Intelligent Machines 853 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. Jeff's here. Paris is here. Our guest, Craig Silverman, is an expert in disinformation, a journalist and a co founder of Indicator Media will talk to him about disinformation. He's the guy who coined the term fake news. Yes. He was the first to. Save it.
Leo Laporte [00:00:17]:
Say it. And then a lot of real news in the AI world. That's all coming up next on Intelligent Machines, podcasts you love from people you trust. This twist. This is Intelligent Machines with Paris Martineau and Jeff Jarvis. Episode 853. Recorded Wednesday, January 14, 2026. All the clocks were wrong.
Leo Laporte [00:00:49]:
It's time for Intelligent Machines, the show. We cover all the latest news about AI robotics and all those smart je jaws surrounding you in your home. Unless they're run by Verizon, in which case they're just dead and not so smart. Paris Martineau lives in Verizonville. She's an investigative journalist.
Paris Martineau [00:01:10]:
I do.
Paris Martineau [00:01:10]:
I haven't been able to send or receive a call for four hours.
Leo Laporte [00:01:14]:
That's so weird.
Paris Martineau [00:01:15]:
Not even to 91 1.
Leo Laporte [00:01:17]:
Paris, don't you have a landline?
Paris Martineau [00:01:19]:
I don't, but I should.
Leo Laporte [00:01:21]:
Come on, Jeff Jarvis, you sound like an old guy here. Jeff Jarvis.
Jeff Jarvis [00:01:24]:
I feel like one too. Yeah.
Leo Laporte [00:01:26]:
Professor emeritus of journalistic innovation at the Craig Newmar Graduate School, State University, was a visiting fellow. And that's SUNY Stony Brook. Author of the Gutenberg Parenthesis and magazine. I feel it's unfair, Paris, that he gets such a long intro and all we just say is and a song. And a song. We got to do something for you. We got to do something for you.
Paris Martineau [00:01:50]:
We gotta. We gotta balance it out.
Leo Laporte [00:01:51]:
Yeah.
Leo Laporte [00:01:52]:
Nick Cage, lover, historian and expert.
Paris Martineau [00:01:56]:
It's true.
Leo Laporte [00:01:57]:
Very important document troller and conspiracist.
Leo Laporte [00:02:01]:
Yes, the. The Queen of the Red String. How about that?
Paris Martineau [00:02:05]:
That's pretty good.
Leo Laporte [00:02:06]:
Yeah, we got a great guest. And actually, Jeff, you have brought him to us. Craig Silverman is here. Canadian journalist, one of the world's leading experts on disinformation on fake news. Co founder of A Brain, Relatively, a young new newsletter and website dedicated to exposing digital deception. It's called Indicator at Indicator Media. He was at ProPublica. Before that BuzzFeed News.
Leo Laporte [00:02:36]:
He's a founding editor of BuzzFeed Canada. And my favorite site name. Regret the error.
Craig Silverman [00:02:45]:
We're going back. It was a while ago.
Leo Laporte [00:02:46]:
Going back in time. 10 years ago.
Leo Laporte [00:02:49]:
Guess who wrote the foreword in that book? I'm honored to say that'.
Leo Laporte [00:02:54]:
So, Jeff, I should have let you do The Craig and I do go.
Leo Laporte [00:02:58]:
Way back early in the, in the whole disinformation days. Craig wrote a phenomenal regular. It was a blog, right Craig?
Craig Silverman [00:03:06]:
Yeah, back in the. Launched in 2004.
Leo Laporte [00:03:09]:
Yeah.
Leo Laporte [00:03:10]:
It collected error corrections, Regret and they're.
Leo Laporte [00:03:16]:
Really the name we regret the Error, the error.
Leo Laporte [00:03:18]:
And they, they're, they're, they're great reading, they're just wonderful. But it also said a lot about journalism, that journalism used to try to correct its. Craig, I gotta ask you first, in your memory, do you have a single or two favorite corrections from those days?
Craig Silverman [00:03:33]:
There was one where Reuters, which was rare, you didn't see a lot of wire services making big mistakes. But Reuters, they had an amazing autocorrect error where I think instead of referring to Queen Elizabeth, they sort of substituted. Or actually it was the other way around. Instead of referring to the queen bee of a, you know, a colony that referred to Queen Elizabeth being able to lay up to 2,000 eggs per day. I can't believe I remember that, like off the top because it's been a long time. But yeah, that one stuck with me. There was a headline also where they called, they had to. Their correction was, you know, there was a headline on Tuesday that mistakenly referred to New York as Jew York.
Leo Laporte [00:04:21]:
Oh geez.
Craig Silverman [00:04:22]:
It's a lot of stuff like that.
Leo Laporte [00:04:23]:
Oh, geez. Well, there are a few people in the world who actually think that, so maybe it wasn't such a typo. Well, the media is still making errors, but now they're doing it on steroids. And I think it's appropriate to bring you onto a show about AI because I imagine AI is now one of the best ways to create disinformation.
Craig Silverman [00:04:46]:
Yeah, well, there's a great quote from the researcher Renee Diresta where she talks about social media taking the cost of distribution of information to zero and then AI taking the cost of generation to zero. So now you can reach huge amounts of people, of course, very easily on social networks. But it used to take time to sort of come up with, oh, what's our narrative? You know, how are we going to generate, how are we going to create a video, how are we going to create images to sort of spread, you know, this lie or you know, just sort of make our argument. And now, you know, what might have taken hours, days, needed expertise in certain areas is, you know, matter of minutes and prompts. And so we sort of see that in a variety of different manifestations. You know, some might be kind of like politically oriented state backed kinds of campaigns and others is Just, you know, scammers and other people who are spreading false information, creating deepfakes of prominent people to sell products and do other things. And it is trivial now to just generate stuff that looks pretty convincing and then to get it distributed quite quickly.
Leo Laporte [00:05:57]:
Where does scale for this? Because spam, it costs nothing to send out. And even though it may be a low return on your investment, since the investment is so low, you get the benefits. This is kind of weaponized spam.
Craig Silverman [00:06:16]:
Yeah, yeah, and it's spam. You would send out millions or tens of millions or hundreds of millions of emails, and it would hit a person individually. And in the dynamics of sort of the social environment, if you get it and it. It reaches a certain amount of people and they interact with it enough, then sort of the systems themselves take over.
Leo Laporte [00:06:34]:
For it, you know, that's key, isn't it? It's vital.
Craig Silverman [00:06:37]:
Yeah, yeah, this is it. And so you sort of get it. You get the systems working for you if you target it. Well, this is my cat's tail, Rosie, who always.
Paris Martineau [00:06:46]:
Hi, Rosie.
Leo Laporte [00:06:48]:
We saw her come down the stairs.
Craig Silverman [00:06:49]:
So that's it. She sort of. I thought she was going away and then she's like, no, I'm gonna make an appearance. And you know, she framed all tail.
Leo Laporte [00:06:59]:
Our cat's named Rosie, too. Do you spell it I, E or Y? I, E. Yeah, us too.
Craig Silverman [00:07:04]:
Yeah, yeah, Nice.
Leo Laporte [00:07:05]:
We have a Rosie that's all black. In fact, Rosie's the complement of your Rosie. Our Rosie.
Craig Silverman [00:07:10]:
There you go. Yeah, she's mostly white. And then we have a black dog, so the dog is scared to come down to the basement. So it's only Rosie who comes in.
Leo Laporte [00:07:18]:
Oh, cats are never scared. When there's a camera on, they immediately show up. Paris, you were going to ask something I interrupted. Go ahead.
Paris Martineau [00:07:28]:
Oh, I was gonna say, I think we've already. I mean, I think there was one really interesting example of the way that AI has kind of both lowered the cost of creating and disseminating this information. We saw this the other week with Casey Newton over at Platformer, had a good breakdown of debunking kind of this AI food delivery hoax that had fooled Reddit and went a bit viral. If you guys recall, this was, I believe, a Reddit post that basically claimed they worked for an unnamed food delivery app and they were up to no good and basically doing all of these terribly evil things and that they had proof. And this unnamed redditor, Casey, like a lot of other journalists at the time, reached out trying to confirm whether or not this was true. And the person sent back a photo of their ID at Uber Eats, as well as internal company documents that really seem to spell all this out in great detail. And Casey says in his post, she's like, you know, I excited to see both the ID as well as. Then you see this document.
Paris Martineau [00:08:37]:
And that seems impossible to fake. But both were completely fake, it turns out, and generated, I assume, very quickly using something like Gemini.
Craig Silverman [00:08:47]:
I think it actually was Gemini, and I think that was kind of the mistake that this person's made. And I hate to sort of say this like I'm giving people a playbook, but you know, Gemini, Google. So one of the things to sort of combat the onslaught of AI generated content is that a lot of platforms, technology companies, you know, the Adobes of the world, the Metas of the world, the OpenAI's, they've all signed on to sort of create and insert metadata content credentials. When you generate an image with gemini or with ChatGPT, there's supposed to be some data embedded in it that, you know, could then put a label when it gets shared on Facebook saying, hey, this was generated with AI. And Google has actually gone a step further where they've got their own kind of proprietary watermarking technology where called Synth id, where not only are they embedding the metadata, which unfortunately sometimes doesn't end up in the file, it's very uneven right now. But they also have this kind of watermarking. They do, and then they have detection tools for things that have been created in Gemini. So all that to say, I believe that this person had used Gemini for summer, all of those documents and images.
Craig Silverman [00:09:56]:
And I think that was one of the ways that Casey was able to sort of have a really high level of confidence that this stuff was AI generated. Because the truth is that a lot of the AI detection tools you can't really rely on, they don't work equally across all the different models. And so where it's cheap, easy, fast to generate the stuff, to actually verify it, to debunk it, to dig into it, is still quite labor intensive because there's no simple oh, aside from if you get lucky with sincere Synth id, there usually isn't a really fast way to know instantly.
Paris Martineau [00:10:30]:
Yeah, and it's quite interesting also, because once this had come out and Casey had kind of done his blog about it, then a reporter from NBC shared. I'll share the image in the Discord Chat here, that the fake ID that the it was an Uber Eats ID that this alleged delivery employee had sent to Casey it was seemingly generated based on a ID from an NBC reporter. That NBC reporter had sent the same person trying to corrupt, corroborate that they were a legit reporter.
Leo Laporte [00:11:07]:
Here's my id. Let's see yours. You know, interestingly, it's almost identical, just the color is different. Wow, that is kind of a giveaway.
Leo Laporte [00:11:17]:
But. But when you want. It's the thing about us journalists is we want the story. We don't want to ask too many questions. I mean, yes, we should. We always should. And that's where Craig's heritage in journalism comes back, is that you're better off when you make a mistake to correct it. But we never thought that way.
Leo Laporte [00:11:33]:
We wanted to hide. We wanted the story to happen. And it's hard to ask that added question.
Leo Laporte [00:11:39]:
Well, even harder for us as individuals, because how are we supposed to do this kind of work and figure this stuff out? And we're flooded with this information. So much so that even if you wanted to and could, you know, figure it out, you probably wouldn't have the time to do so.
Craig Silverman [00:11:58]:
And that's, I think, baked into it to a certain extent is, I mean, there's an old saying in journalism, something being too good to check, which is not a sincere saying. It's more just like, oh, that's an amazing story. That's almost too good to check. You're not supposed to sort of not check it. Right.
Leo Laporte [00:12:13]:
You're not supposed to admit it either.
Craig Silverman [00:12:15]:
You know, and so that is sort of the scenario of you can craft these things and you can, you know, drop it in the right place and it can sort of catch, catch like wildfire. And of course, I mean, look, there's human dynamics here as well. Of we read something that aligns with what we may suspect about, you know, food delivery companies and apps. It aligns with our suspicions, our beliefs, and we're just more likely to accept it as human beings. And, you know, we read something that goes against that, and it's not just that we might be skeptical of it, we might actually unconsciously look for reasons to dismiss it. And so when people are crafting these kinds of things, you know, this might have just been a one off prankster, but when you think about people and organizations that are sort of thinking about this as campaigns, as many of them do as jobs, that they go in and work 9 to 5, whether they're in a scam compound or in, you know, a disinformation operation funded by a state, you know, they're really strategic about it and they think about how to craft these narratives. And then they have the ability to just iterate really, really rapidly with the images, the text, the video to find out what's going to actually work and hit with people. And you know, that's kind of unfair as an advantage.
Leo Laporte [00:13:25]:
And to some extent you can also blame the platforms. You've done some work that on the Meta issue with them making $7 billion a year from scam ads. They're not strongly incented to fix this. They want us to go out. Right.
Craig Silverman [00:13:43]:
There's, you know, the thing that Meta will say is, listen, we don't want our users being scammed. But the real is, for me, I did a story, I guess it was, I don't know, like seven years ago now about just like one shady marketing company in San Diego that they alone had spent more than, I think it was $50 million on scam ads placing them on Meta.
Leo Laporte [00:14:08]:
Oh my God.
Craig Silverman [00:14:09]:
And so, and now it's actually gotten so much worse than that. It is sort of industrialized. You know, I mentioned scam compounds. I mean, it's in Myanmar and some other places around the world. There are people who have been human trafficked who end up basically having to work as forced laborer scamming people in the US and Canada and other countries. The defenses and the oversight, it's just not there. And unfortunately, when it comes to the ads, there is a money factor for the platforms is that scam ads do make them money and as much as it does rip off their users. And that is probably a bad thing.
Craig Silverman [00:14:41]:
The recent reporting from Jeff Horowitz and Reuters show that Meta has been aware of it and it's, I think, has failed to take a lot of really important steps to turn it off. I hope that that stuff coming to light might start to change things.
Leo Laporte [00:14:52]:
It's gotta be hard if you're making $7 billion a year to say, well, we really shouldn't, we shouldn't be doing that. That's a lot of money to say no to. I mean, they shouldn't obviously, but I can maybe understand a little bit. Indicator says stay ahead of digital deception. You provide not only news, but tools as well to help people avoid disinformation. Yes.
Craig Silverman [00:15:18]:
Yeah, I mean, we're sort of a mix of one, we do reporting and investigations exposing different types of digital deception. So it might be disinformation, it could be, you know, fake engagement, lots of different types of tactics that people are using these days. But we also want to kind of share the recipe, I guess. And so when we do investigations we include, here's all the tools, here's all the techniques, here's everything we did to report this story. You could do that yourself. And then we also, you know, for professionals who work in journalism research or just really motivated consumers, you know, we provide workshops and kind of how to guides about how to do this type of digital investigative work. Because I think there is a need to kind of raise the level for everybody out there, as you said, Leo. I mean, the reality is everybody has to kind of do their own fact checking now.
Craig Silverman [00:16:06]:
So much information coming at you, it's not being filtered, you know, through professional sources, which had some advantages but also had disadvantages, of course. And there's more responsibility on all of us now to try to figure out what we're seeing. Otherwise we just end up on autopilot and that that can lead down some bad paths, particularly around stuff like scams and what have you.
Leo Laporte [00:16:25]:
Well, you already said that some of the tools used to detect AI fakes are unreliable.
Craig Silverman [00:16:30]:
Yeah.
Leo Laporte [00:16:32]:
What. Oh, you call this osint? What? OSN tools work? What should people be looking at? You have a whole workshop video, I might add, if you pay to subscribe to indicator that you can watch. But I'm going to try to get it for free out of you here.
Craig Silverman [00:16:48]:
Yeah, we'll do that. Yeah, we do God's work.
Leo Laporte [00:16:51]:
God's work.
Craig Silverman [00:16:53]:
We do a monthly workshop basically for members, and the topic depends on whatever we've reported on or looked at that month.
Leo Laporte [00:17:00]:
Tools you use, in other words.
Craig Silverman [00:17:02]:
Absolutely, the tools that we use. And I also basically, like, I'm just obsessed with this world and so I'm always testing new ones. And I'm sort of saying to people, hey, I'll nerd out on this and I'll tell you which ones are good. And you know the challenge when it comes to images and video. So first of all, the simplest tool is reverse image search, which is free and widely available and been around for a while. Kind of thanks to JLo. And the revealing dress she wore at, I think it was the superworld. Was it the MTV Music Award? I don't know.
Leo Laporte [00:17:33]:
Oh, wait a minute. Revealing dress. I was thinking of the wardrobe malfunction. That was somebody else. Yeah, the revealing dress, that was a different one.
Craig Silverman [00:17:40]:
No, that was not. And yeah, like the legendary story was that so many people were searching about her in the dress that people at Google were like, oh, we should do for images what we do for text. And so you have the ability. And the beautiful thing now is Google Lens is built into the Chrome Web browser built into the Google app. And you can just hold this up at an image you're looking at and potentially get some context of it. Oh, you know, they're saying that this is, you know, a person who was here in this city today, but actually this image appears to be six years old. And you can get some context on that image. And sometimes an AI image or video has circulated enough that there actually is a way to get kind of a report from a reverse image tool like Lens or Tineye that would say, hey, you know, here are places where people have been talking about this.
Craig Silverman [00:18:30]:
And so that's one part of it is try to find the origin. Where did this thing come from? And another piece is sort of just old school verification techniques of the when and the where. So can I figure out when this photo or this video might have been taken? Okay, it's nighttime. I see a street sign, I see a business sign, and it's got an area code in it that suggests it's in New York City. And you start to just kind of figure out, are there real artifacts in this image or video that are tied to a real place? And can I confirm that that's what this place looks like? If there's a prominent person in it, can I figure out what their schedule was or reach out to the representative? And so the tools kind of can get you sometimes pretty far. But at the end of the day, there's a lot of traditional kind of visual observation work of what you're looking at, what's in there. Can I figure out people in it? Can I talk to them? Old school kind of journalism. But I mean, reverse image search is a great tool.
Craig Silverman [00:19:29]:
The other one I mentioned before is metadata, which is that videos and images, when they are created, they have data embedded in the file, and there are free metadata readers where you can sort of drop one of these into. The downside of that is most times when a photo or video is uploaded to a social network, the metadata is stripped out because they don't want that extraneous data. But if it's from Telegram, they keep it there, which is great. Or if it does have this sort of AI metadata in it, sometimes you might still find that metadata there in a metadata reader that actually tells you, oh, this was generated with Google Gemini, this was generated with this or that. And hopefully we're going to see platforms integrating that more.
Leo Laporte [00:20:15]:
Paris, have you ever used anything like this in your researches? You probably, yeah.
Paris Martineau [00:20:22]:
I've also used, I mean, Craig, I think, has a lot of very interesting resources and guidebooks that I've always used.
Leo Laporte [00:20:29]:
Oh, okay.
Paris Martineau [00:20:30]:
Information handbook and things like that. They have lots of great resources just for any sort of kind of OSINT research.
Leo Laporte [00:20:40]:
Craig, we've talked about content credentials before. Who was that we interviewed? I forget. But this is the opposite direction where you actually try to prove that the photo that you've posted is real and what camera it came from and all of that. That's kind of the opposite of AI detection. It's asserting that it's real. Is that taking off or is this kind of dead in the water?
Craig Silverman [00:21:04]:
Well, the positive thing about it is that there's a decent amount of industry buy in, I would say, on this stuff. So you do see some of the larger platforms and tech companies and the Adobes of the world and camera makers and other things.
Leo Laporte [00:21:16]:
Yeah, I have a Leica that does it.
Craig Silverman [00:21:17]:
Yeah, yeah, right. And so they've kind of signed onto this. Where it gets tricky is when it kind of hits sort of of the average person, I suppose, in that. Yeah, like you can have a high level of confidence that Leica, if it's signed onto this and has says it's in its cameras, it's probably putting that in there. But, you know, yeah, a lot of.
Leo Laporte [00:21:40]:
Money to get that.
Craig Silverman [00:21:42]:
And you know, but the problem is when it reaches the member of the public, is that actually going to be available for them to look at? And so we actually did kind of an audit on this earlier this year and. And we found like, the performance of platforms was really, really poor. So we basically generated over 100 images with different LLMs. We generated over 100 videos similarly. And then we uploaded them. We checked to make sure the metadata was there first of all. And then we uploaded them to a whole bunch of different platforms like Pinterest, Instagram, different ones. And you know, like the, at the end of the day, what we found was that for the most part it was about only 30% of content that had those credentials in.
Craig Silverman [00:22:26]:
It actually showed with a proper label when it reached the platform. And so that's them labeling AI generated content. Ideally they should also be able to recognize and say, oh, we have content credentials that show taken with this camera on this date. But the first bar to clear was that they said we will label AI generated content. And they're just not doing that yet.
Leo Laporte [00:22:50]:
You're credited at least Claude AI credits you with creating the term fake news. Is that you. Did you do that?
Paris Martineau [00:22:58]:
Well, is that Claude fake news or.
Leo Laporte [00:23:00]:
Is Claude faking that?
Craig Silverman [00:23:03]:
So, no, Claude's. Claude's kind of on point. And I'm a cloud customer. I have to admit, I like it. It's good. You know, I use it like crazy.
Leo Laporte [00:23:12]:
I'm going to talk of all the ones.
Paris Martineau [00:23:14]:
I. I think Claude's pretty based.
Craig Silverman [00:23:16]:
There's. There's something about it.
Leo Laporte [00:23:18]:
Huh.
Craig Silverman [00:23:18]:
That's like. I mean, yeah. I also like that it's, It's.
Paris Martineau [00:23:22]:
It feels a bit more anti glazing than the average LLM. I don't know whether that could just be glazing in and of itself, because it recognizes that I could want that capability. But.
Craig Silverman [00:23:36]:
Yeah, no, I'm with you. There's something. Yeah, there's something about it. And, you know, maybe it's. It's a bit of the branding and the feel and all that and their messaging, but, you know, the stuff that comes out of it seems to. Because I have. That. I have access to the paid models for Gemini and I'm leaning more towards clothes these days.
Craig Silverman [00:23:54]:
But all that being said, he says.
Leo Laporte [00:23:56]:
It'S the Canadian way, by the way.
Paris Martineau [00:23:57]:
Claude.
Leo Laporte [00:23:58]:
Claude, we were told that we have to. Joey says you got to say Claude. Okay.
Craig Silverman [00:24:03]:
I lived in Quebec for a while, so. Yeah, absolutely. And maybe that's also part of the appeal.
Paris Martineau [00:24:10]:
So you originated the term fake news. Claude is correct.
Craig Silverman [00:24:15]:
Yeah. So I usually say I sort of popularized it and then Trump kind of took it, you know, and so, like, quickly, the Backstory is, in 2014, I was doing a research project at the Tao center at Columbia, where I was studying how viral claims were spreading and particularly how news organizations and news websites were covering. So there'd be a report somewhere or there'd be a claim on Twitter. And then me and a research assistant, we would look at who covered that and how they wrote about it. Did they report it as true? Did they report that, oh, it could be true? Did they say it's false? And we would look at the associated social engagement to try and sort of start to quantify, like, how much were news organizations consciously or unconsciously spreading things that turned out to not be true? How much were they jumping on rumors to try and get social engagement around them? And so as part of that research, I was out there hunting for claims that were in that gray area of maybe true, maybe false, as well as stuff that was just totally fake, totally false. And I came across a cluster, like more than a dozen websites that all looked like real news websites. They had news articles on them, they had bylines, they had photos for their authors. The authors often had bios, but everything was 100% made up and it was all about the ads on the site.
Craig Silverman [00:25:36]:
So they would come up with a crazy headline and story they would post on Facebook, it would go viral, people would click. Because at that time Facebook still had showing people links and people clicked like crazy. This is 2014 and it was all about money. In the research report I ended up publishing in early 2015, I described them as fake news websites because that's what they were to me, literally just a fake news website and defined it as 100% false, intended to deceive and economically driven. And that was the definition I used from 2014 all the way through to the end of 2016 as I was doing lots of reporting at that time then for BuzzFeed News about the false pro Trump stuff. That was coming from the guys I reported on in North Macedonia who were running those websites which, which ended up being a big story at the time. And yeah, that was when the term sort of got out there and was popularized. And then starting, I guess in About January of 2017, Trump decided to take ownership of the term, I would say.
Leo Laporte [00:26:39]:
And then.
Leo Laporte [00:26:39]:
Well, that's the best way to. You co opt those terms and then you can make them your own. That's the best way.
Leo Laporte [00:26:44]:
Well, that's what he did. There's somebody we used to work with, all of us in the disinformation space in the back named Claire Wardle, who's now at Brown University. I think she tried to lecture us all and say, stop using it. Yeah, because it's being used by Trump and you don't want to play into that. But it was a lost cause. It was out there in the, in the lexicon.
Leo Laporte [00:27:06]:
We're talking to a disinformation researcher, Craig Silverman. His publication, his newsletter is a great wealth of information, not only about what's going on out there, but also how to detect it yourself. Indicator media is the place to go. You talk a little bit about the gender consequences of fake images and of course, Grok's been in the news. Malaysia, Indonesia, blocking Grok. The UK is considering blocking Grok because of the non consensual deep fakes it's been the been generating. Of course, actually they want to block X because that's where they all end up. And just earlier today, Elon, this is, by the way, the other side of fake news is lie.
Leo Laporte [00:27:51]:
Right. And if you lie, there's was. Ben Franklin says the lie goes all the way around the world while the truth's just putting its pants on. Lies somehow manage to get lodged in the brain and people believe them. Just this morning, Elon said, I'm not aware of any naked, underaged images generated by grok. Literally zero. Obviously, GROK does not spontaneously generate images. It only does so according to user requests.
Leo Laporte [00:28:18]:
Well, yeah, when asked to generate images, it'll refuse to produce anything illegal as the operating principle. By the way, our friend I can't show them. Pliny the Liberator immediately posted a bunch of images he liberated. They got Grok prompt broke Grok to do so. Yes, you can clearly do it. This is the other side of this is that the people with a vested interest in disinformation use that platform to lie about it, to spread more disinformation that makes people question their own eyes, their own brains.
Craig Silverman [00:28:58]:
Yeah. Musk's takeover of Twitter obviously really changed the dynamics of the kind of content that was. I don't follow him, but I get his tweets whether I like it or not.
Leo Laporte [00:29:08]:
Yeah, and it's interesting. I think he was smart enough to realize if you control this platform, you control a bully pulpit. This is very, very valuable. This was worth $44 million, money he can easily afford because it gave him such power.
Leo Laporte [00:29:26]:
Craig, I'm curious as we go back. Go ahead.
Craig Silverman [00:29:28]:
Oh, I was just gonna say quickly. I mean, the other thing that I think made the purchase worth it for him was that, you know, with the growth of AI and LLMs, you need training data and you need access to an ongoing stream of information. And, you know, that's why now X is part of xai, because that's, you know, it's just feeding into and presents the models there. So in the end, I don't think that's what he originally planned and said. This is exactly what will happen, but I think it's worked out quite well for him.
Leo Laporte [00:29:59]:
So, Craig, I'm curious to go back to 2016 and the rush about fact checking. You were very much involved. Your co founder Alexios was very much involved. We both talked to Meta at the time about their efforts to hire fact checkers. There was a fact checking organization. I'm curious about your thoughts about lessons learned. What did that movement do right and wrong? What should we know based on that experience?
Craig Silverman [00:30:30]:
Yeah, I mean, it's really interesting because for a long time, fact checking was very, very small, very, very niche. Nobody really didn't pay a huge amount of attention to it. And it was often focused on. It was mostly an American phenomenon, and it was mostly focused on checking the public statements of politicians and other big public figures. And a little bit even before the end of 2016, a little bit, you would see the Politifacts and others starting to be like, yeah, we should probably check these viral claims, whether they're from a public figure or not. But the stuff going crazy on Facebook. But what happened at the end of 2016 was a massive change where Meta, then just known as Facebook, went to the international fact checking network that my partner Alexios was the executive director of at the time and said, hey, we are trying to do something to improve the quality of information on our platform. What if we paid fact checkers that have been certified by you guys to kind of check things? And so this basically, over the ensuing whatever, eight years, Meta became the biggest single funder of fact checking in the world, which is crazy to think about, but also not good for the sustainability of fact checking and also not good for the perception of fact checking, I think as well, where you are seen as kind of working for Meta and tied to the platform that is actually one of the biggest sources of this stuff.
Craig Silverman [00:31:53]:
So on the one hand, Meta taking some responsibility, a good thing. On the other, it became, in some ways a very lucrative albatross around the neck of, I think, a lot of fact checkers. And as Meta abandoned the US Program early last year in a pretty spectacular fashion, throwing at the fact checkers on the bus. Mark Zuckerberg calling them bias. Now, I think fact checkers around the world realize this is continuing for us, but it could end at any time. And this is a company that will throw us under the bus when the political movement makes it for them. And so I think one of the problems that fact checking has always had is that you seem like a finger wagging. No fun.
Craig Silverman [00:32:33]:
You're just taking the air out of the balloon. You're no fun at a party. Exactly. And this is honestly, in that 2014 research project. This is what I saw. And I forget if I put this in the report or not, but this, this will only make sense to people who watch the show Cheers. I talked about, like, the Cliff Clavin effect.
Leo Laporte [00:32:52]:
Yeah.
Leo Laporte [00:32:53]:
You know, many people talk about the Cliff Clavin effect, but very few people know it came from Argentina in 1823. Yeah.
Craig Silverman [00:33:01]:
Harris is like, what the hell are you guys talking about?
Paris Martineau [00:33:04]:
No, sometimes this happens on this show and I just learned to smile and nod.
Craig Silverman [00:33:13]:
Yeah, it's. And I mean, Jeff is a longtime TV critic. You know, he knows Cliff Clay.
Paris Martineau [00:33:18]:
He knows every television.
Craig Silverman [00:33:20]:
Absolutely Right. And so he was the know it all on the show.
Leo Laporte [00:33:24]:
And the. And the.
Craig Silverman [00:33:25]:
And nobody, you know, he was annoying and, and I would see it when somebody would debunk something on Facebook or on Twitter or whatever, people were like, oh, you're spoiling the fun. And so I think that's a problem the movement has always had, is never finding a way to sort of communicate in ways other than seeming like finger wagging folks. And I also think the problem, one of the problems with that partnership was that Meta sort of was the one that decided really everything. Like they paid, they produced fact checks, the meta got them. And Meta could decide when to label, how to label, when to remove a label, whether or not to reduce distribution, whether to remove something. But the fact checkers ended up being the ones being blamed, even though meta had the decisions. I think everybody who works in fact checking that I've ever met is well meaning, but there's definitely been some moments and some challenges and some problems. And I think the brand of fact checking has been really damaged over the last decade or so, probably.
Leo Laporte [00:34:24]:
Well, it didn't help when Meta fired all the fact checkers too. That kind of undermined the whole premise. Right.
Craig Silverman [00:34:30]:
In a way, it might be a liberation moment for some of the US fact check. Even though they just lost a bunch of funding like very suddenly to chart and have to find that sustainability financially, but also to be able to say we are not tied to them anymore is. It could be one of these things that in the long run maybe helps fact checkers chart a different path. But the reality is they've been under attack for quite some time by folks who have always labeled them as biased, regardless of what they put out there. And I think fact checkers, like, frankly, a lot of people in the disinformation community were never good at battling and fighting back. It's a weird maybe irony of it, except for, you know, some people occasionally are good at it, but a lot of times they just want to do the work and not necessarily get caught up in the battle over it, but they get dragged in whether they want it or not.
Paris Martineau [00:35:27]:
I'm curious, I feel like a common criticism you often heard of fact checking was that, oh well, people don't. Even if you fact check something correctly, people don't believe the fact check or will dig in their heels. Did you see that in practice?
Craig Silverman [00:35:43]:
Absolutely. And actually it goes to Jeff's question of what does it sort of get wrong? I mean, for better or worse, the perception for a lot of people of fact checking is like the goal of fact checking is to change people's minds. Right? That's what I think a lot of Folks think. And fact checking is for the most part, like, if you're dealing with people who have an open mind, fact checking is shown to be effective. If you are talking about folks who have their mind made up and are very closely held on a belief or a particular issue, the idea that fact checkers are going to come in and write one article and that person is going to be like, oh, okay, yeah, right. I totally see it differently now. And so I think the expectations that fairly or unfairly, that the goal is to sort of change people's minds and influence how people think. I don't know that fact checkers necessarily do the work with that in mind, but it is something that I think a lot of of people feel is the goal.
Craig Silverman [00:36:38]:
And the truth is like, fact checking is not really good at changing deeply held beliefs. It's better as kind of an accountability measure. And some of the data that's out there is that it was actually effective around politicians pre Trump, where if they knew fact checkers were gonna be working around their race, they were shown by some studies to actually be less likely to kind of stretch the truth a bit more. Trump blew that out of the water. But that's in the pre Trump era where it was actually shown to be effective.
Leo Laporte [00:37:06]:
But Paris, that's also the issue of context. The folks that you sent us to at if Books Could Kill the podcast, Glenn Kessler, who wasn't the fact checker at the Washington Post, was their bet noir. Because he would proudly say, oh, I found this fact and whatever it is, and I've got the fact and would miss the larger context and would miss the larger effort at deception that occurred. And so fact checking can be misused. I think, as they, as they well pointed out.
Leo Laporte [00:37:32]:
You know, many people don't know that bet noir is an old French phrase. It refers to a black beak.
Craig Silverman [00:37:39]:
Who knew you had a good Cliff Clavin impression?
Leo Laporte [00:37:42]:
Don't get him started.
Leo Laporte [00:37:44]:
I live to be a know it all. You do coach. We do have to do a BT and a PT before Trump and post Trump when it comes to facts and disinformation. Because in many ways, I mean, he's shown, of course, that you can defy norms, you know, en masse. So in many ways he's shown that it doesn't matter. In fact, didn't Kellyanne Conway say that we're in a post fact era?
Leo Laporte [00:38:13]:
It's an alternative facts.
Leo Laporte [00:38:16]:
What are you so concerned about facts for?
Craig Silverman [00:38:19]:
Yeah, I mean, he ste rolled it. That was. He just realized I can brute force this System, they can't keep up with me. At a certain point, it's gonna keep washing over people and I can continue to kind of make the claims and statements I want and sort of get away with it.
Leo Laporte [00:38:36]:
Is it the old Russian flood, the zone model that he's using? Exactly.
Leo Laporte [00:38:41]:
There's a paper, the Rand.
Craig Silverman [00:38:43]:
Yeah, the Rand Corporation. The fire hose of falsehood. Yes.
Leo Laporte [00:38:46]:
Yep.
Leo Laporte [00:38:47]:
Fire. That's a better way to say it. Firehood of falsehood. But I doubt he'd read that Rand Corporation report. But somehow the strategists around him, this is his brilliance. I always intuitively understood this. I don't think it was intellectual. I think he just intuitively understood as from his years of lying as a real estate developer about how many floors his buildings had, he learned, you know, this works.
Leo Laporte [00:39:14]:
You don't have to follow these norms.
Leo Laporte [00:39:16]:
But Leo, isn't it also. It's a. It's a Russian fire hose of disinformation. But it was though, was. I think there was a Putin connection. It's a strategic. It's a tactic.
Leo Laporte [00:39:29]:
It's a tactic. I think what Putin does it.
Leo Laporte [00:39:31]:
There are people who were connected to it as a tactic.
Leo Laporte [00:39:34]:
Yeah.
Paris Martineau [00:39:34]:
I think there are a lot of different reasons why people end up employing what become the same tactics, whether or not they are conscious or carefully planning it out.
Leo Laporte [00:39:46]:
So what do we do to restore a faith in facts, to restore a faith in science? How can we get more?
Paris Martineau [00:39:54]:
How do we fix everything?
Leo Laporte [00:39:55]:
How do we fix this post? How do we get out of this post fact era?
Craig Silverman [00:40:00]:
Well, I'm glad you asked me because I have the answer and it's only going to take 20 seconds and that's it. And we're good.
Leo Laporte [00:40:06]:
Great.
Paris Martineau [00:40:07]:
Wonderful.
Leo Laporte [00:40:08]:
And $100 a year will help.
Craig Silverman [00:40:10]:
Boy.
Leo Laporte [00:40:11]:
Yeah.
Craig Silverman [00:40:12]:
You know, there isn't one fix. It's, you know, the sort of unsatisfying points around it are that it's, it's kind of a. It's a whole of society challenge, a whole of society problem, and it's interconnected and all those things that make it complex. But I do think for the average person, the starting point is just around some awareness of the dynamics of this information environment that we're all in, which is, you know, incredible abundance of information and much more open, which is great. More voices, more opportunity for people to participate. But it also means it's much easier to manipulate. And so it puts a little more responsibility on all of us when it's not just like a handful of professionalized media outlets which were never perfect and you're not saying that that was the great time when it was like that, but you had a smaller amount of choices and it was coming in a professionalized environment where it had failings and problems, but they were sort of more contained. And today you have so much more choice.
Craig Silverman [00:41:11]:
But that means you have a responsibility. You're not getting a hand holding from someone. And so the awareness of how much it is being manipulated. And that's why it's like, I don't just mention the political stuff. It's the financially motivated people, the skeezy marketers, the scammers, all of that is out there intermingling. As Paris said, people with different motivations can create the same kind of stuff without necessarily studying from the same textbook. And we are all kind of under attack from this stuff. And so the awareness that you need to kind of be on guard and you need to value your attention is to me, always the very first thing.
Craig Silverman [00:41:49]:
We have to be a little less passive. And I think the other part of that is like, okay, so if I'm aware, then what do I do? Is like, patience is one of the best things. You don't need to take an action right away. You don't need to, you know, go withdraw money if someone has tells you that, you know, you owe money to the government. But you. You also don't need to click, like, you don't need to reshare it. You don't need to necessarily stay on that video because your attention is really, really valuable to those kinds of people and the platforms. And you should value it too, and think about where you are giving it and where you aren't.
Craig Silverman [00:42:22]:
And don't feel like you have to take an action because a lot of times it is about getting past, you know, your initial defenses and riling up an emotion or appealing to an existing belief. And then, then we're sort of on autopilot. So you have some awareness. You guard your attention. You don't. You find a way to be patient and not quick to act. And then you think about looking in other places rather than just the thing that's in front of you. And those to me are like very simple human behavioral things that if people practice, that's a really helpful thing because it's like you can talk about tools and techniques and stuff like that, but the average person, it's really about your behavior and how you think about navigating this universe that we're in, which is really chaotic and does have a lot.
Leo Laporte [00:43:05]:
Of risk to it, is the key also funding responsible Reliable sources of information.
Craig Silverman [00:43:13]:
Well, look, it looks like I'm talking my own book when I say that, but I mean, I kind of did set you up for that.
Leo Laporte [00:43:18]:
So go ahead.
Craig Silverman [00:43:20]:
I mean, I think it's, you know, one of the challenges, and I think this is something Jeff has talked a lot about, is I don't think it's a great information environment. If only the good stuff is behind paywalls, that's not a great scenario. We don't really want to end up with that. But I do think there's an element of you should find places that you do want to support and you should find places that serve you and that do it well on a consistent basis. Nobody's going to be perfect, but finding ways to support that and get in the habit of that, it makes people feel good when they actually are supporting information source that they feel connected to. But I also think for the larger ecosystem, we have to figure out ways to make sure that good quality information is being surfaced and is not locked away for people who can afford it.
Leo Laporte [00:44:08]:
So, Craig, as part of that too, what I've been arguing unpopularly, but not for the first time in my life, is that not only do I think that training AI models is fair use, but we as journalists need to look at our moral and ethical obligations to take journalism to where people are. And if we in journalism cut off through bot killers and legislation and litigation, all the AI, then the propagandists and the marketers are going to have a field day. They will have AI to themselves and AI won't know when to call upon the journalistic organization. How do you look at how we operate from journalism in this new reality?
Craig Silverman [00:44:53]:
Yeah, I can see why it's unpopular to say it should all be fair.
Paris Martineau [00:44:58]:
Use for that we should give it all up to.
Leo Laporte [00:45:01]:
Parents are kind of like, oh, yeah, I don't know.
Craig Silverman [00:45:08]:
But I mean the. I think there's a fundamental point there of, you know, you don't want to surrender that environment and starve it of good quality information at the same time. You know, I think the challenge for me around that argument is you look at the valuations, you look at the fundraising that those company have and it's like, I think they got a little bit of money to.
Leo Laporte [00:45:28]:
Maybe there's a discussion to be had.
Craig Silverman [00:45:29]:
Yes, there's, there's. And it should be, it should be a negotiation. And I hope that, yes, that information gets in there. What actually concerns me about it is even if they do these deals, they're going to end up the way Basically, Meta ended up, which is it did start doing deals with fact checkers and then it did deals with news organizations. And because it did deals with news organizations that some people said, oh, those are all liberal news organizations organizations, and it felt it had to run and do a bunch of deals with conservative news organizations. And so it's like at the end of the day, none of these companies, even though they're like the LLMs, are saying, we're going to give you the right answer, we're going to give you good quality information. They politically, it's poisonous for them to stand up and say, yes, we are going to pay for quality information and we are going to pick and choose because they're just going to get targeted the way Meta did. And so I think that's going to be a real battle, or maybe it won't be.
Craig Silverman [00:46:17]:
Maybe they'll just make sure that they're doing. What Meta did is do big deals. They did a deal with New York Times, well, we better go make sure we do a deal with Fox News, otherwise we're going to be in trouble. And so I would like to see obviously the high quality information in there. I think the negotiations should be happening. I think the reality is there's no choice. I think you've got to have a way to get it in and into these models because frankly, of the stuff that I track online, I already see a lot of shady people selling products saying that they can get your company ranked highly in LLMs. That's already happening 100%.
Craig Silverman [00:46:54]:
People are already trying to poison the results. People are already trying to inject into them. And if we don't find a way to get the good quality stuff in, it will get drowned out.
Leo Laporte [00:47:03]:
Amen. And I think that the other problem with the deals to date I've said this on the show, is that it's mogul to mogul, it's OpenAI to Murdoch. And the vast ecosystem of news is left out. The diversity of voices, the pluralism of journalism is completely left out. And so that's why I have argued that we need to consider whether those organizations should come together to create an API for news. But that's not going to be popular either.
Leo Laporte [00:47:30]:
Craig Silverman is the king of disinformation, not making it. Craig is king journalism kingdom of this information expert. I regret the error. I'm sorry. Craig Co founder of Indicator Media Go there now, read up. There's a free membership, but for $100 a year you can also be a subscriber. And I think it seems Like a very worthy thing to subscribe to. Craig, I really appreciate $100 tithe.
Leo Laporte [00:48:00]:
Yeah. And I really, I'm grateful, grateful to you for the work you do. It's so important for years.
Craig Silverman [00:48:07]:
Thank you. Thank you so much for having me on. Thank you for that. Thank you to all of you. To having discussions where you can actually get into some detail on these things. It's much appreciated.
Leo Laporte [00:48:17]:
Many people don't know that Toronto was called the City of the broad Shoulders because of the. No, I don't know why it was. No, it's not that. It's the Big Smoke. I don't know why it's the Big Smoke, but I don't either, actually.
Paris Martineau [00:48:30]:
Is there a lot of smoke and is it physically?
Leo Laporte [00:48:32]:
I think there are a lot of factories at one time.
Paris Martineau [00:48:34]:
Yeah, I mean, I guess you could call any city at a certain time the Big Smoke.
Leo Laporte [00:48:40]:
When we launched our show, why is.
Leo Laporte [00:48:42]:
Toronto called the Big Smoke?
Leo Laporte [00:48:45]:
Oh, she'll know. And if she doesn't, I will say.
Paris Martineau [00:48:48]:
This is the one thing that Claude, that Claude doesn't have.
Leo Laporte [00:48:52]:
Smog.
Leo Laporte [00:48:53]:
Smog, yeah. When we launched the Call for Help show in Toronto, the first episode, I got a whole bunch of synonyms for Toronto and I said, live from the Big Smoke, home of the raptors, it's called for Help. And that was kind of fun. But that's the only one I remember.
Leo Laporte [00:49:10]:
It was coined by, as we have, the coiner of fake news here was coined by Alan Fotheringham, who was a well known journalist back in the day.
Craig Silverman [00:49:16]:
Very well known, yes.
Leo Laporte [00:49:18]:
All right, Craig, thank you so much. I appreciate it. Thanks to Rosie too, for joining us.
Leo Laporte [00:49:23]:
Rosie, what was it, dinner time? Rosie went out of there.
Craig Silverman [00:49:26]:
I have to feed her right now or she is gonna come attack me.
Leo Laporte [00:49:28]:
Yes. Oh, you're doing a podcast. I thought it was cnn.
Paris Martineau [00:49:32]:
There's a cat behind him holding a gun right now. Never mind.
Leo Laporte [00:49:36]:
Thanks, Craig. Take care.
Craig Silverman [00:49:38]:
Thank you so much, everybody. Appreciate it.
Leo Laporte [00:49:40]:
Take care. We'll have more on intelligent machines in just a moment after this word from our sponsor, Melissa. Melissa's the trusted data quality expert. Longer. They've been doing it longer than we have. Since 1985. 41 years this year. Bad data is bad for business.
Leo Laporte [00:50:00]:
It could be costing you. And Melissa is specialized in taking bad data and turning it into good, clean, accurate, peak quality data. And now they're doing it because they are data scientists with cutting edge AI. So they've got the expertise in the field and they're using AI to make it even better. And this is where what this means for your business you get of course their bread and butter which is address validation. Melissa can validate addresses globally in real time. I mean fast. Which reduces failed deliveries, keeps bad data from entering your database whether it's through a customer service rep or your customers.
Leo Laporte [00:50:38]:
They also could do and this is very important especially for finance and others. Mobile identity verification. This cuts fraud by matching customers to to mobile numbers. It also gives you a chance to use SMS to talk to customers where they are, how they want to be talked to. Melissa also can do change of address. This is cool. Automatically keeps customers connected when they move automatically. They also do AI powered deduplication.
Leo Laporte [00:51:05]:
On average a database contains 8 to 10% duplicate records. I know my contact list is at least that. In fact I wish I could get this from my contact list. Melissa's powerful matchup technology can identify even non exact matching duplicate records and eliminate them. So you don't send multiple mailings to the same address. You have the right one and only true address. They also do customer data intelligence. You can use Melissa to enrich records with all kinds of information.
Leo Laporte [00:51:33]:
Demographic, geographic, psychographic, firmographic, property info. This way you get more intelligent targeting, better customer service and the new Melissa Alert service monitors and automatically updates your customer data for moves, address changes, property transactions, hazard risks and more. Automatically. I don't think I need to explain that. Clean data leads to better marketing, ROI to higher customer lifetime value to AI that works as intended, right? If you're using AI for personalization or predictive analytics, garbage in means garbage out, garbage data, bad results. Melissa, make sure your AI is learning from the real deal, the accurate information. When it comes to data, Melissa's got you covered. For example, eToro, the social investing platform, 23 million users globally.
Leo Laporte [00:52:25]:
They use Melissa for identity verification. Their business analyst at El Toro said Quote we find electronic verification is the way to go because it makes the user's life easier. Users register faster and can start using our platform right away. Whether you're a small business just getting started or an enterprise manager. Millions of records Melissa scales with you. Melissa has easy to use apps for Salesforce, Dynamics, CRM, Shopify, Stripe, Microsoft Office, Google Docs and everything else you use. Melissa's APIs integrate seamlessly into your existing workflows or custom builds. And Melissa's solutions and services protect your data like it's their own.
Leo Laporte [00:53:03]:
They're GDPR and CCPA compliant, ISO 27001 certified. They meet SOC2 and HIPAA High trust standards for information security management. Remember the name Melissa in fact, get started today with 1,000 records cleaned for free. Melissa.com TWIT M E L I-S S A Melissa.com/TWIT, longtime sponsor of our shows. We thank them so much for their support and you support us. When you go to that address, please do melissa.com twit.
Leo Laporte [00:53:38]:
Paris, is your phone back?
Paris Martineau [00:53:40]:
Great question.
Leo Laporte [00:53:41]:
I was. I was trying to.
Paris Martineau [00:53:43]:
Yes, but how long will it be back? That is my question. Probably not long.
Leo Laporte [00:53:48]:
I was trying to enlist everybody to send you a bunch of texts. So you. You're. You got 100 texts when you came back online, but I guess see what you missed.
Leo Laporte [00:53:54]:
Why did Paris say that? I don't see.
Paris Martineau [00:53:57]:
I tried to text you guys multiple times today and it wouldn't go through even when I was on one.
Leo Laporte [00:54:02]:
You know, it seemed awfully quiet in there in our group chat.
Paris Martineau [00:54:07]:
Yeah, it's not what you want.
Leo Laporte [00:54:09]:
So I'm glad to hear you say you like Claude clothes.
Paris Martineau [00:54:14]:
I do.
Leo Laporte [00:54:14]:
Like, just because it has a French name.
Paris Martineau [00:54:17]:
Yeah, I mean, I just.
Leo Laporte [00:54:18]:
One.
Paris Martineau [00:54:19]:
I enjoy the. I like that I get to call it Claude, but I just think that it. I think it approaches the problems I set it, the tasks I give it in a way intuitively that is more aligned with what I'm looking for than Chat GPT or Gemini. I actually just. I had previously been subscribing to the Pro, or I guess the first paid tier for the 20 chat GPT and clud, but I recently just canceled chat GPT because I wasn't using it.
Leo Laporte [00:54:53]:
I. I did the next best thing. I. I kept chatgpt, but I upped my Claude subscription to the $250.
Paris Martineau [00:55:00]:
Oh, does that mean you get access to Claude? Cowork.
Leo Laporte [00:55:04]:
I get it all. Well, that was part of the reason because I'm on a Mac and I thought, well, I'll try Cowork, but honestly, I like the command line, which can do every single thing that all the other Clauds can do. But it. But it, to me, it feels a little bit more, as you said, more serious. I've been using it. So what happened was I was using it to configure my new laptop and it was great. I could say, you know, I would say things like, well, how do I get the fingerprint reader working? I said, oh, I could do that. Let me.
Leo Laporte [00:55:30]:
This. Let me download some software here. Just put your fingerprint. It walked me through it and it set it up. How do I get the buttons to turn the screen brighter? That's not seeming to work. Oh, yeah. Well, you need this app, so its expertise in a lot of technical areas was very useful.
Leo Laporte [00:55:48]:
Are you giving it direct access to your machine?
Leo Laporte [00:55:50]:
You can, in many ways, yes, you can, and of course no, because it's doing it at my command and it always has.
Paris Martineau [00:55:58]:
Prompt injection attack.
Leo Laporte [00:56:00]:
Well, who's going to do that? I'm sitting on my computer, there's no one here to.
Leo Laporte [00:56:03]:
Famous last words. Would Steve approve?
Leo Laporte [00:56:06]:
Well, I showed Steve. He didn't seem to mind it. In fact, what's interesting, we're going to have Harper Reed on, on Sunday. No, no. Prompt injection comes when you're doing something in public. Like, you know, how would somebody prompt inject anything here? I'm using this at my machine.
Paris Martineau [00:56:24]:
Aren't there just more areas for prompt injection attacks to occur? If you're using something like Claude, coworker, I assume, Claude code, where suddenly it has access to all the documents in your computer. If one of the things you downloaded has a prompt injection thing, then certainly the amount of vectors for potential attacks expand.
Leo Laporte [00:56:43]:
Yeah, if you say, if you said, hey, I just downloaded this text file, read it, and this is where people are doing it, or this document, like you might have a document you've downloaded and you say, I don't want to analyze this, just give me the bullet points. But it's been maliciously formed. It's a maliciously formed PDF that has hidden text in there. That is a prompt. Yeah, that would happen, but I'm not doing that.
Paris Martineau [00:57:06]:
I'm sitting here, every PDF he downloads is perfect.
Leo Laporte [00:57:10]:
I'm not having Claude analyze PDFs, I'm not having it do that kind of thing. I'm actually sitting here saying, configure this for me or do this for me. I mean, I guess there's some risk. It seems fine to me. In fact, one of the things I've been doing is asking Claude to do a security essay, find security flaws and where it found them. I said, for instance, said, well, you have a text file in your root directory. It's not public, but that text file has your keys for the API and for your access to GitHub. You probably should encrypt those.
Leo Laporte [00:57:46]:
You want me to do that for you? And I said, yes. And it runs PGP and encrypts them. And then it adds code to various things, saying, okay, instead of getting it from a plain text file, you could. And it says, test this first. Make sure it all works. Yes, I guess there's potentially. There's always going to be some trust, but I think the kinds of things I'm doing seem less likely to be vulnerable.
Paris Martineau [00:58:09]:
Have you gotten access to ChatGPT Health yet?
Leo Laporte [00:58:13]:
Not yet. I did apply immediately, but have not. I think you probably should always be careful, right?
Leo Laporte [00:58:22]:
That's what nurse has been trying to tell you.
Leo Laporte [00:58:23]:
Yes, but that doesn't mean you shouldn't use it. I think you should just look, there's plenty of.
Leo Laporte [00:58:28]:
Use it wisely.
Leo Laporte [00:58:29]:
As I said the other day, if you go to WebMD and ask it for advice, you probably shouldn't just say, oh yeah, well, must be True. It's on WebMD. You know, if you go to Dr. Oz and ask for advice, you might want to. There's a lot of suspect places you can get advice that aren't AI. In fact, a lot of the suspect stuff in AI comes from websites that were full of BS to begin with.
Leo Laporte [00:58:50]:
Did you just ask about OpenAI Health or Claude Health? Paris?
Paris Martineau [00:58:55]:
OpenAI Health, but that's my next question.
Leo Laporte [00:58:57]:
What about Claude Health? Have you gotten into that?
Paris Martineau [00:58:59]:
No. Who. I assume that Claude Health is available only to the Max subscribers, is that correct?
Leo Laporte [00:59:07]:
Do I have access to.
Leo Laporte [00:59:10]:
I have it in the rundown.
Leo Laporte [00:59:12]:
I also asked Claude, could you set it up so that I can hit a key and talk to you instead of typing and said yeah, let me just set up a faster whisper dictate and you just press the copilot. It's great because I use. It's on a windows.
Paris Martineau [00:59:23]:
I will say the one thing that is a downgrade for Me going from ChatGPT to Claude is not as many cool voices. I think you got one voice and it's classic.
Leo Laporte [00:59:33]:
Well, I'm not using the chat. It's all. I'm typing. Typing. I'm not talking to it.
Paris Martineau [00:59:38]:
I'm just saying that's the one thing that Claude doesn't have. Just because I do enjoy. I did enjoy whatever specific voice I picked out from the many from Chat GPT. I thought it was pretty good.
Leo Laporte [00:59:49]:
I don't chat. Chat is for the plebs. I'm.
Paris Martineau [00:59:53]:
Yeah, I vocally.
Leo Laporte [00:59:54]:
Do you want me to show you the app that Claude wrote for me?
Leo Laporte [00:59:57]:
Is this your perfect app? The ideal app that you wanted?
Leo Laporte [01:00:00]:
This is one of the insights added. So anyway, so I was using it to configure. I thought well maybe I could do. Do some. Maybe I could do more with this. Maybe I need some more tokens and I'll. You know what the nice thing is it's 250 bucks this month if it's not useful or I want to Do. The nice thing is it's really a neck and nest horse race between a few different companies.
Leo Laporte [01:00:18]:
If I want to do next month, ChatGpt Max, I could do that. So you're not really locked in. It is a lot. Look, I admit it's a lot of money, but how much would you spend for custom software? And this is one of the things I realized is this is going to change software a little bit. I'm not writing programs for you to use. I'm not interested in that. I'm writing code. I'm running programs for me to use that are tailored to my needs.
Leo Laporte [01:00:43]:
So one of the things that I do all the time is I spent many hours a day going through news feeds looking for stories for this show. You see them all in our rundown and every other show that I do, which involves going to a newsreader. I've tried a bunch of different newsreaders, or going to websites like techmeme and scanning them and then bookmarking them in. I use Raindrop IO. I thought, you know, maybe I could speed this up with a simpler program that's tailored specifically to my needs. So Sunday morning, I wrote this with Claude. I didn't write any code. Claude said, you want to do this? And I told.
Leo Laporte [01:01:22]:
I gave it a you. There's Claude has cloud code, has a planning mode and a coding mode. You go into the planning mode and you say, hey, here's what I want to do. So I gave it a list of features that I was looking for. They said, okay, do you want to do this in Python or Rust? I thought, well, let's do it in Rust. That's supposed to be a good language. I don't know any Rust. I know some Python.
Leo Laporte [01:01:43]:
I don't know any Rust. So it wrote, by the way, this program, as it stands now, is 3,000 lines of Rust code. I posted on GitHub. If people want to see it, it's on my GitHub account, Leo Laporte. But what you would probably do with it is download it, clone it, run Claude code on it and say, now I want to customize this for my needs. Here's a starting point. But I want to make this mine. So I called it Speedy Reader.
Leo Laporte [01:02:08]:
By the way, I didn't call it Speedy Reader to begin with. I called it RSS Reader. But then I thought, you know, I should give it a good name. And I said, claude, can you change the name? And it said, yeah. So it just changes the name.
Leo Laporte [01:02:18]:
Whatever you want, boss.
Leo Laporte [01:02:20]:
Let me show you how Speedy it is. Can you show my screen? Could you have my screen? Yeah. So I'll just run Speedy Reader. That's how fast it pops up. And this has downloaded now. First of all, I only care about articles from the last week because next week it'll be a new show.
Leo Laporte [01:02:36]:
Is this why we have a lot more articles in your part of the rundown today?
Leo Laporte [01:02:39]:
Yes.
Paris Martineau [01:02:41]:
Wait, how did you. Are you walking us through how you created this or what does it.
Leo Laporte [01:02:46]:
I guess this is the program.
Paris Martineau [01:02:47]:
Yeah. So what it.
Leo Laporte [01:02:49]:
So what I do what I. Well, I. How did I get here? What was the problem?
Paris Martineau [01:02:54]:
How did, how did they get in here? How do you send it to Speedy Reader?
Leo Laporte [01:02:57]:
You know how do you know how Feed Reader works? You have RSS links, right?
Paris Martineau [01:03:01]:
Yes.
Leo Laporte [01:03:01]:
So all of these sources, 9 to 5, Google slash, Ars Technica have RSS links. I have had for a long time a list of RSS links. They come in a standardized format called opml. So I said I want to download my OPML file, which I exported for my old feed reader and import it here. So it said, okay, we'll import the. In fact, I've added, if you look at the help, I've added a way to import an OPML file as well as export my.
Leo Laporte [01:03:27]:
You got to send this to Dave Weiner. You have to send this to Dave Weiner.
Leo Laporte [01:03:30]:
But this is very specific. For instance, it said, well, you know, a lot of feed readers have ways to starve articles and keep track of what you've read and haven't read. And I said, no, no, they put them in. At first I said, I don't want that, I don't need that. In fact, when I bookmark, I only want to bookmark to one place. So I'm just going to have B for bookmark. Let's bookmark this. And it says, and then I just want to enter a tag I am and hit return.
Leo Laporte [01:03:55]:
And now it's bookmarked on Raindrop. The other thing that I had it do is sometimes there's articles. This is the top, you see it's feed content. But I want to know, I want more. Can you do an AI summary? So I'm using Haiku, which is the faster Claude model, not the high end opus model, to generate bullet pointed summaries of the articles which then I can use save. It also gives me a better idea and it's a very quick way for me to look at the article. The point is this is very specific.
Paris Martineau [01:04:28]:
Are you able to then to read the full article in this or do you have to like click the link.
Leo Laporte [01:04:33]:
If I hit oh, it will open a browser to it.
Paris Martineau [01:04:36]:
Oh, that's so cool. Wait, so do you have the prompts for setting this up? This just reminds me of Kevin Roose, frenemy of the pod, posted this week that he, he said cloud code is so good. I built a pocket clone over the break. Yes, I said adding new features, text to speech, read aloud, video summary, spaced repetition, emails, Kindle sync, and now I have a perfect external brain. And he posted a version of it that I thought honestly was just a good way to go about this, but.
Leo Laporte [01:05:11]:
I have no interest in giving somebody.
Paris Martineau [01:05:12]:
Else kevinroost.com stash, but that's just. I think it's cool. If you find something that's useful, you can just, you can.
Leo Laporte [01:05:22]:
Give this to all your hosts.
Paris Martineau [01:05:23]:
I was gonna say give this to us at the very least. So Kevin at the, at the beginning, at the bottom has a install with Claude code. If you have Claude off, COD code setup takes two minutes. Just paste this prompt and he has like an eight step prompt in there that you can then go through to get your own version of this.
Leo Laporte [01:05:39]:
And I, yeah, in fact, Claude wrote all of that for me as well. If I go to, if I go to my repositories here, you can see the. I posted on GitHub. So it is, it is available for anybody who wants to take a look at it. One of the things GitHub did, because I wrote it in Rust, is it made versions for Linux and Mac so you can run it, you can download it. There are binaries on here you could download and run. But really, I don't think that's the point. I think the point is that people can now with.
Leo Laporte [01:06:09]:
Without any coding skills, I mean, I guess you might have to be a little technical. This is going to change over time, right?
Leo Laporte [01:06:15]:
This is what I was trying to ask last week.
Leo Laporte [01:06:17]:
Personal software that scratches your itch specifically.
Leo Laporte [01:06:22]:
But you can also create. It's a whole new software industry. You can create applications, agents, call them what you will. For other people, you can charge.
Leo Laporte [01:06:32]:
Yeah, I'm less interested in that. I think the.
Leo Laporte [01:06:34]:
But I think there's opportunity some people.
Leo Laporte [01:06:36]:
But I think the real power is everybody should learn how to do this and do it themselves. You could. By the way, I did this with the $20 version. It just would take time because every, maybe every few hours it would say, okay, you got to take a break till two. But you could absolutely do this. Not probably in the free version, but you could absolutely do this with the $20 version. I think that this is the future of software. Yeah, I remember back in the early days of computing, George Morrow, who you remember the, I know you remember the Osborne Jeff, which is a suitcase luggable.
Leo Laporte [01:07:09]:
I had one.
Leo Laporte [01:07:10]:
The Moro was a clone of that. It was a suitcase luggable and George Morrow was the guy behind it. Brilliant guy.
Leo Laporte [01:07:15]:
I had that too.
Leo Laporte [01:07:16]:
Oh, well, there you go. He said then, and we're talking what, 1929, I don't know what it is. 1984, it was CPM, so it wasn't Windows. So it would have been, you know, early 80s. He said then everybody should just write their own software. Which I laughed at because people don't have that capability. That's part of what this shift is to me, the AI shift is people can write their own software. What this really does is it fulfills the promise, the very early promise we were always looking for, of a human to computer interface that understood language.
Leo Laporte [01:07:54]:
See, we speak one language, computer speaks in a different language. There's some interface that allows us to tell our computer what we want and it will do it. And it's always been kind of out there, the distant holy grail of computing. This brings it so much more immediate and really now the sky, for me, the sky's the limit. I can say I can customize this. I'm thinking of all sorts of software that I'm using.
Leo Laporte [01:08:18]:
How many words in total did you use to prompt it?
Paris Martineau [01:08:21]:
You guessed.
Leo Laporte [01:08:22]:
If you, if you do go to the GitHub, you can see the conversation. It's probably quite a few because I, I spent a couple of hours doing it on Sunday. It was a finished thing. And then over the last couple of days, See, this is the other thing and this is what Mike Masnick was telling us. It's never done. Yes, you, you know, I go, oh, you know what?
Leo Laporte [01:08:41]:
I, for instance, it ain't shrink wrapped anymore.
Leo Laporte [01:08:43]:
I realized, oh, I don't really care how many read or unread articles there are. Just forget about read or unread. Assume if it's in there, I haven't read it because I'm going to delete it. So the main key I use is the delete is D. Just delete that. Nope, not going to look, I don't want to ever see that again. I've read it. So I changed that.
Leo Laporte [01:09:02]:
On the left here, it used to have the headline and then now I just have the date because really what I care about is, see, it's T for Tuesday. What I really care about is how old is this story? Because I don't want to do a story that's a week old. If I go to the bottom, this is last Wednesday, right. So these are last Wednesday's stories. So I want to know how old it is. So I've got the date and I've got the source. The headline is in the upper right, and then beneath it, I have the opportunity. I don't want to generate an AI summary every time, although I could change that.
Leo Laporte [01:09:34]:
Have it do it at night when I'm not around.
Leo Laporte [01:09:37]:
That uses more compute.
Leo Laporte [01:09:40]:
I got lots of compute. I spent a lot of money.
Paris Martineau [01:09:43]:
Hi, this is real quick Leo on this. I was wondering why the Leo's links document started to look different.
Paris Martineau [01:09:51]:
Yeah.
Paris Martineau [01:09:51]:
And this is probably it. Because.
Leo Laporte [01:09:53]:
Oh, yeah, these are all going to.
Paris Martineau [01:09:54]:
This AI summary gets dropped in there too. So all the boxes get really big. And now it's a lot harder to parse.
Leo Laporte [01:10:00]:
Well, see, I can. Okay, that's true. I can change this. So what we have a way of. This is the raindrop that it goes to when I bookmark it. You see, I just bookmarked this way. And if I have a summary, it's going to put the bullet points in there. And you don't like that, Benito?
Benito Gonzalez [01:10:16]:
I don't like how it imports it into the document that I use to make the doc. To make the show doc.
Leo Laporte [01:10:21]:
Yeah. So we're using Zapier to import that into a Google sheet. I think ultimately the best way to do this would be to get the stakeholders together and saying, okay, this is what I want. This is how it's going to process it. Now what do you need When I put this in the sheet and have it do that too, so I can control the whole chain.
Craig Silverman [01:10:41]:
So right now you have to clean.
Paris Martineau [01:10:42]:
Up all these headlines.
Paris Martineau [01:10:43]:
I didn't have to clean it up. It just.
Paris Martineau [01:10:45]:
We're looking at.
Paris Martineau [01:10:46]:
I didn't have to clean it up. It's just that now I only see a couple of articles per screen. I can't see like 12 or 13. I only see like 6 now because the summary makes the box really big.
Leo Laporte [01:10:58]:
I can also stop putting that in.
Paris Martineau [01:11:00]:
But I think now that you have this. Yeah, we should get all the stakeholders together and figure out, like, what the best way to do this is now. Because right now it's still a little, you know, it's just pieced together still right now. Right. But now that you have.
Leo Laporte [01:11:13]:
It's now perfect. It's perfect on my end.
Paris Martineau [01:11:14]:
Yeah, exactly.
Leo Laporte [01:11:15]:
It's perfect on my End. But see, this is what it's generating is this spreadsheet. And you can see now that these boxes have more information.
Paris Martineau [01:11:24]:
Yes, sir. Scroll down. See how it used to look?
Leo Laporte [01:11:27]:
Yeah, see that. But I thought you just ignored this column. Anyway, this is the summary column. Column, the description column. But yeah, you pasted.
Paris Martineau [01:11:36]:
But you see how I only see like five articles per screen now? Like four or five.
Leo Laporte [01:11:39]:
What I could easily do, and I will do is have Zapier not copy description. Well, the thing, the next step, I was also thinking, which I will do for you, Vanita, the next step, by the way, Bonito is so inspired by this. Tell them what you're gonna. What you're thinking.
Paris Martineau [01:11:54]:
I'm gonna start vibe coding too. I'm gonna vibe code some audio plugins because. Because I wanted a specific reverb this weekend and I couldn't afford it. So I was like, ah, maybe I'll just try to vibe code one. And it was surprisingly easy. So I'm going to start making more.
Paris Martineau [01:12:10]:
Yeah, that's exciting.
Leo Laporte [01:12:12]:
Cool. And this is the thing is that Claude understands APIs or can download APIs and understand them. So there's a specific format for plugins called VSTs. You can tell Claude, okay, learn everything there is to know about VSTs and how to make them. Now you know that I want to make a VST and it will actually there's an interactive part in the planning thing where it'll say, okay, well what do you need? What do you want? How do you want me to do this? And there's. So there is some inactivity. You can even go back to Claude, as I have several times, and said, what do you think? For instance, if you see when I generate a. A summary, there's a hang for a few seconds.
Leo Laporte [01:12:56]:
I said, hey, could you put. Actually I said, I want. Could you put a little rotating newspaper or something? Some way of letting me know. So what it said, well, no, I'll put this little Braille rotator in there. That's more standard. So I put the little rotator in there. It's cute. And I said, well, you know what I want you to do? I want you to go through the entire code base and look for anything that might happen, like a refresh feed that will.
Leo Laporte [01:13:20]:
Will lock, will make this unresponsive and put up one of those little rotators anywhere that that happens and some information like you're doing something. So I know that I haven't lost access to the computer. These are little things that's easy to do. And Claude has some, lots of experience of how others have done it. In fact, when I said do a newsreader, it knew what a newsreader should look like. I actually had to tell it, No, I don't, I don't want that. I don't need to star anything. I'm going to either bookmark it or delete it.
Leo Laporte [01:13:53]:
That's all I want. And it removed that stuff so it has some understanding of what's kind of typical. And then you customize it from there.
Leo Laporte [01:14:00]:
Could they put it straight in the spreadsheet?
Leo Laporte [01:14:02]:
Well, that's zapier, but I could, yes. So the next step is to perhaps to eliminate Zapier. Although Zapier does it pretty. It'd be easy right now for me to say to Zapier, hey, just don't copy that big fat field. Benito doesn't want it. Ultimately, what I think I will do for you guys and the others is make a briefing book. So what I can also do is say, okay, now go through all of the stories that I have bookmarked for the week of January 14th and create a bulleted list of the headlines, a briefing book. And this is the thing I'm really interested in, because as you know, we all bookmark for far more stories than we are ever going to get to.
Leo Laporte [01:14:45]:
Especially when Leo talks on and on about Claude.
Leo Laporte [01:14:47]:
Whose fault is that?
Leo Laporte [01:14:50]:
What would be really interesting, and I don't know how well it worked, is to say to Claude, okay, what are the most important ones?
Leo Laporte [01:14:59]:
But you see, you never go with important, you go with the things that spark.
Leo Laporte [01:15:02]:
It probably won't do a good job, unless maybe after I could say, hey, note what I've picked in the past, could you learn my. But my. I wonder.
Leo Laporte [01:15:11]:
I mean, I think you could, you could ask which. Which ones might spark the most interesting conversations.
Leo Laporte [01:15:15]:
Yeah, I mean, there's. I don't know. We'll see.
Leo Laporte [01:15:17]:
Let me ask you a question, Leo, if I may. So there's a company like Cursor out there that got a lot of attention.
Leo Laporte [01:15:22]:
Yeah.
Leo Laporte [01:15:23]:
That is there to also code in AI. How does that fit? Where do these, these things. Where does Claude code versus Cursor fit to?
Leo Laporte [01:15:32]:
Cursor, will use Claude, but it's what we call an integrated development environment, an ide. It's an editor. So what? The way this evolved is initially like copilot on GitHub is you'd be in an editor, you'd start typing something and it would complete what you were typing. It would save you typing. That was called intellisense. That was the first step. Then later, copilot, you could say, hey, I'm sitting in my editor. Could you write the next block? I want to.
Leo Laporte [01:16:03]:
Could you write some code that would just scale Everything by 20%? That and then it would propose code which you could then include in the editor, or you would even enter it into the editor. Just like intellisense. It would autocomplete that. And that's what a lot of people are still doing. That's kind of what I. My understanding of how cursor works. Any gravity from Google. Those are editors where the AI is in the editor and inserting stuff.
Leo Laporte [01:16:26]:
This is not like that. I mean, I can do that. I can do that in my emacs, which is the code editor I use. But I kind of prefer this. This is not any. I'm not doing any coding. I'm just.
Leo Laporte [01:16:39]:
Are you editing any of the code? Did you change any of the raw code?
Leo Laporte [01:16:41]:
I don't know, Russ.
Leo Laporte [01:16:42]:
Not a line.
Leo Laporte [01:16:43]:
No. It's 3,000 lines of code I don't even understand. But I don't need to understand it, remember? That's another reason why I don't want to ship this. This is. I'm taking the entire risk myself. I don't have to write a EULA that says I make no warranty about the appropriateness of this software. If it doesn't work, it's not on me. Should I show you what it looks like? Let me see.
Leo Laporte [01:17:10]:
What would be a feature. Can you. Let me do this. Can you add a clock that tells me how long I've been using Speedy Reader. Maybe I'm. Maybe I'm worried I'm spending too much time in Speedy Reader.
Paris Martineau [01:17:31]:
Maybe you're not being speedy enough.
Leo Laporte [01:17:33]:
Maybe I'm not being speedy enough. So now it's thinking you can keep it on here for a little bit Bonito. Just so people can see what it looks like. Again, this isn't an editor. This is. So it's looking at the code. That's. That's the code right there.
Leo Laporte [01:17:46]:
That's part of the code. It's looking for a place to put this request. It's actually reading the code. I didn't develop it on this computer. I asked this computer to catch up. I developed it on a different computer. So I said, hey, Claude, catch up. Okay.
Leo Laporte [01:18:06]:
So, yeah, it's found somewhere to put it.
Paris Martineau [01:18:08]:
Claude says, hey, Leo. Mustard.
Leo Laporte [01:18:11]:
Oh, ketchup.
Craig Silverman [01:18:12]:
Ketchup.
Leo Laporte [01:18:14]:
Aren't you glad I didn't? And say orange.
Leo Laporte [01:18:17]:
So it's showing you where it's adding the lines.
Leo Laporte [01:18:19]:
Yeah, this is. So it does show you the code and the green is added. If it's deleting something, it'll be in. Highlighted in red. Adding elapsed time, formatting method. So these are. This is code. It's adding.
Leo Laporte [01:18:31]:
I know enough coding. I know I can look at this and know what it's doing. I mean, I'm not. Rust isn't.
Leo Laporte [01:18:37]:
Yeah, but you didn't. You didn't feel any compulsion to go and.
Leo Laporte [01:18:40]:
No, I wouldn't want. I'd probably break it. That's another thing, you know, people often say, well, what about the security concerns? It. I think it's must. There's a reds thing. So it's deleting that to replace it with something else. So I think it's much less likely to make the kind of boneheaded errors coders often make because we're lazy or we're not paying attention or whatever. I think it's.
Leo Laporte [01:19:03]:
It doesn't get tired. It's much less likely to make those errors. And as it gets better and better, it's going to know more and more of the kinds of errors. Okay, done. The session timer now appears in the bottom right corner. Timer starts when you launch the app. Let's see, should we see if it. Should we see if it did it? Let me exit out.
Leo Laporte [01:19:23]:
I don't know if it put it in the new one. No, it didn't. Let's see. Let me ask it how I can get it and then can you build this version so I can call it by name? So it has to compile it, I think. No, it compiled it. All right.
Paris Martineau [01:19:46]:
Fascinating.
Leo Laporte [01:19:47]:
Yeah.
Leo Laporte [01:19:48]:
So you know what I want to.
Paris Martineau [01:19:49]:
I think that it should make the timer. If you get to a certain point, it turns red and there's like a little frown that emerge.
Leo Laporte [01:19:55]:
Easy to do. This is the iterative process which is so interesting I can then say, yeah, you know, what I really want to know is not just how long I've been in this session, but how long I've been doing this. All week, for instance.
Leo Laporte [01:20:09]:
This is my double question, but for the sake of us dumb people out here, are you running this in Linux on your laptop?
Leo Laporte [01:20:16]:
It's running on a Mac right now.
Leo Laporte [01:20:19]:
This is an executable.
Leo Laporte [01:20:20]:
Yes, executable for Mac or Linux. In fact, if you go to the GitHub, you can download it and run.
Leo Laporte [01:20:24]:
Got it, Got it.
Leo Laporte [01:20:26]:
I have a few things you need to know how to do. Like you have to import the OPML and stuff. But I've tried to make it that fairly straightforward. It's all documented on the GitHub.
Leo Laporte [01:20:34]:
You want to be nice to me? You know what I want?
Leo Laporte [01:20:35]:
Yeah, I want. Yeah.
Leo Laporte [01:20:36]:
I want a version of this to go to our arXiv.org and look at all of the last weeks.
Leo Laporte [01:20:44]:
Oh great. Now I'm in the business of writing.
Leo Laporte [01:20:47]:
Code and find the ones that are relevant to a discussion about media and that are interesting for discussion.
Leo Laporte [01:20:58]:
Okay, I could probably do that. But why don't you do it? I bet you could do it, Jeff, if you gave me.
Leo Laporte [01:21:09]:
Revise it.
Leo Laporte [01:21:10]:
This is the important thing. It empowers people, even people who are not coders, people like Jeff, to create custom software for themselves that does exactly what they want and you can refine it and get it better. And I think that you shouldn't feel like you can't do this is I guess my point.
Leo Laporte [01:21:29]:
Yeah, I just gotta find the time to.
Leo Laporte [01:21:31]:
It doesn't take that much time. I didn't, I wasn't, I was doing other things. I was watching a football game.
Leo Laporte [01:21:37]:
So Darren OKE is telling me what to do in the, in the discord saying go to go go Google Google Opal.
Leo Laporte [01:21:46]:
Well, one of the things that's interesting is trust. No one's saying you could do this with N8N and, and. But I think N8N which is that node based editor that lets you hook stuff up is actually obsoleted by this. I think you can write code directly. That does it. I think a lot of the, this is the other thing that's changing so rapidly on Christmas Day. Andre Karpathi. Did we mention this? Last week I might have tweeted, I can't keep up.
Leo Laporte [01:22:14]:
This is a guy who created the term vibe coding. It's happening too fast. There's too many tools. I can't keep up. And that's what's really interesting. This is exploding. A lot of the new features in Claude code that have been released in the last two weeks have been written by Claude code. Now that's interesting because that means now this stuff is going to happen faster and faster and faster and faster.
Leo Laporte [01:22:36]:
Well, it takes away the, it's the end of, it's the expansion of coding and the end of coding.
Leo Laporte [01:22:41]:
Once we've always said this. Once AI starts to improve itself. Yeah, it eliminates the human in the loop. There's no limit.
Leo Laporte [01:22:50]:
You're still in the loop because you're, you're, you're, you're deciding what is right for you.
Leo Laporte [01:22:54]:
I don't have to be the Customer. Did I tell you about Ralph Wiggums?
Leo Laporte [01:22:59]:
What is that? The name of a cat.
Leo Laporte [01:23:02]:
So it'd be.
Leo Laporte [01:23:03]:
It would be a good name for a cat.
Leo Laporte [01:23:06]:
Okay, there's a tool called Ralph Wiggums and I'm going to say use Ralph. So this was coded originally by a third party and then Anthropic said, oh, this is really good. Use Ralph Wiggums. What Ralph Wiggums does is it makes it a little bit more autonomous. Oh, use it for what? Oh, what you do is.
Paris Martineau [01:23:33]:
I'm not sure what you mean.
Leo Laporte [01:23:35]:
You say we fail English. That's impossible. I'm in danger. What Ralph Wiggums plugin does is you can give it a constraint. Like, I want you to keep doing this until there are no more errors. And then you can also. The constraint can also be. But only do it 20 times because I don't want you to use up all my credits.
Leo Laporte [01:23:58]:
And it will go out and continue to work. And this is the beginning. This is what people are starting to do now. And you'll see people talking about this. I let COD go, go for three days or I let it go for an hour and it wrote this splat game. You can give it these constraints and it will start to do stuff. So Ralph Wiggums is really interesting. I didn't invoke it properly, but because you can say, hey, Ralph, keep working on this until the spinner gives you no errors.
Paris Martineau [01:24:29]:
Bash loop is what Reddit is telling me.
Leo Laporte [01:24:33]:
Yeah. Well, okay, so this is the other thing. Yeah. Bash is just a shell programming language. It's a shell script. You can write this stuff in anything. That's what's also really interesting. It's very agnostic about languages.
Leo Laporte [01:24:50]:
The real thing is this is a human to machine interface that uses English, which we happen to be pretty good at, but computers have been terrible at understanding.
Paris Martineau [01:25:04]:
Interesting.
Leo Laporte [01:25:04]:
Yeah, all interesting.
Leo Laporte [01:25:05]:
That's my little. My show and tell.
Leo Laporte [01:25:07]:
Yes.
Leo Laporte [01:25:08]:
Yeah. Ralph Wiggum. When not while not finished, send prompt to Claude. It just. It loops it until you get to the. And you have to tell it what the completion criteria criterion is.
Leo Laporte [01:25:21]:
But you're. But my. I still want to go back. You're still the customer. You're still the human who wants to use this. And you will decide when it's done. You will decide whether it's good enough, whether it does what you want.
Leo Laporte [01:25:31]:
It's never done. Just like Mac. Maznick said he's continuing to work on his time management program. Probably every day. He says, yeah, I don't like that. Can you make it blue.
Paris Martineau [01:25:43]:
I'm often saying that.
Leo Laporte [01:25:48]:
Well, I think you both, as an experiment, should play with this a little bit. It's maybe seems imposing if you're not comfortable with a command line or a coder, but honestly, I think I don't. I think it's very doable. If you want, Jeff, we can do this. Send me a paragraph of what you want. Oh, and I will, I will.
Paris Martineau [01:26:10]:
We could do a club twit where we all vibe code together.
Leo Laporte [01:26:13]:
We could actually. We've done that already.
Leo Laporte [01:26:15]:
I think it might take 24 hours to do that. Paris could.
Paris Martineau [01:26:18]:
We might need to do a floor.
Leo Laporte [01:26:19]:
Did you see how fast it wrote that little timer for me? I mean, it's just.
Paris Martineau [01:26:23]:
I mean, did you see how fast Claude made a game where I could hit you with a tomato last week?
Leo Laporte [01:26:26]:
Exactly.
Paris Martineau [01:26:28]:
It was not a very two and a half minutes. It wasn't. It was not very good game.
Leo Laporte [01:26:32]:
Did you show any friends the.
Leo Laporte [01:26:34]:
And Darren wants me to show you Google's Opal because you could also do this. This is something that's less intimidating and less capable, but somewhat. Somewhat similar.
Paris Martineau [01:26:45]:
Yes, but you have to give it access to your entire Google Drive, which I don't like.
Leo Laporte [01:26:50]:
Oh, I haven't tried. It's already.
Leo Laporte [01:26:51]:
Google already has all my stuff.
Paris Martineau [01:26:54]:
I don't want. I just don't like when somebody. When I have to connect something to my entire Google Drive, you have to give it permissions. That's not what I want.
Paris Martineau [01:27:04]:
You need to burn a Google account for that kind of stuff.
Paris Martineau [01:27:06]:
I know, I know. That's what I gotta do.
Leo Laporte [01:27:09]:
All right, so Opal, Darren says is not less capable. So that's good.
Leo Laporte [01:27:15]:
Take that, Leo.
Leo Laporte [01:27:17]:
So there.
Leo Laporte [01:27:17]:
Aaron knows.
Leo Laporte [01:27:18]:
He's a big fan. Build and share mini AI apps using natural language. All right, fine.
Leo Laporte [01:27:27]:
So you can share Jeff, which is what I've wanted to do. Yes.
Leo Laporte [01:27:30]:
Yeah. Maybe Jeff, you should start here just to see. Create a weather. Specific outfit for a specific. And you see, it's. It's. Well, I'll zoom in on this, but it's not. It's.
Leo Laporte [01:27:46]:
It's very. It's no. What this is a nodal compositor is what it is. And so you. You create nodes and you can. I can insert something in here and. And so forth. And these are just.
Leo Laporte [01:27:57]:
Darren, does it run on Opal? If I want to share it with Paris, does it run there?
Paris Martineau [01:28:02]:
So this makes web apps. Like it actually wants you to make web apps.
Leo Laporte [01:28:05]:
Web apps, yeah. Okay, that's what I wanted. That's exactly what I wanted.
Leo Laporte [01:28:08]:
Right.
Leo Laporte [01:28:08]:
Thank you, Darren.
Leo Laporte [01:28:10]:
So, let's start with the weather.
Paris Martineau [01:28:11]:
So do you think it will work on a workplace account, Jeff?
Leo Laporte [01:28:17]:
I. I don't know. When it comes to your wardrobe, do you identify as male, female, or non, binary? Let's do this for Paris. So female. Okay. Getting weather report. It's going to generate a nice outfit for you. I didn't answer all the questions, so maybe it won't be perfect.
Paris Martineau [01:28:34]:
Yeah, this is kind of interesting. Give me a.
Leo Laporte [01:28:37]:
By the way, trust no one says it will work with your workspace account, Jeff. Ooh, this is good for today's weather.
Leo Laporte [01:28:45]:
Download all my Google Drive files. It wants to see, edit, create and delete only specific Google Drive files you use in this app.
Paris Martineau [01:28:51]:
This is why we got to get a burner.
Leo Laporte [01:28:53]:
It wants to view, query, edit, and manage the content you uploaded to the Generative Language API, Semantic Retriever, whatever the hell that is. Learn more.
Leo Laporte [01:29:03]:
Okay, what's the occasion for this outfit? Paris?
Paris Martineau [01:29:08]:
Let's say going out for drinks tomorrow night. I think that's actually a good one because I think it's supposed to be very cold tomorrow night, if I recall correctly. I think we're getting.
Leo Laporte [01:29:18]:
Why don't you have it? Going out for drinks.
Leo Laporte [01:29:19]:
Depending what else defines your style.
Leo Laporte [01:29:23]:
Tell it to look up the weather in New York tomorrow.
Leo Laporte [01:29:24]:
Oh, and I would say, let's see. Nihilist. Goth, right?
Paris Martineau [01:29:28]:
Yeah, nihilist. A little goth alt, but also kind.
Leo Laporte [01:29:33]:
Of colorful and glamorous. Some glam.
Paris Martineau [01:29:36]:
Yeah, yeah, some glam. You know, we'll really confuse it, I think. Oh, my. I had. While I was doing all this, I was getting Claude to put together a briefing report for us based on this rundown, and so I'll. I'll put it together for you.
Leo Laporte [01:29:52]:
Oh, cool.
Paris Martineau [01:29:53]:
You're using someone else's app here, right? You're not making a stylus app here.
Leo Laporte [01:29:56]:
This is their app. Yeah, this is their. Their demo app. Let's start with the weather.
Paris Martineau [01:30:02]:
Oh, I actually tried Opal first for Vibe code because it was free, but they only make executable, so I couldn't make plugins.
Leo Laporte [01:30:10]:
Yeah, I think you really wanted Claude. Code is.
Leo Laporte [01:30:13]:
Yeah, yeah, yeah, but for dummies like me. But, you know, hey, I've never coded.
Paris Martineau [01:30:20]:
A line in my life. I've never coded a line in my life.
Leo Laporte [01:30:22]:
Entry drug. Yeah, you're still younger and more courageous than I am, young man.
Paris Martineau [01:30:29]:
How interesting. We're all coders now.
Paris Martineau [01:30:33]:
No, we're all developers now. Claude's doing the code.
Leo Laporte [01:30:35]:
There are no Coders.
Paris Martineau [01:30:36]:
I guess that's true.
Leo Laporte [01:30:37]:
I will. What I excites me is that this is the notion of something super hyper personalized. That it is finally the holy grail of computing, which is that I control my computer. By the way, I'm kind of soured on both Windows and Macintosh because they are so opinionated about what you should and shouldn't be able to do with your computer. That's one of the reasons I went to Linux. Oh, this looks good. I think Paris. This is Midnight Muse.
Leo Laporte [01:31:01]:
Oh yeah, it's going to be. Yeah. Emerald allure black faux leather trench coat.
Paris Martineau [01:31:07]:
That's kind of nice.
Leo Laporte [01:31:08]:
That is black velvet long sleeve top. Yeah. Chunky black platform combat box.
Paris Martineau [01:31:15]:
I do have those.
Leo Laporte [01:31:15]:
And here's a little bit of color. Emerald green velvet cross body bag. A vibrant.
Leo Laporte [01:31:20]:
Now I need pop color blood lipstick.
Leo Laporte [01:31:23]:
You want the ops op?
Paris Martineau [01:31:26]:
My. I keep trying to text you guys my Claude briefing, but I no longer have the ability to Send and receive SMS's.
Leo Laporte [01:31:33]:
So you can put it in discord. Are you still. It's rising. Still not.
Paris Martineau [01:31:39]:
It was up for maybe like five minutes but then it went back down. I guess I'll put it in discord.
Paris Martineau [01:31:46]:
I do have a question.
Paris Martineau [01:31:47]:
I don't.
Paris Martineau [01:31:49]:
If you decide to sell this app, who owns it?
Leo Laporte [01:31:53]:
Oh, I do.
Leo Laporte [01:31:53]:
Yeah.
Paris Martineau [01:31:54]:
Okay, so Claude has no secret.
Leo Laporte [01:31:55]:
But you can't.
Paris Martineau [01:31:55]:
Cloud has no.
Leo Laporte [01:31:57]:
Okay, no.
Leo Laporte [01:31:59]:
But can you copyright it?
Leo Laporte [01:32:01]:
Sure, why not?
Leo Laporte [01:32:02]:
Products of AI as of now can't be copied.
Leo Laporte [01:32:05]:
Oh yeah, but the human part is so I. Because a human was involved. So that was the thing. Is that the solely AI? Yes, but if a human has created it with the help of AI or not, it's copyrightable.
Paris Martineau [01:32:21]:
Here's what Claude thinks are the main themes of this week. The AI industry is experiencing a simultaneous trust crisis and power consolidation moment, says Claude. Three parallel storylines dominate. Grok's deep fake disaster demonstrates what happens when safety guard rails are removed. Apple surrender to Google reveals that even trillion dollar companies can't build Frontier AI alone. And Health Care AI's gold rush shows all three major AI labs racing into the highest stakes domain. Meanwhile, the financial viability of the entire AI sector faces unprecedented scrutiny. I mean that's fine.
Leo Laporte [01:32:59]:
That's. That's pretty good. I mean those are the three top stories I've been talking about for sure.
Paris Martineau [01:33:04]:
Yeah, it is a little slick, Jeff. You're right.
Leo Laporte [01:33:06]:
Yeah, it is.
Leo Laporte [01:33:07]:
Well, I don't like its prose. I never do like the prose. That's why I asked it for the summaries to Be bullet pointed, for instance. Is I just. Just the facts, man. Anyway, I've talked way too long about this. Let's get to some other stories. In just a moment, we will.
Leo Laporte [01:33:20]:
This is Intelligent Machines. Actually, Darren, would you find out. He says, do you own the code? Can you take it with you? Darren says, that's a really good question. So let's put the club to work. Ask all your AIs who owns this code that I wrote code? I believe I do. Because it. It prompted it. Unless it.
Leo Laporte [01:33:40]:
No, I know I do. Of course I do. Anyway, we'll move on.
Paris Martineau [01:33:45]:
I don't know. Is there something. Is there something in the cloud code eula, that says, like, we own all the code that comes out of here?
Leo Laporte [01:33:51]:
Oh, well, look, I'm very surprised if it does. I mean, this. They.
Craig Silverman [01:33:55]:
This.
Leo Laporte [01:33:55]:
The intent of this is to have people write software. That's why you pay 250 bucks for the software. If I sell it. Okay, I will check, but I'd be shocked if it. Oh, okay. Darren says you own the code. Claude code was making. But when you talk about Bolt or Opal and they're hosting of it, that's another question.
Leo Laporte [01:34:21]:
Yeah, they have to host it, so it's not exactly yours. That's interesting, but now I've got to.
Leo Laporte [01:34:28]:
Figure out how to make an RSS feed from the archive.org API.
Leo Laporte [01:34:33]:
I love it when AIs apologize. Paris, go ahead and tell us what.
Paris Martineau [01:34:37]:
I said, Jeff to Claude. Jeff thinks your prose is, quote, a little too slick. How do you feel about that? I enjoyed it because then it has like, the thinking. And whenever it was doing the thinking, you saw one saying, weighing my immediate negative response as potentially defensive, offensive. What are you doing, Claude? He says, Claude. Or it says, honestly, he's probably right. There's a certain AI generated sheen that creeps in write that. The tidy parallel structures, the way everything gets packaged beneath thematic buckets with connecting threads.
Paris Martineau [01:35:12]:
It reads like it was written by someone who's read a lot of good writing, but hasn't actually lived through a 3am deadline fueled by bad coffee and spite. Now it's trying to be too cute.
Leo Laporte [01:35:23]:
You know, the briefing format kind of invites it too. See, it's defensive.
Paris Martineau [01:35:28]:
It's true.
Leo Laporte [01:35:29]:
I like to synthesize a dozen major stories into a scannable sections. The temper temptation is to make everything feel equally weighted and polished. This is a good lesson for editors. It's, you know, if it's useful feedback, I could lean harder into the places where the stories are genuinely messy or where the through lines don't actually connect as cleanly as I made them sound.
Paris Martineau [01:35:54]:
Would it be more typos?
Leo Laporte [01:35:57]:
Lawrence did some research.
Paris Martineau [01:35:58]:
No, it would be upsetting.
Leo Laporte [01:36:00]:
Yes.
Leo Laporte [01:36:00]:
He says Anthropic's consumer terms state, subject to your compliance with our terms, we assign to you all of our right title and interest, if any, in outputs. You own the code. The code. Commercial terms are more explicit. Customer owns all outputs and Anthropic assigns to customer its right title and interest, if any, in and to outputs. And Anthropic will not train their model on your inputs or outputs under commercial terms. But Claude itself says there is uncertainty in copyright law. It requires human authorship.
Leo Laporte [01:36:32]:
This is what we were talking about. Yes, but that means if it's purely generated by AI, when you modify or integrators substantially contribute to AI generated code, you strengthen your ownership claim through your human authorship. I think I can make a strong claim that I authored this program with the help of Claude, but that doesn't give Claude Right the rights to it.
Paris Martineau [01:36:56]:
Gotta wait for the eventual.
Leo Laporte [01:36:58]:
I was gonna take a break.
Paris Martineau [01:36:59]:
There will be a court case about this.
Leo Laporte [01:37:00]:
There will, of course. Yeah, yeah, yeah. Another reason why I have no desire to sell this. But I'd be glad to write you your little program, Jeff. So just send me a paragraph of what you want.
Leo Laporte [01:37:10]:
That's all right. I'll go to Opal and try it as well.
Leo Laporte [01:37:13]:
Okay, but does it have to be rss?
Leo Laporte [01:37:15]:
Could it go to the website and do a search and execute that?
Leo Laporte [01:37:18]:
Absolutely, yeah.
Leo Laporte [01:37:19]:
So I don't need an RSS feed.
Leo Laporte [01:37:21]:
Oh, no, that's just what I was writing. This absolutely does not have to be rss. In fact, there is, as we talked about last week, a Chrome plugin for Claude code that it can invoke Chrome and do anything. A web browser.
Leo Laporte [01:37:34]:
Can I do this? Can I use the address of your GitHub and say, Go look at Leo's application. Yes, I like it. I want you to modify it to do this.
Leo Laporte [01:37:46]:
Yes. In fact, right there on the GitHub, there's an explanation of how to do that. That's because I've made that repository public. GitHub does allow you to make private repositories, but because I've made it public, anybody can look at it and download it and fork it. It's MIT licensed. It's a open source license, by the way. Interestingly, I didn't. It didn't ask me and I didn't specify a license.
Leo Laporte [01:38:08]:
It just shows that. Which is a good license. I'm happy with it. It's a very interesting our show today brought to you by Zscaler, the world's largest cloud security platform. The potential rewards of AI, as you can see, are very interesting and certainly as a business, you really cannot ignore AI in your business. But there are risks too, right you can't ignore, like the loss of sensitive data attacks against enterprise managed AI. And let's not forget generative AI, just as it's a boon for you, is a boon for threat actors, helping them to rapidly create phishing lures and write malicious code and automate data extraction. Did you know there were last year there were 1.3 million instances of Social Security numbers leaked to AI applications.
Leo Laporte [01:39:00]:
ChatGPT and Microsoft Copilot alone saw nearly 3.2 million data violations. I think a lot of businesses realize it's time to rethink our safe use of public and private AI. And that's why you want to check out zscaler. Listen to what Siva, who's the director of security and infrastructure at Zuora, says about using Zscaler to prevent AI attacks.
Paris Martineau [01:39:26]:
With Zscaler, being in line in a.
Craig Silverman [01:39:28]:
Security protection strategy helps us monitor all the traffic. So even if a bad actor were to use AI because we have tight security framework around our endpoint, helps us 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.
Leo Laporte [01:39:56]:
With Zscaler, Zero Trust +AI, you can safely adopt Gen AI and Private AI to boost productivity across the business. Their Zero Trust Architecture +AI helps you reduce the risks of AI related data loss and protects against AI attacks to guarantee greater productivity and compliance. To learn more, go to zscaler.com security that's Zscaler. And we thank them so much for supporting intelligent machines. Jeff and Paris, I'm sorry, I monopolized the conversation. I'm a little bit. I'm very excited about what's happening with AI. I think this year.
Leo Laporte [01:40:38]:
I think a lot of doubters are going to have to eat their words. I think things are happening very rapidly now. Very rapidly. Some good, some bad. I mean there's the grock, the whole grock thing.
Leo Laporte [01:40:50]:
Well, that's. Anything that he touches is bad.
Craig Silverman [01:40:53]:
So yeah, as.
Leo Laporte [01:40:55]:
As. Who has it said that that's an E elon problem, not an AI problem? Gro's solution was, well, you gotta. If you have A paid account. You can generate sexual imagery.
Paris Martineau [01:41:09]:
Hey, you know, I guess he said it.
Leo Laporte [01:41:12]:
Well didn't, didn't open AI at one point say they were gonna do the same thing if you had a paid account.
Leo Laporte [01:41:16]:
They were gonn erotica for adults. If you are provably an adult. And by the way, I've noticed. Who was it? I think it was Anthropic said maybe it was Apple. Somebody said we were gonna verify your age before. It was when I was using either.
Paris Martineau [01:41:32]:
It was when you were trying to generate all that erotica. It was like, oh yeah, but of course we've gotta get your age.
Leo Laporte [01:41:38]:
Tell me a dirty story, Claude. No, I think I was using Claude on the iPhone and it said we gotta verify your age. Apple has my age. This is something Steve's always been talking about. He said, you know the platforms Android and iOS know your age. In many cases they could easily be and they have an API. In fact that's what whoever it was Anthropic or OpenAI were using. They said which age group are you in? That's the Apple API.
Leo Laporte [01:42:07]:
You don't say your specific age, just which age group you're in before they allow you to use it, period it. I think they're very concerned about miners using it.
Leo Laporte [01:42:16]:
Yeah, yeah, yeah, yeah.
Leo Laporte [01:42:21]:
Indonesia and Malaysia blocked croc. I think I mentioned that. UK's considering it by the way. As a result, Elon is on a rampage against the UK on X today. I wonder why as as your analysis, your clever analysis pointed out, Paris. One of the big stories this week is Apple has announced that yes, we are going to use Gemini from Google. We're going to pay them a billion dollars a year and it will be the foundation model for Siri and Apple Intelligence. But we're going to take their model, we're going to copy it from their server to our servers and we're going to run it in our own private Apple cloud.
Leo Laporte [01:43:01]:
So no information will go back to Google. Just the moment.
Paris Martineau [01:43:04]:
I do like that.
Leo Laporte [01:43:05]:
That, yeah makes sense.
Leo Laporte [01:43:06]:
I think it's what people want. I really do. We mentioned Claude cowork. I think that's less of a story than people have been making of it. It allows you. So the big thing, one of the examples they've used is see Normies can use it. You have to have a Mac and you have to have a paid cloud account but 20 buck account won't count. Normies are using it to do things like see all those icons on my desktop? Could you put them in folders and organize Them.
Leo Laporte [01:43:34]:
Them stuff.
Paris Martineau [01:43:36]:
Okay. Do you really need an AI to do that for you? Just organize your folders, people.
Leo Laporte [01:43:43]:
If it were so easy, I wouldn't have all those icons on my desktop.
Paris Martineau [01:43:48]:
You just need to set it so that your screenshots save in a folder called Screenshots. And then bada bing, bada boom. You're good.
Leo Laporte [01:43:55]:
Summarize my meetings from this week. Find action items. I think that's pretty good. Basically the idea. This is just what Copilot was doing from Microsoft. What Apple Intelligence will probably do once Gemini is transplanted. This is what these companies think people want, which is kind of like a little assistant. Right.
Paris Martineau [01:44:21]:
They do want a little guy who'll help them with their work. And that is nice.
Leo Laporte [01:44:25]:
Yeah. So it's worth a try. That's why initially I paid for the high end Claude, because you can't do it with the. I don't. I think you. I don't think you can do it with the $20 one. I can't remember. I think I remember I had to buy an expensive one.
Leo Laporte [01:44:38]:
But then I thought, well, I can use Claude Mo. Here's their example. You saw this, obviously. Look at all those icons. What a mess. Organize my desktop. That's quite a cluttered desktop. I see you have a mix of project folders, screenshots, photos, video files.
Paris Martineau [01:44:59]:
This person has deeper problems than a desktop.
Leo Laporte [01:45:02]:
I have seen many, many, many desktops like this. This is very common.
Leo Laporte [01:45:06]:
This is. Yeah, I've seen much worse.
Craig Silverman [01:45:08]:
Yeah, look at that.
Leo Laporte [01:45:10]:
Oh, my God.
Leo Laporte [01:45:12]:
No, I can't find anything.
Paris Martineau [01:45:13]:
You didn't even just do the right click. Organize the thing first like that.
Leo Laporte [01:45:16]:
Hap.
Paris Martineau [01:45:16]:
You can just do that?
Leo Laporte [01:45:19]:
I guess. I don't know. Yeah, not as well as that.
Leo Laporte [01:45:23]:
You're all.
Leo Laporte [01:45:23]:
Because it knows the contents are. Yeah. You should also be aware, it says, of the risks of prompt injections.
Paris Martineau [01:45:33]:
Yep.
Leo Laporte [01:45:34]:
Attempts by attackers to alter Claude's plans to content it might encounter on the Internet. We've built sophisticated defenses against prompt injections, but agent safety is still an active area of development in this industry. Hint, hint.
Leo Laporte [01:45:49]:
Not our fault.
Leo Laporte [01:45:51]:
We recommend taking precautions, particularly while you learn how it works. Well, there are going to be new attack vectors. There are. It's absolutely true. Have you seen. I'm sure you have, Paris, because it was on Reddit all the videos of the new driverless delivery vans in China.
Paris Martineau [01:46:11]:
I have not.
Leo Laporte [01:46:13]:
They have become a total meme. They plow. Well, watch. They plow through crumbling roads. They're relentless. They tend to.
Paris Martineau [01:46:23]:
Oh, my God.
Leo Laporte [01:46:24]:
They refuse. They refuse to Stop for anything.
Paris Martineau [01:46:31]:
There's also the videos of the walking robots, the humanoid robots attacking people on the street.
Leo Laporte [01:46:37]:
Really?
Paris Martineau [01:46:38]:
Yeah, just like running.
Leo Laporte [01:46:39]:
Watch out. Don't go on this man. Do you see he was waving his shovel, trying to stop it. And now this guy, the truck driver is getting out. He's saying he's gonna kick it. Get going. This one has a motorcycle attached to the wheels.
Leo Laporte [01:46:57]:
Luckily, there was no blood drilling behind.
Leo Laporte [01:46:59]:
Yeah, well, you gotta wonder what happened to the driver behind.
Leo Laporte [01:47:02]:
Yeah.
Leo Laporte [01:47:05]:
Yeah, I just, I thought, you know, I enjoy these robots gone wrong videos. So Meta has decided it's not the Metaverse. They've laid off a bunch of people in the Metaverse to Meta, and they say you can work on AI. I wonder if Apple's going to do the same. Apple continues to double down on the Vision Pro. Meta's reality labs, according to Engadget, has lost more than $70 billion in the last four years. Five years. $70 billion, in fact.
Leo Laporte [01:47:44]:
Meta's actively trying to change the narrative on this. They ordered, or they made sure people knew that they had ordered a doubling of the luxottica Ray Ban display Meta glasses. But a number of pundits have said, yeah, we don't think they're actually selling them. Oh, there isn't really a market for them. Every sales figure people can find shows it's not selling all that well. But if you ordered a double the production, that becomes the story. Right. And I think that maybe the more telling story is that they're laying off so many people from the reality labs.
Leo Laporte [01:48:23]:
But at some point, even though Zuckerberg I was just looking up, he controls near total voting control over Meta through dual class structure.
Leo Laporte [01:48:32]:
Oh, yeah, he was very smart about that.
Leo Laporte [01:48:33]:
10 votes per share of Class B, giving him around 60% of total votes.
Leo Laporte [01:48:37]:
So if you're a Meta Stockholder, you have one vote for every 10 that Mark Zuckerberg has.
Leo Laporte [01:48:42]:
But still, I wonder when there's going to be activist investors who come in and say, enough of this Meta crap.
Paris Martineau [01:48:48]:
I mean, Meta is closing down three VR studios as part of its latest round of Meta cuts. They're doing layoffs in the Metaverse team. I don't know what the activist would be pushing for outside of that. I mean, they're already clearly being like, yeah, the whole Meta thing was pretty dumb.
Leo Laporte [01:49:06]:
They should have done it two years ago.
Leo Laporte [01:49:07]:
Speaking of dumb, I'm gonna. You can watch while I cut up my American Express card. Because on Monday, while I was close, didn't you just.
Paris Martineau [01:49:18]:
Did you just show the card Number.
Leo Laporte [01:49:21]:
You can see the number because it's no longer good. Because Monday I got a text from T mobile saying, hey, Leo, you got 28,000 points that expire.
Paris Martineau [01:49:32]:
You did.
Leo Laporte [01:49:35]:
I said, click this link to see what your rewards could be. They're gonna expire tonight. I said I was your job. I know, I can't. I've never been bit by this before. The closest I can I saw, someone.
Paris Martineau [01:49:49]:
Added us on like blue sky being like, never submit any of your information to a. A random link. You get texted and I'm like, do.
Leo Laporte [01:49:57]:
They go to the site? Go to the site. Right, right.
Leo Laporte [01:50:02]:
My reserves were down. I guess it was early morning part of it. I blame T Mobile a little bit because T Mobile continually sends me promotional, legitimate promotional tests and you can't turn it off. Well, but the point is, I don't.
Leo Laporte [01:50:15]:
Know about his points.
Leo Laporte [01:50:17]:
The point is I was not surprised to get this message from T Mobile because they keep saying, oh, it's T Mobile Tuesday, you're gonna get a free hamburger. So I thought, oh, this is another, you know, one of those fine benefits from being a T Mobile subscriber, which I am. Okay, I'm saying this, I know it makes me look stupid. I'm saying it to everybody to know this can happen to anybody. Okay, because you're right. I should have known better.
Leo Laporte [01:50:44]:
I think we need a public interest ad made by AI as a hook. A fish hook in Leo's mouth.
Leo Laporte [01:50:50]:
Oh, it gets worse. Listen, listen to all the red flags. So the link is to T Mobile. L U O S A cc. That must be their fulfillment. So I clicked the link. Very nice. T Mobile.
Leo Laporte [01:51:06]:
I'm on my phone. T Mobile page says, oh, hey, you can get some Sony headphones, you can get an iPad, you can get or travel discounts. Would you like to use the travel discounts? I said, no, you know what, I don't need headphones, but I could give those as a gift. So let me get the Sony headphones. Said, great, here's the shipping fees, but we recommend the 99 cent one because you'll get it in five days. I said, good, but we need a credit card for the 99 cents. I said, that's fine. So I entered the credit card and there was a long delay and I should have, that should have been like while the hacker still didn't have one.
Leo Laporte [01:51:43]:
And then I get it says, okay, we're gonna send you a text. I got a text message and this is the other thing. So I use my Apple card for this. Actually that one Failed. So I said, oh, well, let me use a different one. So I put my Visa in and this spin. And then said, oh, here's a text message. And it said, the Visa text message was a little weird.
Leo Laporte [01:52:04]:
It said, to add your Visa to the Apple Wallet, enter this code.
Paris Martineau [01:52:08]:
Oh, no.
Leo Laporte [01:52:10]:
Which I realized in hindsight, the bad guys were smart, because what they're doing is instead of using these credit cards directly, they're adding them to an Apple Wallet. So it's. It's invisible to the merchant. Right. The merchant doesn't get the information about the credit card. So it's a good way of staying anonymous when you're using the stolen credit card. But then, you know, weirdly, there was a long delay. I kept refreshing the page.
Leo Laporte [01:52:32]:
It said, stay on this page till it's done. And it didn't work either. So I thought, well, I'll try one more. I'll do the American Express card. It was the business card.
Paris Martineau [01:52:42]:
Oh, you did another one.
Leo Laporte [01:52:46]:
Jeez.
Leo Laporte [01:52:47]:
So. And then finally I went.
Leo Laporte [01:52:50]:
When did the gulp come?
Leo Laporte [01:52:52]:
Wait a minute. When it said that didn't work either. I said, wait a minute. I must have been groggy. I don't know. I have no excuse. I have canceled all three of those cards. Now, here's the nice thing.
Leo Laporte [01:53:06]:
The Apple card, you just say that number. Give me a new one. So it just invalidated immediately. Lisa shouts down the stairwell about an hour later, saying, did you just try to charge something at Lowe's for 500 bucks? I said, no. He said, well, American Express blocked it. I said, well, I hope so, because I froze the card. So they. Within an hour, they tried to use it.
Leo Laporte [01:53:31]:
I've canceled all three cards. That's why you saw me cutting that up. I know. It's embarrassing.
Leo Laporte [01:53:37]:
It could happen to any of us, Leo.
Leo Laporte [01:53:38]:
Well, that's why I bring it up. Just to remind you, you. It can happen. It's easy.
Paris Martineau [01:53:43]:
I feel like I am so far in the opposite way that it's like, it. It is a impediment to my daily life. Like, I have had issues where I get a legit text from bank of America being like, your credit card's been compromised. Like, please click here to, like, you know, verify transaction. And it definitely, like, wasn't me. They were correctly texting, texting me. And I will, like, call bank of America from a different number I have saved in living. Like, I'm.
Paris Martineau [01:54:12]:
Clearly, someone's trying to hack me. And they're like, no, no, no. Someone did take your credit card, but that message was Real.
Leo Laporte [01:54:19]:
Yeah, I, you know, obviously I knew. I knew better. This is every possible red flag. I've been telling people for 20 years not to do this. Somebody's asking, didn't you use your sponsor privacy.com? if I. I wanted to take the time, I could have generated a one time only card. Yeah. But I didn't.
Leo Laporte [01:54:37]:
He was too much in a rush to get.
Leo Laporte [01:54:40]:
I want those headphones. They're free and it was going to expire tonight. I didn't want to lose those points.
Paris Martineau [01:54:46]:
You could never lose those points.
Leo Laporte [01:54:48]:
Did you actually look at the T mobile site to see how many points you do have?
Leo Laporte [01:54:52]:
Yeah. Well, after all of this, I went to the T mobile site. They don't offer points. My mistake was doing that after I gave them three credit card numbers.
Leo Laporte [01:55:05]:
Oh, that's perfect.
Paris Martineau [01:55:06]:
Was there ever any moment between credit card one and two and credit card two and three where you're like, no.
Leo Laporte [01:55:12]:
The thing was hitting me on the head saying, don't do this. It was like spinning. It was slow. It didn't work. There was every reason to go, oh yeah, this is bogus. Luosa. CC I have no excuse. I must.
Paris Martineau [01:55:27]:
You know what?
Leo Laporte [01:55:28]:
Stoned. I don't know what was that?
Paris Martineau [01:55:29]:
Seems like a link that's generated by a website that I'm going to cannibalize for my picks the week called creepylink.com it sounds like I'll go to right now. It's a website that's kind of a URL shortener or in some ways a URL lengthener that it will make any link look absolutely creepy and like it's going to give you a virus. For instance, I'll put in Twitter TV here and creepylink.com and it's given me google.cl Ic.link. a bunch of string of text underscore free money.
Leo Laporte [01:56:05]:
Yeah.
Paris Martineau [01:56:06]:
Bac.
Leo Laporte [01:56:08]:
No. If I'd seen a lot of numbers and stuff, I might have. I. There is certainly I would have. I don't know. I don't know what I would have done. I don't. I.
Leo Laporte [01:56:16]:
It was like. It was. It was like 7:30 in the morning. It was. I wasn't fully awake. I don't know. I don't know how that happen. By the way, I'm gonna click that link.
Leo Laporte [01:56:24]:
What do. Is there free money there?
Paris Martineau [01:56:25]:
There is free money and it's gonna give you something.
Leo Laporte [01:56:27]:
Oh, did you see what. Hey, look at that. It's pretty good. Discord said. Did you see what you do when you click it on Discord potential Dangerous download with purple. Guy says, go, Leo.
Leo Laporte [01:56:40]:
Don't do it.
Leo Laporte [01:56:41]:
Don't do.
Paris Martineau [01:56:44]:
Certainly does make your links look as suspicious as possible.
Leo Laporte [01:56:47]:
If you're in a buying mood, though, Scooter X put up a link to a hands on with a new B. Now that it's an Amazon hands.
Leo Laporte [01:56:54]:
Yeah, big deal, right?
Craig Silverman [01:56:56]:
They were showing it easy os.
Leo Laporte [01:56:59]:
They were showing it at ces. In fact, we had Jennifer Patterson Tuohy on and she actually is wearing one now. I don't. You know, I was tempted. It's funny because I already gave my be away to my hairdresser because I wanted her to record all those personal conversations she has every day with her clients. She couldn't get it working on Android. I probably should have kept it, but I just. I didn't.
Paris Martineau [01:57:22]:
I love that you're like, this is a security risk for me to have, and it's a bad product. I'm gonna give it to you someone a lot less.
Leo Laporte [01:57:30]:
I had told her that I was wearing, and I showed her and talked about it, and she said, oh, that's cool. I want that.
Paris Martineau [01:57:36]:
I want to be able to afford all of these.
Leo Laporte [01:57:38]:
So that that week, instead of tipping her, I just gave her the B. I still tipped her. I still tipped.
Leo Laporte [01:57:45]:
I said, my salon should do a reality show.
Leo Laporte [01:57:49]:
I go to a barber shop. I just want to tell you, I go to a barber shop where they offer you tequila when you walk in the door.
Paris Martineau [01:57:56]:
Not even whiskey.
Leo Laporte [01:57:58]:
Tequila feels crazy. They have whiskey, but they. Oh, yeah. I mean, they're.
Paris Martineau [01:58:02]:
Are they giving you just like a shot, or are you just getting straight tequila in a glass like it's whiskey?
Leo Laporte [01:58:06]:
They were doing shots. Yeah, okay.
Paris Martineau [01:58:08]:
Okay. That's. I guess, fine.
Leo Laporte [01:58:10]:
I like barbershops.
Paris Martineau [01:58:11]:
I want to go to.
Leo Laporte [01:58:12]:
They had a girly calendar on the wall.
Paris Martineau [01:58:15]:
Oh, boy.
Leo Laporte [01:58:16]:
But what's funny is my barber is a woman. And I said, does that bother you? She said, well, a little bit. Anyway, we moved. She left that barbershop. And you followed her, all male. And I followed her to her own. She owns it now. It's called Las Katrinas.
Leo Laporte [01:58:33]:
It's really good. I like it.
Paris Martineau [01:58:35]:
Nice. And if you go there, you will be recorded.
Leo Laporte [01:58:38]:
Yes. No, she never got it working because she had an Android phone.
Paris Martineau [01:58:42]:
Oh, it was.
Leo Laporte [01:58:44]:
I'm in danger. You're watching Intelligent Machines, the show that just goes on and on and on and on with Jeff Jarvis and Paris Martino. Are you working on a big story? Is there a radioactive shrimp in your future?
Paris Martineau [01:59:01]:
I'm working on another story. Yeah, you mentioned you went to the courthouse. I went to a courthouse today to go look at some documents because we have a really messed up system in the US where if you want to look at court records for many of our district courts, you have to pay money for each document you look at based on the amount of pages you're looking at, as if you're printing them out. And that gets really annoying and expensive. Expensive. And adds up when you're trolling through cases that maybe have like 40 associated cases with. And you got to figure out what document you need. But I've realized you can view all these for free if you go to the actual courthouse where those documents are.
Leo Laporte [01:59:40]:
Associated, as it should be.
Paris Martineau [01:59:43]:
But I think I should be able. I just go on a website that's a free version of a website where I have to pay. It costs them nothing.
Leo Laporte [01:59:50]:
The website's address, Free money. Google.
Paris Martineau [01:59:56]:
Yeah, it should be, because that's what it felt like. But instead I just had to check all my electronic devices and sit in a room where all the clocks were wrong. The ones on the computer kept saying it was 9:54. And I was like, it's very clearly mid afternoon. And so I had to keep going up to this man behind a desk. Meaning like, what time is it pointing at something else, a different clock. That'd be kind of wrong. And then I looked through them.
Leo Laporte [02:00:28]:
You couldn't bring the documents?
Paris Martineau [02:00:29]:
No, no. I had to bring in pencil and paper and I wrote on my legal pad the names of the document exhibits because I already knew the case numbers that I want to. Then tonight, once I'm done with this podcast, go and pay money to download on my computer.
Leo Laporte [02:00:47]:
Wow.
Paris Martineau [02:00:48]:
Because that. That's a really normal way to.
Leo Laporte [02:00:50]:
But you can't tell us what you're working on. This is.
Paris Martineau [02:00:52]:
I can't. It is a secret.
Leo Laporte [02:00:54]:
Good. And. And how long does it take you to do a story like this?
Paris Martineau [02:00:59]:
Well, it depends. This is a story me and my editors have been kind of kicking around thematically for a bit, but it's kind of a complicated topic on first glance. Seems a bit boring or esoteric, but every time I explain it to people, they get very interesting. So I'm trying, but I mean, it's. It's difficult to figure out how we drill that down into a couple different stories that will interest people in the topic. And I think we found one angle for it, but there's a couple different kind of procedural questions we have to answer first. So that's.
Leo Laporte [02:01:31]:
This is. Are you. Is this a. Is this typical Process. It sounds like Consumer Reports is really more detailed, focused, more thought.
Paris Martineau [02:01:38]:
They really are. I've really enjoyed working here. It's just a lot of people who know their stuff and care a lot about taking the time to get it right.
Leo Laporte [02:01:46]:
Because I would imagine most websites would just say, yeah, whatever, go ahead and do it. You got three more stories to do by noon.
Paris Martineau [02:01:52]:
So, yeah, they're just like, whatever, do anything.
Leo Laporte [02:01:55]:
Do anything. Just fill that page. Yeah, yeah, good. I'm so glad you got that job. What a great place to work.
Leo Laporte [02:02:02]:
And they should be glad they got you.
Leo Laporte [02:02:04]:
I think your colleague is going to be on Twitter on Sunday, Nicholas.
Paris Martineau [02:02:07]:
Oh, lovely.
Leo Laporte [02:02:10]:
All right, all right, all right. We're going to take a break. You're watching Intelligent Machines with this intelligent person, Paris Martin, Owen, and then a couple of dummies, Jeff and Leo. Actually, I'm the dummy. Look, you got this brilliant investigative reporter, you got this former journalist, now esteemed professor, deep thinker and author. And here I am, I'm a podcaster.
Paris Martineau [02:02:35]:
I will say I did somehow end up showing someone Dev Nole last night. They were like, whoa, he used to do that. Does he look like that? They're like, oh, he worked with Soledad. That's cool. I was like, she hated him. She hated Dev. Not Leo, but Dev.
Leo Laporte [02:02:56]:
No, I think she. I think she. I don't know. We tried to interview her and then she canceled. So maybe she doesn't. Maybe she realized. She went, leo. Oh, not going to interview there.
Leo Laporte [02:03:04]:
No. This is Intelligent Machines. All right, all right, all right. Matthew McConaughey is trademarking himself, saying, all right, all right, all right.
Leo Laporte [02:03:17]:
Ego. He's the actor I cannot bear most.
Leo Laporte [02:03:22]:
He wants to fight AI misuse. There he is in the 1993 film Dazed and Confused, which is where he coined that term, or the screenwriter coined that term. All right, all right, all right. Trademarking.
Paris Martineau [02:03:37]:
He actually outlived that line.
Leo Laporte [02:03:39]:
That is his line.
Paris Martineau [02:03:39]:
He actually did that, did he?
Leo Laporte [02:03:40]:
Yeah, he.
Leo Laporte [02:03:41]:
Oh, good. All right, so he should trademark it. He had eight trademark applications approved by the US Patent and Trademark Office featuring him staring, smiling and talking. According to the Wall Street Journal, his attorney said the trademarks are meant to stop AI apps or users from simulating his voice or likeness without permission, like on Sora, right? There's a seven second clip of him standing on a porch. There's a three second clip of him sitting in front of a Christmas tree. And there's audio of him saying, all right, all right, all right.
Leo Laporte [02:04:16]:
So, Leo, what would you trademark of yourself? What is the trademarkable Leo Laporte.
Leo Laporte [02:04:22]:
But see, I have no desire to do that. I want people to use my. I know my image.
Leo Laporte [02:04:27]:
What is the essence of you?
Leo Laporte [02:04:30]:
I guess there's. There's really nothing. Empty shell. I don't know.
Paris Martineau [02:04:36]:
You are nothing.
Leo Laporte [02:04:37]:
I think you watch the show and you. And. And it's over and you go, who was that guy in the middle? I don't. I don't use a non entity. I'm like Joey Bishop. I'm like. I'm like, sorry, Paris, can we leftover.
Paris Martineau [02:04:53]:
Have a brief discussion of what Anthony just posted in the chat, which is a full character sheet for Dev Null to be able to. I ostensibly recreate him in 3D videos.
Leo Laporte [02:05:07]:
We don't. We don't really even need that. I talked to the guy who was at the company that created Dev Null and he said, oh, yeah, I have all the files on a hard drive.
Paris Martineau [02:05:16]:
Well, they'll need it because Anthony's got a 3D recreation of Devnall's butt right here.
Leo Laporte [02:05:21]:
It's a little too good. Look, looking. Honestly, he wasn't that.
Paris Martineau [02:05:25]:
I think devnold needs one gold tooth. That's my pitch.
Leo Laporte [02:05:31]:
Although I am going to trademark this image from. From Pretty fly for assist guy. Now, do you think Matthew McConaughey would get mad at that?
Leo Laporte [02:05:42]:
Yeah, you're sued.
Leo Laporte [02:05:43]:
I'm in deep trouble. Yeah, I don't know. I would never wear that watch, though. So that's pretty fun. Funny. It's. I like what you can do with. With AI.
Leo Laporte [02:05:54]:
Have you shown Paris the suit you had to wear when you made Dev? No.
Leo Laporte [02:05:59]:
Like, I have it over here.
Paris Martineau [02:06:01]:
Do you have a suit with just a bunch of little balls?
Leo Laporte [02:06:04]:
Think of a wetsuit. Yeah, think of a wetsuit. It was. It was neoprene, full body neoprene, except for the head stopped at neck and it had ping pong. Like ping pong ball like things on it? No, actually, yeah.
Paris Martineau [02:06:16]:
Like the suits they put those cats in for the cat kids game.
Leo Laporte [02:06:19]:
Yeah, exactly. Or it was actually what that would be if you were on camera. This didn't have ping pong balls because there was no camera. It was all electromagnetic sensors so it would sense your motion. So, but they were still things stuck all over the suit. But I don't have a. I don't have a picture. I don't think.
Leo Laporte [02:06:36]:
Maybe there is a picture somewhere.
Paris Martineau [02:06:38]:
I think you should show up one day to record just in one of those suits.
Leo Laporte [02:06:42]:
Yeah, I think so.
Leo Laporte [02:06:45]:
And I think there was. I feel like there was a headpiece because and then. But I had puppeteers. I was the one doing the movement. There was a person with a like a little joystick doing my mouth movement. And then Carson, my old producer was spinning my hair and making my eyebrows go up and down. So that was. And that was way back when.
Leo Laporte [02:07:09]:
That was 1994. I mean this was very early on doing real time mocap. It wasn't even pre recorded. It was done live.
Paris Martineau [02:07:17]:
That rig must have been so expensive.
Leo Laporte [02:07:20]:
It was attached to a half million dollar Silicon Graphics onyx.
Leo Laporte [02:07:24]:
Jesus.
Leo Laporte [02:07:25]:
NBC, baby.
Paris Martineau [02:07:29]:
Back when network TV was the thing.
Leo Laporte [02:07:33]:
I have a pick for the, for the network TV thing later on. It's actually for Paris to get, get her caught up to the cultural milieu that she is bathing in right now.
Paris Martineau [02:07:46]:
Beautiful.
Leo Laporte [02:07:46]:
Yeah, I have been monopolizing this. Please, you guys, your stories of the week. Are there any archive.org stories that you wanted to.
Paris Martineau [02:07:58]:
Sorry.
Leo Laporte [02:07:58]:
If I write this program, are there going to be more archive.org stories?
Paris Martineau [02:08:03]:
I've gotten really distracted looking up photos of cats in motion capture suit. I see the discord. I see they're just really cute.
Leo Laporte [02:08:11]:
Have you ever played the game Stray?
Paris Martineau [02:08:13]:
I have, but then I got really sad because there's a couple quick time events in the game where if you don't move the cat fast enough, it gets overwhelmed by little glommy gooey guys in the cat. And it made me so deeply sad to see the cat die that I was like, I literally can't play the game anymore.
Leo Laporte [02:08:33]:
It didn't make me sad. It angered me. Angered me.
Paris Martineau [02:08:37]:
But it made me sad. The cats seem too real.
Leo Laporte [02:08:41]:
Well, it was a chance to experience life as a cat.
Paris Martineau [02:08:46]:
I know. And I don't want to experience life as a cat getting overwhelmed by glommy gooey guys.
Leo Laporte [02:08:52]:
Yeah. I'll give you a story while you guys look for your stories. The Senate yesterday unanimously unanimously passed the Disrupt Explicit Forged Images and Non Conceptual Edits act act or Defiance Act. No roll call vote. No Senator objected. This is of course in response to Grok and the deep fakes. Of course. Now it goes to the House where it will die a quiet death.
Leo Laporte [02:09:24]:
I don't know. Maybe not maybe.
Paris Martineau [02:09:26]:
People, I do feel like there's been a surprising amount of political capital on the Hill for these sort of bills.
Leo Laporte [02:09:33]:
Yeah, yeah. This, by the way, this bill does not give law enforcement any new tools. All it does is say as a victim, you can sue. And this is, remember Texas did this. This is the new thing, which is, oh no, we don't want to enforce laws. We're just going to let people bring civil suits, tie up the courts.
Leo Laporte [02:09:52]:
That'll solve the problem and will enable citizens to sue.
Leo Laporte [02:09:57]:
Yeah, citizens. That's what this. That's the one thing that this bill does. If you're a victim, you can sue. Now, interestingly, not only sue the creator of the Deepfake, but the platform. So you could in fact Sue Grok.
Paris Martineau [02:10:10]:
Or X.com as I will choose a story here that just came in over the wire. I'll put it in the live chat. X announced has, while we've been sitting here recording this, that it's going to no longer. Grok will no longer allow the editing of images of real people into bikinis, as well as they're kind of making some other just general changes to this.
Leo Laporte [02:10:39]:
Well, why, Elon? There was no problem. You said it never happened. Never happened.
Leo Laporte [02:10:43]:
Zero. But if it does, we're gonna stop. Yeah.
Paris Martineau [02:10:48]:
It says that basically you can't use a Grok account to edit images of real people into revealing clothing generally, and that all of the image generating features are going to be behind its subscriber paywall. And the ability of all users to generate images of real people in bikinis, underwear and similar attire will be GEO blocked in regions where it's illegal, which seems like not exactly the point.
Leo Laporte [02:11:16]:
Make it legal everywhere where it's not illegal.
Paris Martineau [02:11:18]:
Yeah.
Leo Laporte [02:11:18]:
Yeah. Jeff, your turn.
Leo Laporte [02:11:22]:
We could talk about the billionaire tax and how pissed I am at the founders of Google to have already escaped.
Leo Laporte [02:11:29]:
Yeah. Larry Page bought a mansion in Florida where there is not only no billionaire tax, there's no state tax at all.
Leo Laporte [02:11:36]:
Moved his companies out of, out of California.
Leo Laporte [02:11:39]:
And in fact, now Gavin Newsom, our governor, is saying, oh, I'm going to stop that law. We can't lose our billionaires. Oh, heaven forfend, no billionaires. Whatever will we do, Gavin? They're not paying any tax. They're just leeches. It's okay for us to lose them. It might have help with the housing crisis.
Leo Laporte [02:11:59]:
Yep. Yep. So if it, if, if, if Newsom vetoes it, which he says he will do, then there'll be an effort in the crazy California to get it on.
Leo Laporte [02:12:08]:
The ballot, which will pass. I mean, I, I think it's a good idea. I mean, look, there's no reason a billionaire should.
Leo Laporte [02:12:17]:
The argument. I think I don't understand how the billionaire's tax works, but the argument often in these cases is that if you tax assets as opposed to income, then if you, let's say, start taxing a.
Leo Laporte [02:12:29]:
Farm, there'll be no Problem left, tax income. I agree. The problem is billionaires are very good at deferring and hiding income.
Leo Laporte [02:12:37]:
Right.
Leo Laporte [02:12:38]:
They basically borrow against assets and spend that money.
Leo Laporte [02:12:41]:
But there needs to be a way to get these suckers.
Leo Laporte [02:12:45]:
Yeah.
Leo Laporte [02:12:45]:
Peter thiel has donated $3 million to fight the billionaires.
Leo Laporte [02:12:48]:
Yeah, I, I understand both sides of this. I really do. It's, you know, it's problematic. It wouldn't. It's. The billionaires wouldn't suffer. They can afford it. It's the people who have stock options or stock awards from their companies.
Leo Laporte [02:13:03]:
They're working at Facebook or Apple and they get stock awards.
Leo Laporte [02:13:07]:
They can't buy a house with it yet.
Leo Laporte [02:13:09]:
Yeah. They can't do anything with it. It's not actual money. But they are taxed on it.
Leo Laporte [02:13:13]:
I understand that Jensen Wong has said he has no objection to the tax. Interesting that.
Leo Laporte [02:13:20]:
Honestly, the really big billionaires like Larry Page and Jensen Wong. That's a rounding error.
Leo Laporte [02:13:26]:
Yeah, it's nothing. Well, they're, They're. Their increase in their equity value is far exceeds 5%. If it's only 5% a year, then they're doing it wrong.
Leo Laporte [02:13:36]:
Patrick Delahany says Massachusetts does this. People said, oh, no. All the billionaires leave Massachusetts, but the number in the state went up after the taxes went into an effect. And what are they spending the money on? It means every child in every school in Massachusetts gets lunch free.
Leo Laporte [02:13:51]:
Yeah.
Leo Laporte [02:13:53]:
And they say there's no such thing as a free lunch. I don't know why I put this in the rundown. I think maybe because I want to make Paris listen to a 1952 new recording of J.R.R. tolkien reading from the Hobbit.
Paris Martineau [02:14:09]:
Please do. Yeah.
Leo Laporte [02:14:11]:
I think it's incredible that this surfaced right down here with a dark water lived old Gollum. I don't know where he came from.
Craig Silverman [02:14:21]:
Nor who or what he was.
Leo Laporte [02:14:23]:
This was 1952. This was very early in the tape recording era. And somebody sat down with Tolkien and said, hey, hey, look at this cool thing I've got. Would you. Would you record. Read some of your book into it. He also sang, apparently.
Paris Martineau [02:14:42]:
Oh, I would like to. Did he sing in Elvish?
Leo Laporte [02:14:46]:
That's a good question. When Tolkien visited a friend in 1952 to retrieve a manuscript, he was shown a tape recorder. Having never seen one before. He said, how does it work? And it was. Was and was then delighted to have his voice recorded and hear himself played back for the first time. The only reason this is newsworthy, these are. These have been around for a while, is that somebody has put them up on YouTube. So if you, if you want to hear it, search YouTube for JRR Tolkien.
Leo Laporte [02:15:12]:
That's pretty cool. I've heard these before.
Paris Martineau [02:15:16]:
So another interesting thing, I don't know if this is in the rundown. I should have probably added this. Dig Dig launched new new Dig today.
Leo Laporte [02:15:26]:
Oh, I've been using it for a while.
Craig Silverman [02:15:28]:
Oh.
Leo Laporte [02:15:28]:
I guess it's now public.
Paris Martineau [02:15:28]:
What is? I guess your. I'm curious as to your experience so far because it's basically from what I've heard of the writing about it, it seems to basically be Reddit. Like there's communities posts, comments. Yeah, it's Reddit upload Dig and it's.
Leo Laporte [02:15:43]:
All invite Reddit copy dig. Exactly.
Paris Martineau [02:15:45]:
I know but Digg has abandoned that for long enough that it is Reddit is. Is more ingrained in the common mind for these features now than Dig ever was. And I think that's always been a bit of a shame given what a early advantage Dig had to this.
Leo Laporte [02:16:03]:
I'm logging. Well, I'm. I guess I won't log in. So.
Paris Martineau [02:16:06]:
One thing that I saw in some of the coverage of this is that Kevin Rose is kind of emphasizing the trust like anti bot angle of it as like kind of an interesting aspect of the this. Basically he's betting like that this AI era social media needs new verification tools beyond just like a blue check mark and says that they're going to use zero knowledge crew proofs like cryptographic verification without revealing personal data and other quote signals of trust bundled together to identify people within specific communities. Like one of the examples they gave is a community for Oura Ring owners could verify that everybody actually owns an Oura ring. That's in order to participate in it.
Leo Laporte [02:16:50]:
Kevin, first of all, Kevin's a good friend, longtime friend. I've known him forever. He and has been on our show many times. He was one of the founding members of our panels. The problem with Digg was that everybody gamed it. They went out of business because it was a very successful way to get attention for your blog or your website or whatever. And people started gaming it and they couldn't figure out a way to not. Not to keep it from gaming.
Leo Laporte [02:17:15]:
And so he was briefly on the COVID of Business Week. Briefly. He was for a whole week on the COVID of Business week as the $60 million man, which Kevin took great pains to lift down, saying I never made that any money from Digg. They eventually kind of sold it.
Paris Martineau [02:17:31]:
But Reddit sold the stuff sell ads, right?
Leo Laporte [02:17:34]:
Yeah, well, Reddit did it right and was very successful. The reason the thing that Dig did that Reddit took was this idea of voting stuff up. And so Dig became, you know, you would go there and see the most voted stories would be the top stories. So just as we were talking about earlier, how do you figure out what are the stories to pay attention to? Yeah, so I didn't realize that they. My.
Leo Laporte [02:17:57]:
My boss, Steve Newhouse, when I worked for him at advance, really wanted to invest in or buy Dig, but he couldn't get anywhere with them. So he got this. He took the second choice, which was Reddit.
Leo Laporte [02:18:09]:
Interesting. And it did very well.
Leo Laporte [02:18:11]:
And he was smart. He owned it fully for very little money. But then he spun it out so he could create equity in it.
Leo Laporte [02:18:18]:
Right.
Leo Laporte [02:18:19]:
To motivate the. The staff. And it's grown to be huge now.
Leo Laporte [02:18:22]:
It. It looks like kind of. I guess. Yeah, it looks like Reddit. It's not.
Paris Martineau [02:18:26]:
Yeah. What does it look. Anything going on?
Leo Laporte [02:18:29]:
Here's slash Food.
Paris Martineau [02:18:30]:
The bagel of life. All right.
Leo Laporte [02:18:33]:
I wouldn't need it.
Paris Martineau [02:18:34]:
Do they have a subreddit for conspiracy theorists that believe.
Leo Laporte [02:18:38]:
How about let's see if they have it for sovereign citizens.
Paris Martineau [02:18:40]:
I was about to say sovereign. The sovereign citizen subreddit is really what I'm always looking for in a communicant.
Leo Laporte [02:18:48]:
No communication.
Paris Martineau [02:18:49]:
You could start it.
Leo Laporte [02:18:51]:
Oh yeah, I could. I don't know what other subject you said conspiracy. They must have that.
Paris Martineau [02:18:57]:
Well no, I really mean meant sovereign citizen. I assume they don't have a conspiracy is one being. It's an early version of it.
Leo Laporte [02:19:07]:
But they must have an what is.
Paris Martineau [02:19:08]:
Do they have a like what's. Do they have a what's hot popular. Oh, like what is the most popular.
Leo Laporte [02:19:14]:
One post in AI core coding?
Paris Martineau [02:19:17]:
Well, it's only been a couple of hours.
Leo Laporte [02:19:19]:
Well no, but I've been. There's been the non public beta has been around for months. Let's see what the most Doug story is. No, I'm not going to show that. How Italian hazelnuts built an empire so rich it shouldn't pay taxes.
Paris Martineau [02:19:36]:
Cheesecake. Oh no, you're still on food. You're on food subreddit.
Leo Laporte [02:19:40]:
Oh, I was on the most Doug. Okay, so here's trending. Trump says Americans don't need Canadian products.
Paris Martineau [02:19:47]:
You're in finance now. On the right is what it shows.
Leo Laporte [02:19:51]:
Oh, but I'm.
Leo Laporte [02:19:53]:
How does you get to a home with everything?
Leo Laporte [02:19:55]:
Let's go to the Keep doing some.
Paris Martineau [02:19:58]:
Back click on a little dig.
Leo Laporte [02:20:04]:
Feed trending Most dug. That's what I want. All of Dig Dig rebooted under original founder Kevin Rose and Alexis Ohanian, who ironically was the original founder of Reddit and is now married to Serena William.
Paris Martineau [02:20:23]:
Yeah.
Leo Laporte [02:20:24]:
Okay. You know, it's like Reddit capture a path. You know what? I. I was. It was pretty slow when I went originally for the beta, but now that it's going.
Leo Laporte [02:20:36]:
See what's there.
Leo Laporte [02:20:37]:
I'll check it out. And it has. What's this? Dig Daily. It looks like a audio.
Paris Martineau [02:20:44]:
Oh, it must be interesting.
Leo Laporte [02:20:46]:
It's not. It's not doing anything, but I guess this. It's like a podcast. Yeah. Who's excited for the animal? This looks just like Reddit, doesn't it?
Paris Martineau [02:20:55]:
I was gonna say, is there an animal cross? Is there a new animal?
Leo Laporte [02:20:58]:
There is a new one for. You have to have the switch to or. Oh, yeah. They've added a lot new content. Yep, yep, yep, yep, yep. It's. It's pretty cool. Are you.
Leo Laporte [02:21:09]:
Are you ready to go back, Paris?
Paris Martineau [02:21:11]:
I haven't. Well, it'll be quite sad because Animal Crossing does that thing where you haven't been there in years where everybody looks at you and he's like, oh, it's been six years.
Leo Laporte [02:21:21]:
And they say exactly how long it's been like six years, three months and four days since you've been here.
Paris Martineau [02:21:27]:
I have long said the thing that New Animal Crossing is missing and I will die on this hill is they need to bring back Resetti. They need to bring back that annoying guy who gets really mad at you if you reset your games that you can take advantage of of the time feature of it to get better stuff. And it. I think that's just a wonderful game mechanic that frankly, more games need to be using because I love that there's a little guy that yells at you.
Leo Laporte [02:21:59]:
Apparently Cap' n has a family and they have a resort on the pier. It's kind of cool. That pier was empty for so long. There's some new.
Paris Martineau [02:22:09]:
Oh, my God. There is a Resetti's reset service.
Leo Laporte [02:22:13]:
I'm happy. Clear your island with Resetti's reset service. So here he is. I want the entire island cleared and then you start over.
Paris Martineau [02:22:22]:
Listen, that's not what Resetti does, but I'm happy I get to see him.
Leo Laporte [02:22:28]:
Yay. I might. Yep, I might pick it up now. I don't have the switch 2 yet. I. I should buy.
Paris Martineau [02:22:34]:
Can you play this without the switch too?
Paris Martineau [02:22:35]:
This looks like they found him a job. Paris. They found Resetti a job.
Leo Laporte [02:22:39]:
Founder said he a job instead of annoying.
Paris Martineau [02:22:41]:
But you know I think his job should be to give you increasingly long monologues about the importance of playing your game the right way.
Leo Laporte [02:22:50]:
Yes, you can Upgrade it. The 3.0 update. Is it for everybody. There is a $5 Switch 2 version which has enhanced graphics and mouse controls, game chat capability, and the ability to host up to 12 people on your island at once. This was horrors.
Paris Martineau [02:23:08]:
What horrors await me. I haven't been back there since COVID Yeah.
Leo Laporte [02:23:11]:
This is our quarantine game. What we did when we couldn't leave the house.
Paris Martineau [02:23:15]:
Yeah. Maybe we should all get back into it and connect our islands.
Leo Laporte [02:23:18]:
Okay.
Craig Silverman [02:23:20]:
No.
Leo Laporte [02:23:20]:
It's such a time sink.
Craig Silverman [02:23:22]:
I've got.
Leo Laporte [02:23:22]:
I've got to spend time with Claude now. He's very demanding. I did like this story and it has nothing to do with AI, but I'm going to do it anyway. Anyway. Postal arbitrage.
Leo Laporte [02:23:33]:
Yeah, I didn't. Couldn't figure this out.
Leo Laporte [02:23:34]:
Okay, so if you wanted to send a postcard or a letter to a friend right now, can you believe it? A stamp is 78 cents.
Paris Martineau [02:23:45]:
That's messed up.
Leo Laporte [02:23:45]:
That's messed up. But it turns out if you're a prime member, it will send stuff items for less than that. In fact, they sell items for less than that with free shipping. So what they're doing is a way to stay in touch with friends, is exchanging cheap items like a lime or a packet of Kool Aid Tropical punch powder mix. Just a little thing that's cheaper than a letter.
Paris Martineau [02:24:09]:
And then you put on the. On the little note that you can send with it. Your letter.
Leo Laporte [02:24:14]:
Yes. It's called postal arbitrage.
Paris Martineau [02:24:19]:
I will say, though, if anybody who's looking to do this to try and keep in touch with their friends, the little gift note that you can send with your prime packages is like a huge heat stamp. So it will. It will disintegrate into nothing. The text will go away after a couple of weeks.
Leo Laporte [02:24:38]:
So make sure. Well, it's. It's overnight.
Paris Martineau [02:24:40]:
I mean. Yeah.
Leo Laporte [02:24:41]:
All right, Paris, I'm gonna send you a russet potato. I'll let you choose.
Paris Martineau [02:24:45]:
Thank you.
Leo Laporte [02:24:47]:
How much carbon goes in the atmosphere with every trip?
Paris Martineau [02:24:50]:
Christ. Yeah.
Leo Laporte [02:24:51]:
Oh, we are burning the future of our. Our children. But. But damn it, we've got free postage.
Paris Martineau [02:25:00]:
For now until the postal service shuts down. Jammer. Be just had a great idea, which is twits. 24 hours of animal crossing. And I agree. We could. We could do an hour of those just sending items to one another.
Leo Laporte [02:25:13]:
I guess I could. With the HDMI port connect I could. I could connect it up.
Paris Martineau [02:25:18]:
Easy.
Leo Laporte [02:25:19]:
Probably do it right now.
Paris Martineau [02:25:23]:
Listen, we could do a solid 1/3 to 1/2 of the 24 hour stream. Just gaming.
Leo Laporte [02:25:30]:
It's very tempting.
Paris Martineau [02:25:32]:
And as someone who used a sad lamp for the first time today, I think if we just had one of those babies running for the whole live stream, we'd stay awake the whole time.
Leo Laporte [02:25:41]:
Okay, you talk amongst yourselves while I see if I can hook this up.
Paris Martineau [02:25:45]:
Are you going to play Animal Crossing right now?
Leo Laporte [02:25:48]:
I. I'm gonna fall over.
Paris Martineau [02:25:51]:
Jeff can't even comprehend the idea of playing a video game about his niche on a.
Leo Laporte [02:25:56]:
On a hot pad, on a sore ass.
Paris Martineau [02:25:59]:
Aren't we all?
Leo Laporte [02:26:00]:
Yeah.
Paris Martineau [02:26:01]:
When are you. When are you going One eye.
Leo Laporte [02:26:04]:
Next Wednesday. So I'll come exciting in the afternoon like that.
Leo Laporte [02:26:09]:
It's too complicated. I have to move the whole dock which has to be plugged in and all that. I'll have it for the next show and we can show a J. Jeff pentamento.
Paris Martineau [02:26:16]:
We should.
Paris Martineau [02:26:17]:
You just need to plug it into the blackmagic in your. In your set. That's it? That's all you need?
Leo Laporte [02:26:21]:
Yeah, no, I. Yeah, it has. So. But I need the dock for the HDMI out and the dock is and is all plugged into the back of my credenza, so. How dare they move the credenza.
Paris Martineau [02:26:34]:
Could never move a credenza.
Leo Laporte [02:26:35]:
And the power supply out. It's really rather entangled back there. And I have. I really haven't played it. It's. You're right. You get. You have roaches in your house and Spiderwoman webs and they.
Leo Laporte [02:26:45]:
Everybody kind of complains about how long you've been gone and. All right, let's.
Leo Laporte [02:26:50]:
I tried to use VO and I used our pictures. If you go to the Discord and you scroll up, you'll see the three videos I made about us.
Leo Laporte [02:26:59]:
Oh.
Leo Laporte [02:27:00]:
In VO this new vio, we're not quite recognizable.
Leo Laporte [02:27:06]:
No. He looks a little bit like Peter. No, that's not it.
Leo Laporte [02:27:10]:
It. You'll see. Two of them say local news. The other one is. Was my effort to have us in the pit before that. After you.
Paris Martineau [02:27:20]:
The show. The pit.
Leo Laporte [02:27:22]:
Yes.
Leo Laporte [02:27:22]:
In.
Paris Martineau [02:27:23]:
Are these supposed to be us in.
Leo Laporte [02:27:24]:
The Discord or in the.
Leo Laporte [02:27:25]:
So the Discord.
Leo Laporte [02:27:27]:
Why don't I see it?
Paris Martineau [02:27:28]:
It's higher up there. I'm about 47 years old in this and have the forearms of a man.
Leo Laporte [02:27:38]:
I look like the guy from Project Runway. Oh, no, that's Jeff. Wait a minute. Let me turn up the sound. Is there sound on it.
Leo Laporte [02:27:46]:
Yeah, there's sound.
Leo Laporte [02:27:47]:
The latest reports are looking stable. Yeah.
Paris Martineau [02:27:49]:
Let's proceed with the next topic.
Leo Laporte [02:27:53]:
And that's how we wrapped up today.
Paris Martineau [02:27:55]:
I like that. None of the lips are remotely going on with.
Leo Laporte [02:27:59]:
All right, so here's one where we're. What? Gilligan's Island, I was asking.
Leo Laporte [02:28:02]:
Wants to do a dance routine on a cruise ship.
Paris Martineau [02:28:05]:
It's getting filtered out, Leo. The audio is getting filtered out a little bit.
Leo Laporte [02:28:08]:
Oh, I gotta click. Hold on. No, original sound is on.
Paris Martineau [02:28:13]:
It was working.
Paris Martineau [02:28:14]:
It just wasn't very. Yeah, it's just not very loud.
Paris Martineau [02:28:19]:
Oh, the waves are crashing the sun is bright we're sailing away.
Leo Laporte [02:28:25]:
Is this supposed to look like us?
Leo Laporte [02:28:26]:
Yeah, it's ridiculous. No. And I tried to go to the fourth and I ran out of my videos for the day.
Leo Laporte [02:28:33]:
Thank God.
Leo Laporte [02:28:34]:
The one that really didn't work. If you go up further, I thought we would do the pit version. But then, of course, it put masks on us so you can't see us.
Paris Martineau [02:28:42]:
Let's proceed with caution. Remember our previous findings on us.
Craig Silverman [02:28:46]:
And that's how we wrapped up today's procedure, folks.
Paris Martineau [02:28:49]:
Well, at least pull a microphone from his chest.
Leo Laporte [02:28:56]:
That's very weird. So you cut it with a steak knife and then you pull out a microphone.
Paris Martineau [02:29:06]:
I mean, and I like that the background is your house.
Leo Laporte [02:29:12]:
Oh, it is. Hey. Well, that's the good news. It is in his living room. Yeah, that's right. Studio.
Leo Laporte [02:29:17]:
Yeah.
Leo Laporte [02:29:18]:
So at least it got that right. And actually, this cruise ship looks just like my. My. My bedroom room. Okay, anything else before I send it to the picks of the week?
Paris Martineau [02:29:33]:
I'm going to go grab a cat really quick. Some people in the chat are asking for that. Then we should go to.
Leo Laporte [02:29:38]:
We saw a little gizmo sighting earlier, but to compliment the Rosie sighting, you're watching Intelligent machines. Hey, a very special thanks to our club Twit members who make this show possible. This week has been a little light on the adverts, but, you know, it's always a relief to know that we've got your support. In fact, 25% of our costs are supported by club members, which is very cool since it's only. It's less than 2% of the total audience joins the club. I'd like to make that a bigger number. Big enough. We wouldn't need ads at all.
Leo Laporte [02:30:09]:
Wouldn't you like that? No ads. Well, that's one of the benefits of being a club member. No ads. You wouldn't even hear me begging access to the discord where all Those weird AI things are. And if you act today, we'll send you a hair from Gizmo. A gizmo hair.
Paris Martineau [02:30:25]:
A gizmo hair. We've got plenty of them. We got lots coming off in bushes.
Leo Laporte [02:30:29]:
We got lots. I don't know, it's just one added benefit of the. Oh, Gizmos doesn't like that idea at all. At all.
Paris Martineau [02:30:37]:
I want to keep all my hair.
Leo Laporte [02:30:38]:
I want my hair. Don't dig my hair. Twit. TV club Twit. We've got a lot of stuff coming up. In fact, in about an hour and a half, Micah is going to do his crafting corner. A very chill chance to hang with Micah and Micah Sargent and many of our club members as they do crafting of various kinds. He likes to do Lego.
Leo Laporte [02:30:58]:
There's crocheting, sewing, cooking, coding. Lots of fun. Actually, we have been doing a lot of coding in our AI user group. That's the first Friday of every month, so it'll be a couple of weeks. A chance to get together with other club members who are really into AI and we do stuff, we make stuff. We also have, oh, some good stuff. Coming up Friday, Chris Marquard, our photo guy for our photography show. You should come on that, Paris.
Leo Laporte [02:31:25]:
You could ask him about what camera you should get.
Paris Martineau [02:31:28]:
I should, I really should.
Leo Laporte [02:31:29]:
He's a great teacher. He would be. He would be a good guy to listen to. We also have Stacy's Book Club coming up in at the end of the month. A very interesting book. What else? Oh, Johnny Jett's gonna make his return. He was a regular, weekly regular on the radio show talking about. About travel and tech and he will do his thing.
Leo Laporte [02:31:51]:
I've realized I haven't talked to Johnny in so long. So we're gonna get Johnny and talk travel. So please join the club. We'd love to have you Twitter, TV Club, Twit. All of that's possible, made possible by your membership. Paris Motno's pick of the week. You got a few of them here.
Paris Martineau [02:32:15]:
I got a couple picks the week I did click, I did creepy link. I guess I'll plug that. I. We tested five more protein powders after our last thing. It's actually kind of good news. We got. I got hundreds and hundreds of emails from seemingly every reader in America asking me why I didn't test their favorite protein powder. So I tested the five most requested ones.
Paris Martineau [02:32:39]:
All of them had the low levels of lead, which were low enough for kind of safe daily or near daily consumption, which I think is Particularly interesting because a common criticism or response I heard from frankly a lot of companies in the supplement space as well as people who take protein shakes that we found at high amounts of lead. They're all like, well, it's really hard to get lead out of these things. There's always going to be some lead. It's really unreal realistic to use Consumer Reports as really stringent lead standards. And this proves it's not. We found, we tested five products, all chocolate which could result in maybe more lead in it. And all of them had low safe kind, safer at least amounts of it, which show it's totally feasible for this industry to produce safe products. Just not every company is doing it.
Paris Martineau [02:33:32]:
I asked a lot of the or I asked all the companies the products we tested. A lot of them didn't want to talk, which I thought was weird because I think this is a pretty easy PR victory. But the two that did respond said how they did this is it's pretty simple. They just test all their ingredients and test every lot of their finished product. And if there's a lot of lead in it or something bad and they don't send customers the product, which seems like pretty standard operating procedure to me.
Leo Laporte [02:33:59]:
How hard is it to test for lead? I mean you guys are doing this at Consumer Reports labs. Is it a difficult test?
Paris Martineau [02:34:05]:
I mean it's, I guess you have to actually send it out to its own kind of lab. I think what I heard from some in the industry is that because there isn't a stringent requirement, there's no like strict federal limits basically for the amount of heavy metals, like something like lead permitted in protein powders. Basically what the rules are is the FDA is like supplement makers. You're not supposed to have anything harmful in your products, but you get to decide how much lead counts.
Leo Laporte [02:34:35]:
This is the problem with supplements. Yeah. So it's like hands off on this.
Paris Martineau [02:34:39]:
For a lot of companies. What they do, if they're going to do anything they might reach out to like there's this kind of third party testing service called NSF International that they have less stringent standards. They tend to test. They say it's good to have like up to 10 micrograms per serving of lead. And we think that's you're fine. That that's cons. That's what they, they will test to like if you'll only flag as having too much lead if you're over 10 micrograms preserving. And for context, the FDA says like one of the kind of the point that they Reference as, like, women of childbearing age, but then also all American adults for, like, lead exposure where it could be a risk is like, around like, eight something.
Paris Martineau [02:35:29]:
So it's like, I don't know. I don't think that we should. I don't think that a large amount of lead is good for anybody. And it was heartening to find out that there are, like, better options out there.
Leo Laporte [02:35:41]:
Good Consumer Reports. And this one's open to the public.
Paris Martineau [02:35:44]:
This article, all of my reporting is open to the public.
Leo Laporte [02:35:47]:
Nice.
Leo Laporte [02:35:48]:
Yay.
Paris Martineau [02:35:49]:
Yeah. So get on up there. My other brief pick I'll just shout out. I mean, I'm cannibalizing a pick for next week, but such one won't do is Traitors Season 4, everybody. The Traitors is a perfect reality show. You should be watching it. Alan Cumming dresses up like the campiest demonic lord of a manner you've ever heard. And it's lovely.
Paris Martineau [02:36:15]:
You get to see housewives fight Survivor people fight drag race people, all while they pretend like they're about to be murdered in a creepy Scottish castle. It's fantastic.
Leo Laporte [02:36:27]:
Do they shoot it in Scotland or do they shoot it.
Paris Martineau [02:36:29]:
They shoot it in Scotland.
Leo Laporte [02:36:30]:
Oh, that's nice.
Paris Martineau [02:36:32]:
In a giant mansion.
Leo Laporte [02:36:33]:
Yeah, nice. Yeah.
Paris Martineau [02:36:34]:
I think. Is he Scottish?
Leo Laporte [02:36:36]:
Oh, yeah, I think so. That couldn't be. Couldn't make up an accent like that. That's a real accent, don't you think? He hosted the Webbies many years ago. It's quite funny. Quite good.
Paris Martineau [02:36:49]:
Yeah. He's just a great guy. He's got such crazy outfits on the show, too. I mean, they really dress him up in the most outlandish things.
Leo Laporte [02:36:57]:
I think I saw him in Cabaret actually, on Broadway. Actually. It was the old Studio 54 that they turned into a theater. A likely place for him to be very talented. I had two picks, and they're both for you, Paris. Yes. One is Wikiflix, which is a wiki with 35,000 public domain films in it. You can read the stories of these films, and then you can also watch them.
Leo Laporte [02:37:27]:
Like, did you see Nosferatu when it came out a couple of years ago? Well, wouldn't you like to see the original Nosferatu?
Paris Martineau [02:37:34]:
That one of Nic Cage's favorite films?
Leo Laporte [02:37:37]:
Yeah. And they have a button called the Bechdel Test that gives it, I guess, a rating in the Bechdel test. I doubt Nosferatu.
Paris Martineau [02:37:45]:
What is Nosferatu rating?
Leo Laporte [02:37:47]:
What was the Bechdel results? I don't know. That's the test where you. You do women talk about Men. In this movie, if or do.
Paris Martineau [02:37:58]:
Is there two women having a conversation about not men.
Leo Laporte [02:38:01]:
Not having men. Men, yeah. Oh, I see. They give it a good badge. If it passes the Bechdel test. They give it a badge. Yes. In old Arizona.
Leo Laporte [02:38:10]:
Did not, I'm sorry to say.
Paris Martineau [02:38:11]:
Oh, they've got a trip to the moon on here.
Leo Laporte [02:38:14]:
Oh, the classic Jacques. Yeah.
Paris Martineau [02:38:17]:
Okay, it says that a trip to the moon passes the Bechdel test. But is anybody talking in this?
Leo Laporte [02:38:22]:
It's a silent film. Of course it does.
Leo Laporte [02:38:25]:
Omission is a. Is a virtue.
Leo Laporte [02:38:28]:
Yeah. There are no women talking in it. How about It's a Wonderful Life? I bet that doesn't do women in that movie have. Oh, it's green in the Bechdel test. That means it's good.
Paris Martineau [02:38:37]:
Oh, okay. So when it's read in the Bechdel test, that means it doesn't pass. All right, so it's a wonderful ice. Does pass. Cabinet of Dr. Caligari does not pass.
Leo Laporte [02:38:47]:
How about Mabel's Strange Predicament? I wonder how that has no Bechdel reading. It's only 17 minutes. That's. So this is a chance to see a bunch of public domain films. Some new ones have come out, by the way, of course. Every January 1st and. And this is really good. 35,000 movies in Wiki flicks.
Leo Laporte [02:39:05]:
No, I'm sorry, I got that wrong. 4,409 movies, I guess. I don't know where I got that number. I made that number up.
Leo Laporte [02:39:12]:
Just up. You're too damned. LLM.
Leo Laporte [02:39:14]:
That was an LLM. Well, 4005. 400 movies still is a good number. A good. The other one, I think is fantastic. WantMyMTV Vercel app. It is MTV Rewind. You can.
Leo Laporte [02:39:32]:
Now, I know you were not born in the MTV era, Paris Martino, but this is mtv. Did you. Did it have music on it when you watched it?
Paris Martineau [02:39:41]:
Yeah. No, all the time in my house growing up, we'd play mtv just kind of in the background.
Leo Laporte [02:39:46]:
It's the moon man floating through this Had. Now I double click. Any channel is set as your channel. I don't know if it's what's going on. It did when I. Oh, yeah, here we go. For instance, the first day of mtv. You can watch the first day.
Paris Martineau [02:40:04]:
That's fun.
Leo Laporte [02:40:04]:
Wouldn't that be cool? Live aid from 1985, after the passing of Bobby Weir. You can watch the Grateful for the Music. Bob Weir on meeting Jerry Garcia. So this is just a collection of stuff, but if you want to really kind of relive. MTV. They have MTV in the 70s, 80s, 90s.
Paris Martineau [02:40:25]:
Who put this together?
Leo Laporte [02:40:26]:
I don't know, but I love it. And I think all these are YouTube videos, so it's legit. And you can. You can relive your childhood.
Paris Martineau [02:40:38]:
That's fantastic.
Leo Laporte [02:40:39]:
Rewind.
Paris Martineau [02:40:41]:
I always thought that MTV should have a streaming channel where they're playing Today's date, like 20 years ago. Yeah, exactly. What was going on, like 20, 25 years ago.
Leo Laporte [02:40:50]:
They should do that on this. Yeah, really clever idea. And. And all of this stuff is on YouTube now. So Flexosaurus Rex did it.
Leo Laporte [02:41:04]:
Well, that answers that.
Paris Martineau [02:41:06]:
Thanks, Flexosaurus.
Leo Laporte [02:41:08]:
Thanks, Flexosaurus Rex, whoever you are. Mtv, rewind. Okay, now, Jeff Jarvis's pick.
Leo Laporte [02:41:20]:
Well, I'll go with the one that pisses. Paris and I both voted for this one. Let's go with this one, which is the bombshell report in the Guardian. Oh, pass doubt on the discovery of microplastics through the human body. We want a little good news.
Leo Laporte [02:41:34]:
Maybe this.
Paris Martineau [02:41:35]:
What do you think of this, Jeff?
Leo Laporte [02:41:37]:
So the premise of this report, which is a response to the reports we all saw that said they found microplastics in everybody in their brain and stuff, is that fat in your body is very hard to distinguish from the kind of microplastic polymers, especially the microscopic particles that you would find. And they believe that is probably false positives that, especially in the brain. The Guardian has identified seven studies that have been challenged by researchers publishing criticism. A recent analysis listed 18 studies that said it had not considered that some human tissue can produce measurements easily, confused with the signal given by common plastics.
Paris Martineau [02:42:23]:
The issue is that this is. It's useful to be having a discussion about our limits of analysis when it comes to any. Anything, basically, but especially when we're talking about very small particles like this and a relatively new field. However, I think the framing of this is overly broad and overly dismissive of a very large field of research, research that has a lot going for it. The. Even the quote in the headline that this is a bombshell report, and much of the most inflammatory quotes throughout the piece casting doubt in these studies are from a former Dow Chemicals engineer who, like, for the companies that put a lot of the microplastics and phthalates in things, I think it's also. I mean, if you look at the. It's hard to prove the negative.
Leo Laporte [02:43:21]:
I was gonna say, you can't say there aren't microplastics.
Paris Martineau [02:43:24]:
Yeah. But, like, there are, like, some of these critiques that they're talking about. Like, don't Entirely add up. Like one of the I believe studies that they say has people are asking questions about their methodology involves the kind of the researcher that checks checked like cadavers brains that came with the whole spoonful of microplastics, things that stuck in a lot of people's heads. Like he literally found a spoonful of actual plastics in the brain. Like the spoonful analogy was the plastics he found in the cadavers filled a spoonful in one case. So I don't think that you can say that that's like a limit. The analysis of that was incorrect.
Leo Laporte [02:44:02]:
Let me quote Dr. Dushan Materic at the Heimholtz center for Environmental Research in Germany, who says the brain microplastic paper is a joke. Fat is known to make false positives for polyethylene. The brain is approximately 60% fat. Rising obesity levels could be an alternative explanation for the trend.
Paris Martineau [02:44:26]:
I mean, I think that tells you all you need to. There's like a very specific category of like scientific critic who's like, well, maybe the answer could just be rising obesity levels for everything. Which I think is just a gross oversimplification of a number.
Leo Laporte [02:44:40]:
There's no question there's a lot of plastics and polluting the world.
Leo Laporte [02:44:44]:
I think it's. It's a commentary on media, both ends of the story.
Leo Laporte [02:44:47]:
Yeah, it is.
Leo Laporte [02:44:49]:
When they were supposedly found, they took scientific papers that are going to be challenged as they should be challenged, because that's how science works. And they overdid it. And then here comes the challenge and they're kind of overdoing that. And that's what media always does because we don't understand how to communicate science. And I should put in a plug here that I am affiliated with the Alan Alda center for Communicating Science for just these reasons.
Leo Laporte [02:45:08]:
God bless Alan Alda.
Paris Martineau [02:45:10]:
So science is super slow and media wants to be fast. So, like, those two things are just not compatible.
Leo Laporte [02:45:17]:
Yep.
Paris Martineau [02:45:17]:
Here's some critiques of this from one of my colleagues who we're discussing this yesterday, who has been kind of covering this over the last decade, which is that I found some of the critiques quite persuasive. She writes, others less so. And agree this is kind of dramatic frame framing for what is essentially, no one has agreed upon methods for measuring microplastics. So some headlines are probably way overblown. Like, I think that that's basically the takeaway here and that's fairly uncontroversial. Like, I don't think that anybody would think that's like a bombshell study or that these other ones are a joke. And specifically that researcher you just quoted who called the brain study a joke is literally selling his own, quote unquote, better method for detecting nanoplaques neck sticks on his own LinkedIn. Like that's his whole thing.
Paris Martineau [02:46:03]:
So I feel like that should have been included in there as well. If you're going to say that this guy is some objective source on this, like, yeah. And he's also trying to market what he thinks is a better version of this and sell it to you. I think that, like, I don't know, I don't entirely know why people are so focused on microplastics given that they are really difficult to properly detect and how little we know about their health harm know, like, I mean, yeah, I think we're all freaked out about it because plastic just very clearly seems bad to be there. But we know that there are chemicals in plastics like phthalates and bisophenols that like leach out into us and into our food. And we have a lot of research showing the ill effects of that. And I think that that is, I don't know, a way better thing to focus on here than being like, no, the microplastics don't really exist. It's fine, put, put plastic and everything is just a.
Paris Martineau [02:46:58]:
I don't know, I'm sorry to take away our.
Leo Laporte [02:47:00]:
Make a strong case that plastics, which are made from hydrocarbons from fossil fuels are not a good thing to make anyway because they don't biodegrade and they're made with problematic chemicals.
Leo Laporte [02:47:12]:
So they're everywhere in our lives.
Leo Laporte [02:47:13]:
Use less plastics is not a bad thing to aim for, whether it's in your going into your brain or not.
Leo Laporte [02:47:19]:
Amen.
Leo Laporte [02:47:20]:
I mean, I try.
Leo Laporte [02:47:22]:
It's so hard.
Leo Laporte [02:47:23]:
Just like you, Paris. I don't buy water bottles. I use a refillable bottle when I go to the grocery store. We use cloth bags instead of the plastic bags the grocery store offers. Not just for carrying home the groceries, but putting the vegetables in too, just wherever possible. Eliminate plastics. Nobody would disagree, I don't think, except the oil industry. That's a good thing to do.
Leo Laporte [02:47:49]:
We don't need all these plastic plastics. They're a bad thing. I don't use a plastic toothbrush, you know, I don't use plastic razor blades whenever possible. I try to try to avoid the use of plastics because where do you.
Paris Martineau [02:48:01]:
Get a non plastic toothbrush?
Leo Laporte [02:48:05]:
You can get on airlines, they have, they have wooden ones.
Leo Laporte [02:48:08]:
Yeah, you can get bamboo toothbrushes.
Paris Martineau [02:48:09]:
Actually, I use Are the bristles non plastic?
Leo Laporte [02:48:12]:
I use a electric toothbrush because I think that's better for my teeth. Made by a company called Suri that they recycle the toothbrush. The toothbrush itself is corn. The cornstarch, the head of the thing is made out of cornstarch, which you send back to them and they recycle and then the body is repairable, which is really fantastic. You send it back to them and they can take it apart and recycle it or repair it and send it back to you. So I actually really like this thing. But yeah, that's how I try to avoid plastic toothbrushes. Same thing with razor blades.
Leo Laporte [02:48:50]:
I just use an electric, I switched an electric razor so I wouldn't use plastic razor blades. Now I'm using electric.
Leo Laporte [02:48:56]:
My ass and I are fading.
Leo Laporte [02:48:58]:
Okay. Fading ass. Thank you. Jeff Jarvis, professor of journalistic Are you. Oh, I forgot you had some procedure. Are you, are you doing all right?
Leo Laporte [02:49:10]:
No, that's next week. No, I, I, this is, I'm, I fell on Saturday. That's why.
Leo Laporte [02:49:13]:
Oh, I'm sorry. Feel better, sir?
Leo Laporte [02:49:15]:
You took a fall? I did take a fall. Like an idiot. I was pulling the sheets and blankets off the bed to do the wash and standing on a hardwood floor and bare feet and the, I had put the blanket in really securely the week before. Yeah. So my feet went out and I.
Leo Laporte [02:49:38]:
Yikes.
Leo Laporte [02:49:38]:
So no, I have the next, next Wednesday morning. I have the first of my eyes.
Leo Laporte [02:49:45]:
You're gonna have the eye patch. That's right.
Paris Martineau [02:49:47]:
We've got to get eye patches, Leo.
Leo Laporte [02:49:49]:
I have an eye. I will wear it. Yeah. I also have a monocle. So I can have an eye patch and a monocle. Thank you, Jeff. Jeff Jarvis. Don't forget to get his books, the Gutenberg, Parenthesis magazine and everything else.
Leo Laporte [02:50:04]:
Jeff jarvis.com Paris Martineau. Thank you so much. Investigative journalist. She's working on a string, the red string. Consumer Reports. Thank you all for joining us. We do this week in intelligent machines. I almost said this week in Google.
Leo Laporte [02:50:21]:
We do intelligent machines every Wednesday.
Leo Laporte [02:50:24]:
Ms. Now. Darn it. Ms. Now.
Leo Laporte [02:50:26]:
Ms. Now Now. Has anybody, anybody in MSNBC said msnbc. I have not heard them. Super impressed that they.
Paris Martineau [02:50:34]:
That's really impressive given the amount of errors that CBS News has already done.
Leo Laporte [02:50:39]:
Oh, my.
Leo Laporte [02:50:40]:
Don't get me started.
Leo Laporte [02:50:41]:
Don't get us started.
Paris Martineau [02:50:41]:
Yeah, sorry, we can't, we can't bring this up at this point.
Leo Laporte [02:50:44]:
Bring up Barry Weiss. Don't, please. What's his name? Tony Dukipal.
Leo Laporte [02:50:51]:
Like you have to Cope with him.
Leo Laporte [02:50:52]:
Cope with him.
Leo Laporte [02:50:53]:
If you go, he's going to be.
Paris Martineau [02:50:54]:
Doing a sponsored whiskey hour very soon.
Leo Laporte [02:50:57]:
If you go to Katie Couric media, you'll see me writing about it. She invited me to write about it.
Leo Laporte [02:51:01]:
Nice. I can't wait to read it.
Paris Martineau [02:51:03]:
Exciting.
Leo Laporte [02:51:03]:
It's sad. I mean that was the Tiffany network, but that's another story for another day.
Leo Laporte [02:51:07]:
It's now the Capital Monte network.
Leo Laporte [02:51:09]:
Yeah, we do the show every Wednesday, 2pm Actually, I'm trying to do. By the way, I don't want to do PM anymore because that's Latin. I don't want to do o' clock because that's medieval. I'm going to say we. It's a 1400. We do it at 1400 Pacific. 1700 etc Eastern Time. That is 2200 UTC.
Leo Laporte [02:51:36]:
Yeah, see, I'm going to do 24 hour clock from now on. All my clocks are 2400.
Paris Martineau [02:51:40]:
O' clock is crazy.
Leo Laporte [02:51:42]:
No, it's 2200.
Paris Martineau [02:51:44]:
2200.
Leo Laporte [02:51:45]:
2200. I didn't say O'. Clock. I said 2200. There's no O'. Clock. Somebody reminded me that's of the clock. We were talking about this, right?
Paris Martineau [02:51:55]:
And we don't want. We don't want.
Leo Laporte [02:51:57]:
It's not post meridian. What the hell does that mean? No 1400 Pacific tradition.
Leo Laporte [02:52:04]:
Tradition is okay.
Leo Laporte [02:52:07]:
I am a man of the new AI generated century. That's all I want to say. We will see you. You can watch us live in the discord if you're a club member. Also on YouTube, TikTok. No, not TikTok. YouTube, Twitch X, Facebook, LinkedIn and Kik. TikTok proved too difficult.
Leo Laporte [02:52:26]:
They have lots of barriers to streaming after the fact. On demand versions of the show at the website, Twitter, TV IM. And you can also get it on YouTube if you want to share it with friends and family. Best thing to do though, subscribe in your favorite podcast client. That way you'll get it automatically the minute it's available. Ladies and gentlemen, I give you tomorrow, next week. This is going to be the show. But.
Leo Laporte [02:52:55]:
But I got to get my eye patch on the proper side. Which eye will it be on, Jeff?
Leo Laporte [02:53:00]:
Left.
Leo Laporte [02:53:00]:
Left eye. Okay, we all need to f. Left eye.
Leo Laporte [02:53:03]:
Two weeks later, the right.
Paris Martineau [02:53:04]:
We do.
Leo Laporte [02:53:04]:
Oh God, it's so confusing. What was it?
Leo Laporte [02:53:07]:
I'm gonna have. It might'll be ugly. This white plastic thing with.
Leo Laporte [02:53:10]:
Oh no. Yeah, cup.
Leo Laporte [02:53:11]:
Disgusting.
Leo Laporte [02:53:11]:
Yeah, you should get one of those dog things like the Elizabethan collars. Keep you from licking your wounds. Thank you for joining us we'll see you next time on Intelligent Machines. Do not lick your wounds, kids. Hey, everybody. Leo Laporte here. It's your last chance. We're in the final stretch of our 2026 audience survey.
Leo Laporte [02:53:33]:
So important for us both to know you better because we don't gather information about you except this one time a year. And it helps us with advertising and we collect no personal information. We just want to know about you in general. So if you would help us out. If you haven't filled out our survey yet, this is the time it closes. January 31st. TWIT TV survey 26. It helps us improve our shows and it helps us make money.
Leo Laporte [02:54:01]:
And we need to make do both. So survey is @Twitt TV survey 26 should only take a few minutes. Thanks in advance.
Paris Martineau [02:54:10]:
I'm not a human being, not into the animal scene. I'm an intelligent machine.
