Intelligent Machines 823 transcript

Please be advised this transcript is AI-generated and may not be word for word. Time codes refer to the approximate times in the ad-supported version of the show

0:00:00 - Leo Laporte
It's time for Intelligent Machines. Our guest, Stephen Witt, wrote the book about NVIDIA and Jensen Wong. It's called the Thinking Machine. We will talk to Stephen in just a little bit, plus all the AI news next Podcasts you love. From people you trust this is. Twit. This is Twit. This is Intelligent Machines, episode 823, recorded Wednesday, june 11th 2025. Intelligent with two Gs. It's time for Intelligent Machines, the show where we cover the latest in AI, robotics and all the smart doohickeys all around you. Paris Martineau is here. Great to see you, paris.

0:00:46 - Paris Martineau
Great to be here.

0:00:47 - Leo Laporte
Yeah, I like that. Is that your new signature with the fingers? Like that it is actually an old signature.

0:00:52 - Paris Martineau
I used to be a classic.

0:00:54 - Leo Laporte
This sort of pointer, did you really? Yeah, was it ironic. I was a finger, I mean maybe it started ironic.

0:01:01 - Paris Martineau
It became earnest at a certain point where it all blurred and then I had to stop myself for a period. But perhaps I'm bringing it back I like it.

0:01:10 - Leo Laporte
I good to see you and welcome back speaking of bringing it back, mr jeff jarvis. Hey, authors of, where's your books?

0:01:18 - Jeff Jarvis
oh, uh, well, because I'm, I'm okay, here's where. Here, here's the thing I'm doing the bibliography for the current book. I don't know if you can see over there, whoa, that entire shelf is filled with. That's the bibliography for the hot type book, wow.

0:01:32 - Leo Laporte
Do you have like a library van? Pull up out front every morning and deliver books.

0:01:37 - Paris Martineau
Like a milkman.

0:01:39 - Leo Laporte
Here's your book. Good to see you. Welcome back. Did you have a nice trip?

0:01:44 - Jeff Jarvis
Yes, I did All, all right 40 miles away to montclair, new jersey, but that was okay. Oh, you were teaching. Well, I was, I was no learning.

0:01:52 - Leo Laporte
Jeff is a professor of journal, professor emeritus of journalistic innovation at the craig newmark graduate school of journalism at the city university. It's back, it's back, it's back. Why not? Wow, it had a sabbatical, like every professor, and now it's back.

0:02:09 - Paris Martineau
This is why you review the show people, because you can affect real change in the world.

0:02:12 - Leo Laporte
Yes, I had to listen to all those five-star reviews that said bring back the jingle. Jeff is currently at Mount Montclair State University in New Jersey and SUNY, the State University of New York at Stony Brook, and is the author of the Gutenberg parenthesis, the Web we Weave and Magazine, and we thank you for coming home. Hey, I think you brought Stephen Witt to us. Stephen is a Los Angeles-based writer, television producer and investigative journalist. His latest book, the Thinking Machine, just came out. It's all about Jensen Wong and NVIDIA. Welcome, stephen.

0:02:55 - Stephen Witt
Thank you for having me.

0:02:56 - Leo Laporte
I like your view.

0:02:58 - Stephen Witt
Yeah, it's really nice here. Oh, that's great. Beautiful view. There's a story behind it. We'll get into it later.

0:03:04 - Leo Laporte
Oh okay, it's not fake. That's a story behind it. We'll get into it later. Oh okay, is it? It's not fake. That's the story.

0:03:10 - Stephen Witt
It's real, that's no, it's unusual okay, so, um well, we saw jeff's bibliography, right?

0:03:14 - YouTube
uh, there is all the books that he's reading, for the book he wrote the bibliography for my book was extensive as well.

0:03:21 - Stephen Witt
Uh, I mean, you know 50 or 60 volumes, um. And then, seven days after I turned in, oh no, the manuscript for this book, uh, my home in altadena burned down in the evening.

0:03:33 - Leo Laporte
Oh, I'm so sorry. Oh, I'm so sorry including all my bibliography.

0:03:38 - Stephen Witt
So I have oh no, I don't own any books anymore.

0:03:41 - Leo Laporte
I mean, except for a few digital. Well, you're free at last. Think of it. Yeah, I'm totally free.

0:03:45 - Stephen Witt
It's literally not to schlep those around anymore so this is a, this is a new place, uh, with a nice view, and I'm lucky to have it, but I do miss my oh, I'm so sorry yeah our friend kevin rose, lost his house in the heights as well, and uh, david remnick of the new yorker called me shortly, actually, while I was driving around the burned out um, uh kind of like landscape.

It looked like a nuclear bomb had gone off in my neighborhood and a 10 000 structures burned down. I was like, yeah, yeah, I think my home is gone. And he's like you're the third new yorker writer I've talked to today who lost their home and uh, can't fire. Isn't that crazy? I am so sorry that community was full of journalists, creative types, uh, editors, producers, so I know quite a lot of people I have to go back in your bibliography.

0:04:31 - Leo Laporte
I've got to read how music got free, which is the story of the mp3 and napster and piracy and all of that. Wow, I really want to read that. I wish we'd interviewed you then. But the new one is a little more topical yeah, for sure you couldn't get more topical than nvidia, did jensen wong cooperate for this uh, at first yeah until he found out you were writing a hit piece. What?

0:04:53 - Stephen Witt
I didn't write a piece. Uh, I like jensen. He's really a smart guy, um, but jensen has a terrible temper, uh, as I've heard a great number of times that he sometimes explodes in anger at people.

0:05:05 - Leo Laporte
Oh dear.

0:05:10 - Stephen Witt
And so this was kind of a theme that has followed Jensen for a long, long time, and everybody I talked to was like, yeah, jensen will occasionally just like explode at people, right, but I didn't think it would happen to me you know, because I don't work for Jensen.

But in one of our last interviews I just kept pushing him about kind of the risk that ai poses to our species and the future of humanity and I probably asked this question in different forms four or five times and I think he just thought I was wasting his time. He didn't want to talk about it, he just erupted at me.

0:05:38 - Jeff Jarvis
What is, what is the run? What is the jensen wong eruption?

0:05:41 - Stephen Witt
it's addressing down. It's humiliating. Um, he questions your professionalism, he questions your ability. Uh, he questions why he's even talking to you. Why am I even participating in this if you're just gonna waste my time in this way? It's loud, there's yelling involved.

I'm so sorry that's, and it's kind of like a uh, it doesn't stop. There's no, once he blows his stack, you really have to let him completely like, shoot steam out of his ears for 10 to 15 minutes of just penetrating question Like why would you ask me that question? We gave you access inside our company. What kind of crappy book are you writing? Like, why are these questions so dumb? These are pedestrian questions. I expected more from you. I'm really disappointed in you as a journalist to be asked these questions. So this went on for 20 minutes and I'm giving away the end of the book, by the way and his PR people are there and they're just frozen, yeah.

0:06:38 - Paris Martineau
Wait, how did they react? Do they ever do they unfreeze and try to corral the situation in any way?

0:06:42 - Stephen Witt
or are they just so you know nine out of ten ceos. Yeah, that's what would happen. 99 out of 100 ceos. Right, but jensen has trained his people not to interfere they're just as scared of him as you were they're frozen smiles on their faces and they're watching him.

I interviewed a former chief scientist at nvidia about this and I was like, did jensen ever yell at you? And he was like, well, the only time jensen ever yelled at me is when he was yelling at one of my staff members and he's just torturing this guy in front of like 30 people and I felt like I had to step in. But that's like standing up in the trench and waving your arms at the machine. The machine doesn't stop shooting, it just turns and starts shooting at you. And so, like nobody wants to draw fire and jensen has created an environment in a world for himself where he can engage and what I would consider these almost like self-indulgent kind of like are people loyal to him?

0:07:36 - Jeff Jarvis
I mean so how does that work? Is it? Is it? Is it out of fear? Is it out of success? It's classic.

0:07:43 - Stephen Witt
It's classic cult leader tactics where everyone I talked to at Nvidia had a story where Jensen they just felt so loved by him he feels in his keynotes very affable, very warm, genuine yes. So funny and this dark aspect of him does not appear in public. He's careful about when he turns it on when. I only turns it on, though, when there's an audience at his job, so he'll never just grab somebody and start screaming at them in the hall. Oh interesting, he waits until there's 30 or 40 people.

0:08:14 - Paris Martineau
It's a humiliation thing.

0:08:16 - Leo Laporte
You know what? When I hear that, I think that's what his father did to him. I think it's something from his childhood. He didn't want to talk about that, did he?

0:08:23 - Stephen Witt
he did not tell me much about that, but I will tell you my impression is, his father wouldn't do this okay I, maybe his mother would do something, somebody did um where do you think this comes from?

0:08:34 - Paris Martineau
yeah, I mean, obviously that's total speculation.

0:08:36 - Stephen Witt
I talked to his friend, his old friend jens horseman. I was like, did jensen always do this? And he was like well, I worked with jensen in his 20s. He was not known for doing this and I saw interesting it really started when he became the ceo, so whether this was some kind of like power trip. Uh, jensen defends what he's doing. He's saying well, I'm making a demonstration, an example of people because they failed in some way and I can't just dress them down.

0:09:01 - Leo Laporte
The whole company has to see it everyone has the see even jobs technique kind of yeah, so so failure must be shared, is one of his mottos.

0:09:10 - Stephen Witt
Wow um one time about success the success is shared at nvidia, and this is you asked why people are loyal to him. If you've been hanging out with jensen for 10 or 15 years, you have several hundred million dollars now. There's that, so your belief in this guy and his ability has paid off massively. If you could put up with the kind of abuse like you have, been compensated beyond your wildest dreams.

0:09:36 - Leo Laporte
Do people in NVIDIA feel like they're changing the world? They're making a huge difference.

0:09:40 - Stephen Witt
Yeah, and they are right. Every company says innovators who changed the world. But this company really did days.

0:09:46 - Leo Laporte
Yes, right, this is the. This is the way ai companies train their models.

0:09:52 - Stephen Witt
They buy as many gpus as they possibly can from nvidia yeah, so the the limiting factor, actually, and how fast ai can go, is these microchips. Yeah, they can't make them fast enough. They can't make them fast enough and they sell them for thirty thousand dollars000 a piece. This has made NVIDIA the single most valuable company in the world, or they're right up there with Apple and Microsoft, I think. Actually, today they are the most valuable company in the world. And why is that? Well, if you look for demand for something like ChatGPT, right, chatgpt has eclipsed Wikipedia in terms of number of daily users, and users are coming to it with increasingly complicated requests. Executing those requests requires, like quadrillions of what they call flops, individual arithmetic operations. Okay, so to execute a quadrillion flops for every user in the world multiple times a day, you have to build a giant supercomputer, a mega computer. The one they're building in texas costs 50 billion dollars, so that makes it more expensive than china's three gorges dam and about 10 times as much as the james webb space telescope.

0:10:59 - Leo Laporte
That's how much this thing costs is that the grok, the xai one?

0:11:02 - Stephen Witt
that's a different one and everyone's building them. So this is the thing there's like 40 of these coming online, okay, and you look at that $50 billion like where does it go? Well, it goes to electric power and water and land and construction, but the single largest cost is NVIDIA microchips.

0:11:19 - Leo Laporte
So all of that money Of this H100, this $30,000 chip, they bought $350,000.

0:11:27 - Stephen Witt
Yeah, and that's probably $30,000 each. So I think if you do the math you'll get to a lot. It's a $10 billion order, if not more, and it all just goes right into NVIDIA's pocket and they earn like a 75% gross margin on these chips. So they earn more, basically, uh, income per employee than any company on earth if any, I think actually, craigslist used to have a high, one of the highest, because there were so few employees.

Yeah, that might still be true, actually, but they're up there in video for major publicly traded corporations actually. Yeah, it might be number one yeah, I think.

0:12:01 - Leo Laporte
Uh, elon bought 150 000. I don't know what these new plans in Saudi Arabia involve, but they're building network operations centers bigger than the entire country of Monaco.

0:12:14 - Stephen Witt
Yeah, no, it's five square miles of computing equipment at a rumored cost of well, I don't know what the cost is, but it's going to be five gigawatts of energy power Unbelievable Required electric power required to power it. So, in context, uh, one gigawatt of electricity is about enough to power kansas city so I believe that's pronounced gigawatt. No, I'm just kidding so let's go back in time, though, because this company was originally, uh, made gpus for gaming.

Uh-huh, that's how they started, right yeah, so this sounds like they were making toys, right? Yeah, and this is why nobody took them seriously. But the insider's perspective is consume. Uh, computer graphics are actually one of the very hardest things you can ask a computer to do. You're kind of tracing these 3D skeletons in real time and then painting them with textures and it's just enormously computationally intensive. The amount of arithmetic you have to do. I can't remember the exact figures.

I put it in the book, but to do one second of rendering, if you did it with pencil and paper, would take a human being something like 800 million years, wow, and that's if you work 24 hours a day and never made a mistake. So that's kind of what has to get compressed each second of rendering a 3D game. So it's very computationally intensive and it has to go super fast. And then the thing that really makes it special is that demand is infinite, no matter how well you render the game. There's a certain kind of video game customer who's going to come back and say I want it better. This is going on. There's a certain kind of customer who will spend three to four thousand dollars on a gpu, build their entire gaming computer around it like it's some kind of hot rod car, just so they can render the pixels in the game a little faster than their competitors I've got an nvidia, expensive nvidia card in my gaming rig.

0:14:12 - Leo Laporte
I mean, there you go, that's how you had to do it. I mean I go back even prior to that, to you know, uh, other companies but but nvidia became the champion.

0:14:21 - Stephen Witt
There's amds, radion and stuff, but nvidia really became the the premier gaming coprocessor bit of history here when it started, there were 70 competitors, uh, in the in the mid 90s, um, and jensen kind of intuited or not intuited, but I would say logically reasoned that within five or ten years there would only be a couple companies left, it probably just one, right, yeah, he was determined to be that one. Yeah, so he was absolutely ruthless. I don't know if you've ever seen the movie battle royale, uh, where the japanese teenagers are stranded to an on an island and have to fight to the death, um, but basically it was that, uh, in the gpu space, they all had to just fight to the death the big.

0:15:01 - Leo Laporte
The big guy was 3dfx right.

0:15:03 - Stephen Witt
And so what jensen did? Uh, he absolutely out-competed and destroyed 3DFX.

0:15:09 - Leo Laporte
Yeah, they're gone.

0:15:10 - Stephen Witt
And, along with his team, he would make lists of everybody who worked at 3DFX, everybody who worked at all his competitors. He put them on a whiteboard and then they would strategize on how to poach these people away from 3DFX. They called it brain extraction from 3DFX. They called it brain extraction. Once they got that guy the two or three chief engineers to defect to NVIDIA and they were very successful at doing this the company folded because it no longer had a brain. That was the end. So they conquered the 3D graphics space by maybe early 2000s and they had their technology in the Xbox as well.

And what started to happen is that scientists look at the intense computational density of these chips, what they were able to do for 3D graphics, and the scientists said God, I'd love to get into that architecture myself so I could run my super complicated quantum physics calculation or my super complicated tomography, which is basically MRI. So I'm, which is basically MRIs. I'm going to reconstruct MRIs to higher fidelity using these chips. And they started actually hacking the chips, jerry-rigging them to do these other functions. And Jensen saw this and he said wow, this is like a whole untapped market.

0:16:19 - Jeff Jarvis
So he didn't see that before, he didn't have a grand vision before. That said, this is the most powerful computing chip. Once it was demonstrated to him, that's when it started opening up to him.

0:16:29 - Stephen Witt
He was too busy winning the knife fight.

So, you know, he was too busy trying to survive the battle royale to focus on anything except being the dominant player in graphics. But they started to approach him and there was actually a Stanford, very famous Stanford computer scientist, bill Dally, who wrote a paper talking about the arithmetic intensity or the arithmetic density of these chips. And this had not. You know, jensen had not been deliberately building tools for scientists, but when the scientists saw it they were like were like god, we got to use this. And when jensen heard that, he went to stanford and met dolly and was like, oh my god, I, I have to serve the scientific community with these cards. And so, at great expense, he built this massive platform that would flip the switch on the graphics accelerator and turn it into a scientific high performance computing platform for quantum physics, for breast cancer. What was that called? It was called CUDA. That's CUDA.

0:17:32 - Paris Martineau
Yeah, I was just about to ask you about.

0:17:34 - Stephen Witt
CUDA. Yeah, that's CUDA. Okay, so this is called. Cuda was the name of this platform, described as the switch on the graphics card that you flick and it turns it into a low-budget supercomputer.

0:17:50 - Jeff Jarvis
Switch on the graphics card that you flick and it turns it into a low budget super. Now, uh, wall street hated this. Yeah, why scientists?

0:17:55 - Leo Laporte
don't have money. They don't have money.

0:17:57 - Stephen Witt
That's not the market you want scientists have a ten thousand dollar research grant, maybe right right like if they're lucky, and they won't necessarily show up with it next year. So they're not even necessarily repeat customers and it was extraordinarily expensive to develop. So basically, nvidia was spending something like a billion dollars a year on R&D and then receiving a return less than a billion dollars in revenue from this market.

0:18:21 - Jeff Jarvis
Did he understand the business potential then?

0:18:23 - Stephen Witt
or only the-.

0:18:24 - Jeff Jarvis
Yes, okay.

0:18:27 - Stephen Witt
Yeah, I mean. Obviously that's why he did it right. He's a businessman, he's trying to make the stock price go up, but his reasoning was I have to do this. He reasoned that because of Clayton Christensen's model of disruption the innovators dilemma innovators dilemma. Yeah, which he, which he actually assigned to all of his executives? Uh and, and Christensen worked as a consultant at NVIDIA for a while.

0:18:44 - Leo Laporte
Interesting.

0:18:45 - Stephen Witt
Basically, his thinking was if I don't do this, someone else is going to jerry-rig this and turn it into their own platform and erode our profits away from the bottom. And Christensen had this incredible quote in his book where he said something I'm paraphrasing, but something like sometimes it is right to ignore your customers and to ignore your investors and to invest in low margin, small market products that nobody seems to want. And that's what this was. It was the disruptors play, it was the count, and investors they talk about disruption. They actually hate it when you do this.

0:19:25 - Leo Laporte
Yeah, yeah, they want hate it when you do this.

0:19:27 - Stephen Witt
They want you to keep making money the way you make it. They want you to earn profit margins.

0:19:29 - Jeff Jarvis
McKinsey will come in and kill everybody who does that.

0:19:31 - Stephen Witt
You're lowering your margins on a product that does not, in an even intermediate term, promise to return cash flow to your company.

0:19:40 - Leo Laporte
One of the interesting things about-.

0:19:42 - Stephen Witt
It's absolutely the opposite of the way that a classic MBAba right or mckinsey type would run the company right. Um, thank goodness, the stock price actually went in the toilet oh interesting uh, for 10 years. While they did this, they did not break through for a long time and, in fact, jensen nearly lost his job because an activist investor came in and started writing uh, you know, schmuck oh, these morons letters to the board that were not very friendly.

I actually like activist investors. I think they play an important role in the capitalist ecosystem and the guy who did this has actually a very good track record. He famously turned around Olive Garden. His name is Jeffrey Smith How's.

0:20:23 - Leo Laporte
Olive Garden doing these days.

0:20:25 - Stephen Witt
He's just three bread sticks in a trench coat okay, so they have a whole breadstick strategy at olive garden oh, I bet they do so the idea at olive garden here's what you do okay, you go to um the table and you always give, uh, the same number of breadsticks, as there are people at the table plus one, plus one. Why Let them fight over that last one? Because people are either A very polite and they're not going to take the last breadstick Right If they eat all the breadsticks, they'll just ask for more. But if there's one left, it's actually hard to be the guy who goes and breaks up the last breadstick or if you're not, with people like that and they just start consuming all the breadsticks and want more. It's more opportunities for the server to come and talk to the customers and maybe upsell them on more pasta. So it was this whole thing. But Olive Garden actually got away from this and they were just dumping breadsticks on the table and the servers didn't want to keep going back to the table, so they would just put like 30 breadsticks out there, have all the breadsticks, but that actually destroys the whole concept of the free breadstick.

Okay, anyways, this is Jeff Smith. Thank you, very smart actually.

He understands the operational details of this company quite well, but he completely misanalyzed NVIDIA.

0:21:38 - Paris Martineau
Wait. So how do we get from CUDA being something that potentially could tank the stock of this company because they're spending so much money in it to it being this kind of cornerstone to nvidia, nvidia's big pivot?

0:21:51 - Stephen Witt
right. So who is cuda for? Well, it's not for leading research scientists, because they can afford time on the massive supercomputers. Right, it's for scientists whose research is a little bit out of favor. It's for scientists who want to do things that can't really get funded. It's it's for mad scientists and young mad scientists that's what it's for. I mean, it's for scientists who do have kind of wacky theories.

0:22:24 - Leo Laporte
It ended up being their secret sauce. I don't know if it was the processors. It was really CUDA, because it was proprietary and, as things became written, for CUDA you had to have an NVIDIA chip. That's right.

0:22:36 - Stephen Witt
So these mad scientists maybe the maddest scientists were this research group in Toronto. So this will be hard to even remember or understand for, I think, younger viewers. But there was a time where AI was a career graveyard. I mean, if you went into AI in the first decade of this century, it was kiss your career away, you're going to be working in academia the rest of your life and you're going to struggle to get a $5,000 research grant. Within AI there was even a worse backwater called neural networks. So nobody, even in AI, liked neural networks and almost nobody thought that they would work. They were viewed as a dead technology. Absolutely mad science, just mad science.

Jeffrey Hinton and his group in Toronto acquired two GPUs that was what they could afford Each of which cost about $500. And then they jerry-rigged a low-budget supercomputer in Alex Krasjansky's childhood bedroom at his parents' house and they took one of these dysfunctional neural nets that nobody liked, not even within the AI community, and they trained it in this kid's bedroom. Basically, and at the end of a week, because of NVIDIA's CUDA platform and their hardware, they had produced what was by far the leading AI in the world. In a week Was this ImageNet? In one week, this was ImageNet AlexNet. Sorry, alexnet it's called, but ImageNet was the competition that AlexNet won. Ah, okay.

0:24:08 - Leo Laporte
I interviewed ImageNet in Toronto back in the day in Fay Lee.

0:24:12 - Stephen Witt
You might have interviewed Fay Fay Lee. Yeah, I did. Fay Fay described this moment as it was as if the world land speed record had been broken by 100 miles an hour by someone driving a Honda. So this was the paradigm shift, as I say in the book, and the way to think about this is it's almost like the Wright brothers right, we're two bicycle mechanics. Jerry rig an airplane, right? And if you want to know what Jensen's role about that, it's kind of like he was building engines with propellers attached for 10 years, just waiting for someone to come along and find a use for this propeller.

0:24:50 - Leo Laporte
Put wings on it. Yeah, by the way, jeff, this is Ilya Suskever as well.

0:24:55 - Stephen Witt
Yeah, ilya was the third member, so Ilya, who is now obviously kind of, was one of the leading programmers behind ChatGPT. Now, is it safe? Super intelligence is like omnipresent he's everywhere, he's ubiquitous. He's ubiquitous in AI and he was literally there at the creation including the University of Toronto this last week yeah, this is like. This is like it would be like witnessing the AI version of, like witnessing the Trinity test, basically to be there with this AI um, so that percolated very rapidly, those results when did Jensen?

0:25:32 - Jeff Jarvis
how did Jensen find out about that?

0:25:34 - Stephen Witt
uh, dolly brought it to him. Okay, so in 2013, if you watch Jensen's GTC presentation, he? Uh is in leather already that's that transition is starting.

0:25:46 - Leo Laporte
Famous leather jacket how many jackets does?

0:25:47 - Jeff Jarvis
he own.

0:25:49 - Stephen Witt
And he would. Actually that one was not. So brief side note on Jensen's wardrobe, because I know everybody wants to know. Jensen's not like Steve Jobs, by which I mean he does not have one signature outfit that he wears every day. He's not an artist in that way. He doesn't have like an artistic vision. He's not an artist in that way. He doesn't have like an artistic vision. He's an engineer and so what he does is he A, b tests different items in his wardrobe, so he takes in a piece, he sees if it goes and you watch this and he's like actually constantly making subtle alterations to his wardrobe until he nails it In 2013,. He absolutely had not nailed it.

0:26:24 - Jeff Jarvis
I thought his last one. The sleeves were too long for him, yeah he was experimenting with snake skin.

0:26:29 - Stephen Witt
That's kind of new. I don't think it looked a little too We'll see Anyway sorry.

So yeah, but he experiments. So this one, he's wearing a motorcycle, jackson. Anyway, long detour 2013,. He does not mention AI one time. And NVIDIA puts out a list of what CUDA is good for and it's all these scientific applications. Ai is not listed.

2014, he goes out and he only talks about AI. And somewhere between that, 2013 and 2014, gtc someone at his company was like you look, I think it was brian cottonzaro, I talk about this in the book. He said you gotta look at these, uh, ai results, you gotta see what's happening here, because I think it's big. And jensen looks at it and he gets it instantly and he bets the whole company company-wide email over a weekend saying we are no longer a graphics card company, we are now an AI company and we are going to put 100% of the resources in this company into AI that we can, or every resource we can.

And he goes to Brian Catanzaro, who at this time is like a low level product manager, and he says Brian, I want you to imagine I've marched all 8,000 employees of this company out into the parking lot. You are free to select any one of these employees to join your team and you can have as many as you want and this is like. So he just deputizes this guy to run his AI effort and this is like a low-level researcher who had almost he had bad performance reviews at the company, but basically field promoted this guy into running his entire AI strategy and within a year, they were the dominant, essentially only AI hardware platform and tailoring their chips to meet what they saw very early and this is really Jensen's key insight that he saw before anybody that to make these things more powerful, there would be just as there was infinite demand for the graphics better graphics there to be infinite demand for ai and that's a good question for more intelligence.

0:28:30 - Jeff Jarvis
Kuda is so important and and as we'll discuss later I I, as I said before we got on I'm a connoisseur of jensen's keynotes and it always strikes me he spends a good amount of time bragging about cuda, giving you the sequence, giving you everything In the importance of the company. In his mind, everybody thinks it's a chip company, but it's also a platform and software company. Where does that balance in his mind and the company's mind in terms of the importance to its strategic value?

0:29:00 - Stephen Witt
Yeah Well, it's all integrated, right. It's a little bit like the iPhone. What's more important than the iPhone? The hardware, the software, it's an integrated platform. It's a completely integrated platform.

I will say they view their competitive advantage as being software. Amd can design the silicon just as well as we can it's not a trade secret, right? I can peel the chip off this silicon and I can expect it with a metallurgical microscope and tell you after a little while exactly what it's doing. So anything they put on as a physical object can be reverse engineered, right. So if they're going to have an advantage, it has to be at a layer above that and where the real advantage is they have a lot of low-level software, but their real advantage is actually that they go out and they embrace the scientific community and they will look for the hardest computing problems in the world, and then they will lose money for years, sometimes forever, to build these scientists the tools that they need, just so they know they're always on the bleeding edge. That's the secret of NVIDIA, and I think it's something other firms have a hard time replicating.

0:30:10 - Leo Laporte
Yeah, so it's not just gaming and AI. The other part of NVIDIA's business is chips for self-driving vehicles. Yes, so that's a big business for them too, isn't it?

0:30:23 - Stephen Witt
And it's going to explode. Right now, their big business for them too, isn't it? And it's going to explode. Here's the right now. Their big business is selling data center chips. We call that the cloud cloud computing. What that means is I ask chat GPT something, and it takes my request on a broadband pipe and ships it to a mega supercomputer in Texas, processes the answer and ships it back to me. Okay, great, but there's a lot of applications where we can't wait that long. If we're in a car and we're driving, I cannot wait for the car to communicate with the data center in Texas. It has to move right away. This is called edge computing, and so not only do we need these chips in the cloud in Texas. Increasingly, as more items and things, especially robots and cars, become autonomous, we need edge computing chips that are directly in the car, that don't need to interact with the internet to provide an answer.

0:31:10 - Jeff Jarvis
And this is the digital twin.

0:31:12 - Stephen Witt
That's different.

0:31:13 - Jeff Jarvis
Okay, all right.

0:31:15 - Stephen Witt
Yeah, okay, this is called edge computing. Right now I don't think edge computing is a huge part of NVIDIA's business. I would be surprised if it was more than five percent of the revenue. But jensen is putting all his resources there, because jensen's future world, where nvidia's next three trillion dollars of market capitalization is going to come from, is they're going to ship robots and cars, every right right and factories and they're all going to be ai enabled and we can't have any latency.

If the robots let's say cleaning our dishes, it can't just pause constantly and have to go back to Texas and get answers. Its brain has to be physically located in our house right Right now. The brain for chat, gpt. It's not on your laptop, it's in Texas, it's in some data center somewhere right, and every time you pump an answer, we ship it down to Texas for processing and we ship it back to you. The edge computing platform. They can't wait right. So robots and cars, they can't follow this paradigm. The brain has to be in the thing and that's going to be a huge market in years to come.

0:32:20 - Leo Laporte
I remember and I don't remember what year it was a gdc where jensen was talking about still talking about gaming, talking about ai, talking about cars, and it struck me then, this company's going places, this company is firing on all three cylinders yeah, they are um, especially now. I mean god yeah, this was a long time ago, but uh, it's come out to be that way.

0:32:45 - Stephen Witt
So I asked you know, I asked Jensen what's, what's the next CUDA, what's the thing you're investing tens of billions of dollars and that's losing tons of money and everyone's screaming at you about it that maybe five to 10 years from now might pay off. And this is Jeff's thing that he brought up the digital twin. They call it Omniverse and basically what it is is it's a high fidelity physics simulator of the real world, of our universe. Okay, so why do you want that? Why do you need that? Well, think about it this way If we want a robot to wash our dishes in the sink, right, there's kind of two ways we can train its brain.

One we can physically have it train a bunch of dishes at the sink or wash a bunch of dishes at the sink. It will break 10 million dishes in the real world. While we do this, it'll learn Eventually. It'll learn via reinforcement learning, but it literally has to do tens of millions of training cycles. It'll be a huge mess. It'll be a huge mess. So what if we replicate the dishwashing environment, the sink, in code right In a high-fidelity physics simulation, which, via GPUs and graphics, nvidia already has some experience building these kind of worlds, if we get the fidelity high enough, then we can train the robot to wash dishes.

It's hard, right, right. Surface tension, uh, certain dishes are fragile, certain are harder. Soap it's slippery. It's just not an easy thing to do, right. But if we get all of that right the surface tension, the physics, all of it maybe we can train the robot to, uh, to wash dishes in code in this kind of like digital gymnasium that we built, and then, once it's done, we're going to download its brain into a real world pot body and deploy it to actually in the edge in the edge.

0:34:31 - Leo Laporte
Steven was taught to play go. It played a billion games against itself and then, in fact, a human this is in fact almost, uh, a large number.

0:34:42 - Stephen Witt
Uh, it's called reinforcement learning, yeah right, and it actually goes back to the very beginning of neural nets. There's a great anecdote in my book about the first person to really get to work who was trying to win a backgammon game in the 90s, and he actually got a state-of-the-art backgammon thing and he sold it via floppy disk off his website. His name is Frederick Dahl. It was inspired by the movie War Games, which is very funny, but side note anyway. Yeah, so this is how we train things.

0:35:10 - Jeff Jarvis
So, stephen, there's two things about this, one again coming from his keynotes so that's the extent of my expertise in this is that the cars are going to gather data as they go, and that's part of how they teach, is that? But what's really struck me about the digital twin is that it's my joke, is that it's the matrix but we're not in it. Yeah, precisely that. It's every permutation and computation of possible futures that's right Are being computed, and it really struck me most in a factory warehouse setting.

0:35:43 - Stephen Witt
Yeah, yeah, so you can also do like very advanced autocad layout stuff once you have this environment up and running. So, uh, in addition to training, but mostly it's for training robots, honestly, I mean, I think that's where, at least for now, he really believes robots is the thing oh, yeah, yeah, no he is he says yeah, he talked about that little cute disney robot he brought on, if you see how much he talked about ai in 2014 and 2015. Before it was like a big deal, right.

0:36:06 - Leo Laporte
That's how much he talks about robots now I want to know a little more about jensen wong. What? What's his story? Where did he come from? He was an engineer he was an engineer.

0:36:15 - Stephen Witt
so jensen was born in taiwan uh moved to bangkok, uh with his dad, who was also an engineer. When they were young his dad worked in an oil refinery, um, but his dad was determined to move the family to the United States and so Jensen was actually sent ahead of his family to the United States, arriving in, I think, 1973, in Oregon, and his uncle was taking care of him. His uncle thought that the boys should go to a prestigious boarding school. Do you want to wrap up, are you?

0:36:43 - Leo Laporte
running out of time Because we are coming up on a half an hour.

0:36:45 - Stephen Witt
I don't know, oh my gosh, that's all we got uh, uh.

0:36:48 - Paris Martineau
One last thing you'd mentioned before the show that, while we're on the topic of jensen, that he doesn't like public speaking. Is that ah?

0:36:55 - Stephen Witt
yes, so the key to understanding jensen.

0:36:58 - Paris Martineau
Honestly, he might be the most neurotic person I have ever met in my life what you're saying, the guy who frequently yells at his employees and has trained them all.

0:37:07 - Stephen Witt
This is the thing that will not be clear to you, if you only and it's really explored in my book. This is where I really I feel like I added a lot of insight. The guy who lectures at GTC is so polished, he is so charismatic. You'd think he was born to do this.

0:37:26 - Jeff Jarvis
And says he doesn't rehearse, right says he doesn't doesn't rehearse.

0:37:29 - Stephen Witt
He really was hanging out with jensen in the green room before he had to go speak to a bunch of architects like this is a meaningless presentation 400 people and he starts to rock back and forth on his feet and he starts to get like visibly like agitated, starts cracking all these jokes and talking really fast and I was like, oh my God, he's nervous, he's got stage fright. He's nervous to go on here and give this meaningless lecture and he turns to me, just goes. God, I hate public speaking, Jeez.

0:38:01 - Jeff Jarvis
You'd never know.

0:38:02 - Stephen Witt
So effing good at it. Yeah, you know why he's so good. He's terrified of bombing.

Yeah, he's absolutely terrified that he's going to go out there and bomb, and so he does everything he can to avoid that happening, and he's constantly scanning the audience. This is why he doesn't use prepared remarks to see if they're bored, he'll completely change what he's talking about. Now that's not true in GGC, for example, but if he's speaking extemporaneously, he's reading the room constantly and recalibrating. But Jensen is totally neurotic, by which I mean not that in a clinical sense, he's just totally driven by negative emotions. So much of Nvidia success has come from Jensen being afraid, terrified that it is going to fail and that he personally will feel shame when that happens. It's why they survived. That's how he survived the battle royale. It's why they're so big today.

0:39:04 - Jeff Jarvis
It's like Intel. It's like intel on steroids, right yeah, he wrote the book.

0:39:09 - Leo Laporte
Only the paranoid survive.

0:39:10 - Stephen Witt
Yeah jensen has read that book and absorbed it. Yeah, jensen's a voracious reader, but but primarily of just a business books. Uh, his office is he doesn't use it, he, he. His actual office is a conference room but he has an office complex at nvidia headquarters and it's just filled with stacks and stacks of business books.

There's a famous anecdote about Jensen when he was arguing with one of his executives about how much something should cost and the MBA has some kind of idea and Jensen just breaks them off and he's like what are your three favorite books on pricing? And the guy like sputters, you know, he can can't name any. And jensen rattles off three books on pricing. He's like go read those books and then, when you're done with those books, come back. We can resume this argument.

0:39:52 - Jeff Jarvis
Wow, jesus steven I'm so respect how you're explaining all this, because it's it's it's so complex. Um, how did you train yourself in all this so that you could deal with the likes of jensen wong? You know, jensen is not my my first tough customer.

0:40:09 - Stephen Witt
I've I've done a lot of investigative reporting on ceos. Uh so, and many of the. I will say this about jensen it's true he's brilliant, and I don't. I'm not a guy who walks around talking about how brilliant silicon valley guys are, but this, this guy, is special he's not your typical billionaire. I gotta say that yeah, he doesn't have an mba, he's engineering background. If he if he wasn't running this company, he'd be designing their chips and as a teenage ping pong star, you say yeah.

So not only that, but he basically did not play ping pong at all, started, threw himself into it and within about four months was nationally ranked isn't that interesting, wow so his ability to rapidly absorb and implement new information and this is like a sign of intelligence is maybe the highest of anyone I've ever encountered in my life. It's just extraordinary how fast he can learn things and I think, uh, here's, here's. If we don't have much time, I'll give you an anecdote that I'll close this we have all the time you want.

0:41:09 - Leo Laporte
I know, I just don't want.

0:41:10 - Paris Martineau
We're just trying to be uh cognizant of your time.

0:41:13 - Leo Laporte
Oh no, I, I can do this all day um I saw the nurse came in to give you your shot and I just didn't want to yeah, no I'm just yeah, we're.

0:41:21 - Paris Martineau
He might be running out of juice.

0:41:24 - Stephen Witt
Jensen was asked to throw out the first pitch in a baseball game Okay, in Taiwan. And he gets up there and I don't think he's probably thrown a baseball in four years, if ever. And he just kind of is visibly grimacing and he gets up there. I swear to God, he's like listen, I'm about to throw it like on the mic to a stadium full of people. He god, he's like listen, I'm about to throw it like on the mic to a stadium full of people. He's like listen, I'm about to throw the first pitch. But I'm really bad at pitching, so I ask you all to turn away and not watch me as I throw this pitch, please.

Brilliant that's so engaging and you know it wasn't very good. Uh, the the catcher had to kind of like scoot up to catch it right, and so this is like a meaningless thing, right?

but jensen, like for jensen, this is an irresistible opportunity for self-improvement yeah and so over the next six months he gets his wife lori they've been married for 42 years to go in the backyard of his mansion every night, after running the most valuable company in the world, and throw the baseball around with him. They're both 60s, they're both in their 60s, until until he gets good at throwing a baseball. So the next time and this like six months later he's asked to throw another ceremonial pitch, and this time he nails it. Of course that's that's.

0:42:41 - Leo Laporte
This is who Jensen is. This is a Jensen is.

0:42:44 - Stephen Witt
He's a guy who actually like kind of lives in a mental torture chamber of his own divisement, who is constantly beating himself up for his perceived shortcomings and then creating programs for self-improvement that will kind of recalibrate who he is to fit the needs of the moment. That will kind of recalibrate who?

0:43:05 - YouTube
he is to fit the needs of the moment. Jesus, the.

0:43:06 - Jeff Jarvis
New.

0:43:06 - Leo Laporte
York Times calls it a lively biography Fascinating James Surowiecki and the Atlantic framed as a biography of Jensen Wong. The book is also something more interesting and revealing a window into the intellectual, cultural and economic ecosystem that has led to the emergence of super powerful AI ecosystem that has led to the emergence of super powerful AI the book, the thinking machine, jensen Wong, nvidia and the world's most coveted micro chip, stephen Witt. Thank you for giving us your time today.

0:43:34 - Paris Martineau
I really appreciate it. This is so great. Yeah, can't wait to read the book.

0:43:38 - Jeff Jarvis
Are you saying what you're working on next, or are you one who doesn't say that?

0:43:43 - Stephen Witt
Sure, well, I'm trying to get this pitch hasn't been picked up yet, but I'll just say it as if it's going to be. I think I want to write about Stargate. So Stargate is kind of the next generation of this. This is OpenAI's mega. I mentioned this at the beginning. They're building these hyperscale computers that are just going to. They're just going to be so big it's almost unfathomable. So OpenAI Stargate investment program, alongside Oracle and a couple other companies, is going to schedule to cost $500 billion, which is more than all 17 missions of the Apollo space program combined.

They're going to spend more on this than we spent going to the moon six times.

0:44:29 - Leo Laporte
Yeah.

0:44:29 - Stephen Witt
Okay, and it's just computing equipment and basically all of it is going to NVIDIA. But I want to go this is in.

0:44:36 - Leo Laporte
Saudi Arabia.

0:44:37 - Stephen Witt
No, this is one in Texas. Saudi Arabia is even bigger and there's like 30 or 40 of these coming online across the world. To me, me, the perceived infinite demand for ai and smarter machines and smarter systems I think for sam it's not even the demand that's driving this.

0:44:55 - Jeff Jarvis
He wants to build a super intelligence oh, oh yeah I mean, they would read his essay this week and it's just well, we're going to talk about it.

0:45:01 - Stephen Witt
Yeah, moon into a supercomputer. I mean they um, and the thing is like you've got to think well, there's got to be a more efficient way to do this and like, maybe theoretically there is, but practically this is at least the most straightforward way to do this right now, and I think that's the other thing.

0:45:17 - Leo Laporte
What's unproven is throwing more compute at it will make it intelligent. That's just not clear. Make it intelligent, oh no I think that is proven.

0:45:25 - Stephen Witt
So that has been the thing that has really turned the corner since 2017, 2018. It's called the bitter lesson.

0:45:31 - Leo Laporte
Yes, we've talked about it, yeah.

0:45:33 - Stephen Witt
And so the thing that you can guarantee is more computing power. The thing that will absolutely happen, no matter what happens in AI software, is you'll have more computing power available down the road.

0:45:44 - Jeff Jarvis
Yeah, that may not get them where they think they're going to go. That's what even Sundar Pichai just said, that.

0:45:50 - Stephen Witt
Right? It might not, and in fact, we've had two big software breakthroughs in the past 15 years. First was neural networks on GPUs in 2012., and then it was the Transformer architecture in 2017., and so maybe we need a third one. I think most research scientists do think we need a third one. That's what Leon LeCourne says, Although you know, I don't think Ilya Sutskever thinks we need a third one, no, and they are building the cutting edge things.

0:46:28 - Leo Laporte
That's what Rich Sutton said in the Bitter Lesson. He said if we've learned anything, it's throwing more compute at it makes it smarter, not making it more complex, just throwing more compute at it.

0:46:37 - Stephen Witt
So the way to think about this is the computer, the compute. It's not the brain of it, especially the training. It's more like the evolutionary conditions that led to the development of the brain. So if you think of your brain, it's quite compact and very good at what it does and you think, oh, it's only three pounds in this $500 million computer. It's not analogous. But the brain is the product of 500 million years of evolutionary conditioning. It didn't just arrive, it's the inference engine for this 500 million years of evolution. And so maybe, or maybe even probably, this is what is going to happen with these computers, with these mega computers as well. They're not really the brain, they're more like the evolutionary environment that produces the intelligence. And if you can make that more sophisticated and more complex, the things that come out of the bottom, the intelligence engines that come out of the bottom of that, are better and better. Well, good luck with the next book.

0:47:36 - Benito Gonzalez
Thank you yeah.

0:47:37 - Leo Laporte
I want to bring you back for the last book, but that's another story, yeah anytime, I'm always free.

0:47:41 - Stephen Witt
I mean, that's a history book. Now Whatever happened to Smith.

0:47:44 - Jeff Jarvis
It is the investor. Did Smith make money in the end?

0:47:48 - Stephen Witt
Smith flipped out of NVIDIA in 2013 and then regretted it ever since oh kidding.

At least he wasn't shortened. Yeah, he's trying to shake up. Is he in Intel? He's trying to shake up somebody else right now. I'm not sure he's in Intel. His investment returns are very good. Smith is very smart at what he does and I know people hate this kind of guy, this kind of corporate raider guy, but there's so many companies out there that just don't do a good job and squander investors' money. There's also so much passive money in the stock market via ETFs and Vanguard and basically like index funds that if you don't have someone who's going to agitate on behalf of shareholders for better returns, capitalism isn't going to work that well. Right now he's long on Pfizer.

0:48:37 - Leo Laporte
He wants Pfizer. Yeah, he's long on Pfizer, he's being on.

0:48:39 - Stephen Witt
Pfizer, and Pfizer, despite having come up with several fantastic vaccines for COVID and also Paxlovid, has not performed well for shareholders lately. And so you know what Smith looks for is when CEOs or companies are wasting money and he tries to kind of reset them and put them on the right path, if you like, steven and you like his style.

0:49:02 - Leo Laporte
he reads his own book on Audible. That would be a good place to get a copy of it. I think Stephen would be great doing this yeah yeah, I will say multiple people have already said in the Discord they want to read it. No, they've already said they've purchased your book in the middle of this interview you should see a little tiny bump.

0:49:22 - Stephen Witt
A little micro bump from the podcast. Well listen, the audio audiobook is actually um.

0:49:31 - Leo Laporte
I write my books to be audiobooks, oh nice, oh, that's awesome.

0:49:33 - Stephen Witt
Yeah, I love that when you go going back um, have you ever read the cane mutiny by herman wook? Watch the movie? Never read the book, actually. Um, surprisingly so. And I was like god, I'm flying through this book. Why is it so easy to read? And it turns out we could actually started scripting radio dramas. I was like, oh, this is like I'm gonna have to read it. Yeah, form of writing. If you link it to audio, people can't put it down and you, as you read it aloud, you will see the parts where you, as the reader, get bored. So you just cut those parts out and then you've got a script.

It's like Jensen Wong, so in fact, I find this part of Jensen very relatable, because I am totally paranoid of my audience being bored. I'm absolutely paranoid about it.

0:50:21 - Leo Laporte
Well, all the reviews say it's a lively read, actually along these lines. I I just I wasn't going to mention it but uh, the woman who basically created the audiobook industry passed away, uh, this week, barbara holdridge. She founded kate cademan. Oh, you may remember, jeff, you're old enough. I had this when I was a kid, the recording of d Dylan Thomas reading A Child's Christmas in Wales. Yeah, and this, this was a sold 400000 copies in the United States in the 50s we had one and it. I think I love audio books because I grew up listening to this. They went on and recorded the TS Eliot and many other poets. She and her partner in this enterprise both passed away in the last year. Uh, she was 95, barbara, uh, holdridge.

0:51:13 - Stephen Witt
So audiobooks kind of began in in 1950s well, the original audio book was actually the odyssey right there they read it.

0:51:22 - Leo Laporte
Yeah, that's her point is, uh, but yeah, books, literature was read, was spoken aloud.

0:51:27 - Stephen Witt
Yeah, from the very beginning especially if you're trying to compete. So my perspective as a writer to some extent I have competition, obviously, from other writers, but my, my big competition is instagram. Right, like, why are you not going to put this book down and scroll your feed?

0:51:42 - Leo Laporte
you gotta have a smart point of view.

0:51:44 - Stephen Witt
That's a modern point of view, but I think you're right yeah, I mean, I know from my own life, like, if a book doesn't grab me, I'm I'm in posting, you know, scrolling my doom, scrolling, um. So your competition is like, you know, these algorithmic systems that are designed to hijack someone's uh attention span and direct it towards Mark Zuckerberg's bank account. Right, and so so it's. You know, I have to do that too, except direct it to my bank account.

0:52:13 - Leo Laporte
As we have learned from your book, you can start a company in Denny's and become the most valuable company in the world. If you've just got a great idea. Thank you, it's been a great pleasure talking to you and I look forward to seeing you again. I hope the Thinking Machine, jensen Wong, nvidia and the world's most coveted microchip, stephen Witt.

Thank you so much. All right, we'll take a break, we'll come back. We have AI news and more. Lots, lots, lots, lots. This is Intelligent Machines with Jeff Jarvis and Paris Martineau. This episode brought to you by Agency. Let me spell that for you A-G-N-T-C-Y.

Building multi-agent software is hard Agent to agent and agent to tool communication still the Wild West. So how do you achieve accuracy and consistency in non-deterministic agentic apps? That's where the agency agntcy comes in. The agency is an open source collective building the internet of agents. And what's the internet of agents? It's a collaboration layer where ai agents can communicate, discover each each other and work across frameworks. For developers, this means standardized agent discovery tools, seamless protocols for interagent communication and modular components to compose and scale multi-agent workflows. Build with other engineers who care about high-quality multi-agent software. Visit agencyorg and add your support. That's agn tcyorg, thrilled, as always to be able to promote an open source collective trying to create these standards. It's really important that they be open standards and agency's doing it. Thank you, agency, for supporting intelligent machines. We're going to let Paris oh, she's back, okay, hello, paris.

0:54:12 - Paris Martineau
I know how to time my beverage breaks for the podcast. You're good. I saw you getting towards the end of the ad read and didn't put the the carafe of water back in the fridge so I could hustle back over here.

0:54:25 - Leo Laporte
Oh well, I'm glad you're going to have warm water now, but I'm glad that you got back in time.

0:54:29 - Paris Martineau
I will, but I do it in service for the podcast.

0:54:32 - Jeff Jarvis
Thank you. Did you get a cold tortilla while you were at it?

0:54:35 - Paris Martineau
No, but I did buy more tortillas today and it's, you know, it's always an honor.

0:54:41 - Leo Laporte
I wanted to show you this because there was an interesting post on Reddit this week that I mentioned. But this is the list of companies that by the end of that last year that bought NVIDIA's $30,000 H100 GPU Meta by far more than three times the next, with $350,000. X, I think, actually has gone up to $150,000, but they have $100,000 as the end of last year. Tesla only $35,000. X, I think actually has gone up to 150,000, but they have 100,000 as the end of last year. Tesla only 35,000. Lambda 30,000. Google 26,000. Oracle 16,000 and down.

0:55:16 - Jeff Jarvis
Tell me something, Leo the difference between an NVIDIA chip and a Tensor chip. Does Google have its own stuff? It doesn't need as many many it does, but I think the key is cuda.

0:55:27 - Leo Laporte
I really one of the real lessons of this and it's something probably we should pay more attention to just in general and technology is that software lock-in can trump everything else. Yeah, and because you have a, uh, the software lock-in with CUDA, that's real. I think I got the impression Stephen might have thought this too. That's the secret sauce. Yeah, I think it's yeah, you know that's what keeps them on top and when you listen to Wong, as I do.

0:55:56 - Jeff Jarvis
That's what he spends time on. Sure he does the look at our new chip? Isn't it cool, isn't it wonderful?

0:56:10 - Leo Laporte
And wonderful and he talks about huge uh server farms that are going to build on top of it, but then he really spends time on cuda and it's and it's evolution and what it can do.

0:56:14 - Paris Martineau
And the thing I don't like about it it's it's fully proprietary, so it is, oh yeah nvidia solely yeah, so what are?

0:56:17 - Leo Laporte
the alternatives that are not nvidia you know I mean uh many people think that the apple silicon chips because they have so much ram. You know, one of the things that makes these nvidia gpus expensive is the amount of memory they have in. The more memory means a more powerful ai. But apple has this unified ram architecture that lets you can buy like my m4 over here, with 96 gigs of ram. That would cost you probably 10 or 20 thousand dollars in nvidia. So apple has a great platform. Plus they have built in uh neural processing units but they don't have what's their?

software. They don't. That's the problem.

0:57:00 - Jeff Jarvis
They don't have, what about? What about google? So this is they have metal, yeah, google has its own code.

0:57:05 - Leo Laporte
Go, go ahead, benito.

0:57:07 - Benito Gonzalez
So we use CUDA for rendering video, also rendering video in Premiere.

0:57:11 - Leo Laporte
It's faster, you use CUDA for it, right?

0:57:14 - Benito Gonzalez
So we have to buy NVIDIA cards, no, no, but for Mac it's Metal, so they do have their own thing, but I don't think it's.

0:57:19 - Leo Laporte
It's Metal.

0:57:20 - Benito Gonzalez
It's not as good and it's not as picked up Like people don't use it, so you know you're using it.

0:57:24 - Jeff Jarvis
I thought CUDA was kind of a really high level of stack.

0:57:27 - Leo Laporte
Well, it is.

0:57:29 - Jeff Jarvis
I mean, you know you're using it, but you don't.

0:57:32 - Benito Gonzalez
Yeah, yeah, we're using it for rendering. You use it for rendering a video there are projects.

0:57:38 - Leo Laporte
This is one on GitHub for CUDA to Metal, the idea being maybe you could take these CUDA instructions, these dedicated instructions, and translate them in a way that you could use this on Metal Cause. Look, there's a lot of pressure to do this cause. The Apple products are much cheaper than the Nvidia products and theoretically is capable if you had CUDA. Um, this is a dumb question are you buying?

0:58:05 - Jeff Jarvis
do people buy um uh apple chips for racks?

0:58:09 - Leo Laporte
no, you have to buy the whole thing. You do, but you can serve. You could turn them all into servers. You know, they're not.

0:58:16 - Benito Gonzalez
They're not doing the nvidia play at all you know, one thing stephen didn't really mention, though is like, uh, the role of nvidia in the whole the bitcoin, whole bitcoin thing, because oh, I forgot to ask that whole section of time was all nvidia also.

0:58:28 - Leo Laporte
It was it. Was it that I forgot that drove?

0:58:31 - Paris Martineau
it in what way?

0:58:32 - Leo Laporte
oh, you had to have an nvidia card again because of cuda to do the bitcoin. Uh, all the bitcoin miners had to use nvidia.

0:58:41 - Benito Gonzalez
That's what caused the video card shortages back then, when all the gamers couldn't yeah, so all right in a way.

0:58:46 - Leo Laporte
He's what he's done I wish we had. I forgot to ask steve about this, but he's very smart, wisely leapfrogged from an industry as it was going up to the next thing, and that's that's what, jeff smith was pissed off about, but he leapfrogged to bitcoin, and then leapfrogged to cars, and leapfrogged to AI and at each moment he's caught the wave, as it's beginning to take off and now robotics. So really his genius probably is in his business timing. He's very sharp that way. I don't think necessarily it's technology, believe it or not.

0:59:21 - Paris Martineau
Were NVIDIA chips. What was powering the early wave of Bitcoin adoption and the early Bitcoin boom?

0:59:27 - Jeff Jarvis
or was this recent all of it. Okay, that was. That's what I remember, but you know, I just wanted to make sure that I could stay, you could use, you could use other cards my purchase of uh, a redacted amount of fake ids and redacted year, was powered by nvidia right everything.

0:59:48 - Leo Laporte
I'm ashamed the statue of limitations has expired, my friends were you working in the cellar with a hundred servers pumping out bitcoin?

0:59:58 - Paris Martineau
no, I just had to go through a lot of hoops to buy bitcoin at a time when dying bitcoin was not like I hold it no because I was an idiot. It was just like I. I was foolish, and I was not like I hold it no because I was an idiot. It was just like I was foolish and I was like, oh, I should with it. I bought like $3,000 worth of Bitcoin at a time when that got me multiple, many multiples of Bitcoin, and then used it to buy fake IDs for me and my early college friends so that we could drink in the city that we wanted to.

1:00:28 - Benito Gonzalez
Hey, that's how you're supposed to use that stuff, man.

1:00:30 - Paris Martineau
That's how you're supposed to use that stuff. That's the problem.

1:00:33 - Benito Gonzalez
But the issue was at the time you could not.

1:00:36 - Paris Martineau
Just, it wasn't super easy to buy big. I had to go to a. I had to like go transfer money in person. I had to go to a money gram, but in the town I was in at the time they didn't just like have a money gram place you could go to. I had to go inside a CVS and I was like where's the money gram? And the woman just points to a red wired phone like a corded landline phone. She's like pick it up. Someone will answer and I'm like what, what do you mean? So I did, someone did did. I explained that I wanted to buy a, uh, was using this to buy a bunch of bitcoin and they denied my purchase because they thought I was being scammed.

1:01:14 - Leo Laporte
Um, oh oh, that's fascinating that's really interesting, yeah, so uh, let's um talk that was a great conversation.

1:01:23 - Jeff Jarvis
I'm glad we had that Thank you.

1:01:24 - Leo Laporte
Yes, oh, he was fantastic Good booking. Let's talk about Sam Altman's little blog.

1:01:33 - Jeff Jarvis
I have an annotated version, if you want to email it.

1:01:37 - Leo Laporte
I think that would be great. The post is called the Gentle Singularity. Oh Jesus, we are past the event horizon. The takeoff has started. Humanity is close to building digital super intelligence. Wait a minute. I need some reverb to do that completely right Digital super intelligence, digital super intelligence.

That wasn't very good reverb. It was pretty good. He says robots are not yet walking the streets. Nor are most of us talking to ai all day. I personally am. People still die of disease. We can't easily go to space and there's a lot about the universe we don't understand, but we've recently built systems that are smarter than people in many ways stop right there, right?

1:02:26 - Jeff Jarvis
no, that's just false, that's just false it's right. No, he's defining smarter okay and and and and. Uh, you know. Is a calculator smarter at being certain tasks? Yes, is it an amazing machine? Yes, can it do a lot of fat functions far better than I can? Yes, but to then make the next leap, that it's smarter than us, is that's a poor choice of words. Oh well, he has tons of them in there. A species hubris?

1:02:54 - Leo Laporte
that's beyond he says and, by the way, this was the week they released the new oh three uh, what do they call it? Oh three deep research model that, uh, so far looks extremely good. He says the scientific insights that got us to systems like GPT four and oh three were hard one, but we'll take us very far. He also reassures people that don't worry, you're not killing the environment by using AI.

1:03:27 - Jeff Jarvis
I sent you my, my, and I have my annotated in red version of it, which will be so crazy?

1:03:32 - Paris Martineau
oh well, we can wait. Where is that?

1:03:35 - Leo Laporte
I'll be able to tell you both I will, I will pull it up here people are often curious about how much energy a chat gpt query uses, says allman. The average query uses about 0.34 watt hours. That's what an oven would use in a little over one second or a high efficiency light bulb would use in a couple of minutes and it uses 0.000085 gallons of water one fiftieth of a teaspoon where's, where's the footnote for that?

1:04:05 - Jeff Jarvis
do you think he's lying? Oh, I think it's. It's.

1:04:08 - Leo Laporte
It's funny math, it's all been math well, I look, even if it's accurate math, what it doesn't say is we'll take that number and multiply it by a billion queries a day.

1:04:18 - Jeff Jarvis
Yeah yeah, yeah, the worst line to me in some big sense, chat gpt is already more powerful than any human who has ever lived. The ego of that, the hubris of that, it's just ridiculous.

1:04:35 - Leo Laporte
Well, what you can say now is that he's a salesman and he's over typing it.

1:04:41 - Jeff Jarvis
That's the problem. That's what I think is ruining the technology that I do admire and use and like and root for. Is that this kind of crap is what um not only oversells it, but but uh puts a regulatory target on its back. Uh, it's going to dissipate point people. In the long run it's going to cause another ai winner because he's not going to deliver this stuff that he talks about the stuff that he talks about by the way. Here is, uh, the copy of jeff's annotated version.

1:05:14 - Leo Laporte
In red ink he writes going crazy, not smarter. Uh, something about taylor swift? I don't know, it's taylorist, it's taylorist, he's talking about how that's.

1:05:24 - Jeff Jarvis
That's the efficiency expert of years gone by this is so funny, Jeff. I love this. I wish you were my professor.

1:05:31 - Leo Laporte
AI will contribute to the world in many ways, but the gains to quality of your life from AI driving faster scientific progress and increased productivity will be enormous. The future can be vastly better than the present. Why don't you shout your annotation as I read this? Scientific progress is the biggest driver of overall progress.

1:05:48 - Jeff Jarvis
And science is being attacked.

1:05:51 - Leo Laporte
In some sense, chat GPT is already more powerful than any human who has ever lived. The hubris 2025 has seen the arrival of agents that can do real cognitive work no Period. But the world wants a lot more of both, and experts will probably be much better than novices as long as they embrace the new tools. Gee, thanks, sam. Well, but that's good, don't you? I mean, I've always said ai plus human equals something special. Yeah, thinking myself, but he's putting it above us wait till we get to ilia later people.

By the way, good news, we'll still love their families, express their creativity, play games and swim in lakes was it ever a concern that people wouldn't love their families or swim?

1:06:40 - Jeff Jarvis
in lakes I think he's projecting there he's like I don need my bunker.

1:06:47 - Paris Martineau
People don't need to find out about the bunker. It's going to be okay, we don't need to talk about the bunker. There's no bunker.

1:06:50 - Leo Laporte
What bunker. We do not know how far beyond the human level intelligence we can go, but we're about to find out. No basis for that. Intelligence and energy are going to become wildly abundant. No basis for that either. With abundant intelligence and energy and good governance, solve for trump there we can theoretically have anything else define intelligence already I love solve for trump as a comment talking about and good governments and parentheticals.

So admittedly, we have an administration in washington dc right now. That is somewhat anti-science, yeah, but on the other hand, they are also lifting thanks to elon, maybe this will change now that they've had a falling out, but they're lifting the lid on ai. They they're eliminating regulations statewide on AI, nationwide on AI, so it's a pro-AI administration.

1:07:54 - Jeff Jarvis
It's a pro-big AI company, but I think I want to see more open source stuff and that's not here.

1:07:59 - Leo Laporte
Yeah, trump also is pretty much killing NASA, or maybe I shouldn't say Trump.

1:08:05 - Jeff Jarvis
The big awful bill, or whatever they call it, is pretty much gutting NASA jared the guy he jared isaac man is gone, he's he's from my town, well, so there he dropped out of high school because he started his business.

1:08:19 - Leo Laporte
Yeah, yeah, um, advanced ai is interesting for many reasons, but perhaps nothing is quite as significant as the fact that we can use it to do faster AI research. Well, that's his business model. This is a larval version of recursive self-improvement. Yeesh, this is fun. You've got a lot of red here. Yeah, what's the? I mean, I mean seriously, what's the worst thing?

1:08:53 - Jeff Jarvis
the worst is, the is the better than any human in history. Yeah, this cost of intelligence should eventually converge to near the cost of electricity. It's so anti-humanistic. It acts as if and again, we're going to get to this with ilia's um uh speech, I hope later uh, it acts as if we're nothing but a machine. It's, it's mechanistic. Um, and then there's we probably won't adopt a new social contract all at once. That's so. That's so zuckerberg.

1:09:19 - Leo Laporte
After a sophomore uh lecture yeah, sometimes this does feel like, you know, late night college kid BS.

1:09:29 - Jeff Jarvis
The worst, the worst is down on page six of eight.

1:09:34 - Leo Laporte
I'm here. So the best path forward he says, sam Altman says might be something like this the alignment problem, meaning that we can robustly guarantee that we get ai systems to learn and act towards what we collectively really want over the long term bs.

1:09:54 - Paris Martineau
There's so many things wrong with that sentence.

1:09:57 - Leo Laporte
I don't even know where to begin robustly guarantee a lot to unpack there. Well, maybe we, maybe you feel like we can't, but that's's certainly a goal, isn't it Okay?

1:10:07 - Paris Martineau
but wait, wait, wait. So one is robustly guaranteed.

1:10:11 - Jeff Jarvis
Yeah, it's like very unique yeah yeah, yeah.

1:10:14 - Paris Martineau
Secondarily, that we could get these systems to learn and act towards what we collectively really want over the long term, yeah, good point we as a collective agree on nothing. No, we have no shared goals or wants that we could agree on.

1:10:28 - Leo Laporte
Spice girl said tell me what you want, what you really really want. Yeah, thank you, golly, I stole her joke. Um, yeah, I mean, but I think that is something it would be nice. I mean, this is, it's impossible but that's my own view yeah, this is the alignment. The alignment problem, though, which is that you know, it's the you know, nick Bostrom's paperclip conundrum.

1:10:54 - Jeff Jarvis
Is that Bostrom or is that? No, that's what's his name. That's Sutzgerber. No, that's not Sutzgerber. That's what's his name. That's the other one.

1:11:01 - Leo Laporte
Okay, it's. What's his name's problem? So we had somebody on who brought up a really good example of AI kind of misunderstanding the task in the case of finding landmines. Right, it was supposed to diffuse landmines, but what did it do instead? It found the ones that didn't go off, I think, or something like that it doesn't want to get killed?

Yeah, it's get killed. Yeah, it's alive so it's tricky, let's put it that way, to get an ai to do what you want, what you really really want. Uh, number two then, after we do that, let's focus on making super intelligent, cheap, widely available and not too concentrated with any person, country or company then open source it buster yeah, open source the weights. Well, that was ai's open, ai's original right and that was exactly that's why they were founded.

Uh, yeah, I wonder how committed he is to that point given. No, not at all. Yeah, so that's just kind of bs he just threw in there but what's he?

1:12:09 - Jeff Jarvis
what's he building, leo? What are he in the industry building we?

1:12:12 - Leo Laporte
the whole industry, not just open. Ai are building a brain for the world.

1:12:19 - Paris Martineau
The hubris, you know he had to think of that line and like, be like I got it.

1:12:25 - Leo Laporte
I got the block oh, he didn't think of it. Douglas adams thought of it in the hitchhiker's guide to the galaxy. When it turned out, the entire earth was a giant computer the mice had made to solve the problem of life, the universe and everything. Actually, that's cool, it's a nice uh, it's actually a nice parable in this case, because the super geniuses built a computer to do it and they said what is the answer to life, the universe and everything? The supercomputer said 42, right. So then they had to build an even bigger computer to answer. Well, what the hell does that mean? So you see it, it's kind of big brains all the way down uh, it's open, ai.

1:13:10 - Benito Gonzalez
That's exactly what this is, though, right like that's what he wants, that's what he's insisting yeah, make it up building a brain for the world.

1:13:16 - Leo Laporte
It sounds like the mice are at it again. Uh, we are a super intelligence research company. Yeah, why? Is that pr, why that's exactly what they are a.

1:13:28 - Jeff Jarvis
so if you're gonna buy into the super intelligence BS B, they're calling themselves a research company. What he really said above is look at what we really got going with building server farms. That's PR.

1:13:42 - Leo Laporte
A trillion dollars worth of PR.

1:13:44 - Jeff Jarvis
Yeah, well.

1:13:45 - Leo Laporte
Intelligence too cheap to meter is well within grasp I lost my place it's not that cheap. I mean, it's 20 bucks a month and you could pay even 200 bucks a month, so it's not yet too cheap to meter. That may sound crazy to say. He says but if we told you back in 2020 we were going to be where we are today, that's probably true. It sounded more crazy than our current predictions about 2030.

1:14:08 - Paris Martineau
You know, I've, I've given up, I don't think it sounds crazier than the current predictions about 2020, 2030.

1:14:14 - Leo Laporte
I feel like if you told me, in 2020 there's a pretty crazy one at where we are today, I'd be like, huh, that's kind of wild interesting but.

1:14:21 - Paris Martineau
I wouldn't be like, yeah, we're gonna be all have robots in our blood. Uh sounds, does not sound less crazy than that yeah, there's a lot of sci-fi in this post.

1:14:30 - Leo Laporte
Oh, there is computronium, there's a lot of computronium in this post finally, the coda to all of this, the last sentence. May we scale smoothly, exponentially, exponentially and uneventfully through super intelligence. Oh amen, and you write, and may the force be with us. Uh, I, I love his posts by the way, oh god, can I, can I?

1:15:02 - Jeff Jarvis
I know it's not a democracy, but can we uh segue into satskiver, into one minute of Sutzkever in Toronto?

1:15:08 - Leo Laporte
Yeah, so he was giving the valedictory address at his university. He went there, right, he went there.

1:15:18 - Jeff Jarvis
Yep, yeah Well, that's where all the things happened in the bedroom was in Toronto that Stephen told us about.

1:15:24 - Leo Laporte
Okay, if you missed the interview earlier, earlier he's talking about a computer that was built in the worst it was in a kid's bedroom, but we're called alex net.

Yes, uh, start at 50, 550. All right, ready, let me turn up the sound. Ladies and gentlemen, just a little background, just a minute here. Okay, elia sutskiver we talked about him. He was at the university of toronto, created alex net with two others, including jeffrey hinton. Uh, which solved the, actually proved to people who did not believe it that convolutional neural networks were indeed a valid way to pursue ai a big deal, a big deal deal everybody else, including, by the way, I have to point out, when Stephen Wolfram was on.

He said no, no, I still think symbolic logic is going to end up. You know, he even still was not convinced that neural nets were the be-all and end-all, and he comes from that era of symbolic AI. So here's Ilya speaking to the University of Toronto. He's a recipient of an honorary doctorate. This was just a couple of days ago.

1:16:31 - Ilya Sutskever
Slowly but surely or maybe not so slowly AI will keep getting better, and the day will come when AI will do all of our, all the things that we can do Not just some of them, but all of them. Anything which I can learn, anything which any one of you can learn, the AI could do as well. How do we know this, by the way? How can I be so sure? How can I be so sure of that? The reason is that all of us have a brain, and the brain is a biological computer. That's why we have a brain. The brain is a biological computer. So why can't a digital computer, a digital brain, do the same things?

1:17:10 - Jeff Jarvis
Paris. You want to answer that one?

1:17:14 - Leo Laporte
Oh, I don't even know at this point what would Jeff be shouting in red right now if he were.

1:17:20 - Benito Gonzalez
No Exclamation point Substrate dependence.

1:17:25 - Jeff Jarvis
Anti-humanistic. I talked about this jason earlier today, but you're too high in your own fumes go my god, amen, amen, and, and when. When the telegraph started, it was analogized to our nervous system. When the internet came along, our brain was analogized to a network. Now our brain is called a computer. Ergo, if I can build a brain, it can do everything and more that we did, and I'm the really most powerful person because I built it. And um, it's, it's, it's. It's. Living by a metaphor that is false, I love this substrate dependence.

1:17:54 - Benito Gonzalez
I want, I want benito to explain what he means by that it means that, like what we know of how our brains work might depend completely on what we're made of.

1:18:03 - Leo Laporte
We don't know, it might not be so reducible that a silicon device could do it.

1:18:08 - Benito Gonzalez
It might have to be. It might have to be wet works, like what we're made of. It might have to be that we don't know.

1:18:14 - Jeff Jarvis
Oh, they're going to tear us apart, just for our brains it's.

1:18:17 - Leo Laporte
It's the intelligent version of soylent green I need your substrate in those little people pods you see in matrix they're harvesting substrate this is where you know what is normally a computational question or an engineering question verges into philosophy, because we don't know what consciousness is. We don't know what we're doing when we think it. I think it's completely possible that we are pretty much indistinguishable from an autocorrect on steroids, that we are the sum. Well, think about it, you're the sum of our experience everything that comes out of your mouth every, every thought you have is the sum of all the things you've experienced up to now. Do you think there's something else contributing to that?

1:19:05 - Paris Martineau
yes, okay, that's the question genes biology instinct, uh, your percept, the way that you perceive your experiences is different than the way I perceive my experience.

1:19:16 - Leo Laporte
No, I agree with that, but but I'm saying so. This is really another version of the nature nurture argument. I'm saying, I'm not saying that you and I have the same process by any means, but your process it's. Is it not possible? Your process, every word that comes out of your mouth next is is merely the conditioning of all the sum of your life's experience.

1:19:37 - Jeff Jarvis
How deterministic are you?

1:19:40 - Leo Laporte
Well, I don't know. This is a problem. We don't know the answer to that question.

1:19:43 - Jeff Jarvis
This goes back to a book. I've mentioned a few times over the years how history gets things wrong and that there isn't a theory of mind. The theory of mind is wrong. We don't balance uh, needs and desires. That instead. Yes, we are a big videotape and, um, uh, we play it back and and, and that and that, that we are our own digital twin. What's the? What's the? What are the paths here? What we do, but leo, this?

What really struck me is that you have argued that we are our own digital twin. What's the? What's the? What are the paths here? What we do, but leo, this? What really struck me is that you have argued that we shouldn't. I'm going to get this wrong now. You've argued that we shouldn't expect the computer to be anything more than our brains are. Right, right that our brains are imperfect, so computer is imperfect. But he's he's doing a weird variation on that argument that says that because our brains are computer, it will be uh, as good and better than our brains. And you're saying, you're saying, accept it to be less than our brains, and that's okay. And he's saying it'll be more than our brains.

1:20:42 - Leo Laporte
Well, here's okay so here's and again, I don't know the answer to this. I'm just making the argument that is, I think, a legitimate argument, but we don't know the answer to If you and I and Paris are the sum of our experiences and some sort of mechanistic thing that creates thoughts and words and actions based on all that we have learned in our lifetime. Based on all that we have learned in our lifetime, if something could learn a lot more than us, if something could encopen and come and come encompass all the learning in the world, would it not be a superior thinking? I mean, it's still an autocorrect, but would it not be a better one?

1:21:25 - Benito Gonzalez
I don't think that's quantifiable. I I don't think that's quantifiable.

1:21:29 - Paris Martineau
And I think that's also. I think, trying to compare something that ingests large amounts of data, be it written or visual or otherwise, and then is capable of outputting responses, and response is not a one-to-one comparison with human learning or the human experience like part of the reason why we are the people we are is because, as you say, we are the sum of all these experiences we've had, and those experiences are things that are not just ingesting all the text in the land or, uh, ingesting all the youtube videos and reddit comments you can. They are experiences. I don't know. I think that this we could eventually get to the point.

1:22:10 - Leo Laporte
Admittedly, the current ais don't have all the experiences we do. That's one of the reasons robotics is important so they can have the experience of physics.

1:22:17 - Paris Martineau
But it's conceivable, if that is true, that you could create a situation where a machine would have all those experiences mr in Mett in the chat just brought up a point that I've been trying to articulate throughout this conversation, which is Mr Mett says I think one of the things that makes me so uncomfortable with all of this is all of this dehumanizes actual humans and starts looking at everyone as a version of a robot. This makes it so much easier to dehumanize people, to say, wipe out millions who are not up to snuff, and I think that's a really important conversation to be having, especially right now in our kind of political, I agree, but that's not a, but you still have to tell the truth To have that not happen.

I do think that, purposefully, we need to be centering our conversations around these topics to not dehumanize people Like I do think that's something you have to keep in your brain. You can't just ignore it.

1:23:07 - Leo Laporte
But what if it's the truth that?

1:23:11 - Paris Martineau
humans are not valuable in none of our things.

1:23:12 - Leo Laporte
Oh, I didn't say we're not valuable.

1:23:14 - Paris Martineau
I think that that's the truth anyway. I think that nothing that humans do is particularly unique when you think of the entirety of the universe, but we consider ourselves unique solely because we prioritize our relationships with others and have created this kind of society that is arranged around the idea of individuality and life having meaning.

1:23:34 - Leo Laporte
Look, I believe this is the same thing people say. Well, if you don't believe in God, then how can you have a morality? I believe that you could be moral and not kill other people just because you recognize that what we do is essentially mechanistic. I don't think those are two related processes. Some people might use that. Now I don't know what the answer is, but I have to say if we're not some deterministic mechanism process, then you have to presume some magical process, roger.

1:24:04 - Jeff Jarvis
Penrose would say quantum physics, Magic. Is that what you can't explain?

1:24:06 - Leo Laporte
it's like an accident we can't explain, but you have to conjure magic by that you have to conjure up some sort of special thing that we are. That's distinctive from a mechanistic, deterministic machine you always maybe there is, that's the soul or whatever.

1:24:24 - Benito Gonzalez
Maybe that exists, I don't know no matter how far we push science, though, there will always be that edge, god. That's why they say god is in the edges, like it's always going to be there. There's always something that's a belief.

1:24:33 - Leo Laporte
I understand that and that's a good belief. Um, also, some people believe in god and some people believe in the spirit and the soul. I don't I'm not making a case here, I'm just saying that's. What you have to presume is that there are some sort of process over and above this mechanistic process that makes us special, that a machine could not do.

1:24:55 - Jeff Jarvis
No, it's what anthony said. I mean, I'm sorry, it's what benito said it could be that we're presuming it's a process analogous to a computer, when it could be that there's other, stranger processes going on that we just can't explain. Could be, doesn't make it better, it doesn't make it above, could be, but we don't know. I'm going to read from Mark Twain's really weird book, the Mysterious Stranger Love that book, love that book. This is the spoiler of all spoilers. There is no God, no universe, no human race, no earthly life, no heaven, no hell. It is all a dream, a grotesque and foolish dream. Nothing exists but you and you are but a thought, a vagrant thought, a useless thought, a homeless thought, wandering forlorn among the empty eternities you never finished that book, by the way how do?

1:25:39 - Paris Martineau
you. How do you beat that?

1:25:41 - Jeff Jarvis
paris. Is that nihilistic?

1:25:42 - Leo Laporte
for you it's great yeah I think it's it's a it's it's now. We're in a philosophic conversation, right? Yes, we are. I guess my only point is we just don't know. No, and and I don't think it's impossible, even if we don't like it, that we are purely deterministic, mechanistic creatures, it is possible it's possible but sitz gave her saying that that's what is, that's the problem. He doesn't know it either, no, but he doesn't know, but he's claiming to know.

1:26:13 - Paris Martineau
It's just a very uninteresting world to sort of state that you know something like that as a fact and that not only do you know it, but you know that you're going to build something even better than it.

1:26:23 - Jeff Jarvis
One sentence described for all time I've done it.

1:26:27 - Leo Laporte
There's the answer. I know we need to take a commercial break. I also know that the next topic will be this Apple paper on the illusion of thinking.

1:26:34 - Paris Martineau
It's a philosophical day, man. Before we get there, can we do a brief? You know 10 minutes on WWDC and how stupid some of the stuff is there.

1:26:41 - Leo Laporte
Oh, that's your point of view. Is it young lady? All right, We'll talk about that point of view. Is it young lady?

1:26:46 - Jeff Jarvis
all right okay, wait a minute, I gotta take a break.

1:26:47 - Leo Laporte
We gotta take a break.

1:26:48 - Jeff Jarvis
I come back to this argument after this break, after this message?

1:26:53 - Leo Laporte
uh, of course that's a great conversation. It's not exactly I am. I guess it is because they talked about ai, okay, okay, uh, that's coming up with paris martineau and jeff jarvis. You're watching Intelligent Machines, this episode brought to you by Smarty. Now here's something I do know.

Smarty is the trusted leader in cloud-based automated address data solutions. They power real-time validation, autocomplete enrichment and the way they do it is the right way lightning-fast APIs. With Smart Smarties address validation APIs, you can instantly and I mean they're fast verify and standardize addresses across any system, from customer forms to massive databases, ensuring accuracy and deliverability at scale. Smarties autocomplete API this is great. You've probably encountered it already when you're in a shopping cart and you're typing your address and you notice how quickly it figures out what the rest of that address is. That's probably Smarties autocomplete API. Now here's the nice thing. There are people who use like Google to do this, but unlike Google, smarties API only offers valid address solutions to your forms in real time, no hallucinations, making form filling and checkout quick and easy for your customers and keeping your database filled only with valid addresses. That's pretty important. You don't want to ship to an address that doesn't exist. You know who loves Smartie, ship to an address that doesn't exist. You know who loves Smarty Fabletics. Companies like Fabletics have drastically increased conversion rates for new customers and one of the things Fabletics does is a US company, but they sell internationally. Right, that's been so helpful. With Smarty. For deeper insights, smarty's property data API appends 350 plus property details to address records you can get. You can enrich your database with location, structural financial insights automatically.

Faraday and other. You, by the way, see all these testimonials right there on the Smarty site. Faraday needed a turnkey way to standardize and blend address data from various sources to feed their customer behavior prediction models. The CTO of Faraday said, quote Smarty helped us scale our business and focus on our core competencies, and their technology has kept up with the cutting edge. Even 10 years later, smarty still has the best address API in the market. Still, they're getting better every year.

Smarty is a 2025 award winner across many G2 categories best results, best usability, users most likely to recommend. High performer for small businesses yeah, it's not just for big enterprises. Also, you'll be glad to know, smarty is USPS, cas and SOC 2 certified and HIPAA compliant. So whether you're optimizing user experience, cleansing legacy records, implementing identity verification or scaling data pipelines. Smarty delivers enterprise-grade performance with unmatched speed.

Can you say 25,000 plus addresses a second. You can even get things like rooftop-level level precision, which in some areas is really critical to delivering things. The door is there, not there and easy developer integration. It's so easy. This is a great api. Smarty makes it simple to add accurate, reliable address features right into your product or platform. With powerful apis designed for speed and ease, you'll have address data that works, no matter what you're building. I hope you're interested. Try it for yourself. Get 1,000 free lookups when you sign up for their 42-day free trial. Visit smartycom slash twit to learn more. That's smartycom slash twit. We thank Smarty so much for supporting this week in tech. No Well twit, but I am specifically Thank you All of twit. We thank Smarty so much for supporting this Week in Tech. No Well Twit, but I am specifically Thank you All of Twit. They bought the whole network.

1:31:01 - Paris Martineau
So, paris, you were unsatisfied with Apple's WWDC keynote. We just need to be talking about liquid glass, which is the design rollout that Apple has announced. Well, one. This also is coming at a time where apple's big change is that they're changing their ios numbering from just being numbers to being years, which I guess makes sense. But it's just ridiculous to go from whatever number ios we were at before, whatever it was like. Well, that's the problem. You don't know 26, but why do I need to know leo? I don't. That's not a number I need to have, and it's dumb for it to be I have to look it up.

1:31:31 - Leo Laporte
It's just the year or no, it's I?

1:31:32 - Paris Martineau
was 25 or it's. I was 26, but it's coming out in 2025. I don't like that.

1:31:37 - Leo Laporte
Secondly, well, your new car. When you buy it in september, you get the 2026 model.

1:31:42 - Paris Martineau
You have a 2026 vw, I think that's wrong and dumb too, and it doesn't mean that apple should do it as well. Um, but liquid glass one terrible name. It sounds like it's the name the name of a style of t-shirt that would be sold by guy fieri um two I do have to agree with you on that. It looks, you don't think it's pretty. Not really. It might be pretty in some of the photos that they're showing. You use a Mac, right? Yeah, I use a Mac.

Imagine having that busyness on all of your screens all the time, 24-7. I don't want that sort of visual clutter. Screens all the time 24 7. I don't want that sort of visual clutter. I don't like that. A toolbar will be floating always and then vaguely reflecting what's under it. It just doesn't. It doesn't seem useful. I remember seeing someone doing a long Twitter thread on like. I assume that that uses I'm going to probably describe it wrong some sort of computing power or like memory, in the same way that, like motion effects on the iPhone, also uses something on an iPhone.

1:32:47 - Leo Laporte
They even said they couldn't do this until recently and in fact, by the way, Microsoft tried to do this in Windows Vista. They called it Aero Glass and it didn't. It was terrible because there wasn't enough horsepower to do it.

1:32:59 - Paris Martineau
So Apple believes there's enough horsepower in the modern phones and iPad, but it's like from a company that is already building a phone, like within a company that is already building a phone. That like, within a year of me getting whatever new iphone, my battery power is already like. I'm already losing 50 of battery by the time 2 pm rolls around. I don't need any more doodads and gizmos making that worse just for vibes and not even particularly good vibes did you?

1:33:24 - Leo Laporte
did you put something in her cereal?

1:33:26 - Jeff Jarvis
she's sounding like you, jeff yeah, yeah, I'm standing back watching.

1:33:30 - Paris Martineau
I'm interested in unnecessary visual clutter from the company that's supposed to be, mr design no I'm not, I'm not willing, I'm single. Right now I can't have an android.

1:33:41 - Leo Laporte
This is also I will say, this is something I've noticed now a blue bubble yes, this is unfortunate.

1:33:47 - Paris Martineau
This is a data point I've learned from. I will say there's probably like at least three people that I've gone on dates with in the past, all men in the past six months that have said they've switched from android to iphone because they thought the social stigma was impacting their dating prospects. I realized this the other day because I was like on a date with somebody boys don't oogle girls.

He was showing me in a notes app and I was like, what notes app is this? And he's like, oh, google notes. And I'm like, oh, so you were an Android user. And he's like, don't judge me, I did um.

1:34:18 - Jeff Jarvis
Did you judge him, did you?

1:34:19 - Paris Martineau
just a little bit, I thought. I thought it was kind of funny that he switched to Apple just for the, so I judged him a little bit for that. Instead, um, I have to ask my daughter about this.

1:34:28 - Leo Laporte
Uh, she's a devout uh android user. She hates apple.

1:34:32 - Paris Martineau
She thinks apple's the devil that's like that's a class of android user as well I mean I don't have any strong feelings about either. I just think that apple has done a good job of what they set out to do, which was apply unnecessary and unwarranted social stigma to using an android device and direct was their settings and they've they've accomplished that that was their all.

1:34:54 - Leo Laporte
Right, let's, since we are okay, my one talk.

1:34:57 - Paris Martineau
My one pro for uh wwdc is I think it's great that they're doing.

1:35:02 - Leo Laporte
You can do polls in group chats that'll be really start doing them the minute that arrives. Yeah, we have a group chat. I don't care, I'm fine, I'm happy with my android, I live, love you to google.

1:35:16 - Paris Martineau
Oh, yeah, well, I don't know what will happen if we do a poll you might not see it, you might not get to vote yeah, maybe that's the thing I will say you snobbing jeff in our group chat as an android user does mean that I keep getting strange impacts where it keeps saying that I renamed the group chat, but I didn't.

1:35:30 - Leo Laporte
No, I've been renaming the group chat by the way, as you should Listen.

1:35:34 - Paris Martineau
I knew it was probably you, but I was like I can't you know.

1:35:37 - Leo Laporte
I decided on human beings.

1:35:38 - Paris Martineau
How is that my fault? It seems like it's probably your fault, because you're in there as an Android user because you're green.

1:35:45 - Jeff Jarvis
I have to say he's not, he's not a green bubble.

1:35:50 - Paris Martineau
We're all the same color bubble you guys keep saying this my bubbles are all green.

1:35:56 - Leo Laporte
In the group chat and that's mine, are green. My bubbles and all the rest of you are gray right yes, but that's what I mean by green bubble.

1:36:04 - Paris Martineau
It's a green bubble based message. It's a shorthand for what is now rcs, but that doesn't matter, because rcs is, it's a green bubble based message.

1:36:10 - Leo Laporte
It's a shorthand for what is now RCS, but that doesn't matter. Because RCS is fine? Oh, because we have to be RCS, because one of us is an Android user, we can't use it.

1:36:16 - Jeff Jarvis
What was this show called for?

1:36:19 - Paris Martineau
I don't know what you're talking about.

1:36:20 - Leo Laporte
Hey, by the way, by the way, the new Android 16 came out this week. Are you excited, Jeff? Well, it's fine. It's really not that interesting.

1:36:31 - Paris Martineau
They didn't redesign it or anything.

1:36:34 - Jeff Jarvis
It worked. That's the thing. It's not worth getting upset over.

1:36:38 - Leo Laporte
I think that's one of the reasons we don't do a show about Android anymore. It's because it's just a tool. It doesn't spark debate it doesn't spark joy.

1:36:51 - Jeff Jarvis
Or not even anything emotion it's just like yeah, it's like a hammer. I'm I'm happy as hell because I got I. I got an invitation to next week's chromebook event.

1:36:57 - Paris Martineau
But there's, are you gonna take them to the cleaners with concerns about google workspace?

1:37:02 - Jeff Jarvis
well, there's that. Yeah, I just yelled at two google people about that today good, so there is they're announcing new Chromebooks, or what's the deal?

1:37:09 - Leo Laporte
Yeah, yeah In New York.

1:37:10 - Jeff Jarvis
The Mondo one yeah.

1:37:11 - Leo Laporte
New York oh, it's exciting.

1:37:13 - Jeff Jarvis
I'll fill you in next week.

1:37:15 - Leo Laporte
So I was very so. I thought Apple had a thread and needle. You remember, last year at their developer conference they announced AI features. They never delivered. They still haven't delivered, particularly Siri. They really felt like the world was going towards AI chatbots and they were the original chatbot. Siri should be able to do it, but Siri couldn't do it. They've admitted since then that Siri kind of did it, but it made enough mistakes that it wasn't something they were going to put out.

So they had a big ad campaign. They killed they, just they, they. They said, sorry, we're not going to do it. And so everybody was wondering what are they going to say on monday? Are they going to say, oh, we blew it? Are they going to not mention it at all? Are they going to forget? Ai exists? And I think they did the right thing, which was, instead of focusing on a chatbot which they still don't have, they said you can use open ai. Uh, you can use other chat bots if you want. They incorporated, for instance, chat gpt into their very pathetic image, playground image, design thing. But mostly what they did is they said uh, ai should be in the background, it should help you without being a big ai thing without being a transparent, slightly shiny overlay the design is separate from the ai.

The design is just sorry, I'm not gonna let it go did you have some red bull today?

1:38:41 - Jeff Jarvis
uh, paris, I truly did not.

1:38:43 - Paris Martineau
I will say, coming into this podcast, I was like, wow, I'm a bit low energy, I didn't have much coffee, I don't know. I didn't realize that I had this passion for liquid glass until I started saying it, you can see, I'm turning red.

1:38:55 - Jeff Jarvis
You're kind of your volcano behind you. You're living. I am.

1:39:00 - Paris Martineau
Yeah, I am, that's good, it's bubbling.

1:39:03 - Leo Laporte
I appreciate the passion. You are a lava lamp. Apple's decision was we really want to focus on stuff with our models that you can do on device. That's good for the, you know. For those of you who are worried about the environment, that's good because it's your device, not, you know, some big server farm. It's people worried about privacy. That's good because only you see what you're doing. It is a little limiting, right? That's going to be the question Can their AI do the things they've promised on device? But I think it's a smart move. I think it puts them in a different venue than Google and OpenAI and Anthropic. They're going to do their own thing. They're going to say if you really want Google or OpenAI or Anthropic, you can do as as I have done put one of those apps on your phone and use that on your iphone. That's fine, but we're going to keep it private. We're going to keep it to ourselves, and I think that's an interesting play.

1:39:56 - Jeff Jarvis
I think they're smart so what does johnny I've do?

1:40:00 - Paris Martineau
remember he's doing a hardware device with johnny I've gets 70 billion dollars to stand in a cafe with sam oltman while wearing six and a half billion, there's a big difference.

1:40:11 - Leo Laporte
Okay, actually there's no difference. From your point of view of your lifestyle, it doesn't matter anything above one billion, it doesn't matter. Uh, yeah, no, I think that's the point. Is that open? Ai thinks it's going to be about chat is this?

1:40:26 - Jeff Jarvis
is this an application or a platform? That is ai an application or a platform? I'm coming to think more and more that it's a platform. It's just like the internet and ai is going to do stuff, and what you're going to want is what's built on top of that and that's exactly apple's notebook lm.

1:40:41 - Leo Laporte
That's exactly apple's point of view. Actually, notebook lm is really interesting. Uh yeah, apple doesn't have anything like that. I was just reading a very I was kind of a little bit surprised a piece by a guy I wouldn't think of as being a Notebook LM fan but who puts all of his articles and research into it and then queries it. He says the RAG, that Retrieval, augmented Generation, the ability to have the ai only work on the documents you provide, is really powerful. So are you using it, jeff, for your?

1:41:12 - Jeff Jarvis
books, um no, but what I've decided to do this is why I bought a tablet recently is that's why I could annotate uh, sam's piece, because I'm trying to get in the habit of it is for the next book about, about mass. I want to put, I want to keep everything in p, everything in PDF, so that I can then play with notebook, not to write anything for me, but to find the stuff To query it. To query it, yeah, because I'm writing this on a very big topic and so I need to try to start to organize it in my head, and I wonder what help it will be or not. I couldn't do that in the last books because I showed, know, I showed you the you know not only are there, is there a ton of books there, but there's also a ton of there's. You can't even see it. There's thousands of pages of papers that I've printed out.

1:42:00 - Leo Laporte
Won't it be nice when everything's digital and you could put it into an AI? No, it's not fun.

1:42:04 - Paris Martineau
I feel like I don't like having my resources be digital it's going to be hard for me to get used to this. I might get embarrassed. You crack me up I don't.

1:42:14 - Leo Laporte
You should have been born in the 30s. Where do I put the thread on my file folder?

1:42:19 - Paris Martineau
this is one of three.

1:42:20 - Benito Gonzalez
I just created something new to have in my book. You can't control f that, though you know like I mean, that is the one problem.

1:42:27 - Paris Martineau
but I think when it comes to first pass research for me, like going through primary documents I've perhaps, like already scanned, realized are useful to whatever query I'm doing, I feel like I engage on a deeper level with the text, going through it IRL.

1:42:45 - Jeff Jarvis
I, as you can tell, I like to annotate things, I underline things, but I'm going to hope that that makes some difference. The other thing is that with books on Kindle, this is what Stephen Johnson did. One of the you know the editorial director of Double Kill M is that he's so damned organized, he's written 15 books. When he reads books, he marks them I think it's Kindle. He reads and then all of those snippets get saved. And then his first use case for Notebook LM was to put all those snippets into a queryable database so we could say well, who else said this or who disagrees with that? And so we got those pieces of books.

Now I use a lot of books that are old and that's a little bit difficult, but I can scan them in. It's going to be worth the effort. I don't know, but I'm really interested in playing to see what impact Notebook LN might have, mainly in my research and organization I'm editing. I'm hoping to submit the manuscript for Hot Type as soon as next week and then, once I do that, I'm going to put that whole thing in Notebook LN, just after I've already handed it in, just to see what it does I should point out I'm not a huge fan of gemini.

1:43:51 - Leo Laporte
I should point out that that kind of rag capability is available in other tools. Chat gbt has a has added a project feature. I think oh three, the new oh three, is actually the smartest ai I've used. It's really really good that's that's.

1:44:05 - Jeff Jarvis
It's hard for what I'm doing. How much smarter is smart? You know, I don't know Right. Right, the other thing about notebook LM that they just announced. I don't know if you talked about this last week in my absence, but you can now share a notebook. Yeah, and I think I can imagine that for journalism. I mean, imagine Paris if you were able to put in tons of raw documents from an investigative piece and share that with the public and see what they asked of it and what they found in it.

1:44:33 - Leo Laporte
Wouldn't that be interesting.

1:44:34 - Paris Martineau
I mean, that is a thing that I think a lot of investigative outlets are now not a lot, but some investigative and collaborative outlets are now kind of trialing as a tool and this is also something you're starting to see in newsrooms is it has to come with a lot of caveats, saying like hey, this is an LLM, so like it may be producing answers that are not granted reality. But if you are interested in this as a potential surveying tool, like, use it, ask some questions and see what you get, which I think is interesting.

1:45:04 - Leo Laporte
You may want to use Google's AI, because ChatGPT is facing a little court action. We're going to talk about that in just a little bit. You're watching Intelligent Machines, jeff Jarvis and Paris Martineau. Nice to have you Our show today brought to you by Monarch Money. Oh, I love this thing. I love this, monarch Money. Oh, I love this thing. I love this.

Finances can be messy and confusing. Monarch Money is kind of like your personal CFO, giving you full visibility and control so you can stop earning and start growing. It's more than just your average budgeting app. Monarch Money is a complete financial command center. I put everything in it my accounts, my investments, my property, my cars, everything, everything. Put everything in it my accounts, my investments, my property, my cars, everything, everything you have in it, and then you add your goals. So now you're not just managing this stuff, you're starting to build wealth. Right now, 50% off your first year for our listeners.

Just for me, seeing in one place a dashboard of how much I lost in the stock market or how much I gained in the stock market, how much my house has gained All of that in one place is a very useful tool for figuring out what I could do, what I can afford. I'm saving up for travel because that's one of my favorite things to do. Can I put more away? Can I put less away? It's so valuable to do and because I could do it with Lisa, with my partner, it makes it even more powerful. Start managing your finances to build the life you actually want. You see, without a clear financial picture, those financial dreams can easily feel out of reach.

Monarch makes managing money simple, even for busy lives. With all your accounts, your credit cards, your investments in one place, you'll always know where your money stands. No hassle, really no data entry. It tracks your spending, your savings, your investments, effortlessly auto-categorizes. You can focus now on what matters most making your biggest life goals a reality. It's a financial tool people actually love.

Join over 1 million households using monarch money. Named wall street's journal's best budgeting app of 2025, I would agree the top recommended personal finance app by users and experts, with over 30,000 five-star reviews. Get control over your overall finances with Monarch Money. Use the code I am at monarchmoneycom in your browser for half off your first year. That's 50% off your first year. That's what I did at monarchmoneycom. You please use our offer code. I am monarchmoneycom. Offer code I am. We thank monarch money for their support and for keeping my finances in order, because it's not easy. Uh, open ai is has been ordered. It is in the new york times at all against open ai. They're claiming it of plagiarism. The judge ordered them to save all their chat logs, even the logs of those people who told chat gbt.

Don't remember this oh uh, this seems like a bad idea. Um, in the filing the response, openai alleged the court rushed the order based only on a hunch raised by the New York Times and other news plaintiffs. And now, without any quote, just cause and quote. Openai argued the order quote continues to prevent OpenAI from respecting its users' privacy decisions. This includes ChatGPT Free, plus and Pro. If you use the API, same thing. Thank you, new York Times. Nice job, new York Times. Well, and I got to blame the court on this one.

1:49:01 - Jeff Jarvis
Well, no, New York Times asked for it.

1:49:04 - Leo Laporte
I blame the New York Times. They're looking. It's like a fishing expedition. What are? They going to do go fishing for all of us all of our prompts, all of our responses, looking for a sentence that appeared in the new york times and the times was objecting to open ai asking how did you what?

1:49:21 - Jeff Jarvis
what 100 questions did you ask to get a repeat a paragraph?

1:49:24 - Leo Laporte
yeah, well, we saw what one of the it was in the pleading. One of the things they did is they. They gave it the first two paragraphs in an article and then said what would come next. Well, yeah, uh, so normally this is not unusual, see, this is why the court got confused. It's it's normal in a lawsuit like this to tell the company being sued no, you can't delete emails, you can't delete any communications. You have to preserve everything for discovery. That's normal. This is different because it's not it's us. It's us, it's our material. Uh, all right, I just thought I'd let you know. Maybe you do want to use that book. There was yet another.

1:50:08 - Paris Martineau
There was another lawsuit that just broke today, like while we were recording this right News of, I think, disney and others suing Midjourney.

1:50:19 - Leo Laporte
Right.

1:50:20 - Paris Martineau
In particular, they, disney and Comcast Universal filed a copyright lawsuit against Midjourney today, calling its popular AI-powered image generator a quote bottomless pit of plagiarism for its use of the studio's best-known characters. Look, there's Shrek.

1:50:36 - Leo Laporte
Look there's Darth Vader. Look, there's Spider-Man and whatever those things are.

1:50:41 - Jeff Jarvis
Look, I have to insert here just as an egotistical aside. You know that Murphy Brown once called me a bottomless pit of hate.

1:50:49 - Paris Martineau
I just wanted to throw that out, that's lovely they're not wrong.

1:50:53 - Leo Laporte
Um, mid Journey could, as some other AIs do, say no, I can't do any copyright material. I frequently ask chat GPT to take a picture in my photo album and turn it into a disney cartoon, and I love it and disney hates that.

Well, you know there's no white doesn't appear in it yeah, uh, so I you know, for instance, I asked it to um, I have a. I had a picture of my piano where I play and I asked it, let me see if I can find it. I asked it to take that picture and make it in the style of Disney, with little bluebirds singing around it, like Snow White had those little bluebirds singing around it and it turned it into this. I think that's pretty cool. Yeah, made me feel good about practicing yeah who's the performance in the podcast?

the bust is a bust. My, uh, my piano teacher gave me of Chopin, ah it's not a perfect.

1:52:03 - Stephen Witt
We do want.

1:52:04 - Jeff Jarvis
We do want the recital.

1:52:05 - Paris Martineau
Leo, you're not getting off this yeah, we are going to need you to perform for at least an hour or two of the 24 hours. You understand, I'm I'm just learning right oh yeah, that's why it's all the more by the end of the two hours you're going to be getting really all right, I'm going to play and you can sing along the famous irish song cockles and muscles would you like to hear it?

1:52:24 - Leo Laporte
of course, yes ladies and gentlemen, cockles and muscles, I I once met a girl named sweet molly, I didn't realize this is a real song, to be honest she rolled A wheelbarrow through streets broad and narrow, singing cockles and muscles alive, alive.

1:52:59 - Paris Martineau
I'm a beginner.

1:53:03 - Leo Laporte
Thank you very much.

1:53:04 - Paris Martineau
That was really great I appreciate all the applause.

1:53:08 - Leo Laporte
I don't, yeah, I kind of feel I kind of no golf claps please. I kind of feel like I hate to say it, but disney and and darth have a point here that mid journey is in fact a quintessential copyright free writer and a bottomless pit of plagiarism.

1:53:23 - Jeff Jarvis
That's why we love it but um, so you can be inspired by van gogh, you can be inspired by disney. How did you see each?

1:53:36 - Paris Martineau
they're not, inspired.

1:53:38 - Leo Laporte
They are copying it on the right are actual pictures from uh star wars of darth vader. On the left the generated pictures of darth vader that's a little more than inspired.

1:53:50 - Jeff Jarvis
It's a little. Yeah, I think it's better.

1:53:52 - Leo Laporte
Actually it's it is actually better. And, by the way, at the same time as they're suing, a lot of high hollywood producers are using these tools to make pre-visualizations and storyboards for their upcoming movies. Has become the all the rage. Here is uh, here's uh. Is this elsa in frozen? On the right, the actual image of elsa is slightly different. On the left she looks a little different. That's the ai version, so I think my journey did perhaps, um, it wasn't as cautious as it could be about reproducing copyright material. Here is Homer Simpson in reality on the right, and actually a much better Homer Simpson, actually. It is.

1:54:41 - Paris Martineau
I have an embarrassing admission, which is that I've never seen all the Star Wars movies, and it's gotten to the point where now I feel like if I see them, I shouldn't just watch them in my home, I should do a bit out of it. If Club Twit is ever interested in watching the Star Wars movies with me for the first time, let me know. Oh, I don't know if we're allowed to do that in the club.

1:55:03 - Benito Gonzalez
So which ones have you seen? Which ones have you seen?

1:55:05 - Paris Martineau
Oh, I haven't seen any of them. I don't know why I phrased it like that.

1:55:08 - Leo Laporte
Any of them. How do you expect to date men if you haven't seen Star Wars movies?

1:55:14 - Paris Martineau
I can explain as to why I have not seen any of them in more detail when we're not on a podcast. Honestly, people keep showing me the Star Wars movies and then we end up talking about something else and then I don't finish the Star Wars movie.

1:55:27 - Leo Laporte
I hate to admit it because it's bad for my geek cred, but I never like them. They're childish to me. Well.

1:55:32 - Paris Martineau
I mean, this is the thing is, I hear mixed opinions and then so I'm like should I be devoting my time to watching, catching up on all this poor? Pop culture. I know, but people keep telling me Andor's really single star.

1:55:44 - Leo Laporte
No, you can't oh, wait a minute. And or precedes all the movies chronologically. You can, in fact, if you were to watch them in order, oh this is, you would start with andor.

1:55:55 - Jeff Jarvis
So yeah, she's, she's.

1:55:56 - Leo Laporte
She's a star wars virgin, it's true, where should she start chat room if she's a star wars virgin? Not with episode four, clearly.

1:56:06 - Paris Martineau
What is that? What's the first one that came out, chronologically, episode four. That's why not. Why wouldn't I watch the first one they ever made? Because it's out of order. But it's not out of order if it's the first one that came out.

1:56:19 - Leo Laporte
It's the order that Lucas made it in, but then he went back and did one, two and three. Unfortunately, that's where jar jar binks appears, and that's not the highlight of the show patrick delahunty.

1:56:31 - Jeff Jarvis
Start with and see I knew that this would get people star wars. This is the issue every time I ask people about this.

1:56:36 - Paris Martineau
They give me like 75 tasks I have to do and I'm like that's a lot the other.

1:56:42 - Leo Laporte
The problem with doing it in in the order that they occurred in this imaginary space is the technology will be really really good at the beginning and then go way down.

1:56:53 - Paris Martineau
But isn't that the true fan experience? Isn't that how you guys all experienced it?

1:56:57 - Leo Laporte
in just terms of increasing with episode four, shouldn't she bonito?

1:57:01 - Benito Gonzalez
start with the start okay, so there are two ways you can think of. The way I think about this is that only episode four, five, six, one, two, three are real star wars, because those are made by lucas, those are made by the actual those are the lucas ones everything else is fan fiction. Everything else is fan fiction. Oh, you asked the wrong guy oh my, oh my.

1:57:21 - Leo Laporte
So she should watch four, five, six, one, two, three.

1:57:23 - Benito Gonzalez
She should only watch four, five, six that's what I think that's listen, that's the most those are the three best and it's easier for you it's only six is blowing up yes, I know that's a hot take and I'm gonna get blown up on the internet for that. I never put that out on the internet before, but that's what I believe and or rogue one, then a new hope. Episode four I'm not saying the new stuff is bad, I'm just saying it's fan fiction, it's not made by the guy who made it yeah.

1:57:52 - Paris Martineau
Yeah, everything I've seen of the new stuff seems like it's fan fiction. There's like a baby Yoda. So you've seen enough.

1:57:57 - Leo Laporte
You've seen clips.

1:57:58 - Paris Martineau
You kind of know, I know that Yoda's a baby and that's weird, but it's also not. There's something called grogu.

1:58:04 - Leo Laporte
I don't really know and I don't care to know. Don't watch the mandalorian. I don't jammer, I have tried to watch them.

1:58:10 - Paris Martineau
I somebody showed me one or two episodes the mandalorian and I wanted to shoot myself.

1:58:14 - Leo Laporte
To be honest, um that's not a good netflix and chill yeah, it was not jammer b says andor is right before rogue one, which is right before episode.

1:58:23 - Jeff Jarvis
Yeah, andor is the earliest it's when do you have?

1:58:26 - Paris Martineau
any strong opinions on the Star Wars franchise.

1:58:27 - Benito Gonzalez
Everything is based on four, five, six, so like that's kind of-.

1:58:30 - Jeff Jarvis
No, because I haven't seen one in 20 years. See four, five, six, almost like you, I think Benito's right.

1:58:36 - Leo Laporte
See four, five, six.

1:58:40 - Benito Gonzalez
Those are the first three that came out, and if you, you don't have to watch anything else you can stop that's because you will just have the.

1:58:46 - Paris Martineau
You'll then be up on the lore that's all you need right this has really come to a fruition, because one of my favorite podcast, uh jurors they sometimes on their off weeks we're not, they're not doing like dnd comedy stuff that I like they'll do like movie reviews or like weird tv shows, and normally when they're watching obscure or weird stuff they start the episode before they get into their funny bits, by recapping what it was they just watch for people who haven't seen an obscure cartoon from the 80s but they've been watching the new released, I guess, star Wars movies that have been in theaters lately and they don't give a recap because they're like oh, everybody knows Star Wars, some of us want to know what you're talking about.

1:59:23 - Jeff Jarvis
What kind of idiot wouldn't have watched that. Leo, do you have any star wars imitations in your repertoire?

1:59:30 - Ilya Sutskever
mrs don't like a star wars. I know he would I do know that.

1:59:38 - Benito Gonzalez
I do know that floppy guy also these movies are made for children.

1:59:42 - Leo Laporte
They are made for children yeah, they were made for children and I was an adult when they came out, and that's really the big difference. So the first one came out in 77. I was already 21. It for me it was like it was good, but it was like I like 2001 a space odyssey I thought there were better sci-fi movies out there, um, and it is kind of a. It's kind of a fairy tale it's's not sci-fi.

2:00:04 - Benito Gonzalez
Yeah, it's not sci-fi, it's fantasy. Yeah, 2007?

2:00:06 - Leo Laporte
1977.

2:00:10 - Jeff Jarvis
Thank you.

2:00:12 - Leo Laporte
Okay, so you were not only an adult, you were like middle-aged.

2:00:17 - Jeff Jarvis
I was a columnist in San Francisco.

2:00:22 - Leo Laporte
You're like there's no reason I'd be watching this child movie. This is a movie for children. I think I should.

2:00:26 - Paris Martineau
I'm the most eligible bachelor in San Francisco.

2:00:29 - Jeff Jarvis
I'm going to go watch Dick and Liz in. Who's Afraid of Virginia Woolf? I'd rather see that.

2:00:36 - Paris Martineau
Well, I do think the incest potential plot line could be intellectually stimulating. All the lightsaber effects. Not my speed.

2:00:43 - Leo Laporte
Well, it is kind of based on the hero's journey. It is a little bit.

2:00:46 - Paris Martineau
I mean, I'd hope so. It's a.

2:00:49 - Leo Laporte
It's a big budget movie, but it's a classic example of the hero's journey luke skywalker's story of redemption. And then he falls in love with. Oh, should I tell her now or no? I know spoilers. No, no, let her.

2:01:04 - Paris Martineau
Let her find that stuff out, if she doesn't know, I vaguely know, he falls in love with his sister, right? That's why I said the incest plot line.

2:01:12 - Leo Laporte
Say that okay do we want to talk about the apple?

2:01:21 - Paris Martineau
uh illusion from a very on topic thing or do we care, I want to hear what you have to say about it yeah there's all sanovsky on.

2:01:29 - Jeff Jarvis
It was really good. Futurism was good. There's there's comment on it, so what they did in this apple thing I I think, is I'm I'm not crazy about the what they did, but anyway what they did let me do a preamble, if I may, real quick, Because what strikes me here is that when Apple couldn't conquer advertising, they said that, oh, we're the privacy company because they couldn't sell advertising. And now they can't conquer AI. So this is kind of the way they're shooting AI in the foot.

2:02:00 - Leo Laporte
Honestly, these guys are researchers.

I doubt they're carrying water for Apple's corporate marketing. But Corporate marketing is giggling. Corporate marketing didn't mind it. So they're testing these new large reasoning models right, these ones that are thinking Now. Unfortunately, when the paper was done it was Claude 3-7. It was ChatGPT-01 and 03, deepseek R1. So there's newer stuff. But what they did is they gave it classic problems in not exactly even computer theory, like the Towers of Hanoi. You know the Towers of Hanoi right? That's the one where you got three pegs and a bunch of disks and you're to move the pile of disks from the first peg to the third peg without ever putting a larger disk on top of a smaller.

2:02:53 - Jeff Jarvis
I'm so bad at it I am not AI.

2:02:57 - Leo Laporte
Steve talked about this yesterday and I actually went out and bought it because I grew up with the towers of Hanoi when I was a little kid so I said I got to have another one. So once you figure out the algorithm, it's pretty simple. A human can solve it, a kid can solve it. I solved it when I was a kid and in fact I think probably this is before I did any computer programming, before we even knew about computers. Practically I think it helped me when it did come time to kind of learn how to program. I kind of understood algorithmic solving Because essentially what you're doing is repeating something over and over again.

2:03:32 - Jeff Jarvis
Is this a spoiler, Leo? Are you going to spoil it for me?

2:03:34 - Leo Laporte
No, just look for the patterns, jeff. Okay, anyway, it turns out that what you can do quite well, these reasoning machines fail at miserably. So this first graph on the left along the x-axis is the number of disks one, two, three, four, five, six. You can make this an infinite number of disks. They went all the way up to 20 disks, which for a human, once you've solved a three disk, would be trivial, because you just do what you did for the first three, over and over and over again to get to 20 and in fact up at the top. Here you're looking at claude 3.7 versus the thinking version of claude, both did fairly well, nearly 100 solutions for the first three discs. After three discs, claude 3.7 fell off the map. It its accuracy went way, way down.

2:04:28 - Jeff Jarvis
And finally, got had leo. It had a complete accuracy.

2:04:30 - Leo Laporte
Collapse, yeah, by eight discs it could no longer solve it. The thinking one did go a little farther, but again by 10 discs, same thing. Uh, let's see response length position. Yeah, that's less important. I mean, the main point is they couldn't do it as it got more complex, they just failed. So are they thinking? I don't know, I'm not sure what we learned here, except something that anybody's used ai much knows that it's not. It's gonna fail.

2:05:06 - Jeff Jarvis
It's not always perfect well, here's, here's the question I was thinking about earlier today. Does it know the definition of failure?

2:05:14 - Leo Laporte
well, that's to me a more interesting result that a lot of studies are having. What's very clear is, if the ai is not given enough information to answer your question, if notebook lm doesn't have enough documents to answer your question, it will. Hallucm doesn't have enough documents to answer your question it will hallucinate.

2:05:29 - Jeff Jarvis
That's where the trouble comes, because it wants to please you.

2:05:31 - Paris Martineau
It is written to please because it's understanding of failure is not producing an answer, for failure should be both. If the answer is incorrect, you're both answerable.

2:05:41 - Leo Laporte
All it's doing, remember, is it's putting one token after another. Yeah, if it has, you know information, it'll put it there it will do the best, you know. It will take the best token. Absent any information, it'll just take whatever it comes up because it has no meaning.

2:05:57 - Jeff Jarvis
It has no sense. It's meaning it's just one token after the other, and if success or failure, if it runs out of you know, use useful tokens.

2:06:05 - Leo Laporte
It's got a bunch of crap. It will just put that in there. It's there's no. So that's why it hallucinates more and I mean, I think it's pretty straightforward. Uh, they did checker jumping, they did blocks world, they did river crossing, similar problems. Let's go down to the. They did a lot of this. Oh my god, they did it over and over and over again. Let's see that let's be in a researcher.

2:06:28 - Benito Gonzalez
That's how you do science, leo yeah, yeah, you keep doing it.

2:06:35 - Leo Laporte
It feels fairly like a fairly trivial study. Where's their conclusion?

2:06:40 - Jeff Jarvis
they don't even. It's in the paper itself, isn't it down there?

2:06:44 - Leo Laporte
I thought it was by the way, one of the things they didn't do, which everybody who uses AI knows, you can also say think harder or, even better, write code. Had it been asked to write code to solve the tolerance at Hanoi, I think it could have solved it perfectly forever.

2:07:01 - Jeff Jarvis
Page 11 of the paper is the conclusion. We systematically examine frontier large writing models through the lens of problem complexity. Our findings reveal fundamental limitations in current models. Despite sophisticated self-reflection mechanisms, these models fail to develop generalizable reasoning capabilities beyond certain complexity thresholds. That's valuable, I think.

2:07:27 - Leo Laporte
actually I think it is. Yeah, they can't think. In other words, that's good.

2:07:33 - Benito Gonzalez
What was interesting it's kind of quantifying what we can ask them, right, like that's kind of what I'm saying, right, that's good.

2:07:38 - Leo Laporte
To me, that's what you want to learn. That's what this show's about.

2:07:48 - Jeff Jarvis
What can you reasonably use AI to do and what is unreasonable to ask of it? So then there were reactions to it, which I which I enjoyed more. Steven Sanofsky came out of it and said that the problem here is the anthropomorphization. Yeah. And that and that this is the. The argument I have about things like Sam Alton's sophomore essay is that it it presents a view of this stuff that's just not accurate and it and it's not fair to the technology and what it can do. This is where we get ourselves in trouble.

2:08:19 - Leo Laporte
The person who probably did the most work was the intern who is listed. So good job, parshan Shojay, because I think you probably did all the work. This was the first name in the authors, and and he's an intern. Um, yeah, I think anthropomorphizing is definitely. You know the flaw.

2:08:44 - Benito Gonzalez
That's surely the problem that's happening in the general public.

2:08:47 - Leo Laporte
Yeah, and certainly Sam Altman has encouraged that. No question about it Now.

2:08:54 - Paris Martineau
Futurism says Big beautiful brain.

2:08:56 - Leo Laporte
Apple researchers just released a damning paper that pours cold water on the entire AI industry. I think that's going a little far. And then it has a robot with a dunce cap.

2:09:11 - Paris Martineau
Just to okay, I do like that, drive it home.

2:09:13 - Leo Laporte
That's good, that's a good, it's a good art yeah, it went to getty to buy it, but anyway they got I feel I wonder if they ai generated that. To be honest, yeah, it kind of looks a little like that. Um, yeah, I mean, yeah, I don't feel like it's a great insight, but it's true and it's good to know and it got a lot of response yeah yeah, um, let's see, could ai make a scorsese movie? No only scorsese can make a scorsese well.

2:09:43 - Benito Gonzalez
That's so dumb. Only Scorsese can make a Scorsese movie.

2:09:45 - Paris Martineau
Yeah, I was going to say because AI is not Scorsese, so by definition it would not be a Scorsese movie.

2:09:53 - Leo Laporte
So Demis Asibis and Darren Aronofsky, the director, join hands to make a movie. They met in 1999, uh, when they were to give a speech on the future of storytelling. Nasibius was a video game developer. He's very smart, uh, gaming. He's a chess master, um, he's the ceo of deep mind that does not surprise me yeah, yeah, he was a very good chess player.

Uh, aronofsky, of course, created pie, black swan, the whale, um, and now they're going to do together short narrative films using ai. Three are already in the works. I don't think that's a scorsese movie.

2:10:35 - Jeff Jarvis
It's too short what really strikes me is last week there was a story about advertising. Meta was throwing all of everything in advertising, creative and media into AI and in the meantime the head of WPP quit, stepped down. It wasn't because of that alone, but advertising is freaking. It is not difficult to imagine this stuff making a 30 second creative on a commercial. That will be amazing.

2:11:00 - Leo Laporte
They already have. Last Christmas, Coca-ola had an excellent ad that was ai fully ai generated because it I mean you had to be. It was polar bears and right, right and it didn't they even try to hide that. I mean it looked like it was done in ai.

2:11:14 - Jeff Jarvis
Yeah and so I think we're gonna find this guy and and the art, so to speak, the technology has advanced immensely since then. Right, what we saw out of IO was amazing, so that's where the frontier is gonna be, I think, and then advertising does lead the aesthetic culture in that way.

2:11:32 - Leo Laporte
As does pornography, and we're seeing quite a bit of AI pornography as well, Well, I'm not Leo. Maybe you are, but I'm not.

2:11:37 - Jeff Jarvis
No, I don't know where Jeff has never seen a pornography, ai or otherwise in his life.

2:11:42 - Leo Laporte
Don't know where you go to get it. To be honest, I've tried. Actually, that raises an interesting question. Oh, if no human is involved, it is an interesting question. It's just created in the mind of a computer in the mind of a computer.

2:12:08 - Paris Martineau
Is it exploitation? Oh well, it depends what you're talking about. I would argue pornography generally as an industry, if it's consenting adults, would not be considered exploitation. I think if you're talking about child sexual abuse material, it's still exploitative and probably not something that we should condone as a society to be created in any form, whether it be from a computer. That's a good answer.

2:12:23 - Leo Laporte
Otherwise, yeah, and actually truth, even truth with adults, the the. The problem with pornography is it dehumanizes sex anyway, and so now you've even more dehumanized it, uh, and it's kind of it's the wrong take on uh sex, right it's uh, and so it's kind of training people sex.

2:12:47 - Jeff Jarvis
Right, it's uh, and so it's kind of training people to think about sex. There's debate about that. I mean, that's that's just like. Does video game violence turn us all into murderers? No right. So I think there's a, there's a presumed cause and effect there. But leo, it doesn't matter, because in a deterministic world. You're already going to do what you're going to do, so it really doesn't matter that, that's true.

2:13:03 - Paris Martineau
We are just one big computer brain, created by rats.

2:13:07 - Leo Laporte
I'm not saying we're not stochastic, we are stochastic.

2:13:11 - Paris Martineau
I'm a parrot personally.

2:13:13 - Leo Laporte
By which I mean you don't know what the next step is going to be in the line Because it's too complex. It's complex and it's probabilistic and maybe there are multiple steps and it's an even choice. I'm not saying you know what's going to happen next.

2:13:25 - Jeff Jarvis
I'm just saying what happens next comes out of what happened all the way up but it's like my friend david weinberger said an accident is not an accident, it's something you couldn't explain true like magic I asked chat gpt to make us into star wars characters and I don't understand the references, but the image is pretty good.

2:13:44 - Paris Martineau
I put it in the chat.

2:13:45 - Leo Laporte
I recently upgraded. Is it in the Discord?

2:13:48 - Paris Martineau
I added you in the Discord, it's kind of far off now.

2:13:50 - Leo Laporte
These are not. Is this what you gave us?

2:13:53 - Paris Martineau
No, I did not. Let me see if I can at you again.

2:14:00 - Jeff Jarvis
Pretty flawed for this guy, the diversion.

2:14:08 - Leo Laporte
Yeah, he's very good, by the way, with this, though, oh I like it.

2:14:10 - Paris Martineau
Right, I like it, yeah. So, um, I recently upgraded chat gbt to try it out in the pro version and I've honestly been finding it did you spell intelligent with two g's? No, I did not, I just took a screenshot it did. I took a screenshot of of the show and it had it spelled correctly in there.

2:14:25 - Leo Laporte
I love making images with chat GPT. I gave it. I know I'm probably doing harm to the environment in the world. I told you about.

2:14:37 - Jeff Jarvis
It's only a third of a teaspoon. Oh, that's right.

2:14:39 - Leo Laporte
It's a 15th of a teaspoon, no big. I told you about the um piano thing I did. Let me see if I can find the one I did. This was yesterday.

2:14:51 - Jeff Jarvis
Yeah, you should also go up to. I took a pretty five, four, six.

2:14:55 - Leo Laporte
This is from a picture of a protest that we went to and I thought, yeah, I said, make that a cartoon. And then I said, and add a biplane. Uh, so that's invented. Uh, yeah, there's a few things don't make a lot of sense, but it's, you know, like defend oligarchs, grifters elon, that guy's definitely angry. That's coming out of the top like just gop. Yeah, I don't know what the gop is either. None of those signs were in the actual picture. I sent them, but uh but they're all.

The people look unhappy in the front line the people are angry and they're all wearing sunglasses for some reason I like the guy who's just wearing a shirt that just says go go does it say? Gop or yeah, I think they did something with GOP. I think this is why I kind of find AI fun. You never know what you're going to get. I took a picture out my kitchen window and I said make this a country scene in a cartoon. And it's exactly what the picture contained.

2:16:00 - Jeff Jarvis
Belly and Cat is going to figure out where you live based on that.

2:16:05 - Leo Laporte
Yeah, we might want to not include that. I bet that would be an interesting challenge.

2:16:11 - Paris Martineau
Don't challenge the internet, Leo. They 1000% will know where you live. If you both show them that image and give them that challenge, you'll be doxxed immediately. Yep.

2:16:20 - Leo Laporte
I don't think so.

2:16:22 - Paris Martineau
Don't play this game with them. Think so.

2:16:23 - Leo Laporte
Don't play this game with them, leo don't play this game anyway, so you bought. You're spending 20 bucks on the chat gpt I spent it because I wanted to. I don't know I felt like it's your job.

2:16:36 - Paris Martineau
It's my job. I also just had some problems with using it the free version that. I wonder if it's going to be any different. I will say so far some limited success. I was trying to. I've been reading a lot of nonfiction books lately on various topics relating to a job I'm interviewing for, of new books that I hadn't already found from like the past three years and related to these one or two very specific things, and so I paid for it and asked, like whatever the deep research 4.5 for recommendations and honestly, it recommended like the perfect book, um, from within the past, like six months that's association.

It can do yeah, because I was like I there should be a tool for me to easily check, like all the boxes of categories of type of book plus the release year, and then go through it myself on some website but, like Goodreads doesn't have that, amazon doesn't have that, I don't know where to do so I was like I'll ask chat GPT, and it was quite useful by the way, I want to scream about Google for a minute, if I may, just because I need to yes, we can balance it out with.

2:17:41 - Jeff Jarvis
I got a bill. I just got my bill for the frigging, frigging, frigging what do you call it? Work group thing 62 bucks a month. Whoa, yeah, whoa, because of AI. They said the excuse was because of AI. Well, I can't get to your AI thing. You just released to everybody else in the world for free and you're charging me for your freaking ai google. What have you done? I ask you to bring this up with the chrome. What have you done?

2:18:12 - Leo Laporte
when you meet them I'll bring it with everybody speaking of google, youtube has loosened the rules guiding moderation of videos. Hey content moderators we favor freedom of expression over the risk of harm, so go ahead, promote ivermectin. Who needs vaccines masks? They don't work as long as the videos are considered to be in the public interest.

2:18:39 - Benito Gonzalez
I mean, I have noticed a lot of youtubers self uh censoring themselves certain words because oh interesting because youtube pushes them down, like if you say suicide or if you say kill or anything related to like that kind of stuff. They push the algorithm, put algorithm.

2:18:53 - YouTube
Thanks for screwing up our yeah, well, we got, we got subs great, now nobody will see the show.

2:19:04 - Leo Laporte
All right, we're going to take a little break. Uh, you're watching intelligent machines, many of you watching live. I love it. Thank you to the folks, of course, in our club twitter discord who are watching, uh, in the discord itself also the general unwashed masses watching on youtube and tiktok and xcom and Facebook and LinkedIn.

2:19:22 - Jeff Jarvis
My LinkedIn too.

2:19:24 - Leo Laporte
Huh.

2:19:24 - Paris Martineau
You can live stream on LinkedIn.

2:19:27 - Leo Laporte
Live stream on LinkedIn yeah, hello, gizmo Gizmo are you a generally unwashed mass?

2:19:33 - Paris Martineau
Yes, you are.

2:19:34 - Jeff Jarvis
Gizmo looks like a Star Wars cat, I think.

2:19:38 - Paris Martineau
Gizmo, did you hear that You're a Star Wars cat who wants to explain? To Paris why she was wearing that big cat headdress someone someone sent a gif that I guess kind of says like some person has got two lightsabers. I don't know, I don't come at me in the comments for my lack of knowledge about star wars.

2:19:55 - Leo Laporte
I think that's refreshing that's what I'm saying. I'm just a refreshing person, by the way we are going to launch the Leo Laporte OnlyFans site any day now. When I do porn, it's not dehumanizing. I just want to say OnlyFans isn't just for porn.

2:20:20 - Benito Gonzalez
You know there are celebrities using that to monetize a lot of stuff.

2:20:23 - Leo Laporte
No, in fact, I think it's right. I think that's the really interesting shift Guys. We should be streaming live on OnlyFans, right now, yeah, I feel like there is the assumption that if you're on OnlyFans, it is kind of adult, right?

2:20:36 - Jeff Jarvis
I'm going to unbutton my button now.

2:20:38 - Leo Laporte
Okay. I was going to say. This is why Jeff isn't wearing a long-sleeve shirt today. He's showing out a little how exciting Our show today and they are so sorry Our show today, brought to you by no, I love this company, spaceship.

They are.

This is so cool. They are the modern way to get your domain name and to put your site up on the web. It takes the pain out of choosing, purchasing and managing domain names and web products, and they have them all shared hosting, virtual machines, business email and I love this below market prices for domain registrations and renewals. They also have some pretty fresh ways to deliver simplicity. I like Unbox. You can connect your Spaceship products to your domain and configuring it all in just a few steps.

I wish my daughter had this. She was going I have my domain name and I've got my Shopify, and how do I connect them? And, oh my gosh, with Unbox, it's a click. In fact, I'm using Spaceship's new encrypted messaging platform, thunderbolt, which is tied to my domain name. I registered a domain name for four bucks leosim at Spaceship and then I pressed a button in Unbox and connected it to my messaging and that's all I need. Paris, you should look into this. You don't have to give them your phone number, you don't have to give them a username, just a domain that you control.

Very cool, and unbox makes it super easy. And if, for instance, you're like my daughter and saying help me, dad. I could just say go use alf. Alf is their very own spaceship's very own ai assistant. Alf likes doing the stuff you don't like, like domain transfers and updating DNS records. The other thing I love about Spaceship they're very open to your ideas. They love their customers and they have a place where you can go to vote on new features and products. They call it roadmap. You can explore their roadmap for the future, make suggestions, vote on new features. In fact, that's how Thunderbolt happened the messaging came out of customers asking them for it. So customers in the tech community get what they really need. Visit spaceshipcom slash twit to discover exclusive deals on domains and more Spaceshipcom slash twit. Thank you, spaceship.

2:23:02 - Jeff Jarvis
Spaceshipcom slash twit I am doesn't stand for instant messaging, leo, it stands for what intelligent machines?

2:23:19 - Leo Laporte
yeah, that's right, leo's I am. I didn't even think of that machines. Yeah, that's right, leo's I am. I didn't even think of that, leo apostrophe s I am. I think the im tld probably stands for a country. Let me see, here it's the isle of stands, for it's mine it's the isle of my leo.

2:23:37 - Paris Martineau
Therefore, on the isle of man. That's a pretty, that's a metal, uh tld that's a metal tld.

2:23:44 - Leo Laporte
Thank you very much. Dustin says it took him 30 seconds to get up and running on thunderbolt and the thing is it's completely secure because you, it's only a domain that you control that can then be used as your, as your messaging thing, and that domain can be anonymous. It doesn't have to say your name or anything it can. It can be anonymous, like it could have scoops for paris dot I am or something like that. Uh, let's see what else. What do we got?

2:24:19 - Paris Martineau
uh I I know we've not daddy zuck is paying seven to nine figures for new AI researchers to try and join Meta.

2:24:29 - Leo Laporte
Really Like a bonus.

2:24:31 - Paris Martineau
No, he is personally trying to recruit and steal AI researchers and executives from other companies. And they're so desperate that they're paying jensen wong did uh, anywhere from you know seven figures to nine figures of uh well, seven figures is in the millions, nine figures is in the hundred hundreds millions and that's just cold hard cash. If I recall correctly, that's just like that and I blew it.

2:25:01 - Leo Laporte
In college, I studied chinese. What was I thinking? You're a fool I should have known I'd be a podcaster instead of a multi-millionaire ai researcher hundreds of millions wow, there's a new york times article about it.

2:25:18 - Paris Martineau
This is for their super intelligence lab yeah, I think this is as news has been coming out that they are, you know, falling behind competitors.

2:25:28 - Leo Laporte
But isn't it interesting? Because they own more H100 GPUs by like three times as many. Now what are we going to do with them? Yeah, they just bought them and now we'll figure it out.

2:25:40 - Paris Martineau
Seven to nine figure compensation packages figure it out seven to nine figure compensation packages. Meta has recently grappled with internal management, struggles over ai, as well as employee churn and several product releases that fell flat. So mark zuckerberg is personally, uh, joining the fray and like trying to personally recruit people I guess by throwing them a lot of money.

2:26:02 - Leo Laporte
As much as I hate meta and I will give you a reason to hate meta too in just a second uh, I love it that they at least make llama open weights. Yeah, I mean, why not they're?

2:26:14 - Jeff Jarvis
you know, we're number two and that's, and that's, that's uh, yann lecun. Oh, is it? You think it's?

2:26:20 - Leo Laporte
his. He was the one who really he believes in that good. All right, well, that's, that's good.

2:26:25 - Jeff Jarvis
I'm glad to hear it's also kind of an android strategy too we're not going to be open ai, but we can yeah?

2:26:32 - Paris Martineau
yeah this, uh meta team zuckerberg. This is from bloomberg. Zuckerberg aims to hire around 50 people from the new team, including a new head of ai research, almost all of whom he's recruiting personally. He's rearranged desks at the company's Menlo Park headquarters so the new staff will sit near him. The people said that's crazy.

2:26:53 - Leo Laporte
Sit near me and make $100 million, now how much would you pay? So meta uh got caught big time breaking android's privacy. Meta and yandex, the russian search engine, put code in their software to bypass uh all the cookie blocking, all the third-party tracking, all of the ways that you tried to say don't, don't spy on me, bro. This uh research uh, and ours technically has a story. Uh refers to Android users and their web browsers, but there's reason to believe this would be for every mobile and even maybe desktop devices. If you're interested in the nitty gritty, steve Gibson covered it yesterday on Security.

Now the covert tracking implemented in the Metapixel and the Yandex Metrica trackers allowed and I put this past tense because as soon as this research came out, they turned it off immediately, like that minute. So that means they've got some other way of doing it. It allows Meta and Yandex to bypass core security and privacy protections provided both by the Android operating system and the browsers that run on it. Defenses such as state partitioning and storage partitioning don't work. They opened up ports on your machine, established local host connections so your machines could exfiltrate information.

It was really kind of awful what they did. It is fairly technical. Here's the graphic that explains it, but it was a very sophisticated technique. There was no. There would be no other reason to do this than to bypass your privacy protections and your security protections. There's no other reason to do this than to bypass your privacy protections and your security protections. There's no technical reason to do this otherwise. And the minute the research came out, both companies immediately turned it off without any response. Meta's response did not answer to ours emailed questions, but provided the following statement we're in discussions with google to address a potential miscommunication regarding the application of their policies. Upon becoming aware of the concerns, we decided to pause the feature, while we do never work with google to resolve the issue, yandex said oh, we're just continuing the practice. In Russian of course Yandex strictly complies with data protection standards. It does not. The amount of my user data?

2:29:55 - Paris Martineau
uh, unbelievable, unbelievable, yeah unbelievable, but also kind of believe but totally believable.

2:30:02 - Leo Laporte
So, on the one hand, while I appraise meta for having open weights on llama, they're not a good, nice company. They're not nice people that llama will spit on your face they are not nice people and uh, okay, so they're trying to go for super intelligence. I I hope they don't get it, because what are they going to do? Use it against us. This is the problem. By the way, I'm not worried about the ai, I'm worried about the people who make this.

2:30:33 - Jeff Jarvis
Well, that's what I've been. That's right.

2:30:35 - Leo Laporte
I wrote a book about that exactly you're exactly right yep amazon is testing humanoid robots to deliver packages do we need the robots to be humanoid?

2:30:48 - Paris Martineau
is that? Is that a useful component of it?

2:30:51 - Leo Laporte
well, apparently, if you're working in a factory designed for humans, yeah right I guess wait, but why?

2:30:57 - Paris Martineau
if the robots are delivering packages, why are they at the factory? Because that's not where you deliver, or the warehouse?

2:31:02 - Leo Laporte
it's not where you deliver packages they have an indoor obstacle course used for testing at an Amazon office in San Francisco. The report says it's about the size of a coffee shop. They hope this is what they want to do. They want to. You know, they have those Rivian electric vans. You may not see them in New York, but we see them in town here. They drive around. They've got a real. They'll still have a driver, but when the driver pulls up, the robot will hop out.

2:31:27 - Jeff Jarvis
Well, they were going to do that with with drones. We're going to do that out of a single van. Now they're going to do the robots instead even with a human driver behind the wheel.

2:31:34 - Leo Laporte
A robot could theoretically speed up drop-off times by visiting one address while the human employee delivers to another.

2:31:41 - Paris Martineau
Oh, I can see people tripping the robots someone's going to shoot that robot with a gun, unfortunately I did see and I don't know if it's true are the robots good with fending off dog attacks?

2:31:55 - Leo Laporte
that's a robot? How about robot dogs or porch?

2:31:59 - Benito Gonzalez
pirates like. They don't need to like right outside, they just need to mug the robot I did see this and I again.

2:32:05 - Leo Laporte
I'd love to see if I could validate it, but according to this video, 75 of amazon orders in the warehouse are now fulfilled by robots.

2:32:16 - Paris Martineau
Here's what it looks like that's been the case for a while. This has been. This was a major shift.

2:32:22 - Leo Laporte
They've rolled out, I feel like, over the last five so the the pickers are no longer humans, which is probably a good thing yeah, and so you see these big racks that you're seeing behind here.

2:32:32 - Paris Martineau
This was a major pivot, that amazon that's right.

This was over the past like seven years or so, just because it uh for a while. A lot of the complaints about amazon worker treatment in particular were that the warehouses were giant and gargantuan so in order to get something you'd have to have a person walk all the way over that. So they invested I believe, actually bought the company that makes those little floor robots. So how those big stacks move is it's like a huge stack on top of. You can see in the bottom of the video, little robots scooting along the floor. The bottom of the video, little robots scooting along the floor. Um, those kind of all move in tandem and will. Whenever a picker is standing basically at a station, whenever the next thing comes up, the robot will move that big stack over to them and say, oh, you need to grab what's ever in quadrant 3c.

2:33:15 - Leo Laporte
So you go over three down and like down three well, I want to meet the robot from amazon that when I ordered this, was at an accententure talk, actually for a Fortune talk, so I guess it's true. I want to meet the robot that, when I ordered a synthesizer from Amazon, instead delivered a succulent planter.

2:33:39 - Paris Martineau
So probably what it was is the synthesizer was in one of those cubbies and next to it was a succulent, and either they had the sort of thing that had a robot grab it or just the person grabbed the wrong one. I'm still in a lot of like Facebook groups and other employee groups for Amazon warehouse workers Sometimes it'll show up on my.

Facebook feed. They're just kind of like social media groups. Sometimes it'll show up on my Facebook feed they're just kind of like social media groups and it'll always be things like whenever, I guess, people order sex toys from Amazon and don't realize that that means a human being then has to look at a picture on a screen of your strange dragon dildo and then pick it from a cubby and put it in your box. But there's a lot of, so I assume Over and over again.

2:34:22 - Leo Laporte
After a while it's just old hat.

2:34:25 - Paris Martineau
After a while they all all the dildos blend together.

2:34:28 - Leo Laporte
They all blur together exactly that's hysterical.

2:34:31 - Jeff Jarvis
Dildo on aisle six.

2:34:34 - Paris Martineau
Sorry, should I grab the succulent?

2:34:36 - Leo Laporte
or the succulent. I just thought it was funny that I I mean it's, it was this big heavy metal, succulent.

2:34:44 - Paris Martineau
How small was the box that you got? It was huge, oh whoa.

2:34:48 - Leo Laporte
And heavy.

2:34:49 - Benito Gonzalez
Oh, is that when you ordered the cork?

2:34:51 - Leo Laporte
So what.

No, I ordered it was a little one of those little Akai things. I wanted something that would be more portable, that I could practice on and carry around. It was only 150 bucks. But I also ordered a baking steel which was big and heavy, and they said they both came in the same box. Neither one came, just a succulent planter. I'm trying to convince Amazon. I returned the succulent planter. They said, okay, well, we'll credit you for the steel, we won't credit you for the synthesizer. No, it was the same box but it didn't get anyway. I don't know, probably a robot did it google.

2:35:30 - Jeff Jarvis
Google has a problem. You wouldn't expect. What's google's problem? Foxes like what foxes on the roof of their london new london headquarters. Oh, l, london has a huge fox problem.

2:35:44 - Paris Martineau
Foxes are everywhere in that town. What.

2:35:47 - Jeff Jarvis
This is something.

2:35:48 - Paris Martineau
I've learned when I've had coworkers in London they'll be out in the garden taking a Zoom call and be like. The foxes have arrived. I must move inside. Are they dangerous? I don't know, but foxes are like London's rats.

2:36:01 - Jeff Jarvis
They like to hunt them there. I guess they're just not very good at it. Oh, that's what happened.

2:36:04 - Leo Laporte
They chased them into the city is what happened?

2:36:07 - Jeff Jarvis
All those fox hunters jumping over the hedges, they just chased them into town, but I want to know how did the foxes get on the roof of Google's new building, line 130.

2:36:21 - Leo Laporte
Does it answer the question? No, I don't know line 130.

2:36:24 - Jeff Jarvis
does it answer the question or is it no? I don't know this is a king's cross.

2:36:25 - Paris Martineau
I can't figure it out. No, it's king's cross. It's right next to the garden. They're everywhere. They could probably just climb up foxes have begun to dig burrows.

2:36:31 - Leo Laporte
Oh, it's because they have a rooftop garden. But how do they get up there? So it's they say this is nice, this is nice, we take the elevator up, it's nice well, you can see the trees, they climb the trees and then they leap from the trees to the headquarters. Those are beautiful foxes, by the way.

2:36:48 - Paris Martineau
They're gorgeous I might be spreading misinformation, but I have a memory of like a big thing in london having to be. They had to design trash bins that were like fox proof and it was like a difficult, like foxes were like a huge problem.

2:37:00 - Leo Laporte
The oldest cities in the world and it's so you could see why foxes might love this bill. Oh yeah, look at that. That's. That's like. What a trail for foxes, with a view and everything. Uh, how did they get up there?

2:37:12 - Jeff Jarvis
I don't know. That's what.

2:37:13 - Paris Martineau
That's what that's what I want to know. They walked up.

2:37:21 - Leo Laporte
That's right, iine gear. The rooftop garden has 40 000 tons of soil and 250 trees. They, uh, they think that the, uh, the foxes might be living off rats. So which would you rather have? The?

2:37:38 - Jeff Jarvis
foxes or the rats exactly I would prefer the foxes. Yes foxes are cute until there's a rabid rat in all this do you know that it's called a skulk of foxes?

2:37:50 - Benito Gonzalez
I think foxes are kind of dangerous, though, right, aren't they dangerous like wild if they're?

2:37:54 - Jeff Jarvis
rabid? I don't know there's there you would. This is wrong. You should not make a fox a pet, but but for people who do, there's videos of them. They like giggle when you scratch them foxes make the weirdest sounds they don't make sounds yeah, yeah, there was that song, what's?

2:38:09 - Paris Martineau
can you give me your impression of a fox I?

2:38:11 - Leo Laporte
wonder what the fox? What does the?

2:38:13 - Paris Martineau
fox say yeah I'll play you a fox sound. No, no from from your mouth, leo. I want you to, I want you to try and give us some foley work here I hear this at night in our neighborhood.

2:38:26 - Leo Laporte
I hear that that's a fox oh, you don't have a.

2:38:30 - Benito Gonzalez
We're not hearing it. You got to turn on original. Oh, I have to press the button.

2:38:33 - Jeff Jarvis
We didn't hear your piano before actually, oh, you didn't.

2:38:36 - Paris Martineau
No, we did not, I just heard my singing that's why I was singing because I was like I thought you were just doing a piano yeah, I thought you were mocking yourself no I oh, all right, all right.

2:38:48 - Leo Laporte
Now I have to play the piano for you guys again. If I had my 140 akai synthesizer, I wouldn't have to fake it. I'm gonna, I'm gonna where the foxes. Did you hear the foxes?

2:39:02 - Paris Martineau
I turned. No, it just sounded no that's not a fox, no this is a fox ready.

2:39:12 - Leo Laporte
Actually, that's the waves that's a fox, that sounds like a dog being attacked I know that's their normal sound. They sound like somebody's dying.

2:39:23 - Jeff Jarvis
This is better. I'm putting this in the chat right now. This is what. This is what I'm talking about. Hold on, hold on. Here we go better than that. These are these are like oh, yeah, yeah, yeah, pet foxes, giggling fox yes it's a oh god they're cute little suckers what have we done?

2:39:38 - Leo Laporte
oh, wait a minute, I still have the other fox. This is 11 labs you just sent me no, I didn't.

2:39:43 - Jeff Jarvis
I sent you giggling foxes. Well, how the hell did that happen?

2:39:48 - Leo Laporte
no, no, there it is, yeah, okay hey man, I got the other foxes open, I gotta close the other.

2:39:55 - YouTube
Okay, giggling foxes get ready if you need a laugh boy, do we have one for you, if you need a laugh boy?

2:40:05 - Leo Laporte
do we have one for you? The nine news is going to take us down. We have the copyright on the foxes.

2:40:13 - Paris Martineau
Just 45 seconds of fox sounds from various devices.

2:40:17 - Leo Laporte
Pretty much what's happening.

2:40:19 - Jeff Jarvis
And I love it on the page. Related topics Female fox laughing. Baby fox laughing. Fox laughing gif. It's like fox porn. Let's see the laughing.

2:40:29 - Paris Martineau
Well now we could put this on only fox.

2:40:33 - Leo Laporte
All right, I am going to play for you the actual song. Now you can hear it. This is cockles and muscles. No, that's foxes. Can you hear that? Yeah, no, that's foxes. Can you hear that? Yeah, yes, that's me, this is good. How much are you paying for the lessons? A lot.

2:41:03 - Paris Martineau
Enough that they gave him a bust.

2:41:05 - Leo Laporte
You want to hear the one I'm learning right now. It's fun, right, little bluesy. Yeah, it's fun, I enjoy it. It's uh, it's a past time. I'm never gonna, you know, be on the stage, but uh, you know you want us to disagree with you. No, I know the truth. I'm not. I'm not a fantasist. I'm not sam altman. I don't think ai is going to suddenly teach me how to play piano perfectly. Uh, you are watching intelligent machines. If you are watching, uh, not in the club right now I want to urge you shame on you, sitter you?

2:41:51 - Paris Martineau
what are you doing?

2:41:53 - Leo Laporte
now. Here's the reason the club exists. It's there. There was a reason. Four years ago, you might remember, there was this thing called covet. Advertising kind of dropped off and we thought we are in trouble, we need to survive. We did, we survived.

Here's a picture of me playing the piano. I gotta put that in my uh, in my up city. Uh, the way we survived was we went to you, our audience, and we said can you help us out? We passed the hat, but it's become more than that. Four years later, we're thrilled. Five years, is it five years? Wow, well, no, five years for COVID, but four years for the club. It has taken off beyond our wildest dreams, which is fantastic, because now you're covering 25% of our operating costs, and that's good, because advertising is only covering 75%. So because of you and your support in the club, we can now continue to do our mission and make shows, and make new shows, and we would really like you.

This is, by the the way, an example of the kind of stuff you get in club twit. This is in the club twit discord. What do you get for 10 bucks a month? Ad-free versions of every show we do, including special shows only in the club. All the keynotes, for instance from last week, the microsoft build, the google io, the this week's wwdc keynote all of those were in the club only and you can still get them on the TwitPlus feed.

Stacy's Book Club, chris Marquardt's coming up on Friday. We're going to do our photo show. We do that every month and more and more, lots of stuff like that. The Untitled Linux Show, home Theater Geeks we do it in the club and that's kind of our thank you, our way of saying thank you for supporting what we do. We really appreciate it. If you're not a club member, I would love you to be in the club so we can chat and discord and you know you get all those benefits. Go to twittv, slash club twit, and I thank you so much, all the club members, uh, for your support. It's um, it's really been amazing. We appreciate it. I'm looking so you notice I got a new clock, yeah I did.

Yeah, I'm really happy about this thing. This is uh from a company called uh mit zella m-i-x, m-i-t-x, m-i-t-x-e-l-a. Dot com, I think. Um, I it's. He's actually a youtuber, um, m-i-t-x-e-l-a, yeah, midzilla. So he, he does projects and uh kind of fun projects, the most recent being, you know, you see quite a few of them, but the most recent being this clock. It's the precision clock Mach 4. And I saw the YouTube video and I said this is the coolest thing ever. This uses GPS to set the time, so it is absolutely accurate, down to the millisecond. Uh, I didn't tell it where I was.

That's good, because you always start this show exactly people for some reason one o'clock, and I think it's their timing something or I don't know what it is anyway. So we get the date and time for this. It's kind of cool. It stretches out, so it's like a clapper, like on the on the movie oh, oh cool, oh oh, that's fun.

He did some really cool things. There's an ambient light sensor, the GPS. If it can't get a signal, as the time gets less accurate, will show fewer and fewer tenths or hundreds or thousandths of a second. But right now it's got a lock, so it's showing thousandths of a second. I thought it was really cool.

2:45:42 - Paris Martineau
Anyway, if you see this and want one m-i-t-x-e-l-acom but this website has some cool stuff, such as led industrial piercing earrings oh yeah, well, you should go out and get them, or maybe get this euro rack knob.

2:45:58 - Leo Laporte
I don't know what that's for me, that's for that's for everyone on your podcast yeah, it's crazy, isn't it? Motorized cam oh, that'd be good, I should get that. Oh, is this what you want to paris?

2:46:10 - Paris Martineau
yeah industrial light piercing that seems cool, uh, the world's smallest smitty synthesizer I like that. There's a uv protection amulet, a magical pendant pendant that warns you about ultraviolet radiation.

2:46:24 - Leo Laporte
Yeah, this guy's so cool. Anyway, he shipped it out right away. He's in the UK. I was very grateful.

2:46:29 - Jeff Jarvis
What was the tariff on that?

2:46:32 - Leo Laporte
There was no tariff. Oh, I don't know how he did that magically, anyway. So I don't have the clock above me Now, I just have like a. That's a sound, you know, it's hearing me talk, a little sound thing Spectrum analyzer.

Spectrum analyzer. Thank you, but I want to put, like club membership numbers in there or something like that. In fact, what I was really thinking is, when you join the club, you could send me a text string to put in there briefly, so every time we get a new member, we get a new like the brick, like the digital brick.

2:47:07 - Paris Martineau
Yeah, yeah, exactly somebody does like a super chat in one of the things that like pops up on there yeah, it's kind of a twitchy kind of thing to do so that's the plan we're thinking.

2:47:18 - Leo Laporte
We're thinking maybe doing that that'd be kind of fun that'd be fun gotta write some code. But you know I have a little ai friend who's pretty good at that. Uh, all right, you guys pick something. We've done a lot of these stories.

2:47:32 - Jeff Jarvis
Um 23 and me uh is being sued to not sell the genetic information as meanwhile the purchase price is going up, up, up a company called regeneron pardon me, and wajiski has stepped back in uh, she says 260, 256 million is not enough um, she's made a bit of she's bid 305 million.

2:48:01 - Leo Laporte
She wants to buy it wow, it dwarfs her previous offer to acquire the company for 40 million just ahead of the march bankruptcy file, jesus. So they're, they're, uh, uh, pausing the sale process. Regeneron, we thought was going to get it. This is a pharma, pharma, research uh company and they, I guess, wanted the uh, the data if they wanted. Still, they've got a million more wow, yeah, because the bankruptcy court will send it sell it to the highest bidder. The regeneron gets the last look losing bidder gets a 10 million dollar breakup fee.

2:48:38 - Paris Martineau
So maybe she's just going for the 10 million it's a little risky, but yeah, it's weird to um, it's interesting, I mean her involvement in this whole process has been very odd yeah this was, uh, also the case where basically the entire board resigned, if I'm recalling correctly due to her how she was handling the sale and uh potential bankruptcy process uh, okay, well, we'll watch with interest um I have already deleted my data three and me. Oh yeah, did you do that? My?

whole family did it, did you delete your data no you're just cool with whatever pharma company having it what are they gonna clone me? They can't, that's not enough what are they gonna do clone?

2:49:20 - Leo Laporte
me says man who has 17 clones.

2:49:22 - Benito Gonzalez
I'd be pleased they will charge more insurance to all of your descendants.

2:49:27 - Jeff Jarvis
No, yeah, no, so I had this discussion with um esther dyson once, when she did her, her, her genome and I said, esther, my, my son's, uh, the fact that I had prostate cancer and my, my descendants, she said, get over it, jeff, everybody gets it she very famously gave her uh entire genome to george church's project, the personal genome project.

2:49:54 - Leo Laporte
I tried to I uh and they were like we don't want this but then I later, george, then now has a uh, commercial version of that which I did sign up for and I got, you were like I want to give it to.

2:50:09 - Paris Martineau
Someone could make as much money as they want from my genome why not?

2:50:13 - Leo Laporte
yeah, no, the prices come down dramatically. I think. What is the personal genome project? When esther did, I think it was like ten thousand dollars, uh, it was around a thousand to do it, uh with, uh, what was the name of it? Uh, I can't remember now, george church sequencing would cost a thousand dollars yeah.

So what 23andme did for only a couple of hundred bucks was a partial sequence and then the results they, they. It wasn't. It wasn't what kind of. They implied, like you're getting your whole genome. But what Church is doing with Nebula Genomics that's the name of it is they scan the whole thing and you can download it and, by the way, it's a lot of data, it's gigabytes of data and then you can send it off to, to. There are a lot of companies around that. If you give them, uh, your, your genome will say, okay, you have markers for this cancer or this, uh, you know, health condition or brown hair or whatever. It is a lot of biome companies and stuff. Um, so this is I. Yeah, I did this one. Now I don't know what I'm gonna do with it, but it's more complete than what I got from 23andme. I interviewed george a couple of years ago on triangulation. That's what started the whole thing. Um, it's very it's kind of interesting.

uh, all right, I think we could go to the picks of the week, don't you do? I need need another break, benito, I haven't been keeping track. No, we're good, we're good. So, ladies and gentlemen, thank you for watching the show. And now, ladies and gentlemen, I give you Paris Martineau and her pick of the week. And if it's, Fox's.

2:51:56 - Paris Martineau
Well first, my pre-pick is last week I told people to leave a five-star review on our podcast. Oh, how's that going? Your favorite thing If you have a comment that would be fun to read in the show. And people did leave some good ones. I thought One of my favorites was by PJANSC four days ago. The review says got the bug. This is the only podcast bold enough to review the 1975 movie Bug. I saw this as a double feature with phase four.

Bug is only surpassed by the killer cockroach scene of 1977's Damnation Alley. Thank you, even though technically the movie I referred to was the 1978 movie the Nest, I did see a different cockroach movie called Bug this week and it was not as good as the Nest, but still a different cockroach movie called bug this week and it was not as good as the nest, but still a good roach bug, you made him an honest reviewer.

2:52:50 - Leo Laporte
We have now reviewed both nest and bug and uh.

2:52:54 - Paris Martineau
On a similar note, someone wrote entertaining and educational. Really enjoy the show. I learned a lot about ai and understandable level. Three hosts are smart witty as they banter over sometimes very silly topics. Love the pics the week? Well, maybe, except the roaches and rat movie series your opinion is noted, but I do not value it, to be honest.

2:53:15 - Leo Laporte
Those are my highlights from that uh, so more next week please yeah, I will yes, go to your favorite podcast client, review us and rate us. Give us a good one, give us five stars.

2:53:26 - Paris Martineau
We got to get our rating up. We've had some weird review bombing over the years. That's brought it down. This should be a closer to five stars podcast, so get out there. If they're funny, we'll read them on the show. Do it Do it. My pick of the week is I saw the Phoenician Scheme, which we had watched the trailer for on the show at some point. I saw that last night Good movie Would recommend it if you're a Wes Anderson fan.

2:53:47 - Leo Laporte
It was fun oh that's the new Wes Anderson. Yeah.

2:53:50 - Paris Martineau
I would just say it was like fun, it was delightful. It was just a very funny movie actually, which I think is fun. Are you a Wes Anderson like? Whatever he does, I love it. Fan or well?

2:54:06 - Leo Laporte
no, I watch everything he does, but I don't always like I hated uh the french dispatch for to be. Yeah, I thought this was gonna be a lot like that one.

2:54:10 - Benito Gonzalez
This one looks like a looney tunes this is like no, this one is.

2:54:13 - Paris Martineau
It's like it's a. It's a um globe trotting quad, I guess. Criminal business mastermind who keeps getting people keep trying to assassinate him, and so he keeps getting caught in near death experiences like plane crashes or being shot in the chest or things like that. Meanwhile, he's trying to connect with his daughter, liesel, who is kind of slapstick. A little bit slapstick in a Wes Anderson way.

2:54:45 - Benito Gonzalez
Michael Cera is in it as well and really does a great job yeah that's why I say Looney Tunes like this is kind of a Looney Tunes kind of movie, like the comedy is kind of Looney Tunes-ish.

2:54:52 - Leo Laporte
I love the Royal Tenenbaums. I love Fantastic Mr.

2:54:56 - Paris Martineau
Fox. So the movie also. It kind of explores a relationship betweena absent father and his daughter and I think the themes, if we're extrapolating, are like about faith and legacy, in the sense that it did remind me tonally of um, uh, what movie did you just say?

2:55:13 - Benito Gonzalez
uh, tenenbaum they're all right. I love this in the sense that it's a patriarchal relationship.

2:55:19 - Paris Martineau
Yeah, um, okay, my other pick is I just remember this, when someone uh mentioned tick tock earlier I saw this really cool tick tock of a woman who's uh somehow used code to make like her hands in videos control the vocal production and like production of songs and it's all these cool videos of like if you play it, like her doing different hand gestures, like creates whatever is go, like it controls like a midi board in a really.

2:55:48 - Leo Laporte
She's got a python program called midi cam. That is a camera watching what she's doing here. Let me play this and it's her own songs.

2:55:57 - Paris Martineau
We shouldn't do this instead of the piano under my glasses and like each hand gesture creates something else going on in the thing, and she's a bunch of other videos of how it works. I just thought it was very interesting and I don't know someone who's more technically inclined than me could probably explain how it's.

2:56:15 - Leo Laporte
That's pretty cool. It's pretty simple, it's very it's a little gimmicky.

2:56:17 - Benito Gonzalez
In my opinion, these can just be pedals yeah, and should just be buttons, yeah, yeah probably yeah, but it's neat, yeah, it's just reminding me of the show this week.

2:56:30 - Paris Martineau
Yeah technology.

2:56:32 - Jeff Jarvis
Jeff, what's your stuff? Well, I could do a rant about my favorite topic, the death of mass media, but I thought I would ask unless you talked about this last week, did you talk about mountain head?

2:56:43 - Leo Laporte
I don't know if we talked about on this show. Did we paris?

2:56:46 - Paris Martineau
I don't think we did I don't think so what did you think of it, jeff?

2:56:49 - Jeff Jarvis
um, I liked it a lot. Okay, uh, I think it was. I think it was. You know, it had its weak moments, uh, sitting there talking let's explain.

2:56:56 - Leo Laporte
This is people. This is a movie. Is it on netflix? I can't remember hbo hbo, now hbo max. That's not confusing enough. Uh, starring uh steve carell, written by the guy who did succession, jesse armstrong. It's uh, four of the richest men in the world. Well, three of the richest men in the world and one guy who only has a few hundred million. Uh in a uh in a uh. Where are they?

they're in um utah yeah, they're park city or near, likely place for multi-planers in a in a big house and the world is falling apart, thanks, literally thanks to the guy who owns tram t-r-a-a-m, who is the richest man in the world, worth several hundred billion dollars. He has created it's basically Facebook. He's created a AI image generator that's so realistic that people are believing it and it's causing huge religious strife, civil wars, killings. It's a mess. And they're up there in their mountain area watching the news and it gets kind of crazy. I it's funny. It stayed with me more than I thought it would. I thought it was very wordy, it was just a lot.

2:58:14 - Paris Martineau
You know, they shot it in a couple of months yeah, from what I haven't seen it because I feel like a lot of my film friends are like said it seemed like it was definitely a product of a incredibly tight production schedule, with kind of first thought, best thought, but I do agree that it seems like the sort of thing I would like there's a lot of, a lot of jargon in it.

2:58:36 - Jeff Jarvis
Uh, from the stuff we talk about, well, it makes me very happy because they skewer long-termism and transhumanism exactly.

2:58:42 - Leo Laporte
Yeah, you see all the stuff that we talk about um somebody in the uh chatroom, brandroid, says the primary flaw was that all the main characters are terrible, terrible people I would agree with that, except succession as well there were no heroes in succession and and and and. You still kind of you were, you still were kind of in there rooting for him. You're right, though there was nobody to root for in this.

2:59:04 - Jeff Jarvis
They were, they were but that was the whole point, yeah at.

2:59:07 - Leo Laporte
At one point they I don't know, should I have any spoilers? Maybe I'll leave the spoilers out it it goes gets very dark. Yeah, let's put it that way. Um, I I didn't like it after I finished it, uh, and then I've been thinking more about it. I think maybe I I liked it in an abstract way.

2:59:26 - Jeff Jarvis
I'm going to spoil one line because I just loved it so much screaming I take kant very seriously.

2:59:32 - Leo Laporte
Yes, steve carell, okay actually he plays, the older he's basically peter teal he play, or or maybe paul. Yeah, he's an older VC who's kind of funded. All of these guys, the other three guys, and he's the wise one, the former professor, you know, who apparently loves to quote Kant and Nietzsche. It's very. That part's very funny. It's very pretentious, very pompous. Yeah, uh, and they're constantly saying first principles, first principles. Uh, yeah, the more I think about it, the more I like it.

3:00:08 - Jeff Jarvis
Yeah, I'll be curious so, so, so even while that is made and even though, while you know, we have a little bit of kind of mass media left, there's just a bunch of things here. So warner brothers is going to split off warner brothers, discovery is going to split off its cable companies as Comcast, universal, nbc. It's kind of letting the lifeboats go. Cable has cooties as the ship sinks. It's cable cooties. Yeah, anything alive.

3:00:33 - Leo Laporte
They're keeping the movie studios, but they're getting rid of all the cable stuff. Right.

3:00:39 - Jeff Jarvis
And then, interesting, the Guardian says that user-generated material will eclipse the ad revenue attracted by professional media from TV networks, cinemas and news companies very soon.

3:00:56 - Leo Laporte
Any day now the creator apocalypse, yep, yep. Well, we're in that creator space kind of old school version of that creator space it. You know I I just saw a study that said podcasting is taking off wait for me did you see this flight for a long time, the I can't remember if it was Netflix, I think it was Netflix as a biography documentary of the Call Her Daddy, alex Cooper host. Oh wow yeah. Yeah, the big big. It's interesting how podcasting has changed since we started way back.

3:01:39 - Jeff Jarvis
Content creators are expected to see their revenue through ads, brand deals and sponsorships increase by 20% this year, according to wpp yep the problem?

3:01:48 - Leo Laporte
I'll be honest, the problem is it's not us, it's influencers.

3:01:51 - Paris Martineau
So there are all kinds of creators, but influence going to all the people on love island, usa, right now well, it's also.

3:01:58 - Leo Laporte
I mean, my son is benefiting from this. He's an influencer when's his sandwich shop opening. Saturday.

3:02:04 - Paris Martineau
It's the latest. Wait really, Leo, are you coming?

3:02:08 - Leo Laporte
No, I you know.

3:02:10 - Paris Martineau
Leo, you've got to give Jeff his switch.

3:02:13 - Leo Laporte
He doesn't. He doesn't oh, that's true, I do have the switch for you. He doesn't, I think, want me there. His mom talked him into letting him, letting her come out, uh, cause she said, well, somebody has to be there. But he's really. He doesn't want to focus on us, he wants to focus on his thing, what he's doing, and all that.

3:02:32 - Jeff Jarvis
So I'll go out next month. Yeah, that's what I'm saying, and we're going to do it.

3:02:36 - Leo Laporte
We're going to do a festival fun there, yeah, but I wanted to give him a chance that's right, salt hanks okay well, I'll be there on bleaker street next to john's uh in the beautiful west village. You could walk there. Practically you take the subway and walk the rest of the way well, I mean, you can do that all over new york yeah, it's uh it's

3:02:58 - Jeff Jarvis
amazing what we are I miss new york?

3:03:00 - Leo Laporte
yeah, he doesn't. I think it's normal. He doesn't want to have to worry about us, he just wants to have fun and pay attention to what he's doing. And opening a restaurant is hard enough.

3:03:13 - Paris Martineau
I think it could be hard to not focus on your family.

3:03:18 - Jeff Jarvis
Let's go to Salthankscom and at least give him a good plug. Grand opening Doors open May 31st, 280 Bleeker.

3:03:25 - Leo Laporte
Yeah, they still say may 31st. Yeah, because he had a soft opening last weekend but he decided not to open, uh, right away. Um, so, because they still had some stuff.

3:03:36 - YouTube
I just signed up good, drop your email, as is there a menu uh, yeah, see that sandwich.

3:03:44 - Leo Laporte
What should I get? Get that sandwich? I said would you name a sandwich after me? He said yeah, so I don't know, maybe there will be, maybe there won't be, I don't know if he's there.

3:03:54 - Paris Martineau
Should I tell him that I do a podcast with his dad?

3:03:57 - Leo Laporte
oh yeah, definitely, oh, no say or will he be embarrassed? Say no, no Say hey, it's Paris. I do intelligent machines formally this week in Google on Twitter.

3:04:07 - Benito Gonzalez
You've met him in Paris, haven't you met him?

3:04:09 - Leo Laporte
before. Yeah, she's met him. Yeah, he's been on the show.

3:04:15 - Benito Gonzalez
No he wasn't here. Oh, he was on Twitter right, he was on Twitter, I think, on this show, but when was like on vacation or something I haven't we're going to go in and tell them how hurt we are.

3:04:23 - Jeff Jarvis
They didn't come on our show mm-hmm hank hank.

3:04:31 - Leo Laporte
Well go, yeah, I, I'd be very curious if you have a sandwich. He's um. It's been fun. If you watch his vlog on instagram uh, salt underscore hank he's been showing how he chose the breads oh, he's got some good.

3:04:44 - Paris Martineau
He's got a great bread that he just cracked for that bread looks so good.

3:04:50 - Leo Laporte
It's one of the reasons I can't go you're gonna poison your dad. They just did a festival where they sold thousands of sandwiches at the look at the end, the thing is his oh, I did. I did hear about this festival, yeah, yes, chef, yeah, his sandwiches are juicy as heck, they're just messy. So that's why his cookbook's called a five napkin situation. Let me see, uh, if there's anything else so what would your sandwich be, leo? Well, I told him that I really like Rubens, so maybe he'll do a Ruben for me.

3:05:26 - Jeff Jarvis
I love Rubens.

3:05:29 - Paris Martineau
Oh, he's got an industrial onion slicer.

3:05:32 - Leo Laporte
Oh God, Does he have an industrial? Oh my God, Because a lot of what they do is pickles, right, Pickled stuff. So yeah, it's going to be.

3:05:43 - Paris Martineau
What's it like being? How is he handling restaurant ownership so far?

3:05:47 - Leo Laporte
it's. I think he's fairly stressed it's stressful he's a new bear

yeah it's uh, there's a lot to do. There's, this is the restaurant. Uh, this is why it's not open yet. They're still kind of working on it, although I think this was a little while ago. Um, he's gonna. So that's like the bar where you go up to buy your sandwiches and the kitchen's in the back, but he's going to be making sandwiches there in the window. So the people who are in line to John's pizzeria, we'll see him and one hopes, come in, don't I don't, don't name the zd sandwich after me, that's all I ask look at that, my dad the zd.

My dad loves his zd anyway, yeah, I'm very proud of him and he's so. All the ad deals are going that way. He's got deals. Wow, I mean the amount of I don't even like. It's unbelievable. Wow, yeah, all the stuff that we would love to get. It's going to happen. But I gotta say this show, thanks to the new focus, has, as you may have noticed, really picked up in both audience and ad revenue, so I'm very pleased with that. Yeah, um, we get a lot of the b2b stuff, uh, and for this show we get a lot of the ai companies. So I'm uh, I think we're gonna make it. Yay, and thanks to the club. I gotta tell you the club is amazing, uh really bless you all.

3:07:12 - Jeff Jarvis
Yeah, thank you all. We appreciate all your help ladies, let's all do a thanks for the club, leo thank you, club.

3:07:21 - Leo Laporte
Thank you, I won't play the piano ever again I'll only eat loud crunchy sandwiches into the mic oh man, so that's from frenchette. Mr met, who is obviously a new yorker, says uh, the frenchette is fantastic. We go there a lot for the food, as well as the bread.

3:07:41 - Jeff Jarvis
Yeah, I hope henry will become a regular stop for people I'll check it out and I'll bring my influencer sister I might get you know what next tuesday when I go in for um stop by uh the chromebook.

3:07:53 - Leo Laporte
Yeah, yeah, where's the chromebook?

3:07:54 - Paris Martineau
let me know when you're in, we can hit it up so google new york, is it?

3:07:58 - Leo Laporte
is it a big deal event?

3:08:00 - Jeff Jarvis
or is it just no, it's uh, no, no, it's not just me, but I had to sign up, god now I want a sandwich.

3:08:05 - Leo Laporte
I know I'm starving.

3:08:07 - Jeff Jarvis
I also. You have to come for the American Burger place. I'm telling you.

3:08:10 - Paris Martineau
Yeah, oh all right.

3:08:12 - Leo Laporte
All right, it's amazing and give us a report. It's Tuesday, so next week a report on the Chromebook event from Jeff Jarvis. That'll be good. Who is our guest next week?

3:08:24 - Jeff Jarvis
I should ask oh, this is my friend Matthew Kirshenbaum.

3:08:27 - Leo Laporte
Great Tell me about him.

3:08:28 - Jeff Jarvis
Well, he wrote that incredible book.

3:08:29 - Leo Laporte
Track Changes.

3:08:37 - Jeff Jarvis
Track Changes. He is writing a book about the textpocalypse. He's on the Modern Language Association Task Force for AI in the Classroom and just announced just announced he was just appointed to a he's at University of Maryland. Now Next January he will have a chair in AI and English at University of Virginia, nice, that's right. He's really brilliant. He is a friend and a mentor of me and and and really amazing he also runs a book lab where he teaches students how to make old-fashioned print.

3:09:13 - Leo Laporte
I love this picture of him with all these old computers, the old Macintosh, the Apple 2. Look at that, that's awesome.

3:09:22 - Jeff Jarvis
He sent me some topics I'll send to both of you that he can talk about, because I talked to him and he's a wonderful guy and brilliant and yeah, I think we'll have fun.

3:09:32 - Leo Laporte
Good, matthew Kushnerbaum Next week on Intelligent Machines. Don't you dare miss it. We do the show every Wednesday, 2 pm Pacific, 5 pm pacific, 5 pm eastern, 2100 utc. You can watch us live if you choose. Uh, as I mentioned, on all those platforms discord, youtube, twitch, tiktok, facebook linkedin, xcom and kick but the easiest way to watch is, after the fact, go to our website twittv slash. I am. There's a youtube. You'll find a link there to the youtube channel. A great's a great place to see the video, but also to clip the video If there's something in particular you saw you'd like to share. And, of course, subscribing in your favorite podcast player might be the best way. Leave a funny review Five stars, please, so that Paris can read it next week on the show. Thank you so much for being here, paris Martineau. I don't know what to say.

I'll be out, say reading the reviews, she's the greatest, she's a. She's an investigative reporter give her a scoop, let her, let her yell at the ceos in your life let me eat a large scoop of ice cream parismyc and, of course, jeff jarvis.

I'm not you can't get rid of the whole thing. The emeritus professor of journalistic innovation of the craig newmark graduate school of journalism and the author of the gutenberg parenthesis, now in soft cover what would google do long time ago? The web we weave, which is a really great read about how we got here and what we should be doing going forward, and, of course, magazine about his history and the history of the magazine to be an audiobook yes, with a special announcement by Paris at the beginning.

It's true, thank you guys, thank you everybody. We'll see you next time on Intelligent Machines. Bye-bye.