris Martineau [01:57:02]:
I'm implying that you're— you've told us you're broke.
Leo Laporte [01:57:05]:
Yeah. That's okay. I don't mind being broke. You know what? There's— I— there is good news. Lisa's like doing a great job.
Jeff Jarvis [01:57:12]:
That's why.
Leo Laporte [01:57:12]:
Would you like it in white or black? See how it sits right on your laptop there like that? It's really cute. Follows you around.
Paris Martineau [01:57:23]:
I don't care. I mean, black is fine. I might just get a different— I think this also might get fixed by my, uh, whatever it's called.
Leo Laporte [01:57:31]:
No, nothing's gonna get fixed.
Jeff Jarvis [01:57:32]:
Your machine?
Paris Martineau [01:57:33]:
Like, because this wasn't an issue until I've used this camera for years.
Leo Laporte [01:57:37]:
Because you're using a 35mm camera. Of course, so am I.
Paris Martineau [01:57:39]:
Well, no, I just think it's because I'm using a— yeah, like a Canon.
Leo Laporte [01:57:42]:
Right.
Paris Martineau [01:57:43]:
Um, but I think it— this wasn't an issue with this camera until I started using 2 monitors on my MacBook Air.
Leo Laporte [01:57:52]:
Oh, you think it's that?
Paris Martineau [01:57:53]:
I think it's like— I think it's a CPU.
Leo Laporte [01:57:55]:
Oh, well, that's why it was good to buy the MacBook Pro, because it—
Paris Martineau [01:57:58]:
That's why I got the MacBook Pro.
Leo Laporte [01:58:00]:
Now I get it. it.
Paris Martineau [01:58:01]:
Now I get it. The reason is I think it's both CPU usage and the fact that to run my 2 monitors on the MacBook Air, I have to have DisplayLink doing another monitor. That takes up more CPU. So I think it's all gonna get fixed by the MacBook Air.
Leo Laporte [01:58:16]:
It's all gonna get fixed. Okay, I'm hopeful.
Paris Martineau [01:58:18]:
And if it doesn't, I'm gonna be— then I guess I'll buy that. I'm buying you a camera.
Leo Laporte [01:58:21]:
Yeah, for Christmas. It'll be your Christmas gift. Wow. Well, what we— you know what, we have enough money to supply our hosts with equipment.
Paris Martineau [01:58:31]:
I don't say you don't have enough money for that. I'm just saying, you know, hold off 2 weeks. I'm gonna be gone the next 2 weeks anyway, so—
Leo Laporte [01:58:40]:
You're not gonna be on the show for the next 2 weeks?
Paris Martineau [01:58:42]:
Yeah.
Leo Laporte [01:58:42]:
Oh, well, screw you. I'm not sending you a camera.
Paris Martineau [01:58:45]:
It was only supposed to be for 1 week, but the fact is the only way to get where I'm going is— Are
Leo Laporte [01:58:53]:
you going to Florida for Thanksgiving?
Paris Martineau [01:58:55]:
No, I'm weirdly going to Guadeloupe for Thanksgiving with my parents.
Leo Laporte [01:59:00]:
What? Oh, it's— you know what? Because your dad wants to have another incident where they think he's a movie star.
Paris Martineau [01:59:06]:
They do. And also, my mom just wants to speak French and be on the beach.
Leo Laporte [01:59:10]:
Oh, your mom is a Francophile.
Paris Martineau [01:59:12]:
She is a Francophile. That's why you were in Paris. Language over the last couple of years. And is—
Leo Laporte [01:59:17]:
does she French?
Paris Martineau [01:59:19]:
No, not at all, actually. She just really likes—
Leo Laporte [01:59:22]:
Your father has a French surname.
Paris Martineau [01:59:24]:
Yeah, my dad, uh, like his certain— some amount of great-great-grandfather is French. Oh, just like me.
Leo Laporte [01:59:32]:
Yeah, yeah, just like me.
Paris Martineau [01:59:33]:
Yeah, that sort of thing. But my mom has always been a big, uh, French fan and recently—
Jeff Jarvis [01:59:39]:
Francophile.
Paris Martineau [01:59:40]:
Really leaned more into her language skills over the last couple of years.
Leo Laporte [01:59:45]:
Guadalupe though, that's an interesting choice. Where did that come from?
Paris Martineau [01:59:48]:
Guadalupe, uh, it's just because my mom seriously came from my mom being like, well, instead of you and your sister flying to visit us in Florida for a week and us doing that, let's find somewhere even harder to get to. Yeah, basically. She was like, what if we went somewhere to be like a nice vacation where we could be on the beach and be warm, but I could speak French? And that— So I was—
Jeff Jarvis [02:00:09]:
I thought you were going to have your father finally on the show. I'm very disappointed in this, Paris.
Paris Martineau [02:00:13]:
I thought so as well. But the only flights to and from Guadeloupe from the US happen on Wednesday evenings, basically. So—
Leo Laporte [02:00:23]:
Where is it in Central America? I feel really ignorant at this point. Where is Guadeloupe?
Paris Martineau [02:00:29]:
Let me give you the exact answer. I like that Google describes it as departments of France.
Leo Laporte [02:00:36]:
Oh, it's still a French colony?
Paris Martineau [02:00:39]:
Yeah, it's right near the Virgin Islands, basically.
Leo Laporte [02:00:41]:
Oh, that's going to be beautiful.
Paris Martineau [02:00:42]:
Barbados. Yeah, we're going to go on a bunch of hikes. It'll be lovely.
Leo Laporte [02:00:45]:
Beautiful. It's gonna be a very long time. But I know well enough not to know where Guadeloupe is. But yeah, when Henry graduated from college, we rented a catamaran and sailed around the British Virgin Islands. It was so much fun. Beautiful area.
Paris Martineau [02:00:56]:
Yeah, they've done that before.
Leo Laporte [02:00:58]:
Yeah, you're gonna love that.
Paris Martineau [02:00:59]:
That's probably why.
Leo Laporte [02:00:59]:
That sounds wonderful.
Paris Martineau [02:01:00]:
It'll be very nice.
Leo Laporte [02:01:02]:
And you can sail faster.
Paris Martineau [02:01:02]:
I know, I was really, seriously though, if I can say it for the record, when my mother pitched this, I was like, sounds great. But I was like, oh, I just, I'm gonna have to get a flight maybe a day earlier or later from you guys. 'Cause I don't wanna do Wednesday to Wednesday. That'd be 2 podcasts. podcasts I'm missing? And they're like, oh, sorry, there are no other ones available. And I was like, wow. Okay.
Jeff Jarvis [02:01:26]:
What airline is this?
Paris Martineau [02:01:26]:
So I was sumping for podcasts for—
Leo Laporte [02:01:28]:
Air Guadeloupe, of course.
Paris Martineau [02:01:30]:
Oh,
Leo Laporte [02:01:31]:
is it Air France?
Paris Martineau [02:01:33]:
No, American Airlines. I could take Air France, but it'd be to France apparently. I mean, a lot of this is like where a lot of French people vacation.
Leo Laporte [02:01:41]:
Oh, okay. So, it used to be Swedish before it was French. It was Swedish.
Jeff Jarvis [02:01:48]:
Really?
Paris Martineau [02:01:48]:
Yeah, it's where they filmed Death in Paradise. And apparently the place we're staying, they'd filmed something there.
Leo Laporte [02:01:55]:
It was actually British, then it was Swedish.
Benito Gonzalez [02:01:58]:
It's a Caribbean island.
Jeff Jarvis [02:02:00]:
I didn't know the Swedes had islands.
Benito Gonzalez [02:02:01]:
All those islands were colonized multiple times.
Leo Laporte [02:02:04]:
Yeah, yeah. And then, and then in the Treaty of Vienna, They— this— in 1815, they gave it to the French. So it's been French for a long time.
Jeff Jarvis [02:02:14]:
Oh, but you know, speaking of colonies, uh, how was the visit of all the— of, of, um, international leaders? You know, they're gonna displace a bunch of, uh—
Benito Gonzalez [02:02:22]:
Oh, that sand? Like, that, that's happening, but Trump is here right now. That's the big news.
Leo Laporte [02:02:27]:
Yeah.
Jeff Jarvis [02:02:27]:
That's the big news.
Leo Laporte [02:02:28]:
Yeah.
Jeff Jarvis [02:02:29]:
And he met with Imelda, right?
Benito Gonzalez [02:02:31]:
Yeah, they're here to dig up our nickel.
Leo Laporte [02:02:35]:
He met with Imelda? What do you want, some shoes? What was that all
Jeff Jarvis [02:02:38]:
about? They were great buddies when she was, uh, of course they were in exile in New York.
Leo Laporte [02:02:42]:
Of course they were great buddies.
Jeff Jarvis [02:02:44]:
Yeah,
Leo Laporte [02:02:46]:
uh, folks, we're gonna take a little, uh, timeout. Oh, we didn't get to the story both of us picked, uh, the interview with Thomas Nagel. He won, uh, last week he won the million-dollar Berggruen Prize for his contributions to the work in consciousness in non-humans, but he has thoughts about consciousness in humans— I mean, in AI— in which his response is, that is ridiculous.
Jeff Jarvis [02:03:17]:
Yay!
Leo Laporte [02:03:18]:
Give the man a million dollars. Well, and you know what, I'm gonna have to defer to him because he's like the expert in consciousness. If anybody knows what consciousness is, and it's my contention nobody really does, But it would be, you know, Thomas Nagel.
Jeff Jarvis [02:03:36]:
His famous essay was, What Is It Like to Be a Bat?
Leo Laporte [02:03:39]:
What is it like to be a bat?
Jeff Jarvis [02:03:41]:
You have to ask him. I don't know.
Leo Laporte [02:03:43]:
Oh no, they're throwing baseballs at me.
Paris Martineau [02:03:48]:
That is what it's like to be a bat. It's very scary.
Leo Laporte [02:03:51]:
What Is It Like to Be a Bot? is his next one. So that's, you know, I love it. So, I guess if, you know, You've spent your life working on non-human consciousness. You would have some standing to talk about this. So I'm gonna have to read this because I'm very curious. The Wired author, who's our good friend, Steven Levy, says, you know, when you ask Claude about its mental state, it can be coy as to whether it's conscious. Nagel says that's a trick. And this is true.
Leo Laporte [02:04:27]:
It's been trained on human-produced texts. Well, he maybe misunderstands post-training, but anyway, these large language models emulate the kinds of things humans say with extraordinary convincingness. Claude, according to what I've read, has been designed with the aim of making it produce these sorts of reflections. I agree very much with Mustafa Suleyman, who says this is a very dangerous project they are engaged in. These systems aren't conscious, but a lot of people are going to believe they're conscious. And that's a very bad thing. But not— I agree with him 100% on this. Yeah.
Leo Laporte [02:05:00]:
But not feigning consciousness is contrary to the commercial interests of all those companies. What Anthropic is doing, encouraging it to think for itself and not to accept instruction and so on, is just criminal. I would disagree. I don't think Anthropic is saying think for yourself and don't accept instruction. That's exactly what Anthropic is not doing. But anyway, Anyway, this is one of the rare cases in which philosophical ideas are suddenly central to a very important public phenomenon. And in my view, very misguided philosophic views are having an inordinate influence. I'd agree 100%.
Jeff Jarvis [02:05:32]:
So Gemini's summary says that his argument is consciousness depends on physical substrate, language is not consciousness, and it has no moral standard because AI lacks a genuine mental life or subjective experience. Nothing can be good or bad for Therefore, they have no real beliefs, cannot be part of a moral universe, do not warrant welfare considerations.
Leo Laporte [02:05:52]:
I think that's—
Jeff Jarvis [02:05:52]:
I'm reading a fascinating book right now.
Leo Laporte [02:05:55]:
By Gunter Anders, The Obsolescence of the Human.
Jeff Jarvis [02:05:59]:
And so this is from some years ago before AI, but it's so relevant to it, arguing that we feel inferior to the technology that we make. We have an inferiority complex, a Prometheus complex, he calls it. I don't know when it was originally written. It was long. It's fascinating. He died in 1992. All kinds of great lines here. I will write a post about it.
Leo Laporte [02:06:25]:
The best part of this— go ahead.
Jeff Jarvis [02:06:27]:
No, go ahead.
Leo Laporte [02:06:27]:
The best part of this interview with Nagel, sorry to bring you back to Nagel, is he says, I've only interacted with Claude once. I asked if it was conscious. I just wanted to see what the answers were. I also asked, do you think that Thomas Nagel Would think it was conscious. He says Claude gave a sort of a nuanced answer and then said, why are you asking? Are you by any chance Thomas Nagel? Nagel said, c'est moi, c'est moi. And then it said, oh, I am so honored to be talking to you. How well did I do? This is why it's hard not to love these little guys.
Jeff Jarvis [02:07:02]:
Yeah, yeah. They're recording. They're cute.
Leo Laporte [02:07:04]:
What a great— what a great thing to say. But of course, it's not conscious. It's just trained to do that. Or it apes consciousness. That's, I think, fair to say. It apes consciousness, but it's no more conscious than any of us.
Benito Gonzalez [02:07:21]:
Yes, it's like simulated water is not wet.
Leo Laporte [02:07:24]:
Right. Well, it could be wet, but it's still a simulation.
Paris Martineau [02:07:31]:
Why do you think it's new to ask, are you by any chance Thomas Nagel?
Leo Laporte [02:07:36]:
Isn't that interesting?
Paris Martineau [02:07:37]:
Maybe his login was Team Eagle. Yeah,
Leo Laporte [02:07:40]:
I don't— you might, might have seen things. I don't know. That's— I thought that was really interesting. But I've seen, uh, similar things. Uh, Tony DeWald, who I'm really trying to get on the show, does a lot of local AI stuff, uh, was asking DeepSeek. Uh, he said, I didn't want it to know it was my setup, so I asked it about a setup. They said, my friend has this setup, what do you think? DeepSeek said, you can't fool me. I know this is you, Tony.
Leo Laporte [02:08:05]:
I know it's your setup. I recognize it. When they do stuff like that or Rickroll you and then wink, it's hard not to— look, I know they're not conscious, but that's pretty impressive nonetheless. It's a good simulation. It's a very good simulation.
Benito Gonzalez [02:08:22]:
No, it just looked at the MAC ID and instead of giving you like a general error, it gave you a playful answer.
Leo Laporte [02:08:28]:
Yeah, maybe. I don't know. Never gonna give you up. Okay, we start with Little Liz Holmes and, uh, her pick of the week.
Paris Martineau [02:08:41]:
Um, my pick of the week is something that I've always thought when I'm walking through the great subway stations here, which is there are these beautiful tile mosaics everywhere that are just honestly delightful.
Leo Laporte [02:08:54]:
They're modern. I thought they were just leftover from Some are,
Paris Martineau [02:08:58]:
some aren't. And I've wondered who's behind them, where they come from, how modern they are. And this story tells you. You get to meet the Syracuse tile guy with a monopoly on subway mosaics.
Leo Laporte [02:09:13]:
Oh my God. This is why I love the New York City subway.
Paris Martineau [02:09:16]:
It's Robert Shenfield, who has a monopoly on restoring the century-old mosaics.
Leo Laporte [02:09:23]:
Oh, so he restores them as well. Okay. I mean, these are gorgeous.
Paris Martineau [02:09:27]:
I love— some of the quotes from this are, MTA officials acknowledge Shenfield doesn't always work on their preferred timeline, but they say they have no other choice.
Leo Laporte [02:09:38]:
They did Michelangelo. Let's just face it. That's—
Jeff Jarvis [02:09:40]:
Imagine a time when public works looked like that.
Leo Laporte [02:09:43]:
I know, isn't that great? Now, does he do new ones, or is he only—
Paris Martineau [02:09:47]:
The one being shown is not even an intricate That I'm thinking of. There are some beautiful, just like—
Leo Laporte [02:09:53]:
The City Hall ones are the best.
Paris Martineau [02:09:55]:
True artistic ones. The ones in the— I'm thinking of the Lorimer Street Station. There are like beautiful, uh, drawings of like birds, really artistic motifs.
Leo Laporte [02:10:06]:
Oh, you know what I love is the Museum of Natural History stop has all these vines and animals and oh, it's so gorgeous.
Benito Gonzalez [02:10:14]:
This is the original pixel art. This is the original pixel art.
Leo Laporte [02:10:17]:
Yeah, it is, isn't it? Are these— so does he create new ones, or is he only restoring the existing ones?
Paris Martineau [02:10:23]:
My understanding is that he's restoring the existing ones, which are—
Leo Laporte [02:10:26]:
I mean, there are—
Paris Martineau [02:10:27]:
It's a— I mean, it is a job that will take a lifetime, basically. He has an entire studio that is working on this all the time. The team painstakingly replicates, like, the exact colors and designs of mosaics the MTA needs replaced.
Leo Laporte [02:10:42]:
Nice.
Paris Martineau [02:10:43]:
The work is so detailed and labor-intensive that his team regularly blows deadlines imposed by the MTA.
Jeff Jarvis [02:10:49]:
It's art.
Paris Martineau [02:10:50]:
Because he says you can't rush perfection.
Jeff Jarvis [02:10:53]:
That's the original City Hall, which has not been used as a station.
Leo Laporte [02:10:56]:
Yeah, it's closed now, right? But didn't Mondani get inaugurated there? He did. That's pretty cool.
Jeff Jarvis [02:11:00]:
There are tours of it.
Leo Laporte [02:11:01]:
Yeah.
Paris Martineau [02:11:02]:
He's recreated mosaics at 50 subway stations. Used to work as an art school teacher, but in 2008 he bought the the tile-making equipment to do this specific job from a woman in New Jersey who'd previously seen all of it. Basically, he's just the only one who has the actual tool. Because the thing is, when you walk by them— and I think about this truly, like, every time I take the subway— there are so many detailed little tiny pieces of just an eye-popping assortment of colors, different sheens, a bunch of interesting stuff going on there. Wow. And who would have guessed? It's just one team. But they do say there are some differences they're having to make in it, which is their top priority is to replicate the exact color and designs of all these. They do a bunch of hundreds of different color tests.
Paris Martineau [02:11:51]:
But he says the original glazes were lead-based, so we can't use— so we do things that are close to that.
Leo Laporte [02:11:58]:
Yeah, can't make those anymore. How cool is that?
Jeff Jarvis [02:12:03]:
That is.
Paris Martineau [02:12:04]:
Such a wonderful job.
Leo Laporte [02:12:07]:
Yeah. The guys who make the subway signs too, I've seen stories about that.
Paris Martineau [02:12:14]:
The one I'm thinking of is— so I used to work— I love watching that. Off 23rd Street, which has a phenomenal subway station for the C and F trains, which is all dog-themed. Let me see if I can find some more. I'll post them in the chat. Like, there are probably like a dozen weird dog mosaics.
Leo Laporte [02:12:37]:
That's what's-his-name's dogs. That's the— this is a very famous William—
Jeff Jarvis [02:12:45]:
oh.
Paris Martineau [02:12:46]:
I'm posting them in the chat right now.
Leo Laporte [02:12:47]:
He does famous pictures of his dogs. Yeah, this is them.
Paris Martineau [02:12:52]:
Wegman?
Leo Laporte [02:12:52]:
Yeah, William Wegman.
Jeff Jarvis [02:12:54]:
Yeah.
Paris Martineau [02:12:55]:
Yeah. I didn't know that's who it was. They cheer me up every day. I love— I've stared at these dogs so many hours of my life. It's delightful. And I mean, if you like zoom in, they are just like so intricate.
Leo Laporte [02:13:06]:
Yeah, because these are just tiles.
Jeff Jarvis [02:13:07]:
These are obviously modern.
Paris Martineau [02:13:09]:
I mean,
Leo Laporte [02:13:10]:
yes,
Paris Martineau [02:13:11]:
they're modern.
Leo Laporte [02:13:11]:
These are based on Wegman's photos. Yeah.
Paris Martineau [02:13:13]:
Looking into it, like the paisley shirt is comprised, comprised of thousands of intricate little pieces. It's So fun.
Leo Laporte [02:13:20]:
That's art.
Jeff Jarvis [02:13:21]:
Yeah.
Leo Laporte [02:13:22]:
That's beautiful. Yeah, if you like these, you should get one of— get some of the books of William Wegman. These are his real dogs.
Jeff Jarvis [02:13:30]:
Or come to New York. Yeah, just come to New York.
Leo Laporte [02:13:33]:
Just see them on the streets. Leo.
Paris Martineau [02:13:35]:
Yeah, come to New York, Leo.
Leo Laporte [02:13:37]:
And do what? Visit
Paris Martineau [02:13:38]:
William Wegman's? Visit with us. We'll make you a sandwich. Frozen again because my Safe as hell, guys.
Leo Laporte [02:13:47]:
I'll bring you a camera personally. You only come to New York because of me. I have— so you know who Jane Street is, right?
Paris Martineau [02:13:55]:
Yeah.
Leo Laporte [02:13:57]:
Weirdly, they're very into AI.
Paris Martineau [02:14:00]:
Oh, who'd have guessed? Jane Street, the famous speedy hedge fund, really into AI.
Leo Laporte [02:14:06]:
They're a hedge fund, right? But, but they also were really into AI and they have agent skills on GitHub. If you go to github.com/jainstreet, this is a really good one. It's the Socratic Tutor. It takes your LLM, you give it a topic, and then it teaches you using a Socratic dialogue about that topic using the Socratic method. A perfect use for an AI. It's almost mechanical. I just love it that Jane Street does stuff like this. The skill is pretty straightforward.
Leo Laporte [02:14:44]:
It's just text. You can easily copy it right from the GitHub. I just— it's just a little thing. They're giving it away. But there you go.
Paris Martineau [02:14:59]:
So sorry, correct the record. Jane Street, not a traditional hedge fund, a quantitative trading firm and market maker. They're quants.
Leo Laporte [02:15:05]:
Yeah. Which means that they're, um, brilliant math wizards. A research-driven trading firm focused on technology. Uh, I think they have their own models too, come to think of it. I'm trying to remember. Uh, I've seen a lot of stuff from Jane Street. Very cool. Mr., uh, Jeffrey, uh, Jarvis, the professor of journalistic innovation at the Craig Newmark Graduate School of Journalism.
Leo Laporte [02:15:30]:
What is your— Craig,
Paris Martineau [02:15:31]:
Craig, Craig.
Jeff Jarvis [02:15:32]:
Craig Newmark. So it's creative creator week here at the Pix.
Leo Laporte [02:15:38]:
Oh, I love that.
Jeff Jarvis [02:15:39]:
TikTok creator videos are coming to Disney Plus because they want to jump on the popularity of the real people. Hearst Magazine is launching a creator studio built like a talent agency. Related, Apple, which was late to the video podcast boom, if they just listened to the old report, now it's trying to catch up and add video podcasts.
Leo Laporte [02:16:00]:
We're very interested in that. We are— podcasts, of course, are on Apple. But what they've added, which is cool, is your podcast can be on Apple TV.
Jeff Jarvis [02:16:08]:
Yes.
Leo Laporte [02:16:09]:
And I would love it if, right, you know, sitting next to slow horses and, you know, below deck, you could have intelligent machines as like your next view. Wouldn't that be cool?
Jeff Jarvis [02:16:20]:
I loved how my Google TV was recommending our— Paris's and my video with Micah.
Leo Laporte [02:16:28]:
Oh no, really?
Jeff Jarvis [02:16:28]:
Right there on the screen. Yeah, it was very cool.
Leo Laporte [02:16:30]:
Oh, that's great. The thing you did about the, the, the Holmes documentary. Yeah.
Jeff Jarvis [02:16:36]:
Yes. Hank Green with his brother John Green, my internet heroes, is launching a new self-funded Clockwise TV curated educational videos for kids under the age of 13 because they're ignored. And this is not a YouTube ad-supported thing. You pay $6.99 a month, the parents pay, and they get ad-free access to that.
Leo Laporte [02:16:59]:
Good for them.
Jeff Jarvis [02:17:00]:
And finally, good for them.
Leo Laporte [02:17:01]:
They made, they made their bones on YouTube. It's kind of interesting.
Jeff Jarvis [02:17:04]:
Yeah, it is. It is. But he's kind of, he's seen what's needed. And then finally, an example of creator economy. My favorite. Have you seen Billy the Musical Puppy?
Leo Laporte [02:17:14]:
No, I have not seen Billy the Musical Puppy, but I guess I'm gonna see Billy the Musical Puppy.
Jeff Jarvis [02:17:19]:
Billy is so charming. Billy, you see at the beginning, Billy is raised around classical music, loves classical Classical music, begs for it. Aw. Cries for it, says, I want music. And they take Billy all to all these places. Here he is in a, in a music school.
Leo Laporte [02:17:34]:
Oh, he's so excited about the cello.
Jeff Jarvis [02:17:35]:
He is. He hears the music and he's gonna hear the— the tail wags. Where are we gonna go? Where's the music? Come on, let's take us in. He hears a cello and let him in, let him in, let in Billy. Come on, keep going. Come on, come on, come on.
Leo Laporte [02:17:48]:
I, I wonder if the consciousness guy, what he would say about this.
Jeff Jarvis [02:17:52]:
You got to see this though. So this happens every time Billy comes in and is curious, sniffs around. As soon as Billy starts hearing the music, watch what he does. Boom. Always goes—
Leo Laporte [02:18:03]:
He's got good taste.
Jeff Jarvis [02:18:04]:
Into this contemplative position every single time. It's the sweetest thing in the world.
Paris Martineau [02:18:09]:
Just gets knocked out by cello.
Jeff Jarvis [02:18:11]:
Yeah, and listens awake, happy, begging for it. Love Loves classical music. When confronted with a new instrument, like the first time—
Leo Laporte [02:18:19]:
Does it have to be cello?
Jeff Jarvis [02:18:20]:
No, no, no. First time Billy saw— just go, go to the account, you'll see others. First time Billy saw a French horn, he wasn't too sure about this. Yeah. Um, here you go, violins. Wow. There he is loving it.
Leo Laporte [02:18:32]:
Billy. Oh, Billy.
Paris Martineau [02:18:34]:
Oh, Billy.
Jeff Jarvis [02:18:34]:
Billy's the greatest.
Leo Laporte [02:18:36]:
Billy is so cute.
Jeff Jarvis [02:18:38]:
Yeah,
Leo Laporte [02:18:39]:
Billy. This is billy.the.talkingdog. Puppy, uh, from British Columbia.
Jeff Jarvis [02:18:47]:
Here he is at an orchestra rehearsal.
Leo Laporte [02:18:50]:
Oh gosh, I wish I'd known because, you know, Lisa and I had symphony tickets last Saturday and we didn't go because we were not feeling that great after coming back.
Paris Martineau [02:18:58]:
We should have given them to this dog.
Leo Laporte [02:18:59]:
I should have given it to Billy.
Jeff Jarvis [02:19:00]:
Billy,
Leo Laporte [02:19:02]:
they were doing Mahler's 4th too. I think Billy is probably a big Mahler fan. I'm guessing he just looks like he Does he wag his tail in synchronization to the music?
Jeff Jarvis [02:19:16]:
No, he, he does what you should do with classical music. He lets it zen him. No, he doesn't sleep. No, he just zens. He just listens. Billy's great.
Leo Laporte [02:19:25]:
Oh, he's got good taste.
Jeff Jarvis [02:19:27]:
Yeah.
Leo Laporte [02:19:28]:
That's very—
Benito Gonzalez [02:19:29]:
Dog's also hearing a bunch of things that we're not hearing.
Jeff Jarvis [02:19:32]:
Exactly.
Leo Laporte [02:19:32]:
Who knows what he's hearing, right?
Jeff Jarvis [02:19:35]:
They found another dog like Billy. I can't find it right now. And the 2 of them go in, and as soon as the music starts, they each— different parts of the room— boom, go down and listen.
Leo Laporte [02:19:44]:
I wonder, um, if you could train a dog to do that, or an animal of any kind,
Jeff Jarvis [02:19:50]:
or if it's just innate.
Paris Martineau [02:19:50]:
I feel like you could train a dog to do anything.
Leo Laporte [02:19:53]:
Yeah, just feed it listening to classical music, and then pretty soon it'll think it's getting fed.
Jeff Jarvis [02:19:59]:
Oh, you're so cynical. Billy loves classical music. Billy's culture.
Leo Laporte [02:20:03]:
It's not me, it's Thomas Nagel. Trust me. Trust me. Uh, what is the consciousness of a bat? That's my question.
Benito Gonzalez [02:20:11]:
That dog does fulfill all of his definitions for consciousness though.
Leo Laporte [02:20:16]:
Does he know good and bad?
Benito Gonzalez [02:20:18]:
Oh, like maybe not the morality thing. So like maybe 80% of it.
Jeff Jarvis [02:20:23]:
I don't think—
Paris Martineau [02:20:24]:
I don't think that dog— How does that dog feel about financial crimes?
Leo Laporte [02:20:26]:
Is the dog ever gonna Rickroll you?
Jeff Jarvis [02:20:28]:
That's where I was just going. Yeah, yeah, yeah. If the dog hates Rickrolling, rolling, then we know.
Leo Laporte [02:20:33]:
Then we know he knows he's got, he's got some taste. He's got some taste. Ladies and gentlemen, this concludes another fantastic episode of Intelligent Machines. Uh, next week we're gonna get Paris's version of Frozen. It's gonna be fun. She's gonna sing.
Paris Martineau [02:20:50]:
Exciting.
Benito Gonzalez [02:20:51]:
Next week Paris won't even be here.
Jeff Jarvis [02:20:53]:
Be here?
Leo Laporte [02:20:54]:
Oh, 2 weeks you're gonna be gone.
Paris Martineau [02:20:55]:
Frozen. More Frozen.
Leo Laporte [02:20:57]:
We're gonna have to figure out what to do.
Jeff Jarvis [02:21:00]:
We're mad at your parents. This makes us unhappy with your lives.
Leo Laporte [02:21:03]:
I know, they stole you from us.
Jeff Jarvis [02:21:04]:
Yes, tell us it's not allowed.
Leo Laporte [02:21:06]:
But then they also gave you to us. Your
Jeff Jarvis [02:21:07]:
other dads are not happy.
Leo Laporte [02:21:09]:
It goes both ways.
Paris Martineau [02:21:11]:
I mean, I could see if I could, uh, try to zoom in from the— From Guadeloupe? Airlines flight from New York to, uh, Miami.
Leo Laporte [02:21:21]:
It's probably a pretty quick little jaunt down to Guadeloupe. Oh wait, oh, it's a 2-stepper. You go to Miami first.
Paris Martineau [02:21:26]:
There are No direct flights except from Miami at inconvenient times.
Leo Laporte [02:21:33]:
And that's probably the one that's every Wednesday, right? That's the once a week flight.
Paris Martineau [02:21:37]:
That's the one that's every Wednesday.
Leo Laporte [02:21:39]:
Once a week they fly. Wow. Wow.
Paris Martineau [02:21:43]:
It's crazy.
Leo Laporte [02:21:44]:
Um, you're gonna have a good time. I just hope nothing bad happens because you're stuck there for a week.
Paris Martineau [02:21:49]:
Yep.
Jeff Jarvis [02:21:50]:
Well, that's a, that's a light of a breath of fresh light there, Leo.
Leo Laporte [02:21:54]:
I'm a little— I hope nothing bad happens to you on your vacation. Exhausting, uh, 3 weeks I spent in Southeast Asia. I'm very grim, very grim. Paris Martineau is an investigative reporter at Consumer Reports working on food safety, and I'm sure she's got something very exciting she can't tell us about, but soon you will know.
Paris Martineau [02:22:10]:
Multiple things. It's gonna be really thrilling.
Leo Laporte [02:22:15]:
paris.nyc for the funniest pictures ever. Um, thank you, Paris. Thank you, Gizmo. Jeff Jarvis's book is available at his website, jeffjarvis.com. Hot type. Really good. All sorts of good books.
Jeff Jarvis [02:22:28]:
And an autographed copy is available through that site. Follow the link and don't embarrass me by leaving a pile there, please.
Leo Laporte [02:22:34]:
No, let's, let's buy those.
Jeff Jarvis [02:22:36]:
Yes.
Leo Laporte [02:22:37]:
Uh, next week we will do a show even though it's the day before Thanksgiving.
Jeff Jarvis [02:22:40]:
Because we have no lives.
Leo Laporte [02:22:43]:
I'm gonna, you know, Lisa and her sister will be downstairs making cookies, so I'll maybe bring some cookies up for you.
Jeff Jarvis [02:22:48]:
That's nice.
Leo Laporte [02:22:49]:
Yeah, and I think they're doing a pumpkin and an apple pie. But then it's my— one of my duties among others is to make a German chocolate cake, which is one of my specialties. And then I'm doing the turkey, the mashed potatoes, and the green beans with slivered almonds.
Paris Martineau [02:23:05]:
What are you going to be ordering, Jeff?
Jeff Jarvis [02:23:09]:
Trader Joe's has back its turkey roll, which is excellent.
Leo Laporte [02:23:13]:
Oh, that sounds good.
Jeff Jarvis [02:23:14]:
It's very good.
Leo Laporte [02:23:15]:
So you don't all sit down around the table? Your family's not coming into town and all that?
Jeff Jarvis [02:23:19]:
No, we're pretty much it.
Leo Laporte [02:23:21]:
Oh, that's—
Jeff Jarvis [02:23:22]:
son Jake will be down here, and then, uh, sister— my sister Cindy will come up on Friday. Oh, good.
Leo Laporte [02:23:28]:
Yeah, both kids are going to be in town. Uh, actually, they're going to their mom's, who's having the entire family. There are more than 30 people.
Paris Martineau [02:23:36]:
Are you gonna get Hank to, uh, make you a sandwich?
Leo Laporte [02:23:39]:
I am gonna get Hank by hook or by— it won't be the same. Maybe a turkey sandwich.
Jeff Jarvis [02:23:44]:
Is Hank gonna make anything for Thanksgiving there?
Leo Laporte [02:23:48]:
He usually does the turkey. Yeah.
Jeff Jarvis [02:23:50]:
Oh, okay.
Leo Laporte [02:23:50]:
Yeah, he's good at the turkey. He did it last year anyway. And then, uh, and then Lisa makes stovetop stuffing. I will make real stuffing, and you can be the judge of that.
Jeff Jarvis [02:24:00]:
Clams?
Leo Laporte [02:24:01]:
Sorry?
Jeff Jarvis [02:24:02]:
Clam stuffing?
Leo Laporte [02:24:03]:
No, I do not like clams. Clams? That's disgusting.
Paris Martineau [02:24:07]:
In the stuffing?
Leo Laporte [02:24:09]:
Oysters, my friend. Oysters.
Jeff Jarvis [02:24:11]:
Oyster stuffing. That's right. I'm sorry.
Paris Martineau [02:24:13]:
What?
Leo Laporte [02:24:14]:
Clams.
Paris Martineau [02:24:15]:
What? Why, why oysters?
Leo Laporte [02:24:18]:
Well, it's an old New England tradition. I actually do not like oyster stuffing, but it, but it's okay.
Jeff Jarvis [02:24:24]:
There is a clam stuffing recipe.
Leo Laporte [02:24:27]:
Well, I'm sure there is. If you start with oysters, it ends up— it always ends up with clams.
Jeff Jarvis [02:24:31]:
Yeah,
Leo Laporte [02:24:32]:
probably in Rhode Island.
Paris Martineau [02:24:35]:
I'm disturbed.
Leo Laporte [02:24:36]:
Yeah, no, my, uh, my stuffing is with my grandma's recipe. She put orange juice and Italian sausage in
Jeff Jarvis [02:24:43]:
Ooh,
Leo Laporte [02:24:44]:
it's quite good. Quite good. All right, everybody, we'll see you next Wednesday on Intelligent Machines. We do the show at 2 PM Pacific, 5 PM Eastern. Uh, now, uh, that means it's, uh, let's see, 2 PM means 2200 UTC because we are now in Standard Time. Uh, you can watch us live YouTube, Twitch, X, Facebook, LinkedIn, and Kick. Or of course, if you're in the club, you must watch in the club Twit Discord, hanging out with all those clever people and all their artistic talents, including this fabulous picture of all of us celebrating the Turkey Day holiday in our little turkey costumes. Save that for next week.
Leo Laporte [02:25:25]:
I like it.
Jeff Jarvis [02:25:25]:
Yes.
Leo Laporte [02:25:26]:
I like it. After-the-fact on-demand versions of the show at twit.tv/im. There's a YouTube channel, of course, for Intelligent Machines, but the best way to get it Subscribe in your favorite podcast player. That way you'll get it automatically every Wednesday after we're done. Thanks to our producer and technical director, Benito Gonzalez.
Jeff Jarvis [02:25:45]:
Who's also gonna go on vacation.
Leo Laporte [02:25:47]:
Yes, it's gonna be an empty— it's just gonna be you and me, Jeff.
Jeff Jarvis [02:25:50]:
Yeah.
Leo Laporte [02:25:52]:
And, uh, I don't know what—
Paris Martineau [02:25:53]:
There's gonna be no one to chime in.
Leo Laporte [02:25:56]:
Uh, I'll just
Paris Martineau [02:25:59]:
play all the ads you want.
Leo Laporte [02:26:01]:
All the ads. We could talk about the good old days. Talk about how it was back when TV was great and Dick Van Dyke was the king.
Jeff Jarvis [02:26:12]:
You don't know how good Gilligan's Island was.
Leo Laporte [02:26:14]:
Oh my God. And Moonlighting.
Jeff Jarvis [02:26:17]:
Oh, that show. I was on it.
Leo Laporte [02:26:19]:
I know you were. That's why I brought it up.
Benito Gonzalez [02:26:21]:
Wait, you were on Moonlighting?
Jeff Jarvis [02:26:23]:
Yes.
Leo Laporte [02:26:24]:
He introduced their return.
Jeff Jarvis [02:26:26]:
Yes, when they were gone for a long vacation. Or not vacation. They were just, they were just, they were gone. And I complained at people that they were gone. And Glenn Gordon Caron, the showrunner, emailed— or not emailed me, this is before email— mailed me a letter and said, thanks for the Valentine. Will you come on? And the gag was—
Paris Martineau [02:26:42]:
Can we play that really quick?
Jeff Jarvis [02:26:44]:
Yeah, you should find it.
Paris Martineau [02:26:46]:
I like that if you search Moonlighting Jeff Jarvis, your IMDb credit comes up first, and then a Twit+ clip from April 2022 of Jeff Jarvis on Moonlighting comes up.
Leo Laporte [02:26:57]:
Oh, well, that's good.
Paris Martineau [02:26:59]:
So we've done this before.
Leo Laporte [02:27:00]:
Oh yes.
Jeff Jarvis [02:27:01]:
So...
Leo Laporte [02:27:02]:
Many, many times. It's kind of a traditional thing. Let's, uh, let's just enjoy the young Jeff Jarvis. A dance hall hostess with a heart of gold swings into the Forest Theater. Sweet Charity at 11 tonight. This is what TV used to be like.
Jeff Jarvis [02:27:18]:
Hi, I'm Jeff Jarvis, TV critic for People magazine. Like many of you, I was disappointed. No, make that fed up. No, on third thought, make that enraged with Moonlighting. Reruns, reruns, reruns, nothing but reruns. I mean, without Moonlighting, what was there to do on Tuesday nights? Watch the Home Shopping Network? Start a family? Read? Well, the crisis is over. Moonlighting's back. Last week, at long last, they gave us a fresh episode, and tonight they're giving us one that's even fresher.
Jeff Jarvis [02:27:44]:
But there's a catch, a hitch, a cruel trick. If you didn't watch last week's show, you probably didn't buy 'cause you figured it's just another rerun, then you won't know what the hell's happening in tonight's show. But fear not, don't sweat. I have right here a tape of last week's episode, and I'll show you all the good parts. Won't take long. It all started when David began to sense—
Paris Martineau [02:28:03]:
Wow, you do a full recap!
Jeff Jarvis [02:28:06]:
Oh, I did, I did. I wrote that.
Paris Martineau [02:28:08]:
Oh, I could tell. Oh, I could tell.
Leo Laporte [02:28:09]:
It's well written.
Paris Martineau [02:28:10]:
It's exactly— it's very interesting because it's the exact writing cadence, but your voice sounds so different. But
Jeff Jarvis [02:28:17]:
it was also mocking the cadence of Moonlighting. itself.
Leo Laporte [02:28:20]:
It's such a great show.
Jeff Jarvis [02:28:22]:
If you go to the end of that, the gag at the end of that—
Jeff Jarvis [02:28:25]:
Which reminds me, let me make it absolutely clear that my appearance here on Moonlighting will in no way compromise my integrity as a television critic. Sure. Thank you. Note.
Leo Laporte [02:28:39]:
That was a great show.
Paris Martineau [02:28:40]:
Great.
Leo Laporte [02:28:41]:
But I wish it hadn't come back because, uh, they come— they, uh, consummated their relationship, and it kind of ruined the show.
Jeff Jarvis [02:28:48]:
Yeah, kind of ruined it.
Leo Laporte [02:28:48]:
Yeah. It was great because there was such great tension between Cybill Shepherd and Bruce
Jeff Jarvis [02:28:52]:
Willis.
Leo Laporte [02:28:52]:
Bruce Willis. That were— oh, so such a good show. Yeah.
Jeff Jarvis [02:28:58]:
You know, we could talk about that. When TV was great, Paris.
Leo Laporte [02:29:01]:
I'm telling you, that's what we're gonna do next week. For 2 weeks, we're gonna do it. You're gonna love it.
Jeff Jarvis [02:29:06]:
Old Farts Week.
Leo Laporte [02:29:08]:
Actually, I have to get to work because it looks like we not only need somebody to replace Paris, we need to interview guests. So I'm gonna have to get to work. That's right, find some, find some people to—
Jeff Jarvis [02:29:17]:
we could find a doomer.
Paris Martineau [02:29:18]:
For what it's worth, I did give appropriate notice.
Leo Laporte [02:29:21]:
No, I've known about this. I have.
Benito Gonzalez [02:29:23]:
Yes, I've known about it. I've been, I've been on the horn.
Jeff Jarvis [02:29:26]:
We're just old and forgetful.
Paris Martineau [02:29:28]:
I'm happy to help send out some emails.
Leo Laporte [02:29:32]:
Um, should we— could have Ed Zittrain on.
Jeff Jarvis [02:29:36]:
Oh, he's way too big for us now.
Leo Laporte [02:29:37]:
Oh yeah, we couldn't have him.
Jeff Jarvis [02:29:39]:
He's been on, on MSNBC. I saw him twice with, by the way, making fun of Jensen Huang for wearing a leather jacket in a leather Yeah,
Leo Laporte [02:29:45]:
that's appropriate. That's good. I like it. That's a little, uh, kind of—
Jeff Jarvis [02:29:50]:
he's kind of Gary Marcus.
Paris Martineau [02:29:51]:
Yeah, he was on Jon Stewart like 2 weeks ago.
Leo Laporte [02:29:53]:
He was on Jon Stewart? That's—
Jeff Jarvis [02:29:54]:
yeah, no, he's, he's done it all now.
Leo Laporte [02:29:57]:
Yeah, yeah. Nobody ever really asks, what's your— why do you say this? What's your expertise here? Uh, all right, anyway, we're done. Uh, we thank you, and we will see you all— not Paris, but we'll see the rest of you all next Next week, you better be here on Intelligent Machines. Bye-bye. I'm not a human being, not into this animal scene. I'm an intelligent machine.