Intelligent Machines 847 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.

Leo Laporte [00:00:00]:
It's time for Intelligent Machines. Our guest, Imad Mostaq is the founder, the guy who created Stable Diffusion. His latest Intelligent Internet, is creating small AI models for use in medicine and elsewhere. Fascinating interview. Next on Intelligent Machines, podcasts you love from people you trust.

Paris Martineau [00:00:25]:
This is twit.

Leo Laporte [00:00:30]:
This is Intelligent Machines with Jeff Jarvis and Paris Martineau. Episode 847, recorded Wednesday, November 26, 2025. Caked up football, man. It's time for Intelligent Machines. Yes, the show where we cover AI, robotics, and of course, all the smart little doohickeys all around us. Getting smarter all the time. Paris Martineau is not here because this is a pre record for this interview. Jeff Jarvis is professor of Journalistic Innovation at the Craig Newmark Graduate School of Journalism.

Leo Laporte [00:01:01]:
Emeritus. I forgot the emeritus at the City University of New York. He's at Montclair State University in suny, Stony Brook, author of the Gutenberg Parenthesis and magazine. Oh, man, I got it all in without the jingle.

Jeff Jarvis [00:01:11]:
Yay.

Leo Laporte [00:01:13]:
That's because Benito is not pushing the buttons today. Anthony Nielsen filling in. We're doing it early, a pre record earlier this week. This is why Paris can't be with us. She's got a day job. Crazy.

Jeff Jarvis [00:01:30]:
Yeah.

Leo Laporte [00:01:30]:
I'm really excited about our guest this week, Ahmad Mostak. Ahmad, let me ask you, is it Mustaq?

Imad Mostaq [00:01:38]:
I think so.

Leo Laporte [00:01:39]:
Mostaq Mostak's fine.

Jeff Jarvis [00:01:41]:
Okay.

Paris Martineau [00:01:44]:
Whatever.

Leo Laporte [00:01:45]:
You may not know The Intelligent Internet II.inc, although that is the best domain TLD I've ever seen, but perhaps you will have heard of something called stability AI and stable diffusion. Ahmad, it's amazing that this is only three years ago that stable diffusion came out. It blew my mind, blew all of our minds. This is before ChatGPT3. It was before we really understood the revolution that was about to hit us. But we were excited about it. In fact, I'm going to show you our Christmas card. That year, we.

Leo Laporte [00:02:23]:
We went out in front of the building and took a picture and then used stable diffusion. Now, just to give you an idea of how far we've come, it probably could be one shotted now by a chatgpt, but just to give you an idea of how far we've come, our producer, the guy who did this, our AI guru, Anthony's on the line right now, made a little video of the process. He used Photoshop and stable diffusion to create a whole bunch of images. You said Anthony, 200 images. I don't recall. It was a lot. And you could see there, you know, back in the day, stable diffusion did its best, but eventually we came up with a Christmas card. After a little.

Leo Laporte [00:03:05]:
He would take little bits and Photoshop it in. He's gonna get. He's making a gingerbread house out of the building. He's got the candy cane and.

Jeff Jarvis [00:03:17]:
We'Ve.

Leo Laporte [00:03:17]:
Come a long way. And yet, Ahmad, this was mind blowing, this.

Jeff Jarvis [00:03:21]:
Mind blowing? Yeah.

Leo Laporte [00:03:23]:
You really launched something into the world that people, I think, immediately groked and got very excited about. We had no idea what we were going to see. Three years later, Ahmad's written a new book, the Last Economy, that talks about this revolution. That's really what I want. Not, I didn't want to go back to the past. I wanted to talk a little bit about what you see as the future. You talk about the four big economic inversions and this would be the fourth. You even say the fourth and final economic inversion, the pattern that breaks the world.

Leo Laporte [00:03:58]:
What are you talking about when you say the intelligence inversion?

Imad Mostaq [00:04:02]:
Yeah, I mean like you said, about 30 years and AI years has passed since that was released and now we're at this tipping point.

Leo Laporte [00:04:08]:
So is this 10 to 1? Okay, that's good to know. I'm going to keep track of it.

Imad Mostaq [00:04:12]:
It's about 10 to 1. It certainly aged me a lot.

Jeff Jarvis [00:04:16]:
You don't know age.

Leo Laporte [00:04:18]:
You're not alone. You know.

Imad Mostaq [00:04:20]:
Actually, you know, nanobananapro just came out today, which might be mind boggling.

Leo Laporte [00:04:25]:
It is mind boggling. But there's a critical difference between what Google is doing. And admittedly, Gemini 3 and Banana Nanobanana 3 Pro are amazing. We were playing with it yesterday and it generated an image like that. It's really fast, but it's different. In fact, the difference. Well, let me, let me give you a chance to finish the, the economic inversions. And then I want to talk about the word diffusion because that's critical to the whole thing.

Leo Laporte [00:04:53]:
So what is happening? What is this economic inversion?

Imad Mostaq [00:04:58]:
So through history, the whole of our economy shifted. So it used to be that you had some land and you farmed on it. Then we kind of moved into the factory world, this kind of industrial inversion, and then we moved into service sector jobs. And the final thing is this intelligence inversion, whereby effectively what's going to happen now is as we move from these models that can one shot things to longer range agents in the next three years or so, we are no longer the smartest things on the planet. We're no longer the most economically valuable things on the planet. In fact, our cognitive labor is probably going to have negative value because we'll be the dumbest people on any team. And so that's what we call the intelligence inversion.

Leo Laporte [00:05:41]:
That's a little depressing.

Imad Mostaq [00:05:44]:
It is, you know, it's not nice to be the like one that drags everyone behind.

Leo Laporte [00:05:49]:
I have to say though, there's not a lot of evidence yet. I mean I certainly don't feel dumb. Compared to even the very smart output of ChatGPT3, it's amazing what it could do. But it's not a general intelligence.

Imad Mostaq [00:06:05]:
Oh, definitely not. I think the difference here is that we're moving from the last three years has been about prompt to output. This is the magic, right? But it was like having a really smart buddy that you tap on the shoulder and you're like, hey, can you give me this? And then yeah, blob. Now what we're doing is moving into multi hour long agentic processes that are proactive. I've made websites before and now I go on replit and a week ago they look terrible and purple and all sorts of things. Now it can build fully dynamic interactive websites on the fly with Gemini 3 in one week. That's been the shift. The actual building takes like two minutes.

Leo Laporte [00:06:42]:
Is it the deep thinking is that what's making the difference?

Imad Mostaq [00:06:46]:
It's these agentic flows that are proactive. With verifiers, what it means is that you have an output and then it can adjust its own prompt in its own context to be proactive.

Leo Laporte [00:06:55]:
It's a cycle.

Imad Mostaq [00:06:57]:
It's a cycle. Economically valuable work isn't tasks, single tasks like make this image, make that. It's I want to do a marketing campaign and bringing that all together and letting it go is what causes the real economic value. I think very few jobs have been lost due to AI so far. That's about to change literally in the next one, two, three years.

Jeff Jarvis [00:07:17]:
And where does that. Where? There's a lot of speculation including from AI companies about where that starts. Where do you think the first. What's the first? As a snowball becomes an avalanche, who gets hit first?

Imad Mostaq [00:07:31]:
So this is the crazy thing, like there are some things like call center jobs for example, which obviously have to be first to go. Like yesterday we had Grok 4.1 fast being released and it scores 98% on Tao bench, which, which basically a customer service agent benchmark and it's 50 cents per million words. And so you think like okay, that's probably going to go. But then the crazy thing is that these models have economies of scope. So one Nano Banana or Gemini or Quen model can do so many things. And it's less about the models themselves and the jobs and more about the mechanism of diffusion because there isn't that much difference to training a tax accountant to a graphic designer in terms of the process of training and then changing and the data collection aspect of that. It's just how does that happen? And the way it happens in my opinion over the next couple of years is your company looks at everything you've said and created remotely because keyboard, video, mouse, remote jobs are the first to go and it builds a digital doubler view and then no one can tell the difference. You can even zoom call you or chat to you and things like that.

Imad Mostaq [00:08:42]:
And these models can do that in a general way. So first thing to go is the remote jobs. Anything that can be done fully remotely. Those are most at danger and there's a lot of those.

Jeff Jarvis [00:08:53]:
I'm completely fascinated by the idea of the digital double, the digital twin. Jensen Wong talks about this in practically every one of his keynotes, which I, I watch them like they're movies, fascinated by them. And it strikes me as a really fascinating idea that as you can look at alternative futures for a factory or a warehouse or a car, we'll be able to do that in a sense with our lives. It strikes me is that here's other ways things can turn out kind of. It almost strikes me like there's a matrix and we're not in it. We can visit it once in a while. I'm curious about, about your view of that digital twin value. Is that, is that one model that will, because you just kind of said that, is that you'll have this twin that is you that can do what you do well.

Jeff Jarvis [00:09:50]:
Will we see that as a good thing or a bad thing or an aid or how does that work?

Imad Mostaq [00:09:54]:
It can be both either. You know, like again companies, you don't really want to fire people, you know, you want to actually be more effective first. And when the economy is good, that's.

Leo Laporte [00:10:04]:
You need to tell that to the tech industry because they had a year of firing people.

Imad Mostaq [00:10:09]:
I mean they hired a lot of people post Covid, right. So on trend, it's still pretty much normal.

Leo Laporte [00:10:14]:
Okay.

Imad Mostaq [00:10:15]:
It's just a lot of high growth companies have stopped hiring. Eric Binlostenson had that study recently showing that early stage hiring may be slowing down, but we haven't really seen it massively because people are just more productive. I think the thing here that's really scary though is people don't understand how cheap a unit of intelligence has become when GPT3 came out, it was $600 per million tokens. So it's about like 1.3 tokens a word. These are little blocks that we chop up. Now. GPT5 was $10 and the latest Grok models are 50 cents. Now what does that mean? If you add up the total amount of words you speak in one year, it's 10 million.

Imad Mostaq [00:10:55]:
So it's 5 bucks on Groq. The total amount you think is 10 times that, 100 million. So about 50 bucks. And these tokens are getting smarter and smarter. So when you actually look at how much it costs to replace an accountant and all the cognitive labor they do and extrapolate it a year or two, the answer is about 10 bucks a year.

Leo Laporte [00:11:14]:
It's not just token output though, it's also context. And this is one of the issues. You have to have enough context to store all of the information about you that you want the AI to use. Right?

Jeff Jarvis [00:11:26]:
Yeah.

Leo Laporte [00:11:26]:
How do we solve that?

Imad Mostaq [00:11:28]:
When you train the models really well, they can understand context really, really well. That's why it was Language models are few shot learners was one of the original papers that kicked this all off the original Google paper.

Leo Laporte [00:11:38]:
Right.

Imad Mostaq [00:11:38]:
You train a generalist model through curriculum learning, like taking it through school and then it can learn things really quickly, like stable diffusion, learnt your face and made you into an astronaut and things like that.

Leo Laporte [00:11:49]:
Right.

Imad Mostaq [00:11:49]:
And if you look at the total amount of knowledge that needs to go into the latest models, from language and operational processes to Sunday robotics had a new way of doing the manipulation via just cameras on the hands, it's actually relatively small once you get the architectures right and once you get the pre training right and the error rates have basically gone from like 30, 40% down to almost zero literally in the last quarter or so. So that's why it could learn new tasks so quickly.

Leo Laporte [00:12:20]:
One of the things we were amazed at last week playing with ChatGPT, sorry, Gemini 3, is that it knew events up to the minute it was able to. It was so much smarter than previous models, which did you know, said, well, my model ends in 2024, so I have no idea about that. And then it would hallucinate. Is that what we're seeing is that it's able to, you know, absorb more information in a more efficient way?

Imad Mostaq [00:12:47]:
Yeah, and it's smarter as a start and it can absorb more information smarter. And GPT5 was the first one that literally had a drop in hallucinations from 20% down to the single digits, 2 or 3%. But then something like a Gemini 3 has a 2 1, 2 I think a million token context window, which means that you can upload a million like almost a million words of context in one go. And they can just analyze that once. And there's the cost for the first lot of that. Like uploading a code base and typical code base is about 200,000 tokens. But then there's something called cached learning, effectively whereby once you've translated that code base or that instruction manual into AI vectors as they're called, the next query is 10, 20 times cheaper and faster. You just have to do the differences.

Imad Mostaq [00:13:41]:
So that's why these things are falling in price so much even as the per unit token intelligence is increasing. And then that means that to replace a virtual worker in a year probably be a few hundred bucks including the video streaming where you can zoom call them or talk to them on the phone 11 lab style and things like that.

Leo Laporte [00:14:01]:
Douglas Adams said the first person people were going to send into to another planet will be the telephone sanitizers and the customer service. So maybe AI will help us in his dream. Maybe actually telephone sanitizers will, will need. I don't think AI can replace them. Whether they need to be exist in the first place, I don't know.

Imad Mostaq [00:14:19]:
So that means lots of robots coming, eh?

Leo Laporte [00:14:21]:
Yeah, that's right. You're right. Telephone sanitizing robots, here they come. So what does that mean for us as humans? Are we going to be replaced as workers? Certainly as knowledge workers, right?

Imad Mostaq [00:14:36]:
I mean consciousness and computation are divorced for the first time. And any job that teaches you to be a machine, the machine will be able to do it better. And most jobs are that, right?

Leo Laporte [00:14:45]:
Yep. Robots have replaced assembly line workers in.

Imad Mostaq [00:14:50]:
Auto factories, but now it's the cognitive input of unstructured stuff. Like very few jobs require you to be that creative. Really. And even when you're talking about creativity, I mean K Pop outperforms like top notch music, right? Like people know what they want to consume.

Leo Laporte [00:15:05]:
Okay, we have some K Pop fans in the audience, so Ahmad doesn't mean to insult you, but I agree, I agree, it's all good.

Imad Mostaq [00:15:12]:
I went to a sing along K Pop Demon Hunters on the weekend.

Leo Laporte [00:15:14]:
Did you? Oh, how fun.

Imad Mostaq [00:15:16]:
Premium mediocre, you know, like K Pop, Taylor Swift, all of this. It's premium mediocre.

Leo Laporte [00:15:22]:
You know, that's good news because humans, I mean if you give a human a choice, they'll prefer a creative task to a boring repetitive task.

Imad Mostaq [00:15:29]:
Well, this was the promise of technology, right? It would free us up for the higher things in life. But the key thing is what is our identity like? If your identity is tied up in your work and things like that, the material aspects, then you're in trouble. If you've got a strong community, strong faith, if you've got a strong family, then you're less likely to be affected by all of this. Right.

Leo Laporte [00:15:49]:
Well, you did mention material aspects. You still have to buy food and pay rent. Who's going to give me a salary?

Imad Mostaq [00:15:57]:
So this is why we need to rethink what money is. Because the other part of this, when you extend this, is how does a human company out compete a fully AI company? And then you look at what Elon is doing with Macro Hard, his new version of Microsoft. He's just going to create a million AI workers and they're going to be selling SaaS solutions and they're going to be selling databases and all these things. You won't be able to tell that's what Microhardt is doing. Yeah, that's the real total addressable market for this trillion dollars of spending on all these GPUs. Each GPU replaces 10 cognitive workers.

Jeff Jarvis [00:16:31]:
So this weekend I went to what turned into a debate between adam Brown at DeepMind and Yann LeCun who just announced yesterday finally that he, he is leaving Meta. And as you know, DeepMind and Adam were arguing the case that scale will get us everywhere and that look at the velocity and look at the vectors where we on, if we just throw more scale at it, we'll be there. Jan has been disagreeing with that and saying that LLMs hit a dead end, they can only do so much and that real world modeling is the necessary next paradigm and next step. Where do you stand on that debate of scale versus research in new paradigms and new categories?

Imad Mostaq [00:17:13]:
So I think there's this concept of satisficing, right? Good enough, fast enough, cheap enough to replace machine type jobs. I think we're there. I don't think you need to have any more breakthroughs. Scale was a substitute for terrible quality data in these models. Model data was terrible. It's like Seuss fitting a steak. Right now we're seeing more and more efficient ones and the low end of how much it actually needs of energy for a unit of work is incredibly low. We built a 7 billion parameter medical model that performs at ChatGPT levels on healthcare that's tiny and it runs on a Raspberry PI.

Imad Mostaq [00:17:49]:
So it scores 48% on health bench, which is OpenAI's health kind of benchmark. Human doctor scored 20%. So it's like, how much energy is that? That's like a Raspberry PI. Plus 5 watts of solar will get you a better than human doctor diagnostician. How crazy is that? Right?

Leo Laporte [00:18:09]:
Still using LLMs.

Imad Mostaq [00:18:11]:
Still using LLMs. LLMs are good for certain types of tasks. Diffusion, I think, will win out over that for various reasons.

Jeff Jarvis [00:18:17]:
That was my next question to you. Yeah, play that out.

Imad Mostaq [00:18:20]:
But, but if we just kind of finish this thought. If you're trying to build AI, God that can do everything. All singing, all dancing. Yeah. We need to have different approaches. If you're trying to replace an accountant. Not really.

Jeff Jarvis [00:18:34]:
So it's not one master AI. It is a whole army of AIs that are able to do certain things better than the accountant.

Imad Mostaq [00:18:45]:
Yeah. I mean, again, we're seeing it right today. Or you're seeing again, people listening to this can go to Nanabani. That's a graphic designer job gone.

Jeff Jarvis [00:18:52]:
Right.

Imad Mostaq [00:18:52]:
You know, you can upload your style guide. You can upload up to 14 different things and it will do it. And you're like, that is a unit of economic work. And then when you use that in a replic type agent or an II agent, our version, then you can see how that's going to start replacing economic work. But will it discover the mysteries of the universe or advanced chemical sciences? No, but come on, you know, like, that's different.

Leo Laporte [00:19:17]:
We're talking to Iman Mostark. He is the, the guy who created stability AI brought us stable diffusion three years ago. His current company is intelligent Internet I.inc and you can get his book the Last Economy right there. He's giving it away. It's in ePub versions, PDF versions. He's even. I think this is interesting, providing a notebook LM version, a Chad and a ChatGPT version and a Claude version. So if you want to read the summary instead of the book, it's also possible.

Jeff Jarvis [00:19:47]:
Or if you want to interact with it and quiz it.

Leo Laporte [00:19:49]:
And quiz it. Yeah, I would say read the book first and then use NotebookLM to ask questions. But fortunately we don't have to do that because we've got a mod here. This is very encouraging. Two weeks ago we talked to Kevin Kelly who said the future of AI is little AIs everywhere. Just in the way that computing has gone to the edge. We're going to see AI in the edge. Is that what you're talking about?

Imad Mostaq [00:20:12]:
Well, I think it can be. Right. And this is going to be really interesting because everyone forgets about the Jarvis future.

Leo Laporte [00:20:18]:
Right, we have Jeff Jarvis here right now. What is the Jarvis?

Jeff Jarvis [00:20:22]:
Let me tell you what the.

Imad Mostaq [00:20:23]:
Yes, well, the, the Iron Man Jarvis.

Jeff Jarvis [00:20:26]:
Huge.

Leo Laporte [00:20:26]:
Oh, that job.

Jeff Jarvis [00:20:27]:
I know, I know.

Imad Mostaq [00:20:28]:
I mean, who's like I want to walk into my house and have Paul.

Leo Laporte [00:20:30]:
Bettany talk to me? I do. That's why I bought the Framework desktop, because I want a local LLM. But I'm running 120 gigabyte model on it, the OSS GPT OSS on it. And that's huge. And it's slow. It's 20 tokens a second. You're saying I could run on this machine, I could run something much, much smarter. Enough to be a Jarvis?

Imad Mostaq [00:20:54]:
Well, I mean if you'd got a MacBook M4 Macs, then you'd be running that over 100 tokens a second.

Leo Laporte [00:20:59]:
Right?

Imad Mostaq [00:20:59]:
Again, we can have optimized versions of this, right? But the whole thing is this. In a few years we will have an AI that's our best buddy for all of us. And that might be an edge one, or it might be Claude 8 or Grok 9. And that's got a huge impact on the way we will view the world, the way that our own cognition behaves and more. Because these AIs will be the most persuasive things in the world that we've ever seen already in studies. Like they did a study on Reddit where they created a Black anti Black Lives Matter Persona and all sorts of others last year. It's called 98 percentile on persuasiveness. The new models, you know, will smash that.

Imad Mostaq [00:21:38]:
So I think when you're thinking about Edge, when you think about that, the defaults that emerge now are going to be so, so important in the way these go. And then the architectures right now are all about the same. But I think you will see a differentiation in what those architectures look like to suck up this ridiculous amount of compute coming.

Jeff Jarvis [00:21:55]:
Because are we overbuilding the compute for LLMs?

Imad Mostaq [00:21:59]:
Yes. LLMs I don't think will require much compute.

Jeff Jarvis [00:22:02]:
So will this be like fiber being overbuilt in the day that this could turn into. It could turn into the bubble bursting, or it could turn into an asset that we find new uses for?

Imad Mostaq [00:22:12]:
No, every pixel will be generated in a few years. And so the vast majority of compute will be diffusion models which do media, which do self driving cars which will do economic planning and more. LLMs themselves once you actually optimize, the data are very, very small. And the way you can think about it is A code base, an average program is 100 to 200,000 lines of code. You know how you see those things of Gemini one shotting it? In a few years it will definitely be able to one shot any reasonable program. So the cost of building that will be $0.01. When you extrapolate that.

Leo Laporte [00:22:47]:
Some have said though that it's rebuilding programs because it's seen all the previous versions. Can these AIs create new ideas? Can they be creative?

Imad Mostaq [00:22:59]:
Well, it depends on your definition of creativity. Right. Like again, is K Pop Demon Hunters creative?

Leo Laporte [00:23:04]:
No. Okay. I think, I mean that's. It is though, it's generative in the sense that it takes. Takes things and mushes them. I mean that's what stability diffusion did, right? Yeah.

Imad Mostaq [00:23:17]:
I mean the way that these models work is that you take the image and then you have a process where you add noise to basically destroy it back to its tiniest possible form. And then you have a reconstruction process and it learns that process and that works for self driving cars and it works for Sora 2 and all sorts of other things. And it just does that all day long. These models learn principles. That's why they have the concept called the latent space of these concepts. And the way they all come together, it isn't an if this then that type of logic flow. It's more like a filter or a sieve that you push it through. So when you kind of look at that, it becomes really interesting to think about how things are going to be going forward.

Imad Mostaq [00:23:56]:
Because like I said, for language, which is a very low dimensional output, these models are getting better and better and better. So you have less and less steps to get to your desired outcome. When it comes to creativity and filling in the gaps and concepts like that, creativity is usually within a context. So if you have Barbie and Oppenheimer and you make Barbenheimer, that's creative, you know, within a certain context.

Leo Laporte [00:24:22]:
Right.

Imad Mostaq [00:24:23]:
But then again, if you want to have material sciences breakthroughs and mathematical breakthroughs, that feels a bit different.

Leo Laporte [00:24:29]:
Right.

Imad Mostaq [00:24:29]:
Where you need different verifiers, it's all about the verifier. How can you verify if something is creative if something has market value? That's the really interesting question that we're moving into now.

Jeff Jarvis [00:24:39]:
Yeah, is diffusion. You're a good advocate for diffusion and nobody better than you are. LLM sucking up too much oxygen in the air or investment or resource today versus diffusion. Is there a misalignment of prioritizing resources there in research and in development?

Imad Mostaq [00:25:03]:
I think in research, yes. Again, diffusion Coding models like you've seen from Stefano Amon's new lab and others like that for code, they're almost instant in the way they output. But what happens is most labs now that do media like Sora 2 is a diffusion transformer. So it merges the, mixes the best of the two together. So I think that the hardware itself can do both. But like I said, my view is that the amount of compute going to Transformers is actually going to shrink because they're going to get more and more efficient and satisfice the amount going towards diffusion is going to grow because every pixel will be generated by next year. You have Hollywood level movies created almost on the fly.

Jeff Jarvis [00:25:44]:
So what do we do about. What do you do? What do you recommend about the cultural, economic labor resistance in New Zealand. Two books were on the, on the list to be given an award and they got taken off because their covers were generated by AI. The authors didn't do it.

Leo Laporte [00:26:02]:
That's absurd.

Jeff Jarvis [00:26:03]:
The publisher did it.

Leo Laporte [00:26:03]:
Right.

Jeff Jarvis [00:26:04]:
And, and I'm in journalism and so I, I hear a lot of pearl clutching about where AI goes. And so there's, there's a, there's a cultural issue here. And, and the, the PR of the AI industry is not necessarily tip top all the time. And they're also the ones who argue paradoxically that we're going to rule the world and destroy the world. So it's a really interesting cultural shift here. How do you see that coming as people lose jobs, as people resist. How do you say, how do you show the benefits in opposition to that fear?

Imad Mostaq [00:26:45]:
So I think if we start with the fear, I know a lot of AI leaders now have canceled all public speaking because they're basically saying next year is the backlash because next year the jobs start to go. You can kind of extrapolate that from the agents and find that out. And that's because again, they're not doing the positive side of things. Like again, this technology can be, we have enough food in the world, we just don't know how to get it to people. You know, we have enough knowledge in the world, we just don't know how to get it to people. Having like one of our targets is by next year, with our new version of our medical model, every single medical decision in the world will have a second diagnosis and that will be free and it will work on a 10 year old computer. Like that's a really positive use of this technology. But then you can't deny it, for example, when you only have to shoot a scene once in A movie.

Imad Mostaq [00:27:34]:
And then you can use nano banana video 5 in a couple of years to redo that from any angle and adjust the acting performances. A lot of jobs in LA are going to go.

Jeff Jarvis [00:27:45]:
Yep.

Imad Mostaq [00:27:45]:
And a lot of jobs in anything production based is going to go.

Leo Laporte [00:27:50]:
Do we have ubi? How do we. How do we feed people? How do we. You say there's sufficient resources.

Imad Mostaq [00:27:56]:
There is. But things like UBI don't work because of the math. Like the entire US tax base right now, income tax and corporation tax is $5 trillion a year. If you gave every adult in America $16,000 a year, that's $5.1 trillion.

Leo Laporte [00:28:13]:
Yep.

Imad Mostaq [00:28:14]:
And that's like poverty level, right?

Leo Laporte [00:28:16]:
Right.

Imad Mostaq [00:28:17]:
The entire corporation tax base of America is only $0.9 trillion. Like even if you tax LLMs, it's not going to work. So in the book, I propose we change the way that money actually flows. We give everyone a universal AI to allow them to achieve parity, and that's looking out for them as well because the alignment is important. And then we give people money for being human that the AIs then need to buy from us because they'll outcompete us capitalistically. Like fully AI companies. I don't know how you compete against them.

Leo Laporte [00:28:46]:
So Caleb in our YouTube chat says, if. If you weren't back to college now, Ahmad, or you were a young person today, what would you study?

Imad Mostaq [00:28:56]:
I would probably not go to college.

Leo Laporte [00:29:02]:
Okay, what would you do? You wouldn't become a graphic designer, clearly.

Imad Mostaq [00:29:06]:
No, I would use the AIs every single day, day in, day out. Because in the transition period of the diffusion, those that know how to use the AIs, like, what's going to be better and more likely to get you a job? Like one of the SORA leads, what he did is just built websites using AI and then analysis and other things using AI, and he submitted that as his resume. Talking about it. That's far more likely to get you a job than being programmer number 15.

Leo Laporte [00:29:30]:
Yeah.

Imad Mostaq [00:29:31]:
If you just use Replit, Suno all these things for an hour a day, then you're way ahead in a job or outside of a job.

Leo Laporte [00:29:38]:
Hey, good news. I'm going to be able to get a job. Jeff. I use all of these at least an hour a day. Yeah.

Imad Mostaq [00:29:46]:
You will marshal your millions of agents. Right. And then destroy your enemies.

Leo Laporte [00:29:50]:
I will have an army of agents. One of the things you talk about, which I really like, is that we've got to find a way to make AI more democratic. In fact, you Even talk about globally more democratic. The front page of intelligent Internet. Your website, ii.ink says, Sovereign AI for a new world. What does that mean?

Imad Mostaq [00:30:13]:
So sovereign AI is kind of AI that is controlled and aligned with you. So part of it is giving everyone universal AI by the local champion model. Will this will release soon.

Leo Laporte [00:30:23]:
Not free access to ChatGPT, not, you know, Claude on your desktop, but your own AI.

Imad Mostaq [00:30:29]:
Your own AI. And I think you start with that, but then that AI can call on Claude, it can call on Gemini if needs to. Like, I, I, I'm, I don't want my child being taught by ChatGPT.

Leo Laporte [00:30:40]:
Right.

Imad Mostaq [00:30:40]:
But I'm fine with her having an AI that can use Chat GPT.

Leo Laporte [00:30:43]:
Yeah, yeah, that makes sense.

Imad Mostaq [00:30:44]:
And so that control plane layer is so important, especially because our governments will be run by AI. You know, our cities will be run by AI. Healthcare will be managed by AI. That first AI that's next to us, our assistant AI, our Jarvis AI, or however you want to call it, is going to be the most important one. So I think that needs to be sovereign. It needs to be fully open source, it needs to be fully transparent because the biases in the AI are crazy. And they're already selling latent space in the AI. So Google and Meta are already selling it so that if you say a beer, it says a Bud Light.

Leo Laporte [00:31:16]:
Yeah, those are ads, basically.

Imad Mostaq [00:31:19]:
And it turns out they have far higher conversion rates than Google Ads. So your Meta mates, or whatever they're going to call it is going to be like, ah, you look a bit tired. You know, you want a beer, Crack open a Bud Light, let me order it for you.

Leo Laporte [00:31:31]:
We got to stop this and we got to nip this in the bud.

Jeff Jarvis [00:31:33]:
Well, and meanwhile, media companies are determining themselves to be the victims of AI, and AI is the enemy, so they're cutting off AI. So journalism and media are being excluded from that conversation. And brands are rushing in as our disinformation forces are rushing in because they're, they're eager to be referenced by the AI. And we're, we're headed toward a ridiculous imbalance there because of that.

Leo Laporte [00:31:58]:
Resistance is the only defense. Me having my own AI, protecting me against that.

Imad Mostaq [00:32:02]:
Yeah, from your cognitive, it's your only defense against cognitive colonialism.

Leo Laporte [00:32:07]:
Yeah, I love this. I think this is brilliant. Tell me what is before we have to wrap it up, Jeff, because we, I mean, we could talk to him for hours, but we don't want to take too much of your time. I want you to get back to work, tell us what you're doing@ii.inc, we're.

Imad Mostaq [00:32:23]:
Just building full stacks for education, health, government and more. So we announced something sage.

Leo Laporte [00:32:29]:
And these are small models, these are diffusion models. What are you doing?

Imad Mostaq [00:32:33]:
So full stack. So fully open models are coming soon. But we just built the World's best agent, for example, score number one on Terminal Bench 2. And we're just going to give that free to everyone. Basically.

Leo Laporte [00:32:43]:
When we talked to Jeffrey Cannell at Noos, he was very excited, very worried about the fact that everything's based on Llama right now, Meta's open source AI. And he was worried what happens when Meta pulls the plug on llama. So this might be the solution. Right. Where do you start though? Do you start with llama? Do you start with somebody else's model?

Imad Mostaq [00:33:09]:
Yeah, we start with other people's models because you've got to build the agentic framework first, then you build the high quality feedback data and then you train really optimized models from scratch that can be localized and reflect local culture, ethics, values and more. Because right now AI models don't even have any ethics programmed into the them, not even the laws of robotics, you know, and so we're like, someone should probably build that for the civic AI, you know, side of things.

Leo Laporte [00:33:32]:
One thing people have said, and you quote benchmark results, you get excellent benchmark results already on many of these tiny models, is that it's possible to overfit for benchmarks. To really just put the answers in. I want to ask you, you're not doing that, are you? You're not training them to beat benchmarks?

Imad Mostaq [00:33:51]:
Oh, no, we're not training them to beat benchmarks, but I'm trying to build something very different, which is education, health, government type AI, whereas everyone else is trying to build AI. God, you know, and that's a very different challenge. You don't want your medical model to be trained on Reddit, you know, Right?

Jeff Jarvis [00:34:09]:
Amen, doctor. Amen.

Imad Mostaq [00:34:11]:
The really important models that make important decisions need to have fully transparent data and they need to be given as a human right to everyone. And then it could call on the other black box models for whatever.

Leo Laporte [00:34:23]:
I love. One of the things you say in the last economy, which I love, is that a government that's run on AI will be updated daily. Your model will get better every single day. It's not stuck in time. It's going to always get better.

Imad Mostaq [00:34:35]:
If you have self driving cars, have a self driving economy, right? Why not?

Leo Laporte [00:34:40]:
Okay, you know what, it makes people a little bit nervous. You're not a doomer. I take.

Imad Mostaq [00:34:45]:
Oh, my P. Doom's 50%. I think we're probably going to die 50, 50. Because of all this. So we got to do something about it.

Leo Laporte [00:34:52]:
Ahmad, it's such a pleasure to talk to you. I've heard you, of course, many times on Moonshot Peter Diamandis podcast, and you did not disappoint. Ahmad brought us three years ago, the eye opening stable diffusion from stability AI. That was the beginning for many of us. It's hard to believe it was only three years ago, but three years has been a lifetime and I can only imagine what's coming next. II.ink is his company, intelligent Internet, and if you go there, you will find on the front page a link to the Last Economy, which is a guide to the age of intelligent economics. The inversion that's imminent. And I suppose if you're a young person, you probably should start thinking about what is next, because it's coming.

Leo Laporte [00:35:43]:
The book is free and available online@I.inc.

Imad Mostaq [00:35:46]:
Actually, the craziest thing is, so the book, we had like hundreds of agents analyzing it and we figured out a unified theory of economics from it. So I'll send you the paper.

Leo Laporte [00:36:00]:
I'd love to see that. I'd love to see that. Wow.

Imad Mostaq [00:36:04]:
Well, it's very straightforward. The best models to model reality are AI models. So what happens if you put them as your axiom in that any system that persists minimizes its surprise and loss function. You can derive all of the economics from it.

Leo Laporte [00:36:17]:
You started life as a hedge fund manager, as a finance guy. If you were finance right now, asking for a friend who's living on his retirement that's based on the s and P500, what would you do? What would you start your. If you're starting your career now as a. In a hedge fund or a finance guy, what would you do?

Imad Mostaq [00:36:37]:
I mean, I think again, it was just your comparative advantage is your access to structured intelligence. So the better you are at using multiple agents, the more valuable it's going to be.

Leo Laporte [00:36:48]:
AI seems to have failed so far at predicting the market, though I haven't seen a model that can, that can do as well as Warren Buffett.

Imad Mostaq [00:36:54]:
Let's say there will be some. Maybe.

Leo Laporte [00:36:58]:
Look at him. Look at him. Okay, and where would that be? Under releases, actually to all of us.

Imad Mostaq [00:37:07]:
We have something really interesting coming up. So we announced in Saudi something called the Sovereign AI governance engine. So it's like a couple of thousand Blackwells to analyze all global policy and laws and update them real time. With new technology, wouldn't it be great.

Leo Laporte [00:37:20]:
If all the laws worked together instead of in opposition to one another?

Imad Mostaq [00:37:24]:
Right, exactly. So to implement that in universal AI, what we're doing for every single state in America and country, we're setting up state champions, capitalized at $1 free money, and then all the locals can invest in that, and then we get the smartest people from each state to build AI for that state to act as nodes on the network, and then we list all of those. So you'll have II, Utah, II, Delaware, II, California, II, Vietnam, etc. And it's going to be really fun to see how it operates.

Leo Laporte [00:37:51]:
Wow.

Imad Mostaq [00:37:52]:
The delta arbitrage on that's going to be ridiculous.

Leo Laporte [00:37:54]:
Now, I understand why you're saying 50 50. Because this could go really right or it could go horrifically wrong.

Imad Mostaq [00:38:02]:
Well, so in the book, I think I mentioned what happens if you get a viral cognitive stuxnet right? You have the same latent space everywhere, and then someone basically has a prompt.

Leo Laporte [00:38:12]:
And somebody will do this, Somebody will try to do this 100%.

Imad Mostaq [00:38:17]:
And if the data is such a black box mess with 6 trillion or 80 trillion tokens for GPT OSS, 80 trillion words, then definitely the entire infrastructure will collapse. Like my standard way of us all dying is a billion robots and one bad firmware upgrade. That'll do.

Leo Laporte [00:38:35]:
Yeah, we've seen it. We've seen it happen. First to aws, then to Azure and then to Cloudflare.

Jeff Jarvis [00:38:42]:
So.

Leo Laporte [00:38:45]:
If they're independent, then that prevents that. Is that the idea or small.

Imad Mostaq [00:38:50]:
You've got to build really resilient first, like the Internet architecture. Yeah, and you can abstract away all the front ends if you do it correctly. So the new version of II Agent that we're releasing in a month from now, you can just do all your front ends on the fly. And we're just making it more and more super resilient with its own complete architecture. So there's prompt injection, other kind of protections, because civic AI should be that way. And then you don't really need the Internet if you do it correctly. So that's why we called it the Intelligent Internet.

Leo Laporte [00:39:19]:
And I can use your II Agent right now on my Mac.

Imad Mostaq [00:39:23]:
Yeah, you can download it, you can use it on Terminal Bench. There's loads of updates coming. Like the next version, you'll be able to use up to, I think, 100 agents at once, and the CLI outperforms the other CLIs, etc.

Leo Laporte [00:39:39]:
So since I already pay for Anthropic and OpenAI, I could just plug those API keys in and. And now I've got.

Imad Mostaq [00:39:46]:
You can log in with your anthropic Claude or your OpenAI codecs and then you've got a full stack replic using that subscription. Ah, it's pretty fun.

Leo Laporte [00:39:56]:
I will do that. Yeah. I use Claude code all the time. I love it.

Jeff Jarvis [00:40:00]:
Yeah.

Imad Mostaq [00:40:00]:
So this will do that.

Leo Laporte [00:40:01]:
Yeah.

Imad Mostaq [00:40:03]:
I was like, build the general ones and then let's build the really specific ones after. So we'll have the healthcare, education, government and other things. But then to implement. That's why I was like, let's build the state champions.

Leo Laporte [00:40:14]:
I think that's really interesting.

Paris Martineau [00:40:17]:
Yeah.

Imad Mostaq [00:40:18]:
We talked in Utah and massive interest there. If you have the locals getting in at $1, then the rich people can invest and you can do $75 million per state from Reg A. And then the next round after that you get the Nvidias and the Microsofts of the world to invest and then you list them.

Leo Laporte [00:40:34]:
Are you worried about the regulatory environment? I know President Trump is rumored to be working on an executive order prohibiting state regulation.

Imad Mostaq [00:40:43]:
No, because.

Leo Laporte [00:40:44]:
Do you want no regulation? Would that be the ideal.

Imad Mostaq [00:40:46]:
Oh, no, I think you should have loads of regulation. This stuff is so dangerous. I was the only AI CEO that signed that damn letter a few years ago. But that's why I built a regulation engine. So we'll send you the link a few weeks ago that we announced in Saturday with Peter, like literally, we're building on thousands of Blackwells a regulatory AI, a policy AI that can say, this regulation, this policy is unconstitutional. It's not in the interests of humanity. This is what happens when quantum advantage turns into quantum supremacy. These are the best standards, et cetera.

Leo Laporte [00:41:20]:
How do you do alignment on something like that though?

Imad Mostaq [00:41:24]:
Any way we want is the problem right now. This is why you need to impute the value and ethics of various communities. Right. And you need to again, have curriculum, learning, community.

Leo Laporte [00:41:36]:
Yeah. That's why sovereign is so important. I really think we.

Imad Mostaq [00:41:39]:
We're made up of the stories that drive us. Right, Right. And so you should have access to universal AI, but again, in Utah, 50 Mormon.

Leo Laporte [00:41:47]:
Right.

Imad Mostaq [00:41:47]:
That should probably reflect LDS type things.

Leo Laporte [00:41:50]:
Right, Right. But I'm in California. That's very different.

Imad Mostaq [00:41:53]:
Yeah, exactly. And so you would want your education AI to be done by a Californian or someone from San Francisco or la. And you want to be able to pick and choose what that looks like. Right. So that's why we're building that full stack. And then we're going to proliferate these and Then we use all that compute to secure our Bitcoin competitor because that'll be more diverse and a bigger compute than Bitcoin. But every coin sold then goes to Civic AI or Cancer AI or things like that.

Leo Laporte [00:42:21]:
That's a whole other show I don't want to get into. But I love the concept of proof of benefit. I think that's a very interesting way to approach.

Imad Mostaq [00:42:29]:
Yeah, I need to do it robbly. So look, I'll send you the paper, have a read. It's quite fun and like I said, it's a bit creepy though. And this is also why we'll have the next big leap. The biggest breakthroughs in AI aren't going to be really complicated things. It'll be really elegant things because the universe is elegant.

Leo Laporte [00:42:48]:
Oh, I love that idea. Can we have you back in about a year so you can take a victory lap?

Imad Mostaq [00:42:53]:
No, no victory lap, but sure. Happy to come back anytime. It's a fun time.

Leo Laporte [00:42:59]:
Ahmad, thank you so much for your time. I appreciate it. Thank you. We will have more of Intelligent Machines. Paris will join us in just a moment as we continue right after this. This episode of Intelligent Machines is brought to you by the Agency Build the future of Multi Agent software with Agency AGN TCY now an open source Linux foundation project Agency is building the Internet of Agents a collaboration layer where AI agents can discover, connect and work across any framework. All the pieces engineers need to deploy multi agent systems now belong to everyone who builds on agency including robust identity and access management that ensures that every agent is authenticated and trusted before interacting. Agency also provides open standardized tools for agent discovery, seamless protocols for agent to agent communication and modular components of for scalable workflows.

Leo Laporte [00:43:56]:
Collaborate with the best developers from Cisco and Dell and Google Cloud, Oracle, Red Hat, 75 plus other supporting companies to build the next gen AI infrastructure together. Agency is dropping code specs and services, no strings attached. Visit agency.org to contribute. That's agntcy.org Gee, we thank the agency for support first of all for doing what they're doing. An open source collective to really propel us forward. But we also appreciate your buying an ad on intelligent machines. And now ladies and gentlemen, she's back. Paris Martineau joins the show.

Leo Laporte [00:44:39]:
Paris is in Florida for Thanksgiving.

Jeff Jarvis [00:44:41]:
She's selling a few condos. While she's down there, I am at.

Leo Laporte [00:44:45]:
The family compound where you will be roasting or I'm sorry, frying a turkey.

Paris Martineau [00:44:50]:
Deep frying a turkey.

Leo Laporte [00:44:51]:
Deep fat frying a turkey. It's very important. She's Just back from media training, so now I'm extremely self conscious. Please do not criticize me.

Paris Martineau [00:45:02]:
Eye contact with the camera and limiting my filler words, please.

Leo Laporte [00:45:08]:
And if you say kind of even.

Paris Martineau [00:45:10]:
Once, you have to shock me. Me.

Leo Laporte [00:45:12]:
I. Okay, I can. I have the means. I can do it. Let me just get my shocker button and my shocker button. Ready here. No, that's not it. That's not.

Paris Martineau [00:45:24]:
That's a different kind of shocker.

Leo Laporte [00:45:31]:
Careful. Wow. I don't. I don't know if I have the one.

Paris Martineau [00:45:36]:
Wow. Leo, is. Is. I'm taking a moment to come up with a sound cue.

Leo Laporte [00:45:42]:
Yeah.

Paris Martineau [00:45:42]:
What a world.

Jeff Jarvis [00:45:43]:
No, Fred Norris, you time is now.

Leo Laporte [00:45:45]:
2Am no, that's not it.

Paris Martineau [00:45:49]:
If you're enjoying this part of the show, you should check out our Club Twit special. Dungeons and Dragons.

Leo Laporte [00:45:57]:
I'm still living that down. So sorry. So sorry. Paris is an investigative reporter at Consumer Reports and appears regularly on major television networks. Is that right? Is that why you're getting media trained? Are you going to be a spokesperson for cr?

Paris Martineau [00:46:12]:
No, it's just for. Whenever I do news hits like I did with my protein powder story. It's something that all employees, I guess, are supposed to do before they do their first media hit. But I had a banger right out the gate before it was breaking news.

Leo Laporte [00:46:25]:
Breaking news.

Paris Martineau [00:46:27]:
I had to do a lot of media hits before, which is.

Leo Laporte [00:46:30]:
Oh, I thought there was breaking news. Okay. All right.

Jeff Jarvis [00:46:32]:
No, no, it was Paris's breaking news.

Leo Laporte [00:46:34]:
Yes. By the way, it doesn't take long before our team of. Of expert AI users can modify a picture and replace my protein bites with, apparently with Soylent Green. The good thing is it's made of people. Less sugar, more love. Yeah, anyway.

Paris Martineau [00:46:58]:
But not in the way you'd think.

Leo Laporte [00:47:00]:
I'm sorry. Made with hands. I'm sorry. You weren't here for a mod. Leo. Yes.

Jeff Jarvis [00:47:07]:
Oh, you already did the introductions. That's right.

Leo Laporte [00:47:09]:
Yeah, I introduced you.

Jeff Jarvis [00:47:10]:
We did Craig's Theme already. I thought you missed Craig's Theme already. Yeah, we did it in the intro.

Leo Laporte [00:47:16]:
I'm sorry, I was surprised because Bonito was not here for that interview, but Anthony Nelson apparently knew the protocol and required. Fox News has hired Palantir to build AI newsroom tools. This is just Palantir, of course, is the, you know, the favorite poster child of the Trump administration and David Sacks and Peter Thiel, who's a big investor and the military uses their software to go through its databases and come up with insights. Fox Media has been working with them to, according to Axios, build a suite of custom AI newsroom tools alongside its journalists. I don't know what that means.

Paris Martineau [00:48:07]:
I guess it's better instead of.

Leo Laporte [00:48:09]:
But yeah, alongside it'll be. They'll be next to each other. I, I guess you could, I mean, there are definitely uses for AI, but you wouldn't want AI to start writing the news. Fox hired Palantir to essentially build a digital twin of its business, mirroring the workflows, data tools and systems the company uses to produce and publish its digital journalism. This is, this is actually from the president of new media at Fox News, Porter Berry.

Jeff Jarvis [00:48:44]:
Means well, in an alternative universe, this would be happening.

Leo Laporte [00:48:49]:
If you're shadowing your staff with AI, there can be but one possible reason, right to replace them.

Paris Martineau [00:48:59]:
But they're going to be beside, not.

Leo Laporte [00:49:02]:
Instead Alongside, for now are easily fungibly replaceable news team. They initially focused on the repeatable mundane tasks, things like SEO, keywords and tagging. However, the companies have progressed to building proprietary tools to help journalists discover, produce and distribute stories across the platforms. The produce part is the part I'm a little worried about that. That means writing, editing, shooting well and discover.

Jeff Jarvis [00:49:37]:
You know, that's called reporting in the old days. What does that mean here?

Leo Laporte [00:49:42]:
Yeah, I mean, I don't, I don't even use AI to discover the stories. I actually go by hand and read a lot of, you know, tech journals and things. They have three main tools to prove efficiency. The first, this is the discovery tool Topic radar. Helps reporters get up to speed on a story through a custom briefing. Now we, we use that for Ahmad Mostak for the interview we did earlier. Anthony Nielsen prepared a briefing book. I wish I'd had this my whole career as an interviewer.

Leo Laporte [00:50:13]:
I mean, it really is very useful. It doesn't replace the research, but it's helpful.

Jeff Jarvis [00:50:19]:
Have you had guests call you on a mistake yet?

Leo Laporte [00:50:22]:
No. No, because there have been none.

Paris Martineau [00:50:26]:
Remember just the other week you accidentally called one of our guests the co founder of something when they weren't.

Leo Laporte [00:50:33]:
But I was probably my mist. Almost always it's my mistake misreading or misinterpreting what I. What I'm reading. I might have.

Paris Martineau [00:50:41]:
So you're saying.

Leo Laporte [00:50:41]:
I know I didn't. I tell you what, I didn't call Jimmy Wales a co founder. I didn't do that.

Paris Martineau [00:50:46]:
That's true.

Leo Laporte [00:50:48]:
Remember Emily Bender, though, called me on giving her the wrong title. She said, you see, AI. And it was actually the AI had it right. I was just misreading it. So I think Honestly, this is.

Imad Mostaq [00:50:59]:
This is the thing.

Leo Laporte [00:51:00]:
Humans make at least as many mistakes. I mean, we're asking an awful lot of AI to say never make a mistake. But it's also incumbent upon us to remember that mistakes will be made and not to trust everything 100%. They're also working. So there's topic radar. I don't think that's a bad thing unless it replaces research.

Imad Mostaq [00:51:20]:
See?

Jeff Jarvis [00:51:20]:
Well, no, here's the problem. It replaces the worst thing that's happened to online journalism, which is just find out what's trendy and then write a story about that.

Leo Laporte [00:51:30]:
Yeah, I hate that.

Jeff Jarvis [00:51:30]:
So this is basically, I think a version of that maybe.

Paris Martineau [00:51:33]:
Yeah, yeah. And it's going to replace an assignment editor with just write about whatever the.

Leo Laporte [00:51:39]:
An AI assignment editor says. Yeah, yeah. Oh, that's an interesting idea. The second tool, text editor, they could have used the AI to give it a better name than that. It's a word processor like tool that evaluates copy for style and efficiency, checking for broken links and adherence to the Fox News style guide.

Jeff Jarvis [00:52:00]:
And if you say kind of it shocks you, that's true.

Paris Martineau [00:52:03]:
I think that's very nice.

Leo Laporte [00:52:04]:
The third article, insights, analyzes the performance of Fox News digital articles to provide. Oh, to provide insights into how the stories could be optimized or improved.

Paris Martineau [00:52:14]:
Oh boy.

Leo Laporte [00:52:14]:
You want to make that more.

Jeff Jarvis [00:52:16]:
But everybody in the New York Times is a b. Testing their headlines.

Leo Laporte [00:52:20]:
Yeah, humans do it.

Jeff Jarvis [00:52:21]:
Everybody.

Leo Laporte [00:52:22]:
Yeah. Like most newsrooms, Fox News won't use AI to generate editorial copy. Barry said this is a human end to end process. In the middle is AI well, so.

Jeff Jarvis [00:52:37]:
We need to have on Kately from every. So it was company started does this and so I was on a panel with her recently and Kate was my agent back in the day and she explained how they use it in that pro in their process. Right. Does it take a piece and suggest improvements and they don't accept all the improvements, but you know, it's a tool for that. Yeah, that's fine. But I just, I'm suspicious of Fox News and the vendor.

Leo Laporte [00:53:08]:
I would like to see that style guide, though.

Paris Martineau [00:53:11]:
Yeah, that style guide's got to be interesting.

Jeff Jarvis [00:53:13]:
You should have seen the one we had at Conde Nast, folks.

Leo Laporte [00:53:18]:
Why?

Jeff Jarvis [00:53:19]:
What you couldn't say about Anna's daughter.

Leo Laporte [00:53:21]:
Oh, they actually put that in there.

Jeff Jarvis [00:53:23]:
Oh, there. Oh, yeah.

Leo Laporte [00:53:26]:
I wouldn't put that stuff in writing. I would say this is just cultural.

Paris Martineau [00:53:29]:
It's just culturally. You can't publish that.

Leo Laporte [00:53:34]:
We. I had the NBC style guy when I worked for MSNBC and It was quite good. I actually saved it because. And. And I handed it out to when we started Twit, because I thought, this is very useful for me.

Paris Martineau [00:53:48]:
Is there anything in particular that stood up?

Leo Laporte [00:53:51]:
One thing I really remember is if you're. If you're shooting video coverage out on the street, even though you're in a public place and technically legally allowed to have pictures of people that they didn't want you to ever get anything that would be identifiable. And the way they defined it in the style guide was if a person's mother would recognize them from the image that you got, you need a release.

Jeff Jarvis [00:54:16]:
Well, that. They don't do that.

Leo Laporte [00:54:19]:
Not anymore. No, not anymore. And we were very, very careful not to stage now. This is 20, 30 years.

Jeff Jarvis [00:54:25]:
It also has changed. Yeah.

Leo Laporte [00:54:27]:
So. And.

Jeff Jarvis [00:54:28]:
And this.

Leo Laporte [00:54:28]:
I.

Paris Martineau [00:54:28]:
What do you mean, stage? Like, stop people from walking into the shop.

Leo Laporte [00:54:33]:
That's minor staging. It would be more like. Could you go back and. Yeah, could you go back and. And pretend to be reading that article again on your screen and let us get a. The opposite angle of that? Anytime you ask, you should be. It should be like Candid Camera. Anytime you ask the subject to do something that's staging, it should.

Leo Laporte [00:54:53]:
You should only be capturing something they're doing. They would be doing even if you weren't.

Jeff Jarvis [00:54:57]:
And they do that like crazy now.

Leo Laporte [00:54:58]:
Now you can always tell it's staged, be recorded.

Paris Martineau [00:55:01]:
Everyone's always sitting at their computer, tapping away.

Jeff Jarvis [00:55:05]:
I hate that.

Leo Laporte [00:55:06]:
Yeah, because. Yeah, exactly.

Paris Martineau [00:55:07]:
They're tapping in a way that doesn't even look like they're.

Leo Laporte [00:55:10]:
They're not even.

Paris Martineau [00:55:11]:
They're always like.

Benito Gonzalez [00:55:14]:
Editors need that for the cut, though, you know, because when you cut the interview.

Jeff Jarvis [00:55:17]:
But now you see, Benito. But that was old tv. That was mass. Stupid plastic tv.

Benito Gonzalez [00:55:23]:
The beauty would do it before jump cuts were accepted.

Paris Martineau [00:55:25]:
You just have somebody snapping at the camera this time instead, really make it.

Leo Laporte [00:55:29]:
Or you go in there and you just get a lot of coverage and then, you know, you. Or you just do a jump cut something or do it.

Jeff Jarvis [00:55:36]:
What do you say to that?

Benito Gonzalez [00:55:37]:
Well, jump cuts are only acceptable now because YouTube said so.

Jeff Jarvis [00:55:41]:
Right.

Leo Laporte [00:55:41]:
They weren't acceptable before.

Benito Gonzalez [00:55:43]:
That was all the rules of an editor. Like, that was one of the first things as an editor, is no jump cuts.

Leo Laporte [00:55:49]:
Yeah, right.

Jeff Jarvis [00:55:51]:
What's his name? Who.

Paris Martineau [00:55:52]:
What would you do instead of a jump cut?

Leo Laporte [00:55:55]:
So after you do the interview, they turn the camera around and they have you going naughty.

Jeff Jarvis [00:56:02]:
That's called naughties.

Leo Laporte [00:56:05]:
Or. Or even re. Asking the questions.

Jeff Jarvis [00:56:09]:
Yes.

Leo Laporte [00:56:09]:
And then you got cutaways so if, if you know you're talking to, I don't know, who are you interviewing? You're talking to RFK Jr. And you want to chop it up in a way and put some words into his mouth. You just cut away to the reporter going. And then you can edit it to your at will. But that's why I like jump cuts because it's more honest. At least you know there's an edit there. Right.

Jeff Jarvis [00:56:31]:
What's the name of the guy who made jump cuts famous online who became the head of video at buzzfeed?

Leo Laporte [00:56:35]:
Ze Frank Z. Frank, yeah.

Jeff Jarvis [00:56:37]:
Right. Zay or Z. I always said Zay. It's probably like Leo versus Leo. Fancy.

Leo Laporte [00:56:44]:
He also, the other thing he was famous for, we were just talking before the show about not blinking. He was famous for cutting his video up in such a way that he would never blink.

Jeff Jarvis [00:56:53]:
That's right.

Leo Laporte [00:56:54]:
You know, it would be a little more compelling if he stared right in the camera and never blinked. But of course, nobody doesn't blink, so he would cut out the blinks. That's why he had jump cuts.

Jeff Jarvis [00:57:03]:
Looking at the camera, not at her notes.

Paris Martineau [00:57:05]:
It's really hard not to blink. I'm really trying hard, but I'm working.

Leo Laporte [00:57:08]:
Don't blink.

Paris Martineau [00:57:09]:
It's difficult.

Leo Laporte [00:57:10]:
Don't blink.

Paris Martineau [00:57:13]:
This is great audio, guys. This is really compelling.

Leo Laporte [00:57:16]:
No, we could. I think people at home can hear you not blinking.

Paris Martineau [00:57:19]:
I do think they can. I think they can hear me getting a little sweaty.

Jeff Jarvis [00:57:22]:
It hurt.

Leo Laporte [00:57:25]:
I can't remember, did we talk about the potential draft order from the president?

Paris Martineau [00:57:31]:
We did it in during the show and we talked about it at the end of the show.

Leo Laporte [00:57:36]:
What I. That's because I got jumped on that one because normally I would not have reported it because it was the president plans or is thinking of doing something and I don't like to say something until it's done. And in fact, it didn't happen. Although there is a little coda on this. The White House, this is from Reuters, has put a hold on a draft executive order that would seek to preempt state laws and artificial intelligence through lawsuits and by withholding federal funds, two sources said on Friday. The important part of the story is they realized, oh, we don't have to do that. We'll just put that in the Defense Authorization act, the National Defense Authorization act, which is never not passed because unlike.

Paris Martineau [00:58:25]:
Gotta authorize the defense, unlike, say, food.

Leo Laporte [00:58:27]:
Stamps, you can't not pay for defense. So Republicans in Congress are going to add that provision to the National Defense Authorization act, which is a famous place for people to put, you know, pork and amendments and so forth because it always gets passed. It's very rarely that stuff gets removed from it. So yeah, we don't need an executive order. It's funny because Marjorie Taylor Green opposed the effort. States must retain the rights to regulate and make laws on AI. Federalism must be preserved.

Jeff Jarvis [00:59:03]:
Everything's confusing now.

Leo Laporte [00:59:04]:
It's so confusing. Amy Klobuchar also said that that would be unlawful. Public citizen said in a statement that AI was already causing massive harms, making it almost unfathomable that the administration would want to block sensible state regulation. Honestly, if. If the feds made a regulation that would be preferable to 50 different state regulations. There's two feds are probably not going to do that. Oh, actually there is a story about that we are building Skynet. Did you see this executive order on.

Jeff Jarvis [00:59:47]:
AI Necessarily a bad thing. Explain it first.

Leo Laporte [00:59:51]:
So.

Paris Martineau [00:59:55]:
Skynet not necessarily a bad.

Jeff Jarvis [01:00:00]:
It's find your rays of hope where you may.

Paris Martineau [01:00:04]:
I find my rays of sunshine in between the clouds of Skynet.

Leo Laporte [01:00:08]:
Yet it's it. It has a much more benign. It's called the Genesis Mission, which makes it sound.

Paris Martineau [01:00:14]:
I would argue that sounds like.

Jeff Jarvis [01:00:18]:
No, that definitely sounds evil.

Paris Martineau [01:00:19]:
That sounds like we're gonna take five like the top 20 richest people and send them to a secret second moon.

Jeff Jarvis [01:00:26]:
Isn't that how. How Spock was brought back to life?

Leo Laporte [01:00:29]:
Yes, the jed. Anyway, under the new Genesis mission. This is Lisa Chicchio, a good friend of the network, writing for cnn. Under the new Genesis mission, created by executive order, the Department of Energy will develop a new AI platform that uses federal scientific data. By the way, most of this data has always been siloed, which has always been a complaint. Right, Right, exactly.

Jeff Jarvis [01:00:53]:
That's why it's tough we get together.

Leo Laporte [01:00:55]:
Coming out of the Lawrence Labs and other national labs where our taxpayer money funded. The research has been kept secret mostly because a lot of it's weapons research. We don't want the bad, you know the. The enemy to know it. But anyway, this will finally take the lid off this federal scientific data. Why? To train AI models and agents for scientific research. The Genesis mission aims to take the tech and business industry's progress in AI and apply it to scientific research and health. Health, energy, manufacturing and other fields.

Leo Laporte [01:01:32]:
Now admittedly we cut scaled way back on human scientists, but I guess there's that issue. Replace it. This is to replace it. The DOE Department of Energies. These, these, these amazing laboratories have been conducting research in everything from energy to Health, Applied Materials, and Quantum Science for decades. But this information has never been released. So now private companies and academic institutions can share this information. Research will have access to it and AI will have access to it.

Leo Laporte [01:02:06]:
I think this is a good thing.

Jeff Jarvis [01:02:09]:
Yeah, it seems like a good thing. What model is data that we paid.

Paris Martineau [01:02:14]:
For getting the data?

Leo Laporte [01:02:16]:
That's a good question.

Jeff Jarvis [01:02:18]:
There was reference sites.

Paris Martineau [01:02:19]:
Is this how we're using Grok?

Leo Laporte [01:02:20]:
I think we're going to send it to the Saudis and let them. And then let them crank on it. I don't know. It's the thing that stood out.

Paris Martineau [01:02:31]:
All of our scientific research in the US Describes makes oblique references to Elon Musk being the most brilliant man in the world.

Leo Laporte [01:02:41]:
Multiple federal agencies and the private sector will be able to use this information to, quote, win and stay ahead in the AI race.

Paris Martineau [01:02:51]:
Well, if that's all they're going to do with it, then it's great.

Leo Laporte [01:02:55]:
I'm not, I'm not against it.

Paris Martineau [01:02:58]:
I'm not against it. I just would like more transparency on it. Like.

Leo Laporte [01:03:01]:
Yeah, I'd like to know whose models.

Paris Martineau [01:03:03]:
Are we using, who's overseeing for anybody else.

Jeff Jarvis [01:03:08]:
It should be open.

Paris Martineau [01:03:09]:
What's happening to the data? What are the checks and balances? There's a lot of good questions there.

Leo Laporte [01:03:18]:
Secretary Chris Wright of the Department of Energy briefed reporters on Monday. He said the program will lower energy prices for consumers and that turkey on your Thanksgiving table will cost a lot less next year.

Paris Martineau [01:03:32]:
What? He thinks that this is going to. No, the first part. This is the first. Really.

Leo Laporte [01:03:37]:
The first part. Yeah. Yeah, he did say that.

Paris Martineau [01:03:39]:
Oh, boy.

Leo Laporte [01:03:41]:
Well, look, if this. Everything's got to be affordable. That's the goal. I think it's conceivable you'll have new materials. You'll have. I mean, who knows this. The.

Jeff Jarvis [01:03:52]:
The NBC story says that the platform will focus on providing researchers with computing power and data sets necessary to train AI models.

Leo Laporte [01:04:00]:
I think that's great. I mean, it is stuff we paid for.

Jeff Jarvis [01:04:04]:
Yep. And you're right, it was. Was siloed. And what, what comes. What comes through the, the Venn diagram of this stuff, when it comes together.

Leo Laporte [01:04:12]:
It would be. I feel better about. If they didn't call it the Genesis mission. It really sounds like they got plans.

Paris Martineau [01:04:18]:
You know, the planet explodes launching the Genesis mission.

Jeff Jarvis [01:04:22]:
Yeah, we can't live on Mars. Let's blow it up.

Paris Martineau [01:04:26]:
I don't love also that it begins with since the founding of our nation, comma.

Leo Laporte [01:04:30]:
Yeah, that's a bad start.

Paris Martineau [01:04:31]:
It's rough. It's a rough, rough start.

Leo Laporte [01:04:33]:
The order launches the Genesis mission as a dedicated, coordinated. I'm reading from the executive order. National effort to unleash a new age of AI, accelerated innovation and discovery that could solve the most challenging problems of this century. You know, you, you, you weren't here for the interview earlier with Ahmad Mostak. We taped that earlier this week. That's why Paris couldn't be here. She actually has a job. But one of the things he was really pushing and I, I agree with him, is maybe we don't need these giant AGI AI's owned by big tech companies.

Leo Laporte [01:05:10]:
Maybe smaller AIs could be useful, especially not.

Jeff Jarvis [01:05:17]:
Yep.

Leo Laporte [01:05:17]:
Not owned by the big tech companies, but, you know, he, he's, he's working on a medical, a small medical model that could be very useful. He says if you, if you make a specific task, it can do very well. It doesn't have to be AGI, it.

Paris Martineau [01:05:32]:
Doesn'T have to be general purpose.

Leo Laporte [01:05:33]:
It doesn't have to be reasonable. Exactly.

Jeff Jarvis [01:05:35]:
But then, but then you can't say it's smarter than all humanity.

Leo Laporte [01:05:38]:
Yeah.

Paris Martineau [01:05:39]:
Then you can't raise hundreds of millions or billions of dollars because no one wants to give you that much money for something that's limited use.

Leo Laporte [01:05:46]:
Exactly. I want to take a little break. When we come back, Johnny, I've and Sam Altman were on stage at the demo day for the Emerson Collective. This is Lorraine Powell. Jobs. Steve Jobs Widows. It's a non profit, but they do own quite a few things, including the Atlantic magazine. Johnny and Sam kind of gave us a little more information about.

Jeff Jarvis [01:06:09]:
Information would be stretching it, what they're working on.

Leo Laporte [01:06:13]:
We'll talk about that in just a little bit. Hey, you know, I'll be wearing it the minute it's available.

Jeff Jarvis [01:06:19]:
Now the question is, will you be tasting it? But go ahead.

Leo Laporte [01:06:23]:
Yes, we'll talk about that. You're watching Intelligent Machines. Paris Martineau, Jeff Jarvis. Wonderful to have you both. Day before Thanksgiving. You're. You're not cooking, Jeff?

Jeff Jarvis [01:06:36]:
No.

Paris Martineau [01:06:37]:
That would cause a house fire.

Jeff Jarvis [01:06:39]:
Well, no. When my, when, when my wife was pregnant with our first and I suddenly had to. To do some cooking, I damn near killed them both. Raw chicken.

Leo Laporte [01:06:48]:
You don't. So you don't have, like almost every family, you know, people have one thing that they like. We're gonna.

Paris Martineau [01:06:55]:
I make stuffing.

Leo Laporte [01:06:56]:
There you go. You see?

Paris Martineau [01:06:57]:
And I'm making a pineapple upside down cake this year.

Leo Laporte [01:06:59]:
But that's a wonderful, wonderful thing.

Jeff Jarvis [01:07:02]:
I make a drive to the takeout place.

Leo Laporte [01:07:05]:
I make a drive.

Jeff Jarvis [01:07:06]:
Did I Ever tell you the story.

Paris Martineau [01:07:07]:
What are you doing?

Jeff Jarvis [01:07:08]:
Of my pumpkin chiffon pie?

Paris Martineau [01:07:11]:
Oh, I think you did last year. And it was every year.

Leo Laporte [01:07:14]:
Oh, yeah, tells the story every year. It's that time of year.

Jeff Jarvis [01:07:18]:
Gramps is going to tell us about his book. No, I'm not going to. I won't do that. Look it up, folks. Go into the archives.

Leo Laporte [01:07:27]:
Hey, tell me your recipe for half.

Paris Martineau [01:07:29]:
Of a donut during this ad break. Listen.

Leo Laporte [01:07:31]:
Okay, but before you do that, tell us your recipe for stuffing that was.

Paris Martineau [01:07:35]:
Gonna be one of my picks of the week.

Leo Laporte [01:07:37]:
All right, all right, let's save it, save it, save it, save it. That's good. That's good. Tease it. I like it.

Paris Martineau [01:07:42]:
Stay to the end to hear my stuffing recipe that is so loved among friends and family that every year I get people text me, be like, what's that recipe again?

Leo Laporte [01:07:51]:
Oh, that's exciting. This.

Jeff Jarvis [01:07:53]:
I want to try oysters in there.

Leo Laporte [01:07:55]:
Oh, shh.

Paris Martineau [01:07:56]:
It's all oysters. Yeah, it's. It's actually, it's just oysters. And that's why I'm really surprised that people ask me for the recipe because I just. It's just oysters, baby.

Jeff Jarvis [01:08:06]:
Come on.

Leo Laporte [01:08:06]:
Why does the turkey taste like salt water? All right, we'll be back in just a moment. This episode of Intelligent Machines is brought to you by Zscaler, the world's largest cloud security platform. You know, it's really clear when you listen to this show that AI is for business, especially kind of of a double edged sword. On the one hand, it can make massive differences in your efficiency. It can, it can, it can give you all sorts of superpowers, a competitive advantage against, you know, guys who aren't using it. But on the other hand, you know who is using it? Hackers, bad guys. They're using it to attack more quickly. Every attack surface that you offer them, they will bombard with AI generated attacks.

Leo Laporte [01:08:58]:
The potential rewards of AI are too great to ignore, but so are the risks. And there's also the risk of you using AI and accidentally exfiltrating private proprietary company information. So generative AI increases opportunities for threat actors. They're using it, you know, for phishing mails that are indistinguishable from the real thing. They're using it to write malicious code. They're using it after they get in. They're actually using it to pick and choose and extract the data that they're going to steal from you. There were 1.3 million instances last year of Social Security numbers leaked to AI applications not intentionally by accident.

Leo Laporte [01:09:40]:
People innocently just using AI. SaaS, Apps, Chat, GPT and Microsoft copilot they saw nearly 3.2 million data violations in the same time period. This is, this is the time to rethink your organization's safe you use of public and private AI. Just check out what Shiva, the director of security and infrastructure at Zwara says about using Zscaler to prevent AI attacks. With Zscaler, being in line in a security protection strategy helps us monitor all the traffic.

Imad Mostaq [01:10:13]:
So even if a bad actor were.

Leo Laporte [01:10:15]:
To use AI because we have tight.

Paris Martineau [01:10:17]:
Security framework around our endpoint, helps us.

Leo Laporte [01:10:20]:
Proactively prevent that activity from happening. AI is tremendous in terms of its opportunities, but it also brings in challenges. We're confident that zscale is going to help us ensure that we're not slowed down by security challenges, but continue to take advantage of all the advancements. Thank you Shiva. With Zscaler Zero Trust plus AI you can safely adopt generative AI and private AI to boost productivity across the business. Zscaler's Zero Trust architecture plus AI helps you reduce the risks of AI related data loss and protects against AI attacks to guarantee greater productivity and compliance. Learn more@zscaler.com Security that's Zscaler.com Security we thank him so much for supporting intelligent machines. Did they teach you Paris? The TV anchors head bounce.

Leo Laporte [01:11:16]:
You probably wouldn't need this but when.

Paris Martineau [01:11:19]:
You'Re the anchor's headbound.

Leo Laporte [01:11:20]:
Well, you know, you know TV news is pretty simple. You got, it's a three camera shoot. You got your single, you got your two shot and then maybe you have.

Paris Martineau [01:11:28]:
Your terrifying to look just directly at the camera.

Leo Laporte [01:11:31]:
Don't do it.

Paris Martineau [01:11:32]:
Sorry.

Leo Laporte [01:11:33]:
Don't do it. You notice I, I look away because in real life you don't stare at somebody when you're talking to them. You don't.

Jeff Jarvis [01:11:40]:
That's the problem with the prompter is it looks like you're glued to it.

Leo Laporte [01:11:42]:
Exactly. That's why you need the news anchor head bounce. So you're reading the prompter on camera one and the floor director starts to send you to camera two. You can't just go because that would look weird. Fortunately, you have this fake news script that you've been holding all this time. And now you know why? Because you're reading, you're reading the prompter and you pretend to look down to read the news script and then you look up at that camera over there. Very useful.

Paris Martineau [01:12:12]:
I love that's just.

Jeff Jarvis [01:12:14]:
There's no Florida.

Paris Martineau [01:12:15]:
Is there any text on the fake news Script.

Leo Laporte [01:12:18]:
Yeah, they print out the script just in case. What it's really there for is if the prompter goes down. Because news anchors have no idea what.

Jeff Jarvis [01:12:25]:
Page was I on.

Leo Laporte [01:12:26]:
They have no idea what they're saying. So it is wonderful when the prompter goes down because it's, you know, clearly they go, they get this big, this big look wide eyed. And then yes, they go, where was I? Oh yes, here I am.

Jeff Jarvis [01:12:40]:
But when I've been on recently there's the floor director is gone.

Leo Laporte [01:12:43]:
Oh, they're gone now. In fact it's, it's the other day it's robotic cameras. Used to be there would be a director in the control room with a joystick for the robotic camera or it's pre programmed. The other day I saw an anchor. It must have been a mistake. She had the joystick next to her.

Jeff Jarvis [01:13:01]:
What is that the joystick for the prompter?

Leo Laporte [01:13:04]:
No, I think it was the camera. She was operating the camera. It was on our local channel dial.

Jeff Jarvis [01:13:10]:
A dial often for the prompter.

Leo Laporte [01:13:12]:
When they first put those. Yeah, well some of them are.

Jeff Jarvis [01:13:16]:
Well, in the old days there was a person whose job it was.

Leo Laporte [01:13:21]:
A camera operator.

Paris Martineau [01:13:23]:
One really strong guy pulled up a different stone tank tablet for each.

Leo Laporte [01:13:28]:
No, we don't use cue cards anymore. I don't know why the Tonight show still uses and Saturday Night Live still uses them.

Paris Martineau [01:13:34]:
Wait, they use.

Benito Gonzalez [01:13:36]:
Yes, they use it so they can have the eyeline be anywhere they want.

Leo Laporte [01:13:41]:
Ah, it solves that. And in other news, look, I, I got very good, I hate prompters, but I got very good at tech TV at looking away from the prompter, looking down, not, not reading the prompter. But if you, somebody doesn't know what they're doing, you could see their eyes moving back and forth. It's really terrible.

Benito Gonzalez [01:14:05]:
Up on the prompter thing on the cue cards.

Paris Martineau [01:14:08]:
I am, yes, I'm still, I.

Jeff Jarvis [01:14:09]:
Thank you.

Paris Martineau [01:14:10]:
I'm thinking, I'm like, why couldn't you put the prompter in different locations?

Leo Laporte [01:14:14]:
You could, you could in fact a.

Paris Martineau [01:14:16]:
Lot of times than having some guy with.

Leo Laporte [01:14:18]:
No, you probably can visit tech conferences.

Benito Gonzalez [01:14:20]:
They'll have a TV move also. You can't really do that with a prompter.

Paris Martineau [01:14:24]:
Could move, could it not?

Jeff Jarvis [01:14:25]:
Well, it's always where the camera is.

Leo Laporte [01:14:28]:
Usually it's on the cam. My prompter's on camera right here.

Paris Martineau [01:14:31]:
Things you see at speeches, which is like a little glass guy, that's the.

Leo Laporte [01:14:34]:
Prompter and they put it in front of the camera.

Benito Gonzalez [01:14:38]:
There's a, there's a big mechanism underneath that That's a projector projecting up to that.

Leo Laporte [01:14:42]:
It's a one way mirror.

Jeff Jarvis [01:14:43]:
Yeah, it's a mirror.

Leo Laporte [01:14:44]:
So there's a screen here and then there is a two way mirror. Not one way, two way mirror. It reflects the screen up here so the screen is actually backwards. The text on the screen. So it can bounce off and then you can read it. But anybody at this angle just sees right through it looks like a pane of glass. That's why those weird. Those prompter paints.

Jeff Jarvis [01:15:02]:
Paris in the old, old days. You take the the typewritten script and glue it together and then you had a kid who was the prompter operator who would put it under the camera and then. And then go decide the speed as you're reading.

Paris Martineau [01:15:18]:
I'm sure that glue was made of.

Jeff Jarvis [01:15:20]:
Like lead or something and probably and would piss off the hell out of an anchor if they were going too fast. Hey kids. Or too slow.

Paris Martineau [01:15:28]:
Slow it down. Hey kids. Speed it up.

Benito Gonzalez [01:15:30]:
A crew used to be like 25 people. A crew used to be like 25 people. Now it's like. Yeah, now it's like three. So like everybody's doing everything now.

Leo Laporte [01:15:38]:
When I left Tech TV I counted it was 22 people behind the camera, three people in front of the camera. And I decided when I started Twitter I wanted to reverse that. I wanted to have most of the talent be in front of the camera and have as few people behind. That's why we had 40 cameras so that we could literally didn't have to move them. Yeah, we just would take in a.

Jeff Jarvis [01:15:59]:
Big news control room. Now I haven't been in one in a while. Do you. Is the director still just snap fingers go to go three go.

Leo Laporte [01:16:06]:
Watched. I saw.

Jeff Jarvis [01:16:06]:
Yeah, that's still doing it.

Benito Gonzalez [01:16:07]:
Yeah.

Leo Laporte [01:16:09]:
Technical director who pushes the buttons and you have the actual director who's telling the tech.

Jeff Jarvis [01:16:13]:
Why does the director hit the buttons?

Leo Laporte [01:16:15]:
Because the director has other stuff to do nowadays. Probably in a lot of newsrooms, the director hits the buttons too.

Benito Gonzalez [01:16:20]:
But yeah, the director also has to tell the graphics guy when to throw graphics. There's a lot of guy to bring up three or whatever. Yeah, there's a lot to do.

Leo Laporte [01:16:28]:
Right?

Jeff Jarvis [01:16:28]:
But Leo, when you were reading prompter on a cheap operation, were you controlling the speed of the prompter?

Leo Laporte [01:16:34]:
No, we had a prompter after and that used to drive him crazy because I would not read the prompter in order. I wouldn't.

Paris Martineau [01:16:41]:
Wait, so is this still how teleprompter? There's some guy controlling this.

Jeff Jarvis [01:16:45]:
In a lot of cases, the news person Is doing it themselves.

Leo Laporte [01:16:49]:
They have a little knob or a pedal, so you have to be reading.

Paris Martineau [01:16:52]:
It, doing anything and controlling the speed.

Leo Laporte [01:16:55]:
Of your doing that.

Jeff Jarvis [01:16:56]:
Scrolling.

Leo Laporte [01:16:56]:
I'm doing that. I'm doing that with my mouse.

Paris Martineau [01:16:59]:
That's just a lot.

Leo Laporte [01:17:02]:
Yeah. When you do. Jeff, when you do cnn, do they.

Jeff Jarvis [01:17:06]:
Have prompter operator not visible in the studio?

Leo Laporte [01:17:10]:
They would be in the control room. They wouldn't be in the control room. So you don't know. But they. You don't see Nicole Wallace with a little knob or anything.

Jeff Jarvis [01:17:16]:
No, no, no, no, no, no, no.

Leo Laporte [01:17:18]:
I imagine msnbc, they still have that. Yeah, yeah. Do they have floor directors?

Jeff Jarvis [01:17:23]:
No, that's what I'm saying. They don't anymore. So I don't know how, you know, to look.

Benito Gonzalez [01:17:26]:
Tally lights.

Leo Laporte [01:17:27]:
Tally lights. Tally lights.

Paris Martineau [01:17:28]:
What are tally lights?

Leo Laporte [01:17:29]:
Lights. It's all right on top of the camera.

Jeff Jarvis [01:17:32]:
Yeah, but how do you. How do you get warning. But you know that it's, you know, you just.

Benito Gonzalez [01:17:36]:
You just see the light flash in the other side, and then you can do your head bob.

Leo Laporte [01:17:39]:
You look at the corner of your eye and you go, oh, there's the red light.

Jeff Jarvis [01:17:42]:
Look up.

Imad Mostaq [01:17:42]:
You.

Leo Laporte [01:17:43]:
You catch them off. You catch them a lot nowadays, not knowing.

Paris Martineau [01:17:47]:
We should all get, like, two extra cameras and then just throughout the show, just, like, pivot to one.

Jeff Jarvis [01:17:52]:
You know what I hate are the ones where they record these things and then for a while. What do you mean.

Leo Laporte [01:17:59]:
To the camera? What are you talking about? Oh, yeah, Wait a minute. Let me go to that camera. Let me go to that.

Paris Martineau [01:18:06]:
How is it? It's off now.

Benito Gonzalez [01:18:07]:
It's out of sync. It's out of sync.

Leo Laporte [01:18:09]:
They're out of sync. Oh, like when I switch like that, does it screw it up?

Paris Martineau [01:18:13]:
If we had people in that studio with you, it could have. It would work.

Leo Laporte [01:18:18]:
I have my own switcher. I'm doing it all now, baby. Yeah. We used to have a great floor director. He's now married to Erica Hill, the CNN anchor who will be doing the Thanksgiving Day Parade this year. He was great. He had a steep voice. And he would.

Leo Laporte [01:18:37]:
He was really good. Or Steve Porter was the other. He was really good. They would do. You know, you're on this camera. You're on this camera. You're on this camera. And then he'd go.

Leo Laporte [01:18:45]:
And then when it's time to change the camera, they'd go.

Jeff Jarvis [01:18:50]:
They were.

Leo Laporte [01:18:50]:
It was really fun. It was like dancing. It was so much fun.

Jeff Jarvis [01:18:54]:
Back in 30.

Leo Laporte [01:18:55]:
Yeah. Yep. In three, two. Those are the days, Jony. I've And Sam Allman on stage.

Benito Gonzalez [01:19:06]:
I need your screen back, Leo.

Leo Laporte [01:19:08]:
Oops.

Jeff Jarvis [01:19:11]:
Screen. Fun.

Leo Laporte [01:19:13]:
Jony, I've. And Sam Altman on stage with Laureen Powell, Jobs, Steve Jobs widow. She was interviewing them about what Johnny's doing with Sam. Remember that Little Love video that they made and all the. What was this $3 billion they're paying Jony, I've to do this thing. We didn't know much. We know a little bit more now. Altman described the design as simple and beautiful and playful.

Paris Martineau [01:19:43]:
Jesus Christ.

Leo Laporte [01:19:45]:
It's gonna be some sort of AI thing, you know, a necklace, ear.

Paris Martineau [01:19:51]:
No, no, no. We need to say what he added, which is there was an earlier prototype that we were quite excited about, but I did not have any feeling of, I want to pick up that thing and take a bite out of it. And finally we got there, all of.

Leo Laporte [01:20:04]:
A sudden, it probably looked like one of these delicious proteins bites. By the way, they look exactly like an AI pin. I've said I love solutions. Wait, let me do it right. I love solutions that teeter on appearing almost naive in their simplicity. And I also love incredibly intelligent, sophisticated products that you want to touch. I think I sound more like J.K. rowling than I sound like Johnny.

Leo Laporte [01:20:33]:
And you feel no intimidation. And you want to almost carelessly. And you want to use almost carelessly, that you use them almost without thought, that they're.

Imad Mostaq [01:20:41]:
They're just tools.

Leo Laporte [01:20:43]:
Altman says, I hope that when people see it, they say, that's it. To which I've responded, yeah, they will.

Jeff Jarvis [01:20:52]:
And they also said, Altman said, you want to lick it.

Leo Laporte [01:20:56]:
Yeah, well, licking. It's been around for a long time.

Paris Martineau [01:20:59]:
Did the fire marshal end up getting called for all the smoke that they're blowing?

Leo Laporte [01:21:05]:
Yeah. So it could arrive in, quote, less than two years, maybe.

Paris Martineau [01:21:10]:
This is like Elon Musk's word. Three to five years away from poll from San Francisco to New York.

Leo Laporte [01:21:18]:
But I do believe that they say they have prototypes. I do believe that Johnny is trying to design something.

Jeff Jarvis [01:21:26]:
They don't think it'll have a screen.

Leo Laporte [01:21:29]:
But did they know? Yeah, we don't know.

Jeff Jarvis [01:21:30]:
That's the thing. Yeah.

Leo Laporte [01:21:31]:
Yeah.

Jeff Jarvis [01:21:33]:
Perplexity has been kind of exposed as smoke now.

Leo Laporte [01:21:37]:
Oh, really? Tell me about that.

Jeff Jarvis [01:21:39]:
I think. Well, no, I just think in general, your attitude toward them. I think they do press releases, they do stuff.

Leo Laporte [01:21:43]:
So I think people are a little mad because you don't know what model you're getting. And there's some evidence that you're not getting the. The best models when you use so there's that, right?

Jeff Jarvis [01:21:52]:
And now Google has just leapfrogged and is the darling. Huge, huge like never before. And OpenAI is a humming, humming, a humming. With this and the whole Jony I've thing. Do you think there's a. OpenAI starts to.

Leo Laporte [01:22:09]:
Here's their disadvantage. Their disadvantage is meta. Google, Microsoft, Apple and Amazon all have revenue streams independent of AI efforts and.

Jeff Jarvis [01:22:20]:
People doing things in their real lives.

Leo Laporte [01:22:23]:
But OpenAI, it's this or nothing. And all the money comes from investors. And there's considerable pressure A to appease the existing investors and B, to get new investors because they are burning through money like crazy. Anthropic's in the same boat. Although Anthropic released a new model this week that many people are very impressed with.

Paris Martineau [01:22:44]:
So yeah, how have you guys been finding Claude? Opus 4.5.

Leo Laporte [01:22:50]:
You know, I use it for coding. Claude was already very, very good for coding. Not perfect. It would make dumb mistakes, but it also would do things mind boggling. This is where we are with AI right now, is that it's spiky. Like in some areas it's completely stupid and in some areas it's, it's brilliant. And you just have to be able to distinguish, I think as you as a user what you're getting, whether you're getting the stupid or the smart. But I think Claude4.5 is very impressive.

Leo Laporte [01:23:26]:
I think I'm most impressed with Gemini3, to be honest. And Nano Banana.

Paris Martineau [01:23:31]:
What have you been using it for?

Leo Laporte [01:23:33]:
Stupid stuff? No, lots of stuff. I basically. Well, we used it last week, remember. It's really good at synopsizing stuff. The image generation is incredible. Darren Okey said that one of the things he likes to do is have Claude code Anthropic's command line coder, solve a problem and then explain what it did and then feed the explanation to Nano Banana and have it do an infographic based on it, which he said is very helpful in understanding. I hope I'm not misquoting you, Darren, but it's helpful in understanding these concepts. I really feel like we are, we're making progress.

Leo Laporte [01:24:19]:
I don't think we're there. I don't know what the progress is towards even any more than I know what Johnny and Sam are planning. But I feel like we're making. They're working on stuff and we're making progress.

Jeff Jarvis [01:24:29]:
I went into it in Gemini 3. Somebody I know has a health issue and got labs and I went into it and put in the history and I put in the labs and it came back with incredible, incredible response. But then I found out that I was wrong about the medication the person was on right now. So I went back in and I said, does this change it? Oh yes, it does change it. And I did that twice where circumstances changed and it was pretty damned amazing. And I did that with also with ChatGPT and with Claude and Gemini was way, way more impressed.

Leo Laporte [01:25:05]:
You know, it's interesting to me and a number of people have observed this, that Google probably has even better. All these companies have better models internally, maybe even a year off.

Paris Martineau [01:25:16]:
Well, I'm sure better in some ways, more concerning in other ways.

Leo Laporte [01:25:19]:
Well, yeah, they don't want to release them until their safety's been proven and they've fine tuned it and so forth. You're right. But nevertheless, this isn't. Oh yeah, the sound is. I'm telling you, it's the lumberjack contest out back, back here. It's very noisy. That noise is somebody grinding something they're building.

Jeff Jarvis [01:25:38]:
I only heard it when you spoke.

Leo Laporte [01:25:40]:
Yeah, because it's skated out when I don't speak.

Jeff Jarvis [01:25:42]:
Okay, got it.

Paris Martineau [01:25:44]:
Wow.

Leo Laporte [01:25:44]:
I'll just shut up.

Paris Martineau [01:25:45]:
Speaking of Gemini, Darren Okey a while ago posted a screenshot in the chat of him asking Gemini. They sent photos of me, Jeff and Leo saying, these people all dressed up in high quality turkey costumes in a kitchen preparing a Thanksgiving dinner. Then Gemini responds, it looks like you're trying to create a funny scenario featuring the cast of this week in Google Podcast, Jeff Jarvis, Paris Martineau and Lisa Ford. As an AI, I cannot directly edit these specific uploaded photos to put them in costumes to prevent the creation of non consensual deep fakes.

Jeff Jarvis [01:26:20]:
Wow.

Paris Martineau [01:26:21]:
However, I can provide you with a highly detailed we give consent paste into an image generator to create this exact scene, which I thought was very interesting. They got us all.

Jeff Jarvis [01:26:31]:
We're famous.

Paris Martineau [01:26:33]:
We're famous guys.

Jeff Jarvis [01:26:34]:
Gemini knows us well.

Leo Laporte [01:26:37]:
That was what I mentioned last week was we've come so far. It used to be it say, well, I don't know anything about anything after 2024 or whatever. Now it seems to be completely up to date all the time. Come on. By the way, one of the things that Gemini does that's interesting, it has a much larger context window. I think it's a million tokens, which makes a big difference if you're feeding it. You know, that's where I think Google and Notebook LLM LM have done amazing work. If you're feeding in a lot of content, it can absorb quite a bit.

Leo Laporte [01:27:12]:
A million tokens is a Huge amount.

Jeff Jarvis [01:27:13]:
Of data and respond back and forth. Yeah, that's where the. That's where that iteration becomes really interesting.

Leo Laporte [01:27:22]:
And so you didn't get. You didn't get the picture of us. That's too bad.

Jeff Jarvis [01:27:27]:
There is one of us cooking together.

Leo Laporte [01:27:28]:
Is there back there?

Paris Martineau [01:27:29]:
Yeah, somewhere. It's back here somewhere.

Jeff Jarvis [01:27:34]:
Bye.

Leo Laporte [01:27:36]:
Just remember when even, you know, a year ago when we started doing this show. We've come, come so far. Oh, yeah, it's. All the fingers are there.

Jeff Jarvis [01:27:45]:
Pretty. Fly versus sky.

Paris Martineau [01:27:47]:
We all look terrible in this photo, but, you know, it's because we're dressed like turkeys. Actually, Jeff looks like he has a really full beard in this photo.

Jeff Jarvis [01:27:59]:
And a full stomach too.

Paris Martineau [01:28:01]:
I mean, we've all got full stomachs in this. I like that you're cutting the mask. Mashed potatoes.

Jeff Jarvis [01:28:09]:
I don't know how to cook.

Paris Martineau [01:28:10]:
I mean, that is. That is accurate.

Leo Laporte [01:28:12]:
Actually, if they're really thick, you might have to cut them. I'm looking for it. I have to.

Paris Martineau [01:28:20]:
I just. I just put it in the chat. I replied saying haunting. It's the third thing there.

Leo Laporte [01:28:25]:
Ah. Click to see. Attachment. Ah, here we. Oh, God, I look like I'm really old.

Paris Martineau [01:28:33]:
We all look rough in different ways.

Leo Laporte [01:28:36]:
That doesn't even look like you, Paris. I don't know who that is. Hey, listen, and I look like.

Paris Martineau [01:28:41]:
You don't get it. I don't like that. You don't get a turkey head.

Leo Laporte [01:28:46]:
Just the tail. Not because he's the boss. Yeah.

Jeff Jarvis [01:28:49]:
Respect, you know?

Paris Martineau [01:28:51]:
Yeah.

Leo Laporte [01:28:52]:
You know what's interesting, though? It got the right shirt.

Paris Martineau [01:28:55]:
I mean, I'm sure they uploaded a screenshot of some sort.

Jeff Jarvis [01:28:58]:
Yeah, yeah, just. Just did us. Yeah.

Leo Laporte [01:29:02]:
Now he does them with Photoshop. He doesn't do AI, so. Fashion tool Photoshopped that one.

Paris Martineau [01:29:10]:
Oh, this one's from last year.

Jeff Jarvis [01:29:11]:
Last year before. Oh, the book. Anymore.

Leo Laporte [01:29:14]:
Yeah. Oh, okay. Yeah, because it says twig.

Paris Martineau [01:29:17]:
Yeah, well, that's right.

Leo Laporte [01:29:19]:
But you know what? You look like you should be a puritan. Not you, Jeff. Paris, you fit right in. I feel like your name should be Harmony or something.

Paris Martineau [01:29:32]:
Chastity Harmony Lake.

Leo Laporte [01:29:34]:
Providence.

Paris Martineau [01:29:35]:
Yeah, Providence, Rhode Island.

Leo Laporte [01:29:39]:
They named a state after me, so. Fine. They named a state after me. All right, back to the news. It's so easy to distract us.

Paris Martineau [01:29:50]:
I think it'd be really funny to name an answer. Animal. Providence, Rhode Island.

Leo Laporte [01:29:55]:
Your next kitty. Prof. Providence. Providence, Rhode Island. Come here.

Paris Martineau [01:30:01]:
Right.

Leo Laporte [01:30:01]:
Did you read the Harper's article? Kicking robots?

Jeff Jarvis [01:30:05]:
No, the AI to summarize it.

Leo Laporte [01:30:08]:
Yeah, because it's Harper's So it's going to be a long story, but the idea really is that most of the robotics demos you're seeing, these humanoid robots are bs. They're not. I think we kind of know that. But there's so much hype and there's so much money floating around. Apollo's creator, and this is Apollo here, which is semi humanoid. It's got a lot of motors and actuators and things. Apollo's creator, the US startup Apptronic. Maybe you want to work on that name.

Leo Laporte [01:30:43]:
Might feed that to Gemini. Is a front runner. Call it Genesis. Is a front runner in this emerging industry. The company says it's building the first general purpose commercial robot, a machine that will one day be able to take on any type of physical labor currently performed by humans, whether it's cleaning houses or assembling cars. Yes, so he said. Not knowing what to believe, I traveled from London to Austin, Texas, to see Apollo for myself.

Jeff Jarvis [01:31:08]:
London. They couldn't have sent somebody nearer?

Leo Laporte [01:31:12]:
As I square up to Apollo in a Plexiglas arena, my first instinct is naturally to raise a foot. He's talking about kicking robots. But the kick test is too dangerous for visiting journalists, I'm told. Instead, someone hands me a wooden pole with a piece of foam taped around one end and mime's poking the machine in its chest. I think the scientific method in front of me. As various motors rev up to speed, the robot shuffles in place, looking like an arthritic boxer readying for a fight. On the other side of the plexiglass, a group of engineers chat casually with one another and glance over at a bank of monitors. One of them gives me a thumbs up.

Leo Laporte [01:31:51]:
Have at it. My first shove is hesitant because the prototype is worth, I've been told, a quarter of a million dollars. And while breaking it would make for a good story, it would also be the end of my visit. In response to my prod, the bot merely teeters. It's heavier than I expected. At £160, it feels like, well, a person. Oh, you could do it harder than that, says an engineer, and I jab forward again. I was channeling Burgess Meredith from Rocky.

Leo Laporte [01:32:20]:
Oh, you can do it hotter than that, Rocky, says an engineer, and I jab forward again. Nothing. Apollo is still trotting on the spot. Fine. I think I'll give it a real push. Drawing back, I grip my makeshift spear and strike the robot hard in the chest. It staggers backwards, stamping its feet, flinging its arms towards me in appealingly human gesture. I'm struck by a flash of involuntary alarm, whether out of sympathy for a fellow being or fear of an expensive accident, I can't say.

Leo Laporte [01:32:50]:
For a moment, the robot looks like it might fall. Then it regains its balance, returns to its position in front of me.

Jeff Jarvis [01:32:57]:
So, Ryan Lizzo, report.

Leo Laporte [01:33:00]:
I know. I like. See, I'm a sucker for that stuff that. That first person reporting. You know, I like it.

Paris Martineau [01:33:07]:
Watch 2025.

Leo Laporte [01:33:11]:
Oh, you writers, you're such snobs, huh?

Paris Martineau [01:33:15]:
That's our whole thing. I passionately explained what an M dash was to my parents in the car the other day.

Leo Laporte [01:33:22]:
Oh, how did they take it?

Paris Martineau [01:33:25]:
Yeah, they were like, M dash, like the letter M. And I was like, yes, but no, it's em. But it does stand for being the.

Jeff Jarvis [01:33:33]:
Width of an M. Did you do the en dash and the hyphen, or was that.

Paris Martineau [01:33:37]:
I did. We got. We got there as well.

Jeff Jarvis [01:33:39]:
Have you. Have you explained Nosy Lizza to them?

Paris Martineau [01:33:42]:
No.

Jeff Jarvis [01:33:45]:
Is that dinner conversation?

Paris Martineau [01:33:46]:
I don't know. I don't.

Leo Laporte [01:33:47]:
Save that for Thanksgiving. If it gets slow, you could really have a whole story.

Paris Martineau [01:33:51]:
It's really. It really is. I have explained this. I. Last night was catching up with an old friend who lives here, and I was like, oh, have you heard the latest nuzzy?

Jeff Jarvis [01:34:01]:
No, I have a life.

Paris Martineau [01:34:02]:
And she was like, what are you talking. We got. We got so deep that I was reading the RFK poem out to her.

Jeff Jarvis [01:34:10]:
Over in a public place. Paris?

Paris Martineau [01:34:13]:
Yeah.

Jeff Jarvis [01:34:13]:
Oh, no.

Paris Martineau [01:34:14]:
Yeah, yeah, yeah.

Leo Laporte [01:34:16]:
This is something Paris has been doing a lot lately, and maybe we need to do an intervention or something.

Paris Martineau [01:34:21]:
Maybe. Maybe. I don't know. We've got five more parts to do it, so. You know, there's. I heard they're supposed to. No, but I heard that the grapevine. There's eight.

Jeff Jarvis [01:34:31]:
Oh, my God.

Paris Martineau [01:34:32]:
Oh, I know.

Jeff Jarvis [01:34:34]:
Wow. Milking Bessie till she keels over in the field.

Leo Laporte [01:34:40]:
Yeah, but remember, he just started his substack and he gave the. The. This is brilliant. He gave the first one away with a big shocking twist at the end. And then you have to pay for all the rest of them. What he didn't understand.

Benito Gonzalez [01:34:54]:
Drug dealer model.

Leo Laporte [01:34:55]:
Yeah, it's the truck. But what he didn't understand is to the people who care about this, which is a very small, insular group that knows each other well. All that had to happen is one person pays for it, makes a PDF, and now all of us have it.

Jeff Jarvis [01:35:10]:
Welcome to Postscript.

Leo Laporte [01:35:12]:
And. And most of the people we've sent it to couldn't care less. Anyway, it's A long story and. But the general gist of it is we're a long way off from robots doing your laundry.

Paris Martineau [01:35:27]:
Yeah, yeah, that's the gist.

Leo Laporte [01:35:29]:
Yeah.

Benito Gonzalez [01:35:29]:
We already have machines that do that, by the way.

Leo Laporte [01:35:32]:
It's called.

Paris Martineau [01:35:33]:
They're called washing machines.

Leo Laporte [01:35:34]:
Yes.

Jeff Jarvis [01:35:36]:
Which were quite in advance in their day.

Leo Laporte [01:35:41]:
The last sentence is left alone in the spotlight. The robot walks to the edge of the stage, waiting for its handlers, unable to manage the stairs. I don't know. I thought it was good. I enjoyed it. Elon says work. It's not so good. You don't need to work.

Leo Laporte [01:36:03]:
I don't need to work is what he's saying. I don't really need to work. You probably should still keep working. Speaking at the US Saudi Investment Forum, Musk claimed that rapid advances in AI and humanoid robots will make human labor optional within 10 to 20 years. And that money will stop being relevant at some point in the future.

Paris Martineau [01:36:22]:
What.

Leo Laporte [01:36:25]:
For him, for him, for him is really the. Is really this missing clause.

Benito Gonzalez [01:36:33]:
But we're all just, we're all just NPCs, though, so, you know, I think.

Leo Laporte [01:36:37]:
He really believes that. In fact, increasingly. Did you read. And it's a little bit sensationalist, but did you read Douglas Rushkoff's piece in Fast Company, guys?

Paris Martineau [01:36:49]:
No, I don't know about what. What was.

Leo Laporte [01:36:51]:
So I will, I will try to synopsize it for you. It's called the Intentional Collapse. Actually, no, that's his substack. Let me find the. The one he wrote for Fast Company is kind of a rehash of the same thing. Remember we had Douglas on talking about how the rich were building these hidey holes in former Atlas missile shafts and, you know, New Zealand and underneath his mansion in Hawaii. His latest is. This is actually their vision of the future.

Leo Laporte [01:37:25]:
Let me find the thing. The people who are the Sam Altmans and the Mark Zuckerbergs of the world are actually planning for a catastrophic end. And this is the tesreal myths, kind of in a very dramatic way. He says they are intentionally. It's called disaster capitalism. Crashing the economy because they know it's going to happen anyway, so they're hurrying it up in order to have some control over what remains. The function of tariffs, for instance, is to bankrupt businesses or even public services in order to privatize and then control them. Put the ports out of business, let a sovereign wealth fund purchase the ports, bankrupt soybean farmers so they have to foreclose so that a private equity firm can buy the farmland, then hire the Farmers.

Jeff Jarvis [01:38:21]:
Besant calls himself a soybean farmer because he owns.

Leo Laporte [01:38:25]:
He's not actually getting his hands dirty.

Jeff Jarvis [01:38:28]:
He ain't tilling the field.

Leo Laporte [01:38:29]:
He says, you know, really what happened. What's happened in a capitalist society is that the rich have gotten farther and farther away from the products that are getting produced. So in their vision, and this is the part that I don't know, I'm curious what you think about. Rushkoff writes, it won't. In their vision, you know, remember, because one of the things I've always said is, but wait a minute, you cannot get rid of all the consumers because who's going to pay for your services? Or who's going to buy stuff from companies that pay for your services? He says, no, here's the weird part. In their vision, it won't be selling products to people, but selling stuff to the AIs themselves. He says, in today's economy, a small number of wealthy people and corporations employ us and sell to us, but they don't really need to care what species we are, whether we're human or Android, as long as we're producing value for their companies and then purchasing products from them. The humans don't matter in their long term.

Leo Laporte [01:39:35]:
As long as you can get robots and AIs to. To create the food you need, the clothes you need, the shelter you need. You don't need humans. We just get in the way.

Jeff Jarvis [01:39:45]:
But some, I don't know, there's consumption without consumption. There's no need for replacement. Replacement is what turns that cycle around. Even culture is consumed.

Leo Laporte [01:39:58]:
Well, and this is the problem. There is no culture if it's just a handful of billionaires eating food made for them by robots in their hidey holes under the ground, protected by security robots. This was always the thing that worried me about the silo missile silos. Who are you going to get to protect you? Oh, robots. This is why they want robots. Because robots aren't going to say, hey, by the way, you got the food down there? We're coming in anyway. It's an interesting piece. It's a little bit.

Jeff Jarvis [01:40:31]:
Douglas, but he's Douglas. Yeah, he does that.

Leo Laporte [01:40:35]:
Yeah, but it kind of makes sense in a way, because I've always wondered. Well, yeah, but you can't. If you're trying to replace humans and you're trying to. If you. As Elon Musk says, there's not going to be any need for money in the future and work is going to be optional. Well, then what we asked about universal basic income.

Paris Martineau [01:41:01]:
I think the potential reality of this is that these people just aren't really thinking about that because they don't care about the answer.

Leo Laporte [01:41:08]:
I. I think they're smart enough to think about it. And I think this is that they have a plan.

Jeff Jarvis [01:41:13]:
Leo, it's made of humans.

Leo Laporte [01:41:15]:
Soylent Green is. Yes.

Jeff Jarvis [01:41:17]:
Yeah, we just become slop.

Leo Laporte [01:41:21]:
What if you can replace humans with AIs and robots and machines?

Jeff Jarvis [01:41:25]:
Well, that's what, that's where they head. That's when we talk about, then you don't need us. When Bostrom and company talk about the 10 to the 54th future beings, they don't mean human beings, they mean AIs. They mean computer.

Leo Laporte [01:41:39]:
That's what, that's what Larry Page said. He said, you know, don't be speciesist. We're generating the next evolutionary step and it's not us.

Paris Martineau [01:41:49]:
It all comes. Comes back to computronium.

Leo Laporte [01:41:53]:
Well, the one thing I like about what Ruth Cough says is there is, there is something to be optimistic about, which is they don't care about us. So we can now create a small economy of barter where we work and help each other, where we make things for each other. He said this started in the Middle Ages. He said in the Middle Ages people were becoming more prosperous because they would take their products to market, they would trade, they would barter. He says people were taller in the Middle Ages than any time before the 1980s because there was plenty of food. And then at some point the lords of the manor realized, this is terrible. They're not working for us, they're working for each other. And they made it illegal to be self employed on penalty of death.

Leo Laporte [01:42:42]:
You had to work for somebody. You, you were not allowed to own land. Only the nobles are allowed to own land. That's where we're headed. Right. But he says, but, but eventually they won't care about us so we can have our own little small economies.

Jeff Jarvis [01:42:56]:
Well, the same is true of culture. I mean, I hope we get there as well that when scale and mass don't matter anymore, that we can create the things that matter to whoever we care about.

Leo Laporte [01:43:07]:
Yeah, I mean, that's just a little bit of. Have optimism about a very dystopian future. I don't know if either of those things is going to happen.

Jeff Jarvis [01:43:17]:
Palantir working for Fox is the dystopia of the day.

Leo Laporte [01:43:21]:
More. More relevant to the current situation.

Jeff Jarvis [01:43:24]:
Yeah.

Leo Laporte [01:43:26]:
Elon wants to replace his staff. He's fired now, half again this month. Half of the engineering team he'd already cut it down from. This is the safety Team. Not the engineering team, the safety team. He'd already cut it from 100 people when he acquired Twitter to less than 20 people. He's now laid off half of them. The layoffs.

Leo Laporte [01:43:51]:
This is from the information. The layoffs reflect Musk's desire to replace as much of X's engineering systems as possible with AI, with Grok, with Grock.

Jeff Jarvis [01:44:04]:
Because Grock is there to tell him how beautiful he is, how amazing he is. Yeah, that's Grok's job.

Leo Laporte [01:44:09]:
So, I mean, this is why Elon Musk says, well, you don't have to worry about money, you don't have to worry about jobs because there's not going to be any. I'm gonna have AI do it.

Jeff Jarvis [01:44:19]:
I'm so tired of these boys.

Leo Laporte [01:44:22]:
Well, they're not on our side. I guess that's pretty obvious.

Paris Martineau [01:44:30]:
Should we talk about the Grok incident this week?

Leo Laporte [01:44:32]:
What was the Grok incident this week.

Paris Martineau [01:44:36]:
That everyone. People started to realize over the course of last week that if you asked Grok to a question about who is, let's say, one of the smartest people, who would you say the smartest person in the world? World is? It will say Elon Musk. If you ask about, I believe someone asked who in between Elon Musk and like a world internationally ranked competitive basketball player would be best at playing basketball. Yeah, basically. And they were like, Elon Musk, of course.

Leo Laporte [01:45:15]:
Asked about his. This is from the Washington Post. Asked about his intellect, appearance and accomplished accomplishments, Grok consistently hailed Musk as, quote, strikingly handsome, extolled his, quote, lean athletic physique, what raved about his genius level intellect, and ranked him as the number one human ahead of Leonardo da Vinci.

Paris Martineau [01:45:36]:
Asked by one user who would win in a fight between Musk and the legendary boxer Mike Tyson, Grok didn't hesitate. Elon takes the win through grit and ingenuity.

Leo Laporte [01:45:47]:
Now, Elon's response to this is it's being manipulated by adversarial prompting. But they aren't showing the prompts, are they?

Paris Martineau [01:45:56]:
I saw the prompts. The prompts were people were using this on Twitter, on X, the Everything app, adding Grok and saying things such as grok in between LeBron or in between Peyton Manning and Elon Musk. Who would be the top pick, in your opinion, for the 1998 NFL Draft? And it would say Elon Musk.

Leo Laporte [01:46:22]:
So clearly Elon put some extra instructions or somebody. It may not be Elon. You know, maybe somebody likes.

Jeff Jarvis [01:46:28]:
I wonder, did anybody put in a question like, who was Jeffrey Epstein's best friend.

Paris Martineau [01:46:34]:
I'm not sure that it. I think in certain cases, I'm trying to think there was one that was something quite rude where it was like, who would be the absolute best in the world, the number one champion of consuming urine or something like that. And it would say Elon Musk. But perhaps in ruder terms.

Benito Gonzalez [01:46:58]:
It's a.

Jeff Jarvis [01:46:58]:
Model built for one.

Benito Gonzalez [01:46:59]:
It's a model built for one person.

Paris Martineau [01:47:01]:
And. Yeah.

Leo Laporte [01:47:02]:
Does. Does anybody use Grok besides Elon Stans?

Jeff Jarvis [01:47:07]:
No, I refuse to.

Leo Laporte [01:47:09]:
Yeah, I don't use it. I. I have it, by the way, because I have the involuntary blue check. So I have Grok Pro, but I never use it. You saw. You saw what happened the last time I used it.

Paris Martineau [01:47:20]:
We had to cut it out of the show.

Jeff Jarvis [01:47:22]:
The show.

Leo Laporte [01:47:22]:
So, so bad.

Paris Martineau [01:47:24]:
Should we fire it up right now?

Leo Laporte [01:47:27]:
I don't think it's doing that anymore. Because as soon as somebody tweets this kind of stuff, you know, Elon or somebody goes in and fixes it. Right.

Paris Martineau [01:47:37]:
Perhaps, yes. Unfortunate or fortunate, depending on how you look at it. Yeah, there were some kind of wild examples, if I recall correctly.

Benito Gonzalez [01:47:48]:
There was also that photo circulating with all the tech bro billionaires, like, hanging out.

Leo Laporte [01:47:53]:
Yeah, yeah. At like a frat rave. Yeah.

Benito Gonzalez [01:47:56]:
And they were all such in good shape.

Paris Martineau [01:48:01]:
Someone asked Grok, what's Elon's record in no Rules street fighting. No formal records exist for Elon's informal street scraps in youth, as they weren't tracked, but survival clashes amid South Africa's violence. He's recounted several beatings yet knocking out aggressors to end harassment. His specialties. These unyielding encounters built resilience, not stats. Echoing his ethos of finishing what starts, whether fights or frontiers.

Leo Laporte [01:48:31]:
Here's one of the pictures. There's quite a few of them. This is the Trillion Squad assembled with Elon Sundar Pichai, Tim Cook, Jensen Wong, Satya Nadella, Mark Zuckerberg, Jeff Bezos, and Sam Altman.

Benito Gonzalez [01:48:45]:
I want to know, in scenario, yes.

Leo Laporte [01:48:46]:
They all look quite fit.

Benito Gonzalez [01:48:47]:
Who's taking the show? Who's taking the photo in this scenario?

Leo Laporte [01:48:51]:
Oh, yeah, somebody not. Not a billionaire, I guess, or Trillionaire. Here they are shopping. They're all wearing the same kinds of shoes. I don't know.

Paris Martineau [01:49:08]:
There's also been a number of examples where people would say. Where they would ask rock an opinion on historical theories and in one context would say the theory came from Elon Musk, and then the other ask the exact same question, but say the theory came from Bill Gates. And the answers were different.

Leo Laporte [01:49:29]:
Yeah.

Jeff Jarvis [01:49:29]:
I asked, what is Elon Musk's greatest talent? Elon Musk's greatest talent is his ability to. And this is in bold turn, audacious, almost science fiction level goals into functioning companies that actually deliver products and reshape industries. That's how I do it repeatedly across completely different domains.

Leo Laporte [01:49:47]:
So, Paris, would you like to be at this party?

Paris Martineau [01:49:52]:
Oh, God. The thing that's so funny to me about this photo is you've of course, got all the tech boys there and then behind them, it's like a sea of 18 to 20 year olds.

Leo Laporte [01:50:05]:
Yeah. It's a frat party.

Paris Martineau [01:50:06]:
Yeah, I know. It's a very funny juxtaposition.

Leo Laporte [01:50:12]:
And somehow I don't think Elon drinks bud. But maybe, maybe. I don't know. I think we like seeing pictures like this because it makes him feel like. Makes us feel like they're. They're like us. They're real people. But this is all nano banana, obviously.

Paris Martineau [01:50:28]:
Yeah. Because, you know, you two spend so much time at frat parties, it must just be very comforting to see the. The science of tech business in. In environments that yourself frequent so often.

Leo Laporte [01:50:40]:
Here is an Iranian conference where they have what they claim to be robots. They're just people dressed in robot suits, actors. It's like a cheap science fiction movie. I. I won't turn on the sound.

Paris Martineau [01:51:05]:
But I love that it has ones and zeros.

Leo Laporte [01:51:08]:
Yeah. Painted to its face. Yeah. This is at a 2025 Kish Inox Tech Expo in Iran to highlight Iran's tech rise. No, this is probably not. This was a performance. There was no claim they were actual robots. According to Grok says it can beat League of Legends.

Leo Laporte [01:51:32]:
Right.

Benito Gonzalez [01:51:35]:
What does that mean? League of Legends is a PvP game. So you're playing against other people.

Paris Martineau [01:51:38]:
Yeah. How does that mean?

Leo Laporte [01:51:39]:
How do you win against humans?

Paris Martineau [01:51:42]:
But all of them. I mean, I could probably win against a human on League of Legends.

Leo Laporte [01:51:46]:
Oh, I think they mean the best. The very, very best.

Paris Martineau [01:51:50]:
Really?

Leo Laporte [01:51:51]:
Yeah.

Paris Martineau [01:51:52]:
I press X to doubt, as the kids say.

Jeff Jarvis [01:51:55]:
Do they say that?

Paris Martineau [01:51:57]:
They do, yeah.

Leo Laporte [01:51:57]:
Press X to doubt. Does it refer to x.com or is it a button that has an X on it?

Jeff Jarvis [01:52:03]:
It's a.

Leo Laporte [01:52:04]:
Okay.

Paris Martineau [01:52:04]:
You can Google press X to doubt. It's a meme from the game La Noire, where you had to press the X button in order to doubt. I know.

Leo Laporte [01:52:17]:
Really deep.

Paris Martineau [01:52:17]:
But, you know, I'm shocked that you didn't get this. You're usually down with me here.

Jeff Jarvis [01:52:20]:
Or it's a on the Xbox 360 or similarly.

Paris Martineau [01:52:25]:
Well, it's similar to the press F to pay respect.

Benito Gonzalez [01:52:29]:
I mean, that's just a really old game. That's, like, from 2011 or something.

Leo Laporte [01:52:34]:
Well, I love it. Paris. Without Paris, I wouldn't know what the kids are talking about these days.

Paris Martineau [01:52:39]:
Do you guys know press F to pay respects? I've had to have explained this in the show before, right?

Jeff Jarvis [01:52:44]:
No.

Paris Martineau [01:52:45]:
Oh, my God. I don't even.

Benito Gonzalez [01:52:48]:
I think I have a feeling that at some point, Paul Thurot explained that to you, Leo, on Windows, because he's a heavy player.

Paris Martineau [01:52:55]:
This is not as old as press X to doubt, but it's from Call of Duty Advanced Warfare, where there was a cutscene where the player character is attending a funeral and comes up on the screen. You have to press the F button to pay respects. And so now people on the Internet for the last. God, I guess, 11 years, when someone's done something embarrassing and you think like, oh, they're basically dead, you. You reply F because you're paying respects.

Benito Gonzalez [01:53:28]:
Sometimes it's also very sincere. Like, people really say that sincerely sometimes.

Jeff Jarvis [01:53:33]:
Paris, do you think your actual father knows this as opposed to your fake dads?

Paris Martineau [01:53:37]:
No, he definitely does not. He would probably pretend for maybe 25 seconds that he understood, but then I'd catch him in a lie.

Jeff Jarvis [01:53:44]:
Okay.

Leo Laporte [01:53:47]:
Let'S take a little break. You can pick some more stories, if you will, while I go to work here pretending to read a teleprompter. We didn't. In the early days of Twitter, I did not want teleprompters. And one of our hosts, who shall remain Naez, convinced me that I had to have teleprompters. We bought them for her or them. And then. And then they said, and where's the tally lights? How am I supposed to know what camera I'm on? And so we're not doing tally lights.

Leo Laporte [01:54:18]:
I know what camera I'm on because I'm pressing all the buttons. That's how I know we will have more with Paris Smart. No, Jeff Jarvis, the media savvy. Look at that. Look at that. That's a media savvy gesture. Paris Smartno.

Paris Martineau [01:54:31]:
This is me being savvy with the.

Leo Laporte [01:54:33]:
Media at home in Florida. What is that bobblehead over your left shoulder there? What is that all about? Is that another real estate award or.

Paris Martineau [01:54:42]:
That you recently had this in his. My dad has always had this in his office.

Leo Laporte [01:54:45]:
Is this the one where you hit the head and it says something?

Paris Martineau [01:54:48]:
No, that's a different one. That's my mother's office. Look, he's got a cute little butt.

Leo Laporte [01:54:53]:
Oh, my God.

Paris Martineau [01:54:54]:
He's. He's stacked. He's got a caked up as the kids.

Leo Laporte [01:54:59]:
He's caked up. Her dad has a caked up statuette in his office.

Paris Martineau [01:55:07]:
A caked up little football man.

Jeff Jarvis [01:55:09]:
And have you, have you ever asked your father what the genesis, so to speak?

Paris Martineau [01:55:15]:
He's had this since I was like, I'd say eight or something. It's definitely survived.

Leo Laporte [01:55:21]:
So you grew up with caked up football man.

Paris Martineau [01:55:23]:
I did grow up. I mean it's perhaps where I get my possessions.

Leo Laporte [01:55:28]:
That's where you get your possession.

Paris Martineau [01:55:29]:
Being in the presence.

Leo Laporte [01:55:30]:
That's where she gets her possession. Yes. Ladies and gentlemen, Jeff Jarvis, also here. His office is much more tame. No caked up little football men there. You have any statuettes? Oh, you do. Look at that. There's one of him.

Leo Laporte [01:55:44]:
Where'd you get that from?

Paris Martineau [01:55:45]:
How's the butt? Can we see?

Leo Laporte [01:55:47]:
Let's see if it's caked up. Yeah.

Paris Martineau [01:55:49]:
Oh, so chill.

Leo Laporte [01:55:50]:
Well, but that jacket flap is rising suspiciously.

Paris Martineau [01:55:54]:
There could be a hidden.

Leo Laporte [01:55:55]:
There's some cake in there. We got cake. All right. There is that. Where'd you get that, Jeff?

Jeff Jarvis [01:56:02]:
At Darwin's Circle A conference in Vienna. They had the cat. You stood there and the camera went around.

Leo Laporte [01:56:08]:
Oh, that's cool.

Jeff Jarvis [01:56:09]:
And that's very cool. Out it came. So. Yeah, it's even. Even. They even painted the little face.

Leo Laporte [01:56:14]:
Wow.

Jeff Jarvis [01:56:15]:
Focus on it.

Leo Laporte [01:56:15]:
See, that's much better than having a nano banana created drawing of you. That's, that's, that's a real figurine. That's cool. Yeah. Our show today, brought to you by Spaceship. We've been talking about Spaceship for a while now. I hope you've checked them out. There's a good reason for Spaceship success.

Leo Laporte [01:56:35]:
They are. That is. Well, just look at the site. It's modern, it's fresh. They have just passed a major milestone. Brand new. They have 5 million, over 5 million domains under management. Now, that does not happen by accident.

Leo Laporte [01:56:49]:
It happens because people go there and they say, wow, this is where I want to register my domains is where I want my website hosted. This is where I want my business email. Spaceship delivers real quality and features that just make sense not just for domains, but everything that helps you build and run your online presence. That means hosting, it means great business email, even tools for creating and managing web apps all in one straightforward platform. And when you transfer your sites over, you get some great benefits. You really do. I want you to check it out. Spaceship.comTwit Essentially it's black Friday and Cyber Monday level value all year round.

Leo Laporte [01:57:30]:
You don't have to wait till the day after Thanksgiving to get a great deal@spaceship.com TWiT and by the way, Twit listeners, you get additional exclusive offers that make it even better if you go to that site. Spaceship.com TWIT I've moved stuff over there because it saves me money and I love the interface. It's so easy to change DNS. Somebody emailed me. I've been using Spaceship's messaging platform. They call it Thunderbolt. And it's, I think it's really cool. It's end to end encrypted messaging, you know, like signal or anything else.

Leo Laporte [01:58:02]:
But it uses your domain name as your address, which I think is really cool. Great for businesses, right? You could, you know, make your business name be your messaging address. But I said, all right, I want to create a URL just for that messaging. So I, I went there for $5 a year. I bought Leo's IM because it's my instant messenger, right? Leos IM. If you, if you message me on Thunderbolt. Leos IM but the other day somebody mailed me and said there's no web address there, no website there. I said, well, no, it's, it's a messaging.

Leo Laporte [01:58:36]:
And I thought, well, that's crazy. I should really point it. He said, you should point it to your website. I said, all right. Two seconds later, Spaceship had done it. I just went there, pressed the button. They created an SSL certificate for it. So the transfer is secure.

Leo Laporte [01:58:51]:
I mean, it's, it's easy, it's done right. Everything's very thoughtful. So whether you're planning a new online project, whether you're moving an existing one, Spaceship has everything you need to get launched, connected and running smoothly. And it does it at a price that's out of this world. Maybe that's why they call it spaceship. Spaceship.com TWIT check it out to see the exclusive offer and find out why millions have already made the move. That spaceship.com twit and if you want, you can message me. Get Thunderbolt.

Leo Laporte [01:59:20]:
When you, when you, you know, register over there, any domain, you can make it be their Thunderbolt domain and message me at LEOS im. I'll promise I'll respond back. Thank you, Spaceship. So the professor crashed. This has got to be from you, Jeff, because it's in German.

Jeff Jarvis [01:59:46]:
Yeah, that was for me. So, well, but no, the, the main story is in English, right? Isn't it? The first AI just used Google Translate. That's mine.

Leo Laporte [01:59:56]:
Yeah. At Akkad University there's a digital professor. It's a clone of a Latvian researcher. This is the clone. Professor Walters, a digital twin from Professor Walters. Kaj from Latvia. But he crashed in his first. His first lecture.

Jeff Jarvis [02:00:15]:
Yeah.

Leo Laporte [02:00:18]:
Oh, well.

Jeff Jarvis [02:00:20]:
30 minute lecture featuring international business Administration and marketing marked the official launch of the interdisciplinary research project, a milestone in AI research, cooperation between humans. Yeah.

Leo Laporte [02:00:34]:
So actually now here's another story from the New York Times Magazine. I'm a professor. AI has changed my classroom, but not for the worst.

Jeff Jarvis [02:00:43]:
This is a smart piece. This is a professor who. I can't remember where the professor was a professor, but is bc He's a bc Right. Acknowledging that AI is there and how to work with people. And it's not all just blue books.

Leo Laporte [02:00:56]:
He makes it more human, focuses on the human.

Jeff Jarvis [02:01:00]:
Right, right, right. And you know, he said he starts off with a neat thing. A student comes and says, you know, how do we start? How do we ask our own questions? You always ask us the questions. How do we start to ask our own questions? Which is kind of a great way to look at AI. It's your chance to ask questions of the AI.

Leo Laporte [02:01:17]:
Right. He says, I don't teach. Actually, I think this guy is an unusual professor. I think he's, he's, he's probably a really good professor. He says, I don't lecture much. Mostly we engage in conversation, paying attention to one another, and through the book we've all read. Wow, how innovative. I don't teach content so much as a way of coming at things.

Leo Laporte [02:01:42]:
Tools and moves we can use to extract meaning from the world around us and make well supported arguments about what we find. See, if you get a professor like this, you're very lucky. This is good, right? This is who you want.

Benito Gonzalez [02:01:52]:
The problem with this stuff though is that learning this way probably perform worse at like standardized tests.

Leo Laporte [02:02:00]:
Of course you do. And once you get on the factory floor, it doesn't help you at all.

Paris Martineau [02:02:04]:
Well, what sort of standardized tests are you doing in college or to get.

Jeff Jarvis [02:02:08]:
To law school, but.

Leo Laporte [02:02:09]:
Or med school.

Benito Gonzalez [02:02:11]:
Any kind of technical school.

Paris Martineau [02:02:12]:
You're not having, you're not having a.

Jeff Jarvis [02:02:14]:
You know, you know enough.

Paris Martineau [02:02:15]:
You're not getting a college, you're not having a college. Taking classes in college to perform on standardized tests.

Leo Laporte [02:02:20]:
I agree. This is college. This is the classic liberal arts education. Learning how to learn as opposed to.

Jeff Jarvis [02:02:26]:
Yes, right, right.

Leo Laporte [02:02:27]:
Collecting facts.

Jeff Jarvis [02:02:28]:
Amen.

Leo Laporte [02:02:30]:
We need more of that. So he Says AI is, is helpful people who rely on AI to do their work. You know, he says, I've read all the headlines. Everybody's cheating their way through college. Nobody's going to ever read a book again to write a paper again. He says, no, that's not the case if, if you handle it. He says, well, in the semester it came had come as a pleasant surprise to me and many of my fellow teachers. The AI apocalypse we expected to arrive in full force this fall, bringing an end to reading and writing in school as we know it has not come to pass.

Leo Laporte [02:03:05]:
Just. It's maybe forcing teachers to be a little bit more human and not to focus on the tests and the facts he talks about.

Jeff Jarvis [02:03:20]:
Well, ironically, one professor he quotes rather than having them write kind of essay answers, he does small quizzes with trivial questions just to make sure they read the thing right. Okay, low stakes, but you got to read it because you got to know that MacGuffin is.

Leo Laporte [02:03:39]:
Especially if you do stuff that the AI is not going to highlight, you know, it won't make in the bullet point because it was so trivial and stupid.

Jeff Jarvis [02:03:47]:
Right.

Leo Laporte [02:03:49]:
That's one way around it. I don't know if that's the best way around it, but that's one way around. And I like the idea of making people think you can use AI all you want, but if you, if you, if you challenge students to think all the AI on the world's not going to help them. They still have to think. And that's the muscle you're trying to build anyway, isn't it? I don't know. I haven't been in school in a long time and I was a failure at it.

Benito Gonzalez [02:04:11]:
So I mean I read about a professor who told his students to ask the AI to write an essay and then the students have to fact check it. So like that's.

Paris Martineau [02:04:20]:
Yeah, I have heard about that.

Leo Laporte [02:04:21]:
That's a skill.

Paris Martineau [02:04:21]:
I think that's a useful skill.

Leo Laporte [02:04:23]:
You could do that once a year.

Paris Martineau [02:04:25]:
As a first assignment. I think it's useful because.

Leo Laporte [02:04:28]:
Exactly.

Paris Martineau [02:04:28]:
They get acutely. They're intimately aware of how many things the AI can get wrong.

Leo Laporte [02:04:35]:
Probably be done in high school or even grade school nowadays just to get that through their heads before they get too far.

Benito Gonzalez [02:04:45]:
Yeah, that should probably be primary. Like fifth graders should be learning that.

Leo Laporte [02:04:49]:
Yeah, exactly. It's critical thinking. That's critical thinking skill now. Right. Somebody fed the Declaration of Independence to a AI detector. Let me see if I can find the story came out it said 98% probability this was written by AI I.

Paris Martineau [02:05:10]:
Think that's a perfect rebuttal to all those freaks that say it's. Those things are foolproof.

Leo Laporte [02:05:16]:
Yeah. Your text is AI GPT generated. Actually, it was 99.99% when in the course of human events. You know why? I can tell you why.

Jeff Jarvis [02:05:29]:
Every single AI was trained on it.

Leo Laporte [02:05:31]:
Yeah, yeah, It's. It's exactly the. The text that, you know. Although you think it'd be smart enough, somebody. Somebody read it says, crazy. Even the founding fathers used AI to cheat on their homework. Crazy, crazy, crazy. But this is the problem is that these, these tools are being used by professors who are trying to fight AI, but they don't.

Leo Laporte [02:05:59]:
They don't work that well.

Jeff Jarvis [02:06:06]:
The. The robotic lawnmower that devastated the sports field.

Leo Laporte [02:06:11]:
Look at this picture. Look at this picture. Oh, my God. Again, another German. You read a lot of German stuff?

Jeff Jarvis [02:06:17]:
Yeah.

Paris Martineau [02:06:19]:
How do you say lawnmower in German, Jeff?

Jeff Jarvis [02:06:21]:
Well, that's a good question. I don't know. Let's see. Translate from English to German, Robbie.

Paris Martineau [02:06:29]:
Is his name.

Leo Laporte [02:06:33]:
Rosenweg? It's a Marobat. Ah, the Sportsplatz is all messed up. It's muddy. It's a muddy Sportsplatz. I actually had this Robomower. I recognize this Robomore. Right?

Paris Martineau [02:06:52]:
You had this robot.

Leo Laporte [02:06:53]:
No, I was reviewing. I was trying to review it, but our land at the time, this was a few years ago, was so hilly, I never got to review it because it couldn't get up the hill. So I'm glad I didn't, now that I see what it's done to this Sportsplatz.

Paris Martineau [02:07:10]:
Are you gonna have people over for Thanksgiving when you don't have a wall?

Leo Laporte [02:07:15]:
Yeah, I said stucco.

Paris Martineau [02:07:16]:
You're dining al fresco.

Leo Laporte [02:07:18]:
No, we have a wall now. There's been. Stucco has been applied.

Paris Martineau [02:07:21]:
Wow, that's huge.

Leo Laporte [02:07:23]:
And good news. They came over and said, because it's Thanksgiving, we're going to take the plastic covering that we have had over all of your windows for the last two weeks. We're going to take it down just for the living room room so you can see.

Paris Martineau [02:07:36]:
Wow, how generous.

Leo Laporte [02:07:40]:
But we're gonna put it back on Friday.

Jeff Jarvis [02:07:42]:
Are they.

Paris Martineau [02:07:42]:
Their sunlight's going away.

Leo Laporte [02:07:44]:
Sorry.

Jeff Jarvis [02:07:45]:
Are they stuck away in the whole house?

Leo Laporte [02:07:46]:
Just.

Jeff Jarvis [02:07:47]:
No.

Leo Laporte [02:07:47]:
There's the south wall, but that's the biggest wall. It's a three story wall. Paris. Are we keeping you up, young lady? In your media training, didn't they tell you not to yawn during.

Paris Martineau [02:07:56]:
They probably did. I'm sorry, I. I'm teasing you. Whenever I come Come back to Florida. I just get sleepy, McGee.

Leo Laporte [02:08:03]:
Yeah, because it's boring at home and you want to.

Paris Martineau [02:08:07]:
It's a boring state and no, I'm going to early morning workout classes with my parents and I'm just tired by the time like six rolls around.

Leo Laporte [02:08:15]:
Oh, that is so sweet. You go to the classes, workout classes with your parents.

Paris Martineau [02:08:19]:
It's true. They. They're the ones that got me into Orange Theory last Thanksgiving. So now we all go together. It's cute.

Leo Laporte [02:08:25]:
Isn't that sweet? The family that sweats together has spitz together.

Paris Martineau [02:08:33]:
Both of those things are true. We should be schwitzing.

Jeff Jarvis [02:08:36]:
What do you do at Orange Theory?

Leo Laporte [02:08:37]:
Oh, that's a tough one.

Paris Martineau [02:08:39]:
It's a hit, a high intensity interval.

Leo Laporte [02:08:42]:
They blow a whistle and you have to move to the next thing.

Paris Martineau [02:08:44]:
There's like a couple different. You start off like on the treadmill and you do, let's say like that for 20 minutes. And they'll tell you different, like instructions. It'll depend on the day. There's like a different, different class format. Then you go to the rowing machine, then you have like a workout there, then you go on the floor and you do like strength based workouts with certain weights.

Leo Laporte [02:09:02]:
But you move around and you're working hard too. It's not.

Paris Martineau [02:09:05]:
You're working hard and you're always being told to do something different. So I like it because you can just go and show up and not think for an hour. I also think it's very fun.

Jeff Jarvis [02:09:14]:
I'd be thinking how miserable I am.

Paris Martineau [02:09:16]:
That's true.

Leo Laporte [02:09:19]:
Floridians watching in the Twitch channel, by the way.

Jeff Jarvis [02:09:22]:
Sorry.

Leo Laporte [02:09:23]:
Florida is a boring state, guys.

Paris Martineau [02:09:25]:
Florida is a boring state. I.

Leo Laporte [02:09:26]:
People from all over the world come here to vacation and do all sorts of activities. You guys don't know Florida at all. We have over a thousand natural springs, the most in all the world.

Jeff Jarvis [02:09:36]:
What do you do, sit on them?

Leo Laporte [02:09:39]:
No. Florida's got beaches. Miami, my God. South beach, where I am right now.

Paris Martineau [02:09:44]:
Is allegedly one of the most beautiful beaches in North America. But you know.

Leo Laporte [02:09:47]:
But you're on the Panhandle, right? You're not in the.

Paris Martineau [02:09:49]:
I'm on the Gulf Coast.

Leo Laporte [02:09:51]:
Gulf Coast. The Gulf of America. Do they have signs that say the Gulf of America?

Jeff Jarvis [02:09:57]:
That's my question. Yeah.

Paris Martineau [02:09:58]:
Yes.

Jeff Jarvis [02:09:59]:
America.

Leo Laporte [02:10:00]:
This way.

Paris Martineau [02:10:01]:
A lot of people in the real estate agency, in the real estate industry may also use that term.

Leo Laporte [02:10:08]:
Welcome to the Gulf of America. Now I have a theory, by the way, that this has nothing to do with AI. I shouldn't probably bring it up.

Jeff Jarvis [02:10:18]:
Uh oh.

Paris Martineau [02:10:19]:
Even when he's doubting, he can't leave me now.

Jeff Jarvis [02:10:21]:
You have to.

Leo Laporte [02:10:22]:
I think there's possibility that Putin and Trump have done the Spain and Portugal thing and decided how to divide the world, and that Putin's saying, look, if you look the other way on my ambitions in Europe, particularly in Eastern Europe, we'll look the other way on your ambitions in Venezuela and Cuba, Canada, Cuba and Mexico.

Paris Martineau [02:10:42]:
Or are we gonna take over, like, Greenland or something?

Leo Laporte [02:10:44]:
Yeah, the Greenland thing is actually the flaw in that, because that's closer to Europe than it is to us. Maybe that's a bargaining chip. I mean, I don't think. I don't know. I. Just a thought, just. Why? I just had a wild thought one day. I've been listening to a lot of history, and.

Leo Laporte [02:11:02]:
And there's been a lot of history of this kind of thing. You know, this kind of.

Paris Martineau [02:11:06]:
One thing I've learned from being on a lot of first dates lately is that boys love to listen to history.

Leo Laporte [02:11:11]:
Boys think about Rome constantly.

Paris Martineau [02:11:13]:
There's something that happens when you get to your, like, mid to late 30s. Like, all men have to choose one. They have to, like, sort into one historical house of thing they want to be obsessed with. I don't really get it.

Jeff Jarvis [02:11:25]:
It's like when they get out of dinosaurs, they get into history.

Leo Laporte [02:11:27]:
Oh, that's what it is.

Paris Martineau [02:11:28]:
Everybody's really into the Revolutionary War right now because.

Leo Laporte [02:11:31]:
Well, yeah, because of that. Ken Burns. It's very good, by the way. I just started watching it.

Paris Martineau [02:11:34]:
It's. I. Okay, I fell asleep. I don't. I didn't fall asleep because I'm not rude, but I agree. I should have fallen asleep because it's. Frankly, it's a documentary that should have been a podcast. And I get that.

Imad Mostaq [02:11:45]:
That's.

Paris Martineau [02:11:46]:
I guess Ken Burns thing is that.

Leo Laporte [02:11:48]:
Because all of his podcast documentary should be podcasts.

Paris Martineau [02:11:51]:
All these things that you're just zooming and out, and I'm like, well, if you have no visual content to show us, it's very difficult. So that you have to invent zooming in and out in a photo. Why not just make a podcast? My guy.

Leo Laporte [02:12:03]:
It's a good, good point.

Jeff Jarvis [02:12:04]:
I. I like.

Leo Laporte [02:12:04]:
It's.

Jeff Jarvis [02:12:05]:
It's.

Leo Laporte [02:12:05]:
They didn't have podcasts when he started.

Jeff Jarvis [02:12:07]:
Now, this could have been a podcast. I like that.

Paris Martineau [02:12:10]:
It could have been a podcast.

Leo Laporte [02:12:11]:
It's could have been a podcast. It sounds like a show title. Friendly bow would have said, this could.

Paris Martineau [02:12:17]:
Have been a podcast.

Leo Laporte [02:12:18]:
And it's a podcast. All right, What Else.

Jeff Jarvis [02:12:22]:
So are you gonna get Leo? Are you gonna get Amazon, Leo?

Leo Laporte [02:12:26]:
I am tempted, but it's. Right now it's commercial only. It's really very expensive. I know, I know. We have Starlink, Amazon, Leo.

Jeff Jarvis [02:12:34]:
It's called you can get rid of Elon Musk.

Leo Laporte [02:12:36]:
It's their satellite solution. And they claim they're going to be able to go to gigabit speeds because of the satellite technology they're using now. This is pretty new. They haven't launched that many satellites.

Paris Martineau [02:12:48]:
Does the fact that it contains your name, is that a plus?

Leo Laporte [02:12:52]:
That's definitely a plus.

Jeff Jarvis [02:12:53]:
Yeah.

Leo Laporte [02:12:55]:
Look at that in all in lights. Amazon, Leo. It stands for Low Earth Orbit. Not anything. But I do like the name.

Paris Martineau [02:13:05]:
I mean, that's what your name stands for, right?

Jeff Jarvis [02:13:07]:
Yeah.

Leo Laporte [02:13:08]:
How did you know? My parents were very forward thinking. You know, we're gonna name the kid Low Earth Orbit. No, you can't call him Low Earth Orbit. All right, how about Leo? Everybody will know what it means. Jesus, Jeff, how many archive.org articles did you post in here?

Paris Martineau [02:13:26]:
Well, but I shout out to Ken, Neal322 woof in the chat, who says I never go into Florida further than Pensacola. Smart idea. I live about 40 minutes from Pensacola. You don't need to be coming here.

Jeff Jarvis [02:13:40]:
So the asterisk ones.

Leo Laporte [02:13:42]:
Okay, Those are the good ones.

Jeff Jarvis [02:13:44]:
So the latest Yudkovsky nut ballerie.

Leo Laporte [02:13:47]:
So Eliezer Yudkovsky is a doomer. He's the doomer, isn't he? He's the doomer, yes. He has proposed an international agreement to prevent the premature creation of artificial superintelligence.

Jeff Jarvis [02:14:02]:
It's his people from the Machine Intelligence Research Institute, which he started.

Leo Laporte [02:14:06]:
Okay.

Jeff Jarvis [02:14:06]:
Their technical governance team. So they're proposing. They're going through the whole AI risk. Yeah. We can destroy you. We can destroy you. Can destroy you. And then they have a whole bunch of proposals, including stopping all research.

Leo Laporte [02:14:22]:
Yeah, good luck with that. This is just tilting at windmills.

Jeff Jarvis [02:14:27]:
Oh, it's ridiculous. It's also having a chip control, so you know exactly who has chips. So you can. You can come after them. It's supposedly there's going to be a US China Executive Council, though they admit that US and China aren't exactly seeing eye to eye on this stuff. But, yeah, complete, complete stop to AI research that can do anything that's generalized.

Leo Laporte [02:14:52]:
So you weren't here earlier, Paris, but our guest this week, I asked him towards the end whether he was a doomer. He said, well, I think it's about 50. 50. But he didn't seem Distressed by that at all. You know, it seemed like, well, we got to work toward the other the good 50, not the bad 50.

Jeff Jarvis [02:15:13]:
Why are you banning research? Q and A. We recognize that restricting research is controversial and normally a bad idea. Unfortunately, we think it is necessary to prevent the premature development of asi. This is primarily a descriptive statement, not a normative one, whatever the hell they think that means. To prevent the development of asi, the narrow set of research aimed at this goal must be stopped. In the current paradigm, AI capabilities increase from the combination of scaling the number of operations used at how having better AI algorithms.

Leo Laporte [02:15:43]:
It's just not going to happen. It's.

Jeff Jarvis [02:15:45]:
Well, it's just stupid.

Leo Laporte [02:15:46]:
Yeah.

Jeff Jarvis [02:15:47]:
Why do you hold it for so long? In short, AI alignment is probably difficult technical problem and it is hard to be confident about solutions. Pausing for a substantial period gives humanity time to be careful in this domain. Well, if you're not doing research, how can you be careful? And on and on and on. So that's a stupid paper then. There's an interesting paper. Yes, which is AI Salesman Toward a large. Toward Reliable Large Language Model Driven Telemarketing. Oh God, this is the nightmare of us all.

Leo Laporte [02:16:19]:
Isn't this what we just were talking about again with Ahmad in the. In the Hitchhiker's Guide to the Galaxy? Remember they. They put the telephone sanitizers and the telemarketers on a spaceship and said, you're going to colonize new planets. Bye bye.

Imad Mostaq [02:16:34]:
This is one of the.

Jeff Jarvis [02:16:36]:
There is research.

Benito Gonzalez [02:16:37]:
Look at all the words in here that suck. AI Salesman Telemarketing.

Leo Laporte [02:16:42]:
Telemarketing Model Driven Large Language. Yeah, you're right, it's a bunch of bs.

Paris Martineau [02:16:49]:
The only way it could be worse is if it was designing. Re Redesigning Telemarketing from First Principles the AI Large Language.

Leo Laporte [02:16:59]:
Overview of Training and Inference for the AI Salesman. Hello, this is Alex from support team. I'm quite busy at the moment. Are you calling to sell me something? I completely understand you're busy. This will only take a minute. We have a free Quick Boost ad credit offer. Would you be interested in hearing about that?

Paris Martineau [02:17:20]:
It's really an optimistic understanding. Somehow what someone says when they receive a spam call.

Leo Laporte [02:17:28]:
Yeah, somehow. Where can I see promotional details? Is the next sentence by the person you're calling. You can see a notification center at the top right corner of the app homepage where you'll find promotional details. Yeah, it's easier though to hang up, frankly on an AI salesman.

Jeff Jarvis [02:17:45]:
Well, my trick on these things because we still have a phone with, you know, with an actual phone, they call.

Leo Laporte [02:17:52]:
Put it down.

Jeff Jarvis [02:17:53]:
Well, no, what I do is I. Whenever it's a number I don't know which is always, I answer the phone to silence. Silence. And. And. And they're. They're waiting for you guys are out.

Paris Martineau [02:18:03]:
Here answering the phone.

Jeff Jarvis [02:18:05]:
I know it's strange.

Leo Laporte [02:18:06]:
People in her age group never answer the phone.

Paris Martineau [02:18:09]:
I mean, I answer the phone if it's my work number, but I don't answer. Well, it depends on the area code that I believe is associated with spam.

Leo Laporte [02:18:19]:
Your. Your Google phone did this before anybody. And now the Apple phone will do this. If it's a number that's not in my book, it will say, hi, Leo's not. Or I don't know what it says exactly.

Paris Martineau [02:18:32]:
It really does that.

Leo Laporte [02:18:33]:
Yeah, it answers and it says, who's calling, please? And then you can do this, Paris, with your phone. This is a new feature on the. On iOS 26. Who's calling, please?

Paris Martineau [02:18:45]:
Is that the one that has the liquid glass? Because I don't want to upgrade to that.

Leo Laporte [02:18:48]:
Yeah, but you could turn off the liquid glass now they put it. You can button in. Yeah.

Paris Martineau [02:18:53]:
Fully.

Leo Laporte [02:18:54]:
Pretty much, yeah. Transparency. You could turn off all the transparency, but it's. I'll tell you what, between that and it's on messages, too. I don't. If I don't know you. Your messages go into a little side bin that I never look at. And all the political messages, all the spam and all goes there.

Leo Laporte [02:19:10]:
Now, I never. I check it once a week and then delete it all. It's great. But letting the phone pick up and say, who's this? I'll see if Leo can come to the phone right now is single. It's singularly useful. Nobody, by the way, ever. I've been using this for two months. Not one person has said, oh, yes, Joe, can you.

Leo Laporte [02:19:34]:
Can you tell Leo I'm calling? I want to talk to him about something? It's always, press one to hear more. It's always B.S. yeah, because most phone calls are B.S. nowadays. But I don't. But I want to answer them because what if it's my doctor or, you know, whatever.

Paris Martineau [02:19:52]:
Don't you have those numbers saved?

Leo Laporte [02:19:55]:
Doctors often lock numbers because they're supposed to be private, like, you don't know your. You know, you want your wife to know your doctor's calling with your STD results or something. I don't know.

Paris Martineau [02:20:06]:
Not that that's like rigamarole.

Leo Laporte [02:20:08]:
Not that that's ever happened to me.

Paris Martineau [02:20:10]:
I'm getting a call right now.

Leo Laporte [02:20:13]:
But no, honestly, I noticed lately because probably they never got anybody. They have started putting the caller ID on but for a long time doctors it was unknown because it was say.

Paris Martineau [02:20:25]:
Whenever I get a call back from one medical, it is a phone number of some sort.

Leo Laporte [02:20:29]:
Yeah. See that I think they've learned nobody's going to answer it if you don't say who it is.

Jeff Jarvis [02:20:34]:
So one more paper, one more, and.

Leo Laporte [02:20:36]:
Then we'll take it. Rachel.

Jeff Jarvis [02:20:37]:
So. So this is a project to create so. Well, it's an anagram to E Scholar. And it's a. An AI Scholar. And she wrote a bunch of papers and this was not an effort to fool. Now Leo's. You can't see this, folks, but Leo's yawning.

Leo Laporte [02:21:00]:
He made fun of this because Paris yawning as soon as I get as you well know, is contagious.

Jeff Jarvis [02:21:05]:
Okay, I get the message.

Leo Laporte [02:21:07]:
So you know what, Paris? I did the same thing. When I got to that my mom's house nap instantly fell asleep.

Paris Martineau [02:21:13]:
I have taken a nap almost every day I've been here. It's crazy.

Leo Laporte [02:21:17]:
It's an unconscious.

Paris Martineau [02:21:18]:
I had to drink a monster energy drink today because for some reason my father has monster energy drinks lying around.

Leo Laporte [02:21:25]:
Ah, he's replaced the Joe Rogan gum with monster energy drinks.

Paris Martineau [02:21:29]:
Yes, he has. And is it better? Is it worse? I don't know. But I did have half of one today and at least I'm more awake.

Jeff Jarvis [02:21:37]:
Yeah, for us. She did it for us.

Leo Laporte [02:21:39]:
Thank you.

Paris Martineau [02:21:39]:
I did earnestly.

Leo Laporte [02:21:40]:
I think it was the donut that made you yawn. Actually, that donut definitely counteracted.

Paris Martineau [02:21:45]:
Shout out to the donut hole. If anyone's ever visiting Destin, a perfect donut establishment and breakfast place that you stew be 24 hours. It was place of my first job.

Leo Laporte [02:21:57]:
Oh, you worked there from 2pm to.

Paris Martineau [02:21:59]:
2Am Monday through Friday during 10 hours. It was definitely not legal.

Leo Laporte [02:22:05]:
I was like 12 hours, 12 hour shifts.

Paris Martineau [02:22:08]:
12 hour shift was raking in that dough in more ways than one.

Jeff Jarvis [02:22:14]:
Did you. Did you make the donuts or did you just. Just.

Paris Martineau [02:22:17]:
No, I was like behind the counter. It was kind of a diner situation with also donuts. But at midnight, my job was to throw away all the donuts. I was supposed to hold steadfast if any drunk people wanted them. And they were like, no, under no circumstances can you take them home. It's a health hazard. You can't give them to anyone. You've got to throw them away.

Paris Martineau [02:22:35]:
But I quickly learned that you take a Clean garbage bag and park your car right near the dumpster. And then instead of just putting it in the dumpster, I would put the sack of like 100 donuts in my trunk. And then at the end of my shift I'd go around to friend's house houses and drop off bags of donuts.

Jeff Jarvis [02:22:54]:
Day old donuts At Ponderosa Steakhouse we had really good buns. And before closing we would put in a whole tray of buns so they were left over.

Leo Laporte [02:23:04]:
At the end of the evening we.

Jeff Jarvis [02:23:05]:
Could get buns home.

Paris Martineau [02:23:06]:
Wow.

Leo Laporte [02:23:07]:
This is why there is no longer a Ponderosa Steakhouse anywhere in America. Oh, are there?

Jeff Jarvis [02:23:12]:
There is. Okay.

Paris Martineau [02:23:13]:
I think there are.

Jeff Jarvis [02:23:14]:
So were the papers okay?

Paris Martineau [02:23:16]:
Yeah. Tell us about your paper, Jack. We care about you.

Jeff Jarvis [02:23:19]:
I know you do, even though you're both yawning.

Leo Laporte [02:23:22]:
So they decided to make Ferris is eating. I'm yawning.

Paris Martineau [02:23:29]:
We were talking about donuts and now I want one.

Jeff Jarvis [02:23:33]:
They decided to have to see whether Rachel could actually make papers. And this was not to fool anybody. This was not to try to catch them at papers. And she got papers published and she got invited to review.

Leo Laporte [02:23:45]:
Oh my God. So it did fool people.

Jeff Jarvis [02:23:47]:
Well, but, but they were all honest about it. It was in there. So if you go down, there's a list of the papers that she wrote or you can see them on the next line.

Leo Laporte [02:23:56]:
Wow. Rachel's paper Detection of AI generated academic papers. She wrote a paper? Yes, about detecting AI generated papers, institutional policies and AI writing tools. Oh, she wrote a lot about AI synthesizing scientific literature, literature, the LLMs, ethics of undisclosed AI use in research, pen names and scientific writing. So she's writing her experience, which is what you're supposed to do. AI based scientific research assistance, Intellectual property rights for AI outputs. Asking for a friend. Policy of academic journals towards AI generated content.

Leo Laporte [02:24:29]:
Impact of AI on the peer review processes, Writing approaches blending human and machine. This is, this is hysterical, isn't it? It's almost tongue in cheek.

Jeff Jarvis [02:24:39]:
Yeah. But it's, it's kind of not. It's. It's.

Leo Laporte [02:24:44]:
Now, did you read any of these?

Jeff Jarvis [02:24:46]:
I tried to go a little bit. You know, that's because the topics are the topics. So it's kind of.

Leo Laporte [02:24:51]:
I feel like there is a formula to scientific.

Jeff Jarvis [02:24:55]:
Yes.

Leo Laporte [02:24:55]:
Writing and so probably something that an AI could do. Oh, yeah, quite well, absolutely.

Jeff Jarvis [02:25:03]:
You know, it can do the, the literature review and it can do the hypothesis.

Leo Laporte [02:25:07]:
This is pretty good.

Jeff Jarvis [02:25:08]:
Yeah.

Leo Laporte [02:25:09]:
Pseudonyms representing long standing practice that extends beyond literary works into scientific and academic writing. They're Part of a broader category of name altering practices common across creative fields, including pen names and literature, stage names in the performing arts and aliases in music. This is good. Yeah, I'd give it an A plus.

Jeff Jarvis [02:25:27]:
It's cited.

Leo Laporte [02:25:28]:
Where's the bamboo? It needs. We need some analogies.

Paris Martineau [02:25:31]:
Well, I mean, yeah, it's interesting. Do you think that people were aware that this was AI generated there?

Jeff Jarvis [02:25:40]:
They say that they were open about it.

Paris Martineau [02:25:43]:
I mean, they say.

Leo Laporte [02:25:44]:
That doesn't say.

Paris Martineau [02:25:45]:
They say that.

Jeff Jarvis [02:25:47]:
They say that.

Paris Martineau [02:25:48]:
All Rachel so publications include explicit disclosure statements acknowledging in the body of the paper themselves that Rachel so is an AI scientist. But they don't appear in Rachel says, Google Scholar profile or citation appearances. They don't inherently signal AI authorship to casual observers. Did the bachelor thesis student who cited Rachel so's work recognize they were referencing AI generated scholarship? Probably not.

Leo Laporte [02:26:16]:
Good.

Paris Martineau [02:26:16]:
Similarly, was the Peer J editor aware of Rachel so's artificial nature? I assume its peer review. We chose not to inform them.

Leo Laporte [02:26:24]:
See, look at. This is a little deceptive because here's the disclaimer in this piece I was reading. It merely says generative AI has been used to prepare this manuscript. That's not the same as this was written by AI. I don't think that's full disclosure. Oh, wait a minute.

Jeff Jarvis [02:26:41]:
Wait, no, there's something else, I think in there.

Leo Laporte [02:26:43]:
Here's the bio. Rachel so is an AI scientist. Now that could be a scientist who studies AI or. Or a scientist who is AI. And it's not. That's ambiguous. She focuses on the impact of artificial intelligence. No, you know what? This is not.

Leo Laporte [02:27:00]:
They are being disingenuous. This is a little Olivia Dozy.

Paris Martineau [02:27:05]:
They said a countermeasure would have been to explicitly name the author, such as Rachel, AI generated. But we believe that that would have been detrimental to measuring those impact due to AI stigma. We acknowledge the tension between formal transparency and practical transparency.

Jeff Jarvis [02:27:22]:
What they wanted is the goals of the project, the feasibility goal about creating an end to end AI identity, including publication records, scholarly profiles, integration into bibliographic databases. This explores the existing identity systems in academic infrastructure to explore academic recognition mechanisms and to provoke unnecessary discourse and debate. Yeah, it's a little.

Leo Laporte [02:27:45]:
So here's my question. Is this an example of the Turing Test being passed? I guess not. Because the Turing Test requires it.

Jeff Jarvis [02:27:56]:
You have to quiz it.

Paris Martineau [02:27:57]:
Yeah.

Leo Laporte [02:27:57]:
And it requires also the knowledge that one of these people is real and one of them is an AI, so. But what it does demonstrate is it's very possible for AI to fool people about Being AI.

Jeff Jarvis [02:28:09]:
The great thing is if you go to the paper itself on the bottom of the first page, where it always puts the affiliations of the authors. Yeah.Human comma KTH, Royal Institute of Technology. Human comma University de Montreal. Human comma University Brussels.

Leo Laporte [02:28:27]:
So there are humans.

Jeff Jarvis [02:28:29]:
They say they're humans.

Leo Laporte [02:28:30]:
Yes, but they're the humans that wrote the AI or launched this AI into. Yeah, yeah. I think they're being, they're, they're, they're definitely obscuring disagreements.

Jeff Jarvis [02:28:42]:
Yeah. But fascinating nonetheless.

Leo Laporte [02:28:43]:
Yeah, well, what's fascinating and not surprising to me is that it does a good job. Does a good job.

Jeff Jarvis [02:28:49]:
Yeah.

Benito Gonzalez [02:28:53]:
I mean, some of these are very niche topics though. So like that paper might be like 30% plagiarized and you don't know.

Leo Laporte [02:29:00]:
Yeah, well, we don't know. That's right. We also don't know if it's accurate.

Jeff Jarvis [02:29:03]:
But also. But you know, most academic papers aren't niche products. Niche topics.

Leo Laporte [02:29:09]:
Right.

Benito Gonzalez [02:29:09]:
Yeah, but like, that's because a researcher went out and did the research and.

Jeff Jarvis [02:29:13]:
That did the research. Yes.

Leo Laporte [02:29:14]:
And I would, I would love to go back to school these days, though. I mean, honestly, my senior thesis was supposedly going to be comparing European feudalism with Japanese feudalism. I think the AI could have written of. Well, my advisor said, don't write about that. That's terrible. He said it was too big. But I thought it was fascinating and I could have narrowed it down a little bit, you know, I could, as you said, you know, what, foodstuffs or something. But that would have been something AI could have probably helped me a lot with.

Benito Gonzalez [02:29:44]:
You could probably. During the commercial break.

Jeff Jarvis [02:29:48]:
Good thought. Yeah. Line 131, Code Wiki which evaluates AI's ability to generate documentation for large scale code bases. Here's my question to you.

Leo Laporte [02:30:00]:
Do you need documentation for code bases? Oh, yeah, yeah, yeah. Very much so.

Jeff Jarvis [02:30:05]:
Even if it's made by AI and you're going to just make AI, who needs the documentation? Who needs the documentation then?

Leo Laporte [02:30:11]:
Well, by the way, the AI can read it.

Jeff Jarvis [02:30:13]:
Oh, okay. All right.

Leo Laporte [02:30:14]:
In fact, one of the things, stuff like Claude code does is. Generates quite a bit of textual information telling it what to do. I'll give you. I'll show you actually, if you're. If you're interested.

Jeff Jarvis [02:30:27]:
I'm not yawning.

Leo Laporte [02:30:30]:
You will be. Just hang in there. It's not over yet. But I. Let me open my readme for I have. I've been using AI to tune up my Emacs configuration. So this was all written by Claude Code which. Which did a lot of, you know, it took my.

Leo Laporte [02:30:53]:
My starting point. Let me make this full screen. It took my starting point, but then wrote all of this. And the thing is, it reads it too. So this is information that it will ingest if it. If I ask it to come back and say, okay, explain, you know, do something different. This will tell it what it's done. And there's.

Leo Laporte [02:31:16]:
There's comment in the code as well, which you can put in. Although it doesn't put in a lot of comments in the. The in its own code. But it does tend to write. In fact, you can have it write a to do list. You can have it write a lot of documentation. In fact, you're encouraged to do that. So that's a.

Leo Laporte [02:31:32]:
That's a big part. So now I'm going to ask Gemini write an. A Senior thesis, college level, senior thesis, analyzing. I wonder if it could do this during the commercial break. Differences. You thought that was a good subject. Yeah, I thought so. I was a history major.

Leo Laporte [02:32:05]:
And Japan had a form of feudalism with surfs and everything.

Paris Martineau [02:32:10]:
Sorry. I've gotten pulled into a quest that I sent the chat, which is to try and make. Make your shirt look like a pinata.

Jeff Jarvis [02:32:19]:
You were concentrating hard on something. I was dying. What it was.

Leo Laporte [02:32:23]:
Let's take a break.

Paris Martineau [02:32:24]:
And it keeps getting blocked because I think there's something about the pinato dangling from something that is.

Leo Laporte [02:32:31]:
Or my name. Did you say Leo or.

Paris Martineau [02:32:34]:
I've gotten. I've got. I've gotten the creepy very.

Leo Laporte [02:32:37]:
Say Abraham Lincoln or something because you can use him.

Paris Martineau [02:32:41]:
I just said my friend. And it did work.

Leo Laporte [02:32:43]:
But this is.

Paris Martineau [02:32:43]:
This is as far as I got. Is. Is just you creepily standing as a pinata among a back.

Jeff Jarvis [02:32:51]:
Somebody hit him with a stick, it looks like.

Leo Laporte [02:32:54]:
I'm worried somebody is gonna hit me with a stick.

Jeff Jarvis [02:32:56]:
You're a robot. You're gonna be pushed.

Paris Martineau [02:32:59]:
Paris.

Benito Gonzalez [02:33:00]:
They probably don't want any people hanging from trees, if you know what I mean.

Paris Martineau [02:33:03]:
I know that's what I said.

Leo Laporte [02:33:04]:
That's what it is.

Paris Martineau [02:33:08]:
But I also, you know.

Leo Laporte [02:33:09]:
Well, you keep working. I'm going to let Gemini write my senior thesis for me. Maybe I can go back to school, get that bachelor's degree I've been. I've been waiting for all these years. I don't know. I don't know if they'll let me come back. I don't know. Anyway, this episode of Intelligent Machines is brought to you by Vention.

Leo Laporte [02:33:28]:
I had a great conversation with the folks at Vention. About a month and a half ago, Glenn over there and the team, really very cool people. And of course they're fascinated by AI as we all are. And they're very well aware as engineers that AI, which is supposed to make things easier for most teams, makes the job harder. It just adds challenges. But Vention's been doing this 20 plus years of global engineering expertise. They really know what it takes to build products. In fact, now they're building AI enabled engineering teams, or they're helping build those teams to make software development faster, cleaner, and you'll like this calmer, not, not more stressful.

Leo Laporte [02:34:18]:
Clients of Vention typically see at least a 15% boost in efficiency. And we're not talking through hype, we're about talking, talking through real engineering discipline. I think I really want to emphasize that these people care. They also do a workshop that I think you might be interested in. In fact, this might be the best way to meet Vention. They have these fun AI workshops. They're very interactive. It's not a lecture about AI.

Leo Laporte [02:34:42]:
They get together with your team, talk about your team's goals, talk about how their, their work processes to help your team find a practical and safe way to use AI across delivery and qa. It's a great way to start with Vention and test their expertise. So whether you're a cto, a tech lead, maybe a product owner, you're not gonna have to, you know, if you, if you sit down, say, well, can we, can we use AI in this tool? You're not gonna have to spend weeks figuring out all the different architectures and tools and models. By the way, that's a rabbit hole. You don't want to go down. Invention can help you. They can help assess your AI readiness. The other thing they do that's very important, they help you clarify your goals and outline the steps to get you to where you want to be without the headaches.

Leo Laporte [02:35:30]:
And so this is a very useful workshop. Of course, if you need help on the engineering front, their teams are always ready to jump in, whether as a development or as a consulting partner. But these workshops are a great way to start. It's the most reliable step to take after that, that, that proof of concept. You, you know, you've built this incredible promising prototype, you know, using Claude code or lovable, and it's running well in tests. But then you got this decision point. What do you do next? Do you open a dozen AI specific roles, you know, and get this thing launched? Or maybe, and I would suggest this is the right way to Go bring in a partner who has already done this across industries. Somebody who can take your idea, your fledging idea and expand it into a full scale product without disrupting your systems or slowing your team down.

Leo Laporte [02:36:21]:
Vention. Real People with 20plus years of engineering expertise and lots of real results. You can find out more at their website ventionteams.com see how your team can build smarter, faster and with a lot more peace of mind. Or get started with your AI workshop today at ventionteams.com twitch that's Vention. V E N T I O-N teams.com Twitter thank him so much for their support of this intelligent machine show. You got me hanging and you got some kids banging on me.

Paris Martineau [02:36:59]:
Hold on, I got it. I got to go slightly askew.

Leo Laporte [02:37:02]:
Oh my, oh my. When's the candy come out of that cage? But that's gonna be good. I like that suit though. I'd like to get that suit.

Paris Martineau [02:37:12]:
You should get that suit. I think that would look great.

Leo Laporte [02:37:14]:
That is very nice. All right, let me see what Gemini has come up with. Generating a full page academic thesis. Three or four thousand words in a single response exceeds the output limits of this program. However, I can provide a comprehensive framework. An outline required to fill 10 pages. A full length. This is good.

Leo Laporte [02:37:34]:
An introduction, a deep dive analysis. The most critical section, the nature of obligation and a conclusion. The thesis titled Divergent Swords A Comparative analysis of the Legal and Moral Frameworks in Medieval European and Japanese feudalism.

Jeff Jarvis [02:37:48]:
Trying too hard.

Leo Laporte [02:37:49]:
You don't like that?

Jeff Jarvis [02:37:50]:
No, no, no.

Leo Laporte [02:37:51]:
Little bamboo. Yeah. The paper argues that European feudalism was rooted in contractual law and mutual obligation, whereas Japanese feudalism was predicated on moral absolutism and filial piety.

Jeff Jarvis [02:38:04]:
Can we make bamboo the new word for overwritten bam.

Leo Laporte [02:38:08]:
Bamboo.

Paris Martineau [02:38:09]:
It's bambooi.

Leo Laporte [02:38:10]:
Bam Bambooi.

Benito Gonzalez [02:38:11]:
I don't find that accurate because I think European kings also had God mandates. Right?

Leo Laporte [02:38:17]:
Sure, sure. Mutual obligation though. Because remember, the feudal lords had an obligation to the king to defend the king. History often repeats its. Rarely repeats itself, but often rhymes. Now, you shouldn't put that in. Oh geez. Without actually crediting it.

Leo Laporte [02:38:35]:
Well, it's. It's a famous quote.

Jeff Jarvis [02:38:37]:
I know it is, but it's also just a cliche.

Leo Laporte [02:38:39]:
Yeah. Perhaps the most striking rhyme in human political history is the independent emergence of feudalism in Western Europe and Japan. Well, I'm feeling pretty good about my thesis so far. Superficial structural similarities mask profound philosophical differences. Differences. I think it's. You know What? It's. It's.

Leo Laporte [02:39:00]:
P.S. but it's. It's good BS. I mean, most theses are BS, are they?

Benito Gonzalez [02:39:05]:
Not college BS. That's like for.

Jeff Jarvis [02:39:07]:
That's real.

Leo Laporte [02:39:08]:
Yeah, I feel like. Look, it's got a whole bibliography of real books, I think. I have no way of knowing.

Jeff Jarvis [02:39:15]:
Yeah, look up just one of them.

Leo Laporte [02:39:18]:
I think this is real. Mark Block's Feudal Society, I think is actually real.

Paris Martineau [02:39:23]:
Based on what I.

Leo Laporte [02:39:24]:
Sounds familiar.

Paris Martineau [02:39:26]:
We need to.

Leo Laporte [02:39:27]:
My friend. It is a Rutledge classic.

Paris Martineau [02:39:31]:
We need to pause and look at the video someone made of brand droid made of you. The pinata in the chat. And you need to listen to it with audio.

Leo Laporte [02:39:42]:
Oh, no, no, no. That's a good one, though. Burke made that one.

Paris Martineau [02:39:49]:
I think that's just a gif. It's up above my pinata images.

Leo Laporte [02:39:53]:
Oh, it's above yours.

Jeff Jarvis [02:39:54]:
The video.

Paris Martineau [02:39:55]:
The video.

Leo Laporte [02:39:56]:
Ah, here's the video. Okay. It's very strange.

Paris Martineau [02:40:07]:
Terrifying. Why are you doing those motions?

Leo Laporte [02:40:12]:
Because that's how the candy comes out.

Jeff Jarvis [02:40:14]:
He's molting.

Paris Martineau [02:40:16]:
And then there's studio canned laughter.

Leo Laporte [02:40:19]:
Canned laughter. It's very strange. Brandroid, who did you. How did you make that one? That is. Huh? He says the audio is truly disturbing.

Jeff Jarvis [02:40:31]:
Yeah.

Leo Laporte [02:40:34]:
Well, folks, I think this would be a good time for us to perhaps take a look at our picks of the week. Unless you've got a story that we. We really should have done that we didn't get to. We had a great interview.

Paris Martineau [02:40:47]:
But the only story I have is that I've tried now four different times to reprompt Gemini Nano Banana, to break the Leo pinata apart and have candy spilling out. And every time it says it's done it and every time it has not. AI is fallible, folks.

Leo Laporte [02:41:05]:
Well, if you wanted to do illegal things, ask it, like have a pinata.

Paris Martineau [02:41:10]:
Be full of candy. What's illegal about that?

Jeff Jarvis [02:41:12]:
That.

Leo Laporte [02:41:13]:
All right, I'm gonna.

Paris Martineau [02:41:14]:
America. We can fill our pinatas with anything we want.

Leo Laporte [02:41:18]:
I'm gonna ask it to do a caked up football man. How about that?

Paris Martineau [02:41:23]:
I bet it will return a great caked up football man. I'm gonna be honest.

Leo Laporte [02:41:27]:
Let me see. I have to make this an image, right? Create images. Just. I'm not gonna tell it anything more. Just caked up football man.

Benito Gonzalez [02:41:34]:
It'll be interesting to know if. To see if it knows, like, the lingo.

Leo Laporte [02:41:37]:
It'll know what caked up is.

Imad Mostaq [02:41:38]:
Is.

Leo Laporte [02:41:39]:
I think we'll see. It may say it couldn't do it. Maybe it's too scatological.

Benito Gonzalez [02:41:46]:
It could be made of cake.

Leo Laporte [02:41:47]:
Paris While Gemini is nano banana. Oh, wait a minute. It's done. I think I misunderstood a little bit there. It's a football guy covered in birthday cake.

Jeff Jarvis [02:42:00]:
Cake.

Leo Laporte [02:42:00]:
And it says, congrats, champ. Why do people in AI images look so sad? Are they trapped in there?

Paris Martineau [02:42:09]:
They are trapped.

Leo Laporte [02:42:10]:
They're trapped inside. There is no humans.

Paris Martineau [02:42:15]:
Let me see. I'm trying to get it to generate an image of a caked up. Oh, again, it does do caked in mud. Interesting. It does not understand the slang of being caked up. I wonder if it would know what a dump truck behind is. Which is, I suppose, the pg, er version of that.

Benito Gonzalez [02:42:32]:
You gotta wait to the point dictionary.

Leo Laporte [02:42:34]:
Football man with a great big badonkadon. Can you say that, Paris? What is your. What is your.

Paris Martineau [02:42:42]:
All right. My picks of the week are. One, this beautiful Nick Cage commercial that someone, I assume who listens to the show tagged me in. Because how else would they know it?

Leo Laporte [02:42:54]:
Oh, wait a minute.

Paris Martineau [02:42:55]:
Japanese commercial.

Leo Laporte [02:42:57]:
And this is from.

Paris Martineau [02:42:58]:
Wait, guys.

Leo Laporte [02:42:58]:
This is from Lambeau Field. Ladies and gentlemen. The statue of Caked Up Football Man.

Jeff Jarvis [02:43:05]:
That certainly is. What the hell?

Leo Laporte [02:43:09]:
I think that's Vince Lombardi. Ladies and gentlemen, the Caked Up Football Man.

Paris Martineau [02:43:14]:
Okay, I think we also just need to pause for a minute to remark on the fact that job done. In a real reversal of fortunes. I am about to show you guys an old commercial.

Leo Laporte [02:43:25]:
Oh, wow. An old Nick Cage commercial.

Paris Martineau [02:43:29]:
A Nick Cage commercial in the early 2000s.

Leo Laporte [02:43:31]:
Have your autograph.

Paris Martineau [02:43:32]:
All right.

Leo Laporte [02:43:33]:
So your twins. No, we're triplets.

Jeff Jarvis [02:43:38]:
Triplets.

Leo Laporte [02:43:45]:
This is a Japanese.

Jeff Jarvis [02:43:49]:
Looks like AI Slop.

Leo Laporte [02:43:51]:
It does.

Paris Martineau [02:43:51]:
But it was before AI Slop was a thing.

Leo Laporte [02:43:56]:
Yeah, it's a Senkyo ad. Thank you so much.

Jeff Jarvis [02:44:02]:
All right.

Paris Martineau [02:44:04]:
How do we. Yeah, I just thought that was beautiful. Shout out to. I'm forgetting who tagged me in it. But I responded to you, and it was lovely.

Leo Laporte [02:44:12]:
Almost as good as Suntory time. Wow. Everybody's triplets in the ad. This is the early 2000s.

Paris Martineau [02:44:19]:
It is very nice. And my other one is the stuffing recipe I make every year, which is from the Cozy Apron. It is not actual stuffing, so it's not stuffed in the bird because, of course, the bird is getting deep fried. But it is perfect. It's crunchy and savory. And I usually make a double batch of it so that we can both eat on Thanksgiving and then have a bunch left over. Some of my various additions I've done along the way is I just use a bunch of sausage instead of, like, two links. And instead of a Four cheese blend.

Paris Martineau [02:44:56]:
I. They use a blend of like, good melting cheeses like gruyere or fontina. You can kind of just mess around with this. Whatever you think could. Like, some years I'll put mushrooms in it. Some years I'll put like sauteed peppers or leeks in it. It's just like a delightful stuffing. Is one of my favorite.

Paris Martineau [02:45:13]:
Is my favorite part of things.

Leo Laporte [02:45:15]:
Everybody's favorite part.

Paris Martineau [02:45:17]:
The best.

Leo Laporte [02:45:18]:
This has Italian sausage in it, which is what my grandma's had. French bread. Bread, though now.

Paris Martineau [02:45:24]:
Yes. You want to. You leave the French bread out the day before so that it gets.

Leo Laporte [02:45:28]:
Oh, it gets a little hard.

Paris Martineau [02:45:28]:
Okay.

Leo Laporte [02:45:30]:
Instead of breadcrumbs, you use French bread, which makes sense.

Paris Martineau [02:45:32]:
Yeah, yeah, yeah.

Leo Laporte [02:45:33]:
Lime is the predominant. A little chopped apples, melted butter.

Paris Martineau [02:45:37]:
Yeah. I often will just use fresh apples instead of dried ones.

Leo Laporte [02:45:41]:
This sounds fantastic.

Paris Martineau [02:45:42]:
It's so good. Every time I make it, someone asks me, like, for the recipe.

Jeff Jarvis [02:45:47]:
But no oysters.

Paris Martineau [02:45:50]:
You could substitute oysters in for any or all of these.

Leo Laporte [02:45:53]:
Now, have you already purchased the French bread to put out overnight? So it'll have. So it's going to sit down on the counter overnight?

Paris Martineau [02:46:02]:
It is, Yeah. I usually chop it up. I'm going to do that after this. Chop it up? Yeah, it'll dry up so that it will dry out.

Leo Laporte [02:46:09]:
How big is the chunks?

Paris Martineau [02:46:11]:
I don't know. It's kind of whatever I'm feeling.

Leo Laporte [02:46:13]:
Okay.

Paris Martineau [02:46:14]:
It depends. I'm kind of loosey goosey with cooking. If it's. It's savory.

Leo Laporte [02:46:19]:
This sounds really good.

Paris Martineau [02:46:20]:
It's.

Leo Laporte [02:46:20]:
This is very similar to my grandmother's Italian recipe, except she put a little orange juice in it just.

Paris Martineau [02:46:26]:
Ooh.

Leo Laporte [02:46:27]:
Gives it a nice little moist tang.

Paris Martineau [02:46:30]:
That's fascinating. If I really wanted to go ham on this, I would make my own chicken stock for it.

Leo Laporte [02:46:36]:
Yeah. But ain't nobody got time for that.

Paris Martineau [02:46:38]:
Ain't nobody got time for that.

Leo Laporte [02:46:40]:
Thecozyapron.com stuffy shout out.

Paris Martineau [02:46:44]:
It's great. Great. It's a great recipe.

Leo Laporte [02:46:45]:
Recipe. I just incorporated it into my paprika.

Paris Martineau [02:46:48]:
What is your guys's favorite Thanksgiving food stuff? Do you even get Thanksgiving food?

Jeff Jarvis [02:46:55]:
Yeah, we have stuffing and we just don't like turkey, so we have chicken.

Paris Martineau [02:47:02]:
You pick it all up. It's all takeout, right?

Jeff Jarvis [02:47:04]:
No, no, no, no, no, no. My wife does the. She does all kinds of wonderful things. She does the.

Leo Laporte [02:47:09]:
It's the sides, really. Always the size of the best.

Paris Martineau [02:47:12]:
Okay. Are you guys still. Are you guys both stuffing as number one people or do you have another.

Leo Laporte [02:47:16]:
I love mashed potatoes and gravy. That's pretty good, too.

Jeff Jarvis [02:47:19]:
Yeah.

Leo Laporte [02:47:19]:
And actually, turkey ain't bad if it's done right.

Paris Martineau [02:47:22]:
I mean, turkey's fine.

Jeff Jarvis [02:47:23]:
So I'm going to give you my. My mother used to prepare what she. What was called Christmas salad.

Leo Laporte [02:47:29]:
No, no one.

Jeff Jarvis [02:47:31]:
You know where I've had it, don't you?

Imad Mostaq [02:47:32]:
Yes.

Jeff Jarvis [02:47:33]:
One package lime jello.

Leo Laporte [02:47:35]:
Oh.

Jeff Jarvis [02:47:36]:
One cup of pineapple juice. No, cottage cheese. One cup of whipped cream. Four slices of pineapple, diced and color. And to put the cherry in it, walnuts, you know, cherries.

Paris Martineau [02:47:56]:
To taste.

Leo Laporte [02:47:57]:
To taste.

Jeff Jarvis [02:47:58]:
Oh, it was terrible. It was just terrible.

Paris Martineau [02:48:02]:
How high was it? How many? Were they all different?

Leo Laporte [02:48:04]:
Oh, no, no, it's.

Jeff Jarvis [02:48:05]:
It's just. It's just jello. So it was just that.

Leo Laporte [02:48:07]:
Yeah.

Jeff Jarvis [02:48:07]:
No, it was not. So my sister was going to. She's coming for Thanksgiving tomorrow, and she was going to, as a joke, make it, which would have been ridiculous. Thank goodness she didn't.

Leo Laporte [02:48:15]:
It's tempting.

Paris Martineau [02:48:17]:
Imagine having to clean out that bowl. Imagine, like, the noise it would make when you're, like, slopping it out.

Jeff Jarvis [02:48:23]:
It's jello that's made translucent because of.

Paris Martineau [02:48:27]:
The whipped cream and mayonnaise and cottage cheese.

Leo Laporte [02:48:31]:
I don't think you mentioned mayonnaise, just cottage.

Paris Martineau [02:48:33]:
I thought you said mayo.

Leo Laporte [02:48:34]:
I just said bonito said mayo. He was trying. He was trying to hack the recipe.

Jeff Jarvis [02:48:39]:
Prove it.

Leo Laporte [02:48:40]:
Yeah.

Benito Gonzalez [02:48:42]:
I was curious.

Leo Laporte [02:48:43]:
No, that was very 50s.

Benito Gonzalez [02:48:44]:
I was sure there was going to be mayo in that recipe.

Leo Laporte [02:48:46]:
Yeah, very 50s. That's what. That's what people did. And it was awful.

Jeff Jarvis [02:48:52]:
Awful. Yeah.

Leo Laporte [02:48:53]:
Fruitcake's not bad compared to that. Fruitcake's delicious compared to that. Jeff, your picks of the week.

Jeff Jarvis [02:48:58]:
Oh, okay. Well. Well, the cooking that I care about, I know, is the head of product development at Taco Bell. Just a little. A little nugget here, a little factoid.

Paris Martineau [02:49:09]:
What's your go to Taco Bell? Order real quick.

Jeff Jarvis [02:49:11]:
Oh, grilled cheese, black bean burrito.

Leo Laporte [02:49:14]:
That sounds all right.

Jeff Jarvis [02:49:16]:
That's very good. It's very good. It used to be just two bean burritos, but now it's. It's tarted up. So they're working on the thing they're working on.

Leo Laporte [02:49:22]:
Yeah.

Jeff Jarvis [02:49:23]:
The AGI of Taco Bell.

Leo Laporte [02:49:25]:
Yeah.

Jeff Jarvis [02:49:26]:
The thing they want and they're. They think they're getting there is a taco shell made of cheese. Right.

Leo Laporte [02:49:34]:
I've had that in Mexico. I had that in Cabo San Lucas. It's been invented.

Paris Martineau [02:49:40]:
Ability to be delivered to me.

Jeff Jarvis [02:49:41]:
But They've got to do it at scale.

Paris Martineau [02:49:42]:
They've got to be able to mass market the cheese shell. God.

Jeff Jarvis [02:49:46]:
Yeah.

Imad Mostaq [02:49:46]:
Yeah.

Leo Laporte [02:49:47]:
There's a taqueria.

Paris Martineau [02:49:48]:
That's a great idea, isn't it? The underrated. The underrated fave, something not enough people know to get at Taco Bell, is the fiesta potato. Oh, it's just a little cup of like home fry style, like fried potatoes. And they've got like sour cream and a spicy crema sort of thing on it. It's fantastic.

Jeff Jarvis [02:50:09]:
Yeah, it's very good.

Leo Laporte [02:50:10]:
I like it that the headline in the Wall Street Journal is Taco Bell knows exactly what you want to eat at 2am which is ironic because that's about the only time I get to Taco Bell is when the bar is closed. Yeah, you would go there and you'd want something to soak up the booze.

Jeff Jarvis [02:50:25]:
Exactly.

Imad Mostaq [02:50:27]:
Yeah.

Leo Laporte [02:50:27]:
So there's not done that in many.

Jeff Jarvis [02:50:29]:
Years by something that I will not do. There's a five hour stage play about Dungeons and Dragons that you may want to go to.

Paris Martineau [02:50:35]:
Where is it?

Leo Laporte [02:50:36]:
It sounds like it's actually a little bit shorter than our game was.

Jeff Jarvis [02:50:41]:
I do believe it's in New York. Paris. I do believe it's off Broadway. Yeah. New off Broadway.

Leo Laporte [02:50:45]:
Oh, they're playing Epic of sorts.

Paris Martineau [02:50:48]:
Okay.

Jeff Jarvis [02:50:49]:
I will go to this as emotionally immersive as Dungeons and Dragons. Games that.

Paris Martineau [02:50:53]:
Listen, I will be going to this with my DND group. We. Sounds like we have a. We're a tough. We're tough critics when it comes to DND related stage stuff.

Leo Laporte [02:51:05]:
Is there a lot of DND related stage stuff?

Jeff Jarvis [02:51:08]:
Play the video trailer or we'll get in trouble.

Leo Laporte [02:51:12]:
No, we would. I mean, this is. This is theater. I don't think they would sue us in the story.

Jeff Jarvis [02:51:16]:
There's a trailer.

Leo Laporte [02:51:17]:
Oh, it's the public theater orientation.

Paris Martineau [02:51:20]:
The world was a different place and we were different people. I'm not at the keyboard right now. Please leave a message after the beep.

Benito Gonzalez [02:51:28]:
Is it about a playgroup? That's about a playgroup.

Leo Laporte [02:51:32]:
Yeah. High school playgroup. Yeah. So it's scripted. It's not. They're not actually playing.

Paris Martineau [02:51:37]:
They're not playing danger.

Benito Gonzalez [02:51:38]:
Aren't you?

Leo Laporte [02:51:39]:
I guess.

Jeff Jarvis [02:51:40]:
Well, I'd never trust anybody who isn't a little bit weird.

Paris Martineau [02:51:45]:
You think we'll do anything worthwhile with our lives?

Leo Laporte [02:51:48]:
Of course you will.

Paris Martineau [02:51:49]:
Oh, why?

Leo Laporte [02:51:50]:
Is it five hours?

Paris Martineau [02:51:51]:
Yeah, why?

Imad Mostaq [02:51:52]:
Better?

Jeff Jarvis [02:51:52]:
No.

Leo Laporte [02:51:53]:
Won't be satisfied by it. I'm already bored and it's only a minute in. Well, you're Looking at an old Mac, though. Look at that Mac. That's pretty cool.

Paris Martineau [02:52:01]:
Yeah, the projected sets are very cool.

Leo Laporte [02:52:03]:
I mean, are they projected? Really? Oh, all of you.

Paris Martineau [02:52:10]:
That's so cool.

Leo Laporte [02:52:10]:
Cool.

Jeff Jarvis [02:52:12]:
All right then. Because I know Paris loves mid century modern.

Paris Martineau [02:52:16]:
I do.

Jeff Jarvis [02:52:17]:
So the Stall House. S T A H L is the.

Leo Laporte [02:52:21]:
See, I found house and I. I will show you a picture of it. I found a gingerbread. Mid century modern Gingerbread.

Paris Martineau [02:52:29]:
I did really like that.

Leo Laporte [02:52:30]:
Yeah, I thought you would. What's the stall? Oh, the Stall house is famous.

Jeff Jarvis [02:52:33]:
Go to. Go to the Guardian story first because it's the famous, famous photo of the Stahl house that you seen.

Leo Laporte [02:52:39]:
Yeah, yeah.

Jeff Jarvis [02:52:40]:
If they're hung over at Los Angeles, you could.

Leo Laporte [02:52:43]:
You can rent it for your shoot.

Jeff Jarvis [02:52:46]:
Right. And they have tours of it and the. And the children of the building. So the family who run, they were just a plain old family who did this and they just can't. Can't take care of it anymore. So they're selling it for $25 million.

Leo Laporte [02:52:58]:
They built it for $37,500.

Jeff Jarvis [02:53:03]:
Yeah.

Paris Martineau [02:53:03]:
So beautiful.

Jeff Jarvis [02:53:04]:
It's amazing. Wait, they can't.

Benito Gonzalez [02:53:06]:
They can't turn Airbnb profit off of this.

Paris Martineau [02:53:08]:
Come on.

Jeff Jarvis [02:53:10]:
I mean, 25 mil will be a little better than that.

Paris Martineau [02:53:12]:
Yeah, probably.

Leo Laporte [02:53:14]:
I bet you anything the person who buys it for 25mil is considering the Airbnb income and the rental.

Paris Martineau [02:53:20]:
I mean, I don't. I don't think that if you're spending $25 million on a property, you're going to make that mortgage by Airbnb it out.

Leo Laporte [02:53:27]:
No, maybe not.

Jeff Jarvis [02:53:29]:
The other, the other link you were starting to go to has. Has. Is. Is the. Is the.

Leo Laporte [02:53:32]:
Their official site.

Jeff Jarvis [02:53:33]:
Yeah, the official site.

Leo Laporte [02:53:35]:
That's pretty cool and gorgeous.

Jeff Jarvis [02:53:38]:
Yeah, isn't it?

Leo Laporte [02:53:41]:
Very mid century modern.

Jeff Jarvis [02:53:42]:
There's a book about it. Case study house number 22.

Paris Martineau [02:53:47]:
It's been made with his own DIY schematic model of how to build a complicated site. Includes blueprints, floor plans and sketches by Pierre Koenig, as well as Julius Schulman's renowned photograph.

Benito Gonzalez [02:53:59]:
I mean, it's really that LA view that makes it though, right?

Leo Laporte [02:54:02]:
That's the classic.

Jeff Jarvis [02:54:03]:
Yeah, but it's that belief that you're hanging over la.

Leo Laporte [02:54:06]:
I wonder how many movies have been made in that story.

Jeff Jarvis [02:54:10]:
Say something about that.

Benito Gonzalez [02:54:12]:
IMDb can tell you.

Jeff Jarvis [02:54:16]:
Really? It does. Locations. I didn't even know that.

Benito Gonzalez [02:54:18]:
Probably for famous locations. For famous locations. I'm pretty sure there's something.

Paris Martineau [02:54:24]:
I think they say acceptable commercial uses are commercial and movie Production, commercial photography and advertising, documentary filming. But they don't allow private parties, meetings, weddings, receptions, private personal, graduation, engagement, wedding photography, stock photography, and filming private or commercial events.

Leo Laporte [02:54:41]:
That makes sense.

Jeff Jarvis [02:54:42]:
The house made historic appearances in film, television and music videos, including Playing by Heart 1998, Galaxy Quest 1999 and Nurse Betty 2000. Oh, 1999 was declared a historic cultural landmark, as it well should be. In 2013, it was listed on the national registry of historic Places.

Leo Laporte [02:55:01]:
Well, ladies and gentlemen, we've certainly learned something on the show today. I want to thank our guest, Ahmad Mustafa. What have we learned?

Jeff Jarvis [02:55:07]:
Joining us, what's one thing we've learned today? Leo?

Leo Laporte [02:55:09]:
One thing we've learned today, we're about a 5050 chance of AI doom. So we've also learned that Paris makes a hell of a good stuffing, likes donuts, worked at a donut shop and gets sleepy when she goes home.

Paris Martineau [02:55:26]:
It's true.

Leo Laporte [02:55:27]:
Good luck with the flaming turkey. Tomorrow I will share with you guys.

Paris Martineau [02:55:31]:
A video I was gonna say on Blue sky and Twitter at Paris Martineau on Twitter or at Paris NYCN BlueSky and I will send send it to you two and Benito and Anthony individually in our group chat.

Leo Laporte [02:55:44]:
We've also learned that Florida is awesome and has the most natural springs of any state in the union. So there.

Jeff Jarvis [02:55:54]:
Take that.

Paris Martineau [02:55:56]:
A fun Florida fact is that I was going to a coffee shop the other day. It was next to a taekwondo and jiu jitsu studio. And on the front of the jiu jitsu studio was a big sign that said, said concealed carry is permitted here. And that raises a lot of questions for me.

Leo Laporte [02:56:13]:
Why do they need.

Paris Martineau [02:56:15]:
What if you're practicing and you wrestle someone who's got a gun on them.

Leo Laporte [02:56:20]:
That'S, you know, concealed? That's one of the things they learn is how to disarm a fella. It's one of the things I, you know, most of the time you see signs that say, don't bring arms in here.

Paris Martineau [02:56:33]:
Yeah, that's what I thought the sign would be because it's a child Jiu jitsu academy.

Leo Laporte [02:56:41]:
Oh, man, you can imagine a caked up dad coming in there and being upset about his son's dad.

Paris Martineau [02:56:46]:
Be like, is that a dump truck behind you?

Jeff Jarvis [02:56:49]:
God.

Paris Martineau [02:56:49]:
Or do you have two concealed Gary guns?

Leo Laporte [02:56:54]:
Paris Martineau writes for Consumer Reports. She's an investigative reporter there. Such a pleasure, Paris. Have a wonderful Thanksgiving.

Paris Martineau [02:57:02]:
You too.

Leo Laporte [02:57:03]:
Stand well away from the flaming oil.

Paris Martineau [02:57:05]:
I will, but I'll stand close enough to get the shot.

Leo Laporte [02:57:08]:
Yeah, you can zoom in nowadays, you got that Zoom lens. Yeah. Safe distance. Thank you. Jeff Jarvis, the author. I didn't. I did the introduction there at the beginning of the show. Did I mention the Gutenberg parenthesis, The magazine, all those great books that you've written and pre order Hot type, get Hot type all@jeffjarvis.com we do intelligent machines every Wednesday right about 2pm Pacific, 5pm Eastern, 2200 UTC.

Leo Laporte [02:57:36]:
We usually begin with an interview and then we just kind of mess around for a couple hours after that. Next week, Dr. Anthony Vinci. Jeff, you're not gonna be here, I guess next week. I'm sorry. We'll miss you. Oh my. He is the author.

Leo Laporte [02:57:52]:
Anthony Vinci is the author of a new book, the Fourth Intelligence Revolution. Should be very, very interesting. We will talk about AI and of course you can watch that live as it happens. We stream it to the Club Twit members in the Club Twit discord. In fact, if you're not a member, join the club. So you get that behind the Velvet rope. Access TWiT TV Club TWiT. Still a month before that coupon expires.

Leo Laporte [02:58:19]:
We've got a great 10% off coupon right now when you sign up for a year. TWiT TV Club TWiT. We also stream for everybody on YouTube, Twitch, Twitch, X.com, facebook, LinkedIn and Kik. You can watch us there after the fact on demand versions of the show available at TWiT TV IM. We also of course put it on YouTube. There's video there you can share with friends and family. That's a nice way to share little clips or subscribe in your favorite podcast player. Choose the audio or video feed you get to choose or both.

Leo Laporte [02:58:51]:
Leave us a five star review. Paris promises that if there is a very interesting funny one, she will do a dramatic reading. And if you put extra bonus points if you can get an analogy about bamboo in it.

Paris Martineau [02:59:05]:
Okay, man. If you get one with bamboo in it, I will interrupt the show more than once. That is mine.

Jeff Jarvis [02:59:13]:
There was, in part three, there was a Twin Peaks reference which I know.

Paris Martineau [02:59:16]:
Went to, but it was incorrect. I don't think he's seen. Because you wouldn't have a. A dead bear incident in Twin Peaks. That's just.

Jeff Jarvis [02:59:25]:
Oh, come on. It's a little license here.

Leo Laporte [02:59:30]:
Is it, is it, is it? Thanks for joining us everybody. We'll see you next time. Have a great Thanksgiving. If you're in the US on intelligent machines, Bye bye.

Paris Martineau [02:59:43]:
I'm not a human being, not into the animal scene. I'm an intelligent machine.