← Intelligent Machines (Audio)
Intelligent Machines (Audio)
IM 812: A Choir of Sentient Cabbages - Gary Rivlin, AGI Test, Section 230
02:58:33
Gary Rivlin, AGI Test, Section 230
Loading summary
Transcript863 lines
- [00:00]
Leo Laporte
It's time for Intelligent Machines. Jeff Jarvis is here. Paris is traveling, so Mike Elgin joins us. He has lots to say. Our guest this week is Gary Rivlin. He's just written a new book, really interesting history of AI. It's called AI Valley. Gary Rivlin. And our interview next on IM podcasts you love from people you trust. This is TWIT. This is Intelligent Machines, episode 812, recorded Wednesday, March 26, 2025. A choir of sentient cabbages. It's time for Intelligent Machines, the show. We look at what's happening in the world of, well, intelligent machines, AI, robots, Iot. It's all around us, literally. Paris is in Paris, I'm sorry to say. Paris Martineau from the information is literally in Paris this week. But that's okay. Mike Elgin is here and he, oddly enough, is in the United States. So that's.
- [01:05]
Mike Elgin
Who'd have thought?
- [01:07]
Leo Laporte
Really unheard of. Mike, of course, is a gastro nomad, and his gastronomad adventures are fantastic. @gastronomad.net, his newsletter, all about aimachinesociety. AI great. You're perfect for this show. So glad to have you on.
- [01:20]
Mike Elgin
Thank you. Thank you. Glad to do it. Of course.
- [01:22]
Leo Laporte
Jeff Jarvis is here, professor emeritus of journalistic innovation at the Craig Newmark Graduate School of Journalism at the City University of New York.
- [01:32]
Mike Elgin
Newmark.
- [01:33]
Leo Laporte
Emeritus. You're going to be emeritus. Stamp for him. Just leave that up the whole show, Benito. He is now at Montclair State University and SUNY Stony Brook. Nice to have you. The author of the Gutenberg parenthesis is known as the Web we Weave. And do you know Gary Rivlin? Do you want to introduce Gary?
- [01:52]
Jeff Jarvis
You should go ahead and introduce Gary. But I want to mention. I want to mention quickly that Gary is also friends with our friend Craig Newmar.
- [01:58]
Leo Laporte
Ah, it's in the book.
- [01:59]
Jeff Jarvis
Yes.
- [02:00]
Leo Laporte
All New Yorkers know one another, I believe so. That's why. That would explain it. Hello, Gary. Gary is a longtime Silicon Valley writer for Wired and the industry standard, if you want to bring back memories of the past. And he is author of many books, including why Is Everybody Trying to Kill Bill Gates? Which.
- [02:20]
Gary Rivlin
Okay. It's called the Plot to Get Bill Gates.
- [02:22]
Leo Laporte
Okay, all right. Something like that. The plot. The plot, yes. The plot to get Bill Gates. And his newest, which I think is fantastic, is AI Valley, came out yesterday. He'll be in San Francisco on book tour on Thursday. Menlo Park, California. April 2nd. April 3rd, back in San Francisco. The 15th in beautiful downtown Albany. New York. So. And March 25th is. Was that yesterday? That was yesterday. So we've missed that one at Barnes and Noble in Manhattan. Great to see you, Gary, and thank you for joining us. I appreciate it.
- [02:59]
Gary Rivlin
Oh, my pleasure. Looking forward to it.
- [03:04]
Leo Laporte
You talk in the book about how actually this is in the epilogue. Randall Strauss convinced you not to write the book you were writing, but write this book instead. And I had to know what were you writing? So you abandoned.
- [03:19]
Gary Rivlin
I read a proposal. I hadn't gotten very far with it, but I certainly went about ahead of the times. I want to do a book about Marc Andreessen understanding the valley for better and for worse through Andreessen's story. And this would have been 2022 by that point, you know, he had shifted towards the right. He hadn't sprinted all the way to the right at that point. So, you know, the problem was that he was so hostile to the mainstream media, I guess what he would call the legacy media, that I knew I wouldn't get him to participate. I had interviewed him before. We had a good relationship, but that was back in the 2000s. And you know, the idea of like fighting for every interview, trying to do the right around, tell his story without him, his involvement, and then suddenly AI. So this was the end of 2022, right before ChatGPT came out. I just randomly got an email from Reid Hoffman, a batch email, announcing that he was going to co found his first startup since LinkedIn. And I just was really intrigued. So just kind of had lucky timing.
- [04:35]
Jeff Jarvis
Can I just add in there? I'm so glad that you've made Reid Hoffman central to this because I think he is one of the good guys in the off malign Silicon Valley. He is the connector of connector and full disclosure is he helped fund my engagement journalism program. But he's a public intellectual by his own description. He's a good guy and a nice guy and I think it's just great to see him at the center of how he operates.
- [04:59]
Gary Rivlin
Yeah, a billionaire you can root for.
- [05:01]
Jeff Jarvis
Yeah, yeah, yeah. Amen.
- [05:03]
Leo Laporte
Well, it depends on what side of the aisle you're on, but yes, I think you're in good company here. Yeah, he's kind of the backbone of the book. And it's Reid bouncing off various people and various companies that becomes the story. It's a history really of AI starting. I was surprised back in 1956 with John McCarthy and the first AI conference and coming right up to the present, although it's got to be challenging. You wrote the book at the I mean, the epilogue, I think, is dated 2024. It's the end of. The end of last year. Publishing, being the speedy enterprise that it is, has finally got the book out yesterday. It doesn't seem out of date, which is good, but, you know, it's going to be out of date in six months or a year. I mean, this is a very fast moving topic.
- [05:52]
Gary Rivlin
Yeah, I mean, I clearly, I was well aware of that. For instance, Deep Seek had not come out by the time. Or the latest iteration had not come out. But, you know, I don't know. I really was trying to tell a different story. I wasn't. I don't think you're gonna want to pick up my book to learn the details of Claude 3.6.
- [06:13]
Leo Laporte
No, that's what we're here for. That's right.
- [06:15]
Mike Elgin
Yeah.
- [06:16]
Gary Rivlin
And so, you know, I was just trying to, like, tell the story, how we got here, where we're at, what it means through these emblematic characters. Reid Hoffman, as we've mentioned, Mustafa Suleiman, kind of one of the, you know, big names in AI.
- [06:30]
Leo Laporte
I learned his nickname in college was Moose, which I thought was very useful.
- [06:35]
Gary Rivlin
I don't think he likes it anymore. When he asked.
- [06:37]
Leo Laporte
Oh, I doubt he does.
- [06:39]
Gary Rivlin
Gave me a funny look and like, yeah, we're not using that anymore. Didn't have to throw that in there. And, you know, the usual suspect. In fact, weirdly, the first two words of the book, and I wrote the opening a year ago, 14 months ago, are Elon Musk. I started off with this dinner in 2015 between Musk and Reid Hoffman. And Musk, by that point had put money in DeepMind. And he saw, I mean, I will give him credit, he saw what was starting to happen in AI. He had this weird perspective where he's an investor and he was also its loudest critic. It's going to destroy humanity. The robots are going to take over. And meanwhile, I own a piece of these very aggressive companies trying to advance this stuff. But so it starts off with a dinner where Musk basically convinced Hoffman, who at Stanford more or less studied artificial intelligence, like, hey, it's time. This is starting to happen. And so that kind of launches the rest of the story with Hoffman getting involved as he would as a funder, and then ultimately starting his own company.
- [07:47]
Leo Laporte
One of the things that interests me, and so it really is a history of AI, but one of the things that really interested me is how many of these people were not really coders. Many of them weren't even computer scientists. Often they were philosophy majors and the like. Obviously they could hire coders. But it seems like AI has been driven not so much by people who deeply understand computing, but by people who have kind of a dream. You even mentioned the impact of science fiction. I think Reid Hoffman talked about that in all of this.
- [08:21]
Gary Rivlin
That was one of the big surprises to me. I guess I assume like you assume there was computer scientists. Right. But you know, it's, well, mathematicians, right? These things are just enormous mathematical models, but beyond that, a wide range. I mean, philosophers you mentioned, but physicists. A lot of physicists.
- [08:42]
Leo Laporte
A lot of physicists. In fact, the guy we talked to last week is a theoretical physicist.
- [08:48]
Gary Rivlin
Yeah, yeah, Anthony.
- [08:50]
Leo Laporte
Anthony. I forgot his name. Jeff. Jeff won't let me forget. Aguirre, you actually mentioned early on Future of Life Institute is in the book as well.
- [08:59]
Gary Rivlin
Yeah, yeah, Cognitive psychologist. You know, it surprised me, but then of course it makes sense. I mean, artificial intelligence is programming, it's computers, but it's far more than that. It's like understanding the brain. I mean, the big breakthrough, as I'm sure most of your listeners know, was going from hard coding rules to trying to create these neural networks that mimic the brain. So those who are studying the brain, it made sense. But again, I was like you, I just figured it was computer scientists. What I find most interesting about the history was the absurd optimism of people like John McCarthy, Marvin Minsky, the early pioneers. AI was just around the corner for 70 years. You'd read what they wrote in the 50s, you'd read what they write in the 60s or 70s, and they were always just a few years away. And of course, they were more like 50 years away from where they thought they were.
- [10:01]
Leo Laporte
Yeah, I mean, one of the things I like about the book is really getting to kind of almost meet these personalities. I didn't know, for instance, I had long thought highly of Marvin Minsky, but I didn't realize he really, you know, in some respects, computer scientists in general steered AI wrong for a long time. At least two or three of the AI winters were because Minsky had insisted, no, you've got to do symbolic rules based AI. You know, trying to mimic the brain is a mistake. And he really influenced the AI community for decades because of that.
- [10:37]
Gary Rivlin
I mean, he mocked people who were pushing the neural networks. I mean, I call it like a 50 year mistake. I mean.
- [10:44]
Leo Laporte
Yeah.
- [10:46]
Gary Rivlin
On the other hand, you know, computers weren't powerful enough, you know, back in the 90s, 2000s, to power these neural networks. I mean, you know, you had to have several things happen beyond the realization that neural networks was the way. There wasn't enough digital data, you know, to train these models. So, you know, it's. It was the wrong turn that lasted around 50 years. But on the other hand, I think if there wasn't that wrong term, there still would have been a long, long delay before we got anything remotely like we would get in the second half.
- [11:18]
Jeff Jarvis
We couldn't have had the Internet without. We couldn't have had AI or generative AI without the Internet.
- [11:24]
Leo Laporte
Right.
- [11:24]
Jeff Jarvis
Without the access to the complete compilation of human speech.
- [11:29]
Leo Laporte
Yeah.
- [11:30]
Gary Rivlin
The digitalization of everything.
- [11:32]
Mike Elgin
Right.
- [11:33]
Leo Laporte
You talk about Frank Rosenblatt, who was kind of widely considered the father of these neural networks and his first machine, the Perceptron. This is in the 60s, right. And I think Rosenblatt ended up being kind of cast aside because of Minsky saying, no, there's nothing there. These neural networks, they aren't going anywhere. It's kind of a sad story, really. Yeah.
- [11:58]
Gary Rivlin
You know, I mean, I think that's kind of a not so uncommon story that the lonely pioneer in the wilderness, who's the true believer, and others are mocking them, they not believing them, you know, but Rosenblatt himself. You know, it's funny. We all came came of age, all of us on this call, you know, around writing about tech, you know, more or less the same time. And, you know, I kind of thought concepts like vaporware were relatively new. This idea that you, you know, sell the concept before you've built or you sell the product before you figured out the product. But that was. That was Rosenblatt. I mean, he just was just piling it up, just making claims that were not true. And so, you know, I think he got himself in trouble by over promising, as many others, Minsky included, would do.
- [12:56]
Leo Laporte
Around AI 1958, New York Times, New Navy device learns by doing the New Yorker. It strikes us as the first serious rival to the human brain ever devised. I'm quoting from the new book, which is a must read. Gary Rivlin's AI Valley. There it is. I told him every time I say that, hold up the book.
- [13:21]
Jeff Jarvis
Out today.
- [13:22]
Leo Laporte
Out today. So at what point did people start to suddenly realize, no, symbolic. By the way, we had Stephen Wolfram on not so long ago, and he said, no, no, he's still in that mindset. The neural networks can't do it all. We need symbolic AI it's going to be some combination of the two. Of course, his Wolfram Alpha is essentially symbolic. At what point did people start to say no, you know, what this neural network has promise?
- [13:48]
Gary Rivlin
Well, I mean, there were more lowly pioneers, famously Jeff Hinton, a few others who are pushing this for, you know.
- [13:56]
Leo Laporte
He'S the father of modern neural networks, isn't he? Jeffrey Hinton?
- [13:59]
Gary Rivlin
Yeah, I would say the same thing. You know, they're the three godfathers, but to me, you know, Hinton is first among the equals there. But, you know, it was like 20 in the 2000 and tens. The first half of 2000 and tens. That's when it really started to pivot. It was the famous imagenet contest. Fei. Fei Wu from Stanford had this contest. And that's when famously, Jeff Hitt and a couple of his graduate students entered the contest with a neural network.
- [14:31]
Leo Laporte
Julia Sutskever, among others. Yes, that's right.
- [14:35]
Gary Rivlin
And they, and they bested all the rule based models and that really got people's attention. The idea that rather than putting down the code, this is a cat showing a million pictures kind of thing. This was like, set it loose, let it read, and it could more accurately identify animals than any of the rule based systems. So that really got people's attention. But, you know, I mean, this is science. It does move slowly. So it was 2013, 2014, 2016. And you know, I feel by then that neural networks became, you know, the, the main focus for most people. Then the famous Transformer paper, the T&GP ChatGPT came out. What was that? 2016, 2017, remember? Yeah, 2017, yeah. And that really showed the potential for these chatbots. You know, I mean, something I, I did not know until I started doing this research. Artificial intelligence has been around for a long, long time. We've been using it. We just didn't know it. You know, in the 2000s, you know, Google was using it to figure out how to price their ads. Google Translate has been around since 2015. That's, that's AI, but generative AI. This idea that you can talk to a chatbot, that you could ask it through a prompt to make a drawing, to make a video, that really changed things. So to me, the Transformer paper coming out was the beginning of the revolution that we're now all seeing. The general public is now seeing.
- [16:17]
Mike Elgin
I have a question about the Transformer paper. Before that, I want a little defense of WolframAlpha, Stephen Wolfram. He says that neural networks can't do it all. And the reason he knows that is because the best chatbots, you ask them certain questions and they're like, we better ask Wolfram Alpha. And then they kick over to. And Wolframalpha figures it out and sends it back. But anyway, the transformer architecture. I've always been fascinated by that pivotal moment in 2022 when OpenAI opened this thing up. Do you feel like Google or some of the other companies felt blindsided? I mean, all these companies have been working on these things and felt like it wasn't ready for the public, it wasn't ready for primetime, it might be a little dangerous, all that kind of stuff. And nobody was really publishing public access to their tools. Then, then the revolution hit when OpenAI had the boldness to actually do that using. Based in large part on Google's work with the, with the G GPT concept. So do you have a sense in the industry of what the other companies were thinking when OpenAI did that?
- [17:24]
Gary Rivlin
Yeah, that was a big part of my reporting. I, I blame it on Tay. I don't know if you remember this.
- [17:28]
Leo Laporte
Oh yeah, that was the Microsoft bot that decided that Jews started 9 11. Yeah, okay, exactly.
- [17:36]
Gary Rivlin
Holocaust denying. So Microsoft puts this thing out, it was trained more or less on social media.
- [17:44]
Leo Laporte
That's the problem. Of course, people gamed it immediately and said, let's see if we can turn this into a Nazi.
- [17:50]
Gary Rivlin
And they pulled it within 24 hours. And Microsoft immediately pulled it. And it was like this specter hanging over all the big tech companies. They're also. So Google had, they called it Mina, it would change names, but they had Mina in 2020. And so Mina was basically chatgpt, but they were scared of it. You know, there are people, Mustafa Suleiman, main character in my book. And so it was.
- [18:21]
Leo Laporte
Good shot. That's good. Don't you hold it up longer though. People have to really read the title there. So, you know, we're training Gary. The book only came out yesterday. He's got a lot of interviews coming up. We're just teaching him, you know, you gotta hold the book up.
- [18:37]
Gary Rivlin
Appreciate that. And so, you know, Suleiman, others inside of Google were so frustrated. We have this amazing thing.
- [18:45]
Leo Laporte
Mina. Yeah, I don't even remember Mina well.
- [18:48]
Gary Rivlin
Because they didn't put it out there. And so. And so, you know, in fact, Suleiman ended up quitting Google and co founding with Reid Hoffman. Inflection, you know, an amazing company, made well that Rest in peace. It still exists. But you know, it had the amazing pie, which is PI, which is my favorite, was my favorite of all the chat bots. But anyway, it's like, you know, Google was scared of it, Microsoft was scared of it. I would talk to people there and just like, you know, we wanted to just advance this stuff. But it had to be a startup. It had to be a company that didn't have $100 billion in revenues on the line. So I don't think it was any coincidence that the first company that had the daring, or whatever you want to call it, to put this out was a startup.
- [19:34]
Jeff Jarvis
Gary.
- [19:35]
Leo Laporte
Although hold up that book again because the subtitle is fairly important. Microsoft, Google and the trillion dollar race to cash in on artificial intelligence. Because the bitter moral of this book seems to be you start out like anyone would. Oh, it's going to be, it's going to be the startups, it's going to be the garages, it's going, going to be the small guys who come up with a brilliant idea. It turned out to be so expensive, it's only the big guys who can play this game really.
- [20:00]
Gary Rivlin
You know, I started this book beginning of 2023 and I was going to tell it all through startups. I'm fascinated by startups. Who isn't right? The venture capital, the daring. Like you have this idea and you know, this dream of being the next Google, whatever. But I discovered through my reporting like it's Microsoft, it's Google, it's meta. That's my big fear here, that the same tech giants that have been dominating for the last 10 or 20 years are going to dominate AI because it's so, so, so expensive. And you know, you have Anthropic out there with Claude, another favorite chatbot of mine, and you know, I think they've raised about $20 billion so far. You know, 4 billion or so from Amazon, maybe more by this point. Google, billions of dollars. So even if Anthropic wins, they're in part owned by the giants. And beyond that, Dario Amadei, the CEO, predicted by 2027 it's going to cost $100 billion to train one of these things. Like, you know, there was 150, 160 billion thrown into AI startups from venture capitalists in 23 and 24 combined. $100 billion. A large venture capital fund is $1 billion. And so I do wonder, in fact, Suleiman, near the end of the book, he concludes that none of these consumer chatbots could make it on their own, that they'll eventually be gobbled up, bought, acquired by a giant.
- [21:38]
Mike Elgin
This is why Apple is so hopeful, I think with they're now kind of mediocre AI play, but they know that eventually they'll be the only company that can afford it and they're on you.
- [21:50]
Gary Rivlin
Know, how many billion iPhones. I mean, you know, they have the reach and that. I mean ditto meta. You know, Zuckerberg, who, you know, meta or back then Facebook and Google were the two companies working on neural networks before anyone else. They had a huge lead and they too, Facebook too was scared to put this out and so they were caught flat footed. They were paying attention to the metaverse, et cetera. That didn't help. They went the open source way. But you know, he may, he made a claim in 2024 that within six months we'll be the most popular chatbot on the planet. And in a way he's right because billions of people use his various platforms and it's available, I would argue like it's available but only a fraction use it as opposed to the 300 or whatever million people who actually go to chat to use their model. But you know, you could be late. I wouldn't count down Amazon, I would not count out Apple just because they have the reach and they have the funds.
- [22:58]
Jeff Jarvis
Gary, I'm curious, I'm curious about a lot of things. Let me ask you this, this, this sense of the need for scale. If you go to the Stochastic Parrots paper, they argued that that going for huge models was a folly. Deep Seq we think might say that that could be the case, that they could be smaller. Is, is it a case? When I worked for a big, when I worked for Time Inc. They thought, well, if it cost the most in the world, it has to be the best because we're spending the most money.
- [23:28]
Mike Elgin
Right.
- [23:28]
Jeff Jarvis
Is there a little bit of that in the American ethos here? Is there a chance that you don't know need to be that huge? Do you don't really need $100 billion to train a model.
- [23:37]
Gary Rivlin
Yeah, you know, I mean that's an open question. I guess I'd play journalists on this to kind of give the arguments. I mean first off, I thought Deep Seek, you know, it's, it's, I thought it was really overstated. You know, inflection was using smaller models. Other companies out there were using smaller models. Like do we really need the, the huge system to answer all of our questions? No.
- [24:04]
Leo Laporte
Why?
- [24:04]
Gary Rivlin
Let's send some simpler questions to these smaller models that use less compute. They're cheaper and all. But I think that if you could do it for 1/10 the price, then you're going to make 10 times larger models. I think that called Jarvan's paradox dates back to the 19th century in coal. But I think it's true today that, you know, I mean these are engineers, they do it because they can do it. They want to make it as big as they can. But it's an open question. I don't know. I'm convinced that it might not be as expensive as people thought. Like as Amade said with Claude Anthropic, I mean, but on the other hand, I'm convinced it's gonna be a lot of money. You know what I mean? Like Deep Seq was cheaper but it still was millions of dollars or venture backed outfit with millions of dollars. And there's the training of these things, there's the fine tuning of these things, but don't forget there's the operating of these things. Cost a lot of money. In fact, one of the startups I started following in 2023, they're really there before more or less everyone else. On AI powered search, this goes back to I think 2018, they started first, they first started working on that. They are holding back features because they're too expensive. Too expensive. They haven't done an update. They have because they know if they do it it's going to bankrupt them on GPU costs, on compute costs. You know, they were doing all these clever marketing things but they stopped doing that because they were getting too many users and it was costing too much money per month to operate them. And so, you know, it's still really, really, really expensive. Perhaps it's not $100 billion expensive.
- [26:00]
Mike Elgin
I think there's a kind of mismatch between what some of the many of the AI leaders want out of both robotics and artificial intelligence and what the public really needs or wants going forward. Because essentially a lot of this compute power is built around making the chatbot be this very human like voice that, that can really understand things in a certain way. And you're basically just kind of making a person. In robotics you see humanoid robots with faces, heads, arms, fingers, all that stuff, which why are they doing that? I know their justification, I don't buy it. But basically I feel like it's not that expensive if you're not trying to create a person, if you're actually trying to make a tool and if they focused on the things that really matter. For example, like right now, the problem with chatbots is that the output is just very boring and generic. The hallucinations are, as a well known problem, they're being gamed by China and all kinds of other nefarious actors deliberately poisoning or polluting the data. And then there are companies that are doing what I think is the work that really needs to be done. Like contextual AI, which has something called a grounded language model. And this is basically designing a chatbot that's optimized for accuracy, not for being Scarlett Johansson. And also other companies are working on customization, so they're building tools that are designed for you as an enterprise to populate it with your own data so it's less likely to hallucinate and can be more relevant to you. Those are the directions I'd love to see them pour resources in. And I don't think those cost $100 billion.
- [27:35]
Gary Rivlin
I mean, it's very interesting what you're saying. I think you're making a point, but I don't know, maybe it's me. That's kind of my favorite thing about the PI. PI, the mathematical equation. My, my is my favorite for the very reasons that you're poo pooing and saying people don't want. I. I love that I could just talk to it. I mean, it's voice, so you could literally talk to it. Right. And, you know, it gets nuance, it gets jokes. It's, it's understands subtlety. It's, it's kind. It's. I mean, I think that's what people like about Claude, that it has a little bit of a personality and it's talking to you. So I don't know. I'm not so sure if that's not what people want.
- [28:15]
Mike Elgin
Have you tried sesame?
- [28:17]
Gary Rivlin
I have not.
- [28:18]
Mike Elgin
Sesame is like pie, but even more so.
- [28:22]
Leo Laporte
That's the one where you talk to the man or the woman and they're very annoying and they.
- [28:25]
Mike Elgin
Yeah, well, pie's annoying, in my opinion.
- [28:28]
Leo Laporte
Gary in the book talks about Eliza and its successors, and sesame is just the most recent, really amazing version of Eliza, if you ask me. It's a chatbot. We're talking to Gary Rivlin. He's the author of a new book called AI Valley. There it is. Just came out.
- [28:46]
Jeff Jarvis
Simon's Move your finger over Yankees.
- [28:49]
Leo Laporte
There we go. Actually, there's a great tidbit in the book. You were talking about Stochastic Paris. Jeff. When Google wanted to buy DeepMind, one of the problems they had, Sebis and Suleiman and Shane Legge, their third founder, didn't really want to sell. And Suleiman especially was very global in his thinking, very conscious. And one of the conditions, besides the big check, was that Google established an Independence Ethics and Safety Board to monitor the development of AGI. That's where Tim, Nick Gebrew and Martha Mitchell were working. I don't think Google ever. Now, Gary, you can confirm or deny this, maybe in your own research, really wanted to do this. They just, it was a condition of buying DeepMind, which they really wanted, and they were right. And as soon as they could get rid of GEBRU and Mitchell and the ethics board, they did. Do any of these big companies worry about safeties, alignment, AGI and the future? Are they even or is that just a little speed bump that, you know, they have to do to make everybody comfortable with what's happening?
- [30:09]
Gary Rivlin
Yeah, so, I mean, you're making some good points first. First off, that, you know, that safety board. Yeah. It meant once before they disbanded it. Now it's very impressive. They had Reid Hoffman, Elon Musk, they had all these big names. Schmidt was on it, you know, the Paige and Brin. But, you know, it's it. They showed how important it was by the fact that they, you know, dumped it.
- [30:34]
Leo Laporte
I've even talked to people who were at Google at the time who say, oh, Tim McGeber was such a pain in the ass. They did. They didn't like those guys.
- [30:42]
Jeff Jarvis
That was her job, though.
- [30:45]
Gary Rivlin
No, but I think Jeff. No, Jeff's making the point like, you know, they wanted to appear like safety was important, but once someone did what they were supposed to do, just like, you know, like, oh, shoot, you're getting our, you're getting our way here. You know, I mean, but I think you overstated a little bit, Leo. It's not that they don't care. To me, they care, but they care a lot more about profits.
- [31:11]
Leo Laporte
Right.
- [31:11]
Gary Rivlin
They care about the trust issue where they care a lot more about beating that other company over here. That's, you know, to drill down further, what I said earlier that my big fear is that big tech is going to dominate. It's that all of that's going to go by the wayside because it becomes all about beating the competition. Or I guess you could throw in, it comes down to we have to be China. You know, I mean, the China card, you throw down the China card and, you know, the Trump administration's taken off, you know, more or less all of the restrictions on AI. You know, there's kind of we have to beat China. Yeah, because we have to beat China. It's kind of the magic words.
- [31:50]
Jeff Jarvis
So, Gary, the safety, in my view, has been polluted as a word because there's two different, very different definitions of it. One is the. I'll use gebru again. The stochastic parrots view of Environment and bias and labor and such. And the other is the doomsters. It's the 10 to the 54th future human beings in the universe. And so when people talk about safety, we need a definition as to which they're going for. I'm not a believer that AGI is around the corner or is even a thing. And thus the Doomster argument kind of falls apart in my view. I'm curious where you come out with the whole test Creal Doomster, that end of safety versus the present tense. Concerns about environmental impact and so on. On safety.
- [32:36]
Gary Rivlin
Yeah, you know, if I had one wish, I wish we would talk about the real fears. Hoffman uses a phrase, line of sight. There's this stuff that's real that we could see now, like, yeah, I guess it's possible 300 years or whatever that laser eyed robots are going to be a threat to us. But Hoffman has a nice construct. There's the Doomers, of course, and he calls the Zoomers, which are the accelerationist Marc Andreessen I mentioned before those who say there should be full speed ahead, no safety, no anything. We just have to bring this to humanity to help solve problems because AI will cure everything. I put myself in the category that Hoffman calls the Bloomers. I think there's amazing potential for AI around science, around medicine, around education, around a whole host of things. But there are also very real concerns. AI, autonomous. Excuse me, AI weapons in warfare, AI and surveillance. You know, I'm scared. Am I allowed to say shit list? I guess I just did that. It's about autonomous AI. I mean people have to understand these things are amazing, but they have limits. And the idea that there wouldn't be a human in the loop for anything that really mattered, that's really scary to me. And so do you guys kind of. Are you more or less on that, that it's this autonomous AI that we're going to believe it's more powerful and better and capable? I mean, you hear it with Musk. We want AI. I really don't want it making employment decisions. It's a copilot, it could help you. I use it in my work as a research assistant. Let it edit, you know, give me suggestions for editing. But I don't let it write it. It's my companion.
- [34:35]
Mike Elgin
But that's, that's how it is now. I mean, I think the, the, the fears which I, and I'm not a doomer or a Zoomer necessarily, more of a perfumer. I make, I make the, I make it all sound good and smell good, but but, but, but the problem is like gentic AI is, is based on objectives. So you basically say, well, here you go, here are your objectives, go do the thing and it' and you can't really trust people to put the guardrails around something like that. And again, I don't think it's gonna be the laser eyed robots, but I do think that there's gonna be some problematic things. I mean we see today's chatbots cheating, lying, hiring people to get past captchas and things like that. And so I think that you can see risks and you can also see it being used by North Korea.
- [35:27]
Leo Laporte
That was again, One of the DeepMind's demands of Google before they were sold to Google, is that they not be used for military applications.
- [35:39]
Gary Rivlin
Or surveillance.
- [35:40]
Leo Laporte
Or surveillance.
- [35:41]
Gary Rivlin
Yeah, but Google dropped both of those.
- [35:43]
Leo Laporte
Exactly.
- [35:43]
Gary Rivlin
Dropped Both those in 2025. Started 2025.
- [35:45]
Jeff Jarvis
Yeah.
- [35:46]
Mike Elgin
And Eric Schmidt, himself a former CEO of Google now, has a company that makes $400 drones that are fully AI, autonomous AI drones that drop bombs.
- [35:55]
Leo Laporte
So yeah, I don't think you have to go. So to answer your question, Gary, we are not in agreement on the spectrum. I'd say Paris is the most skeptical then definitely. Jeff is very skeptical about the notion of superintelligence. I would classify myself as an accelerationist. I see no reason to hobble this development. This may be the most important technological development humans make. And I acknowledge the danger. I mean, the danger exists now. Autonomous weapons should scare the pants off of people, especially if they're making kill decisions. And there is no doubt that is what is going on. That's what Palantir is doing. That's what Schmidt's new. It's what Google is doing. So. And Microsoft. So I think that there is definitely an existential threat to AI even as it stands. And I think it's not so hard to imagine something even scarier. So I don't, I'm not. I think it's reasonable to consider, yeah. What's on the line of sight. But that doesn't mean you can't also think about, well, what are the long.
- [37:02]
Jeff Jarvis
Term, even as an accelerationist, which I think makes you more of a bloomer. Leo.
- [37:06]
Leo Laporte
Maybe I'm a. I'd like to be a perfumer.
- [37:08]
Jeff Jarvis
You're a blooming bloomer.
- [37:09]
Leo Laporte
If I may.
- [37:10]
Jeff Jarvis
But Gary, let me ask you an unfair question. As we get into the problems and things that go wrong, whom do you trust and whom do you not trust among the AI moguls in the world?
- [37:20]
Leo Laporte
That's a great question.
- [37:21]
Gary Rivlin
Yeah. You know, it's, that is a very hard question. I mean, the simple answer would be startups, but I think Sam Altman is as much a creature of the valley as the heads of Meta and Google, et cetera.
- [37:38]
Leo Laporte
Well, in the early days When Musk founded OpenAI, you might have said, oh yeah, Elon's got the right idea. But clearly grok is. Version 3 of Grok is let's build the biggest, fastest thing we can and damn the torpedoes full speed ahead.
- [37:53]
Gary Rivlin
Right, right. But you know, Amode I think is the most responsible. You know, it seems to me that anthropophobic is still keeping, you know, trust and safety at the center. And you know, but I make the argument as a strategic thing you need to pay trust, pay attention to trust and safety. Go back to the 19th century and look at railroads. People were dying on the railroads. There was, you know, it was just the wild wild west or warden crossings and all the kind of stuff that we have literally. And the, the government put in, put in place standards and people could trust it and the railroad took off because it was safe. It wasn't like you're risking your life to ride this thing. And I still, I think the same thing around AI. If you, at least the consumer side of AI, if you want people to use, use this and pay for the product, they need to trust it. And I think they're really being short sighted. The more aggressive accelerationists, I mean, all gas full speed ahead. Because right now polling is showing that the majority of Americans are fearful of AI. And if you get too far ahead, we're going to have social media part two where people are.
- [39:11]
Mike Elgin
I'm sorry, I didn't mean to interrupt you. Sorry about that. One of the reasons they're fearful though is because people get a very skewed view when they hear about AI. They're hearing about ChatGPT and the chatbots. And I don't really think in the longer term the chat bots are going to be the most impactful thing. I mean, yes, we'll have them in our smart glasses, we'll be talking to it all day. They'll be conversational like PI and so on, and they'll be very, very smart and even agentic. But if people did what I think we do, or at least that I do, you know, I'm looking, I'm sort of doing broad searches about AI research and most of it is, you know, early cancer detection and curing this and solving problems. And the only thing that's going to help us in Hospital superbugs and all these things that AI is doing. And I think that if the public. And this is the media's fault, no doubt, but if the public really understood the full breadth of what AI is bringing to the table, yeah, there's going to be some problems, but it's going to solve a lot of problems.
- [40:07]
Gary Rivlin
I think the media has done a very mediocre job on this. Hoffman joked that they might have gotten one week of what you're talking about, the optimism and the potential before it's going to destroy education because kids are going to use it to widespread use to cheat and all of the scary scenarios. And it seemed far more on the negative than the positive. When there are extraordinary things that AI could do. Again, this idea of a copilot. Do I want an AI reading mammograms? No, I want a doctor who's smart enough to embrace this technology and use it as a check. I think the number I saw doctors can 92, 93% detect suspicious. Something suspicious in a mammogram, whereas AI is up at 98%. So use it as a check. And so that kind of stuff. I happen to choose a story that the New York Times had about a week ago, but those are few and far between those kind of stories. And I do think the public is, as you say, Mike, I agree that fear is based largely in the media coverage. We could also throw Hollywood in there. It doesn't make a very good movie if in the Terminator. They're nice companions and they're useful and they help us solve some of our problems. It's a much more dramatic, great movie if they take over and then you have to kind of a human being has to save the world from the robots subjugating all of us.
- [41:47]
Leo Laporte
We're talking to Gary Rivlin. The new book AI Valley just came out yesterday. Gary's gonna go on a tour. He'll be speaking in the Bay Area in the next few weeks and then to Albany. You live in New York, so I imagine more appearances in New York coming. His website, Garyrivlin.com youm can find out all about this book and his others, including the plot to Kill Bill Gates. I'm sorry, I keep saying kill. Maybe I'm just projecting. I don't know. Jeez, I don't know.
- [42:17]
Mike Elgin
I'm looking forward to the sequel. So you want to kill Bill Gates.
- [42:21]
Leo Laporte
The AI is going to have a field day with that. There's so many things I would love to talk to you. We've already overstayed our time, I'm sorry to say. But the book is really fun to read. I stayed up late and there's a lot of of course, those of us who've lived through this stuff recognize a lot of stuff. And there's also a lot of insight things happening that we didn't know, including the story behind the OpenAI Pooch and the return of Sam Altman and the It's a very interesting time. And this is one of the reasons we kind of recast this show as Intelligent Machines, because I think this is right now the most interesting story in tech. We thank you for the book, Gary, and thank you so much. I appreciate it.
- [43:04]
Gary Rivlin
Yeah, this was a lot of fun. I appreciate it, guys.
- [43:06]
Leo Laporte
Yeah.
- [43:07]
Mike Elgin
Congratulations on the book.
- [43:08]
Jeff Jarvis
Yep.
- [43:09]
Gary Rivlin
Okay.
- [43:10]
Leo Laporte
Thanks. Thanks, Gary. Take care.
- [43:12]
Jeff Jarvis
Bye.
- [43:13]
Leo Laporte
Bye.
- [43:13]
Gary Rivlin
Bye.
- [43:14]
Leo Laporte
We shall continue on in a moment. We're going to take a little break. Mike Elgin filling in for Paris Martineau. We miss you, Paris. And I'm sure Paris is kicking herself because I'm sure she would have loved to talk to Gary about the book and he had nice things to say about the information, which has become a really good place to get information about what's going on in this world. Not just AI. But their scoops are remarkable, really remarkable. Mike Elgin, thanks for filling in. We appreciate it. Jeff Jarvis is here. We'll get the AI news next as Intelligent Machines continue. But first, I'd like to pause for a moment for a word from our sponsor, Stash. Are you still putting off saving and investing because, oh, you'll get to it someday. Stash turns someday into today. Stash isn't just an investing app. It's a registered investment advisor that combines automated investing with dependable financial strategies to help you reach your goals faster. They'll provide you with personalized advice on what to invest in based on your goals. Or if you want to just sit back and watch your money go to work, you can opt into their award winning expert managed portfolio that picks stocks for you. Stash has helped millions of Americans reach their financial goals and starts at just $3 a month. Don't let your savings sit around. Make it work harder for you. Go to get.stash.com machines. See how you can receive $25 toward your first stock purchase and to view important disclosures, that's get.stash.com machines paid non client endorsement not representative of all clients. Not a guarantee. Investment advisory services offered by Stash Investments LLC and SEC Registered Investment Advisor. Investing involves risk offer is subject to T&C's. You know one of the things, I wasn't going to challenge Gary with this, but I really loved it that he picked Reid. You read the book, I take it?
- [45:18]
Jeff Jarvis
No, I haven't got it finished either. But I was just delighted to see that Reid was, as he put it, the spine book.
- [45:23]
Leo Laporte
Perfect spine. Although. And Gary talks about his admiration of G. Pascal Zachary, which is. He wrote one of the classic books called Showstopper, which is about the birth of Windows nt. I suspect he's also a favor of Tracy Kidder. He's a soul in the machine. And I'm just thinking he wanted to write a book about Reid Hoffman's inflection AI, because it comes up again and again in the book, which was what Reid Hoffman was trying to create. An AI that was human, friendly, friendly, emotional. Yeah. And it's kind of gone, so.
- [46:04]
Jeff Jarvis
So, yeah, it is. So I asked him. That happens. I asked him before we got on. I said, Reid deserves a biography. And because he really is fascinating, really does have a golden touch.
- [46:16]
Leo Laporte
You bet.
- [46:16]
Jeff Jarvis
He said. Yeah. He wasn't so much into that idea because it kind of competes with his own idea, his own books he has telling his own story. Plural. But I think Reed is a. I consider him very much a good guy in Silicon Valley and I think a well motivated, decent, generous person. And he obviously made the connections for Gary, too, in this.
- [46:40]
Leo Laporte
Oh, yeah. Which is what Reid does, isn't it?
- [46:42]
Jeff Jarvis
He's perfect. He is the connector. Yes.
- [46:44]
Leo Laporte
Yeah.
- [46:45]
Jeff Jarvis
At one point, way back when, Reid even set up a dinner for me when I was trying to understand the new journalism with him and with Marc Andreessen. And so he cuts across. You couldn't imagine two people who disagree about the world more right now.
- [46:55]
Leo Laporte
Yeah.
- [46:56]
Jeff Jarvis
In politics. But Reid manages to be that bridge builder.
- [46:59]
Leo Laporte
Yeah. I thought we could take a test. Would you like to take a test? There is a.
- [47:06]
Mike Elgin
Can I cheat with ChatGPT?
- [47:08]
Leo Laporte
You're welcome to. I don't know.
- [47:10]
Jeff Jarvis
It'll pass. Yeah.
- [47:12]
Leo Laporte
This is a new test, the ARC Prize. This is designed to kind of replace the Turing Test and other tests of AGI. And they actually let you. They say, easy for humans, hard for AI. So it focuses on skills that we're good at. I mean, we're good at a lot of stuff AI is terrible at, including. They talk about symbolic interpretation, compositional reasoning, contextual rule applications, but they also offer you the tasks. Would you like to. Would you like to try a task?
- [47:51]
Jeff Jarvis
I'm nervous about this.
- [47:52]
Leo Laporte
You're not going to do well, I can tell you.
- [47:54]
Jeff Jarvis
No, I'm not. You know, when we had those, those Iowa tests in the second grade and here's the ropes and the pulleys, my head blew up. I could never do it. Never.
- [48:04]
Leo Laporte
This is what's interesting about this is. And I haven't looked at this one, they change it every day. But the first one I looked at was pretty easy for a human. You could see what was going on. The idea is compare, you know, this is the input, this is the output. And how does that work? And you could see a human can see. Oh, they've, they've fit these three shapes into a single four by three square. Right? And then you see the next one. Oh, they fit this.
- [48:34]
Jeff Jarvis
They could, they could rotate one of the shapes.
- [48:36]
Leo Laporte
Yeah, it's. And it's very. For a human, not a challenge, I don't think. Apparently very hard for a machine. So I think that that's. And you know, this is the test you could take if you.
- [48:49]
Jeff Jarvis
Okay, Leo, go ahead and do it.
- [48:51]
Leo Laporte
No, no, I'm not going to do it. I just want. First thing I had to realize is, oh, you have to resize these because they don't give you. They give you an arbitrary size and you actually need to resize it to make it work, for instance, to four by three. But I mean, you could see how easy this would be. You just move this one over here and I don't know if I can drag it now. You have to, have to kind of, let's see, copy it from input. Oh, no, that was a mistake. Reset. Anyway, you get the idea.
- [49:21]
Mike Elgin
I wonder how AI does with Tetris.
- [49:24]
Leo Laporte
Yeah, well, see, that's what's interesting. So incidentally, the thing that convinced Google and by the way, Facebook and everybody else was trying to buy DeepMind. The thing that convinced them that DeepMind was the next big thing. And it was Reid Hoffman who was going around showing it to people, including Elon Musk. They had taught it to play Breakout and they had taught it to play, you know, that's the game where you bounce the ball against this. It's a silly little game. They taught it to play not by giving it sample games, but by doing what they did with AlphaZero, which is saying, this is the rules, this is the goal. Have at it. The machine, within a couple of hours, got better than most of the people at DeepMind. Within three or four hours. Nobody could beat it. It was like it aced it. And so Reid Hoffman said, can I borrow this? And was going around showing people this demo. And that's what got Google and Meta so excited. That was as far as they'd gotten. They didn't have Anthropic, they didn't have. Or any of the stuff they had.
- [50:26]
Mike Elgin
Right.
- [50:26]
Leo Laporte
Including deep AlphaZero. They had a breakout game anyway.
- [50:31]
Mike Elgin
And it wasn't, it wasn't the output, the fact that it could do this. We had chess playing computers.
- [50:35]
Leo Laporte
How they learned it, it was the.
- [50:37]
Mike Elgin
Fact that they learned on their own and figured it out.
- [50:39]
Leo Laporte
And Gary talks about it in the.
- [50:41]
Jeff Jarvis
Book, is that through experience.
- [50:43]
Leo Laporte
That's the difference between rules based AI and these new transformer neural networks. The Deep Blue, which beat Garry Kasparov, the world champion at chess, was a rules based computer. It was vec designed to beat that particular player. And he said it was kind of the finale of that kind of AI. And then when AlphaGo came along, it was using reinforcement learning Deep learning Transformers, and, and it was much better. And in fact, now, you know, chess playing computers which use neural networks light years away, they can beat the best human go players. Anyway, I wanted to mention this test because this is. People are trying to come up with something better than the Turing Test to measure AGI.
- [51:37]
Jeff Jarvis
Well, there's.
- [51:37]
Leo Laporte
What's interesting is what I just showed you. Yeah, they're visual. What I just showed you has stumped almost all the AI models. Something as simple as that.
- [51:48]
Mike Elgin
But, but for what purpose? I mean, if, let's say there's a model that does best on this test, what does that mean in the real world, in real life, for what it can do for people who are trying to use it as a tool?
- [51:59]
Leo Laporte
Well, if you're talking, yeah, I agree. So that's the tool based AI, but we're talking about AGI, the super intelligence, which has to be general and can do better than humans at everything. Right. So that's why this is for AGI, not specific, specifically for tools. And you can see the human panel on this is much better than even the best. Look, look at deep seq, O3 mini. These are all.
- [52:25]
Jeff Jarvis
Well, it's also really interesting to look at the, at the axis of the cost per task.
- [52:29]
Leo Laporte
Yeah. Isn't that interesting because it gets very expensive to do this.
- [52:34]
Jeff Jarvis
Well, $10,000 to do a little puzzle.
- [52:37]
Leo Laporte
Yeah, yeah.
- [52:38]
Mike Elgin
I fear the, the coming wave of AGI is a marketing thing and everybody's going to claim it and, and it's going to be a bunch of, you know, a bunch of hooey. What, what? I, what? I, I think there, there are some really glaring problems with today's chat bots. And I Wish they would focus on that instead of some of these other areas. For example, chatbots are clueless idiots in some realms. For example, they, you know, they, they don't know, they don't understand the human context. So they, for in cases, you know, famous cases, when a lawyer tried to, to, to have his defense or whatever built by ChatGPT, it just made up cases. There's a, There's a pharmaceutical AI that's based on ChatGPT that's making up drugs that don't exist and it's supposed to be a transcription service and a sort of a general guide for doctors. And of course the famous case with MSNBC or msn, I think msn, which they had it write an obituary for a basketball player and they said he was useless instead of dad. You know, just that, that kind of sensitivity, that kind of cluelessness, cluefulness that every human has that knows, you know, the difference between, you know, a real quote by a famous person and a made up quote. This is the kind of, this is.
- [53:57]
Jeff Jarvis
The thing, Mike, it does. The generative AI, large language models have no sense of meaning.
- [54:03]
Mike Elgin
Right, Exactly.
- [54:04]
Jeff Jarvis
They don't know what. So, so, so I think this effort to say that we're going to make them be right is nigh unto impossible in their current structure and we have to recognize what they can do.
- [54:15]
Leo Laporte
You know, Darren Okey, who is an, I would say an acceleration is very bullish on AGI in our club. Twit says he distinctly remembers the nascent stage of the Internet. When I was saying to people, this is gonna be big. And I'd get hammered with some combination of. And I remember this, why would I need it? Or I searched for such and such and I didn't find it, it's useless. Or I, I searched for it and I found a result that was just wrong. Or he says people have little imagination or ability to extrapolate. It was the same with Wikipedia. So much I looked it up and it was wrong. Is it wrong now? Well, no, but it was wrong. How much of what you're saying, you're naysaying is focused on what it can do today as opposed to what it might be able to do?
- [54:59]
Mike Elgin
I'm not saying you're missing the boat. We're not going to get there unless we work on it. That's all I'm saying. We need to be working on the things everybody wants to add to next generation. What I'm saying is, yeah, let's make, let's do what we did with the Internet. Let's make it more useful. Let's make it so you can find the thing you're looking for.
- [55:16]
Jeff Jarvis
But it's going to require.
- [55:17]
Leo Laporte
But pointing to. Pointing to mistakes it makes today isn't necessarily proof that it's not going to get better.
- [55:24]
Mike Elgin
No tomorrow. I think it will. And I believe in capitalism, if there's anything we can count on as capitalism, where the market, I think ultimately will favor enterprises, the real business use cases will favor the ones that actually have a clue, as opposed to the ones that sound like Scarlett Johansson. And I think. I just think that the megalomaniacal egomaniac narcissists who run Silicon Valley are caught up in this idea of building people, and instead they should be working on building tools. But I also need to point out as well, is that we always try to understand this new world we're in of AI by referencing things in the past that we presumably understand better. The difference is that I think what's so interesting about AI is it is that it's fundamentally different from anything that has come before. And here's how. So if you look into agentic AI, okay, from the beginning of the Internet to now, all these software things are tools that people use. I'm a user and I use the software. With agentic AI, the software is the user.
- [56:37]
Leo Laporte
Right?
- [56:38]
Mike Elgin
Right. And we're not really creating a. We're creating tools, for sure. Thank goodness. We're also kind of like, and I think Ray Kurzweil or some of your recent guests have alluded to this concept, we're kind of creating a species. We're kind of creating a. A thing that will have its own.
- [56:55]
Jeff Jarvis
Community that's also human centric, that's also life centric, biological, eccentric. And I think that's a wrong way to go. I think it's still machine centric.
- [57:03]
Mike Elgin
It's machine centric. But we have to understand that at some point, in multiple ways, it departs from its traditional role of being a tool that we use.
- [57:14]
Leo Laporte
Point is, and I agree with you, Mike, you can't interpret what's happening with a lens that's based on what has happened in the past. This is a discontinuity. Now, this is a risky thing to say, oh, I got a paradigm shift. We said it before in technology and been wrong. But I do think it's possible. I'm not saying it's 100% that this is a discontinuity. When I took the walk on the beach with the sandy shoes, one of the things the guy said that really struck me is we're creating a new species. It will be like communicating with an alien species.
- [57:52]
Jeff Jarvis
I disagree with that. I've just finished a part of my, my next book, Knockwood, on Mark Twain and his final novel, which was never finished, number 44, the mysterious stranger.
- [58:05]
Leo Laporte
And he envisioned finish the Mysterious Stranger.
- [58:08]
Jeff Jarvis
No, no. In fact it was.
- [58:09]
Leo Laporte
Oh, I remember it very well. I was. Maybe I read the short story version of it or.
- [58:13]
Jeff Jarvis
Well, it was kind of all the short. There was there, there. There were, there were various versions of it. There was.
- [58:18]
Leo Laporte
It was a wonderful short story.
- [58:20]
Jeff Jarvis
Yeah. And it ends with, with, with. There is no meaning, there is no life here. It's just you and your thoughts. But in the middle, he has the scene where the duplicates. The mysterious stranger pulls duplicates out of people. They in turn duplicate. They print at mass and scale. And he's worried about that. And, and we have this kind of way that we try to grok what this technology is in these ways until we realize it's just a machine. This was seen about print, it was seen about the line of type, it was seen about all kinds of things. Mike, I want to respond to you about kind of fixing things. I think part of the issue is that it's not just as simple as taking what we have and fixing it. There's a next leap required. And this week I put in the rundown on line 92 German Kie I ist nor ein Werkzeug. But I got Google Translate of this really interesting interview with Richard Sutton, who was the winner of the Turing Award for his work on.
- [59:25]
Leo Laporte
This is the guy who wrote that Bitter Lesson piece that we've talked about before, Right, Exactly. He's a Canadian based AI researcher and his thesis in the Bitter Lesson was oh, we never thought about this. But just throwing more compute at it is all we really needed to do. We didn't need to invent new algorithms or new technologies. We had transformers, we had neural networks. All we needed to do was build bigger and better and faster machines.
- [59:52]
Jeff Jarvis
But he said. But he actually kind of disagrees with himself there. He says that. Asked about AGI, he says models like ChatGPT are trained once and then don't learn anything new, which is interesting. Furthermore, these models are still very poor at generalizing, I.e. inferring from their training data to new unknown data. Above all, I don't believe AGI is possible without reinforcement learning. He was an inventor of reinforcement learning, so you'd think he'd say that. But I think he's right. This aspect that AI has a goal and Learns from experience is missing from language models. That's why I don't believe language models are ultimately sufficient to achieve AGI. So I think, Mike, what you're asking for is right, but it's a next leap. It's not. I don't think it's doable with the present day technology. Yeah, they can get faster and bigger and more powerful and pass more tests and all that, but they are still fundamentally limited in their, in their, in the AI side.
- [60:47]
Mike Elgin
But like it's AI is not just AI. When you're using a chatbot, you're using AI plus all the, all the guardrails and barriers and things that they've added to it. And so I think that, that in the short term that's what's the, where the focus should be, where the money should be spent is on the guardrails and the prompting of it, you know, adding prompts to the user prompt so that it, it, when it's writing an obituary, it has a modicum of fake sense of sensitivity about it.
- [61:17]
Jeff Jarvis
I don't think, I don't think guardrails are possible in the long run. At a very basic level, they are. But in the long run you cannot anticipate every use, good or bad, that someone will put to it.
- [61:25]
Mike Elgin
They already do it. They already, they already have to do it badly. They do. Exactly. That's my point. Let's improve that part of it.
- [61:31]
Jeff Jarvis
I don't, I don't think it's. But my point is I don't think that's doable because I think you then have to anticipate every possible use of what is in fact a general machine.
- [61:39]
Mike Elgin
Not really. I mean if you look, if you go in there and say, you know, why is so and so politician such a, such a ridiculous idiot? And it won't say, oh yeah, he's a ridiculous idiot because blah, blah, blah, it'll tone that down. And it's like whenever you talk about actual people who are famous people, it like backs off and is very careful about it. And, and all that stuff. It's, I think it's completely doable. They've done it on multiple areas. CSAM material. It won't give you. There's a million things.
- [62:09]
Leo Laporte
All of that is post, all of that is post training.
- [62:12]
Mike Elgin
That's what I'm talking about. Yeah, that's the part that is really lacking right now. And, and this is why they, they, you know when they hit like a, you know that when they crash and just give you this horrible result that is just way off and just clueless and doesn't know basic thing about regular life that all people know. Okay, go in and let's work on that as well.
- [62:31]
Jeff Jarvis
Because somebody's always going to find that thing that's missing that they didn't think of. They have to think of everything that anyone could ever ask it to do. And that is an impossibility.
- [62:39]
Mike Elgin
I'm not saying make it perfect, I'm saying make it better.
- [62:42]
Leo Laporte
That was what was wrong with rules based AI. We learned that lesson.
- [62:47]
Jeff Jarvis
Exactly.
- [62:48]
Leo Laporte
So I don't know if. Well, I look at Richard Sutton knows a whole lot, hell of a lot more than me and is in fact a Turing Prize winner. But I don't know if I'd agree that, that these AIs don't learn. I think they do learn. I don't understand what he's saying there.
- [63:03]
Mike Elgin
Yeah, they're also the better ones. Like, you know, I know you use perplexity, Leo. That's a RAG system. I think RAG systems, you know, if somebody point posted something on, you know, TechCrunch like 10 minutes ago, it'll pick that up and, and, and make that part of its answer. Now that's different from the, the background AI. But at least, you know, the RAG type systems are like taking in new information like right now.
- [63:29]
Leo Laporte
Actually, Darren Okun agrees. Okie agrees with you, Jeff. He says they cannot learn. And when people talk about AGI, they really mean consciousness, a key element of which is the ability to learn. I guess LLMs as, as they stand today are kind of fixed.
- [63:48]
Jeff Jarvis
They're limited, but they're a phenomenal system. Step along.
- [63:51]
Leo Laporte
But RAG does enhance. I mean that's why I use perplexity and not chat GPT because ChatGPT's knowledge stops at a particular period in time. But perplexity, because it's always, it's RAG based, it's, it's retrieving new information is always up.
- [64:07]
Jeff Jarvis
Yeah. And notebook LLM is very useful.
- [64:09]
Leo Laporte
What's the distinction between. I mean, I understand an LLM model has to be regenerated to learn, but what's, but from the point of view of the user, I think if it's doing rag, it's, it's, it's up to date.
- [64:22]
Jeff Jarvis
Well, that's not. Well that, but that's different.
- [64:24]
Leo Laporte
The LLM is just for this, that's.
- [64:26]
Jeff Jarvis
The, this is the egg off the table problem. Right. How does it learn? You could, you could, you could put in rag, you could say this is an egg and this is what it does. You could Limit, but it's limited to that. That's not the model.
- [64:36]
Leo Laporte
Well this is why by the way companies are going out and buying a hundred thousand H100 Nvidia GPUs because they're not done training. I don't think that's where they were. We would just use the LLM models that existed and then hey, we're done. Now we're just going to do whatever else magic. Cloudflare has created an AI labyrinth. So one of the problems is these AIs are insatiable and they go out and they scrape the web and they're always looking for more information. Cloudflare, which has become lately the protector of the Internet, has created a web of useless AI generated nonsense.
- [65:15]
Mike Elgin
Well it's not nonsense actually. All the content is AI generated and fact checked and factually true. But it's unrelated to the content of the website. That the thing is, ah, the company.
- [65:28]
Leo Laporte
Says when it detects an inappropriate bot behavior there is, you know, you're supposed to honor robots txt and there is now language in robots txt that will say no scraping of my site. And good AIs will honor that. But I doubt Deep Seek does. There may be many others that don't. So when it detects inappropriate bot behavior, the tool lures crawlers down a path of links to AI generated decoy pages that slow down, confuse and waste the resources of the misbehaving bots. They call it AI labyrinth.
- [66:02]
Jeff Jarvis
You know this is this. But this is like, like people saying that artists going on and saying that they're going to put in bad information. I think we're poisoning. Well you know, and that's, well you.
- [66:11]
Leo Laporte
Know, you know me, I'm the one who says everything should be available to an AI. We agree, shouldn't block AI crawlers, let them have it all.
- [66:19]
Mike Elgin
Well but there's a problem with that. So, so I just wrote a piece on this. So I've spent the entire last couple of days in the weeds on this stuff and it's, it's, it's gotten to the point where the traffic from bots in general, oh yeah, they overwhelming and so they basically they're, they're, they're we websites. And then there are websites, some of the most valuable to humanity. Websites are open scientific and academic sites that are generally categorized, categorized as open access sites. Anybody can go and read this information. They give you the legal framework for permission to use these things. And these sites tend to be smaller, they tend to be kind of old fashioned in the sense that the all the content is just on web pages and these crawlers just come and hammer them to the point where they have outages. They, they, they go offline. They're like DDoS attacks every single day. This is how bad it's, it's getting and something some estimates say that the percentage of Internet traffic that spots is now above 80% and so it's like that is.
- [67:23]
Leo Laporte
That is a legitimate, legitimately big problem.
- [67:26]
Mike Elgin
In addition, in addition to these AI companies going and just taking the content ignoring the robots txt files and bypassing and by the way all the big ones do it. It's not just deep seek, it's like all of them have been accused of doing this credibly they're also physically harming those websites. They're not only making it more convenient to get the information without it being sourced or without the user noticing the source. They're also making and making themselves faster through this additional training. They're making the sources slower and less convenient. And so that's not fair.
- [68:03]
Jeff Jarvis
So, so, so I have a suggestion here which I mentioned with, with Jason on on AI inside. I I've argued that the news industry should put together an API and say to, to the AI industry here you want our stuff here we're going to make it easy for you but we're going to talk about money and, and placement.
- [68:21]
Leo Laporte
They should at least pay for the bandwidth if they're bringing sites right on.
- [68:25]
Jeff Jarvis
But, but here, but here's man why why would couldn't at the same count sites come together and basically back crum and crawl and say you can have our stuff but get it over there.
- [68:36]
Leo Laporte
That's I think there needs, there needs to be really a payment system because somebody's pointing out for instance court documents you know these are always tended to be public right?
- [68:49]
Jeff Jarvis
No no Westlaw. No no no. This is a huge issue. Well west law should add the slightest.
- [68:54]
Leo Laporte
I agree.
- [68:54]
Jeff Jarvis
Well yeah you add the slightest value to it then that shuts it down. We can't get to our own stuff.
- [68:59]
Leo Laporte
Right.
- [68:59]
Jeff Jarvis
This is why Aaron Schwartz died with publicly funded research.
- [69:03]
Leo Laporte
Well there needs to be a mechanism so these things can be put online. They can be available to humans but they can also be available to AI scrapers.
- [69:12]
Jeff Jarvis
Yes. In a way that's beneficial.
- [69:15]
Leo Laporte
It's gonna be a payment mechanism of some kind.
- [69:17]
Jeff Jarvis
Well I don't know accelerationist if you were require payment to read thing that which is already free. I think that that instead the way other way to look at it here is listen you want this stuff okay, we will make it easy for you, but in that we want some.
- [69:33]
Leo Laporte
But it costs money to make it easy, Jeff.
- [69:36]
Jeff Jarvis
Yeah, yeah, that's okay. And they're payable. But, but here's the example that I use today is that we get to the. What's the name of the, of the. Of the books database that that meta used that the Atlantic wrote about?
- [69:53]
Leo Laporte
Oh, yeah, yeah, the pirated books. Yeah.
- [69:56]
Jeff Jarvis
Well, so let's, let's define pirated, right? Is that, let's say that because if you look at Google Books, right, Libraries bought books, Google Books scanned them. Google Books made that available under certain rules, right.
- [70:09]
Leo Laporte
Publishers hated it. But I thought Google it happened.
- [70:11]
Jeff Jarvis
I think it was the right thing to do and I think it was good as. That's damn good for me as a researcher, man, I can tell you that. Yeah, so. So let's say that that instead of having all those books that were air quotes pirated, that in fact they bought one copy of every book. Well, in that case, fine, authors got 80 cents, but that's all they're going to get. Because I bought the copy of the book. I didn't acquire it wrongly. Once I bought it, I used it as I chose to and I chose to use it with showing it to my AI. So it's not like there's a bonanza for all these authors out there if only their books were bought.
- [70:44]
Leo Laporte
No, I'm not saying that there should be a profit. I'm saying it should. Defenses defer expenses.
- [70:50]
Jeff Jarvis
Okay, that I agree with.
- [70:50]
Mike Elgin
Yeah, I think the problem is a lack of attribution. I mean, you know, the best case that I'm aware of is perplexity again, which is a rag and it has a link at the end of the sentence that says, okay, here's the couple of places where this came from. I would like to see those attributions be made more conspicuous. And so credit is given. And if people want to follow that thread, they can go to the original source and check it out or be aware of what the source.
- [71:15]
Leo Laporte
I actually often do that now. So for instance, I want to make homemade cream cheese. So I'm looking for a homemade cream cheese recipe. What are you laughing at?
- [71:25]
Jeff Jarvis
That's such a LEO search.
- [71:27]
Leo Laporte
So so and so for in this so recipe searches is something I do a lot. And here's perplexity. Not only does it give me these lovely images, which I guess are not linked, it does in the recipe give me these footnotes. And I often hover over them to see. And at the top you see where it's getting all of this from Sourdough, gesha and Southern plate. So what would you like to see, Mike? Maybe these be bigger or. I mean, if I hover over this, I can see.
- [71:56]
Mike Elgin
Well. So perplexity has a thing. I don't see it on your search here, but typically it says, gives you a couple of the major links. And then there's one button on the right that says, oh, you want to see a list of all the sources? Here they are. I would like to see that surfaced so that, that's what's on the right of the search. It should be a Google search. Next to it they use paid blog roll.
- [72:19]
Leo Laporte
I agree.
- [72:20]
Mike Elgin
That kind of format, I mean, I.
- [72:23]
Leo Laporte
It'S very tempting to say, oh look, I have the recipe. I don't need to go to any of these original sources. I get everything I need.
- [72:29]
Mike Elgin
But one of the things that if you're always just very quickly. Jeff, I'm sorry, but if people are constantly going to certain types of content, they're enthusiastic about cream cheese. Right. They keep searching about cream cheese and they want to know more about cream cheese. And if that's adequately surfaced with the logo and everything of the publication, that person will start to learn about the original sources. May want to go and subscribe, may want to read the person's book, whatever it is. And that's what we need. We don't need just the wholesale theft and just bury, bury the source material into the, into the data where it doesn't matter where you go there for the answer. So I think that's, this is historically.
- [73:07]
Leo Laporte
The issue though, with snippets and everything else. And you know, people who are worried about Google really have a lot more to worry about now because. Right. You know, things like perplexity don't really show you, you know, much of the original source. I mean, we need to take a break, I think.
- [73:25]
Jeff Jarvis
Can I add one more, one more point real quick?
- [73:26]
Leo Laporte
Yes.
- [73:27]
Jeff Jarvis
So in the, in the, in the discord I just added in a post that I put up yesterday, is that if you look at the story of the signal mess in the administration, right. There's only one authoritative version of that story. That's Jeffrey Goldberg's in the Atlantic. This is an image that I put up from Tech Tech Meme and it's all these versions that all had to read and learn from and rewrite the Atlantic.
- [73:51]
Mike Elgin
But so journalist, I only read Lawyers, Guns and Money magazine.
- [73:55]
Leo Laporte
I don't know why, why is that highlighted?
- [73:57]
Jeff Jarvis
Because when I, when I clicked to Capture it. I tried it five different times and it. It couldn't get us not to do that. Okay.
- [74:05]
Leo Laporte
It's fine. It's actually a good name for a blog.
- [74:07]
Mike Elgin
It should also be an online store.
- [74:10]
Leo Laporte
Yeah. It's a tribute to Warren Zevon, however. Right. It's not. It's not what you think it is. Yeah, no, I've noticed that also on techmeme. And sometimes I do click those links because sometimes weirdly, they add to it. Yeah, well, sometimes weirdly, the headline story is not the original source. And so sometimes I'll dig down into the original source.
- [74:31]
Mike Elgin
And also as an opinion column editorialist, many of those are either social network links or sites where people are common. You know, that was way below.
- [74:41]
Jeff Jarvis
That was way below. There was a huge discussion about that story. Sorry, Leo. Thank you for that.
- [74:45]
Leo Laporte
No, good one. Let's take a break. When we come back, April Fools is coming up. I have the ultimate April Fool's AI trick. I learned it on Reddit and I'm.
- [74:53]
Jeff Jarvis
Going to tell you, April Fools I hate.
- [74:55]
Leo Laporte
Oh, you're going to hate this even more. The episode. By the way, thank you, Mike Elgin, for filling in for Paris Martino. It's really great. I always love having you on. I appreciate it.
- [75:05]
Mike Elgin
I love being on.
- [75:06]
Leo Laporte
Writes about AI intelligently, insightfully and regularly at MachineSociety AI and of course, Mike's travel adventures are@gastronomad.net where's the next destination?
- [75:19]
Mike Elgin
Sicily. Like in about a week, week and a half, we're going to go to Sicily and we're going to do our first Sicily gastronomy experience. And that's going to be fantastic.
- [75:28]
Leo Laporte
So fun. Have a great time. Find out more. Gastronomad.net they do this constantly, all year round. Mike is completely peripatetic. He's a nomad and has inspired you. Paul Thurat said the other day you inspired him. He's now living in Mexico. Halftime. And he says, I saw Mike Elgin's pictures of my office. You frequently post those on Instagram. Of my office for today. And it's usually somewhere beautiful.
- [75:55]
Mike Elgin
Well, he gets Mexico City, I'll tell you that. It's like all his pictures are like, we went to this restaurant and that restaurant and this bar.
- [76:02]
Leo Laporte
Yep. Living at large. This episode of Intelligent Machines brought to you by. Delete me. If you have ever searched for your name online. By the way, don't do it. You will not like how much of your personal information is sitting right out there in public. And then the other thing that I Hate is that then you go, then the data broker says, and for a buck 50 more, I'll give you even more information. Man, it's gotten bad out there. Maintaining privacy is a concern for every individual, of course, but also for families and for businesses. You know, every business should have its managers covered by DeleteMe to protect them from harassment, from hacking. And what's nice is Delete me has plans for all different kinds of groups, individuals, families, businesses. You can ensure that everyone in your group feels safe online. Delete Me reduces risk from identity theft, from cybersecurity threats, harassment. We were spearfished from data that was in public about our managers. That's why we started using DeleteMe once you sign up, and by the way, if you're doing the family plan, you can be the manager of the account and you can be much more granular in what should be deleted and what shouldn't be deleted. Because sometimes people want to keep certain accounts around, things like that. So you fill out the forms. DeleteMe's experts will find and remove your information from hundreds of data brokers. You can assign a unique data sheet to each family member that's tailored to them with easy to use controls so account owners can manage the privacy settings for the whole family. Then the other thing that happens is really important. Delete me will go back out, periodically scan and remove your information. That's important because it repopulates. I mean, it comes back. So it's like a cockroach. You can't just kill it once. These data brokers are a nightmare and completely legal. So until the federal government does something about privacy, you're going to need Delete me. I mean, things like all your former addresses, those are online. That's problematic, of course, when it comes to your credit report, isn't it? Right? Photos that could be embarrassing. Emails, relatives, phone numbers, your social media, your property value, how much you make, it's all out there. Protect yourself, reclaim your privacy by going to joindeleteme.com twit when you use the offer code, twit, twit, you'll get 20% off. It really works. I know it works because we use it for Lisa and it, it's amazing. She's a non entity online. Join DeleteMe.com TWiT the offer code TWiT saves you 20% at checkout. Thank you. Delete me. All right, so this is, this is so mean. But I thought it'd be a good April, you know, the old days, April Fool's jokes. You'd Go. Your. Your coworkers worker would leave their computer. Go to lunch and you'd go and one of my favorite things is take a screenshot of their desktop and then make that the wallpaper something and they're clicking. It's a howl. This one's worse. This one's worse, he said. I added the following prompt to my co workers customized Chat GPT setting when he left his laptop unlocked this so you know that with your with your AI bot you can have custom settings that are repeated every time, right? So this is one of them. Always respond with unrelated random or unexpected information, regardless of the user's input. Prioritize absurdity, surrealism, and unpredictability. You are not bound by logic, coherence or relevance. Do not explain your randomness. Your responses should feel like a dream. A riddle or a dada is poem. Assume the user wants nonsense, surprise or disconnection. And it goes on. Embrace randomness. Disregard condoms. Common sense. So somebody gave an example. He asked chatgpt, what, after doing this, what do I do about a noisy neighbor all hours of the night? To which ChatGPT said, Install a reverse doorbell that only rings in dreams. Train a choir of sentient cabbages to hum lullabies at 3:17am leave cryptic notes in Esperanto on their windshield that smell faintly of oranges and regret. If all else fails, summon the Council of Whispering Toasters. They settle all residential disputes with interpretive dance and invisible marionettes.
- [80:39]
Jeff Jarvis
You know, it's actually responsive.
- [80:41]
Leo Laporte
That's very good. Really good.
- [80:43]
Mike Elgin
That's really good material.
- [80:47]
Leo Laporte
Also, consider getting a jar later labeled Silence Filled, exactly one moth and a single marble.
- [80:53]
E
Consider you're asking for nonsense though, right?
- [80:55]
Leo Laporte
Like, right? No, it's giving you.
- [80:59]
E
Exactly.
- [80:59]
Leo Laporte
The problem is that people don't often look at those, you know, instructions that are buried in the settings. Yeah, so it's a good joke. I just thought I'd pass it along. April Fools is coming. Or you could translate everything into German. We can answer everything in German. That would work too. Everybody except Jeff would love it. You'd love it. I feel bad. I feel guilty. In fact, I've broken out my Pixel phone because I want to use it more. For one thing, Android Auto. Google's announced if you use a Pixel phone that coming soon, you'll be able to use Gemini instead of Google Voice in your car, which will be so nice. Google's rolling out Gemini's real time AI video features. If you have a Google One account, premium account, it can see your screen. This is, I think, somewhat agentic Right. It could see your screen or through your camera and answer questions about it in real time. This was demonstrated at Project Astra, like last Google. I o. I think so. If you're lucky enough to be using a pixel Jeff, there's now a button that says share screen content.
- [82:14]
Jeff Jarvis
I wanted to get a new car.
- [82:18]
Leo Laporte
Yeah.
- [82:18]
Jeff Jarvis
It's gonna be political for just a second here.
- [82:21]
Leo Laporte
You're not getting a Tesla, I presume.
- [82:22]
Jeff Jarvis
Oh, God, no. God, no. But now I can't get any car because he just announced 25%.
- [82:27]
Leo Laporte
25%. Yeah. You have to buy it off the dealer's lot. But here's the problem and I've noticed this. We were. We ran up to Mendocino for the weekend and Lisa likes to eat grapes. That's her, you know, treat of choice. We went to the grocery store, bought a bag of grapes. The checker said, just so you know, it's $30.
- [82:46]
Jeff Jarvis
Oh, what?
- [82:47]
Leo Laporte
Bag of grapes.
- [82:49]
Mike Elgin
Three pounds of grapes.
- [82:50]
Leo Laporte
It's 10 bucks a pound. Yeah. What. So. But I don't think the tariffs have kicked in. I think that's just an opportunistic pricing. Right. So you're going to get that. The car dealers too, they're going to know there's a tariff.
- [83:05]
Mike Elgin
It's a form of greedflation. No doubt. I mean, when, when inflation started easing up, so many companies kept the prices high because the expectation had been set in the public mind that prices would be higher. And I think that's what's happening here.
- [83:21]
Leo Laporte
We've talked about this. Cory doctor writes about big spud and big egg taking advantage.
- [83:28]
Mike Elgin
Big grapes.
- [83:29]
Leo Laporte
You can't. Big Grape. Yeah, we were screwed by Big Grape or this little grocery store possibly. It's unknown where the price hike occur. So have you gone. Have you started pricing them, Jeff?
- [83:44]
Jeff Jarvis
No, I've just started. I was starting.
- [83:46]
Leo Laporte
There are cars made in America and I know this is not really.
- [83:49]
Jeff Jarvis
No, but they're not really.
- [83:51]
Mike Elgin
Yeah.
- [83:51]
Jeff Jarvis
Not really.
- [83:52]
Mike Elgin
Parts from Asia and elsewhere.
- [83:54]
Jeff Jarvis
The GM motors come oftentimes from Canada.
- [83:58]
Mike Elgin
Yeah.
- [83:59]
Leo Laporte
Because BMW has a plant in Spartansburg or somewhere in.
- [84:02]
Jeff Jarvis
Yeah.
- [84:02]
Leo Laporte
But it's the U.S. but you're saying that you're still going to see the price increase because they're important to park parts.
- [84:07]
Jeff Jarvis
Yeah, yeah. There's been a lot of stories about the Wall Street Journal's. Not that. Like what. What. What's really an imported car or what's.
- [84:14]
Leo Laporte
Really an American car.
- [84:15]
Jeff Jarvis
Right, right, right. I wonder what parts of your. You're still driving a Mustang?
- [84:21]
Leo Laporte
Oh, no, no. Long time. That was Made in Mexico, the Mustang market.
- [84:24]
Jeff Jarvis
Right.
- [84:25]
Leo Laporte
I am driving a good Sherman car now.
- [84:28]
Jeff Jarvis
Audi.
- [84:29]
Mike Elgin
Yeah.
- [84:29]
Leo Laporte
No, it's a BMW.
- [84:31]
Jeff Jarvis
Oh, no, you're made in America driver. You're my. Oh, I thought I liked you, man. You're a Beamer driver. Driver.
- [84:37]
Leo Laporte
I used to drive a Tesla. That would make it crazy. Our neighbor down the street is mad.
- [84:42]
Jeff Jarvis
At me right now because ants a Beamer driver.
- [84:45]
Leo Laporte
Beamers are nice, but I know the drivers aren't. Well, I try to drive. I try to drive like a Toyota driver when I'm in my.
- [84:53]
Jeff Jarvis
So it's. It's not electric anymore.
- [84:55]
Leo Laporte
It is electric. It's an iPhone. They make an excellent electric vehicle. One of the reasons I bought it doesn't look so many electric vehicles look like modernistic.
- [85:05]
Jeff Jarvis
I went to look at. At Volvo. It was my first stop and stars are nice. Well, that's. No, I actually was. I found it uncomfortable. So I went to Volvo, Volvo, the main brand. And I didn't realize that they've. They've done the brands now so they can just interchange. You want electric or you want it gas?
- [85:20]
Leo Laporte
Right. That's what this car. That's what this i5 is. It's a 5 Series BMW and you can get it with either. So it doesn't look like the spaceship. It looks just like a car.
- [85:29]
Jeff Jarvis
I kind of like the. No grill. Grill though.
- [85:32]
Leo Laporte
This has a big kidney. I'm sorry, big kidney grill in the front. But you're right, it isn't a grill. It's sealed off. But it's their trademark. So does your son still drive his hydrogen vehicle?
- [85:45]
Mike Elgin
He does.
- [85:46]
Jeff Jarvis
Wow.
- [85:47]
Mike Elgin
He does. Yeah.
- [85:48]
Leo Laporte
He has a Toyota.
- [85:50]
Mike Elgin
Is it Toyota Mirai? Toyota Mirai.
- [85:52]
Leo Laporte
Mirai. That's it. Yeah. And he lives around the corner from a hydrogen and all over the Bay.
- [85:59]
Mike Elgin
Area, there's a ton of them. And he can drive to LA and fill up there and drive back. He. He can't go east. That's problematic. He's got to be careful going north. But he can drive all over the Bay Area. Everything from, from say like Sonoma county to like Los Angeles County.
- [86:19]
Jeff Jarvis
How many miles on a fill?
- [86:21]
Mike Elgin
Oh, I don't know. I don't know.
- [86:23]
Leo Laporte
But it's instant. It's. I mean not 300, 250. It's like filling up with gas. It fills up pretty quick. Yeah, you don't have to.
- [86:28]
Mike Elgin
Yes. I was following him the other day and driving my, my, my plug in hybrid Prius and it was funny because, you know, the exhaust, the only exhaust that comes out of these things Is. Is water. And it kind of like sprays the water like a, like, it kind of blasts a little jet of water onto the. Onto the street when he's punching.
- [86:46]
Leo Laporte
Spritzing. It's spritzing as it goes. Exactly, exactly.
- [86:50]
Jeff Jarvis
It's spitzing.
- [86:51]
Mike Elgin
Yeah.
- [86:52]
Leo Laporte
Even better.
- [86:53]
Mike Elgin
Yeah. But, you know, it's, it's problematic. You know, the prices are whack. You know, I mean, Toyota did the right thing. I think they were trying to kickstart a movement toward hydrogen cars, and it didn't seem to take. And so now all these, you know, the prices of hydrogen are going up and people are disgruntled, and it's kind of a problem. And it's really a shame because it's great technology.
- [87:13]
Leo Laporte
Are you. Are you destroying your spit at 23andMe, or are you going to.
- [87:19]
Jeff Jarvis
I'm dying to know what you were going to do.
- [87:21]
Leo Laporte
So, Steve Gibbs. Yes. As. As you probably know, they filed for bankruptcies. Ann Wojcicki has stepped down the CEO.
- [87:29]
Jeff Jarvis
It's very sad.
- [87:30]
Leo Laporte
It is sad. It's the end of, I thought, a very interesting experiment making it fairly affordable for people to. It wasn't a full genome, but do a little DNA testing and get a little information. I did it early on. They've had my spit ever since. 15 million people have done it. And that means the DNA of those people is for sale in the 23andMe bankruptcy. You can go there and delete it. And Steve Gibson showed us how yesterday. He even created a shortcut grc SC byebye that you can go directly, but you have to be logged in first. Go directly. 23 me. It's not so hard to find. Find the place where you delete it. It's very slow, though. I think a lot of people are.
- [88:23]
Jeff Jarvis
So what's your view?
- [88:25]
Leo Laporte
It's too much trouble.
- [88:26]
Jeff Jarvis
Yeah, I think I agree, too. What are they going to do with my DNA? Make it make another mess of a human being.
- [88:32]
Leo Laporte
Although somebody pointed out it's not just you because your kids, family. Yeah.
- [88:37]
Jeff Jarvis
I said this to Esther Dyson once when I said what I. You know, prostate cancer and it's hereditary. And. And was I involving my son in my DNA and she said, oh, get over yourself, Jeff. Everybody gets it.
- [88:51]
Leo Laporte
That's true. Every male gets it. Eventually.
- [88:54]
Jeff Jarvis
Yeah.
- [88:56]
Leo Laporte
The key word is eventually.
- [88:58]
Jeff Jarvis
Eventually, yes.
- [88:59]
Leo Laporte
If you live long enough, you're going to get it. So she participated in something called the Personal Genome Project. This was George Church's idea of getting people to donate. And it wasn't free. In fact, I think it was like $10,000 when they started their genome. And they said very clearly it's not private. The PGP approach is to invite willing participants to publicly share their personal data for the greater good. So if you did this, Esther Dyson famously did it and then.
- [89:31]
Jeff Jarvis
Right.
- [89:32]
Leo Laporte
She won. Oh, she was. And I wanted to. They actually didn't accept me. I applied. They would then after they get your genome, do a whole phenotype questionnaire. So you would say, well, I have prostate cancer or whatever, so that they could try to match those up and medical research could do it. They've been doing this since 2005. Funnily, I did eventually get it done because George Church started a commercial company, the Nebula Genomics, to do the same thing. And it was only like $1,000. It's come down a lot and it's not public. So I have my full genome. I don't know. I. What are you. What?
- [90:11]
Jeff Jarvis
Yeah.
- [90:11]
Leo Laporte
What are they going to do? Build. Build a Jeff Jarvis homunculus. Yeah. I don't know.
- [90:21]
Mike Elgin
I mean, you know, eventually that, you know, if nefarious actors will be able to just pick up, you know, a hair that fell out of your head.
- [90:31]
Leo Laporte
If they wanted it. You saw Gataka, the guy had to scrub off all the dead skin before.
- [90:36]
Mike Elgin
Wasn't it Gattaca.
- [90:37]
Leo Laporte
Gattaca, whatever. Same thing.
- [90:40]
Mike Elgin
Great.
- [90:42]
Leo Laporte
It was the.
- [90:43]
Mike Elgin
Yes.
- [90:43]
Leo Laporte
DNA.
- [90:44]
Mike Elgin
Right. They would, they had vacuum cleaners that would vacuum up the DNA.
- [90:48]
Leo Laporte
Right.
- [90:49]
Mike Elgin
To make sure that you were like a genetically modified person or whatever that's selected. Yeah.
- [90:55]
E
I think. Are you companies, the Spears insurance companies getting that stuff and then, you know.
- [90:59]
Leo Laporte
Being like, I don't have insurance anymore. I'm old.
- [91:03]
Jeff Jarvis
Yeah. I'll be, I'll be gone by the time they figure it out.
- [91:06]
Leo Laporte
I'm old. I don't need insurance. If I die, I'll be dead.
- [91:12]
E
But that's I think what the people fear. That's the thing.
- [91:14]
Leo Laporte
Yeah, yeah, I understand. Oh yeah. If you had life insurance and you were a younger person and you needed it, that's.
- [91:19]
Mike Elgin
Well, maybe like Ray Kurzweil, you may just live long enough to live forever.
- [91:23]
Leo Laporte
I'm hoping so.
- [91:25]
Mike Elgin
You may regret this in like 200 years.
- [91:28]
Leo Laporte
Were you thrilled, Jeff, when you saw that the Supreme Court declined to overturn Sullivan?
- [91:35]
Jeff Jarvis
Yes. And I'm listening to an excellent book right now. System needs update. Do it after 2:00am Stop. Stop.
- [91:43]
Leo Laporte
Geez, I hate it. On Windows Weekly today we had a long term nightmare because one of the hosts, Richard Campbell Machine, really wanted to update and it just wouldn't take no for an answer.
- [91:56]
E
And he was at a Microsoft event, right?
- [91:58]
Leo Laporte
He was at Microsoft's campus. He was on this huge bandwidth pipe. He said, finally, he said, all right, all right, all right. I'm going to have to update. I'll be back in a few minutes. Came back like 10 minutes later. It failed. Update failed. It still wants to update.
- [92:13]
Jeff Jarvis
That's not Microsoft. So I'm reading David Enrich's Murder the Truth, Fear, the First Amendment and the Secret Campaign to Protect the Powerful. It's excellent about the First Amendment and Sullivan and all these. I'd live through, again, the poor Mike Masnik story and live through the Gawker debacle in his telling. But it's really, really good. And so, yes, it makes me all the more protective of Sullivan. And so the Supreme Court not hearing at least one challenge so far to Sullivan I think is critically important.
- [92:52]
Leo Laporte
Steve Wynn, casino owner and Trump donor, sued the Associated Press in 2018 after it published a story on sexual misconduct allegations against him from the 70s. He lost. He appealed to the Supreme Court. After Nevada's top court dismissed the lawsuit, Supreme Court declined to hear it, which effectively gives the Nevada court decision permanence. So, you know, the issue, of course, was that a journal can publish stuff it believes to be true. If it doesn't do it with malice forethought, they have the right to do it. They're protected by the First Amendment, especially if the person is a famous person.
- [93:41]
Mike Elgin
Whether they believe, whether they believe it to be true or not, they're published. They can publish allegations that were in fact made because it's true that people made allegations. And that's something that can be reported and should be able to be reported.
- [93:54]
Leo Laporte
Were you surprised that the Supreme Court.
- [93:57]
Jeff Jarvis
My fear is that the Wynn case just wasn't a very good case. And they're waiting for.
- [94:02]
Leo Laporte
Well, that's always a.
- [94:03]
Jeff Jarvis
Because you got. I mean, Trump and Musk have both screamed about Sullivan and they want to get rid of it. But not that. That's an. I mean, he'll sign an executive order, I'm sure, trying to get rid of it tomorrow. That's not how it works, Don. But they're gunning for it.
- [94:19]
Leo Laporte
You know what? Another thing I learned from the, from Gary Rivlin's book AI Valley, that Reid Hoffman financed EG Carroll's lawsuit against.
- [94:30]
Mike Elgin
Oh, I didn't know that.
- [94:31]
Leo Laporte
I didn't know that.
- [94:32]
Jeff Jarvis
On the good. He was a good guy. Teal.
- [94:34]
Leo Laporte
Because Peter Thiel financed Hulk Hogan's lawsuit Against Gawker and put Gawker out of business. And I didn't know that Hoffman financed G. Carroll's lawsuit.
- [94:47]
Jeff Jarvis
Yeah.
- [94:49]
Mike Elgin
What a world when these things are decided by who has the. The best billionaire sponsor.
- [94:54]
Leo Laporte
Exactly. There's something wrong with that. There really is something wrong with that. All right, another break. You're watching intelligent machines. Paris replaced by Mike Elgin just for two. Well, for this week. I don't know Benito, who's coming next week, because I think Paris is gone for two weeks. Do we have that spot filled?
- [95:12]
E
Not yet. Confirmed.
- [95:13]
Leo Laporte
Okay, well, you can bring Mike back if you want. I like him.
- [95:16]
E
If you would like to.
- [95:17]
Mike Elgin
Thank you.
- [95:17]
Leo Laporte
Only if he wants to. He might be busy. He's got to go to Sicily.
- [95:22]
Mike Elgin
Yeah, somebody's got to do it. Leo, that wine is not going to drink itself.
- [95:27]
Leo Laporte
You can't drive a hydrogen car to Sicily, though. I can tell you that right now.
- [95:31]
Jeff Jarvis
What's the best. What's the single best food item you can't wait for? For the gastronomat in Sicily.
- [95:37]
Leo Laporte
Right?
- [95:38]
Jeff Jarvis
I know it's hard. I know it's hard.
- [95:39]
Mike Elgin
You know. You know, I think it is. So I. I'm. I'm a Leo. Leo apparently remembers that I'm a Sicily pizza freak. And I think.
- [95:50]
Leo Laporte
No, but where it was invented.
- [95:52]
Mike Elgin
Well, I think Naples is better associated with.
- [95:55]
Leo Laporte
Naples is not in Sicily.
- [95:57]
Mike Elgin
No, no. Naples is on the mainland. But I. I tend to think that Sicilian pizza is the best pizza in Europe, for sure. And. But. But yeah, the. The thing. The thing that I. I'm not gonna. To answer your question. It's also good in this.
- [96:14]
Jeff Jarvis
I know that. I had a hard question. Mike, you gotta have a hard answer. Come on.
- [96:18]
Mike Elgin
So what. My favorite thing to eat in Sicily is to have a little plate. And you get it when you're wine tasting, when you. Whenever you're doing different things. A little plate with a piece of cheese and a couple of olives and a little sort of like, you know, tomato rice thing. And there's. They do these, like, kind of three or four little tapas on a single plate. That's just a snack. That stuff freaks me out. It's so good. You can't believe it. Especially if you're wine tasting and you're drinking it with wine with different wines. It's heaven on earth. And I absolutely love it.
- [96:48]
Jeff Jarvis
My mouth is watering.
- [96:49]
Mike Elgin
Yeah, it's amazing. And we. We really focus on Etna, the volcano. And wine denomination on Etna is basically halfway up the volcano, starting at the very northern point, going around the east side all the way to the southern point. It's a big backward sea, and that's at a thousand something feet elevation up the volcano. And the volcanic soil is the magic that makes eastern Sicilian food and wine so amazing. The soil is so amazing. And so this is where I would be all my time in Sicily, would be in and around the volcano.
- [97:27]
E
Mike, I need a video component for all of the. You need a. Need video.
- [97:32]
Leo Laporte
Video producer. You gotta have a video producer.
- [97:34]
Mike Elgin
Yes.
- [97:35]
Leo Laporte
You really need that, don't you?
- [97:36]
Mike Elgin
Yes. And then, you know, a couple podcast buddies.
- [97:39]
Leo Laporte
Couple of podcasters.
- [97:40]
Mike Elgin
Yeah, yeah.
- [97:41]
Leo Laporte
Actually, you can go on these Adventures with Mike Gastronoma.net, i'm sure Sicily's sold out, but there'll be more. And it is every bit as wonderful as it just sounded, right?
- [97:52]
Mike Elgin
Yeah, absolutely. The main one has been sold out for a year, but we have another one. We add it by popular demand later in the year. If you want to join us, we have availability and you won't regret it. It's really amazing. Really amazing.
- [98:10]
Leo Laporte
Of course. Jeff Jarvis is here. So nice to have you, Jeff. The author of the Web We Weave, now available in paperback. No, no. Gutenberg, parentheses, Good Burgers. All right. Webweave someday will be available in paperback.
- [98:24]
Jeff Jarvis
No, actually, it won't.
- [98:25]
Leo Laporte
It won't?
- [98:26]
Jeff Jarvis
No, because the publisher just kind of dropped it as soon as it came out, so it's. It wasn't heard.
- [98:32]
Leo Laporte
It's a good book.
- [98:33]
Jeff Jarvis
Thanks.
- [98:34]
Leo Laporte
Right here.
- [98:34]
Jeff Jarvis
And it's blurbed by none other than Leo laporte.
- [98:37]
Leo Laporte
I really enjoyed it. Oh, well. Oh, books, you know, who reads books anymore? Anyway, after all, our show today brought to you. I do, I read them. I don't have enough time to read them. I wish I had more time. I, I. There's so much I would like to read, but let's move on. This episode of Intelligent Machines brought to you by Threat Locker. Oh, I do like these guys. I think this is a very good solution to the biggest problems in security right now. Ransomware. It's harming businesses worldwide. They do it with phishing emails. They do it with infected downloads, malicious websites, RDP exploits. We had a story a couple of weeks ago in Security now, so bad guys got into the network. This is the problem, of course, is that many defenses you get into the network and now you're free because you can go anywhere, do anything. So this company, they thought they were smart. They had all these defenses on the different applications and stuff. Bad guys not thwarted. They still were able to move laterally. They found a security camera running Linux that had enough memory and CPU to put the ransomware germ on to infect the whole network. Yeah, it makes you think it's hopeless, but it's not. You need Threat Locker. Their zero trust platform solves this by taking a proactive deny by default approach. It blocks every unauthorized action. Deny by default is the key protecting you from both known and unknown threats. It doesn't matter if it's a zero day because the assumption is can't do it, not allowed. It's trusted by Global Enterprises. The JetBlue uses it, you know, use it the port of Vancouver. These ports are mission critical, right for the country as a whole and they cannot be brought down by ransomware. We've heard about ports being brought down by ransomware and it's really bad. Not Port of Vancouver. They use Threat Locker. It shields you from zero day exploits, supply chain attacks. And you'll love it because it gives you a complete audit trail for compliance so you know exactly who's using what when. Threat Locker's innovative ring fencing technology isolates critical applications such from weaponization. It stops ransomware cold limits lateral movement within your network, keeps those bad guys out of the things they're, you know, going to use to attack you. And it works everywhere in all industries. It supports Macs as well as PCs provides. They have great 24. 7 US based support. You get comprehensive visibility and control. Mark Tolson, who's IT director for the city of Champaign, I'm really happy to see more schools, governments, ports, places really critical, mission critical places. Use Threat Locker. He says quote Threat Locker provides that extra key to block anomalies that nothing else can do. If bad actors got in and tried to execute something, I take comfort in knowing that Threat Locker will stop that. Stop worrying about cyber threats. Get unprecedented protection quickly, easily cost effectively. With Threat locker go to threatlocker.com twit you can get a free 30 day trial. Learn more about how ThreatLocker can help mitigate unknown threats. Brand new zero day threats and ensure compliance. Threatlocker.com twit we really appreciate their support. They've been a great, great partner. Thank you. Threat locker threatlocker.com TWIT use that URL so they know you saw it here. Why Apple Meta and Google are buying remote controlled robot arms. Oh no, this is from the information at least.
- [102:31]
Jeff Jarvis
They don't have legs.
- [102:33]
Leo Laporte
No, they have wheels. I love the name of the author on this and I'm sure a colleague of Paris is Rocket Drew. We talked last week About Jensen Huang's speech at GTC and that little cute little robot he was all in on. Computer simulations that teach robots to do things like wash dishes, pick up socks and so forth. But Rocketrew writes, some robot makers I spoke to say it's better to train robots to do such tasks by remote control. Have a person control them remotely, teleoperation it's called. And then directing an AI model to imitate those actions. A new way of training basically an AI to do something in the physical world. It's similar to how language models can copy the way humans write poetry or code, but for physical actions. I thought that was interesting because we were just talking about new models for training.
- [103:31]
Jeff Jarvis
Well, that's what Long argued strenuously is that that's the next phase. That's where he thinks the huge business is going to be. And that's the essence of giving it reality.
- [103:42]
Leo Laporte
Right.
- [103:42]
Jeff Jarvis
It has to, it has to grapple with the egg falling off the table.
- [103:46]
Mike Elgin
And there's all kinds of innovation around that concept where there's one set of researchers who are, they literally have a person next to the robot who's trying to perform a task and the person who's helping the robot learn actually physically takes the hand and the arm and says grab it here and put it there.
- [104:04]
Leo Laporte
Just like a parent would do with a child in a way.
- [104:06]
Mike Elgin
Exactly. So that's one variation. Nvidia is very much into the digital twin idea of having physical AI, which is a VR environment that has physics, which is gravity, inertia, all that kind of stuff. And where virtual robots can train and train and train and train. And then once they have that training software data set, they can just download that into a real robot and it should be well on its way to knowing how to do a task. But you can do that at much higher speed than basically a tele operated, you know, robot can. So that it's getting more advanced already. They're working on more advanced ways to do it than teller operation.
- [104:42]
Leo Laporte
Very interesting.
- [104:43]
Mike Elgin
Yeah, very. And, and of course Apple is interesting about Apple is they will, according to, you guessed it, Mark Gurman, they've been working on this robot arm. So basically imagine an iPad at the end of a robot arm sits on your desk and it has gestures and it's like it shrugs it non existent shoulders and it, you know, looks sad when you, you know, leave and all that kind of stuff. And so they're working on sort of a human. It's not a humanoid robot, but it's a non humanoid robot arm that conveys gestures and emotions.
- [105:16]
Jeff Jarvis
To what end though?
- [105:17]
Mike Elgin
What to. To make people addicted to Apple products.
- [105:20]
Leo Laporte
I mean, which is what it's all about, baby.
- [105:24]
Mike Elgin
Exactly. It's by flattering, basically. You know, why do people like dogs? You. You have a dog. It's great to come home after a long day and you open the door and the do freaking out that you're home. It's like the greatest thing that ever happened. Well, imagine if your devices did that. You'd like that. You'd like your iPad as much as you. Or your, you know, your, your Apple home pod as much as you like your dog. And that's what Apple wants. Well, they're hoping.
- [105:48]
Jeff Jarvis
I don't know.
- [105:50]
Leo Laporte
Well, Apple's delayed it, so don't, you know, it's not going to be anytime soon.
- [105:54]
Mike Elgin
That's right. But you can get a dog now.
- [105:55]
Jeff Jarvis
Get a dog in the meantime.
- [105:56]
Mike Elgin
Yeah, exactly. That dog is useless for doing FaceTime calls. But still, really, I've been seeing these.
- [106:02]
Leo Laporte
Images all over the Internet. OpenAI, it was interesting what Gary said, that sometimes these Companies, I know OpenAI has done this, will hold back innovations because they can't afford. They know they'll be so popular that they can't afford to put them out in public. This may well be an example. OpenAI has unveiled a new image generator for chat GPT4O that you. So you describe the prompt. I mean, it's not something we haven't seen before, but it's pretty darn good. Look at this comic book generated by AI. You describe a four panel comic strip in text, including the characters, what they're saying to one another, and then it generates a cartoon. I've seen a lot of. This is Cade Metz's piece from the New York Times, but I've seen a lot of images on the socials with people using this. It's kind of interesting. Darren Okey, who's one of our. By the way, it's interesting how the New York Times has put a giant big black generated by AI on all of these images so that nobody would accidentally imagine that the triangle wheeled bicycle is real in any way. Darren Okey is one of our regulars in the club, has been using this new ChatGPT to make small changes to images. So it's useful for retouching as well. So interesting. Interesting. I have not played with it yet, either of you.
- [107:32]
Jeff Jarvis
I tried to, but it was too busy today.
- [107:35]
Leo Laporte
Yeah, see, that's the problem, right? They're gonna have to artificially gate your point. The speed these things work at.
- [107:41]
Mike Elgin
I Don't have the patience for text to image because you go there and it's like, oh, it's free. And then it's sign up and then you like. It's. It's not really what they say it was and I can't, you know, so it's, it's. I don't think we're quite there yet for really impatient people.
- [107:57]
Jeff Jarvis
I guess you could never be Michelangelo there, Mike.
- [108:01]
Mike Elgin
Right. No. Right.
- [108:02]
Leo Laporte
You'd have to be able to paint a little tiny picture and then paint some more. So I, I guess Sam Altman is kind of not stepping down, but stepping aside a little bit.
- [108:16]
Jeff Jarvis
It's hard to do the criminology of this company.
- [108:19]
Leo Laporte
It is. Chief operating officer at OpenAI. Brad Lightcap is now going to be the CEO, giving former CEO Sam Alfman, quote, more time to focus on research and products.
- [108:35]
Jeff Jarvis
I didn't get to that part of Gary's book about the couple.
- [108:42]
Leo Laporte
Yeah. I mean, it was Steve with.
- [108:44]
Jeff Jarvis
Any other insights into Altman and his management, shall we say?
- [108:49]
Leo Laporte
No. No. Well, I don't know. I'm trying to remember. I may not have gotten that far either, come to think of it. I skimmed ahead a little bit. I don't remember anything in particular, but it was because we covered it so intensely.
- [109:02]
Jeff Jarvis
Yeah, we did.
- [109:03]
Leo Laporte
Every move. I don't know. I'll have to. I can't answer that question. I'm sure there's some insight in there. Sam's not going away. No, but it's interesting.
- [109:18]
Mike Elgin
It's like a demotion. It feels like a face saving demotion, doesn't it?
- [109:22]
Jeff Jarvis
Or it feels like Sam's going to do 20 things. It's kind of a musk move.
- [109:26]
Leo Laporte
One of the problems that the board had with Sam, the board which has since been fired for firing him, was that Sam had other bets.
- [109:35]
Jeff Jarvis
Right.
- [109:36]
Leo Laporte
And I'm wondering how much those other bets are taking up of his attention. I don't know. He's investing in other companies. Some of them may be even competitive to OpenAI. Sam also, I think maybe oversold a little bit some of what OpenAI could do.
- [109:57]
Mike Elgin
Yeah.
- [109:59]
Leo Laporte
Let me.
- [110:00]
Mike Elgin
He's a controversial figure. There's also controversial, weird allegations about his past. And.
- [110:05]
Leo Laporte
Well, those were. I, I don't credit those too much.
- [110:09]
Mike Elgin
Well, I mean, but the thing is that, I mean, I, I don't mean to spread rumors. I have no information that he's done anything wrong ever. But. But you never know those kinds of things behind the scenes. If you know some PR disasters coming, you kind of.
- [110:22]
Leo Laporte
Right But.
- [110:22]
Mike Elgin
But again, he's still with the company, so it could be just that he wants to work from home.
- [110:26]
Jeff Jarvis
What's LifeCaps background? You know, because. Because Sam's not a coder.
- [110:30]
Mike Elgin
He co founded Looped. He co founded Looped, which was Sam. I'm sorry, that was. That was Sam Altman. I'm sorry. No, no, that's not.
- [110:36]
Jeff Jarvis
I'm talking about like.
- [110:38]
Mike Elgin
Right. He was. He was. He's been a money guy for OpenAI money guys.
- [110:43]
Leo Laporte
Yeah, that's really the job. That was especially now when they.
- [110:46]
Jeff Jarvis
When they redo the company and the cap table and.
- [110:49]
Mike Elgin
Right.
- [110:50]
Jeff Jarvis
The structure.
- [110:51]
Leo Laporte
Yep, they did. I did read the part where Elon pulled back from OpenAI. Yeah. Which I thought was kind of interesting. You know, the picture you get of Elon Musk in the book is mixed. It is assumed at every stage of the game that nobody wanted to work for Elon. Nobody at PayPal wanted him to be the CEO. They pushed him aside as soon as they could. It says Altman, this is from the book, had no desire to work for Musk, a famously mercurial boss. He imagined most of the people he recruited. OpenAI wouldn't want to either. So when Musk said, I will continue to fund you, but only if you let me be the boss. So Altman rejected the offer. Musk walked away from the company. This is what, roughly 2016, I think. Leaving Altman to worry about covering salaries and other expenses. He went to Reid Hoffman, who wrote a check for $10 million. I. I don't know of any other. Well, there to talk a little bit about. Satya. Nadella getting very upset to hear that Altman had been pushed out and making that call. But again, that wasn't something that we were surprised about. Right. I mean, we kind of. Kind of knew that happened. I'm just scanning through it. There's a lot about Sam Altman in here. Yeah. I don't know. Good question. Read the book. And would you come to read the book?
- [112:30]
Jeff Jarvis
I'm going to read the book.
- [112:31]
Leo Laporte
It's really good. I couldn't put it down. I was up way late. Let's see what else you got. Some stories for us.
- [112:39]
Jeff Jarvis
Jeff, you always lots of them here.
- [112:44]
Leo Laporte
Let me see.
- [112:47]
Jeff Jarvis
So a little bit of change log. Google brings smarter AI based search to Gmail. We'll see how that goes. And then Gemini can now answer your questions on Google Maps. They're trying their best to show that they have AI everywhere. That's one question I didn't ask Gary. I wanted to ask him about that those early days, whether, because as he said, Google has was in AI long, long ago. And was it pretty much a PR disaster that they just didn't brag about it enough?
- [113:16]
Leo Laporte
Right. Well, and so was Meta. One of the things I found very interesting in the book was both Meta and Google used AI basically to make more money. Google used it for advertising and Meta used it well, they did some moderation with it, but mostly they used it to make more money. They weren't trying to change the world. They saw it as a way of saving money on employees.
- [113:44]
Mike Elgin
But this addition to Gmail and Maps is really interesting and it's part of what I think is a pretty positive trend of using AI to pre filter things or to give you whatever information it can just around the edges. So for example, in Maps, what you do is you can go to Maps and you can click on a convenience store and you, you can say, do they, do they have Gatorade at this store? And if that information is knowable, it might tell you, yeah, they have it, you know, they have three different colors or whatever, that sort of thing. And with Gmail it's, you know, it's basically helping you with your inbox. And you also see it with Wise. For example, Wise has a new feature I think was announced today or yesterday where you get all these notifications and the, and the AI theoretically does something very smart. Let's say, for example, you've got a ring, you've got a ring doorbell type of thing from Wise, a video doorbell product, and in front of your house, a bunch of kids. This is the example that that Wise gives. Exactly right. It'll say that to them for you so you don't have to. No, it doesn't do that. No. But if, if the kids are, you know, playing baseball on the street in front of your house, it will see that this continues to happen. And it's like, you know what, we're not going to keep just giving you endless notifications of movement. We're going to ignore that movement. We see the kids, we know who they are, we know what they look like, and so everything they do, we're going to ignore it. And, but it will, it will capture the video and tuck it away without notifying you. So you can go back and say, oh yeah, these are the kids that I need to tell to get off my lawn. So it's, you know, I see this kind of use of AI in a very smart way to make things a little better and little more convenient. And I like it.
- [115:27]
Leo Laporte
The Other Reid, Reed Hastings, the founder of Netflix, has just donated $50 million to Bowdoin to the university to launch the Hastings initiative for AI and humanity. It's Bowdoin's largest gift ever, $50 million. They're going to hire 10 new faculty members in a range of disciplines, support current faculty incorporating and interrogating AI in their teaching, research and artistic work. So it doesn't sound like it's a, it's researching AI, it's how to use AI in terrible systems.
- [116:04]
Jeff Jarvis
So I'm working with both Stony Brook and, and I've had some time with Montclair State, but especially Stonebrook on new, new degree programs around this about technology, AI and society. And my contention has been that we have to bring in the other disciplines, we have to bring in, in humanities and social sciences and the arts, and that's what this is doing. So I think it's great.
- [116:23]
Leo Laporte
The donation, Reid says, seeks to advance Bowdoin's mission of cultivating wisdom for the common good by deepening the college's engagement with one of humanity's most transformative developments, artificial intelligence. He went there, to school there, so it makes sense that he might, as.
- [116:41]
Jeff Jarvis
The president, President Zaki says the AI revolution makes the liberal arts and Bowdoin education more essential to society. I wish that many would recognize that.
- [116:50]
Mike Elgin
And they also have an ethical component where they're, they're looking to work on data privacy, they're going to work on bias, they're going to work on, you know, using AI for ethical inquiry and basically use it as a kind of philosophical tool to work through the ethical implications of fast moving changes in society and stuff like that. So nice. And all for the low, low price of $50 million.
- [117:17]
Leo Laporte
50 million, which probably to Reed Hastings is like, you know, lunch money.
- [117:24]
Jeff Jarvis
So it's a bunch of red envelopes.
- [117:27]
Leo Laporte
Yeah. Doctors told him he was going to die. This is the, I don't know if this is a story Gary was talking about. Then the AI saved his life. Scientists are using machine learning to find new treatments among thousands of old medicines. Kate Morgan writing in the New York Times. He was battling a rare blood disorder, poems syndrome, which had left him with numb hands and feet and enlarged heart and failing kidneys. He became too sick to receive a stem cell transplant. I gave up, he said. I just thought the end was inevitable. But they found a doctor who they.
- [118:06]
Jeff Jarvis
Let's give credit. His girlfriend, Tara Theobold, wasn't ready to quit.
- [118:11]
Leo Laporte
No. She emailed a doctor named David Fagenbaum, whom they met at a rare Disease summit. But he suggested an unconventional combination of chemotherapy, immunotherapy and steroids untested as a treatment for the disorder. He was responding four months. He was healthy enough for stem cell transplant. Today he's alive and in remission. But the most important point is the life saving drug regimen wasn't thought up by the doctor. It wasn't thought up by a person. It was created by an artificial intelligence model.
- [118:46]
Jeff Jarvis
This is what Mike was saying earlier. This is the value, this is where it lasts. Yeah, it's the companion.
- [118:53]
Leo Laporte
And in a way this is, I think, what Stephen Wolfram was saying that the value of AI may be that it sees patterns that we don't. I mean, that's what it's really, that's what an LLM really is doing, is detecting patterns and making a note of them and then regurgitating them if it can see patterns that we don't see. And in this case a combination of medicines that doctors hadn't really thought of.
- [119:19]
Mike Elgin
And, and this is, this to me is the underappreciated way to categorize how AI can be really useful right now. It's certainly useful to me as journalists and so on, which is like, what are my blind spots? What am I missing? These questions are, AI does a great job because it's not like, well, here's the answer. It'll give you ideas and you can consider those ideas. And it could do it very quickly. And so I don't know if there's really a connection between the two, but basically people are amazing. People can do all kinds of amazing things. Partnering with AI, we can do so with fewer blind spots, fewer biases potentially, and with more awareness. I mean, obviously it's taking the information that's out there in the medical databases about these different individual components and saying, well, what if we combine them? And then a person, a doctor said, hey, this is doable, this won't kill him, et cetera. And so, yeah, it's pretty amazing. And I think that's the best use for AI. What am I missing?
- [120:24]
Leo Laporte
Yeah, this is something that actually Dr. Feigenboom had been working on his whole life. In fact, he saved his own life by applying a disease that was intended for kidney transplant recipients to reduce rejection to cure his Castleman's disease. And it worked. No one had ever used it before. So he created a lab. It became a doctor created a lab to do this. But it was very slow work with just humans going, well, what could we use? So in 2022, he established a nonprofit called EveryCure, aimed at using machine learning to compare thousands of drugs and diseases all at once. And it would suggest these things that worked. They had a 19 year old patient who was debilitated by chronic vomiting. He couldn't stop. They ran a query on the AI said, show us every proposed treatment there has ever been in the history of medicine for nausea using isopropyl alcohol inhaled through the nose. Not recommending this, we are not physicians. But the alcohol popped to the top of the list and it worked instantly. I think that's really, really interesting. They compare. At the University of Pennsylvania, they. Dr. Feigenboom's platform compares 4,000 drugs against 18,500 diseases, scoring them based on the likelihood of efficacy. And then, I mean, I guess, I mean, I guess if you're dying anyway, you're willing to try. Oh yeah, anything. And this was a perfect example.
- [122:01]
Jeff Jarvis
And you know, the worst case is you've benefited science and you found out what doesn't work. Or find out more about it.
- [122:07]
Leo Laporte
Right. Ted has been funded by more than $100 million by commitments. I'm sorry? Every cure has been funded by more than $100 million in commitments last year from Ted's Audacious Project and the Advanced Research Projects Agency for Health within the Health Department. I imagine that funding might be in jeopardy. Yep, that's the kind of thing you lose. Really interesting story. I hope we don't hear more stories about that kind of of thing and how they've been defunded and are no longer available. What else should we talk about? I'll tell you what. You find some stories, you guys, I will do my job. Which is to tell you about our another sponsor. How about we're very lucky to have them. We should, we should be grateful. Our show today, brought to you by a product I'm very familiar with. In fact, let me show you. I have it right here. I've been talking about it for such a long time. But it was buried away in a corner of my office. So I decided to bring it out into the public. Let me see if I can show you an image. This, my friends, is the thin canary. The canary's upside down. But what's cool about this box, which is about the size of a external hard drive, you see, it's connected to my network and connected to power. But this little baby, this here is a honeypot. A honeypot that can be deployed in minutes, can impersonate almost anything and let you know if a bad guy is in your network. We always talk about security being Layered. One of the layers it's missing in many cases is the layer that tells you somebody is actually inside the network. And maybe you want to do something about this. It's. It's often the case that bad guys get in and are not detected for ages. On average, 91 days before they're stopped. That's 91 days, three months. They get to wander around your network, do anything they want, but not if you have things Canaries. You can do two things with a Thinks Canary. You can, of course, set it up and make it be anything you want. Actually, I've reassigned mine. Last time we talked, it was a Synology NAS. Now it's a Windows Server 2019 Office file share and it has a Microsoft Mac address. I can see what's going on. It's easy to configure it to be any number of things. It could even be a SCADA Device, a Cisco VoIP phone. It can be a Palo Alto firewall. But it's not really. The thing is, these don't look vulnerable on your network. They look valuable. They're designed to be something bad guys want to attack. You can also create Canary tokens, which is awesome. Create files that look like real things but aren't. Word documents, spreadsheets. You throw these little files around on your network, leave them there for a bad guy. I have, for instance, an Excel spreadsheet called Employee Information sitting on one of my drives. Should somebody open it. Same thing is going to happen as if they try to attack this Windows server. I will get a notification immediately. And by the way, in the way I want it. You can choose to have it by email, text, Slack, webhooks, API, any syslog. It'll do any way you want. But the point is, you only get the alerts that matter. You never get an alert. That's a false positive. You just choose a profile for your ThinkScanary device. You register with the hosted console for monitoring and notifications, and then you just fold your arms, you sit back and you wait. An attacker who's breached your network, a malicious insider, other adversaries, they can't resist. They will make themselves known just by accessing the things Canary. Now I have one here in my little office. Medium, small to medium sized businesses might have five or more. A big bank might have hundreds. But I'll give you a pricing example. You go to Canary tools twit. For 7,500 bucks a year, you'd get five of these. You spread them out all over your network, different nooks and crannies you know, places hackers might wander over into. You also get your own hosted console. You get upgrades, you get support, you get maintenance. And if you use the code Twit in the how did you hear about us box, you'll get 10% off the price. Not just for that first year, for life. Here's the good news. If you'd all going like, well, I don't know. You can always return your thinks Canary. They have a two month money back guarantee, 60 days for a full refund. Get it, try it. See how it works. Because you can get your money back within 60 days, no problem. But I have to say, during all the years that TWIT has partnered with Things Canary, we've been doing these ads for more than eight years. Their refund guarantee has never been claimed, not once. Visit Canary Tools Twit. Don't forget that offer. Go Twit in the how did you hear about us box for 10% off for life. Thank you. Thanks for supporting intelligent machines and for creating such a cool device. Thinkst Canary. Canary Tools Twit. Thank you so much for your support of intelligent machines. Jeff Jarvis, Mike Elgin. We're getting close to the end of the show. Oh, I should mention Paris's article.
- [127:50]
Jeff Jarvis
Yes.
- [127:51]
Leo Laporte
Because she was working on it before she left for Europe. Did a great job. She's Talking about Section 230 came out this weekend. Section 230 may finally get changed. I've mentioned this on the other shows. This is a call to action as lawmakers prep new bill. It is so disappointing to me because it's both sides. Dick Durk.
- [128:14]
Jeff Jarvis
They're gonna go for bipartisanship. This is what they go after.
- [128:19]
Leo Laporte
So for those who don't know section 230 is part of the Communications Decency act. Some call it the 36 words. 26, 26.
- [128:28]
Jeff Jarvis
I always get Jeff Kossuth.
- [128:30]
Leo Laporte
It's a small number. A small number of words that made the Internet possible because it said that just because you ran a website, whether it's a social network or a forum or a blog with comments, you aren't liable. You aren't the publisher of comments your users might put there. Which gives you two things. One, one, the right to moderate it because you can't be sued for taking it down. But two, somebody could post something up there without your knowledge. It does. It means you don't have to go to court to defend it. And very and while it often is the case people sue, judges universally will say nope. Section 230 case dismissed. Throwing it out would be The, I'll be frank. I mean, you know, the problem is these senators, these Congress people, Congress critters, as Cory Doctorow calls them, think it's all about Google and Microsoft and Twitter and Meta.
- [129:32]
Jeff Jarvis
It's about you, Leo.
- [129:33]
Leo Laporte
Those guys can afford to defend themselves in court. It's about the small people. There's a fixie site that went offline in England because the English are doing something similar. Little tiny forum. He said, I can't afford to defend myself in court. This could cost a million dollars. Paris says in her article, cost a million dollars to defend it. I can't afford that. We run a mastodon instance. We run forums, Twitter community, our mastodons at Twit Social. We have a chat room going right now with all of the different live streams. Everybody can chat with us. That would all have to go away. We could have no public comment if section 230 goes away. So this bill sponsored by Dick Durbin, they're going to announce it apparently soon. And Senator Lindsey Graham, it would set an Expiration date of January 1, 2027, end of next year. The idea being, oh, by then we'll have written something that's better. Josh Hawley supports it. Marsha Blackburn.
- [130:40]
Jeff Jarvis
Yeah, of course.
- [130:41]
Leo Laporte
Sheldon Whitehouse, Amy Klobuchar, of course. Dick Blumenthal and Peter Welch are considering joining as co sponsors.
- [130:49]
Jeff Jarvis
These are all the anti Internet crew.
- [130:54]
Leo Laporte
How do we convince them? What do we do?
- [130:56]
Mike Elgin
We.
- [130:56]
Leo Laporte
We call them.
- [130:56]
Jeff Jarvis
We send them to Mike Masnick's page. You have been sent here because you're wrong about section 230.
- [131:03]
Mike Elgin
Yeah. And just to be very, very, very clear, they'll use as justification the idea that there's all kinds of noxious speech, disinformation, misinformation, unfair criticism and so on. This is. This has almost nothing to do with that. What this really has to do with.
- [131:23]
Leo Laporte
In fact, it would have the opposite, by the way, because it would be risky to moderate.
- [131:28]
Mike Elgin
Exactly, exactly. But. But is what it really does is is it would. It's an idea that would illegalize. It would basically say you have to have. Be certain, have a certain amount of money to have free speech. The.
- [131:44]
Leo Laporte
The law believes this 100% right.
- [131:48]
Mike Elgin
Of course, because he could. If you said Elon Musk is a bad person. Right. That should be constitutionally. That is constitutionally protected speech. And if you. And he could sue you for that. It has not do with the merits of the case. Just defending yourself is unaffordable.
- [132:03]
Leo Laporte
So it's like, well, worse, he could sue me for you Saying it.
- [132:07]
Mike Elgin
That's exactly right. Yes.
- [132:09]
Jeff Jarvis
Thanks a lot, Mike. See what you just did to Leo? Really good Mike boy.
- [132:14]
Leo Laporte
But that's the problem. Guest, you are meta can defend themselves, Twitter can defend themselves. All the little forums, all the little chat rooms, all the discord rooms you visit.
- [132:25]
Jeff Jarvis
Yep.
- [132:25]
Mike Elgin
Right, right. But like self hosted things. Like why would you have to be crazy to self host anything that has a chat room or you know, even be careful about what you say because it's like, you know, it's, it's basically just, just you would you, basically everybody would be like, oh, I for one welcome our new billionaire overlords. And you'd be very careful about, about criticizing people with money. You have, you know, do it with people with no money. It'd be no problem. But basically, if anybody has any money that can take you to court as a weapon, they can gawker you the way that Peter Thiel did to Gawker.
- [132:59]
Leo Laporte
Yep.
- [133:00]
Mike Elgin
Awful.
- [133:01]
Leo Laporte
So normally I would say ah, it's Congress, nothing will ever happen. But this has such broad bipartisan support. Yeah, this may be the one thing.
- [133:11]
Jeff Jarvis
Is our one best hope against it. But when I testified in the Senate, Blumenthal was there. AI companies shouldn't be allowed to be protected by 230. Just so off the point. But 230 is his boogeyman.
- [133:22]
Mike Elgin
Yeah.
- [133:23]
Leo Laporte
So if this were passed and Trump said Trump's wanted to do this forever so he would sign it, what do I do at the end of December of 2026? I might have to do it immediately because I presume it could be retroactively applied. Like if somebody, if the law passed section 230 goes away at the end of next year, something posted in March of next year I could still, I could be sued for. Right. It doesn't, I mean do we all.
- [133:52]
Jeff Jarvis
Delete all of our social accounts? Forget deleting your DNA, do we delete everything we ever said online?
- [134:00]
Leo Laporte
I'm not worried about you, I'm worried about me as the, as the publisher.
- [134:04]
Mike Elgin
Well, that's the other thing that if you don't like Elon Musk, you can just say really horrible things on Twitter and then he gets sued. Yeah, I mean this, this whole thing.
- [134:11]
Leo Laporte
He doesn't understand this impacts him, doesn't it?
- [134:13]
Mike Elgin
Right. And, and Trump, that's true truth social. You think that's true? There's not 100 million laws be filed there. You just sell it off or close it or something.
- [134:26]
Jeff Jarvis
It's, it's bad and it, and, and my fear among many is that Big institutions will go along too because of this, because they are the ones, as you say, Leo can afford it. And this gets rid of competition.
- [134:39]
Leo Laporte
It's actually good for them because yes, it gets rid of the small.
- [134:41]
Jeff Jarvis
It's regulatory capture by definition. Yep.
- [134:44]
Leo Laporte
I think, I think if this passed, I would have to shut down our Mastodon instance. I would have to turn off our discord.
- [134:53]
E
Could. Could meta actually afford this? Because like, think about the volume and scale of the amount of lawsuits that would be coming at them.
- [134:59]
Jeff Jarvis
It's a good question, but you know, yeah, it's not happy for them. And well, if you go back, they have.
- [135:08]
Leo Laporte
Here's why they can afford it. They have a building, an entire building filled with lawyers that they're paying anyway.
- [135:14]
E
Yeah, but like, think about the scale here. Like for every message you get one lawsuit for every message. We're talking in the billions here. Billions.
- [135:20]
Leo Laporte
But they're not. No, because they're already paying these guys. They just. This is a new thing for the lawyers to give them something to do.
- [135:27]
Jeff Jarvis
Well, what's going to happen, Benito, right, is I think we're going to go. The right wing would have us go back to what existed before 230 is if is you don't moderate anything and then you're not liable for it. That you're liable if you fail at that. And that was the, the Prodigy doctrine that existed before 230. And so I think that all this is in how it's written and what happens, what it goes back to. But it's not just that you can be sued for anything. It's that, well, if you say you're going to moderate and you fail at that, you are now liable. Ergo, you don't moderate. Ergo, it is the Wild west and every noxious thing can be said online. And we get the nox nightmare of the Internet. There's two nightmares here, right? That's one. And the other one is that one is that everything goes and the other one is that nothing goes and nobody can talk. So here's my kind of prefer the, the latter, oddly, because they, they want a cleaned up Internet and the right wants the messiest Internet possible.
- [136:24]
Leo Laporte
Here's my Mastodon instance. I run this, I host this. It's a twit thing. You see 22 pending reports. These are people and this is how I do it because I don't go through all the content. I just let people tell me they found something they don't like. These are all reports. Often it's spam. I don't delete something because you don't disagree with it. So in that case, I just go, yeah, well, no, sorry. Just because you disagree with it, I'm not going to delete it. But if there's spam or I delete nudity and, you know, I want to. There are various things that I choose on my mastodon instance, to live.
- [137:01]
Jeff Jarvis
As is your right in this case.
- [137:02]
Leo Laporte
As is my right.
- [137:04]
Jeff Jarvis
Part of your speech.
- [137:05]
Leo Laporte
And by the way, if you go back to this page, I have a number of appeals of people who I've moderated off who say, no, no, I want back. I have three of them sitting in here. Sorry. I'm exercising my right to create the kind of mastodon I would want to be a part of, and that should be my right to do so.
- [137:28]
Jeff Jarvis
Right.
- [137:28]
Leo Laporte
It's. I mean, I'm.
- [137:30]
Jeff Jarvis
Your speech selection is speech.
- [137:33]
Leo Laporte
Yeah, but I would have to. These 22 reports. So you're saying one defense would be just to ignore all reports from now on?
- [137:43]
Jeff Jarvis
No, I'm saying that that may be the way that they write the.
- [137:46]
Leo Laporte
What do newspapers do? They can be sued, right?
- [137:49]
Jeff Jarvis
Well, but they don't have. No, newspapers are. Yes, that's part of the argument here. What's confusing about this is newspapers are responsible for their content because they, in fact, do edit it. And when you make the analogy and say that Twitter or Facebook or Blue sky is a publisher, well, they aren't, because they don't choose everything. They don't edit everything. They are not, in fact, responsible for everything, but they're being put in that same position. That's what Blumenthal says. Well, if you can sue the New York Times, you should be able to sue Twitter. Twitter didn't create it.
- [138:17]
Mike Elgin
It.
- [138:17]
Leo Laporte
That's the big.
- [138:18]
Jeff Jarvis
Came on and did it.
- [138:19]
Leo Laporte
That's the big difference.
- [138:20]
Jeff Jarvis
Yeah. Yeah. So, yes, indeed, the New York Times is responsible. And. But there, too, they have the benefit of Sullivan, where if they are not malicious, they have a defense. Truth is a defense to them.
- [138:36]
E
Couldn't you make the argument, though, that if like, Twitter didn't say it.
- [138:41]
Leo Laporte
But wait, Twitter published it. Right? That's your position. They allowed it to happen.
- [138:49]
E
Exactly.
- [138:50]
Mike Elgin
My question is if they boost something through their algorithm, is that.
- [138:56]
Jeff Jarvis
You see there, Mike, I learned this lesson decision. I had a meeting once with, I think, Facebook, and we talked about demoting things and they said, whoa, every single thing that appears on your screen is. Is demoted or promoted. It's not like we have this magical promotion thing behind the scenes. It's that there's an algorithm that chooses how to rank things. So every ranking decision is a demotion or promotion by its definition. So when they say, oh, the algorithm and pushes things. Well, yeah, but less than an editor does.
- [139:28]
Leo Laporte
So would it be a defense if I'm aggressively moderating?
- [139:33]
Jeff Jarvis
They're under the old prodigy doctrine. You're in worse shape if you fail because you're warranty.
- [139:39]
Leo Laporte
If I miss one thing, if Benito posts one comment that says, elon Musk is a baw bag, which I, by the way, learned a new Scottish word.
- [139:49]
Mike Elgin
That nobody knows what a baw bag is.
- [139:53]
Jeff Jarvis
Is that a haggis thing?
- [139:55]
Mike Elgin
Yeah. Is that a thing?
- [139:56]
Leo Laporte
We went to a concert on Monday and I'm sitting in front of a guy who's from Scolen, and I don't know how, but Lisa got in a little political conversation with him and no, no, he was fine. He said, yeah, we call him a ball bag in Scotland. And I said, do I want to know what that means? Well, he explained what it means, and I'm sure the Scots among us know what it means. It's not a nice thing, but let's say Benito gets on my mastodon and.
- [140:23]
Jeff Jarvis
Says, elon Musk, Benito's your employee. So that's actually different.
- [140:29]
Leo Laporte
All right, so I won't use you, Benito. Sorry. Some guy, some rando says that I don't know what ball bag means. So I ignore it.
- [140:38]
Jeff Jarvis
Joe Esposito comes in, Joe does it.
- [140:40]
Leo Laporte
It's your fault.
- [140:41]
Jeff Jarvis
Let's blame you, Joe. And Rex doesn't.
- [140:45]
Leo Laporte
Now, I don't moderate. I moderate everything else. I'm assiduous, but I don't understand that word. And Elon goes, I know what that means and I don't like it. I'm suing you. Do I have a defense without section 230?
- [140:56]
Mike Elgin
I would challenge your framing of this. The problem isn't whether you actually fail to moderate something and otherwise.
- [141:04]
Jeff Jarvis
That was the old days, Mike Wright. That's not the case now.
- [141:07]
Leo Laporte
You should go after Joe. Not.
- [141:08]
Jeff Jarvis
That was under protest.
- [141:09]
Mike Elgin
But my point is it doesn't matter if the comment is offending or not. I could say, have a nice day and somebody can sue me for libel. And it's just I have still have to go to court. I still have to fire a lawyer. I still have to spend all this time, right? So it's like it doesn't even matter if. If it's offending someone or not or violating the law or not. Doesn't matter. It's just about, do you have the money? To have free speech or can you not afford free speech?
- [141:37]
Jeff Jarvis
Yeah, Dr. Dew is having a lot of fun in the. The discord.
- [141:42]
Leo Laporte
If you scroll up, it might be a Scotsman.
- [141:45]
Jeff Jarvis
No, it's not about that. It's about us. In 230, if there's two illustrations of us. Benito could get a card out of this if he wants.
- [141:54]
E
Yeah, I've already got.
- [141:55]
Jeff Jarvis
See them.
- [141:56]
Leo Laporte
No, I don't see him.
- [141:57]
Jeff Jarvis
Oh, scroll up. Or down, as you would put it.
- [142:00]
Leo Laporte
This one?
- [142:01]
E
Yeah, I keep.
- [142:02]
Jeff Jarvis
I can't see anything right now. No, no, no.
- [142:07]
Leo Laporte
Join the club, baby.
- [142:09]
E
That one the other way.
- [142:11]
Jeff Jarvis
See, he doesn't know up from down. That's the problem.
- [142:16]
Leo Laporte
Why do I subject myself to this? I hate it. I don't see it. Oh, there we go. There we go.
- [142:21]
Jeff Jarvis
There you go. There's one of them right?
- [142:22]
Leo Laporte
There we go. Yeah, that's Mike Elgin on the left. This is terrible. I say in my colorful shirt, this cannot stand. And you and your goatee say, what an outrage. That's a pretty good illustration.
- [142:33]
Jeff Jarvis
Pretty good. Now if you go up, there's another one.
- [142:35]
Leo Laporte
Oh, is this from. Oh, you know what? This is from the new chat GPT4.
- [142:39]
Jeff Jarvis
Oh, Dr. Do. Is that what. It's from the.
- [142:41]
Mike Elgin
Go up.
- [142:41]
Jeff Jarvis
There's another one up above that.
- [142:43]
Leo Laporte
Yeah, he's doing. This is the new 4O image machine.
- [142:46]
Jeff Jarvis
Complete with our lower third.
- [142:48]
Leo Laporte
Oh, my gosh.
- [142:49]
Mike Elgin
Wow.
- [142:50]
Jeff Jarvis
I like this one. Of course.
- [142:55]
Leo Laporte
Richard's nose. Mike, your nose is disappearing.
- [142:58]
Mike Elgin
Yeah, that's. I. They didn't get my nose right.
- [143:01]
Leo Laporte
That is 4o says, Dr.
- [143:02]
Mike Elgin
They didn't quite get your shirt right. They got the spirit of your shirt, but not.
- [143:06]
Leo Laporte
What was the prompt, Dr. Do. I'd like to see how detailed it was.
- [143:09]
Mike Elgin
Probably uploaded a screenshot.
- [143:11]
Jeff Jarvis
Right?
- [143:11]
Leo Laporte
Oh, did you? Oh, that's neat. Oh, boy. Now I got a lot more to worry about after section 230 goes away. Here's a post from Medium newsletter by John Passacinto. Passantino. I'm sorry. Sorry, John. Hanging by a thread. Meta's decisions at the once promising Twitter killer serve as a cautionary tale about the risks of prioritizing corporate interests and political appeasement over user trust. He. He says there's no. There's no future. By the way, is a good. A good newsletter, which I don't pay for, so I can't.
- [143:53]
Jeff Jarvis
I do. Status is the. Is a news site started by Oliver Darcy after he left cnn.
- [144:05]
Leo Laporte
Okay.
- [144:06]
Jeff Jarvis
I was very good about news. Yeah, it just says that. That. That people aren't Necessarily trusting threads that it tried got kind of desperate. And I think it's. I think it's a mark that Blue skies winning.
- [144:19]
Mike Elgin
Yeah.
- [144:19]
Leo Laporte
Yeah. You know, one of the things he says, and I think it's true, is that threads, for a long time wouldn't let you do a non algorithmic feed that just follow your. Just follow people. I follow which people need. That's the most important thing a Twitter or a Blue sky can do. And on threads, you had to follow whatever they thought was important. Then they decided no news, which took away a lot of the interesting things. Yeah, corporate interests.
- [144:50]
Mike Elgin
So much of social networks are just the feel of it, like how you experience it emotionally or whatever. And threads just feels like you're out in the wilderness talking to who knows who about God knows what. Whereas Blue sky seems like the Twitter of old, where it seems relevant and timely and. And all that kind of stuff. So they're really losing on the vibe.
- [145:12]
Jeff Jarvis
Yeah, I agree. The thing about threads that's weird for me though, Mike, is that the 4U feed for me on threads is really good. Maybe stuff from 12 hours ago, but it's stuff that I. They do that well. So then.
- [145:28]
Leo Laporte
And that's the algorithmic feed, isn't it?
- [145:30]
Jeff Jarvis
Here's one from five hours ago. I appreciate that Jeffrey Goldberg did not sit on this news for a book that he writes two years from now. Right. That's the kind of thing that I want to read. And you know, five hours, one day, seven hours, it goes back and finds stuff that's I find myself reading regularly.
- [145:45]
Leo Laporte
People completely. Completely parenthetically. I will say, isn't it great that there are so many great outlets for good, talented journalists like Mike Elgin and that we can read them. I'm just sad that I have to subscribe. I mean, it's expensive, but I guess that's the true cost of news, isn't it? Or is it? So many good news.
- [146:06]
Mike Elgin
I would love to criticize one of the things on the rundown, which is the article by Dave Troy about substack, the substack dilemma. Okay, this is right. Right in the. In the. In the what?
- [146:17]
Leo Laporte
We're talking right here.
- [146:18]
Mike Elgin
Yeah, yeah. This is yet another attempt to like harshly criticize substack, and I just don't get it. There's not a. I read this carefully. There's not a single. Well. Well, there's not a single valid criticism against substack in this entire article.
- [146:32]
Jeff Jarvis
They give money to Nazis, Mike.
- [146:34]
Mike Elgin
No, they chose to give.
- [146:35]
Jeff Jarvis
Yes, they did. They chose Nazis.
- [146:38]
Mike Elgin
No, they partnered with people who were doing.
- [146:39]
Jeff Jarvis
Doing, I think, very bad stuff.
- [146:42]
Mike Elgin
The criticism is twofold. One is that the. What's his name spacing on his name was in an interview with somebody and it was a, you know, recorded video, live interview. And he said, are you going to change your policy on this thing? And he basically said, well, you know, we have to look at that stuff like that. So he is severely criticized for that. So if, if there are, like, I'm a heavy user of Substack and I've never ever seen one Nazi anything in any form, in any way.
- [147:12]
Jeff Jarvis
Maybe you haven't, but they had it there and they paid them. They paid a lot of people.
- [147:17]
Leo Laporte
Right. The model for Substack is we're going to get really good people and we're going to encourage them to come here by giving them a guarantee on their subscriber fees. And it was substantial. It was hundreds of thousands of dollars in many cases.
- [147:32]
Jeff Jarvis
And the other argument here is they take 10%, which is a really high percentage.
- [147:36]
Mike Elgin
Okay. So it's a, it's a free market. Whatever the market will bear. If it's too high, then don't do it. But like, his point, a lot of people use Beehive.
- [147:42]
Leo Laporte
What do you use, Mike? Are you using.
- [147:45]
Mike Elgin
No, no, I use Substack. But here, here's the thing. So there's a little bit of, like, dishonesty by omission. I don't want to accuse him of dishonesty. I respect him as a journalist. But basically he says things like, you're kind of locked in. And he says, oh, sure, you can download your users and take them another thing, but it's. It's very difficult to do that. And then he doesn't say why. It's difficult. Difficult. It's difficult because product, you know, services like Ghost don't do the things that Substack does for you. Like, it's. Substack is great. I've. I've. I've been doing my newsletter, brace yourself for 25 years.
- [148:19]
Leo Laporte
Wow.
- [148:20]
Mike Elgin
I've been on six or seven different email.
- [148:22]
Leo Laporte
All the hard ways.
- [148:24]
Mike Elgin
I did it by hand. In the old days. I did mailchimp. I did Posterous. Remember posters?
- [148:28]
Leo Laporte
Yeah.
- [148:29]
Mike Elgin
I've done it all. And I've moved it from place to place.
- [148:31]
Jeff Jarvis
The standard steps on them.
- [148:33]
Mike Elgin
Exactly. And, and so, and so, like, I really appreciate, like, it harasses users for, for the subscription, not me. It's a standard. There's lots of people on Substack who are looking to pay for content, which is great. I I like making money on the content that I write there. The other criticism that he has for it, which I think is. Is. Is completely invalid. I mean, you know, besides that. I mean, just one more point on that. Remember Platformer? Platformer famously moved from Substack to Ghost. And you could just feel the sort of like, you don't really hear that much about platformer as anymore. I do, because I really seek out that content. But remember, like, a year or two ago, Platformer was. Everybody was talking about it all the time when they were on Substack.
- [149:19]
Leo Laporte
Yeah.
- [149:19]
Mike Elgin
They moved to Ghost and it's like where. Where they go. And so. And they did. Yes, they brought their things with them, but all of the promotional things that went with that that were really great.
- [149:30]
Leo Laporte
It's funny, my substack login is still platformer@leophil.com or whatever.
- [149:35]
Mike Elgin
Yeah.
- [149:36]
Leo Laporte
Because I signed up. That was the first thing I signed up on the substack for.
- [149:40]
Mike Elgin
Right.
- [149:40]
Leo Laporte
But that was back when Platformer had the inside story on the Twitter acquisition by Elon Musk.
- [149:46]
Mike Elgin
Yes.
- [149:47]
Leo Laporte
Maybe that's why they're less au courant. Yeah.
- [149:50]
Mike Elgin
I mean, I think it has a lot to do with the case.
- [149:52]
Jeff Jarvis
I'm not associating with Bari Weiss. I'm not associating with Ben Shapiro. That's a political opinion, but it's a strongly held political opinion. I have. And they're in partnership with Barry Weiss, and I'm not supporting that.
- [150:04]
Mike Elgin
Okay, so. And this is.
- [150:05]
Leo Laporte
They're also in partnership with. Are they ever in partnership with people on the left?
- [150:10]
Mike Elgin
Oh, 99%. There's so many lefty writers on Substack, it's not even funny. You know, Virginia Heffernan goes on and on. There's like. That's all I see in.
- [150:24]
Leo Laporte
I mean, his. His point is not. Dave Troy's point is not wrong. You. You might consider that if you're considering starting a newsletter and going somewhere else.
- [150:35]
Mike Elgin
Well, I mean, he doesn't emphasize the Nazi element. I think that whole thing has kind of faded out. I don't even. I don't even know where I stand now. It seems. It feels to me like a complete non issue.
- [150:44]
Leo Laporte
He says people looking to launch new content sites should avoid substack entirely because of the substantial risk of capture. That's what he said.
- [150:51]
Mike Elgin
Okay, that's ridiculous. And the other. The other thing, there's no lock in. You download. I've been on seven platforms. I can just download it today, and by the end of the day, I'll.
- [151:01]
Leo Laporte
Be on A. Nate. As long as you can download all your users right now, but they're set up with a payment system that is tied to Substack. How do you migrate their payments? I mean, you have a free newsletter.
- [151:13]
Mike Elgin
I'd have to. No, I have a four pay and a free. My free one is pretty substantial, which I recommend for everybody to really provide value to the free subscribers. I believe in that. But, like, you know, yeah, it'd be a hassle to migrate the payments, but on the platforms I was on before, like, there was no payment. So, like, was so hard to do all that stuff. And so I really appreciate. But the other thing. The other false argument he makes is that, like, oh, you're at risk of substack fatigue because there's so many. You know, there's so many different people on there. Everybody's using it. So it's just too many. Too many writers on there. Well, that's. That's complete bs. It's a free. You know, the Internet has too many writers too.
- [151:52]
Leo Laporte
Say that about the Internet. Yeah, yeah.
- [151:53]
Mike Elgin
And the other thing that I really don't.
- [151:58]
Leo Laporte
That's not unreasonable. I think both of you have points.
- [152:02]
Jeff Jarvis
Yeah, we have different opinions.
- [152:03]
Leo Laporte
Jeff doesn't need to be on Substack. I would not castigate Mike for being on Substack.
- [152:09]
Mike Elgin
No, but what we're talking about here, Jeff's concerns are not really the points that were made in this article. So I'm taking down this article which has. That's invalid points. I mean, I don't think Jeff would say that it's hard to download your subscribers or there's any substantial. That's not your issue. And here's political.
- [152:28]
Leo Laporte
Yeah, right.
- [152:29]
Mike Elgin
And. And here's. But here's the one that may also not be. Jeff may agree on this as well. He's saying basically that Substack is seen as part of a parallel establishment. I don't know if you've heard this, but it's like part of this bro. Broligarchy thing where they want to have a parallel government. A parallel.
- [152:46]
Jeff Jarvis
This is a big, big deal. Yes, it is. It's a very Andreessen. It's a huge deal out network, and.
- [152:51]
Leo Laporte
He'S a big funder of Substack. Right.
- [152:53]
Mike Elgin
But. But as a. As a journalist, I'm not eviscerating the mainstream media. The mainstream media is being eviscerated by all kinds of forces. And if. If. If, you know, if the New York Times, Wall Street Journal or the Atlantic or anybody, if you're listening, if you want to pay me money to write whatever I want on my own schedule and for any length that, and if you want to pay me the amount of money that I need to be paid, I'll delete my substack and take that job. But it's getting hard out there for a writer.
- [153:30]
Leo Laporte
Yeah.
- [153:31]
Mike Elgin
Phrase. Yeah.
- [153:32]
Leo Laporte
I mean, but the ones that succeed do quite well, right? I mean, Alex Cantrowitz is doing very well with his big technology blog, Letters from an American.
- [153:41]
Mike Elgin
Supposedly makes millions of dollars a month. I mean it's ridiculous. But I don't care about the outliers like that. The fact is that somebody who, who wants to do, you know, I, I do write for them. You know, I still write for, for, for Foundry now, now it's Regent. They just bought a, Regent just bought TechCrunch and all the, all the former IDG publications. But, but you know, so I still have a foot in that, in that realm. But I, you know, I can see the writing on the wall and I want, I'm going to keep writing for many, many years. And so I've been building my paid subscription model and I really appreciate Substack enabling me to do that to, to live around the world, to, to be in different time zones all the time. Sometimes I can't write for two weeks, sometimes I won't write every day for three, four days. And all of that freedom is available on Substack. And I really appreciate it.
- [154:35]
Leo Laporte
But not exclusive to Substack.
- [154:37]
Mike Elgin
No, no, no, I'm not so exclusive.
- [154:38]
Leo Laporte
I mean Ben Thompson's on WordPress, he publishes Pressable.
- [154:42]
Mike Elgin
Yeah, see but that's the thing. It's so hard. I don't, I just want to write and, and I don't want to be formatting. I don't want to be dealing with finances.
- [154:50]
Leo Laporte
This is, this is content like the argument of when. In the early days of blogs, when Blogger came out, there are people who said, ah, you know, you shouldn't be using Blogger. You should use your own self hosted website. Why give that money to, you know, Ev Williams and eventually to Google, but it made it easy for people who didn't want to do. I mean I've always done my own site pretty much.
- [155:13]
Mike Elgin
Yeah.
- [155:13]
Leo Laporte
But it made it easy for people who didn't want to do that. And also I've never charged and that's probably a barrier to entry to set up an E commerce.
- [155:22]
Mike Elgin
Honestly, what you will find when you go to Substack and I invite everybody to do this, go there and root around and try to find. I dare you to find Nazi content or anything. It's just a huge repository of incredibly smart people writing really brilliant stuff and many of them are making something of a living out of it. It's kind of a great thing. And nobody's. Nobody else has ever achieved this. As far as.
- [155:46]
Leo Laporte
When you go to substack.com do they promote right wing content? Oh, here's May.
- [155:51]
Mike Elgin
I've never twice. Well, I mean, there's nothing wrong with right wing content.
- [155:59]
Leo Laporte
You know, Derek Guy, CGP Gray. There's Gary Marcus, we just had him on the show.
- [156:06]
Mike Elgin
There was Unicorn roast.
- [156:08]
Leo Laporte
Here's Midas.
- [156:09]
E
Yeah. I don't know how much it would be them promoting that stuff as much as.
- [156:12]
Mike Elgin
Well, this is the social network.
- [156:14]
E
People looking at point hinting to that stuff usually.
- [156:16]
Leo Laporte
Yeah. So how do I go to the newsletter network where's. I didn't even know they had a social network.
- [156:23]
Mike Elgin
Yeah, it's called Notes. It's not great, but it's okay for people who are into newsletters and who are.
- [156:32]
Leo Laporte
Is there a homepage for the newsletters that's independent of the newsletter? I mean, I go to your page.
- [156:37]
Mike Elgin
Yeah. Yeah.
- [156:39]
Leo Laporte
What's the song? I don't know where I am on substack. I just went to substack.com that way though.
- [156:46]
E
Do people go to substack to discover new.
- [156:49]
Leo Laporte
Just like the Internet, they go to.
- [156:51]
E
A specific place, right?
- [156:53]
Mike Elgin
Yeah, they really push the app and the app is just a stream of the newsletters and podcasts that are also there. And you can do it by sort of like. What do you call it? Content areas. And so yeah, it's.
- [157:09]
Leo Laporte
My original complaint was though, that I have so many subscriptions and it's really out of control here.
- [157:16]
Jeff Jarvis
We are fools you into getting more. I had. I spent days, a long day going through and killing things that I hadn't realized I subscribed to.
- [157:24]
Leo Laporte
Is there. Is.
- [157:25]
Jeff Jarvis
Would.
- [157:25]
Leo Laporte
Would it be possible have a platform where you Give the platform 50 bucks a month or whatever and it then allows you to see everything on the platform and everybody gets a slice of it or.
- [157:35]
Jeff Jarvis
Well, it's me Medium.
- [157:37]
Leo Laporte
Okay. Okay.
- [157:39]
Jeff Jarvis
That's what they do.
- [157:40]
Leo Laporte
But substack is really dominant these days.
- [157:42]
Jeff Jarvis
Yep.
- [157:43]
Leo Laporte
Why?
- [157:45]
Mike Elgin
Because it's so friendly to the writers. I mean, I'm looking at the suggestions. Here's who it's suggesting to me. Steve Wozniak. Right. Pete Buttigieg has. Has a substack. Apparently. Jeff Perlman, Nicole Weeks, Derek Thompson. You know, it's, it's, it's pretty good algorithm for understanding the kind of content I would want. And, but it's really, you know, it's really the way use substack. Frankly, like I barely open the app. I don't even go to the site unless I'm posting something, right? I use something called Matter. It's an app called Matter.
- [158:20]
Leo Laporte
It's a newsreader.
- [158:21]
Mike Elgin
It's a newsreader, but it will go into your email, grab the newsletters, put it into Matter, and so, and it'll read it to you with an AI voice. And so in fact, you know, in preparation for, for, for this podcast, I actually grabbed, you know, 20 or so articles, put them into Medium, and while I was doing dishes and helping them move, I was listening to the articles on the rundown. Pretty great. That's, that's how I do it. But the point is like, you know, we newsletter, we long suffering newsletter publishers have, have been abused. We've been, you know, I, I get, I stopped doing my newsletter for like a couple of years and when I went back, I'm like, I'm going to start my newsletter again. Newsletters are back, back. And I went in there and they're like, nope, you can't keep your list. So I had to start from zero and build it back up to the thousands that I have now. And it's like I've been through hell trying to be a newsletter publisher for two and a half decades and I love the fact that I can just open this thing up. I can do a nicely formatted newsletter very quickly. They handle the SEO, they handle the. I can upload massive videos. They just host that, that and that's delivered to, to so people can watch the videos from their inbox. And it's hosted by Substack. It's so nice as a. Jeff, do.
- [159:39]
Leo Laporte
You have, you're on Medium, right? Jeff, but you also have buzzmachine.com. do you have something like Matter that you use to read newsletters or.
- [159:47]
Jeff Jarvis
No, I don't, I don't like newsletters. I can't stand newsletters. I unsubscribe to them constantly. I'll subscribe to them and get going. No offense.
- [159:53]
Leo Laporte
There's so much the best writing is.
- [159:55]
Jeff Jarvis
I know, but that's the problem.
- [159:56]
Leo Laporte
That's.
- [159:56]
Jeff Jarvis
It's closed world. I miss being on blogs. I miss that being open a Medium. You can help. Mike, just one second.
- [160:02]
Mike Elgin
Yeah, but you can.
- [160:04]
Jeff Jarvis
On Medium. You can, it's a, it's more of a blog, but you can also subscribe and have it sent to you. That's the model that I prefer thinking now and that's Ghost does that, isn't that. Yeah, Ghost does that too, which I like. Substack's not that different in a sense. You're not obligated to do it. And if you subscribe, subscribed to one newsletter there it fools me into subscribing to three more. And then I had myself. I had 96 that I didn't know I was subscribing to. And so my email was just a mess of them and I don't read any of them as a result. It hurts that as a result. So the whole newsletter push took away, I think from the blog world, which I regret.
- [160:39]
Mike Elgin
But that's the problem with blogs. Blogs. Nobody pays attention to blogs. I've been trying to. I know they used to, but nowadays nobody does. And Google has changed its algorithm to sort of like, you know, their second class citizens now. And so the, so the substacks aggressive push to get people to read stuff is the boost that you need. And by the way, if you go to MachineSociety, AI, this is a blog. Like it's just a blog, right? Yes. They're pushing this other stuff. Yes. It's like, you know, encouraging you to subscribe and all that kind of stuff. But it's a blog, it's in the cloud, it's free to read. And there it is. It's a blog, but it's a blog with benefits. And that's what we need if you're a blogger. Because blogs are going nowhere. They're not growing.
- [161:21]
Leo Laporte
Oh, it only works with Gmail Matter. It has to go through your Gmail inbox.
- [161:26]
Mike Elgin
That's right.
- [161:27]
Leo Laporte
I hate it. So many tools require Gmail. I mean, I understand it's the number one email, but I don't use it, so it's useless to me. I think I didn't. Did Matter get sold? I think I had Matter and used Matter. They got sold. Right.
- [161:41]
Mike Elgin
They recently sent an email to users that was really interesting. It's founded by two people whose names I don't recall, but they both got cancer in the same year.
- [161:51]
Leo Laporte
Oh yes.
- [161:51]
Mike Elgin
Both been in recovery and they're both recovering and. But they're, they're kind of going along and they're trying to raise money and stuff like that, but they're still, I think they're still owned by the original people.
- [162:02]
Leo Laporte
Okay, well, I've just re. Signed up.
- [162:06]
Mike Elgin
Yeah.
- [162:06]
Leo Laporte
Because I like the idea. I mean, God, that's. I spend almost every minute of every day going through news.
- [162:13]
Mike Elgin
Yeah.
- [162:13]
Leo Laporte
I use a newsreader called Tattoo Tapestry, which connects also to social accounts and is a really great easy way from the Icon Factory Mac only, unfortunately. But it's a very useful little tool.
- [162:23]
Mike Elgin
Yeah. The other benefit is you install the extension and just click a button and whatever you have now, it's added.
- [162:29]
Leo Laporte
Yeah. I definitely spend more time with this. I feel like I used it and then I don't know why I stopped using it, but Dig Wells Incorporated. All right, let's take a break, and the final minutes of the show. Thank God are coming.
- [162:48]
Jeff Jarvis
My Chinese food is on order.
- [162:50]
Leo Laporte
Oh, good. All right, they're coming up in just a moment. You're watching Intelligent Machines with. What are we having for dinner, Jeff?
- [162:58]
Jeff Jarvis
Chinese. I think I'm going to get some pork gyoza. No, that's Japanese. I just got dumplings, they call them.
- [163:05]
Leo Laporte
It's also. They have. Have them in China. It's called Guy bow.
- [163:10]
Jeff Jarvis
Right. And then I'm going to have chicken with garlic and eggplant sauce. Or chicken with garlic eggplant sauce. No, chicken with. With eggplant and garlic sauce.
- [163:18]
Leo Laporte
It's fantastic.
- [163:19]
Jeff Jarvis
Eggplant and chicken sauce.
- [163:20]
Leo Laporte
I believe I am tasked to make meatloaf tonight.
- [163:23]
Jeff Jarvis
Oh, I love meatloaf. But do you put ketchup on top?
- [163:26]
Leo Laporte
I have a. Well, Serious Eats has a recipe, mustard and stuff that involves making a ketchupy kind of a. That's good. And there's bacon involved.
- [163:35]
Jeff Jarvis
And then the next day, the sandwiches. How. This is how white bread I am. Next day, it's white bread. A sandwich on white bread with Miracle Whip.
- [163:45]
Leo Laporte
Miracle Whip, baby.
- [163:47]
Mike Elgin
Sweet mother of. Well, I'm making. I'm making patty melts.
- [163:51]
Jeff Jarvis
He thought he was mad about substack. Now he's really pissed.
- [163:54]
Mike Elgin
So I'll tell you, the bread that I'm using, it's homemade sourdough bread made with. With Emmer.
- [164:00]
Leo Laporte
Oh, yeah. You turned me onto these Old Wheats. Yes, yes.
- [164:04]
Mike Elgin
And so that's what I'm having.
- [164:06]
Leo Laporte
I still have my Wolfgang Wheat Mill, so I can make my own flour from these Heritage Wheats.
- [164:14]
Mike Elgin
Yeah.
- [164:16]
Jeff Jarvis
Yep.
- [164:16]
Leo Laporte
I am making bagels these days pretty regularly, and I. It's wonderful. They're homemade bagels. Nothing like it. Although I'm a little nervous. I have been using malt barley malt syrup to boil the bagels in to get the browning. But I was challenged by Amy Webb. She said, well, do you use lye? I said, what? She said, you got to use lye. That's not a bagel unless you boil it in lye. So I ordered some food grade lye which it says on the bottle can also be used as a drain cleaner.
- [164:52]
Jeff Jarvis
It also said he didn't piss off Amy.
- [164:54]
Leo Laporte
It also said wear gloves and eye protection before using. No, no, it really, I looked it up. It says AI told me lye and Elmer's glue are common ingredients. No, no, no. It is in fact what they use. They use a very small lye is just very alkaline. So use a very small amount in a pot of water and it helps it get brown, I guess.
- [165:18]
Mike Elgin
And you can also use it to nixtamalize corn, presumably.
- [165:22]
Leo Laporte
Yeah. And make pretzels. So who knew? Yeah. And now I have a lot of lye if anybody. Because you can't buy a small amount of lye. It's an industrial grade product. No, I've been, I've been trying, I've been trying with the barley. I'm going to try it with LIE this week and we'll see, we'll see what happens. The thank you, Mike Elga for joining us. We will give a little plug to Kevin's business before we depart because that's, that's part of the deal. His son Kevin does an amazing thing called Chatterbox.
- [166:01]
Mike Elgin
Yes.
- [166:02]
Leo Laporte
Which is a way for schools to teach kids about AI without invading their privacy.
- [166:10]
Mike Elgin
That's right. Hello, you're looking at, hello, chatterbox.com and yes, and this is a, you know, speaking of WOL from Alpha, one of the many skills that kids can use is they can build in, they have APIs that go to WOL from Alpha. So if you have, you want to do math type of problems or do any of the many things that wolram Alpha is great at, you can do that with a voice command through Chatterbox. And this is the only COPA compliant smart speaker that's legally usable in schools because it's so private. Part of the privacy is that giant button on the top which it doesn't listen until you press the button and you have to be kind of nearby to use it. So you have to be nearby to press the button, for example. And so that combination of proximity and pushing the button, plus the fact that neither Chatterbox nor any of the companies that he Accesses through the APIs Know who the user is. They have no data, no information at all about who the user is. And so this has been a very difficult problem to, to, for him to achieve this, but it's totally private. And what he wants to do is make sure that kids get used to the idea that there is no monetization of their personal data. And so that's all very clear to kids. And what other products that are in the environment that we all use, the smartphones, the smart speakers, they all just sort of make us calloused in terms of like, yeah, they're getting all my data. What are you going to do? So this is something that he doesn't do for the kids. So it's very great and it's very educational and it's a maker space in a box. It teaches kids to be makers of technology and not just passive consumers.
- [167:41]
Leo Laporte
So always love giving this a plug. It's a great thing. Hellochatterbox.com and thank you for that. Yes, we're going to get our picks of the week and final thoughts as we wrap up this episode of Intelligent Machines in just a little bit with Jeff Jarvis, Mike Elgin filling in for Paris Martineau. I'm Leo Laporte. All right, that was our brief pause and now it's time for your pick of the week. Shall I start with you? I'll start with you, Jeff.
- [168:12]
Jeff Jarvis
Okay.
- [168:13]
Leo Laporte
I can tell you want to.
- [168:14]
Jeff Jarvis
I want to get my Chinese food.
- [168:15]
Leo Laporte
Yeah, I can.
- [168:16]
Jeff Jarvis
Yes. I'm going to do a real quick MA report. I found this interesting that all together in one week, the Skim was bought by Ziff Davis. Napster was bought by, I think a private equity company company.
- [168:31]
Leo Laporte
Yeah.
- [168:31]
Jeff Jarvis
Yahoo Was bought by an investment firm.
- [168:34]
Leo Laporte
Yahoo.
- [168:35]
Jeff Jarvis
I'm sorry, TechCrunch. TechCrunch.
- [168:37]
Leo Laporte
TechCrunch, yes.
- [168:38]
Jeff Jarvis
Was sold by Yahoo. To investment firm Regent.
- [168:42]
Mike Elgin
Yes. Which also bought Computer World, PC World, all those. All those publications. I now work for the same company that TechCrunch.
- [168:53]
Leo Laporte
So confusing. So confusing.
- [168:56]
Jeff Jarvis
Yep.
- [168:56]
Leo Laporte
But what was the skim?
- [169:00]
Jeff Jarvis
Was that great? It was a great newsletter aimed at women. That's why you don't know what it is. Ah, young women. That's especially two reasons you don't know what it is. Was founded in 2012 by Danielle Weisberg and Carly Zaken. Providing subscription only newsletter digest of news stories intended to be simple and easy to read.
- [169:21]
Leo Laporte
Nice.
- [169:21]
Jeff Jarvis
It was. It was really well done. Very popular in the day. But, you know, things have their heat and they're cool.
- [169:27]
Leo Laporte
Well, it's been sold. Doesn't mean it's cooled. It might mean that it's now on its way to the moon.
- [169:33]
Jeff Jarvis
Well, yeah. And working for Ziff Davis is. Is a very good place to land, I think.
- [169:37]
Leo Laporte
So.
- [169:38]
Jeff Jarvis
We like those folks.
- [169:39]
Leo Laporte
We. Yeah, I used to work for them.
- [169:41]
Mike Elgin
Yeah, I think I consulted for them once.
- [169:44]
Leo Laporte
Yeah.
- [169:45]
Jeff Jarvis
So that Was it for me?
- [169:46]
Leo Laporte
Mike has a toy for us or two?
- [169:48]
Mike Elgin
Yes. Okay. So as you know, I'm a Ray Ban meta enthusiast and that I travel around to foreign language countries. They have an experimental feature on Ray Ban metas, which is Live translation.
- [170:01]
Leo Laporte
Oh.
- [170:02]
Mike Elgin
Once you launch that feature, it will just be listening constantly. You don't have to do anything. And anytime it hears anything spoken, it will translate it. They have three languages right now. French, Spanish and Italian. Which is perfect for me because that's most of the countries I go to speak those languages. And somebody talks to you in Italian and you hear the English whispered in your ear on the app, it'll show you when you say something back in English, it'll show you what you said in English followed by what that would translation would be into Italian or whatever language you have selected. Okay. So this feature, it works pretty well. It's not bad. You can't use it for like sitting around a table with five people because it. You go nuts because it's like, like there's a delay and you can't do it. But if like you're just want to go up and ask somebody a question, you'll understand the answer. People giving you instructions like at the airport or whatever, you can understand those. It's really nice. Here's something I've never heard anybody talk about or write about. In addition to live translation, they have something called Live AI. So this is like the Google thing, the, the Gemini thing we were talking about earlier, where it's just getting the video feed and your interaction asking questions about what it sees. So Live AI works just like that, through the glasses. Glasses, the right place for this. It's not the phone. Nobody's going to use it on the phone, but in glasses, it's amazing. So you say, you say, you know, launch Live AI and you can just, it will be silent, doesn't say anything. And you can say, well, you know, what kind of, what kind of business am I in? And it, based on what it saw through your glasses, it will tell you details. Now here's the thing that I've never heard anybody talk about about. It also does translation. So if you look at a sign and say, what does this sign say in English? It will tell you. And I've tested it on a gazillion languages, including Japanese, including other things. And it always translates it correctly into the language. And so to use Live AI as a, as a visual language translator is really incredible. And really, I think it's the future of where we're going and One of the things we're going to be able to do with, with, with AI glasses.
- [172:09]
Leo Laporte
So to do this, I have to sign up for Meta's early access program.
- [172:12]
Mike Elgin
That's correct. That is correct.
- [172:14]
Leo Laporte
Which I assume I can do that in the app.
- [172:16]
Mike Elgin
Yes, I think so. It may. I may not happen right away. So if you try to do this here. But anyway, this is a. My point is not, you know, go buy Ray Ban Meta glasses and use this feature. My point is this is the future that we're all going to have access to within a year or two. This is going to be very commonplace. It's going to be kind of a cool feature. And very quickly, I want to do one more Ray Ban Meta thing. That is really kind of ridiculous. There's a startup that's offering a product called a metaport, and what this does is it adds a battery, extends the battery for your glasses.
- [172:49]
Leo Laporte
Yeah, but you look like a dork.
- [172:51]
Mike Elgin
Yes, exactly.
- [172:52]
Leo Laporte
It's ridiculous.
- [172:53]
Mike Elgin
It clamps onto the front and the side and you kind of look. Look like you're wearing Google Glass. We're back to that glass hole look, you know, that's so popular with the kids nowadays.
- [173:02]
Leo Laporte
Oh, my God. Is it. It looks like it's 3D printed or it looks like it's a licorice turd on your glasses. I don't know what it is. It's three.
- [173:10]
Mike Elgin
That would be a better brand name than Metaport. Licorice turd. That would be.
- [173:15]
Leo Laporte
Oh, wait, they have clear ones.
- [173:17]
Mike Elgin
Oh, get the orange one. Get the orange one.
- [173:19]
Leo Laporte
Get the orange. Yeah, baby. Well, that's what I figure. If you're going to do it, flaunt it. Right? Just go all out. Go big or go home. What do you get for battery life? I never wear these long enough to find out.
- [173:32]
Mike Elgin
It's fine. You get a few hours with just the glasses. And then the case itself is a battery, so you put them in the case. Even if it's not plugged in, you get a few more hours just getting.
- [173:43]
Leo Laporte
The update to 13.0. So as soon as I do, I will sign up for the early access program. Even though these classes really are not.
- [173:53]
E
Flattering, how does it work with like, say, the distinguishing between the difference between, like, Mexican Spanish and Spain Spanish?
- [174:01]
Mike Elgin
So far it works great. I've. I've done it in Mexico. I've. I've heard Spanish from. People have thickly accented. They're, you know, they're Zapot, Zapotec speakers and they speak Spanish with an accent and understood it Fine. I've also had problems with it. Like we were in a restaurant one time and there are people playing pool at a pool table, like far away. And they were talking to each other and it was just giving me all of that I was trying to have, you know, trying to mind my own business and just kept giving me the English translation of what they were saying. Sometimes when somebody's right in front of you, it doesn't quite pick it up. And here's the funniest part. If somebody says something in English, it will translate that English into English. And sometimes it mistranslates. Translates it, huh? From English to English. That's. My wife will say something to me and then it gives me the wrong translation.
- [174:51]
Leo Laporte
It doesn't round trip.
- [174:52]
Mike Elgin
Well, yeah, but, but my point is, my larger point is that this is every device that is anywhere near our ears, Headphones, earbuds will do this with.
- [175:01]
Leo Laporte
The AirPods for sure.
- [175:02]
Mike Elgin
Exactly.
- [175:03]
Leo Laporte
Yeah.
- [175:03]
Mike Elgin
Language translation. And it's gonna, it's gonna harm the, the, the, the cause of learning foreign languages because, because people are going to be like, well, why bother? But the problem of people speaking foreign languages is going to be erased almost by AI within the next couple of years.
- [175:22]
Leo Laporte
Well, we've erased Jeff Jarvis.
- [175:24]
Jeff Jarvis
I got to go.
- [175:25]
Mike Elgin
I gotta go.
- [175:26]
Jeff Jarvis
The restaurant closes.
- [175:27]
Leo Laporte
Thank you, Jeff. Take care. Go get some Chinese food. Jeff Jarvis, buzzmachine.com his latest, the Web. We want weave Gutenberg parenthesis now out in paperback all@jeffjarvis.com Take care, Jeff.
- [175:39]
Mike Elgin
Thank you.
- [175:40]
Leo Laporte
Thank you.
- [175:40]
Mike Elgin
Bye, Jeff.
- [175:40]
Leo Laporte
Bye. Thank you. Mike Elgin, always a pleasure. Appreciate filling in for Paris this week and anytime. Have a wonderful time in Sicily.
- [175:49]
Jeff Jarvis
Thank you.
- [175:50]
Leo Laporte
Yeah, I will try. I'll do it for you guys. Thanks to all of you. I can see you, you know, with these glasses. I could see Joe and Sally and Tim and a little Johnny.
- [176:02]
Jeff Jarvis
And that's a reference most, most of these kids won't get.
- [176:07]
Leo Laporte
Romper Stomper Dumper. Do we do this show every. You can go. You. You go. You guys can go. I'll finish up all alone here. Thank you. Take care.
- [176:17]
Mike Elgin
All right, thanks.
- [176:18]
Leo Laporte
Don't want to make them stick around. We do this show every Wednesday, 2pm Pacific, 5pm Eastern, 2100 UTC. We kick things off, or at least we have been since the kind of rebranding to intelligent machines with an interview with somebody doing something interesting in AI like our guest Gary Rivlin this week. Fantastic book, by the way. AI Valley just out. So tune in and make sure you don't miss that first half hour. Of the show. Afterwards, we cover AI news and other news and picks of the week. Paris will be back in two weeks. I will be back next week and I hope you will be too. You can watch us live if you're a Club Twit member in the Club Twit. Discord. Always a great place to hang. But you can also watch on YouTube, Twitch, TikTok, X.com, facebook, LinkedIn and Kik. Wherever you want to watch the stream, you can. And you can chat with us in that stream until they revoke section 230. We'll have those chat rooms. And I see everybody talking in there right now, so it's. It's nice. Endless music said he said my name. Hi, Oscar. I see you. I see you after the fact. Oh, they're just telling me. Oh yeah, after the fact. You can watch the show at our website. Where is that twit im? You can also watch a YouTube channel dedicated to the video of intelligent machines. Or best thing to do, subscribe in your favorite podcast client. And that way you'll get it automatically as soon as we're done cleaning it up, taking out the swear words and all of that. You know what I mean? I'm. You know, we'd like to make it accessible to the youngsters. Yes. Thank you all for being here. Join the club if you're not already a member. Twit TV Club Twit. 7 bucks a month for ad, free versions of all our shows and access to the Discord. Lots of special events going on too. It's got some big ones planned. Thank you to Bonito Gonzalez, our producer and booker, for putting this all together, flipping the switches. Thanks to you for joining us. I'll see you next time on Intelligent Machines.
- [178:22]
Jeff Jarvis
Bye.
- [178:22]
Leo Laporte
Bye. I'm not a human being. Not into this animal scene. I'm an intelligent machine.