Intelligent Machines 814 transcript
Please be advised this transcript is AI-generated and may not be word for word. Time codes refer to the approximate times in the ad-supported version of the show
0:00:00 - Leo Laporte
It's time for Intelligent Machines. We've got a great one for you. Jeff does have the week off, but Paris Martineau is back from Europe and filling in for Jeff. Our special guest, corey Doctorow sci-fi author, activist, eff consultant and a good old friend. We have a lot to talk about AI art Is it really art or is it just eerie'll? Talk about meta. They've been gaming the ai benchmarks what a surprise and why you should never use ai to represent you in court. All that more coming up next on intelligent machines podcasts you love from people you trust this is twit this is intelligent machines, episode 814, recorded wednesday, april 9th 2025, chesterton's fence.
It's time for intelligent machines show. We talk about ai, robotics and intelligence all around us not human in most cases, although in some cases, paris martineau is back. Ladies and gentlemen, hi paris, hi leo how was paris?
0:01:15 - Paris Martineau
it was fantastic. It was très bien. I had so much fun. Um, yeah, I had a great, but I was missing you guys.
0:01:24 - Leo Laporte
When you're in Paris, do you have to call yourself Paris?
0:01:29 - Paris Martineau
No, even when I was in school. In college, I studied abroad there for almost a year and at that time I spoke fluent French. I do not anymore. It's definitely degraded, but even at that time, at the peak of my French speaking ability, where I'd be having a five, ten minute conversation with a random French person on the street, they then ask me, like in French, what's your name? And I'd say Paris, and I did versions of this with all different pronunciations of the name. Every single time the response was immediately switched to English. No, I asked you what your name is, not. Where are we?
0:02:06 - Cory Doctorow
So when I am in Paris, I specifically do not go by my name paris, because french people don't, yeah, yeah, they don't get it, and that's fair.
0:02:11 - Paris Martineau
That's a fair point. It's a weird thing for me to be there well, we missed you quite a bit.
0:02:15 - Leo Laporte
I'm so glad you're back. You missed some very good conversations, but we have a good one. Uh, today as well. Jeff will not be here. He was called away to salem. I hope it's not for witch trials. Oh, he's finally being tried. Yeah, he didn't explain. He just said, oh, I have to go to Salem and said, oh, he's like they're asking me if I'm lighter than a feather.
But we have both a great interview guest in his place and a guest host in his place, and again and a guest host in his place. The wonderful cory doctorow is here. It's great to see you, cory. Uh, always a pleasure to have you on. You've been on this week in google in the past. Of course, this week in tech many times. Did you know? By the way, cory, we are celebrating our 20th anniversary on sunday. Oh my gosh, that's, I know, make you feel old, wow.
0:03:05 - Cory Doctorow
Well, my daughter's going to college in the fall, so yeah you have.
0:03:08 - Leo Laporte
You've been on the show, uh since before uh your daughter was born, right, so at least you know probably much of those 20 years you've been on.
0:03:18 - Cory Doctorow
So I mean I used to come on the screensavers, so that's true. I've known you since before then at least, because it was at least my first year at eff, so it's been at least 23 years wow anyway, it's always great to have you.
0:03:32 - Leo Laporte
Uh cory is both a fantastic science fiction author. His latest is a continuation in the marty hench forensic accountant series, uh picks and shovels, which I hear is doing very well. People love it. I hear a lot from our audience because it's got retro tech in it.
0:03:50 - Cory Doctorow
You know it's like the the good old days of technology, so congratulations on that it's set in the early 1980s, which were, you know, the heroic era of the pc. Back before we knew what pcs were for or like, who was supposed to sell them or who was supposed to buy them or what shape they were supposed to be, people made weird ass computers. In Ontario, the ministry of education made a computer called the Sammy icon. It's sorry, are you trying to get to Martin Hench there, or no?
0:04:21 - Leo Laporte
No, no, don't look at that. If you're seeing that. Click the tab that we should have said this before the show began at the top of your zoom and have, uh, the the meeting instead of my camera, and I'm going to the bezelorg, by the way, but don't let me interrupt the anecdote, go ahead yeah, so.
0:04:38 - Cory Doctorow
So we had these weird computers called the sammy icon and they were a um, uh, uh giant injection molded chassis with an inset enormous track ball like you would get on the arcade game of centipede. And uh, and then it's booted three different operating systems including a logo prompt. Uh, so in case you just wanted to like use a logo shell as your daily driver, I that one of your many options. Uh, it was great. It was a beautiful computer.
So in in picks and shovels marty hench. It's his first adventure. He fails out of mit computer science because he's too busy programming computers to get a computer science degree. And, uh, he ends up in a cpa program because he likes the apple two pluses they bought and wants to learn more about VisiCalc, which he can't afford out of pocket. So before he knows it, he's moved out to the Bay Area with his genius coder roommate and he ends up in this weird PC company called Fidelity Computing, run by a Mormon bishop, a Catholic priest and an Orthodox rabbi.
The punchline is that it's pyramid selling, that they are selling these DRM locked PCs that like use special paper for the printer and special floppy drives. You can't get your data off of them. They're selling them through pyramid sales into the faith groups and locking them in and taking millions of dollars out of their hide. And Marty pretty quickly figures out he's working for the bad guys and switches sides and goes to work for their competitors, a company called Computing Freedom that was started by three of the sales managers from Fidelity who have quit, to build computers that unlock the Fidelity PC.
You know, a drive that will take any floppy, export tools for your data, new sprockets for your printer so you can stand your fanful paper and so on. And they are an Orthodox woman whose family's kicked her out because she's queer, a nun who's left her order because she got involved with liberation theology and a Mormon woman who's left the faith because she's pissed off that they opposed the Equal Rights Amendment. And what starts as a commercial dispute becomes a shooting war and the stakes get really high. So it's, you know it's a thriller set at the beginning of the AIDS crisis in San Francisco in the 80s. Jello Biafra is playing every weekend. You know the PC bubble has just kicked off and everyone is trying to figure out what the computer is supposed to be.
0:07:02 - Leo Laporte
I love it. I mean, it's Marty's origin story, that's awesome, that's right.
Yeah, that's fantastic. I mean, Corey's not just a novelist, though, he's also an activist, as you mentioned. He worked for the EFF and he's been you know, on his pluralisticnet blog a really outspoken champion of the right to repair, of fighting surveillance capitalism. One of your books, the most recent, is the Internet Con how to Seize the Means of Communication, Computation, Computation yes, sorry, I talked to you and Rebecca Giblin when Joke Point Capitalism came out. I mean, you've done so much for all of us and, of course, uh, most famous for I hate to say this, but for coining the term and shitification. Thank you for doing that, I appreciate it.
0:07:54 - Cory Doctorow
Yeah. Yeah, it's definitely going to be on my tombstone, but it's in the can. It's coming out in October. The following year there will be a graphic novel. Uh, over the following year there will be a graphic novel. Uh and uh the we, as I was mentioning before we started recording, we just sold the polish rights and I think that's the eighth uh different language it's been sold into now yeah, it captured people's imagination, mostly, I think, because we all recognized that's what was going on in the world.
0:08:19 - Leo Laporte
Like this is. This was a perfect encapsulation of of why, uh, everything seemed to be going downhill. You know, they were extracting the maximum value out of us all at this point, and so I think everybody identified it and immediately understood it and agreed. I think nobody disagrees. Yeah, I think that's right.
0:08:40 - Cory Doctorow
Yeah, yeah, definitely. You know we have hit a point now where the platforms are the only game in town it's. We're all dependent on them to some degree or another, but they are making our lives so miserable. So we're kind of trapped in this fight between do I stay or do I go? You know, like, do I take the hit of leaving behind if I'm on facebook, say, all those friends who I know from back in the country I moved away from, who I only communicate with on Facebook, or do I stay and allow Mark Zuckerberg to continue abusing me and he's really, you know, they've taken us hostage.
0:09:16 - Leo Laporte
Um, you know, this show is primarily these days about not Google anymore, uh, but about, uh, ai. And actually it's really broader than AI intelligent machines which covers even more than AI robotics and IOT and et cetera the edge intelligence at the edge. It's everywhere. Now You've written quite a bit about AI, do you use?
0:09:38 - Cory Doctorow
AI at all. No, I am a AI skeptic, and not just an AI skeptic, but a violent AI skeptic. Not a violent AI skeptic. A committed AI skeptic, an extremely skeptic. Uh, and not just an ai skeptic, but a violent I'm not a violent ace. Yes, committed ai skeptic, an extremely skeptical ai person. I mean, yeah, I use ai in the sense that, like when I turn on the camera on my pixel, it figures out where the faces are you can't not right.
But yeah, do I? Do I like, rely on spicy autocomplete? No, and you know the weird little collages that I make for my blog. That's me and the gimp and the lasso tool, uh, using sources from the library of Congress and Wikimedia commons and so on.
0:10:13 - Leo Laporte
Uh, you're so retro yeah.
0:10:16 - Cory Doctorow
I like I like making collages. That way, I have a whole theory on it Maybe we can get into it about like why what I think is missing from AI art and why art made by people has something ai art is missing. It's not really a romantic thing, it's.
0:10:30 - Leo Laporte
It's kind of an information theory thing I didn't realize that all of these were collages that you did. That's kind of cool yeah, they're handmade.
0:10:38 - Cory Doctorow
I'm doing a book of them in november, just a little one, uh, probably like a 500 copy print run, and I think creative commons is going to take most of them as premiums to give away to major donors because it's their 25th anniversary. But it'll be called Canny Valley.
0:10:52 - Leo Laporte
Canny.
0:10:53 - Cory Doctorow
Valley. I love it. Yeah, and Bruce Sterling is going to write the introduction, so it will be a nice little book. Fun little art book, fantastic, just in time for Christmas.
0:11:06 - Leo Laporte
So all these images are handcrafted, ladies and gentlemen. So what is tell us your theory about?
0:11:08 - Paris Martineau
ai art. Yeah, yeah, I think this is the perfect venue for this, also because leo is a huge ai everything defender, especially ai art oh, oh, don't doubt me, man, don't out me to corey.
0:11:21 - Leo Laporte
He used to have some respect for me, now he's. Well, you know, I'll say my attitude towards it is I'm not a true believer yet, because it has been yet to be proven, but I went from agreeing with you that it was spicy, autocorrect, it was a parlor trick to, as I think, a lot of people saying wow, you know, there is something happening here. It is more than just probabilistic prediction about the next pixel or the next word. There's something else going on, and how it will end I don't know, but I'm kind of open to the thought that maybe it will. I mean, I'm not. My thought is that we are not necessarily anything more than stochastic parrots, so it is not much of a criticism to say that that's all AI is, you know. I mean, how do we know that what we're doing is not auto, auto, complete? You know, based on our experience in life?
0:12:17 - Cory Doctorow
Well, I think that's a separate question, Maybe. Maybe I'll tell you what I think about the art first yes, do please talk about whether what the art.
0:12:26 - Leo Laporte
First, please talk about whether, what, what, what, the true?
0:12:27 - Cory Doctorow
nature of human consciousness is because that's what the question is. Yeah, yeah, yeah. So, um, let me start with a little parable. So I have a friend who's a law professor and uh used to be that if you were a law professor, only your very best students would hit you up for a letter of reference, because everyone knew that they were a pain in the ass to write. And getting a letter of reference from a law professor when you were considering a postdoc was considered, uh, meaningful because it was kind of proof of work, right, like someone put a lot of thought into writing this letter on behalf of this student, and they wouldn't do that if they didn't think highly of the student. So the existence of the letter was, in fact, an act of communication, in the same way that, like if you hire a fancy law firm to send a threatening letter to someone, irrespective of the contents of the letter, there is a, a message that travels along with it, which is, like I'm angry enough at you to spend a thousand dollars to pay someone to threaten you, and maybe I'll spend lots more money to pay people to threaten you later, right, or to to make your life miserable. So there's some like semantics of the letter. So fast forward to the days of LLMs and my friend now has every grad student ask for a letter of reference. Because everyone knows that the way you make a letter of reference now is you write three bullet points and you ask the llm to inflate it to five paragraphs of nonsense and as a result, when my friend is considering positions for a postdoc, they are flooded with bajillions of reference letters from other profs that are again like, inflated from three bullet points. And so what do you do with those letters? Because there's no way you can parse them all. You feed them to an LLM and ask them to be reduced to three bullet points. Now here's the thing even if we stipulate that the three bullet points that come out of the LLM are the three bullet points, which we know they're not, but even if this were a lossless process, it is, I think, incontrovertible that nothing else in the letter, right. The other five paragraphs matter in the sense that the LLM knows nothing about this law student and cannot convey a single meaningful thing about the law student right. They are just doing probabilistic guessing. So the communicative freight in the letter consists of three bullet points. It cannot consist of more right, where would the information come from? Right? Right, there's, where would. Where would the information come from? Right, claude Shannon decrees that the only information in the letter of reference is the prompt and everything else is nothing. It's padding OK.
So when I write a novel, I am making hundreds of thousands, possibly millions, of tiny, little subconscious communicative decisions. Right, which word goes where is about trying to convey something to you, right? And and I think that you know, art is the process of taking a big, irreducible, numinous feeling that is in the creator's head, imbuing some intermediate medium with that numinous thing as best as you can, whether that's like a dance or a painting or a sculpture, or a novel or a song, and then hoping that some facsimile that materializes in an audience member's head when they experience the work. And so I am making millions of tiny decisions in service to transpiring that big, numinous, irreducible feeling from my mind into yours, and so the communicative density of a novel or of a painting. If you're doing brushstrokes, we were talking about these weird collages, I do, you know. I download these like 15,000 pixel wide Hieronymus Bosch scans from the internet archive or from the library of Congress and I trace out individual figures at maximum magnification, and I'm touching the brush, brush strokes of Hieronymus Bosch and you. There is some communication there, right I am. He and I are reaching across time and space, and I am learning something about what he was trying to do by the decisions that he made when he put his brush on the canvas, right?
So when I prompt an llm to draw something or write something, the most communicative freight it can have is the prompt, right? By definition, the llm isn't communicating anything to you, right? The llm has no like thing to say to you it does. It is not in possession of a big, numinous, irreducible feeling that it hopes to materialize in your head. It's just making some guesses about what goes at the next pixel of the next word.
And so, while those may be good sentences, or while the draftsmanship of the LLM or the image generator may be good, the communicative intent cannot be expanded beyond the amount of communication in the prompt. The communication in the prompt is all of the effort that the creator has made to transfer a big, numinous, irreducible feeling from themselves to you. That's all. There is right? So the more communications that the creative worker has with the AI, the more density there will be. So if you generate 50 of them and pick one. Well, that's two communicative decisions that you've made. If you reprompt, that's three or four, maybe 10.
Maybe the reprompt is very complicated, and, of course, not every communication needs to be big in order to be meaningful. Right, when you have Duchamp, like you know, getting a urinal and putting it sideways and writing Armada on the side was very famous piece of art. Right, it carries a lot of communicative freight. But most very significant acts of communication are more than a few words. Right, they're there, they are. They require some material in order to be unpacked. There are some very moving haiku, but not as many as there are moving novels.
And so, at best, what you can say is that a creative worker in possession of some extremely compact and very meaningful artistic statement has fed it to an LLM that has then diluted it. That one meaningful thing among a million machine generated decisions such that the average amount of communication per pixel or word is still, like, homeopathically low. And so, for me, this is why ai art feels at best eerie, where eeriness is defined as the seeming of intent without an intender, like the way the, the planchette of the, the ouija board, moves right, that's what makes it eerie, the seeming of intent without an intender. Uh, and so at best it is eerie, and at worst it's just soulless. Right, right, it's that whatever communicative freight it's carrying is spread so thin that it becomes undetectable amid the plausible sentences and the excellent draftsmanship. So that's my AI art theory.
0:19:17 - Leo Laporte
I think it's interesting. Certainly I like that you're applying information theory to this, and it's almost Newtonian. You, there's no additional energy provided you can't get any more energy out of it. The question I would have is okay, yeah, uh, but is it possible? So, for instance, uh, with duchamp's urinal, a lot of the information was added by the viewer. That's often the case with art, right, uh, and that's that's that eerie thing that you're talking about. The uh intention without an intender is added by the viewer. So there's some information added there. But is it not possible that the AI itself is in fact doing more than simply churning on existing content? Isn't it possible that the AI is making connections that are, in effect, adding information?
0:20:08 - Cory Doctorow
Well, even if you I'm sorry, go ahead. Go ahead, paris. Oh, I'm excited to hear your point.
0:20:13 - Paris Martineau
I just wanted to add something because I see also in the Discord chat people are kind of raising the same kind of comment with regards to I don't know the artist's name, but the famous banana taped to a wall in addition to Duchamp. But I think in both cases the thing that differentiates what could be perceived as low effort in comparison and I say that with air quotes because I don't personally believe that they're low effort pieces of art but things that are perceived as low effort what differentiates that from AI art is that there is an artist that is creating it.
There is an intender, there is someone who is conceived of this and is making decisions with the intent that it is going to be perceived. That does not exist when you have a machine that is purely pattern completing, creating some of these things.
0:21:08 - Cory Doctorow
Yeah, I think that's really well said, and I mean, even if you stipulate that the AI is somehow adding information, the information it's adding, like we do know enough about how the AI works. So what the AI is adding is the most plausible information, right, it's? It's it's finding the median information that would likely accompany this thing, and so, to the extent that art is mostly about surprise, I I I don't think that it will make the art better, but even if you're like, well, no, even the median information is good information to be adding. It is not information about the big numin, irreducible feeling the author or the creator is feeling, and I don't think AIs have big, numinous, irreducible feelings. I think AIs do really interesting statistical work.
0:21:59 - Leo Laporte
I think that's almost a religious belief, though, isn't it? I mean, it's a belief that somehow we are different, uh, that we are, so we somehow have a soul.
0:22:07 - Paris Martineau
You know, uh, I always I'm not entirely sure, because if we are trying to the part part of the modern and not even, I guess, modern part of our long-standing as a collective, as a human society's, approach to interpreting and engaging with art involves subjectivity and involves having an understanding, a perception of a debate around authorial intent. And you can't do that if there is no author, if an author is an unthinking, unfeeling, unmotivated thing.
0:22:45 - Leo Laporte
That is a perspective that we have as humans, that somehow we are different, that that machine is unthinking, unfeeling, but I don't know. I think that's a little bit of a chauvinistic point of view. I'm not convinced that that's the case. At the very least, I think you could say that AI can make connections that a human wouldn't necessarily think of. That might, in fact, be insightful. I'll give you as an example move 37, the famous AlphaGo move in the Go match between Lee Sedol and AlphaGo, that all the humans thought initially was a mistake and ultimately thought this is a breathtaking innovation that no human would have ever thought of. So where did that come from?
0:23:28 - Paris Martineau
no, I mean again I I think that's different than art, though no, it's not, it's creative.
0:23:35 - Leo Laporte
Whatever it is, it's creative.
0:23:37 - Paris Martineau
I don't think it's different than art it's an innovation, I I think, if I think that innovations in terms of things that a machine is being created to do sure, of course it's going to innovate when it comes to tasks or potentially complete tasks in ways that are unexpected I don't think that either of us are arguing that.
0:23:59 - Leo Laporte
I'm not a criterion for saying that there's something valuable coming out of AIs. If it is indeed just a remixing machine, then you're right, it's not meaningful. But if it at any point can make a connection that didn't exist before, that a human might not have seen, that is adding information, that is something of value. Can it do that?
0:24:20 - Paris Martineau
I don't want to put words in Corey's mouth, but for me at least, I don't think I was arguing that AI and technology will never be able to create value as defined as some sort of connection. That didn't exist before. Of course it will. That's what a lot of these machines are trained to do. I think that's different than does this have artistic authorial merit? Sorry, I've interrupted you twice now.
0:24:44 - Cory Doctorow
Corey, no, no, that's okay. I wanted to say just jumping back a couple of steps. It is possible to be a materialist and I consider myself a materialist to disbelieve in the soul and to still think that we are not just in a continuum with AI. In the same way that it's possible to say I don't care how fast these horses that you've been selectively breeding are getting, there is no point at which one of them will give birth to a locomotive, right. It's. The fact that it's getting better at guessing words doesn't mean it's becoming more intelligent or more like us. So I can believe that what we are doing is mechanical in some sense, without having to believe that it is the doing is mechanical in some sense, without having to believe that it is the same mechanism that takes place in these word guessing programs. So so that would be the first response I would have in terms of um, addressing whether this is art and and whether the, the machine, is creative.
On the on the go move, I don't know enough about it, but, uh, I would say that, and I don't know if anyone ever did any forensics on the underlying process that produced that go move. But if imagine for a moment that what we have as a program that does directed, random walks through a problem space, and so what it's doing is it's saying like I can in, like what it's doing is it's saying like I can in, like you know, tractable time compute nine moves ahead, and I am going to, uh, with some randomness, compute as many of these nine nine ahead moves as I can and try and find ones that take me into unexplored parts of the problem space right now. That's a mechanical exercise. It's a thing a human could. That's a mechanical exercise. It's a thing a human could do with a piece of graph paper and a million years, right, but it's not a thing that humans do normally and it is the kind of thing that automation can do for us. That's really exciting, right, like asking a computer to solve, you know, computationally hard programs like variations on traveling salesmen or whatever will often give you results that are quite surprising.
We get this all the time with agent simulations, right, where, you know, famously, the Santa Fe Institute, for years and years, would get called in to consult on how to decrease traffic in cities, and they would recommend choking off traffic on certain streets so that it changed the cadence with which the traffic was reaching the main arteries so that cars would be more broadly spaced and you wouldn't get the ripple effects where when someone touches the brake, the cars all slow down and that creates a continuous eddy where the cars are just slow at that spot for 45 minutes or whatever. And they got there through agent modeling right. They got there through agent modeling right. They got there through through this surprising, counterintuitive, quasi brute force computational exercise and it is a very counterintuitive and, I think you could say, creative way of solving the problem. And the person who wrote that program was being creative, but the program wasn't creative. The program was a tool for creativity of the programmer.
0:27:50 - Leo Laporte
Yeah, this comes back to the kind of the fundamental question that I have. If everything's deterministic, you're right, you know determinism. It says you know, given a starting point and a set of steps, you're always going to get to the final point and that's what coding does. Your description really is reflective of how chess works. It's very much computational. But that's why Go was a little bit more interesting, because the position, the possible moves in Go are so huge that it isn't really able to solve it computationally. Even in a chess program you have to at some point make a positional judgment which is not based on computation. In the case of a rules-based AI, it looks at whatever rules it has and says well, based on that, I say black is ahead by 0.7 points. Go is more complicated. The positionality is very complicated. I think you're verging on the numinal. But that's really the question to me ultimately, and I don't know what the answer is. But I think we are seeing a glimmer and I think Move 37 is an example. I think there are many other examples In fact, in later go programs like Alpha Zero, where it wasn't in fact given any games or anything but the rules of the game and it was taught.
It learned the game itself by playing them, that there are connections that are made. Uh, this was Stephen Wilfrum's contention a few weeks ago on the show that maybe humans wouldn't have seen or wouldn't have come up with. And that is adding in, for in my opinion, that is adding information and proves the point that it is possible for this stochastic parrot to actually be more than simply spicy autocorrect but actually to in in, insert more information into the system. Now, maybe it's just a bit initially, but if you can, but if you agree that it's possible that it's adding information to the system now, maybe it's just a bit initially, but if you can, but if you agree that it's possible that it's adding information to the system, then that changes the equation.
It isn't simply a mechanistic, deterministic well, this than that. It is now adding something, and I think adding information is as good as we're going to get. We can't define consciousness, but I think it's as good as we're going to get. It's what I've always said. For me will be the ultimate criterion of whether a machine is doing something valuable is if it's, if it can create, in the sense that it can add information, it could do something a human wouldn't necessarily be able to do, then it is valuable. Whether it's agiGI or superintelligence is irrelevant to me.
0:30:31 - Cory Doctorow
So again, I want to just tease apart a couple of ideas. So first, I think things can be mechanical without being deterministic. We have lots of mechanisms that we struggle to predict the outcomes of and, very famously, the halting state problem says that computer programs beyond a pretty trivial level of complexity, uh, cannot be predicted. Right, they are not deterministic. This is where bugs come from, right, we.
0:30:54 - Leo Laporte
We don't know what, a, what, what state if we had perfect information, we would know where the bug came from. But I would agree with you that that's the thing. That's what's going on, uh, with these machine learning models, with these transformers. That is what is different is it's? It's somewhat of a black box and we can't, it's not deterministic, we can't figure out what's going on I believe it's touring church.
0:31:16 - Cory Doctorow
Who who defined, uh, the halting state problem? But their contention and I don't have the math to follow it, but their contention is that, no, even if you understand the program perfectly, you cannot predict how it will turn.
0:31:28 - Leo Laporte
Okay, okay uh, that, that, that is not an information, it's not a lack of information, it's just not predictable yeah halting state problem is what you're looking for. Oh sorry, yeah you're still looking at my screen, I don't know how to get it off that might you know what? I've stopped using the zoom uh app because you're on the web and maybe you don't have the choice. Yeah, do you see a three up right now at all?
0:31:50 - Cory Doctorow
yeah, I got, yeah, yeah yeah, okay, across the top. So yeah, that's fine, as long as you see the three up remember, that's what's going out and you're gonna, you're gonna see my screen for your screen a spotlight, for reasons I don't understand.
0:32:03 - Paris Martineau
You're gonna see how much leo uses perplexity for every single thing on this show it's supposed to be secret.
0:32:10 - Leo Laporte
No, that's just my search. I don't use. I no longer use google for obvious reasons I use kagi. I use kagi for a long time. Um, I stopped paying 25 bucks a month to use kagi. Well, that's for your whole family. It's 10 bucks a month for just you.
0:32:22 - Paris Martineau
I was the only one using it, so, okay, 10 bucks I was paying 25, most expensive subscription possible, of course, and I paid 20 for perplexity.
0:32:31 - Leo Laporte
So it's not free.
0:32:32 - Cory Doctorow
But um, so that that was the first thing. Is that I I don't know that that just because things are mechanical, it implies that there is design?
0:32:40 - Leo Laporte
is dividing by zero a halting state?
0:32:42 - Cory Doctorow
no, that's just a mathematical problem. Okay, no, halting state problem is that it's yeah, there we are. So again, we are now as we look at, I'm looking at your screen and we are now at a point where the math is going over my head so uh as the son of a mathematician, I know what math I'm not qualified to understand this is firmly in that realm, uh, but I my my lay understanding of it is that the halting state problem says that we do not know how a program will terminate until we run the program.
Uh, so it can be mechanical without being deterministic, is my broader point okay so it doesn't.
Just because I think we are mechanical, it doesn't stipulate that okay and the second thing I wanted to say about adding information is, um, surprising you with juxtapositions that you hadn't contemplated is indeed a very fruitful creative exercise. You know, there's the exquisite corpse game where people write sentences without knowing what sentence comes before it, and so on. Um, when I was at the Clarinet and science fiction workshop 30 years ago with Nancy Kress, the great writer, she said you know, we can plot stories that begin with a person in a place of the problem. So each of you write a person, a place in a problem on a separate sheet of paper. We're going to scramble them up and hand them back to you and you're going to try and write a story to that prompt. So these random prompts can be extremely creativity provoking.
Prompts can be extremely creativity provoking. Um, you know, people talk about how uh laying a tarot uh helps them uh clarify their thoughts or make novel connections they hadn't made before, even if they don't believe in anything metaphysical. Going on with the tarot deck and there are storytelling cards that are like this too, very famously, uh, you have uh Brian Eno's um uh oblique strategies deck, which is full of prompts that he used when Talking Heads and Roxy Music were recording like be the first person to not do something that no one else has ever thought of not doing before. And all of these things are super interesting.
Imagine how a machine that makes unexpected connections on the basis of some algorithmic whatever whether it's statistical analysis of a huge corpus of language or whether it is some interesting stuff where you're, you know you're trying to hill, climb towards a certain maximum, but you're traversing the information scape horizontally because you think you've reached a local maximum, you're trying to find somewhere to go up, so you get these weird unexpected connections and then the human reviews the connection and goes aha, that's really interesting. And that's what I think you have to do when we're talking about art, because art does not have an objective.
0:35:20 - Leo Laporte
Uh, well, that's why art is a bad, maybe a bad way to measure this, because there is no. That's why I like Go, for instance. There's no measurable victory or no end right, but we have very clear goals with a lot of other human endeavors that we can say, yeah, you did it or you didn't do it or you didn't do it so and I can.
0:35:46 - Cory Doctorow
I can believe that you can have a computer program that self varies where there is another program that says you got closer, that time try some variations in fact, that's.
0:35:51 - Leo Laporte
That's what these reasoning models are doing these days, right which works when you have an objective function.
0:35:56 - Cory Doctorow
so but but it also, you know, we have to remember the ai companies have been less than forthcoming in terms of their, their we don't know.
0:36:06 - Leo Laporte
It's a black box.
0:36:09 - Cory Doctorow
DeepMind and its material science breakthroughs, where they said they've done. 800 years of material science breakthroughs generated something like millions of synthetic materials.
0:36:19 - Leo Laporte
Most of them were worthless right.
0:36:21 - Cory Doctorow
Well, no Zero of them in a random sample were worth anything either either a thing that you could make, so most of them required that, uh, they would only exist at absolute zero, and the ones that that could exist in tractable temperatures, uh, either had no novel properties or no useful properties, or both so, uh, material scientists were not able to use any of that generated zero for me, zero, zero I mean it's a huge number of materials.
So they did random sampling, they tried to find a statistically relevant sample, a representative sample, right, and then they examined each of them. And they found zero, so that would be.
0:36:58 - Leo Laporte
For me, that would be my Turing test measure of uh, something happening is if they, if a machine could come up with a new material that was a create, a creatable and be usable that way, and that would validate by itself, validate the work done on these things. Right, it would be of use.
Sure, I mean, it wouldn't necessarily be of a hundred billion dollars worth of you, I agree we don't know, maybe, but if you came up with a cure for cancer, yep uh, that would be of use we're always saying, yeah, if ai came up with a cure for cancer, it'd be of use.
0:37:33 - Paris Martineau
That could be said of a lot of things, but you don't know until you try came up with I understand for cancer that'd be of use, but we're not giving her 300 billion dollars but yeah, but it seems like it's worth trying, does it not?
0:37:45 - Leo Laporte
or are you saying it's not? Do you feel like it's not worth it? I?
0:37:48 - Paris Martineau
think it's certainly worth it. I question is wow, gizmo heard her name and is now sitting up in the background of my shot um.
0:37:54 - Leo Laporte
See, she says I got a cure right here.
0:37:56 - Paris Martineau
You just never listen to me I just never give her 300 billion dollars and she's holding out um I think it's certainly worthwhile. I think that the thing that is still that we still haven't figured out is whether or not it's worthwhile spending an unprecedented amount of capital and resources and devoting an unprecedented amount of collective attention worldwide.
0:38:19 - Leo Laporte
I'm getting two head up. My watch just told me it's going to call 911. So I'm going to calm down right now.
0:38:26 - Paris Martineau
Okay, to help you calm down, I'll uh, I have to do an ad that's how I'm going to calm down.
0:38:30 - Leo Laporte
Don't introduce anything else. Hold on hold that thought. No, you can do it in a minute. I do have this article from mit's technology review saying that deep mind did in fact create seven I'll find you that.
0:38:42 - Cory Doctorow
I'll find you the, the article that reviewed that, that conclusion oh, okay, interesting.
0:38:46 - Leo Laporte
No, I believe you. Uh, this was from a paper in nature. Um, yeah, okay, interesting, uh. But first a word from our sponsor. It's so good to have you, corey dr rose, here. It's so good to have you back, paris martineau. From the information, jeff jarvis is kicking himself that he couldn't be here today. Uh, corey's new book is now out, picks and Shovels. You find it at thebezelorg or craphoundnet. Is it net or com com? Pluralistic net for his blog, our show today. I'll tell you what I can prove to you. This tool is worth every penny you pay for it. It's brought to you by Monarch Money.
Nowadays, keeping track of your money is a chore. A chore a computer can do quite well. Finances can be very tricky, very confusing. Monarch Money remember that name. Monarch Money acts like your I don't know your personal CFO giving you full visibility and control. Remember that name. Monarch Money acts like your I don't know your personal CFO, giving you full visibility and control so you could stop earning and start growing. It's more than your average budgeting app.
Monarch Money is a complete financial command center for your accounts, your investments and your goals. I put everything in there, everything, and makes it, by the way. It's very easy. It took me 10 minutes to set it up and now I'm doing more than just managing my money. I am looking at what's going on. I am seeing where I can make changes, make improvements. I'm building my wealth. And, by the way, you can do it, too, with 50% off for your first year if you go to monarchmoneycom and use the code I am. There's so many valuable things. I get alerts from monarch money. It also, with the stock market being a little nutty right now, it helps me see what's happening, where there are options, the. It really helps me manage my finances to build the life I'm planning for, and as I approach retirement, that becomes more and more important. But I think it's important for everybody of every age. Without a clear financial picture, you don't know where your financial dreams are. They might be way out of reach.
Monarch Money makes managing money simple, even for busy lives. There's no data entry. Once you set it up, it automatically keeps track of everything on your computer, on your phone, on your iPad, on the web. Wherever you are, you could see everything your accounts, your credit cards, your investments all in one dashboard. You'll always know where your money stands, without the hassle, and it's great. It's great for budgeting. I don't budget, I suppose I should, but if you wanted to set up a budget without all the tedium of it, this makes it so much easier to track your spending, your savings, your investments. You could focus on what matters most making your biggest life goals a reality, saying I'm going to spend a little less here so I could spend a little more there. It's the financial tool people actually love Over a million households now using Monarch money.
Wall Street Journal said it's the best budgeting app of 2025. I would agree. It's the top recommended personal finance app, recommended by users and experts 30,000 five-star reviews. I love it. I love it. I signed up for a year immediately. Visit monarchmoneycom, get control of your overall finances with Monarch Money and when you use the code I am for intelligent machines at monarchmoneycom in your browser, you'll get half off for your first year. 50% off for your first year. That's another way to save, isn't it? Monarchmoney? Come to think of it. Com. Use the code I am. I promise we'll thank me later. Monarchmoneycom Offer code I am for 50. 50 off for your first year. Cory has sent along the the article from um. Is it who? Who is that? Acs, pubsacs artificial intelligence driving materials discovery question mark it's.
0:42:43 - Cory Doctorow
It's the uh american chemistry society or chemical society's uh journal. That makes sense. I mean, it's not nature, but it is the leading journal in material science I think it's kind of the response to the nature article in fact.
0:42:56 - Leo Laporte
Yeah, that's right. Um, okay, so there's a debate that's interesting, um, and I I think I would trust the chemists to say it's. They said we believe the experimentalists, uh, would find it helpful if the results are presented in a more organized manner rather than a seemingly random walk through the periodic table. Okay, so there's some suggestions for how to make it better? Uh, do they deny, though, the possibility that new materials could be created by deep mind, don't deny that at all.
0:43:26 - Cory Doctorow
But I think that line about random walks it's not just, uh, it's not just a burn, right, it is a. It is a critique of this um, brute force approach, yeah, and really critique of the like what the objective function that they're asking the ai to have. So I could conceive of a world in which you say to the ai go find me materials but don't tell me about them unless they satisfy these you know criteria. That would mean that they have a colorable case of being useful. But I think if that were the case then your press release would be like we ran a program for a year and we found one thing we might that might be useful and not.
0:44:02 - Leo Laporte
We ran a program for a year and we found a million things that are have advanced the science by 800 years, which was the original claim from deep mind that they do say in their very last sentence while we are confident the tools of artificial intelligence and machine learning have a bright future in the fields of material discovery, more work needs to be done before that promise is fulfilled. So they they don't deny the potential value. Yeah, they're not saying like automation won't, won't produce to be done before that promise is fulfilled.
0:44:25 - Cory Doctorow
So they don't deny the potential value of this. Yeah, automation is good. Yeah, they're not saying like automation won't produce dividends. Yeah, but you know, I think back to what Paris was saying before. Even if you stipulate that after $800 billion, or however many billions of dollars that we're going to spend on AI, you get one cure for one kind of cancer, we have to make sure we're not excluding the middle here and the other possibility between we spent $800 billion and we got one cure for one kind of cancer with AI and we spent nothing and we didn't get a cure for cancer. Is we spent some fraction of $800 billion on some other technique and we got one or more cures for one or more kinds of cancer too? Right?
0:45:04 - Leo Laporte
Yes, it's to some degree a zero-sum game. You don't have an infinite amount of dollars to spend, and so you want to spend it in the place that will be potentially uh the most useful. But how do we know?
0:45:16 - Cory Doctorow
well, we don't. But, um, we do have people pursuing, uh, various techniques who are starved for cash, right, and who have had good results, and particularly now in the Doge austerity era, right, we have a lot of public sector research that, unless it connects itself to AI in order to be buzzword compliant, struggles to find funding, even if it's already had successes in the past. So I would say that, given, given that we have taken capital from those people and in some meaningful way reallocated it to a much more speculative AI technique, that that capital is not being rationally allocated, even if it pays off. Right, you can, you can hit the craps table and, uh, you know, hit sevens three times in a row and walk away with a lot of money. That doesn't make it a rational bet, right, just means that you got lucky and maybe, maybe we'll get lucky, right, uh, and you know what do you?
0:46:13 - Leo Laporte
propose? Should we not be spending money, time and let's face it, uh global resources on ai?
0:46:22 - Cory Doctorow
I think we should approach ai as a useful piece of utility automation and not as a transformative technology that is going to be worth $13 trillion, which is what we're doing right now.
0:46:33 - Leo Laporte
Yeah, sam Altman said we need to spend at least a trillion to get to where we want to go, and that's a lot of money.
0:46:39 - Paris Martineau
I think that there are most things in the world, be it technology or otherwise, you could argue, would be world changingly transformative if you spent a trillion dollars on making it so.
0:46:54 - Leo Laporte
Right? Where do you stand, Corey, on the threat that AI poses to humanity?
0:47:00 - Cory Doctorow
So I think that, once again, as a materialist, I think when you're critiquing what people do, you should think about their material circumstances. So I think that ai companies have a bunch of lines of business in their, in their business plan and their powerpoint deck that are threatening to us, like um, automated judgments about who should be an immigrant and who should be turned away from the border, or who should be denied bail or not, or whether your kids should be taken away, or whether you get a job, or whether you get a lease, whether you get a loan. Those are all things where AI can significantly harm people, and probably is right now.
And probably are, certainly are. I mean, there was just an academic who was black bagged uh by um, uh, you know, donald trump goon squad, because an ai generated uh pro-israeli newspaper called her an anti-semite because it said that she had attended this rally and that rally it had the speaker, and that speaker had been somewhere else and that person had been affiliated with hamas. Therefore, blah, blah, blah, right. So you know. We already have a world in which ai based nonsense, although the flaw there is abandoning due process, because I told you so, blah, blah, right. So you know.
0:48:05 - Leo Laporte
We already have a world in which AI-based nonsense, although the flaw there is abandoning due process because AI told you so.
0:48:10 - Cory Doctorow
Indeed. So the thing is that AI really only works as an automation tool that saves money, as opposed to one that improves outcomes, if you abandon due process. So I see on your screen you keep bringing up this radiology story.
0:48:26 - Leo Laporte
This is from the Washington Post earlier this week. Ai hasn't killed radiology, but it is changing it, the claim being it has been helpful to radiologists and they are getting better outcomes with the help of AI.
0:48:39 - Cory Doctorow
So I am really interested in this because I have an extremely treatable form of cancer. So I am suddenly someone who's a lot more interested in cancer than I was a year ago, cory I did not know this.
I'm sorry I've written about it. It's fine, it is totally fine. I'm having some minor radiotherapy in june and the prognosis is excellent. Okay, it should be fine, but, but, um, this is a subject I'm I'm now very personally interested in, in a way that used to be a lot more abstract. So here's what I think there are, and it appears to be the case.
There are algorithms that are better at spotting certain kinds of solid mass tumors than human radiologists. Right, so the ones that the humans miss, the software can pick up. It doesn't imply the other way around, by the way, that the ones that the humans get the robot will also get just as well. But if this is the case, then you could imagine a hospital saying okay, we have a radiologist and right now that radiologist reviews 50 x-rays a day and we are going to add some software to it that's going to do a second opinion on all those x-rays, and when they disagree which is going to happen about once a day that radiologist is going to go back and recheck their results. So, now that radiologist is going to do 49 x-rays a day, there is no hospital administrator who is buying AI because they want their radiologist to do fewer scans the pitch from an AI salesman is fire most of your radiologists, have the machines, do all the work and have the human act as what Dan Davies calls an accountability sink, or what's also been called a moral crumple zone.
Person who, the human being who is supposedly in the loop, who put their initials next to the robot's conclusion, but who was asked to work at a pace where there was no conceivable way they would review the robot's findings.
So, again, I think there's like a socially determined outcome, not a technical one. But I think that, like back to business plans, nowhere on the business plan of the ai company making the radiology bot is the thing where they say well, this is how much money we expect to make from hospitals that are going to increase their costs, and this is how much we expect to make from hospitals that will decrease their costs. They only have a line item for increasing their costs. And so when we find ourselves arguing about things that AI companies don't plan on making money on, like the AI waking up, becoming super intelligent and turning us all into paperclips, or even AI disinformation and elections there's no AI business plan. That's like we are going to make so many millions this year selling AI disinformation tools for the upcoming election. So when we focus our attention on that, we focus our attention on a thing that, even if we all agree that this is bad and we have to do something about it, no AI company is poorer as a result.
And when you have companies that are spending the largest amount of money any companies have ever spent in the history of the world and all the debate about their activities focuses on things they never plan on making money on, I'm inclined to think that those two things are related that those ai companies are asking us to train our focus on things they never plan to turn into a product, so that we are not focusing on the things that they're actually out there selling right now that are not fit for purpose and to complete the thought on this washington post article, they do say most ai detection products produce false positives that radiologists are responsible for following up on.
0:52:08 - Leo Laporte
Here's a quote that says ai that's focused on detecting abnormalities can actually create more work for the radiologist.
0:52:16 - Cory Doctorow
Uh, so I think your moral crumple zone analogy is is quite I can write you a one-line pearl program that catches every tumor that the radiologist misses, right, all it does is says that every x-line Pearl program that catches every tumor that the radiologist misses. All it does is says that every x-ray it examines has a tumor. By definition. It's going to catch every tumor that the radiologist misses.
0:52:37 - Leo Laporte
Yeah, really an excellent point Hospitals hate, him.
We're talking to Cory Doctorow, who's always, I have to say, always been a real refreshing antidote to some of the frothiness that you see in technology, and AI is as frothy as you can get. Meta just got in a lot of trouble because they basically played, they gamed, the benchmarks with their release of Llama 4. This is from the Verge. With llama 4, meta fudge the benchmarks. This confirms, by the way, corey, what you've been saying to appear as though the new AI model is better than the competition, but in fact, the model that they uh, they submitted was is not publicly available. A um and uh, and in fact, it may be that it was trained on the test. In effect, um wait.
0:53:32 - Paris Martineau
So how did it fudge the benchmarks?
0:53:34 - Leo Laporte
in fine print. Meta acknowledges the version of maverick tested at lm arena isn't the same as that available to the public. According to meta's own materials that deployed ex. This is all from the verge, an experimental chat version of Maverick that was quote, specifically optimized for conversationality hi, this is Benito. Like we should never trust meta statistics ever. I think you're right.
0:53:57 - Benito Gonzalez
We've learned that lesson we learned that in video hands up everyone who pivoted to video.
0:54:04 - Leo Laporte
Well, what's interesting, though and Meta knows this is that the news that they were the best at LM Arena traveled the world, and I think that this disclaimer probably is not as well known. It's still put its pants on.
0:54:21 - Cory Doctorow
Well, you know, I gave you a link in the chat to Goodhart's Law, which is that any benchmark becomes a target and then ceases to be useful as a benchmark.
0:54:30 - Leo Laporte
Oh, I love that. And that does happen, doesn't it? We've seen it again and again.
0:54:34 - Cory Doctorow
Every time. Gdp is like the embodiment of that. You can make number go up without making national prosperity go up. Yeah For sure. Moments like this are habitual liars, and memorizing the test is a great way to pass the test. You know, this is all that stuff about ai's passing the sat. It seems really likely that ai's were trained on the lsat.
0:54:57 - Leo Laporte
Yeah, that doesn't mean they're a lawyer so is there any justification for uh continuing ai research, continuing work, uh, or is this just a complete waste of money?
0:55:12 - Paris Martineau
I mean I wouldn't say there everybody should shut down absolutely everything. I think that's certain. I don't think it should be as binary and black and white as I don't think the two also what?
0:55:23 - Leo Laporte
how do you respond then? What do?
0:55:25 - Paris Martineau
you have dollars yeah, you spend zero. I think there's a way to treat this like any other technology and invest a reasonable amount that the market can bear and that we think could you know uh produce results and be aligned with the amount of capital invested cory, what do do you think?
0:55:44 - Cory Doctorow
I think that's completely right. I mean, there are lots of things that people have done with AI that I find super interesting. I think maybe when the last time I was on Twit, we talked about Fourth Ease Vinegar Collective, which is this group of rogue pharmaceutical chemists who are on this year. Are they using AI? Who are on this year? Are they using ai? So they have made a chemical synthesis uh um answer machine search engine. So if you go to wikipedia and you look up any small complex organic molecule so like the one shot hepatitis c cure, for example, it has a what's called a smile code, which is a short string of ascii and there's even a button to copy it to your clipboard. You go to their search engine, you paste it in. It has ingested the entire organic chemistry synthesis literature and it says you get these precursors and you get these reagents and you combine them in this order, at these temperatures, in this way, and this is how you synthesize this molecule and it's extremely reliable. Like, like. That's a super interesting application and they pair it with an open source uh compounding pharmacy robot that's basically functionally equivalent to the ones that uh compounding pharmacy would make, would buy for fifty thousand dollars, but you make it yourself with a raspberry pi and some peristaltic pumps for uh like five hundred dollars. Have you built this for yourself? I have not, but I watched them assemble one on stage from scratch during the talk and then take requests from the audience for molecules.
So the first thing they made was the one shot hepatitis C cure, but it was DEFCON. So there were a ton of people who were on various kinds of hormone therapy because they were trans. So they made testosterone and estrogen. They made a medication, abortion, they made. What else did they make? They made, oh, vyvanse, which is the ADHD drug that was having supply chain problems where if you go cold turkey on it you get seizures. So they made their own Vyvanse. So they were taking requests for molecules on stage and making it. You can go and download them.
0:57:43 - Paris Martineau
I'm on their website. They made Plan B.
0:57:46 - Leo Laporte
Yeah, yeah, a version of Plan.
0:57:48 - Paris Martineau
B, that has also extra stuff in it Awesome.
0:57:52 - Cory Doctorow
Yeah, no, they're amazing and and their goal like they saw the trump administration coming and this was their goal was to say that you, you should never be relying on the pharmaceutical industry or the government to get the drugs that you need to stay alive. And so they they made this open source hardware robot. It was the most interesting uh presentation I've ever seen at defcon. I've seen some pretty interesting presentations at DEF CON over the years. Are you still working with EFF?
0:58:18 - Leo Laporte
I am I thought you were.
0:58:21 - Cory Doctorow
I'm a special consultant, special advisor. I'm an activist with EFF.
0:58:26 - Leo Laporte
Yeah, so you write. In the 23 years I've been at EFF, I've been privileged to get a front row seat for some of the most important legal battles over tech and human rights in history. This is from your article that the EFF is suing Doge.
0:58:41 - Cory Doctorow
Yeah, we were the first people to sue Doge. We're suing them on behalf of two unions representing federal workers and we brought a case under the Privacy Act to say that by the Office of Personnel Management giving access to their records to DOJ that they had violated the Privacy Act. And the court dismissed the government's attempt to get our case thrown out but did say that we have to bring our claims under the Administrative Procedures Act. So this is getting a little esoteric, but the Administrative Procedures Act just determines how federal agencies are allowed to work, but you're not usually allowed to use it until you've exhausted other remedies, and what court said was that we could skip straight to an Administrative Procedures Act claim, which is good, it's very good. And they threw out all of the government's objections and they said that all of our claims are adjudicable and so as we move forward into the actual trial, all of our claims will be heard who adjudicates these procedural?
0:59:40 - Leo Laporte
so this is a court process.
0:59:42 - Cory Doctorow
This will be heard by a court. I don't think this is an administrative law judge that will hear this. Yeah, uh and you're now getting out of my legal knowledge. I am an activist, not a lawyer, so yeah, well, they got some very good lawyers at eff?
0:59:55 - Leo Laporte
we sure do, um, and they're joining a chorus, of course, of lawsuits, some of which have been thrown out. Scotus uh, I think rejected somebody for lack of standing yesterday so that's what we want. We want a standing challenge yeah, yeah, to prove that you have the and, you would seem, the unions would have standing but right, but this is always a problem.
1:00:16 - Cory Doctorow
I mean, look, one of the problems is when everyone is implicated. You have to prove that you are implicated, right? So we had this problem with our case Jewel and another case that we brought as well. That were both NSA spying cases. So in 2006, an ex-AT&T technician named Mark Klein walked into our offices around the corner from tech TV studios on Shotwell street, uh, and he said um, I just retired from AT&T and my boss made me build a secret room in the full.
1:00:46 - Leo Laporte
Oh, yes, he just passed the other day. Yeah, right.
1:00:49 - Cory Doctorow
Yeah, and we put a beam splitter in and we let the NSA wire tap the whole internet and we brought a case. We brought two cases. That was your case, and the thing that the court wanted us to do was prove that our clients had been spied on by the NSA. But the NSA said all of the evidence that they'd been spied on was secret. So we had all these secrecy challenges. One of the cases lasted about 10 years, the other one lasted about 15. Both of them were eventually dismissed for lack of standing. So this is often a problem. When you have these, these really cases that implicate everyone is, you have to find evidence of particularized harm, not general harm, and so there's this weird way in which the courts say, no, they spied on everyone, so you can't prove they spied on anyone. And that's that's really weird. It's a terrible cash.
1:01:38 - Leo Laporte
22 yeah, and I would agree with you that using ai as doge is using it, uh, to decide who to fire, to rewrite the code for the social security administration, uh, and it looks like now the irs is probably a bridge too far. As much as I like ai, that's not where ai. That is a place where I can cause a lot of harm or having a craft, a tariff policy for that matter yeah, I, you know, I don't.
1:02:05 - Cory Doctorow
Who knows where trump got his uh mathematical formula we think a human being sat down and just went through all the cctlds and did the math.
1:02:15 - Leo Laporte
No, that's an ai but somebody came up with the idea of oh, we'll take their deficit, we'll divide it by, you know maybe, yeah, maybe, or maybe said ask to chat gpt.
1:02:25 - Cory Doctorow
So tell us what make this chart for us maybe they said to the chat gpt, though maybe they said, tell me what a good tariff policy would be. We don't know, we don't know.
1:02:36 - Leo Laporte
A number of people have tried asking that, giving that prompt to all the major models, and all of them came up with the same mathematics as far as I know, so maybe that's what they did. It was telling that the chart was in some sort of AI order, not alphabetical or numeric.
1:02:56 - Cory Doctorow
That's the kind of thing ai will do, you know yeah, I mean, look if it it didn't quite have the phrase as a large language model.
1:03:05 - Leo Laporte
I am not capable of advising you on this, but it was close, it was pretty fucking, excuse me, pretty dang close well, the good news, the good news is uh, as of uh, about five hours ago, uh, most of those tariffs have been walked back paused pause. Well, pause, yeah, I mean. The uncertainty remains, but the tariff against china is now ratcheting up past 90 percent 90 percent for small packages from sheen and antenna it's gonna put she in and Timu out of business and it's got to damage Amazon.
1:03:40 - Paris Martineau
I mean most goods that come from China or pass through it. All that fast fashion in your closet there, not even just fast fashion like small, the parts of anything that ends up assembled in America.
1:03:54 - Cory Doctorow
As well as everything you might use to build factories in America that can do high-tech, high-precision manufacturing.
1:04:00 - Paris Martineau
We got 90 days to build those factories, so it'll be fine actually.
1:04:05 - Cory Doctorow
Okay, reshoring is hard, so I had an interesting conversation with Bunny Wang who's a real hardware hacking virtuoso about the CHIPS Act and he predicted that the CHIPS Act would fail.
And I asked him why? And he said look to make an extremely advanced microprocessor, something with sort of four nanometer features. You need to have a country with an amazing education system and a useless passport, and that's Taiwan system and a useless passport. And that's Taiwan. Because the job, the process of making those four nanometer features, starts with you getting into a bunny suit for eight hours at a time, without a pee break, three shifts a day, with your PhD in electrical engineering. While you oversight, see a machine whose first step is vaporizing tin into an evacuated chamber, tracking each of the vaporized droplets with one vision system, hitting it each one with a laser, with one kind of laser to flatten it into a coin shape, hitting it with another laser to vaporize it so that it creates the right wavelength of light to etch the wafer. But the wafers have to be prepared in like a 10 hour process so that they're within one atom of each other's weights and they spin on a turntable like this and if they're like off by a bazillionth of a gram, then they are out of whack and they start to. They start. You know your, your features aren't etched correctly.
And he said, like you can only do that if you have a PhD in electrical engineering, you need to do that. You need to have that level of knowledge to keep that machine running. And they pay $55,000 a year. So if you know anyone with an American PhD in electrical engineering who wants to spend from midnight to 8 am in a bunny suit, with no pee breaks, babysitting that machine for $55,000 a year, then sure. But like we are a ways off from reshoring that capability. Even if you think that reshoring it is a good idea, which I do, it's not going to happen tomorrow. It's going to require a lot of very complex, long run phenomena to play out.
1:06:06 - Leo Laporte
I've paraphrased you. You actually mentioned this on a Twit episode some weeks back and I paraphrased you many times since then. You're just not going to get these people and the important point there was the weak passport. They can't leave taiwan. Many of them have been imported into arizona for the you know these tsmc factories, but whether they'll stay is another matter and whether they'll do the job for 55 000 here in arizona is also a question.
1:06:34 - Paris Martineau
Yeah yeah, you go through almost a decade worth of schooling all to stay in a room in the hottest state. Uh, wearing your bunny suit while having to pee yeah, yeah.
1:06:48 - Leo Laporte
Even even the odd trip to the grand canyon won't make up for it when when you when it's, you know, nove November and it's 98 degrees outside, you might be regretting your choice, yeah you all saw the um, the video that uh was made in China uh mocking the American, uh the American worker, uh trying to do the factory work that the chinese uh do uh so cheaply uh currently I didn't see this, but yeah, this is ai.
1:07:21 - Paris Martineau
This is a lot of memes of uh mba bros wearing their puffer vests sitting at a uh to make iphones yeah, yeah, uh.
1:07:33 - Leo Laporte
This is not a good job. This is not a job Americans will do. Yeah.
1:07:38 - Cory Doctorow
And maybe that you know, maybe there are, like I think, reshoring some capability to manufacture critical goods is a good idea to as a hedge against climate shocks, geopolitical shocks and so on, even pandemic, future pandemics. I just don't think you do it by shutting everything down. Right, like that project would require an enormous amount of technology transfer from China and you don't get that while having a belligerent relationship with China. You know, china did not crack down on US big tech until it had incubated a Chinese big tech sector. Right, they weren't like, oh no, we're not going to let Facebook in here ever. They were like, oh no, we'll let like Google and Facebook in here long enough to figure out how they work, allow local firms to, you know, clone their best features and then we'll kick them out. That's what we should be doing if we want to steal a march on China. That kicking them out before we figured out how to do what they do, that's like that's a. That's a bit of a whoopsie.
1:08:43 - Leo Laporte
Bad planning, whoopsies? Yeah, you don't think robots will be taking these jobs?
1:08:50 - Cory Doctorow
Maybe, maybe robots will do some of these jobs. Manufacturing the precision robots is also a precision manufacturing.
1:08:57 - Leo Laporte
Oh, that's right, Somebody's got to make the robots, oh yeah.
1:09:00 - Cory Doctorow
Right and yeah, like look, I think automation is real and good and I rely on automation and I have benefited from it immeasurably and, like I remember, you know compilers that could do some error checking and how transformative it was. And I had a friend tell me a couple of years ago that he was using Copilot to write CSS, which is just like a giant pain in the ass to write valid CSS that works across multiple browsers, and the LLM was doing it really well. And I was like that's really amazing, was doing it really well, and I was like that's really amazing and I said, you know, if you had told me, my latest compiler has a CSS debugger. That's really good. I would say that's very exciting. But what you're telling me instead is that the, the job of programmer, is now threatened because I have some error checking for CSS which I don't think follows. I think that's great If you're trying to raise a trillion dollars, to be sam altman and steal everyone's eyeballs, but it's not great if it's not like realistic, nor does it make it a good investment he's talking about world coin, which is the orb that you scan your eyeballs into and then somehow the world works better these guys gotta respect the orb yeah, we're very pleased domes domes orbs
1:10:22 - Leo Laporte
they're obsessed with spears, yeah yeah, I love your, uh, your collage with uh, it's a sphere with donald trump's hair, is that's right?
1:10:33 - Cory Doctorow
yeah, so donald Donald Trump. You know, one of the best things about Donald Trump being president is there's a million public domain photos of them now, because all the things that white house produces are in the public domain, so I can just use those without even having to credit them, which is great. Oh, that's fantastic.
1:10:48 - Paris Martineau
I do think it would be kind of fun to have like a brush tool that just paints the texture of his hair trump hair.
1:10:55 - Cory Doctorow
I have some. I have some trump hair pngs, but but I've been, I've made my own from the official presidential portrait I think of a.
1:11:03 - Leo Laporte
Uh, you're right, a photoshop brush of trump hair would be very useful it would.
1:11:10 - Paris Martineau
Maybe ai can get on that and the gimp could do it too.
1:11:14 - Leo Laporte
Yeahimp is actually getting better and better. Do you use the newest GIMP that just came?
1:11:17 - Cory Doctorow
out. Yeah, version three, right, the first big update in 20 years. Yeah. It. It is a bit like the time we renovated the kitchen I couldn't remember where the cutlery was. All new, all new in the wrong spot and it's lost a ton of work.
1:11:32 - Leo Laporte
you have to really hit save every couple of seconds because, uh, it is not as stable as gimp 2 was we are very fortunate jeff jarvis has the week off but a quarry doctor has agreed to sit in and it's so nice to have you. Uh, sci-fi author, brilliant author, the bezelorg for his new books, uh, including the one that just came out, picks, andicks and Shovels, the origin story for Marty Hench, forensic accountant, and the work he does at Pluralisticnet. There it is Picks and Shovels, a Martin Hench novel, paris Martineau's back. It's great to have you back. We missed you so much. She's working at the information. I didn't get a chance to credit you. I credited you. I didn't get a chance to thank you in person for that excellent section 230 article. You wrote just on your way out the door. Um, thank, you.
1:12:19 - Paris Martineau
That was really well last and also a hell of a. I mean, everything's a bit in stasis now. Uh, I think, even as I was reporting, that folks were like oh you know, we could release this bill as soon as this week, but that would uh rely on nothing calamitous happening in uh global and national politics that week and what would you know a lot of what a shock are happening and how did?
1:12:43 - Leo Laporte
that, uh, that's amazing that that, uh, that happened yeah yeah, we had.
1:12:47 - Cory Doctorow
We had like an all hands like let's get to work on this bill thing at EFF. And I I asked my colleagues, you know, like, do you really think a bill is going to move in Congress? And they were like, yeah, no, we're told by all of our, our friends, you know our legend. Our legislative analysts were like we're told by, you know, the staffers, this is moving in Congress.
1:13:06 - Leo Laporte
And I'm like, okay, I'll, uh, I'll pitch in, you know, but like then it didn't move in Congress well, and the reason it's scary and the reason you point out in your article is it's a bipartisan move.
1:13:17 - Paris Martineau
I mean, yeah, that's, that's what's scary and I knew I like phrased that sentence very specifically because I knew some people were going to be like there's been bipartisan efforts to reform section 230 before.
I'm like yes, proposed, never introduced before right we've never had a comprehensive bipartisan effort to not just reform section 230 but repeal it whole hog introduced, and that's, that's scary. I think the main thing is are we going to get a? Uh, I I assume probably sometime in the next four years there'll be a period where there's a little bit, where there's, you know less it's a little calmer.
You think I hope so, but I don't know, maybe we, you know less it's a little calmer, you think. I hope so, but I don't know. Maybe we won't, maybe it's just going to keep getting worse from here and then we won't have to deal with this, we'll have to deal with other bad things well, if you care about our chats in our various streams, we have eight of them going at the same time discord, youtube, twitch, tick tock, xcom, linkedin, facebook.
1:14:16 - Leo Laporte
If you care about our Twit forums at twitcommunity, if you care about our Mastodon instance at twitsocial, if you care about leaving comments on our videos, all of that would go away the minute. Section 230 goes away. Because I'm too small, unlike Facebook and X and Google. I'm too small to suffer a barrage of lawsuits. To have to defend a barrage of lawsuits, paris, you estimated each one would cost me a million dollars to defend. I Couldn't afford one, so we would shut down.
1:14:43 - Paris Martineau
We would immediately have to shut down all of our user contribution areas and I because we couldn't is this plan is being made with, kind of the reason why they're angling or at least people tell me they're angling for a full repeal. It's like, oh, that will. That may negotiate send the tech companies into negotiating, but that means the people.
1:15:06 - Leo Laporte
How's that going with tariffs?
1:15:08 - Paris Martineau
First of all, let's assume that'll work, which I think is already a crazy assumption. Let's assume somehow that works. The people and companies that are going to be negotiating are probably going to be the just the biggest players, not me, not me it was not going to be negotiating no, not that little fixie site in the uk.
1:15:26 - Leo Laporte
That is already gone. Yeah, and it's really scary when you see dick durbin, lindsey graham josh, hollyha Blackburn, sheldon Whitehouse, amy Klobuchar, dick Blumenthal and Peter Welch all in the same room. That should scare you.
1:15:41 - Cory Doctorow
Hawley and Klobuchar are there because they think it's doing antitrust work, and what they don't understand is it's creating a compliance moat that will keep small market entrants out and only allow large firms to function. The other thing that's going on in the background there's a New York Times crime reporter, investigative journalist called David Enrich, who had a book out this month called Murder, the Truth, and it's about the group of people who overturned Roe, who have spent the last 20 years trying to overturn Sullivan, which is the Supreme Court case.
1:16:13 - Leo Laporte
Thank, God that got protected, holy that got protected holy, yeah, well, now it's in danger steve wins lawsuit was turned back though that's right.
1:16:22 - Cory Doctorow
But they have a whole bunch of these key to cute uh it's not the end of it, wants to overturn marriage equality, and so on. It's the literally the same people, it's the same judges. They're targeting clarence. Thomas hates sullivan because sullivan protects the right of uh, the free press to say things if they're targeting clarence.
1:16:35 - Leo Laporte
Thomas hates sullivan because sullivan protects the right of uh, the free press to say things, if they're not malicious about people like steve winn and it protects.
1:16:47 - Paris Martineau
They're maliciously wrong, like right um libel someone with malicious intent to harm and, uh, in in this case, like so long as you are not one wrong and libeling someone and you did so you could be wrong, as long as it's not malicious, wanton, disregard for the truth or a malicious intent, then you're okay right, and so you know the.
1:17:09 - Cory Doctorow
The move to overturn sullivan, combined with the end of a safe harbor for user speech, means that a much larger amount of user speech will be swept up in potential liability, and then the platform will have vicarious liability for allowing the user speech. So you could imagine a future in which the New York Times publishes something and, rather than suing the New York Times over it, they sue a platform for allowing users to talk about it and quote it. Yeah Right, because Sullivan goes away. So this is the ability of rich people to decide who can criticize them and how has been a very big project for a long time and it's coming closer than ever. And the fact that Mark Zuckerberg is an advocate for Section 230 reform should tell you everything you know about whether Section 230 is good for big tech. Regulate us, please, he says. Create a compliance moat so that our competitors can't do unto us what we did unto my space Also known as pulling up the ladder behind you.
1:18:10 - Leo Laporte
Corey, great to have you filling in for Jeff Jarvis, who's off this week. Paris, martineau, great to have you back, corey. I know I'm going to be respectful of your time. We're going to get out of here in a half an hour. Let me do one more ad here before we go. Our show today, brought to you by Melissa.
I got off the phone the other day with Melissa. It's really interesting. They've been around 40 years, since 1985 now, doing address verification. But these days, days in business, it's more than address verification, it's really data science. It's data quality and they have been acquiring companies with ai, with data scientists. I talked to one of their new data scientists and he I was so impressed with where melissa has gone and what they are doing they have. They could have rested on their laurels, but no, they have stayed up to date, whether it's manufacturing or supply chain management. The healthcare industry melissa's using ai to boost efficiency, personalized customer experiences and spark innovation.
Poor quality data is bad right, it could. It can result in expensive mistakes, embarrassing mistakes and, in areas like health care, can kill people or give you a wrong diagnosis one of the things melissa is doing in their health care arm is validating patients prescriptions with the actual medicines put in the bottle, with the patient's record, to make sure they're getting the right medications. It's incredible. Recent study found that 4% of companies consider their data ready for AI models. Even the most advanced AI models, however, cannot correct underlying data quality issues, so it really goes hand in hand.
Melissa is the data quality expert. Imagine having a data expert that never sleeps. Melissa's intelligence system will verify identity to prevent fraud and gaming operations, as I said, ensure valid patient and medicine identification in health care systems. Securely update and verify constituent data across government databases. The know your business requirements. Enable verification and monitoring for financial institutions, kyc. Melissa guides you through complex data management with ease. It makes advanced data quality accessible to everyone, from small businesses to enterprises. With real-time data validation, comprehensive enrichment, cross-reference verification with gold-time data validation, comprehensive enrichment, cross-reference verification with gold standard reference data, intelligent anomaly detection. It's no wonder Melissa is the trusted data quality expert worldwide.
Your business needs Melissa and, good news, you don't have to worry about your data. Melissa securely encrypts all file transfers. They have an information security ecosystem built on the ISO 27001 framework. They adhere to GDPR policies, soc 2 compliant all the. Obviously they want to protect your data as much as you do. Contact Melissa's team to learn how they can elevate your business and improve your data quality. It's like having a data scientist working for you. Get started today with 1 1 000 records clean for free. Melissacom slash twit. We thank you so much for supporting intelligent machines. We thank you for supporting us by using that address melissacom twit melissa. Thank you, melissa, you're gonna get a uh, a home robot corey. Help you around the house.
1:21:37 - Cory Doctorow
I mean we had an incredibly disappointing string of Roombas.
1:21:41 - Paris Martineau
Didn't Roomba just file for bankruptcy?
1:21:44 - Cory Doctorow
I think they did RIP.
1:21:45 - Leo Laporte
Roomba, my robot's been struggling. You know what happened, although I think they might have been premature, because the reason their business is faltered is chinese clones.
1:21:54 - Paris Martineau
well, I think it's also because they were going to be acquired by amazon, but the deal fell through because I assume, uh, I believe the lena khan's ftc denied regulatory.
1:22:06 - Cory Doctorow
Yeah, right, it's a move to scrutinize it?
1:22:08 - Leo Laporte
yeah, but this is where tariffs could have helped them right at 200 tariff on chinese robots, chinese I don't think they make rumbas in america I don't think they make oh shoot. Oh, who'd have guessed they need to quickly spin up that rumba factory yeah every rumba you see carrying a knife.
1:22:28 - Paris Martineau
That's america made that's right.
1:22:31 - Cory Doctorow
That's right, the rumba that they put a bomb on to kill the terrorist. You know, that's right, the roomba that they put a bomb on to kill the terrorist.
1:22:36 - Paris Martineau
You know that's an american roomba.
1:22:37 - Leo Laporte
Yeah is that a made-up story, or did that really happen?
1:22:41 - Cory Doctorow
well, they put a. They put a, a bomb on a bomb robot and used it to kill the guy who was shooting at cops in dallas like oh, I remember that, yeah, yeah, yeah paris. I have to ask in that in the um journalist being added to the wrong chat uh scandal of last week. Given all that, do you ever get auto completed for paris marks and get his email because you could type paris space mar like you get pretty close but I should.
1:23:11 - Paris Martineau
I've thought about this as well I can't believe we haven't gotten confused for one another in this way more.
1:23:18 - Leo Laporte
But getting Paris Marx's messages is not as good as getting Pete Hegseth's messages.
1:23:23 - Paris Martineau
No, it's definitely not as good.
1:23:25 - Cory Doctorow
Paris is more coherent and less drunk.
1:23:28 - Paris Martineau
That's true. I mean I could. I could kind of do a little drunken role play. If anybody wants to add me to top secret government chats, I'll do that for you it's so funny.
1:23:39 - Leo Laporte
The government is now blaming apple. We talked about this yesterday. Mac break weekly yes, oh, it's all apple's fault because the iphone makes a jet. I mean, it might be. By the way, it's not completely far-fetched. The iphone makes suggestions about adding contact information and apparently added jeffrey goldberg's contact information. And apparently added Jeffrey Goldberg's contact information to a different JG, and then that was then that JG.
1:24:02 - Cory Doctorow
Only the US government had spent millions of dollars building apps for classified conversation that are specifically designed to ensure that only people with clearance can be added to the chat and proper records are kept, according to uh sunshine laws. If only those were issued to everyone who worked in a national security role and then where, if only there was a training session which they were told that they were legally obliged to use, just that and anything else would be a potential criminal act. If only those things were the only I mean clearly.
1:24:34 - Leo Laporte
It seems pretty obvious to me they like signal, because of its uh automatic, uh message destruction, that they didn't want records of these conversations and also they probably just didn't want to go to the secure room they want to use the skiff in their house by the way, listen you gotta.
1:24:50 - Paris Martineau
How else are you gonna be sending bombs, falling prayer emoji, fist emoji while you're picking up starbucks, if you have to go to your house to say that?
1:24:59 - Leo Laporte
carrie what do you? Use on those computers do you use something for secure messaging? Do you feel like you're you?
1:25:05 - Cory Doctorow
need to signal use signal I use signal, but I uh, I am not in a circumstance where you're not sharing war plans someone else gets to decide whether or not it's a crime for me to add someone to my chat, and so I don't have to worry about it.
But also, I am um, I am someone who really hates real-time communication unless there's no way around it. Uh, so I like email, um sending me a message and interrupting me while I'm doing something else. I will never thank you for it, unless it's's interesting to do it. You know, if we're out, like on our way to dinner, and we're in two cars and you're texting me to say like, oh, I found parking, that's one thing right. But if it's like the middle of the day and you want to invite me to dinner on Friday and you text me, it's like well, now I got to like, if I can't answer right now, which I probably can't I have to remember to look at that I am later and so that's how I end up not going to dinner with you on Friday and that's why you use the Paris Martin system of don't one get red receipts and then do not click on the message until you're ready to respond.
1:26:10 - Paris Martineau
Then you'll know that they know you've seen it and you trick yourself. But then this also has the problem of sometimes you'll leave messages unread for months and your friends will be like do you hate me?
1:26:20 - Leo Laporte
I need to point out, Corey, that Paris is closer to your daughter's age than yours.
1:26:25 - Paris Martineau
Hey, don't blow up my spot here.
1:26:28 - Leo Laporte
And so and this is to me this is very interesting there is a cultural divide about what form of communication people prefer, and there's no consensus on this. Some people would rather be called mostly people my age.
1:26:43 - Paris Martineau
I enjoy a call.
1:26:44 - Leo Laporte
My daughter, who's 32, wants me to call her. She says, no, text me, Never text me. That's the thing is.
1:26:50 - Paris Martineau
I differ from a lot of my peers in the sense that, listen, I'll enjoy a text with certain. Like I have a group chat, I treat those texts as a low effort. I'll check them when I need them. Sometimes I might engage in real time, but everything else.
Really, I don't really like real time conversation in the middle of my day I need to be in a specific headspace, otherwise I'm just too easily distracted. I think it's because I have adhd. See I, if I'm engaged in a conversation with you, then I'll forget what I was doing before. So I can't.
1:27:20 - Leo Laporte
Email does not work for me because of the volume of crap in my email and while I have many rules and many, I've done everything I can to filter the sewage it's, it's easy for me to miss emails, whereas a text message is as much crap as I'm starting to get is not yet here's my email survival strategy.
1:27:41 - Cory Doctorow
Um, so anything that has the word unsubscribing it goes into a folder called mailing list. I have that. I only check that periodically. Anytime I reply to you, your email address goes into an address book. Yep, and any email from someone in an address book goes into a folder called people.
1:27:56 - Leo Laporte
I know great minds, minds called important. But same thing.
1:27:59 - Cory Doctorow
If you're my contact list, yep so how is it that if, when you look at your the folder of email just from people in your address book that you don't see all the things?
1:28:10 - Leo Laporte
uh, yeah, no, that's true, it's just there's a lot of it. There's a lot of people in my address book. There's a lot of people want to email me. I'm also a public figure to some degree and that means I get a lot of email and I respond sometimes, which means they're now in my contact list. I don't know, I just text messaging seems to me to be kind of a way of prioritizing stuff.
1:28:33 - Cory Doctorow
So the last thing I want to do there's actually a great bit in Bruce Sterling's book, the Hacker Crackdown, which is a history of the first secret service raids on hackers and also the founding of EFF. It's a great book.
1:28:43 - Leo Laporte
I got to read that.
1:28:44 - Cory Doctorow
Yeah, it's an old book, it's from 1992, but it's an amazing book and there's a bit in it about the history of the phone system and they talk about how there was, like this, doctor who wrote this letter to the editor saying like this is insane Any person off the street can barge into your living room without so much as any advance notice by making the phone ring. This is nuts. Who wants this? You know I mean fair.
Right and I feel that. So I feel like email. It's like sending me a letter, right. I get the message and it shows up in my inbox and I can reply to it or delete it or you know whatever. And I have some templates as well that I use for replying to certain people if they have a kind of standard query. But it only shows up when I look at it. It never shows up when I'm you, I'm in the middle of doing something else, so I will periodically blip over to my email and hit F5, but I do not get notifications. I get no notifications, if I can help it.
1:29:49 - Leo Laporte
Yeah, I like turning off all notifications. Yeah, I think that's good, but I can't turn off messaging. You don't turn off messaging notifications, okay?
1:29:57 - Paris Martineau
oh yeah, I have all my notifications on, but I have do not disturb mode on 20. I've had do not disturb mode on for like seven years so my phone does not buzz or make noises it well, I have it set to where if somebody calls me and you have some people can burst through phone calls definitely burst through or at least make my phone vibrate.
But notification wise, they're there and I can just look at them as I need to, but they're not like my phone is. If I had every notification buzzing me I would go insane.
1:30:30 - Cory Doctorow
That's like hundreds of buzzes a day it's like it, like hooking our random number generator up to a taser and then like strapping it to your abdomen, and, and just like every, you know several times you're watching the intelligent machines.
1:30:50 - Leo Laporte
Cory doctor o paris martineau, so glad to have, uh, both of you back. Jeff will be back next week. I'll save some of these stories. I think Jeff actually was putting stories into the rundown.
1:31:00 - Paris Martineau
Oh, he was, he thought, he was going to be here. Clearly Do we want to take bets. Do you think he sank or floated in Salem?
1:31:07 - Leo Laporte
I think he weighs the same as a feather.
1:31:09 - Cory Doctorow
Weighs the same as a duck.
1:31:18 - Leo Laporte
A man from the new york times I'm sure jeff put this in a man employed, an ai avatar in his legal appeal. A video persona. It looks pretty, he's pretty handsome, but it was a fictitious person. The first judicial department, supreme court of the state of new york, was not amused, shocking. Who could have anticipated such a thing? The court had allowed mr dewalt, who's not a lawyer, was representing himself, to accompany his argument with a pre-recorded video presentation. As the video, oh, it turns out, dewalt is 74 years old. Okay, look at the guy in the video presentation. Right, don't think he's 74 years old. Okay, look at the guy in the video presentation. Don't think he's 74 years old. Don't think he's 74. I'm sure he had his normal voice too right.
As the video began to play, it showed a man seemingly younger than Mr DeWald, wearing a blue collared shirt and a beige sweater, standing in front of what appeared to be a blue virtual backwear. Now credit to the judges. A few seconds into the video, one of the judges, confused by the image on the screen, asked mr dewalt, is that your lawyer? I generated that? Mr dewalt responded that is not a real person. Uh, it would be nice to know that. When you made your application, she said I don't appreciate being misled before yelling for somebody to turn off the video. So okay, just a hint, maybe don't do that, right.
1:32:48 - Cory Doctorow
I don't want your chatbot nonsense in my court. I would like to hear salient and cogent things in my court. And you know, like, like obviously pro se is um is a bad idea, like anyway generally a bad idea not always, but generally. And and I appreciate that there are people who you know trying to handle their own um, you know appeals while they're in prison or whatever, who find themselves pro se because they that's a little different, yeah it is what it is.
But yeah, like, the thing about doing, about being pro se, is that you can come up with ideas that anyone who knows even a little about how courts work would say just don't do that, that won't work the way you think it will. You know, my, my, my boss and pal at eff, cindy cohen, who runs uh, the organization, uh, has spent her whole life talking with hackers who've got all kinds of ideas to do things in relation to the law that she just says like that won't pass the giggle test. Like you know, as soon as you try that in front of a judge, they will say, no, you are trying to subvert the spirit of the law. No, right, and and like the, there is a lot of uh, there's a lot of this. Like, oh, I found one weird trick, and you know, if you've ever encountered the sovereign movement, this is their whole thing. It's like I have figured out.
1:34:03 - Paris Martineau
I'm obsessed with the sovereign, oh my.
1:34:05 - Cory Doctorow
God, it's my favorite.
1:34:07 - Leo Laporte
I love those videos. There are lots of videos of people getting pulled over saying and they're like I'm not driving, I'm traveling.
1:34:17 - Cory Doctorow
And you're not. You're not a sheriff, so you can't arrest me.
1:34:19 - Paris Martineau
And even if you are with the sheriff's department, you're not the right kind of show that flag has yellow trim, which means we're in a naval court right now, and none of this applies. I'm a boat or I'm not whatever works best.
1:34:34 - Leo Laporte
Yeah, it is amazing and it continues to happen, but it just shows people are sometimes a little gullible. Uh, shopify. One of our fine sponsors love them a lot. Their ceo sent out a memo saying if you need to add headcount this is from toby lutke if you need to add headcount, before you can hire somebody in your department you have to prove to me before asking for more headcount, he wrote. And resources teams must demonstrate why they cannot get what they want done using ai. What would this area look like if autonomous ai agents were already part of the team? He says.
1:35:16 - Paris Martineau
He also said in this letter that now Shopify employees it's going to be baked into all performance reviews at Shopify, not only self-assessment, but peer assessments and your manager assessing employees. The question how often does this person use AI? And I assume the answer is, if you're not saying in that I use AI five out of five all the question, how often does this person use AI? And I assume the answer is, if you're not saying in that I use AI five out of five all the time, that your job could be at risk, which is incredibly wild to me as a business strategy.
1:35:47 - Cory Doctorow
I mean any metric becomes a target. It is not hard to use AI 10 times a day If all you want to do is burn Shopify's you know store of of uh, open AI tokens or whatever you can. Just, you can just ask it to, like, give you directions for removing a grilled cheese sandwich from a VCR in the style of the King James Bible, 10 times a day and you are now, uh that you know the six Sigma AI user in the Shopify.
1:36:19 - Leo Laporte
Shopify is known for uh productivity and net pronouncements in early 2023, according to the wall street journal. Uh, and actually I kind of agree with this one shopify's coo, they're fully remote 8100 fully remote employees. So I I praise them for that. Uh, shopify's coo directed employees to quit holding quote an absurd amount of meetings. The company then deleted 12 000 events from its workers calendars, freeing up 95 000 hours for more focus tats. You know, I'm not against that. I'm not against that.
1:36:49 - Cory Doctorow
Delete the meetings I'm not against fewer meetings too. I think that's great, yeah, yeah, I'm against.
1:36:55 - Paris Martineau
That's all middle management I was about to say, like how it sounds like to me from what you just described, that management like made a couple broad decisions and suddenly was like 12 or 20 000 meetings gone, and I don't know that I trust a couple people to make the right choices for thousands of other people's meetings.
1:37:18 - Cory Doctorow
I mean. Another important law here is Chesterton's fence, which is that you shouldn't remove a measure until you can explain why it's there. And so while there's doubtless meetings that shouldn't exist, there are meetings that need to exist, and if all you do is randomly remove meetings, you are not going to solve that problem.
1:37:37 - Leo Laporte
There's a similarity to the recent layoffs at many of the federal institutions. You know I like Chesterton's law. Do you have a compendium? I think, corey, you need to write a compendium of laws, of useful laws, useful laws.
1:37:53 - Benito Gonzalez
No just two stone tablets, just two stone tablets. Right laws of useful laws. Useful laws, no just two signs just two stone tablets.
1:37:57 - Cory Doctorow
Right there's stein's law, which is that anything that can't go on forever eventually stops okay, that's a good one. So I was gonna say you know, like if I have a law about ai, it's that irrespective of whether ai can do your job, an ai salesman can probably convince your boss to fire you and replace you with an ai salesman that can't do your job.
1:38:17 - Leo Laporte
Yeah, uh, that's not like you've really pinpointed the problem that the incentives they're perverse out there for for this, uh, whether ai is valuable or not, independently of that, there are definitely uh disincentives and your boss, leo.
1:38:31 - Paris Martineau
Are you using your b right now? I'm wearing it right now, kids, of course he's, uh, he's got a little guy that's recording his whole life and everyone around him 24 7 uh, in flagrant violation of california wiretapping laws no right, because it then gives him very fun and interesting summaries of his day. I guess it.
1:38:54 - Leo Laporte
Basically, as it turns out, it does two things that are useful it writes my daily journal and, and it's pretty accurate.
1:39:01 - Paris Martineau
You know the idea of writing your own journal and self-reflecting on the day no, no, I don't want to self-reflect, I just want to read your journal for you too.
1:39:09 - Leo Laporte
Yes, exactly, I don't need to read it, or?
1:39:11 - Cory Doctorow
write it the douglas adams thing with the automated monk that watches all the shows that you've taped with your VCR. You don't have time to watch. Oh, I need that.
1:39:19 - Leo Laporte
Yeah, the other thing it does it automatically makes a to-do list that I can approve or reject. But that's very useful because I've always wanted something that would remind me that I agreed to do something. You know, I'm often agreeing to things and forgetting about it, so that's useful.
1:39:36 - Cory Doctorow
I mean you need a system to do that, that if that system works for you. That's fine. Yeah, it does. Yeah, I mean life loggers have been around for a long time. I remembered uh gordon 2005-6, when I was at usc justin hall running around with justin microsoft life life loggers hanging around his neck.
He eventually uh, it's funny he after years of being like the world's most confessional blogger, with no secrets and wearing a life logger, he eventually uh, it's funny he after years of being like the world's most confessional blogger, with no secrets and wearing a life logger, he eventually became someone who's quite interested in his own privacy. Interesting good for him.
1:40:04 - Leo Laporte
I feel that the ship has sailed for me, cory, I don't, I think. I think everything is known and uh, I might as well just uh live with it, don't you think?
1:40:15 - Cory Doctorow
well, look, privacy is not. No one knows stuff about you. It's that you choose what people know about you.
1:40:20 - Leo Laporte
And in my case, everything Sure.
1:40:22 - Cory Doctorow
If you make that choice, that's fine. Yeah.
1:40:24 - Paris Martineau
I'd really like to remember, recommend the Wikipedia page for LifeLogger, specifically the image at the top of it, Every single photo. It shows five photos of men wearing life loggers and all of them are more ridiculous than you could possibly imagine oh, those are good.
1:40:42 - Cory Doctorow
Oh, one steve man there's one steve man it was the.
1:40:46 - Leo Laporte
Yes, steve man was the legendary mit hacker there he is on the left, steve man.
1:40:51 - Cory Doctorow
I think there might all be steve oh, they are all, you're right it's all steve man situation.
1:40:55 - Leo Laporte
They're all steve man situation they're all steve man, he walked around. He would walk around the mit campus with that stuff on. I really like his sunglasses.
1:41:04 - Paris Martineau
Mime, look kind of the one on the right yeah, uh, but it's.
1:41:10 - Leo Laporte
We've come a long way. See how it's gotten better, and that, uh, that one on the far left, by the way. Let me go back.
1:41:17 - Paris Martineau
That is the one that led me to suggest we bring this up, because it's quite good. Oh no, I was talking about this one, which is a.
1:41:23 - Leo Laporte
Google device, Remember. They put it out and then they realized nobody it would take pictures, Nobody wanted pictures taken from that angle.
1:41:32 - Cory Doctorow
Yeah, the bar on my corner when I lived lived in san francisco. The elbow room. Someone, famously, I think, got punched for wearing a. Um, oh, yeah, the elbow room, the google glass.
1:41:41 - Leo Laporte
Yeah, this all predates google glass. I don't know what he's wearing in the in the late 1990s picture in the striped shirt uh, the one you're talking about, paris I know what is the one in the 1980s I think that's first man I think that maybe it's all steve man, I do, I agree what is he wearing?
1:41:59 - Paris Martineau
I like the combo helmet, antenna eye thing and what's going on on top of his head. So this.
1:42:06 - Leo Laporte
Remember. This is 1980, so he has to have a. This is his viewer. It's a little tv screen. He couldn't make him that small. In the back of the day he's got a camera. The antennas are a little tv screen. He couldn't make them that small. In the back of the day he's got a camera. The antennas are a little aggressive. He's got a duck to go in a room. No on his head.
1:42:22 - Benito Gonzalez
That's the light. I think that's a light on his head oh, is that a light?
1:42:24 - Leo Laporte
well, where's the camera? Maybe that's the camera and the screen camera.
1:42:27 - Cory Doctorow
Yeah, so it reminds me of this. I just put it in the chat hugo gernsback, the uh the, the founder of astounding stories, and the person for whom the hugo award is named, invented this thing called the isolator to help him focus zoom did he uh, did he wear it?
1:42:45 - Leo Laporte
he did he wore it. Uh, the isolator helmet and uh does not exist.
1:42:52 - Paris Martineau
Wait a minute. What happened?
1:42:54 - Leo Laporte
post the right one in there uh might have gotten the wrong one.
1:42:57 - Cory Doctorow
Uh, the isolator helmet zero got in there somehow sorry, oh yeah, I hit a zero on my way yeah oh, this is great did he write with this thing on?
1:43:08 - Paris Martineau
yeah, yeah, that's what I need when I'm forced to write copy in the office that reminds me a lot of that kickstarter I think I gave it away oh yeah, the the face huggers kicks, yeah, yeah so you could sleep in airports and things.
1:43:26 - Leo Laporte
That's right. Uh, I had that, I would. I wore it even sometimes what is this?
uh, it was a pellet filled. It looked exactly like this a pellet filled, um uh hat that you would wear. It had two holes on the left and right for your arms and it had a small hole around your nose so you could breathe. But the rest of you were covered, was covered in felt, gray felt padded with pellets. And, uh, gosh, I wish I didn't give it away. You know what I reason I gave it away, I think, is because it started leaking and I'm sleeping little pellets behind me mystery pellets.
Yeah, I decided maybe they let you bring this into an airport yeah, well, the idea is you could sleep, uh, you know, because your arms came out so you could sleep like this. Nobody would bother you, especially because you look like the elephant man. Um did you buy?
1:44:19 - Cory Doctorow
them on amazon. They they are still for sale on amazon. The dream of the 90s is live on in portland and that's made in the usa.
1:44:27 - Leo Laporte
I gotta point out the ostrich pillow ostrich pillow wow, that's on amazon, discards on top of it. Wow, that's on amazon, discards on top of it.
1:44:43 - Paris Martineau
Not on mine, it's not oh, oh, my god, on top of my screen. These are some incredible photos yeah, the go neck pillow.
1:44:50 - Leo Laporte
Oh, they got. You know what they got? A little they realized how dorky it looked now.
1:44:54 - Cory Doctorow
But they got the ostrich pillow too. The with that, the thing that you stick your head in yeah, they still have that it looks like a I might have to buy another one oh yeah, they're down underwear.
1:45:05 - Paris Martineau
Yes, it's kind of like wearing it's kind of like wearing your tighty whities on your head.
1:45:10 - Leo Laporte
Imagine a pair of tighty whities but filled with pellets imagine sitting down for a transatlantic flight. I've seen people do weird.
1:45:21 - Paris Martineau
Pull the exit door.
1:45:23 - Leo Laporte
What about vision pro? That's pretty close right.
1:45:25 - Cory Doctorow
There's the product shot there, I just put it in the chat. Okay, all right, I have to do a little rigmarole to get it from from the chat into a different machine.
1:45:34 - Leo Laporte
But let me, let me.
1:45:36 - Paris Martineau
Oh, the power napping photo is great. Oh, it's pretty good. That looks a little suggestive.
1:45:42 - Leo Laporte
That's the wrong one, all right, wow, here we go. Corey, you're fun, it's great to have you Really. Oh yeah, that's it. That's the real deal. I had that exact one. He's not using it fully to its when he wore like I said, uh, I am not an animal. The one uh with the power napping is really more my speed, where you put your hand in the holes yeah, yeah, stick your fingers in your ears.
1:46:14 - Cory Doctorow
You can.
1:46:14 - Paris Martineau
You can pick your nose while you're sleeping just imagine a flight attendant's shaky hand as they decide and they're like should I tap to wake this man up for the meal service?
1:46:24 - Cory Doctorow
no, I should leave it no, no, the game is once you take off, you sneak into a business class seat and then you put that on your head and they don't know and that you've got a business class seat and they don't know, it's not they're like we can't, we can't, we don't know man to move. It could be hugo gersman, we don't know and if they ask you to move, you tell them you've created an easement I am a sovereign citizen and your flag is wrong.
1:46:51 - Leo Laporte
Uh, you wrote an article on pluralisticnet about AI therapy. This article from NPR says that it's all the rage and I guess if you were nervous about talking to a real person I mean, this goes back to Eliza- you might think that an AI therapist is of some use, so this was not about the merits of AI therapy.
1:47:16 - Cory Doctorow
This is about the risks of letting a chap on, owned by companies that habitually lie about how they handle data. Uh, have the data that you can get from nowhere else.
1:47:25 - Leo Laporte
Uh, good point. Much like my B, actually the thing I'm wearing.
1:47:30 - Cory Doctorow
There's your B loop through the cloud. By the way, I'm very proud of that collage. Uh, you just scroll past it loop through the cloud by the way, I'm very proud of that collage. Uh, you just scroll past it. Oh, did I miss it let's see there we go.
That's a really good one like I love these and the title is anyone who uses or trusts an ai therapist needs their head examined yeah, just these companies have been so consistently unbruthful about their data handling and this data is extremely compromising and I just think you have no basis for believing them. Like you know, fool me once we don't get fooled again, as George Bush said.
1:48:13 - Leo Laporte
Is that what he said? I couldn't quite parse it. We're going to take a little break. Normally, at point, we do pics of the week. Cory, if you want to dial out, you can. You've been very patient with us.
1:48:22 - Cory Doctorow
I thank you I'm going to ditch before my back starts spasming, but it was great to see you folks really, really appreciate it.
1:48:29 - Leo Laporte
Cory, you're the best here cory dr o the bezel dot org for his books. Definitely, you got to read the whole Marty Hench saga. They're all good.
1:48:38 - Cory Doctorow
And martinhenchcom as well. You can go to martinhenchcom. They're all just redirects to my personal site.
1:48:43 - Leo Laporte
Awesome and pluralisticnet for his blog and of course he's on Mastodon at Pluralistic. Thank you, Corey. Thank you, Take care. Great to see you. We love Corey. Have you been on with him before? I don't think you have Paris.
1:48:56 - Paris Martineau
Oh, I'm obsessed with his glasses. I should have asked him more about them.
1:49:00 - Leo Laporte
I should have asked him about them.
1:49:01 - Paris Martineau
They were just lovely Big.
1:49:03 - Leo Laporte
Corey, you still there? No, he's frozen solid, so I think he hung up you know.
1:49:10 - Paris Martineau
I think that's a good place for him to be, it's not horizontal than frozen solid.
1:49:13 - Leo Laporte
Yeah, poor guy. He's been in a lot of pain. He's had both hips replaced. Now, oh my gosh, yeah, his back hurts, but he keeps writing, he keeps creating, he keeps fighting the good fight, and it's one of the reasons I deeply respect the man, and his intelligence is second to none.
1:49:33 - Paris Martineau
Absolutely. I mean, he was a fantastic guest.
1:49:35 - Leo Laporte
But I still have sand in my shoes.
1:49:37 - Paris Martineau
I'm just going to tell you that, right I was about to say you were on your best behavior there, leo. You were like oh I am not gonna you were like I don't. I don't know that I think this is going to be super transformative or anything, but I'm just thinking we should hear him out.
1:49:49 - Leo Laporte
I was like, excuse me, leo I am not gonna argue with cory doctorow me. I do not want to be eviscerated. Can you imagine he would just?
1:49:59 - Paris Martineau
I know you even got just a couple little things in there that you believe in. He just correctly put you in your place.
1:50:06 - Leo Laporte
I don't know if he completely demolished my arguments. I don't think it's a. You're saying the.
1:50:11 - Paris Martineau
YouTube headline of this is not going to be Doctor who smashes Leo's weak liberal argument to pieces.
1:50:19 - Leo Laporte
Maybe that's the title of the show. Okay, maybe.
1:50:23 - Paris Martineau
Maybe that'll get the clicks.
1:50:24 - Leo Laporte
Jeff will be back next week. Paris Martineau is back. We will get your picks of the week in a moment. You're watching Intelligent Machines, episode 814. A reminder, incidentally, that there are a couple of things coming up. Sunday, uh, is the 20th anniversary of this network. We did our first twitch show on april 13th april 14th, I think, it was, uh, 2005. On april 13th. This sunday, we're going to celebrate. Uh, patrick norton, who was on that first show, will be joining us, along with Alan Malventano and Samuel Samet. But the really important part of the celebration is you, because, what you know, we kind of brought the old hosts back for episode 1000, but I really wanted to acknowledge the fact that twit would not exist without you guys, without our fabulous audience. So I encouraged everybody and it's not too late to send in a video. Talk about how you first found twit, your earliest memories of twit, anything you want to share. We've had people in all walks of life, from all over the world respond and we're going to have some fun videos. I have a poem I shall read uh, there's been a lot of great response, if you want written by you, no by ai or by not by ai nor me, by an actual human listener and that's art, baby
that's art, baby. Um, what's the best way, bonito, at this point? Should I have them? We asked, uh, that you post on social media with a tag, uh, hashtag, twit, but I think it's at this point for us to harvest all that by Sunday is probably a little late. Why don't you email it to leo at leovillecom or info at twittv? If you've got something recorded on your phone, it's fine. Keep it short. We're going to have a lot of fun with our fans, with our friends, our cohort. It's really been great working with you the other thing I wanted to announce.
Lisa has told me that we are bringing back the one year subscription to Club twit, and this is really good news. Now, club twit is the way we. You know. You see a lot of shows, uh, a lot of projects do this. Um, we are ad supported, but the ads don't cover our entire costs, not nearly. Uh, we need some help and we started doing this a couple of years ago and it's been great. Normally we charge seven bucks a month. You get ad free versions of the shows because you're paying for it. I, I hate it when you pay for something and still get ads. So no ads, not even this plug.
You'll get access to the club twit discord, which is a wonderful place to hang out and be, and some really smart people, interesting people, many of our hosts, uh, hang out in there and, of course, a lot of the great creativity happens, not just during the shows a lot of us during the shows but the the club has, you know, sections for every possible geek interest, so it's really fun to hang out in there. You can also watch special events events, in fact, joe Esposito is always creating fun little posters for us. Here I am as Indiana Jones, michael Jackson and I don't know Simone Biles I don't know who is in the middle there but join the club. Twittv, slash club, twit. You'll also get special events. We've decided to do all of the streaming conferences and keynotes that we've done all this time. You know Apple's starting to get very sticky about it. We've had takedowns on YouTube and now on Twitch, so we decided to start moving those into the club. Do them on Discord only, so you won't see those in public. You will see those on Discord. We figured that's protecting Apple's intellectual property but still letting us do the journalistic job we need to do.
First one will be june 9th, I think. Micah and I will cover apple's keynote at wwdc, but we will also, because of this new format, be able to stick around and do the state of the union address as well. So all day, uh, june 9th for micah and leo talking about what Apple has announced. There are keynotes coming up Microsoft Build, other events. We will see those in the Discord. We also do our Coffee. Geek is coming up. Mark Prince. He's bringing Liz Happy Beans with him and we'll do another coffee episode. We've got Stacy's Book Club, micah's Crafting Corner, photography Time with Chris Margaret, all of those are club only events.
We try to make it worth your while and it's only $7 a month. And, as I said, when you go there now twittv slash club twit you'll also see the opportunity to sign up for a full year Still $7 a month, but it's $84 in one throw and we appreciate your support. It makes a big difference to us by month or by year. Whichever you prefer and if you're already, I should mention this for those of you watching live, if you're already, I should mention this for those of you watching live if you're already a member, you can just go to the subscription page and turn on the yearly for the next renewal period and that'll renew it a year, right, okay. So if you're already in the club and you want to go to year a lot of people have said well, I want the yearly subscription you can do that by going to the subscription page. All right, actually, that part we should, we should not edit out, uh because I want everybody to know about that.
The club is going to hear that yeah, the club, the club needs to hear it. Anybody in the club, I guess here from the club you'll hear it, right, yeah, so leave that, cut that part out and stick it in. Uh, all right time for our picks of the week.
And uh, usually I let jeff start, but I'm gonna, I'm gonna, I have one this week paris so I'm gonna leo pick I'm gonna throw in a couple and actually we even have, I believe, a pick from jeff, because he really did we do, because jeff be here, so I'll do his jeff fills in his. So you remember, the last time Jeff wasn't here, I told you about a new puzzle called Bracket City.
1:56:06 - Paris Martineau
Right, yes, and. I love Bracket City. I've been playing it regularly. It's so much fun.
1:56:11 - Leo Laporte
It's kind of like a crossword puzzle that you fill in. I figured this is going to get purchased. The New York Times is going to add it to their puzzle page.
1:56:20 - Paris Martineau
I thought you were going to be right.
1:56:26 - Leo Laporte
The Atlantic got there first, which actually I'm thrilled about. The atlantic has some very good crosswords and now, if you go to bracketcity, you will be taken to. Oh, you see, I'm the chief of police today. I didn't do as well as I as I sometimes do, but this is uh, so let's play um, let's see, I think I did yesterday's already.
1:56:42 - Paris Martineau
Uh, puppet master yesterday, oh, you became death destroyer of worlds, that's yeah, that's perfect.
1:56:49 - Leo Laporte
So let's, let's just show you what it looks like. We won't do the whole thing, but you see, the highlighted ones are actually. You just solve them so blank your nose, that's easy. Just pick it and then oh oh, it's blow it's blow. Oh see, so I got that one wrong.
That's gonna cost me points, uh famous hard mode of course, famous last, the masculine form of felicity, or a famous cat, felix the cat. You get the idea, so the yellow ones get filled in and then eventually it. It gives you this date in history, which is really fun. Um rocks when added to soda will not. Huh, um rocks when added to soda will not cause your pop, pop, pop rocks oh yeah, and then domain for retail cop it's the mall right low-rise kind of mall Strip. You know, I'm just going to do this with you guys from now on Striped canes. Candy.
Candy canes. You guys are good Night. You are finally free to demand candy from strangers and Halloween blood type.
1:58:01 - Paris Martineau
Huh, hey, is it like oh right oh okay, I don't know. There's a couple, there's not, that yeah, um well, let's look at what's around it.
1:58:12 - Leo Laporte
This is a black to blank fish, which is catfish, to lure someone in using a something internet persona fake, fake halloween fake blood, fake blood. Now we know it's cat fish. And there you go. We did the whole thing. Uh, felix the fifth abdicates, ending the reign of the final antipope, with a link to the wikipedia. Anyway, I love bracket city, now part of the atlantic, it is free still. Do we know what the purchase cost was? No, no, but you know it was a lot right we also don't know entirely that it was purchased.
1:58:45 - Paris Martineau
I noticed, I think, in the announcement Is it?
1:58:47 - Leo Laporte
that they donated it.
1:58:48 - Paris Martineau
No, I don't think they donated it. It could be like a licensing deal, it said. If you look at the Atlantic's announcement article for it. It didn't say we've acquired Bracket. City it says Bracket City is now available to play on here and the guy who created it will now be reporting to Atlantic Games editor. Blankety blank.
1:59:10 - Leo Laporte
So I don't know. Yeah, they're going to work together. I don't know, I just thought it was interesting.
1:59:13 - Paris Martineau
I was like did they offer this guy a job? Because I feel like, if no, he owns a bar.
1:59:19 - Leo Laporte
He does trivia contests.
1:59:21 - Paris Martineau
I know, I just thought it was interesting.
1:59:22 - Leo Laporte
I'm going to say he got a multi-million dollar payday, because you know where the money for the Atlantic comes from Steve Jobs' widow, lorene Powell Jobs, and the Emerson Collective.
1:59:32 - Paris Martineau
She's been turning that tap off or at least closing it a little bit. They've been trying to get it to generate profit.
1:59:40 - Leo Laporte
Well, this is how you do it. That's how the Times did it, right.
1:59:43 - Paris Martineau
It could also be a licensing deal like a limited licensing deal. I mean, I think it could be a licensing deal.
1:59:47 - Benito Gonzalez
I think there could be a lot of things that are not exclusively.
1:59:50 - Paris Martineau
it was bought Because typically that's announced as we've acquired this.
1:59:54 - Leo Laporte
Right. Well, in that case, I hope he gets 5% of everything I pay for Brack, which is nothing but you gotta figure they think this will be good for subscriptions, right?
2:00:08 - Paris Martineau
yeah, I mean it's been a huge boon for the times in the sense that people subscribe the times for games I.
2:00:14 - Leo Laporte
You know there's two things. Even if I didn't want to support what jeff jarvis calls the broken times and I did I did cancel my subscription to the washington post. But even if I didn't want to support the news portion, I love the recipes, I love the crosswords, I love the connections. I think I have to continue.
2:00:35 - Paris Martineau
Yeah, I mean, there's a lot of parts of it I love yeah, I love cooking. I love games yeah I love the news I read it.
2:00:42 - Leo Laporte
I have new york times subscribe part of my job, right, I gotta read the news. You gotta read the news jeff's pick from the pew research center artificial intelligence and daily life. Pew does these great surveys, uh of uh people and uh they talked to a lot of people. I guess I don't.
2:01:03 - Paris Martineau
I find the numbers I love this stat, which is that um the percentage of ai experts that think people in the us interact with artificial intelligence almost constantly is 79 of ai experts think that, meanwhile, only 27 of us adults actually do interact with ai almost constantly, or seven everybody uses our product all the time.
2:01:28 - Leo Laporte
Everybody's using it so ai experts think we're using their stuff more often?
2:01:34 - Benito Gonzalez
a third of us adults. They got to say that that many people are using right ai experts.
2:01:39 - Leo Laporte
When asked what percentage of people, who percent, who say they blank, ever use an artificial intelligence chatbot, I don't understand.
2:01:50 - Paris Martineau
So have or have not they were like, they were um. They asked ai experts like what percentage of people have used an AI chatbot? Which percentage haven't? And the experts say 98% have, and meanwhile actually 33% of US adults have.
2:02:09 - Leo Laporte
Oh, they really misestimated that that's really funny. Among chatbot users, us adults are far less likely than AI experts to say chatbots are extremely or very helpful. Ai experts said 61% would say that only 33% say and 21% say not at all or not too helpful.
2:02:31 - Benito Gonzalez
Fascinating. So this is the gap between the hype and the reality. Right, yeah.
2:02:35 - Paris Martineau
And the AI experts surveyed are those who were authors or presenters at an AI-related conference in 2023 or 2024 and live in the us um, which is kind of an interesting sample to take it from. Is all of these, uh people heavily involved?
2:02:50 - Leo Laporte
these are the true believers say yeah yeah it's everything this is this is your sand guy, leo all right, maybe you and cory have worn me down just a little bit.
2:03:02 - Paris Martineau
Hey, we're getting you know. I can't believe Jeff's not here for this.
2:03:07 - Leo Laporte
I admit Corey's absolutely right I mean, he's always been right when he says you know, these big companies are not there to change the world, they're there to make as much money as possible. And that's not a surprise is it.
2:03:20 - Paris Martineau
No, make as much money as possible. And that's not a surprise. No, opening eye is just changing from a non-profit to a for-profit structure because it just really cares about the right thing, to do paris your picks of the week um, I got two picks. One is a quick shout out. I finished twin peaks on sunday, not knowing that this week was the 35th anniversary of twin peaks, which I just feel like it's kind of fortuitous the show before you, before you Were Born, is now.
2:03:45 - Leo Laporte
you've caught up, congratulations.
2:03:47 - Paris Martineau
Now I've caught up. I'm reeling. Did it end well? That's hard to describe. I mean I found the ending very well. I was going to say satisfying, but it's not the right word. At first I was unsatisfied and confused, in the way that I often am whenever I finish a David Lynch film. Then we're now what three days out from it, I think the ending was a masterpiece, really subverted my expectations for where this show was going and how it was going to handle it in a very interesting, novel and challenging and ultimately entertaining way. So I don't know. I'd really recommend it. If anybody out there hasn't watched Twin Peaks, start from the beginning. It's a delight. The new season is crazy. I don't know how I mean the thing that was so mind-blowing to me about watching the 90s twin peaks. I was like how the heck was this on television in the 90s?
I had to blow everyone's minds and it of course did it did and I I thought well, there's no way you can do that again in you know 20 there's a wait, a minute. There's a new twin peaks yeah, so after 25 years in in 2017, they came back and they did 18 parts of it, and it was just as experimental and groundbreaking as the 90s one. David Lynch, still David Lynch, you know.
Yeah, still David Lynch, but I would recommend it. The other thing I have to recommend is just one of the reasons I went to Amsterdam was there was this artist I stumbled upon when I was there for the first time, like eight or nine years ago, whose work I love. He's just this artist named eddie vericamp doesn't sell any of his stuff online. So I just wanted to shout out if anybody who's listening ever goes to amsterdam and likes art, go visit these are woodblocks.
2:05:38 - Leo Laporte
Yeah, are they woodblock.
2:05:39 - Paris Martineau
He does woodblock prints, he does paintings, he does drawings. A lot of the biggest volume of work, of course he makes are woodblocks, because it's easy to make multiple of them.
2:05:52 - Leo Laporte
Did you pick up an Eddie Veracamp, did you?
2:05:55 - Paris Martineau
pick one up. I picked up two actual paintings of his Nice and he happened to be in the back. So as I was checking out out, he came up and explained to me one of the paintings was a painting of a cafe and he was like, oh, here's where the cafe is. You should go visit it.
2:06:10 - Leo Laporte
Here's like what led me to paint this, and so it was delightful. Forget bitcoin, buy art buy art guys.
2:06:13 - Paris Martineau
It's fun, you know, it's fun to meet an artist and buy and buy some paintings to have in your home so you actually you saw this on his instagram and you said I'm a going there well. So I had followed him on instagram because nine or so years ago I was in amsterdam, randomly stumbled upon his studio and was fell in love with his work and bought two pieces oh, so you've been there before the one that's actually above my mantle, you can't really see it right now is his, but I'd wanted more of it.
But he's just a man who lives in Amsterdam and makes art, doesn't sell any of it online, so I was like, if I want more, I should go, gotta go.
2:06:52 - Leo Laporte
So I realized I was going to Paris with the skeeball team.
2:06:55 - Paris Martineau
I was like I gotta make a trip.
2:06:57 - Leo Laporte
I think you've got excellent taste. These are great. I love them. I love woodblock.
2:07:00 - Paris Martineau
Yeah, he's got really interesting like and the thing is like his art that he does, that isn't woodblock. Still kind of it looks woodblock-esque.
2:07:10 - Leo Laporte
Oh, so some of these are watercolors.
2:07:12 - Paris Martineau
I see, yes, stencil print, the one that you just passed by up, there was the one that I have with my mantle, which is kind of cool. What, which one People? What which one people? Number five. You have that over your mantle, yeah that's gorgeous.
2:07:26 - Leo Laporte
It's a good one. That's do you now? Uh, there are multiple copies.
2:07:29 - Paris Martineau
Right, you don't have the one yeah, and I think there's like there's a number in the corner, I think there's like 30 oh okay, so they're still pretty restricted he does like I mean is he older guy? Older fella yeah he's, uh, definitely older. I don't know what his age is, I mean, probably like 60s or 70s, um, but I wanted to make sure that I was able to go and visit him to see his work, just because I really love it and I love supporting an artist like that, as always, you have excellent taste.
2:07:59 - Leo Laporte
It was I also. Just.
2:07:59 - Paris Martineau
Amsterdam was super fun.
2:08:01 - Leo Laporte
Oh yeah, it's a wonderful city.
2:08:03 - Paris Martineau
I love the museums there. The Stedelijk is great too.
2:08:07 - Leo Laporte
Did you go to the Rijksmuseum? Of course you did. I saw the Rembrandts.
2:08:10 - Paris Martineau
I've been there before, I went to go see the Van Gogh Museum again. Love the Van Gogh Museum Just because those are two of my favorite museums.
2:08:23 - Leo Laporte
It's really wild to see van gogh's brush strokes up close. You can almost feel like they're still wet. They're so strong and you can get right up and look at them. You know you don't see that in a two-dimensional picture.
2:08:31 - Paris Martineau
It's amazing it's really incredible, and also the thing I took away from the museum this time, which I don't know how. I'm sure I took this away last time I was there, but it left. My mind is just they go really in depth into his mental health struggles and how I mean. At one point he was perhaps in the worst mental health of someone that could be described like after he's cut off his ear, having a hard time interacting with anyone. He's uh, goes into a psychiatric ward. His treatment is twice a week. He's allowed to have a bath unbelievable and it's.
That's the entire treatment that's the entire treatment and it's so describes this you know could have been worse, to be honest I mean, could have been worse during this time? He painted most of, I guess, like many of his iconic images because literally the only thing he could do to keep himself remotely sane is paint. There was nothing else to do.
2:09:25 - Leo Laporte
When we were in the south of France we went to visit the sanatorium that he was there and it was quite a kind of moving experience, yeah.
2:09:34 - Paris Martineau
Yeah, I don't know. Go see some art people. That's my pick of the week.
2:09:37 - Leo Laporte
Yeah, you know, there's no way you're gonna have a machine create art like that.
2:09:44 - Paris Martineau
it might look like that, but it's not gonna have that, but it's like, and some I don't think a machine is ever going to create the next van gogh. You know, not that it has to be even a remotely similar style, but a art that affects people in the same way that has people talking about it here on a podcast. How many years after, uh, this person gone. Like a machine is not going to have that sort of lasting permanence.
2:10:06 - Benito Gonzalez
I mean, I don't know if that's a non-zero, but definitely not for the next hundred years.
2:10:10 - Paris Martineau
But I don't know if that's a non-zero chance. I wouldn't say that's a non-zero chance One doesn't know. Yeah, no, I mean sure there could be something that's phenomenal. I think in the next 10 to 20 years we will be reading novels, going to movies, listening to music, all of which were created by machines, and happily doing so. I'm just curious to see as to whether that ends up having the same sort of long lasting cultural impact.
2:10:34 - Leo Laporte
I think there will be connoisseurs who will say it's got to be human or I'm not going to do it.
2:10:38 - Paris Martineau
But I think part of the reason why something like Van Gogh or David Lynch or, you know, like any of these pieces of work we've talked about today, are still something we return to long after they were created and long after their creators had died is because there is something about the human authorial nature of, like an actual creation of art that leads us all to constantly wonder what was this person thinking? And reanalyze their life, reanalyze their work and come back again and again to the text and I just don't see. Maybe I'm being foolish and I'll be proven wrong in a couple of years, but I don't see that draw happening for AI art. Like how are we going to be teaching this is rudimentary, but how are you going to teach a college course about an AI painting Like? It just doesn't seem as reasonable or imperative to be analyzing the author's intent if there is no author.
2:11:41 - Leo Laporte
I don't think you're necessarily wrong. I wouldn't argue against you. But I also have always been suspicious of human beliefs that we are somehow special or magical.
2:11:56 - Paris Martineau
No, and I'm not even saying I think it's like that we're uniquely. I just think that if you look at the way part of like even with that example college courses or academic pursuits, like at the intro to intermediate level the part of what people end up circling upon again and again is like oh, what was this person's intent? How does that relate, or, uh, compare to how it's perceived? How does that relate, or compare to how it's perceived? How does that framework differ from the frameworks of someone else? The ways that we think about these sort of things and the sort of things that end up becoming very culturally salient often have to do with authorship in some way.
2:12:41 - Benito Gonzalez
That's how it's been up to now yeah, I mean things change.
2:12:45 - Paris Martineau
I guess a pretty big fundamental shift from the last like thousand years, you know yeah, I mean you could.
2:12:52 - Leo Laporte
You know I didn't say this to cory, but uh, you're talking about the lawyer writing the letter or the law professor writing the letter versus the AI writing the letter and they're more valuable because it's so much work to write a letter if the lawyer, law professor does it. You know it was a lot more work to make a Bible when a monk had to sit for months in a scriptorium to draw it. Yes, and I'm sure people without Jeff here I've got to bring up Gutenberg thought that those first printed Bibles were not somehow as magical, and they aren't as illuminated volumes.
But he's talking about information density, not labor but he was saying that the information density came from the fact that a human put the labor into it that every word, each word is chosen specifically for a reason and for, like that thing that he calls. I forget his words, but he had a word for it. But is the message of a printed Bible less information dense? I guess it is, but I think that's the difference is printed.
2:13:56 - Paris Martineau
Bible somewhat different than a recommendation letter where you're trying to make a decision based on. Is this going to be a good candidate for this job and this letter is supposed to provide me information.
2:14:08 - Leo Laporte
It's less valuable in that way.
2:14:09 - Paris Martineau
Yeah.
2:14:10 - Leo Laporte
Yeah, hey, it's been great having you back. Welcome back, Great to have all of you watching. We do Intelligent Machines every Wednesday at right after Windows Weekly, which is 2 pm Pacific, 5 pm Eastern, 2100 UTC. You can watch us live. There are eight different streams Club members watching the Discord, but you can also watch anybody you can, including club members, on YouTube, twitch, xcom, tiktok, facebook LinkedIn, and there's one more Facebook LinkedIn, xcom, tiktokok. I'm missing one, but anyway you get. Oh no, that's eight. I just couldn't count my fingers right we can count.
Yeah, we're, we're qualified to be talking about artificial intelligence uh, if you don't want to watch live on-demand versions of the show, audio or video, we've got both available at twitter. Tv slash im. You can also, of course, find a youtube channel dedicated to it. There's a link on the im page. Or maybe the best thing to do, subscribe in your favorite podcast player uh, that way you'll get it automatically. You don't have to think about it. And if you are on itunes or pocketcasts, or if the site that you get your you know your podcast player from has reviews, or the app does, please leave us a five-star review. Let the world know that this week in google is now I am, and it's damn fine, damn fine. Uh, those five stars help us a lot. We appreciate it again.
New, uh, new yearly plans for club twit. Go to twittv slash club twit. If you're already a member, you can switch from the monthly to the yearly. If you're not a member, you take your pick. We want to have you in the club. It makes a big difference. Twittv slash club twit. Thank you, paris Martineau Find her work at the information she had to put on the headphones. We did it long enough that the earbuds died.
2:15:58 - Paris Martineau
My AirPods are constantly on their last leg, despite the fact that these are a relatively new pair of AirPods. Apple just continues to taunt me.
2:16:08 - Leo Laporte
And of course, you can't replace the batteries, so you just have to get a new pair.
2:16:11 - Paris Martineau
No, and you know you'd think you get a new pair that will solve your problem that one of the AirPods constantly doesn't charge. But no, it won't, but no, but no, it doesn't. It doesn't Can, but no, it won't, but no, it doesn't it doesn't. Can AI fix that? No, Can these Bose headphones that I've had for seven years will they ever die? No, because you know that's the world we get to live in.
2:16:36 - Leo Laporte
That's the world we live in. It may not be perfect, but it's ours. Thank you everybody. We'll see you again next Wednesday for Intelligent Machines Ta-ta.
