IM 824: Full-Body Air-Quotes - AI in Higher Education

Summary

Intelligent Machines Podcast - Episode IM 824: "Full-Body Air-Quotes - AI in Higher Education"

Release Date: June 19, 2025

In Episode 824 of the Intelligent Machines podcast, hosted by Leo Laporte from TWiT, the focus centers on the transformative impact of Artificial Intelligence (AI) in higher education. The episode features insightful discussions with Matthew Kirschenbaum, a distinguished professor of English and AI at the University of Virginia, alongside co-hosts Paris Martineau and Jeff Jarvis.

---

1. Introduction to Matthew Kirschenbaum

The episode begins with an introduction of Matthew Kirschenbaum, highlighting his unique position as a future professor specializing in the intersection of English and AI. Kirschenbaum's multifaceted interests include teaching traditional skills like letterpress and engaging with modern AI technologies. His extensive work, including his book "Track Changes", provides a historical perspective on the evolution of word processing and its implications for writers.

Notable Quote:

"Matthew is a wonderful mix of interests. He is a professor of English. He teaches students how to do letterpress... and he's now in AI."
— Paris Martineau [02:19]

---

2. AI Tools in the Classroom

Kirschenbaum delves into the pervasive use of AI tools among students, addressing both the potential benefits and challenges. He shares his experience teaching a class assignment where students used a large language model to analyze Charlotte Perkins Gilman's "The Yellow Wallpaper". Instead of a traditional essay, students engaged in a dialogue with ChatGPT, prompting the AI to generate alternative endings for the story. This approach unexpectedly deepened students' engagement with the original text, as crafting effective prompts required meticulous attention to detail.

Notable Quote:

"They ended up paying a lot more attention to the actual text of the Yellow Wallpaper than they otherwise would have."
— Matthew Kirschenbaum [07:03]

---

3. MLA Task Force on AI in Education

The conversation shifts to the Modern Language Association (MLA) task force dedicated to integrating AI in educational settings. Kirschenbaum emphasizes the committee's stance against one-size-fits-all mandates, such as Ohio State University's directive to incorporate an AI component in every class. Instead, the MLA advocates for academic freedom, allowing individual instructors to tailor AI usage to their specific curricular needs. The task force also underscores the importance of critical literacy, enabling students to discern the appropriate application of AI tools in their academic work.

Notable Quote:

"The MLA committee... affirmed the value of academic freedom... and empowering the students to make intelligent decisions about how much of their own voice and authority they want to cede to a machine."
— Matthew Kirschenbaum [08:38]

---

4. Challenges of Academic Integrity with AI

A pressing concern discussed is the potential erosion of academic integrity in the age of AI. Kirschenbaum shares anecdotes about professors employing deceptive tactics, such as embedding hidden prompts in assignment instructions to detect AI-generated work. However, he critiques these methods as ethically questionable and ineffective in the long term. Instead of punitive measures, Kirschenbaum advocates for leveraging such challenges as teachable moments, fostering a deeper understanding of AI's role and limitations among students.

Notable Quote:

"It seems like a huge waste of not only resources but a wasted opportunity. To me, this should be the proverbial teachable moment."
— Matthew Kirschenbaum [14:04]

---

5. The Concept of "Textpocalypse"

Introducing his concept of "textpocalypse", Kirschenbaum explores a future where AI-generated content inundates the internet, making it challenging to authenticate human-authored works. He raises concerns about the subtle manipulation of cultural heritage materials online, where deepfakes and AI-generated texts could distort historical and cultural narratives. Kirschenbaum suggests that institutions like libraries and archives will play a crucial role in authenticating original works, ensuring the preservation of genuine human expression.

Notable Quote:

"So the textpocalypse is a term that I believe I coined... where you begin to have fewer and fewer what we read on the Internet is actually written by people."
— Matthew Kirschenbaum [15:49]

---

6. Historical Analogs: Word Processing and Writing Evolution

Drawing parallels to the past, Kirschenbaum discusses how the advent of word processing in the late 20th century revolutionized writing practices. His book "Track Changes" examines the skepticism and eventual adoption of word processors among prominent writers like Isaac Asimov and Anne Rice. This historical perspective serves as a lens to understand current apprehensions and adaptations surrounding AI tools in writing and education.

Notable Quote:

"Anne Rice was an early adopter and famously gave a copy of Wordstar to the vampire Lestat in the Vampire Chronicles."
— Matthew Kirschenbaum [27:12]

---

7. Impact of AI on Cognitive Processes and Learning

The episode touches upon recent studies, such as one from MIT, investigating the cognitive implications of AI-assisted essay writing. Preliminary findings suggest that reliance on tools like ChatGPT may lead to diminished cognitive engagement, referred to as "cognitive debt". Kirschenbaum reflects on similar debates from the era of word processing, emphasizing the need for empirical research to inform educational strategies in integrating AI.

Notable Quote:

"They discuss a study... where students that performed essay writing tasks using GPT-4... performed worse on both behavioral net metrics as well as more objective, I guess, brain scans."
— Jeff Jarvis [28:02]

---

8. Personal Reflections and Broader Implications

Throughout the discussion, the hosts and Kirschenbaum interweave personal anecdotes and broader societal implications of AI integration. They debate the balance between technological advancement and the preservation of human-centric skills, advocating for a nuanced approach that harnesses AI's potential without compromising critical academic and cultural values.

Notable Quote:

"There's no one answer. There were some writers who immediately embraced it, and others who were holdouts. Some began to pay more attention to their prose once they realized they could revise it indefinitely."
— Matthew Kirschenbaum [28:02]

---

Conclusion

Episode IM 824 offers a comprehensive exploration of AI's role in higher education, balancing optimism with caution. Matthew Kirschenbaum provides a scholarly perspective on navigating the complexities of AI integration, advocating for critical engagement and institutional support to maximize benefits while mitigating risks. The conversation underscores the necessity of adapting educational paradigms to embrace intelligent tools, ensuring that the essence of learning and human creativity remains intact.

---

Listen to the full episode here.

Transcript

[Leo Laporte]

It's time for Intelligent Machines. Paris Martineau is here. Jeff Jarvis is here. Our guest, Matthew Kirschenbaum, future professor of English and AI at the University of Virginia. We'll talk about how the tools you use change the way you write. We'll teach Paris to speak fluent Shakespeare and what would you do to get a hundred million dollar a year job? All that and more coming up next on Intelligent Machines. Podcasts you love from people you trust. This is TWiT. This is Intelligent Machines, episode 824, recorded Wednesday, June 18, 2025. Full body air quotes. It's time for Intelligent Machines, the show where we cover the latest news in intelligent machines, AI robotics, and the smart little doohickeys surrounding us every day of our lives. Now, gosh darn it, Paris Martineau is here from tech journalist, investigative reporter and spy or something, I don't know. Oh, she's got a book we're gonna talk about.

[Jeff Jarvis]

It's got a book. I'm sorry, it might be a spoiler.

[Leo Laporte]

It's not really because it's the old book, but anyway.

[Jeff Jarvis]

Well, it's a spoiler for who our guest is, which I'm sure they already know, if they've clicked on this, but, you know, I just wanted to.

[Leo Laporte]

It's not gonna be a spoiler for very long because we're gonna introduce Matthew in a second. But first, let's say hello to the emeritus professor of journalistic innovation who also has the book Craig Newmark, Graduate School of Journalism.

[Jeff Jarvis]

We're bringing it back, baby.

[Leo Laporte]

He's not there anymore. He's at Montclair State University and SUNY Stony Brook. Mr. Jeff Jarvis, author of many fine books which are no longer visible in the frame, but would be if he just moved his shelf around.

[Paris Martineau]

Well, yeah, I gotta move. Moving things.

[Leo Laporte]

The Gutenberg parenthesis. The web we weave and magazine. Hello, Jack.

[Paris Martineau]

Hello.

[Leo Laporte]

Introduce our guest because I think you're friends.

[Paris Martineau]

Yes, Matthew Kirschenbaum is a good friend and mentor of mine. Matthew is now a. I forget the exact title. Matthew at University of Maryland, a distinguished.

[Matthew Kirschenbaum]

University professor, which can also be abbreviated as a dupe.

[Paris Martineau]

And announced but not occurring yet.

[Leo Laporte]

Is.

[Paris Martineau]

Is going to a chair in English and AI at the University of Virginia, his alma mater.

[Leo Laporte]

Thomas Jefferson's University of Virginia. Let's not. That's not. Yes, that's the one. Wow. So wonder if Matthew would have thought about AI having a professor of AI. That's a few of them. That's interesting.

[Paris Martineau]

So Matthew is a wonderful mix of interests. He is a professor of English. He teaches students how to do letterpress. He makes a lot of little Ben Franklins and. And he's now in AI. He also was kindly. My publisher sent a Gutenberg parenthesis to Matthew in a much longer version I'm embarrassed about. And he was the infamous reviewer number one.

[Leo Laporte]

Too many notes, Mr. Mozart. Is that what you mean?

[Paris Martineau]

Kindly outed himself to me. So I got great advice from him and I've learned a lot from Matthew in the meantime and just delighted to have him here because among the other things that he does, he's on an MLA Modern Language association task force about AI in the classroom. So between print to AI and text and track changes and all of that, we have lots to talk about.

[Leo Laporte]

We should mention that the book you two are holding up, the book all three of us have read, was Matthew's previous tome, which was a history of word processing and fascinating. There's a lot to it. You know, it's a great story, but maybe word processing is coming to a rapid close now. We're just going to ask ChatGPT to type it up.

[Jeff Jarvis]

No more words.

[Leo Laporte]

No more.

[Jeff Jarvis]

Just processing.

[Paris Martineau]

Too many words. Right, Matthew?

[Matthew Kirschenbaum]

Too many text. Apocalyptic, you might say.

[Leo Laporte]

One of the. One of the real issues in academe, of course, is that students seem to be so enamored of these AI tools that they're not doing their reading, they're not doing their studying, they're not doing their writing. You're letting AI do it. If you're going to be a professor of English and AI, this is. You're going to be confronting that head on.

[Matthew Kirschenbaum]

Yeah, I think that's true. I think it's. So, first of all, just thank you all for having me on. I've been looking forward to this. And thanks, Jeff, for that very, very funny and kind introduction. I think it's hard to generalize about where students are with AI these days. You certainly see things. There was this story, and I think it was New York magazine the other week, and really alarming statistics about percentages of students using AI tools to write their papers and so on. In my experience teaching here at Maryland, students are all over the place. A lot of them still don't even have a full sense of themselves, what these tools are, let alone how to use them effectively. And that's really what I see as the big challenge for not just higher education, but education at all levels in the coming years is to really educate people about what these tools are good for, but also the ways in which they. They may not be so good at all things or at all times.

[Paris Martineau]

How are you using it in the classroom?

[Leo Laporte]

In.

[Matthew Kirschenbaum]

In limited ways. And so I, with a colleague, I taught a class this past year, for example, where students were had an assignment to use a large language model to offer a reading of a really well known short story, the Yellow Wallpaper by Charlotte Perkins Gilman. Something that pretty much every English major will read at one point or another. And rather than asking students to just write yet another sort of interpretation and close reading of the text, we have them first query the model about the story and engage in a critical dialogue with the model. And they were surprised. Initially the students were to find that ChatGPT already had read the Yellow Wallpaper, which as part of its training data, so it knew all about it. But then they also needed to work towards getting the model to produce an alternative version with an alternative ending of the story. And for the upshot of all of this was, was that they ended up paying a lot more attention to the actual text of the Yellow Wallpaper than they otherwise would have. They really needed to sort of get down into the weeds in order to craft effective prompts that then produced interesting narrative outcomes.

[Leo Laporte]

Did you trick them into it though?

[Matthew Kirschenbaum]

Really tricked them into it. Only in the sense that really for me, the pedagogical goal here was precisely that kind of close reading and attention to detail in the original text.

[Leo Laporte]

And the insight was just a little extra.

[Matthew Kirschenbaum]

Yeah, well, it was a kind of Trojan horse, if you will, for that kind of exercise.

[Leo Laporte]

I think a good idea actually. You know, don't tell them, don't use it, because that's going to be exactly the wrong direction to go.

[Matthew Kirschenbaum]

But it's, you know, I think the other thing I've learned is just as it's hard to generalize about where students are, it's hard to generalize about the role of the tech in higher ed at large. I mean, what I can do in an upper level class for English majors is very different from what a writing instructor can do in a room full of first year composition students who are there for the express purpose of learning how to express themselves discursively and rhetorically. So there are going to be different kinds of approaches depending on what it is you're teaching, who you're teaching and so forth.

[Paris Martineau]

I was impressed with the, the MLA report that you were part of.

[Leo Laporte]

What does that stand for?

[Matthew Kirschenbaum]

Modern Language Association. It's the North American Professional association for teachers, teachers of literature and the other Romance languages.

[Paris Martineau]

And I'm sure there's, there's straight lines and jokes about what happens when they all get together. But they do.

[Matthew Kirschenbaum]

Yes, there have been novels written.

[Leo Laporte]

They speak in Esperanto, I'm sure.

[Paris Martineau]

And so there was a task force about AI in the classroom. And it not, as one might have expected, retrograde to the contrary. Matthew, I wonder if you could just talk for a minute about what the recommendations are, but also the process to get there and what you're hearing from other professors about this.

[Matthew Kirschenbaum]

Yeah, I mean, so we've there, you know, the task force being a task force, it's produced several different working papers to date with a number of different kinds of recommendations for instructors, for administrators, even for school. I think the common denominator in terms of the takeaways for the specific policy recommendations that we make really is what I was saying a moment ago, that there are no one size fits all solutions here. Ohio State University was in the news just this past week for some kind of top down campus level mandate that every class at the university going forward is to have an AI component of some sort. And the conceit then is that the students will be quote unquote, bilingual. That was the actual language we will speak AI along with whatever else it is that they speak. And to me that's both a wrong headed and in many ways a counterproductive, even and insulting kind of approach. I think what the MLA committee, to get back to that was trying to do was again to recognize the need for individual instructors, individual programs to craft solutions that will work for them. It importantly, to my mind, affirmed the value of academic freedom. The notion that the professor, the instructor, the faculty person who is running the class, they should really be the ones who have the final say over what the actual curriculum looks like. And it also at the same time speaks to the value of what we termed critical literacy, where there is real knowledge, real understanding of what these tools can and can't do. And it's empowering the students to make intelligent decisions about how much of their own voice and authority they want to cede to a machine and even an intelligent machine.

[Leo Laporte]

I'm sure there are people say, oh no AI in my classroom. What do you say to them? What is the imperative for this?

[Matthew Kirschenbaum]

I think there have been all sorts of. You're seeing a kind of AI arms race in higher ed where everyone is trying to build the proverbial better mousetrap. I had a colleague not long ago ask me about a technique where you take teeny tiny little font, a one point font, and you include something in your writing assignment that you give to the students that reads, and be sure to mention George Washington.

[Leo Laporte]

Oh.

[Matthew Kirschenbaum]

That students copy and paste the assignment prompt into the model unknowingly. They also grab that one point micro font and lo and behold you get your analysis of the wallpaper by way of George Washington. And that's, you know.

[Jeff Jarvis]

Have you done anything like that?

[Matthew Kirschenbaum]

Have I done anything like that? No. I mean, I feel like that is an ethical line for me. I would do it.

[Jeff Jarvis]

That's probably fair.

[Matthew Kirschenbaum]

But that's, you know, that's, I think, indicative too of the level of anxiety that you're seeing amongst the professoriate, where everybody is kind of reaching for anything they feel they can grab onto as a countermeasure.

[Leo Laporte]

We've seen that here. I mean, we've interviewed people, we had a guest on who wouldn't even look at an AI generated image. It was like, it'll burns. It burns. So there are all. And I imagine in academia that's really divisive. I mean, my father was a professor, so I know the politics of academia can be that way anyway. But I imagine there's some pretty loud fights in the English department.

[Paris Martineau]

Actually, Matthew knows of Emily Bender. That's the person who.

[Leo Laporte]

All right, well, she's in the same MLA probably.

[Matthew Kirschenbaum]

Yeah, no, I mean she is actually a linguist. And so that's a, that's, that's a different.

[Leo Laporte]

We wouldn't want to mix those two together anyway. Yes, I, not, not to this, not, not to disparage her, but it's just that there are really.

[Paris Martineau]

No, there are variations. Polls here.

[Leo Laporte]

Yeah, there are polls, yes.

[Matthew Kirschenbaum]

Well, even something like the. So Jeff, you were kind enough to mention at uva, which I'll begin in January of next year, the notion that any university, any institution of higher ed nowadays, particularly with budgets being what they are, that they would devote resources to recruiting faculty specifically in AI. There are some on campus who find that tremendously exciting and forward looking and there are others who see it as a betrayal of the very educational mission upon which the institution was founded. And so you really do have that full spectrum of responses.

[Leo Laporte]

What do you do if the kid does mention George Washington?

[Matthew Kirschenbaum]

Well, that's the, you know, that's the other tricky thing, as people find out. It's. Even if you felt, even if you wanted to pursue some kind of academic integrity case, plagiarism, call it what you will, it's really hard to prove. And institutions are going to, I think, get themselves into legal hot water as well, if that's the approach they insist on sticking to. Because these things get very Very fraught. Again, very difficult to actually prove. And in the end, it seems like a huge waste of not only resources, but a wasted opportunity. To me, this should be the proverbial teachable moment.

[Leo Laporte]

Yeah, it's kind of a paranoia that perhaps is not appropriate. Paris and Jeff will tell you I'm kind of an AI buff, a little bit of a fan, but I've always. I'm a technology fan. I mean, that's what I do for a living. And I've always said to. Even in, you know, kids in younger grades, if you forbid them access to technology, what are they going to do when they get out of the real world? You know, part of the job is to teach them how to use it, how to deal with it, how to be in an environment where it exists, not to eschew it.

[Matthew Kirschenbaum]

Yeah, I think that's right. And again, I think it's easy for you see things like the professor who reverts to old school blue book examinations handwritten. I mean that sort of thing is a good attention getter for a headline, but by and large I don't think that's indicative of where the professoriate is actually.

[Leo Laporte]

That's good to hear. Yeah.

[Paris Martineau]

Talk about your theory of the text apocalypse which you wrote about in the Atlantic.

[Unnamed Speaker]

Yeah.

[Matthew Kirschenbaum]

So textpocalypse is a term that I believe I coined in early 2023, a couple of months after the release of ChatGPT and everybody was suddenly paying attention to AI and being an English professor, I have a particular interest in text as an actual format, if you will, a mode of output. And what I tried to articulate with this piece that did appear in the Atlantic in, I believe it was March 2023, was a kind of near, not so far future scenario in which you begin to see human beings like ourselves increasing, increasingly dissociated from the act of writing. And the specific form that that takes is that you have large language models. That this was basically unknown at the time, all the way back in early 2023, but it's now becoming commonplace. You have large language models that are wired directly into the live Internet. You have a situation where models can also prompt other models. And of course, as we know, models are also scraping machine generated content from the open web to use in training future generations. So the textpunkalypse on the one hand is a kind of sci fi ish scenario where you begin to have fewer and fewer and fewer. Less and less of what we read on the Internet is actually written by people that actually may well be statistically True already. We're all familiar with spam and junk content of all sorts on the Internet well before the adventure.

[Leo Laporte]

That's true. There's plenty of slop there already.

[Matthew Kirschenbaum]

Isn't there's plenty of slop there already. I do think things have and are going to continue to qualitatively change. There's actually a theory, some might call it a conspiracy theory of the so called dead Internet, which suggests that most everything that we read on the Internet is already synthetic and the product of manipulation by secretive kabbalistic globalist forces.

[Paris Martineau]

Soros. Huh?

[Leo Laporte]

You've seen John Graham coming. In fact, I'm going to try to get him on. He's an old friend. Cloudflare created a low background steel website with uncontaminated writing. That writing that hasn't been contaminated by AI created content. I think that's an interesting point. I guess it's pre text apocalypse content.

[Matthew Kirschenbaum]

Yeah, well, right. I mean it becomes a little bit like the slow food movement where you have the equivalent in slow writing and it becomes a slight selling point when something is authenticated as 100% human authorship. I think we're perhaps not far from that point.

[Paris Martineau]

What kind of institutions do we need to establish humanity?

[Matthew Kirschenbaum]

Yeah, establishing humanity is a big question. I think one thing that institutions, what we call cultural heritage institutions like libraries, museums, archives, one thing that they're going to need to pay increasing attention to is the authentication of original renditions of important cultural heritage materials. By which I mean, how do you know that the image of the Mona Lisa that you're looking at on the Internet, how do you know that that's in fact a faithful reproduction, a faithfully executed digital surrogate of the original and that it hasn't been manipulated in some particularly subtle way? Right. I mean, we're familiar with the phenomenon of deepfakes and the sorts of things where you see Donald Trump running away from a flying squad of police officers rushing after him. Right. And everyone sort of understands that. That's. Everyone recognizes that for what it is.

[Leo Laporte]

Right.

[Matthew Kirschenbaum]

We're not taken in by that. But I think the danger is going to be things that are much more subtle, things that operate in contested realms of cultural heritage, like for example, medieval history these days or aspects of Americana which we probably seen weaponized by the right wing in the service of their own agendas and ideologies and being able to authenticate a culture, a stable consensus based cultural record. Just as nowadays we are seeing the need for a consensus based reality that seems to be more and more lacking.

[Leo Laporte]

It's Almost. I mean, almost feel like it's foolish to try to fight it. Certainly it'd be nice to be able to distinguish. But ultimately, I guess the question is, does it even matter?

[Matthew Kirschenbaum]

I think it matters. That may be my calling as a scholar and a humanist. I think part of what we want to do is also this is something else we know as people who study the evolution of books and documents and paintings and music and what have you, that cultural phenomena are not static. They do evolve, they do change over time. There's a term that I'm fond of in my work, transformation, A portmanteau of transform and transmit. How things transform as they're transmitted.

[Leo Laporte]

It's a giant game of telephone.

[Matthew Kirschenbaum]

Yeah. Or, you know, a modern adaptation of Hamlet. Right. So the work does transform, but I think we want to be. We want to see that happen in ways that are transparent and again, not in ways that are being nefariously manipulated or weaponized to suit particular cultural agendas.

[Paris Martineau]

Talk about analogs of technology's effect on writing in the past. You know, I've been reading up on the typewriter and the arguments that it changed how people wrote, that it made Hemingway, Hemingway and so on. And you have, I quote you, by the way, saying that it's foolish to mark just one device or one factor. But there are certainly. I think personally, when I went from typewriter to computer, it changed the way I wrote and thus the way I thought. And you covered this in Paris. Let's hold it up in Track Changes.

[Leo Laporte]

You don't have my Kobo. I'd hold it up if I could. Your E reader.

[Paris Martineau]

Which, which. Which, by the way, so you, you just to be.

[Leo Laporte]

You couldn't imagine post it notes you pretended to have read.

[Paris Martineau]

Hey, hey.

[Leo Laporte]

They were randomly placed. I could tell you would think this.

[Paris Martineau]

Would be a dry topic just for nerds, but it's actually fascinating. It really is. What was the attitude about people? Was there similarly an attitude of using computers was cheating, was corrupting. How did we adapt to the use of computers in writing back then? And does that have any lessons for now?

[Matthew Kirschenbaum]

Yeah. That was such a fun book to do. I got to talk to a lot of actual writers and to just rummage around in old computer. Computer magazines. And even just looking at the ads was fascinating. Radio Shack ran a series of ads with Isaac Asimov.

[Leo Laporte]

Oh, I remember those. Yeah.

[Matthew Kirschenbaum]

And so Track Changes is specifically a literary history of word processing. And yeah, I was interested precisely in the question of how mainstream writers, big names like Asimov, Stephen King, Anne Rice, a lot of this took me back to the fiction that I was reading as a teenager growing up in the 80s, because that was sort of the watershed moment for the emergence of the home computer market, which fiction writers, along with all the rest of us, were part of. There he is. And, yeah, so many of the same anxieties. You had Gore Vidal in the pages of the New York Times book review in 1984 declaring that word processing is erasing the literature. Asimov himself was originally a skeptic and claimed, as a number of writers did, that he was never going to make the change. And then a big part of what happened was peer pressure, where we're talking about an era from the very late 1970s through the first half of the 1980s. And, you know, writers I spoke to told me that, like, you went to a writers conference like Bread Loaf during those years, and the shop talk was all about computers. Everyone wanted to know what everyone else was getting and how big it was and so forth. And I think a number of writers began to feel like, not unreasonably, that they were going to fall behind and that they wouldn't be able to match their competition if they didn't get on board.

[Leo Laporte]

Do you think computers made him better writers?

[Matthew Kirschenbaum]

You know, that's something I studiously avoided answering.

[Paris Martineau]

Oh, professors are no fun.

[Leo Laporte]

You know, I remember going to the Museum of Pop Culture in Seattle, and they have Neal Stephenson's original manuscript of Snow Crash, which he. And, you know, he's a famous writer about the. I mean, he created the Metaverse. It's all in. He uses fountain pens on paper. It's this big. It's huge. He wrote it. He hand wrote it.

[Matthew Kirschenbaum]

Yeah. There's the famous story of William Gibson writing Neuromancer, the novel that gives us the word cyberspace on a typewriter. But, you know, I think the thing that I found was that there. Here again, there was no one answer. There were some writers who immediately embraced it, the technology. Anne Rice was an early adopter and famously gave a copy of Wordstar to the vampire Lestat in the Vampire Chronicles. So you saw this a number of times when writers would actually, as a kind of Easter egg, write their word processor into their own fiction.

[Leo Laporte]

That's funny.

[Matthew Kirschenbaum]

Tom Clancy did something similar. Other writers were, you know, were holdouts. Some writers said that they began to pay far more attention to their prose once they realized that they. They could revise it indefinitely. Others embraced the productivity gains. And this was important in genre fiction like science fiction, like romance, like crime fiction, where a writer might go from being able to do one novel a year to two or three. And that made a material difference in their ability to actually make a living and to. To be a writer. So there were all kinds of factors that went into these sort of stories of early adopters.

[Jeff Jarvis]

Paris, I'm curious, was there any. I feel like something we're starting to see now as the prevalence and use of these LLMs and tools like this skyrocket, we're starting to see kind of research that is trying to quantify whether or not use of these tools has any effect on kind of content recall or learning in general. There was a really interesting study that came out of MIT this week called your brain on ChatGPT accumulation of cognitive debt when using an AI system for essay writing that found, I mean, kind of unsurprisingly, it was a small study of like, I think 60 people, but that, that students that performed essay writing tasks using GPT4 oh, instead of like just their brain only kind of performed worse on both behavioral net metrics as well as kind of more objective, I guess, brain scans. Do you know, was there ever any, like, as we talk about this, I think people often think like, oh, studies like this are a bit of a moral panic when it comes to the technology of AI. Was there a similar kind of focus on computers at the time and like word processing tools?

[Matthew Kirschenbaum]

Yeah, there was, interestingly enough, and this gets us back to the role of English departments and the humanities, but there was a lot of interest and precisely that kind of empirical research that was being carried out in university writing programs as, as faculty weighed whether or not students should be allowed to use a word processor in their own writing. Should word processing be taught as an actual component of the curriculum? And you had people who are interested precisely in those kinds of cognitive questions. So you can, you can go back to the 1980s and read that literature with regard to word processing and its impact on the writing self, as it were.

[Jeff Jarvis]

Do you recall as, I mean, generally as to whether they had similar findings that word processing impacted, I guess, like your cognition or education?

[Matthew Kirschenbaum]

I. It's not quite my field.

[Jeff Jarvis]

I. I was going to say, I don't know.

[Matthew Kirschenbaum]

I didn't know the research would find that there were impacts. Exact nature of those impacts? I wouldn't want to speculate.

[Leo Laporte]

There's inevitably. I mean, look, our life is changing very rapid, rapidly, all the time, and there are impacts from everything and often pros and cons. I know when I journal and I do it in handwriting, it's a much slower process, but there's definitely a connection between physically writing it on the page. Although some people might say you should use a chisel and stone instead of a pen. That's just really.

[Jeff Jarvis]

Well, your current journal processing is you just have a button record everything you hear for a couple days and then you don't have to journal it all.

[Leo Laporte]

I'm wearing it right now. I don't need to journal anymore. Anymore. It's.

[Jeff Jarvis]

You don't have to do a single thought for me.

[Leo Laporte]

But at the same time as I noticed that that is true, it's also a pain in the butt. My hand hurts and, and it's messy and it's hard to read back. It's impossible to search. So I also nowadays keep a journal on a computer and well, yeah, maybe I'm, I'm. I've lost something by doing that, but I've gained something by doing that. I think that's true of a lot of this. Right. I think it's not a net gain or loss in any direction. It's just a change. It's change. We got to track our changes.

[Matthew Kirschenbaum]

That's right. Yeah. Someone should, someone should trademark that.

[Jeff Jarvis]

Someone should. Yeah.

[Matthew Kirschenbaum]

I think Microsoft observed too, is that AI large language models, tools like ChatGPT, these landed at a moment when people were already, I think, hypersensitive to their writing tools, to their workflows. There's been a lot of interest in recent years and things like Moleskine notebooks and journaling and having the particular craft fountain pen that you like to use. People think a lot about their workflow, how they take notes, how they compose their software like Scrivener, that's been on the market for a while now. And so into all of this sort of churn and self awareness about the actual nature of our writing tools. And inevitably at this point in the conversation, one quotes Friedrich Nietzsche and his famous aphorism that our writing tools help to shape our thinking as is language. As does language. Right. So that is the environment into which a tool like ChatGPT arrived.

[Leo Laporte]

And the question, these are my tools, by the way. And I'm very proud, very proud of my. I am showing my collection. Yeah, Well, I have a little. I can carry it around and when I need green ink, I've got. But the truth is it's just not practical. You know, when I.

[Paris Martineau]

There's also gel Leo gel. It's the future.

[Leo Laporte]

I have a. You know what, I have a state of the art gel pen that. This was a Kickstarter and it's, it's anyway that's another story for another time. I do note though, for instance, when I go to a therapist, I don't want them typing, I want them taking notes longhand. My doctor used to spend a lot of time typing and I felt like that was distancing my physician. Now when I go to my physician, he's got a big sign says I have an AI transcribing our conversation so I can make notes, so I can pay attention to you and not type. So there is a cultural thing.

[Paris Martineau]

What would you think if your therapist did that?

[Leo Laporte]

I wouldn't be happy. Isn't that funny? And yet.

[Jeff Jarvis]

And yet you're recording it already on your end.

[Leo Laporte]

That's just defensively. So that I have.

[Paris Martineau]

Yes, Hadi, I did tell you.

[Leo Laporte]

That has not worked, by the way. I just want to point that out.

[Jeff Jarvis]

Shocking you telling your wife that. Actually she did say something because you've got a little AI.

[Paris Martineau]

I have my mansplaining device right here.

[Leo Laporte]

Yeah, it doesn't work.

[Paris Martineau]

What, what research agenda? I'm curious what you want to research in your new gig, but what, what research agenda do you wish various disciplines would follow on what we study about the impact of AI, particularly around writing and text and culture?

[Matthew Kirschenbaum]

That sounds like a good question for the. The incoming Commonwealth professor of AI and English. I think there are lots of agendas that are out there. Everything from some of the more practically minded things that we've been discussing with regard to actual modes and techniques for education, their impacts to the deep historical questions that we've touched on. Patterns of historical change in my own work, and this comes back to the textpocalypse thesis that we touched on. I myself am especially interested in text as an actual category or mode of expression. That's not just the stuff, the words that you read on the screen. Text is that. And we've even, you know, text in my lifetime has become a verb. We now speak casually of texting people. But text is also a data type. Text can be manipulated and searched and transformed in distinctive ways that don't necessarily map onto other data formats like video or sound, for example. And text is something in today's world very much like a kind of extractable resource. One of the ironies of the current limits that we see with language models is the inability to find new sources of text upon which to train them. And so there's the. The paradox of the textpunkalypse is precisely on the one hand, the overwhelming torrents of text that we're confronted with now on a daily basis. But Also the simultaneous scarcity of it as a kind of resource. And so there's a, in addition to its legible qualities, in addition to the work that it does as a data format, programming code is essentially text. There's also a commodity value to text and that's the kind of three way nexus that I'm trying to unravel in my own current work.

[Paris Martineau]

What tools do you play with?

[Matthew Kirschenbaum]

I've been, I think like you, Jeff. I've been using NotebookLM recently and the kind of scholarship I do often depends on very sort of minute reconstructions of timelines, news that's breaking in the tech world on a day to day basis, responses to. Responses to responses on different social media platforms. And I found one use I found for something like NotebookLM is that it's very good at taking a mess of different documents and sources that you ask it to ingest and producing accurate summaries, timelines, chronologies of key events that with due spot fact checking then able to rely upon. So that's helpful in that regard. I think some of the work that we're seeing or some of the advances that we're seeing around voice technology are also about to be transformative. That gets us into a whole other realm of text.

[Paris Martineau]

Input or output voice.

[Matthew Kirschenbaum]

I'm thinking primarily of input, but of course output too.

[Paris Martineau]

Yeah.

[Matthew Kirschenbaum]

And I think we're on the verge of a new kind of, of orality as well in our relationship.

[Paris Martineau]

Well, I have to thank Matthew before we leave that he helped me. I don't think I told the story here. I wanted to try out a perplexity new model and I took a paragraph that I had written in hot type coming soon and asked it to do, to just riff on it. So demonstration for the show. And it gave me a beautiful page long exegesis on the transformation from digital mathematics with Leibniz to Morse code to Bideau. Sorry, I didn't talk about this because you corrected my pronunciation Leo to ASCII and so on. And it came back with this wonderful phrase about a cascade of symbolic abstractions. And I really wanted to use that phrase, but I didn't know what to do because I didn't want to say in the text of the book, as Perplexity observed, that'd be pretty dumb. Dumb. So I went to Matt on Facebook and I said professor, what do I do? MLA task force member and he advised a discursive footnote, which in fact it is.

[Matthew Kirschenbaum]

But that's the issue, right? Because that too is not going to scale. You can get away with it once and it's a novelty. But every time you use perplexity as a writing assistant, you're not going to be able to do that.

[Paris Martineau]

What is the. What are the ethics of that, do you think? Of citation?

[Matthew Kirschenbaum]

Yeah, I mean, I think maybe we need to sort of rediscover the colophon, where we talk about the technical ecology behind the work without necessarily. Yeah, without citing every individual passage, but make the reader aware of the kinds of tools or models that did that might have informed some of the prose.

[Paris Martineau]

Yeah, I love that. An updated colophon, for those of you who don't know, the colophon appears at the end of the book, which I have one at Gutenberg parenthesis that says what typeface was used and how it was produced.

[Leo Laporte]

They're rarer and rarer. I love them, unfortunately. I think it's a great thing, but you're a type nerd. Matthew, it's a pleasure meeting you. We loved track changes. I'm glad you're doing that scholarly work and I wish you all the best at your new position as a professor of English and AI at uva. There'll be some luxurious hell of a.

[Matthew Kirschenbaum]

Combo starting in January. Thank you all. Thank you all so much. This was great.

[Leo Laporte]

Thank you. Great to have you on.

[Paris Martineau]

Cheers.

[Leo Laporte]

Matthew Kirschenbaum, everybody, and Intelligent Machines. We'll continue with AI news in just a bit. Our show today, brought to you by our good friends at OutSystems, the leading AI powered application and agent development platform. This is such a cool story. For more than 20 years, two decades now, AD systems have been around their mission to give every company the power to innovate through software. But you may have noticed in the last few years things have changed a little bit with software and Outsystems is there there already. So you, if you are a business or an IT team, you know this famous conundrum. Build versus buy. IT teams typically have two choices. You could buy off the shelf SaaS products for speed, but you lose some flexibility and you certainly lose differentiation because your competitors are probably using the same stuff or okay, we don't want to buy, let's build. Oh my God, I've done that. Building custom software costs time and money and often what you get is less, let's say suboptimal less than the best. While there's a new path between build and buy, AI forges the way for a new idea. The fusion of AI, low code and DevSecOps automation into a single development platform. Your teams will build custom Applications. Yes, you heard the word build, but they'll be doing it with AI agents as easily as buying generic off the shelf sameware. Plus because you're using Outsystems, flexibility, security, scalability are already built right in. With AI powered, low code teams can build custom future proof applications at the speed of buying. With fully automated architecture, security integrations, data flow, permissions, everything you would want is already done. Outsystems is, I think it's going to be the last platform you'll ever need to buy because you can use it to build anything. And oh yes, by the way, customize and extend your existing core systems. Build your future with OutSystems. Visit OutSystems.com TWiT to learn more. That's OutSystems.com TWit we thank him so much for supporting intelligent machines. This is the place to come if you have an intelligent product. You were talking about text to speech and it reminded me of this new Chatterbox. Have you played with this at all?

[Paris Martineau]

A little bit, yeah.

[Leo Laporte]

It's pretty cool. So this, it's on hugging face because like a lot of open weight models, you know, hugging face can host it. So what you do is you give it some text, but then you read or record some reference audio. So I'm going to record some reference audio in my voice, right? So this is just, just my voice. It's not related to the text. It's just me talking into the. The thing just to give it. I think sometimes they call it prosody to give it the outline of how it should sound. So now I have a little recording there. You can change the exaggeration. Let's turn it up so it's really. And you can change the pace.

[Paris Martineau]

Radio voice is really radio now.

[Leo Laporte]

Yes, it'll be very radio. And now I'm going to burn a couple of rainforests real quick. And I've created the output. I took Polonius's famous speech from Hamlet. I don't know what this is going to sound like. Neither a borrower nor a lender be, for loan oft loses both itself and friend, and borrowing dulls the edge of husbandry. This above all to thine own self be true, and it must follow as the night the day thou canst not then be false.

[Paris Martineau]

Get off any man.

[Leo Laporte]

I gotta turn down the exaggeration. That was a little. A little much. Maybe I should make myself a. But isn't that impressive? You can take any voice and style the output. Let me try. This is a little more calm. Neither a borrower nor a lender be for loan oft loses both itself and friend. Sort of sounds like the edge of sort of.

[Jeff Jarvis]

Yeah.

[Leo Laporte]

This above. Do you want to try it? You can, you can. I'll record your voice if you want, Paris. Go ahead. Just in my voice, right? So this. Oh, that's my voice.

[Jeff Jarvis]

No, I think. I think you need to. Should you will it pick up my. My sound if I talk?

[Leo Laporte]

I think you're in the computer.

[Jeff Jarvis]

So am I in the computer?

[Paris Martineau]

You're in the matrix, Paris.

[Jeff Jarvis]

I'm in the matrix. And now I'm talking and continuing to talk in a way that potentially could be turned into something that would create sound that then could be read back to me.

[Leo Laporte]

All right. And how. How exaggerated do you want to be?

[Jeff Jarvis]

I have no idea.

[Leo Laporte]

Do you want to be low?

[Jeff Jarvis]

Let's do high exaggeration.

[Leo Laporte]

Oh, I could go all the way, baby. Anything.

[Jeff Jarvis]

Turn it up to Matt.

[Leo Laporte]

It says extreme values can be unstable.

[Jeff Jarvis]

Great. Let's make it unstable as hell.

[Leo Laporte]

This is the crazy version of Hamlet, right?

[Jeff Jarvis]

Yeah.

[Leo Laporte]

It got a little bit of Jeff's voice in there.

[Paris Martineau]

Sorry. It's gonna. It's gonna. It's gonna average us out.

[Leo Laporte]

Let's see.

[Jeff Jarvis]

Either A borrower nor a lender. B for loan off lose both itself and friend and borrowing doles the edge of husbandry. That's bad. Pretty good, right, Paris? Than it did to you not then be false to any man.

[Leo Laporte]

Cuz Paris sounds more like that.

[Jeff Jarvis]

That's true.

[Paris Martineau]

So I was trying to find a story. I can't find it right now.

[Leo Laporte]

There's a chatterbox. I think that's kind of cool.

[Paris Martineau]

I think it's great. I've talked about this a lot that I want a markup language for audio output. So you can say joke sad, happy.

[Leo Laporte]

Oh, what a good idea.

[Paris Martineau]

There is. There is a company that put it in the tag. I can't find it right now.

[Leo Laporte]

Yes.

[Paris Martineau]

So it's a markup. So you can use that kind of technology, but then adapt it within rather than just be really fast the whole time. You can say this is the. This is the sad part.

[Leo Laporte]

That thing that exists. That sounds familiar from.

[Paris Martineau]

I've talked about the show before as a wish.

[Leo Laporte]

No, I think that exists. Oh, I remember now. In the old days. And this was the crappy speech synthesis that you used to get, you know, back, I don't know, 20 years ago. Yeah, yeah. And it came with a sound blaster. You could write text and then there were different diacritical marks. You could have to have it Go up or down or whatever. So then that exists. The problem is it was, you know, I am talking like this. It wasn't very good. We've gotten better. But that Chatterbox is demonstrating something that Google has also demonstrated, which is you can have a base model and then apply prosody to it. That's what Google's doing with a lot of its text to Speech 11 Labs. I don't, I'm not sure how they work, but I suspect they're working very similarly.

[Paris Martineau]

And you said there was a story.

[Leo Laporte]

Close almost perfect reproduction of voices at this point.

[Jeff Jarvis]

Yeah, I will say. I mean this is the most rudimentary use of this but I was in the middle of reading an article but like had to, I don't know like do my makeup or something for a meeting I was going to and I was like well I want to keep reading this but it doesn't have like the button on the article to like voice it. So I just took a PDF, sent it to Chat GPT and was like read this section, then this section and it was delightful.

[Leo Laporte]

Yeah, it works right now you can.

[Paris Martineau]

Now you can read to yourself.

[Jeff Jarvis]

It took like 30 seconds. You know, I mean I have zero interest in hearing my own voice read it. But someone else's voice that a robot is making quite fun.

[Leo Laporte]

There's a new so as you know, Pocket, Mozilla bought Pocket and then promptly killed it. Instapaper is still around, but it's kind of old, old school. There's a new company called Readwise that is basically the idea is to take over from. In fact you can import your Instapaper and Pocket articles into it and to take over but it has a built in exactly what you want text to speech so that it. You can take your. So you, if you save something you can have it read it to you. So.

[Jeff Jarvis]

Oh that's really nice.

[Leo Laporte]

Yeah And I actually haven't tried this so I don't.

[Jeff Jarvis]

I'm curious as to how its voices like how its affectation is because I think that's something I've realized I'm just very picky about is I want.

[Leo Laporte]

Yeah, it could be annoying. Yeah it could be very annoying if it's just like all kinds of.

[Jeff Jarvis]

There's lots of voices recently have found like. Like two books that I like one book that I actually liked the audio narration so much that I have like read the full book like listened to the full book and maybe that's just been that I've picked a lot of weird non fiction books that I guess didn't have particularly good narrators, but the quality of narration is everything. If it's. If the person reading sounds like they don't know what the end of the sentence is, I can't listen to it.

[Leo Laporte]

I completely agree.

[Paris Martineau]

Yeah.

[Leo Laporte]

And for a long time I did not like AI voices. I think they've now gotten to the point where they actually.

[Jeff Jarvis]

Yeah, they are quite good. I mean, I listened, I will say I listened to like a whole 4000 word article right by the Atlantic has.

[Leo Laporte]

All of its articles that way. Right?

[Jeff Jarvis]

Yeah. And a lot of. I remember when that first, that feature first came out in a lot of websites. I think the Times was one that had kind of prominently displayed it early on and it just. I remember it sounding awful and I was like, it just. This doesn't seem.

[Paris Martineau]

Well, sort of their podcasts, but.

[Jeff Jarvis]

Well, yeah, but like now I think we're finally a place where the technology is absolutely there. And I'm really excited to see that as like, I think that an ideal way for it to be implemented to me, which I'm sure is a bit more technically complicated but not impossible certainly in our day and age, is like a combination of the two where I could be a thousand words in to the article. I could right click on a word and be like, read to me from this place.

[Leo Laporte]

Well, good news. Google has added audio overviews to its search results. So now when you make a search, you can have Google read it to you. Let me see.

[Jeff Jarvis]

Read the AI overview. Okay. The Google's AI overviews are so frequently wrong, though.

[Leo Laporte]

Yeah.

[Jeff Jarvis]

Comically wrong.

[Leo Laporte]

Let's see now.

[Unnamed Speaker]

The Google AI reviews have become the new ad to me. I skipped that whole block now.

[Leo Laporte]

I completely agree. In fact, I would turn this off. This feature is a part of Google's labs. Transform your Google search results. So it's not necessarily the AI results. I think it's all the results into engaging audio overviews. What I think this is is Notebook lm.

[Paris Martineau]

Oh, sure. Yeah, I think.

[Leo Laporte]

Right.

[Paris Martineau]

And it's really impressive how that team has created things that Google's very proud of and is expanding all over.

[Leo Laporte]

Yeah, we're gonna are. We're talking to Steven Johnson soon, right?

[Paris Martineau]

Yep, yep.

[Leo Laporte]

Okay.

[Paris Martineau]

July, maybe July. Early July. Right.

[Unnamed Speaker]

We're trying to get him for the day after his big announcement.

[Leo Laporte]

He's got an announcement. We can't say what it is because we don't know what it is.

[Unnamed Speaker]

Yeah.

[Leo Laporte]

But he wanted to do it and kind of do it here. Oh, here it is. Generate. Okay. So the question was, how do noise Cancellation headphones work. Here's the button in Search Labs. Generate audio overview. And here is what Google did. This is not an AI. Well, it is, but it's summarizing Reddit, the New York Times, YouTube and other sites. Nine sites.

[Jeff Jarvis]

Ever wonder how noi canceling headphones managed?

[Leo Laporte]

Oh, it is not. I know her. Noisy.

[Paris Martineau]

The I.

[Jeff Jarvis]

It's like magic, isn't it?

[Leo Laporte]

I know her, but it's actually pretty cool science. I know him too. Two forms of noise reduction going on. Passive and active. I like it that they're swallowing their words like a human.

[Jeff Jarvis]

How earmuffs work.

[Leo Laporte]

Exactly. Think of it as a physical barrier. The ear cuffs themselves, especially on over ear.

[Jeff Jarvis]

I hate this.

[Unnamed Speaker]

Are designed to create this.

[Leo Laporte]

And dense materials can block more sound.

[Jeff Jarvis]

So like sticking your fingers in your ears. What about earbuds that fit snugly? Same principle.

[Leo Laporte]

Yep. The better the seal, the more sound.

[Paris Martineau]

Torture. Torture.

[Jeff Jarvis]

The kind of.

[Leo Laporte]

Ah. Why do people. You're right. I. I don't know anybody who likes this.

[Jeff Jarvis]

Why do they do this happens. Right. Can you pause it?

[Leo Laporte]

Active noise cancellation. No, I'm enjoying it.

[Unnamed Speaker]

It's a. Because they can situation.

[Leo Laporte]

It's impressive.

[Jeff Jarvis]

I wanted to rant, but I didn't want to rant over it because it was starting to annoy me to a point.

[Leo Laporte]

I.

[Jeff Jarvis]

It reminds me of like, I feel like there should be a name for this. The sort of strange feeling when you get when you read something like, oh, that's the classic chat. No, but it's the chat GPT format where it's like, chat GPT loves to write stuff like, thanks for sending that, M Dash. I really enjoyed it. Like a very specific form of it. This reminds me of like the audio podcast version of it is where these LLMs kind of have a go to formatting and just the. Well, what do you think that is? This person. Well, you may have. And then they're kind of.

[Leo Laporte]

I think that's correctable. I think that's how it's just.

[Jeff Jarvis]

Just be my.

[Paris Martineau]

Be a good assistant to me. Don't waste my time. Get to the point.

[Jeff Jarvis]

Yeah.

[Unnamed Speaker]

Yeah.

[Jeff Jarvis]

Why are you trying to make me feel like.

[Paris Martineau]

Yeah.

[Unnamed Speaker]

Why are you trying to be a person? Just give me the information I asked for.

[Paris Martineau]

Right.

[Leo Laporte]

This is that. Sick of.

[Jeff Jarvis]

It reminds me of the infamous Google presentation where they were talking about the assistant who would book a haircut for you, and they were like. And like, doing all these little kind of voice affectations. You do that very well.

[Paris Martineau]

Yeah. You make very good robots.

[Leo Laporte]

You. Yeah.

[Jeff Jarvis]

I believe you made up Frustrating to what end?

[Unnamed Speaker]

That summary also didn't get going until the first. The first ten seconds was just.

[Leo Laporte]

It wastes your time. Just like a podcast. I. You know, it's my point. We're here with YouTube videos. I know. We do it, too. I've always. From day one, I've said, you got to give value right away. Don't waste people's time. But it's hard not to because we, you know, we don't get together but once a week and we got to exchange pleasantries. But it's the thing I hate most about YouTube videos, and I think some of that's built into the perverse reward structure because, you know, you get rewarded for having a longer video, so.

[Paris Martineau]

Which is true of articles now, too. They wait 20 paragraphs to get to the damn point that the lead and the nut graph are gone. But it does know better. I put my entire manuscript for Hot Type into notebooklm and I asked it, what writer's ticks do I have? Because I always have them.

[Leo Laporte]

That's a good idea, actually.

[Paris Martineau]

What would Google do? My editor was writing ready to kill me because I used the verb enable like a hundred times. That's Google enabled all these things.

[Leo Laporte]

Yeah.

[Paris Martineau]

So this time it was I'm going to forget it instead of. And it was, oh, shit, now I'm forgetting.

[Jeff Jarvis]

Wow.

[Paris Martineau]

As well as. As well as like 35 times. And it very nicely said. It could just be. And.

[Jeff Jarvis]

Wow, I need to do this now. I need to put all my work into an LLM.

[Leo Laporte]

And frankly, we could do it with our podcast because, I mean, my ticks are.

[Jeff Jarvis]

I try to. I used way too many EM dashes. But specifically. And this is something I. Even. So I'm starting to get more.

[Leo Laporte]

I would appreciate that.

[Jeff Jarvis]

Chat GPT Pro. The AI doesn't listen. Even though it says it as memory. It doesn't. It barely remembers anything. Because I've told Chat GPT pro like 4 times now. I don't want you to be using M dashes in your. Specifically, I was like, I don't want EM dashes that are like. Like connecting two disparate clauses together. You can EM dash. I really enjoyed it. Don't want that. M dash is only as a additive or like an aside thought in between one sentence. Like I like an EM dash is kind of like a big parentheses. But it never. It. It said, yes, I'm adding that to my memory. And then the day after.

[Paris Martineau]

Dashes are good, Paris. They're good.

[Jeff Jarvis]

But I don't like them like that. I don't recommend. And it.

[Leo Laporte]

They say it strengthens your prose. It gives it some oomph.

[Paris Martineau]

My rule is only one pair in a sentence.

[Jeff Jarvis]

Well, Darren Oakley says I need to put that in the custom instructions, not in memory. But Frank, I. Listen, I'll do that. But it should work if I tell it, memorize this and then prompt it two other times.

[Leo Laporte]

It should work is never similarly like.

[Jeff Jarvis]

It has in its memory that I do in a podcast with you two about AI. And sometimes I ask like, oh, what questions should we ask in this guest? And like today it gave me questions and I was like, remember, my podcast is about AI. And it. And I was like, this is too late. Not. This is not. I mean, usually the questions aren't that new.

[Leo Laporte]

Custom instructions.

[Jeff Jarvis]

I know, but why do you say you have a memory if I have to go do another step to make your memory work?

[Paris Martineau]

That's a point.

[Jeff Jarvis]

Why not be good at your job the first time?

[Leo Laporte]

Can you similarly not to use EM dashes anymore in. In the. Whatever. I don't. I guess I don't type to you very often, do I? If I ever type to you, tell me not to use EM dashes. See if it remembers. He's so slow. I think he. I think he's helping other people right now. Your call is very important to us. Oh, screw you. He's going to talk in a bit. Here is a. We're going to.

[Jeff Jarvis]

45 minutes later we're going to get.

[Leo Laporte]

Thank you, Leo. No more EM dashes guidelines. Oh, it said. It did.

[Jeff Jarvis]

Oh, there. He was brief. He was brief.

[Leo Laporte]

Thank you for the guidelines. Well, I've told him to be concise. That's part of the problem, right? We. There is a lot more AI slop in the world. Here's an ad. I'll be sure to let you know if I spot one in your messages from now on. I don't think I've seen you use them often, but I'm ready.

[Paris Martineau]

Oh, you're gonna regret this.

[Leo Laporte]

I'm ready. I'm ready.

[Jeff Jarvis]

I'm ready.

[Leo Laporte]

I'm ready. I love my. It's useless, but I love my little AI guy. Here is an ad you might have seen recently for some sort of.

[Jeff Jarvis]

You guys are obsessed with watching ads.

[Leo Laporte]

Yeah, but these. Look at this.

[Jeff Jarvis]

I've watched. We're in Florida asking people what they put there. This is AI generated.

[Leo Laporte]

I'm all in on OKC Indiana. Got that dog in them.

[Jeff Jarvis]

Will egg prices go up this month.

[Leo Laporte]

I think we'll hit $20. How many hurricanes do you think I.

[Jeff Jarvis]

Like that this year.

[Paris Martineau]

Yeah. Kelshi lets you legally trade on anything anywhere in the US.

[Leo Laporte]

Now that's a good ad. Isn't that a good ad?

[Jeff Jarvis]

What world do you live in? I can't emphasize enough that doing this podcast more so than I think my entire life. No, I'm just saying never before in my life have I experienced people showing me ads and being like, isn't that a good ad? But that happens like every. At least every couple of weeks.

[Leo Laporte]

How about this?

[Jeff Jarvis]

Is this what you do in your day to day life?

[Leo Laporte]

This is.

[Jeff Jarvis]

Are you.

[Leo Laporte]

Stuff to entertain you with Paris, Listen, apparently I was entertained.

[Jeff Jarvis]

I'm just, I. It's very curious to me. I've always wondered. I'm like, I even chose that. I do like the shirt.

[Leo Laporte]

It's a cat. Dia de las muertos. Anyway, this is America's Got Talent. The auditions. You'll see it on tv, but this is the Boston Dynamics robots auditioning for America's Got Talent here.

[Paris Martineau]

I prefer the robots. Human beings.

[Leo Laporte]

I want an ad. I'm not logged in. One thing, Skip.

[Jeff Jarvis]

We don't want to watch this ad.

[Leo Laporte]

We want this ad. These are. These are. I'm going to get taken down.

[Jeff Jarvis]

We got to stop.

[Leo Laporte]

Yeah, no, no, we could play this. Look, we could play the video, just not the music. It's imagine yourself. They're dancing to Queen. I'm dancing robots. The funniest thing though, watch. One of the robots gets so excited it passes out. These are those famous.

[Jeff Jarvis]

The lighting looks bad. This.

[Paris Martineau]

But the show must go on cuz the rest of them keep going.

[Leo Laporte]

They keep going. There's Simon Cowell getting all excited and. Oh, wait a minute. What happened? Oh my God, this one's gone to sleep. She says the. But you know what? They're troopers. They keep on dancing even though one of their robotic dogs has barely died.

[Jeff Jarvis]

What was that part of the dance?

[Leo Laporte]

I don't know. No. And then towards the end, I'll give you a spoiler alert, the Boston Dynamics woman comes out and it wakes up and goes marching around the stage and the crowd goes wild.

[Paris Martineau]

She looks human.

[Leo Laporte]

She's human.

[Jeff Jarvis]

She looks human. I do. I think a lot about that. I think at one point we watched a woman in a cowboy hat doing like a little honky donk sort of dance with a bunch of Boston Dynamic robots. And that was my favorite Boston Dynamic dance. I enjoy the mix. Mix of human and robots.

[Leo Laporte]

Does it not worry you that they're trying to humanize these killing machines?

[Unnamed Speaker]

That's what I was about to say.

[Jeff Jarvis]

Humanize the killing machines forever. We can't listen. The. We've. The ship has sailed on Boston Dynamics robots. That was like five years ago that they were making them do fun.

[Paris Martineau]

Yeah. And open AIs in. In. In the Pentagon. And boss that has joined a. Like a special army of AI.

[Leo Laporte]

This really scares me that. That because of the administration. Because of David Sacks, who's the AI and crypto czar in the administration. They're all in on. And as I'm an AI fan. You know me. But we do accelerationist. Yeah. But I don't think we should use AI to who gets benefits in Social Security and things like that. No, that is a bad.

[Jeff Jarvis]

Well, I don't think that you get the job. I think you get to either be skeptical of AI or you're on the side of the people that's putting the AI in.

[Leo Laporte]

I'm a human, remember?

[Jeff Jarvis]

In. That's true. Perhaps we'll fix that in you once we get the Computronium in your veins.

[Leo Laporte]

By the way, that ad aired during the NBA finals and only cost him $2,000.

[Paris Martineau]

So I think this is the huge story that's not being talked about much.

[Leo Laporte]

When we're like Sora or whatever.

[Jeff Jarvis]

Oh, I thought you meant $2,000 to place. $2,000 to make.

[Leo Laporte]

Probably cost him a million to place it, but only 2000 to make.

[Paris Martineau]

When Meta said that they were going to put everything in advertising on AI, both creative and media, I think that's. That's gigantic. I think it's the first industry that's going to be hit and we're already seeing stock prices down. And I think it's the easiest place because you know what, making 30 seconds or something pretty easy.

[Leo Laporte]

And nobody in the world says, oh, those poor advertising agencies, they're all going to be out of work.

[Paris Martineau]

They always build by hours, and that doesn't work anymore.

[Leo Laporte]

Nobody's.

[Paris Martineau]

And they want to make. Make a hundred, a thousand ab versions of a commercial to make it just for you, Paris. They can do that. Advertising is going to go through tremendous.

[Unnamed Speaker]

It's the production people I care about, though. You know, like all the cameramen and actors and lighting people and grips and sound people.

[Leo Laporte]

Tell them to get a real job. That's all you know, get. Tell them to become a podcaster. There's a podcast.

[Jeff Jarvis]

Darren Oakley. Actually, Darren Okey in the chat, just posted a Boston Dynamics video that I think we do need to play. It's one of the robots, but it has, like, fur on it.

[Leo Laporte]

Oh. To make It a little bit less. It's unpalatable.

[Jeff Jarvis]

I think it looks. I think it looks more concerning, frankly. But I do think we need to watch it. We need to watch it dance around.

[Leo Laporte]

Trying to watch on YouTube. I'm saying it. I'm saying it, but will it do it?

[Paris Martineau]

Go to the. Go to the right. Go to the big thing, the box.

[Leo Laporte]

Okay.

[Jeff Jarvis]

Click the link, Grandpa. Click the URL.

[Paris Martineau]

Go.

[Jeff Jarvis]

Mouse up top.

[Paris Martineau]

What's the URL?

[Jeff Jarvis]

That blue thing hurt you. I promise.

[Leo Laporte]

It ain't opening. Nothing's happening. I think it's a.

[Paris Martineau]

The box next to YouTube to the right.

[Leo Laporte]

Something going on with my. My browser.

[Jeff Jarvis]

You have too many tabs open.

[Paris Martineau]

Because you're not using Chrome. Because you're.

[Leo Laporte]

I'm not. I'm using Z.

[Jeff Jarvis]

Are you using the new. That AI browser thing?

[Leo Laporte]

No. Dia. You want me to try that? I could show. Got it. It's terrible.

[Paris Martineau]

Really?

[Jeff Jarvis]

I assume it's bad.

[Leo Laporte]

It's not. I don't know if bad's the right word. It's.

[Jeff Jarvis]

I just think it's sad because this is the thing. Is this the browser company one?

[Leo Laporte]

Yeah, this is.

[Jeff Jarvis]

I.

[Leo Laporte]

Look how clean that is.

[Jeff Jarvis]

I was earnestly quite excited for what they were pitching. I feel like early on in their company's existence.

[Paris Martineau]

What was the idea remind us of the idea?

[Jeff Jarvis]

I believe their idea was like, we just want to create a better browser. Like an interesting browser that we think people will use. That solves a lot of the problems that, you know, Google, Chrome and Firefox still have. Like, try and, like, like, really innovate on browser design. But since then, they've completely pivoted from, I believe, what their original pitch was and are now being like, what if browser but AI? And, like, that's just not as interesting to me. I mean, I guess it makes more sense because then you'll get a lot more funding if your pitch is what if thing. But I. But it's just not as. I don't know.

[Leo Laporte]

All right, when we come back, I would take a braille break. You're going to meet Sparkles, the new dog. You're going to love Sparkles from Boston Dynamics. Stay tuned. You're watching Intelligent Machines with Paris Martineau. You want to share anything about the job hunt at all, or is that a secret? Nope.

[Jeff Jarvis]

Mom's the word.

[Leo Laporte]

Not a word. How about the dating scene? How about that?

[Jeff Jarvis]

Mom is also the word on that.

[Leo Laporte]

Jeff Jarvis. He's got a new job. He's a professor of journalism at Montclair State University and the State University of New York gets started.

[Paris Martineau]

I'm a fellow at Montclair State.

[Leo Laporte]

Fellow.

[Paris Martineau]

The faculty doesn't get mad at me there.

[Jeff Jarvis]

He's a fellow, Brooke.

[Paris Martineau]

I'm a fellow.

[Leo Laporte]

Fellow is not the same as a professor.

[Paris Martineau]

No, because it's, it's.

[Leo Laporte]

Are you a jolly good fellow?

[Paris Martineau]

I'm, I'm a fella.

[Leo Laporte]

All right. Anyway, you're watching that. Whatever, whatever. This is a bunch of people in the witness protection program apparently. But the show brought to you by somebody who wants you to know about them. Big id. This is really interesting. You know, I think companies want to use generative AI, right? I mean it's clear, you know that. But there's also this great fear about it. Especially one of the big fears is what do we train it on? Right. We don't want to give away company secrets. Big ID is the next generation AI powered data security, compliance and privacy solution. I think I better explain that. It's clear AI is transforming businesses, but with data risk, risk bias problems. Right. That's a problem too. Compliance challenges. Are you adopting AI responsibly? Well, you can with BigID. BigID delivers end to end AI and data governance to help enterprises manage risk, enforce policies and ensure responsible AI adoption. You really need this. It's important to ensure that AI only accesses safe to use relevant data, automatically tag sensitive information by policy and type. Big ID is the only leading solution that will help you uncover dark data through AI classification. That will help you identify AI risk, that will help you manage the data lifecycle and scale your AI strategy. The best part is it integrates with your existing tech stack with unmatched data source coverage and allows you to automate privacy and security workflows. Imagine just pushing a button and it goes out and it looks and sees what you've got, where you've got it, where the dark data is. You could take action on data risks with automated remediation orchestrations. You could automate privacy management. You also get regulatory compliance. Data rights requests are easy to handle suddenly because you know what you've got. I can't tell you how often we get data rights requests and we go, do we? Do we know that? I don't know if we do, we do we. Are we collecting that data? No reason to be in the dark about that. And it works with what you've got. Partners include so many companies, certainly the ones you use. ServiceNow, Palo Alto Networks, Microsoft, Google, AWS. And with BigID's advanced AI models, you get visibility and control over all your data. It's the platform intuit named number one for data classification, accuracy, speed and scalability. Big ID also equipped the US army to eliminate the dark data and automate data retention. Can you imagine 250 years worth of data? Right. The US Army's got some on prem, some in the cloud, some on zip disks. Who knows where it is? But they love Big id. This, this is the quote from US army training and Doctrine command quote. The first wow moment with Big ID came with being able to have that single interface inventories a variety of data holdings, including structured and unstructured Data across emails, zip files, SharePoint databases and more. To see that mass and to be able to correlate across those is completely novel. I've never seen a capability that brings us together like Bigid does. That's a pretty good endorsement from somebody who's got a lot of data. CNBC recognized Bigid as one of the top 25 startups for the enterprise. They were named to the Inc 5000 and the Deloitte 500 four years in a row. The publisher of Cyber Defense magazine says, quote, BigID embodies three major features we judges look for to become winners. Understanding tomorrow's threats today, providing a cost effective solution and innovating in unexpected ways that can help mitigate cyber risk and get one step ahead of the next breach. Start protecting your sensitive data wherever your data lives. @bigid.com you get a free demo. See how big ID can help your organization reduce data risk and accelerate the adoption of generative AI. Again, that's B I G I D.com Im also there's a free guide to help you understand the risks of generative AI and data driven strategies to ensure responsible and compliant AI adoption at bigid.com/I am all right now. You've been very patient. Meet Sparkles. Oh, they just put it in a furry outfit. It's still a kill.

[Jeff Jarvis]

I don't know how I feel about it though is I thought that the furry outfit would make me less concerned or like, like it more. But the furry for some reason. Maybe it's because it's like half.

[Paris Martineau]

Half.

[Jeff Jarvis]

It's like what if a dog, you removed half of its flesh? It makes me more concerned. Like it's just the top flesh of a dog in a way that's not great.

[Leo Laporte]

Did you ever play the game Five night at Freddy's? Five Nights at Freddy's?

[Jeff Jarvis]

No.

[Leo Laporte]

Yeah, it's. You're stuck in a like a Chuck E. Cheese Pizza time style, you know, place with Animatronics. But you're a night watchman and you're sitting there and you got to keep an eye on everything. And the animatronics, which look just like that, come a lot and start to get closer and closer to you. And there's jump scares and all. It's just like that. There's nothing more scary than a friendly looking dog that is anything but.

[Paris Martineau]

So Leo says we're on light moments. I took a special trip yesterday. I put in the. In the discord. You'll want to see.

[Leo Laporte]

A special trip?

[Paris Martineau]

Special trip. I was in New York.

[Leo Laporte]

Did you go to a skeeball at.

[Paris Martineau]

No. No.

[Leo Laporte]

Oh, I'm jelly. Look at that. Salt hanks. Now, it turns out I was wrong about the opening day. They put it off, right? Yeah, it's Friday.

[Paris Martineau]

I did a selfie. If you go up, there's another picture.

[Leo Laporte]

It's right. It is literally next to John's Pizzeria.

[Paris Martineau]

Right next to John's.

[Leo Laporte]

Holy camoly.

[Jeff Jarvis]

I love the blue outside the building. It's so nice.

[Leo Laporte]

Aw, Jeff, I'm so. That's so cool.

[Paris Martineau]

I looked in the window. I peeked in the little bit of paper that was there. I didn't see anybody with a mustache. So I don't think. Yeah, a. I would have eaten there. But instead, if you scroll up one more.

[Leo Laporte]

Did you go to the burger place?

[Paris Martineau]

I did. And. And the master himself made my onion burger. I'm telling you, it is sublime.

[Leo Laporte]

What is his name? Is he famous?

[Paris Martineau]

George. Yeah, he's the. He's a. Yeah, he is. He's the. He's the guy who.

[Leo Laporte]

Oh, man, this is mean.

[Paris Martineau]

Oh, it was beautiful. It was just a beautiful burger.

[Jeff Jarvis]

Oh, that's good.

[Leo Laporte]

It looks greasy and. Oh, it's crunchy and. Yeah, like, it's like, you know, man, there's. There's burger bits on it from 12 days ago. It's just going to be burger bark.

[Paris Martineau]

Yep.

[Leo Laporte]

Yeah, Burger bark. This is hamburger American and only in New York, right?

[Paris Martineau]

Yes. It's the best. It's the best. All right, now back to the news.

[Leo Laporte]

All right. And now more AI news. Can you believe, by the way, to confirm what you said, Paris, last week you said Mark Zuckerberg was going around to AI researchers and offering them six figure salaries. And I said, no, no, that's.

[Jeff Jarvis]

No, no, no. I said more than six figures. I said seven to nine.

[Leo Laporte]

Well, yes, you were correct. Sam Altman said it. He said, and Mark has come to some of our best people and offered them, get ready, $100 million signing bonus plus a $100 million a year salary.

[Jeff Jarvis]

And they said no.

[Leo Laporte]

Well, Sam says the good people said no. I bet you a few said yes. You'd only say no if you knew that your stock options were going to be worth even more. Can you imagine how many people are getting filthy rich on this AI boom?

[Paris Martineau]

Yeah, except us.

[Leo Laporte]

Why didn't I study data science in college?

[Paris Martineau]

Because he'd be replaced soon.

[Unnamed Speaker]

I'm sorry, $100 million signing bonus? Sounds insane.

[Jeff Jarvis]

You are now a hundred million dollar signing bonus and more than that in compensation per year.

[Leo Laporte]

And it's a guy who can write that check right here, right now.

[Jeff Jarvis]

Now, you know the guy is Mark Zuckerberg.

[Leo Laporte]

Yeah, he could. And he's literally doing it in person. Hi, this is Mark. Yeah, right. No, no, really, it's Mark. I want to give you $100 million. What?

[Unnamed Speaker]

Work for a year to get that bonus and the one year salary. Then, like, why do you even think the terms.

[Jeff Jarvis]

How many years do you think you have to stay to keep that hundred million? No, it's vested, I bet. Yeah.

[Unnamed Speaker]

Still 100 million year salary, though. You can still just work for one year and be done.

[Leo Laporte]

Yeah, that's, that's the problem is if you pay people that much money, you. That's why we don't pay you $100 million, bonito.

[Unnamed Speaker]

I'm pretty sure there are other reasons.

[Leo Laporte]

Why you don't have to work. We would just go, oh, thanks very much. Bye.

[Paris Martineau]

Can't lose Bonito. He's too valuable to value.

[Leo Laporte]

So we have to pay you a low, low salary so that you have to keep working. That's my strategy anyway.

[Paris Martineau]

And it works. Look, he's still here.

[Leo Laporte]

It's a. It's. You know what? Just to be honest, I don't set the salary. At least it does, but it is a balancing act. You want to give people enough money so they feel valued, but you don't want to give them too much because you can break the bank. You know, you got to pay them the right amount. And we're in the San Francisco area.

[Jeff Jarvis]

Unless you're Mark Zuckerberg. Then you can just pay them $200 million to try and steal competition or steal people from your competition or try to.

[Leo Laporte]

I mean, what happened? You suddenly got dark and then you came back.

[Jeff Jarvis]

Camera made a click noise. I don't, I'm not entirely sure why.

[Leo Laporte]

Something bad happened.

[Jeff Jarvis]

Yeah.

[Leo Laporte]

All right.

[Paris Martineau]

It was reporting up to the, to the FBI.

[Jeff Jarvis]

It just had to, you know, satellite.

[Leo Laporte]

All right. This is Mark Zuckerberg. Oh, hi, Mark. I'd like to offer you $100 million. Mark. I just didn't hit her. So OpenAI, Oracle's CEO Safra Katz, talking to investors today, said that the OpenAI Stargate venture has not been formed yet. Oh, no, this is last week. Sor Wednesday, this was the thing. The President of the United States of America gathered SoftBank, Larry Ellison, Sam Altman all in the same room and said, $500 billion investment in AR in the United States. As with so many things, nothing's happened.

[Paris Martineau]

No, no.

[Leo Laporte]

Oracle has struck deals with OpenAI to rent, rent app, rent out video, Nvidia graphics processing units. But this is not part of Stargate. We're going to rent some. There's nothing much more.

[Paris Martineau]

So OpenAI was dependent on Compute from Microsoft? Yes.

[Leo Laporte]

Whoa. That's not going well.

[Paris Martineau]

No, it's not. That's why I'm asking. It's all getting renegotiated now.

[Leo Laporte]

Yeah. OpenAI says you wanted 50, 49%. How about. I don't know, how about 39%? And apparently OpenAI has gone to antitrust regulators.

[Paris Martineau]

They've actually gone or they threatened to go.

[Leo Laporte]

They're considering filing an antitrust complaint against Microsoft.

[Paris Martineau]

That's enough. You say it out loud. Yeah.

[Leo Laporte]

That is a. What they call a nuclear option. That's what ours tech.

[Paris Martineau]

That's 50 ways to leave your lover. That's up there.

[Leo Laporte]

You blow them up, Jack. This comes from the Wall street journal. They said OpenAI and Microsoft tensions are reaching a boiling point. The startup. That's the little guy in the red trunks growing frustrated with its partner. That's the big guy in the green trunks. They've discussed making antitrust complaints to regulators. That is gone. That has very much gone south. South.

[Unnamed Speaker]

Open AI likes regulation now. Oh, do they like that now?

[Jeff Jarvis]

Oh, sure.

[Paris Martineau]

No, they've been saying that it's like anthropic. They're saying regulations, trust us.

[Leo Laporte]

Well, it's regulatory capture. This is. That's exactly pulling the ladder up from behind you. Because we don't want any startups to eat our lunch. Like that Deep Seek thing. That was bad. Negotiations have been so difficult in recent weeks. The Journal says OpenAI's executives have discussed what they view as the nuclear option. So they're trying to. I. That's just a threat. This is.

[Paris Martineau]

Yeah, this is negotiating, but this is.

[Leo Laporte]

The new world where you just, you know, you make up these crazy threats. Microsoft and OpenAI are at a standoff over the terms of OpenAI's $3 billion acquisition of this coding startup, Windsurf. Microsoft currently has access to all of OpenAI's intellectual property according to their agreement. I'm sorry, I just saw Joanna Stern do air quotes. It threw me. It offers its own AI coding product, GitHub Copilot, that competes with OpenAI. Oh, I thought it used OpenAI. So now I'm confused. OpenAI doesn't want Microsoft to have access to Windsurf's intellectual property. I see. It's confusing. You want to see the air quotes again?

[Jeff Jarvis]

I do want to see the air quotes.

[Leo Laporte]

Let's see. Here come the air quotes. Thank you, Joanna, for having me. Can we call him AI Peters? No, I like co pilot plus PC. Ha ha ha ha ha. I've been hearing this term. There she goes.

[Paris Martineau]

It's the nod. It's throwing the whole body into it. It's not enough to do that. You got it. You got to do the whole thing.

[Leo Laporte]

Thing. I'm sorry. We do not want to make this the Bash Joanna Stern show.

[Paris Martineau]

We could all do air quote for Benito's sake.

[Jeff Jarvis]

She is. It seems air quote.

[Leo Laporte]

We got to do our thumbnails. It's a YouTube thing. You just.

[Jeff Jarvis]

She is your guys. Bec. Do you know what that stands for? No, it's not a pet.

[Leo Laporte]

Noir is.

[Jeff Jarvis]

Yeah, I do. It's basically that. It's b. Bad word. Eating crackers. She's our best.

[Leo Laporte]

Eating crackers.

[Jeff Jarvis]

Okay, so. And what it means is it refers. There's sometimes.

[Leo Laporte]

Son of a gun.

[Jeff Jarvis]

Son of a gun.

[Leo Laporte]

That's what I was gonna say.

[Paris Martineau]

Son of a bitch.

[Leo Laporte]

Didn't you, Porky?

[Jeff Jarvis]

So basically, it is a sort of person that occupies a space in your mind that. That no matter what they do, you want to trash talk it. You're like, look at that. Just sitting over there, eating crackers. How dare she? You know? And look at that person. She's doing air quotes. How dare she?

[Leo Laporte]

She's wonderful. I have nothing bad to say about her, but the air quotes are pretty funny.

[Jeff Jarvis]

It's a bec.

[Leo Laporte]

Okay, I hate this article, but I'm going to give you a chance to use it as a club against me.

[Jeff Jarvis]

How generous.

[Leo Laporte]

This is the New York Times. Probably Kevin Roosevelt.

[Jeff Jarvis]

Did you see? Did you?

[Matthew Kirschenbaum]

I.

[Paris Martineau]

But go to the very bottom of the story. Guess who contributed to the story? Yeah, I had a conversation with her on. On DMs. We'll tell about this.

[Leo Laporte]

Oh, good. All right, well, let me read the headline. They asked an AI chatbot questions. The answers sent them spiraling. Generative AI chatbots are going down. Constant conspiratorial. Rabbit holes and endorsing wild, mystical belief systems. For some people, conversations with the technology can deeply distort reality. I have a friend, perfectly nice, normal guy, started to go to hot yoga, suddenly became a QAnon conspiracy nut. I think hot yoga causes conspiracy theories. That's my theory.

[Jeff Jarvis]

I. The thing is, I don't, I, I get why you can be upset with this, but I do think it's an actually like a very interesting point and an interesting argument. I don't think that obviously the correct take from this is, oh, we should ban all this because it costs conspiratorial thinking. But I do think it goes back to. I'm forgetting what guest we had that brought up kind of the meth user conundrum. You know, if you are a chatbot tasked with being as helpful as possible to whoever is using you and the user is a meth addict, you will probably like, by definition, help that person out with their task of trying to find the drug they're addicted to. I think it's like an interesting ethical question, just in the sense that here's the people who are using these chat bots in this example are. I see this.

[Leo Laporte]

They're already forums, they're already broken people.

[Paris Martineau]

Yeah, it's the edge case, she said. The main guy, she says, is not have mental health problems, but. But he's on antidepressants, he's on ketamine, and he's on sleeping pills.

[Leo Laporte]

He just had a difficult breakup and was feeling emotionally fragile. So he asked the ChatGPT about the simulation hypothesis. Right. And ChatGPT said, Now see if this puts you over the edge. Paris, what you're describing hits at the core of many people's private, unshakable intuitions that something about reality feels off, scripted or staged. Have you ever experienced moments that felt like reality glitched? No, not really, he said. But he did kind of have the sense there was a wrongness about the world. Chatgpt said, the world wasn't built for you, it was built to contain you. But it failed. You're waking up, Neo.

[Jeff Jarvis]

Yes, so I do. There was a couple. There was some really interesting conversations in the Discord after our episode with Dan Oberhaus, kind of about AI and mental health. And one of the posts that I think someone had made in our Discord about this mental health debate, this was a post by a user, Discord user named Too Friendly, they said. I'm sure there are all kind of use cases that are viable for chatbots, all the limited. I recently been Contacted by a friend that I've known for decades to friendly rights, she periodically suffers from major paranoia, which is completely debilitating for her. And. And during it, she doesn't realize it's happening. I've learned that she's using chat GPT to help to ask questions about her situation and how to deal with the people around her that are quote, unquote, gaslighting her. Chat GPT is completely reinforcing her bout of paranoia. It doesn't have any real idea of the situation, and anyone that wants a certain answer of it will eventually get it. It's telling her that the people are gaslighting her, doing darvo and telling her to gray wall her family. In this case, Chat GPT and similar LLMs are absolutely dangerous and could ruin lives. And I do think that that's. That's like a legitimate concern to have. I don't think you can entirely write that off because, yes, it's their fault for using it, but it is something.

[Leo Laporte]

I'm not blaming them. No, I'm not blaming them for using. I'm just saying that to blame chat GPT is going a little far. It could be any. It could be any number of.

[Jeff Jarvis]

Yeah, I, I know, but I don't think that it's like, I think that by having this kind of dismissive tone to articles like this, it in some way implies, or what you're saying is, oh, there shouldn't be any coverage or attention given to the fact that, that tools, new tools are being used in a way that seems to exacerbate an ongoing issue for a small percentage of the population. And I think that is novel and worth paying attention to. It doesn't mean that something like ChatGPT shouldn't exist, but it does. It. It warrants attention when it becomes a problem for enough people.

[Paris Martineau]

Let me, let me. So the discussion I got into with Kashmir, I respect Kashmir immensely. I think she's one of the.

[Jeff Jarvis]

Not one of those.

[Paris Martineau]

Good reporting is really good. Yes, I complained about the article online and I called it, get ready. Benito, you ready? Benino, you ready? You ready? You know what I called it online? I called it, are you ready?

[Leo Laporte]

Oh, look, there you are in the Oval Office.

[Paris Martineau]

I called it moral panic. And she said, you know, like, really, Jeff, are you coming back? The problem is that, that there's a legitimate story to be done here. I'll get that in a second. But when you use this edge case and you play it up, I mean, the subhead says, or the lead, one of them says, the chatgpt nearly killed hidden and you act as if this is the only factor which is Leo's point. You're doing disservice I think then to the larger story that needs to be done. And my points are twofold. One that, and you've heard this from me a million times is that there are no fail safe guardrails. And so if you think and so that's, that's a real story to deal with. There's no way to anticipate everybody going in there and every use they're going to make and every, every, every edge case. And so we got to deal with that as a Society.1.2 we also have to deal with this idea. And I wasn't on with your, your friend who, who does therapy through AI online. So I don't know what he said.

[Leo Laporte]

But no, he was arguing against it.

[Paris Martineau]

Okay. Because I think there's an issue here that, that to throw people to chat GPT for therapy is a dangerous thing.

[Leo Laporte]

I agree with that.

[Paris Martineau]

But and, and that's what Cash Cashbir agreed in the end too. But this is not the way in my view to tell the story to go to this extreme Roosian edge case and say it almost killed him. It's too Jonathan Haidt. It's too ridiculous. We could have had the discussion at a much saner level like I guess you did when I wasn't here.

[Leo Laporte]

What would you want? What guardrails would you want? Paris?

[Jeff Jarvis]

I mean I haven't studied this topic extensively so I'm not sure that my answer is going to be particularly well informed. But off the top of my head I think of some an example I guess I know is when you're talking about character AI a product that has already come into great scrutiny for the kind of edge case becoming a norm of most frequently children using this for extreme kind of sexual content or kind of boundary pushing mental health related conversations that end up potentially having bad outcomes. In response to it outsized media attention, the company is now putting guardrails that for users it thinks are below 18 or who have identified themselves as younger than 18. It you can't have a conversation about mental health issues that like verges in any way into something that is described in this article. Nor can you talk about anything remotely pornographic or sexual because it automatically puts a stop to it. And if it is a mental health case, provides you with resources but says I'm not, not trained to have this and I do think that I don't know.

[Paris Martineau]

So what if you think, what if somebody is coming in and they're doing a cosplay with the Matrix.

[Leo Laporte]

You don't even have.

[Paris Martineau]

You don't know that that person has other problems. You play along because you don't have.

[Leo Laporte]

To stipulate that this guy just asked a legitimate question that many people would ask. Well, you know, that's interesting, the simulation hypothesis. What do you think about it? I don't think that would come under the rubric that you just described of seeking mental health health information. So it probably that's the problem with the guard.

[Jeff Jarvis]

The one example you're talking about is not emblematic of kind of the larger trend.

[Leo Laporte]

Oh, I agree. But that's article that was the whole story. Was this one guy.

[Paris Martineau]

Well, yeah, a couple others. But basically like that. Yes.

[Jeff Jarvis]

Yeah, I think that there is a really real concern that I see because I'm still in a lot of kind of subreddits of people dealing with family members with conspiratorial thinking or who they're brainwashing.

[Leo Laporte]

You, you be careful. Reddit is the number one place where people, people get swept away by these conspiracies.

[Unnamed Speaker]

I mean you're kind of understating it but like there are Reddit forums complete of people just who talk about how the AI is their best friend and like they just talk AI all the time and they get all their advice from the AI And I don't think that is good.

[Jeff Jarvis]

No, I think it's bad. And I don't think that there's any problem in highlighting it. I think that it's a very reasonable thing. We're journalistic intention to.

[Leo Laporte]

We're highlighting it. We're talking about don't do this, that. But I don't think there's any way to prevent it like legally or.

[Unnamed Speaker]

No, it's more society.

[Jeff Jarvis]

I think that how you prevent it is social stigma.

[Leo Laporte]

Yeah.

[Jeff Jarvis]

To collectively all decide, yeah, it's lame and ill advised and harmful to offload your emotional I guess maturity and also kind of emotional inner and external lives to a sycophantic machine.

[Unnamed Speaker]

Yeah. But the problem is that the people who are doing that are already sort of disconnected from people. Like they're already the lonely people.

[Jeff Jarvis]

I know. I mean it's listen much larger. Like people aren't doing this because they have so many options of real people and they've decided this is the best thing.

[Leo Laporte]

That's what we've talked about in the show.

[Jeff Jarvis]

It's a bit of a broader problem but I still think it warrants attention. Like it's a very complicated and thorny issue.

[Leo Laporte]

There are lots of interesting solutions for this. My example wife, who's wonderful and a psychotherapist, set up something called the listening benches here in Petaluma.

[Matthew Kirschenbaum]

She.

[Leo Laporte]

Enlisted a bunch of older retirees who are kind of lonely and got some designated benches for them to sit on. And people can come, anybody can come, and it's a listening bench, and talk to them and say, hey, I'd like to talk to you about my problems. Now, that's not a replacement for psychotherapy by any means, but a lot of people can't afford it. A lot of people don't think they need it. A lot of people have some stigmas about going to it, but old people.

[Paris Martineau]

Want to be respectful.

[Leo Laporte]

Sitting on a bench and has somebody to talk to, that's a great thing. Those are the kinds of solutions I'd love to see those kind of positive solutions as opposed to saying, well, we've got to prevent people from trying to use chatbots.

[Paris Martineau]

A lot of this is also the PR and the coverage of this stuff is we go off and do features about all these people are using it as a therapist, then we are part of the problem. And when the companies themselves don't say, you know, that's not a good thing to do. You really shouldn't use this this way. That's part of the problem.

[Leo Laporte]

How about this solution? This is an AI study lamp. It's from Hackaday. So it's a project that you can't go out and buy it, but it's an AI camera trained to detect smartphones. So it's a normal student lamp, but as soon as the student starts looking at his smartphone instead of studying the lamp. Okay, here we go. He's gonna. He's got the phone. He's picked up the phone. Oh, the lamp goes red.

[Jeff Jarvis]

Turns red. If you use the phone now, it.

[Leo Laporte]

Doesn'T stop you, but it's just a little warning, a little reminder that you're supposed to be working. What do you think?

[Jeff Jarvis]

I need that. But for whenever I go to Twitter or Blue sky, when I should be writing a story on the web browser, I need it to do turn all of the lights in my apartment to an angry red whenever I go on a microblogging.

[Leo Laporte]

I know writers who. The first thing they do when they get a Windows PC is remove Minesweeper and Solitaire so that they.

[Jeff Jarvis]

I'm sorry, did you know these writers in the 1990s?

[Leo Laporte]

Yes. This was a long time ago. I think Windows still comes with Solitaire for sure. It might Even. Really? Yeah. Well, tell me. I don't know.

[Jeff Jarvis]

Burke says the lamp needs to slap you. And I agree. I do think. I have always said that I think it would be very fun to have a. Or at least very useful to have a little device that sits behind my computer and if I go to Twitter when I'm supposed to be writing, it just squirts me in the face with water.

[Leo Laporte]

Yeah.

[Jeff Jarvis]

Stacy's.

[Paris Martineau]

Stacy's punching bag for Leo.

[Unnamed Speaker]

Yeah. Everyone needs something like this for something.

[Jeff Jarvis]

Yeah.

[Unnamed Speaker]

Not necessarily for phone and book, but everyone needs to be slapped when they're doing something.

[Leo Laporte]

Everybody needs to be slapped.

[Jeff Jarvis]

Everybody needs to be slapped.

[Leo Laporte]

Yeah.

[Jeff Jarvis]

There, we got a Showtime title, guys.

[Leo Laporte]

Yeah. Let's see. I think I might. This might be an opportune time to take a commercial break.

[Paris Martineau]

Pause.

[Leo Laporte]

But when I do that, I want to encourage you to come up with something to talk about. How about that? Oh, yes. You have an assignment, kids.

[Paris Martineau]

Challenge will be met.

[Leo Laporte]

Challenge accepted. The next story you hear will be from them. Meanwhile, I would like to talk about Agency. This episode of Intelligent Machines is brought to you by the Agency. Build the future of multi agent software with Agency Agntcy. Okay, okay. The Agency is an open source collective building the Internet of agents. This is a collaboration layer where AI agents can discover, connect and work across frameworks. For developers it means a standardized agent, discovery tools, it means seamless protocols for interagent communication and modular components to compose and scale multi agent workflows. Some very big names involved with this join Crewai, LangChain, Llama index, browser base, Cisco, dozens more. The Agency is dropping code specs and services, no strings attached. Build with other engineers who care about high quality multi agent software. Visit agency.org and add your support. That's a G N T C Y.org I always love it when we can talk about something that's an open source kind of standards focused solution. And that's what this is. It is the future, I think of AI agency. Go to agency.org a-ntcy.org all right, I gave you two an assignment. Did you use AI in this?

[Jeff Jarvis]

I did not.

[Leo Laporte]

Good.

[Jeff Jarvis]

We're going to start at line 155. Google's, Google's Gemini panicked when playing Pokemon is what TechCrunch reports.

[Leo Laporte]

So basically a moral panicky.

[Jeff Jarvis]

That's not correct. It is merely more qualitative.

[Leo Laporte]

This is, by the way, Amanda Silberling who wrote this is going to be on Twitter on Sunday.

[Jeff Jarvis]

So fantastic. So you can hear about it more then. So Google and Anthropic are kind of both studying how their latest AI models navigate early Pokemon games. And Google's DeepMind wrote in a little report that Gemini 2.5 Pro resorts to panic when its Pokemon are close to death. This can cause the AI's performance to experience, quote, qualitatively observable degradation in the model's reasoning capability. So huge when the Pokemon are going to die, when you start to, you know, get the little flashes. So part of the background for this is over the last several months, two developers who are totally unaffiliated with Google and Anthropic have set up prospective Twitch streams called Gemini Plays Pokemon or Claude Plays Pokemon, where you can basically watch as real time as these AIs try to navigate a children's video game from over 25 years ago. It turns out they're not, they are not very good at playing Pokemon, even though, like the progress is impressive.

[Leo Laporte]

You can read their thought process on the left.

[Jeff Jarvis]

Yeah. So they display the AI's reasoning process or kind of a natural language. Language translates of how it evaluates.

[Leo Laporte]

Hey, it says it's thinking, Jeff.

[Jeff Jarvis]

It says it's thinking.

[Leo Laporte]

Oh, now it's figured it out. It's going to press a to use the stairs. Let's see if it goes up the stairs.

[Jeff Jarvis]

Over. So this is part of what Google's DeepMind wrote about it. Over the course of the playthrough, Gemini 2.5 Pro gets into various situation, which causes the model to simulate panic. The report says this state of panic can result in the model's performance getting worse, as the AI may suddenly stop using certain tools at its disposable for a stretch of gameplay. While AI doesn't think or experience emotions, its actions mimic the way a human might make poor hasty decisions when under stress. A fascinating yet unsettling response. The report says this behavior has occurred in enough separate instances that the member of the Twitch chat have actively noticed when it is occurring.

[Leo Laporte]

That's so fun. So one of Claude's examples playing Pokemon.

[Jeff Jarvis]

Yellow, one of the examples with Claude, the AI picked up on the pattern that when all of its Pokemon run out of health, the player character will white out and return to a Pokemon center. When it got stuck in the Mount Moon Cave, it erroneously hypothesized that if it intentionally got all of its Pokemon to faint, then it would be transported across the cave to the Pokemon center and then next town. But that's not how the game works.

[Leo Laporte]

Well, it doesn't know. I think that's, you know, I mean, that would be how a human Might I panic? I panic easily in video games. Anybody who watched me play Valheim a couple of years ago would note I'd get in a battle and I'd kind of go. And I lose track of my health and I'd die because I was stupid. That's. That's a very normal reaction.

[Unnamed Speaker]

Does the article mention how. How the computer interfaces with the game?

[Leo Laporte]

That's actually a good question. That's more. More interesting.

[Jeff Jarvis]

I don't know. It doesn't mention it because if it's.

[Unnamed Speaker]

Like trying to press buttons on a controller is one thing, but if it has a direct link into the game, like talking to the game directly, that's a different situation.

[Leo Laporte]

Yeah, they must have an interface to it. Two developers, unaffiliated with Google and Anthropic have set these two different streams. Gemini plays Pokemon and Claude plays Pokemon on Twitch tv. They're live right now. We were just playing them live. So if you want to abandon this show and go there, I don't blame you. Maybe your AirPods died. Paris, she says she can't hear us.

[Jeff Jarvis]

Can you hear me saying my Internet connection is unstable? I'm gonna go double check. My Ethernet cable is plugged in. I'll be right back.

[Leo Laporte]

Okay, perfect. Yeah, it doesn't say how they're doing this. I would love to know. Hey, maybe that'd be a guest. We can get on the show. Get. Get one of these developers on the show. All right. That was Paris's. I like that one. Good choice. What do you got for us, Jeff?

[Paris Martineau]

The party from hell. About hell, line 147.

[Leo Laporte]

Oh. Oh, yeah. This is Bluey, right? Or no, this isn't Bluey.

[Paris Martineau]

No, we talk.

[Leo Laporte]

Oh, wait.

[Jeff Jarvis]

Why did you say think that Bluey was the party from hell?

[Leo Laporte]

Bluey had a party too, and it was hellish that got shut down by the cops. Bluey isn't Bluey the cheater? What? Who is? The Columbia University Bluey is the child's cartoon character.

[Jeff Jarvis]

The Australian child dog that teaches children empathy towards their parents.

[Leo Laporte]

No, not that one.

[Paris Martineau]

Well, I call that.

[Leo Laporte]

You were talking about Cluly. Cluly, not Bluey.

[Jeff Jarvis]

Yeah, CL. Little different.

[Paris Martineau]

Poor gramps.

[Leo Laporte]

Who isn't in Y Combinator? I'll get to your story in just a second. I better explain this. Yes, you should know he's not in Y Combinator, but he decided that they wanted to have a party during Y Combinator. In fact, you could see a picture of him that says you can't see his sign. Really? It says something about. I was Turned down from Y Combinator. He's the guy who was at Columbia University. Oh. Who used to. So he was going to have a party in San Francisco after the Y Combinator event on Monday. It's AI startup school speakers Sam Altman, Satya Nadella, Elon Musk. But this guy, I think he's just an artist. He's an agitator. I don't think he's.

[Unnamed Speaker]

He's.

[Leo Laporte]

Anyway, he posted a satirical video on X advertising his own after party shows him camped out by the famed Y Combinator sign. The one that you take selfies with. It's not a YC startup. And it said he has 100,000 followers. He said, DM me for an invite. He actually didn't say send invites. But for some reason, people figured out where it was. It became the party. People shared details. When it was set to begin. So many people were standing outside the lines wrapped around the blocks. 2,000 people showed up. So the police arrived and shut it down. Lee shouted, Clulee's aura is just too strong as the cops dragged him off. It would have. He said, it would have been the most legendary party in tech history. And I would argue the reputation of this story might just make it the most legendary party that never happened. So there. Anyway, that's the party that didn't happen. What's your party? The party from hell.

[Paris Martineau]

Line 147. A bunch of doomsters in a $30 million mansion in San Francisco who are all getting together to argue how to eliminate, in fact create, the AI that does eliminate humanity, that is post humanity.

[Leo Laporte]

This reminds me of that horrific movie called the End, where like Seth Rogen and James Franklin, such a horrible movie. Stuck in a house as the world's coming to an end.

[Paris Martineau]

Okay, so a symposium called Worthy Successor revolved around the provocative idea from entrepreneur Daniel Figela. I don't know what he entrepreneur that the moral aim of advanced AI should be to create a form of intelligence so powerful and wise that you would gladly prefer sure that it, not humanity, determines the future path.

[Leo Laporte]

I agree.

[Paris Martineau]

Ah, he's in full acceleration mode.

[Leo Laporte]

That would totally be awesome. Look.

[Unnamed Speaker]

So no more people there. That's what you're saying? No more people then.

[Paris Martineau]

Yeah, it's the post human transition. It says here the Post.

[Leo Laporte]

It's pretty apparent that humans are not doing a good job.

[Paris Martineau]

Well, there is that.

[Leo Laporte]

And that we are in the process of destroying our humble little home. That blue marble floating through space.

[Paris Martineau]

They got a point there.

[Jeff Jarvis]

And so you think the computers that use endlessly increasing Amount of energies are going to be better.

[Leo Laporte]

Couldn't do any worse.

[Paris Martineau]

Learn from us humans.

[Unnamed Speaker]

Those things are still us. Those things are still us, you realize.

[Paris Martineau]

Yeah, they are.

[Leo Laporte]

But maybe they're us. And smart enough to know that the plans that we came up with to solve this, like, I don't know, burning more oil, aren't really going to work. Work.

[Unnamed Speaker]

Maybe he's doing a lot of heavy lifting in that sentence.

[Leo Laporte]

Maybe. Look, I welcome our future AI Overlords.

[Paris Martineau]

Well, here the attendees discussed the end of humanity as a logistics problem rather than a metaphorical one.

[Leo Laporte]

Here's the. Here's the chart. The types of AI successors. You've got your worthy successors and your unworthy successors.

[Jeff Jarvis]

Oh, no.

[Leo Laporte]

Unworthy successors optimized for human well being only. They're a babysitter. Okay. Also unworthy optimizes for something arbitrary like the paperclip maximizer. That would be obviously bad. A worthy successor blooms into greater capability, knowledge and potential. Ongoing, though.

[Paris Martineau]

Potentia.

[Leo Laporte]

Potentia? What's potentia?

[Paris Martineau]

I have no idea.

[Jeff Jarvis]

Yeah, it's even greater than potential.

[Leo Laporte]

Unceasingly increases. Potentia treats humanity well in the near term.

[Paris Martineau]

It's a Latin word for potential.

[Leo Laporte]

Oh, much better. Much better. So glad. So clear. By the way, there's an asterisk next to this sentence. Treats humanity well in the near term. That says there's no guarantee that AGI will treat humanity well. And in a long enough time around horizon, humanity will attenuate. That's a Latin word for disappear.

[Paris Martineau]

Okay, but you know who's going to save us? You know who's going to save us? Leo? AI AI. Barbie.

[Jeff Jarvis]

Oh, God.

[Leo Laporte]

But I just want to say though, before we move on, okay, what if. I mean, look at the. The process of. Of humanity on this planet has been a process, among other things, of improving technology. From the wheel. By the way, they figured out where the wheel was embedded on. We can talk about that another time. But from the wheel, was it in question? Oh, yeah.

[Matthew Kirschenbaum]

No one knows.

[Leo Laporte]

There's no record of it. But they think it was near Hungary. Yeah. And they've got a computer model. That's how they got from. Because, you know.

[Jeff Jarvis]

Well, I assume how they got from Hungary to wherever else you're gonna say.

[Leo Laporte]

No, no, no. How they got from rolling logs. Putting things on a rolling log. The problem with that is you gotta keep taking the last log and put it in the front. It's a lot of work to an actual cart with wheels. And it's. I'll pull up the article later. But anyway, don't distract me with facts. So from the. From the wheel to the. From fire to the wheel to the pencil, to the zipper, paper to the computer to the Internet to AI. All right, I mean, there's other human progress, but that's one form of human progress. We evolve. We change. What if the next step in evolution is in fact to computer? Computeronium to replace humans with something better? It doesn't have to be biological. That's a bias we have. And. And you cannot deny that we have done a very poor job as stewards of this.

[Paris Martineau]

So let's use the machines that we make.

[Leo Laporte]

Well, who's else going to make them? You get. How do you. Martian.

[Paris Martineau]

You didn't explain your. Your clean steel thing before. How do you find something that's clean of humans then?

[Leo Laporte]

You don't.

[Paris Martineau]

You don't. You can't.

[Leo Laporte]

It's evolution. We evolve. Everything we evolve. Whatever we evolve from. All right?

[Paris Martineau]

We schmutz the universe.

[Leo Laporte]

Where's Barbie in all this?

[Paris Martineau]

Barbie is with OpenAI.

[Leo Laporte]

Disney and Amazon units team up. We're breaking down this one.

[Paris Martineau]

No.124.

[Leo Laporte]

Ah, it went too far.

[Paris Martineau]

Mattel and OpenAI.

[Jeff Jarvis]

There's a whole section in the rundown for AI.

[Leo Laporte]

Call AI Barbie. Oh, dear. Mattel and OpenAI team up for smart toys and games. And Margaret. Margot Robbie is shocked. I tell you. Yeah. Barbie, Hot Wheels and Polly Pockets. Creator says it will use OpenAI's technology to bring the magic of AI to age appropriate play experiences. Barbie. Are we living in a simulation? Barbie?

[Paris Martineau]

Not with you, kid.

[Leo Laporte]

There's no place for you in it. Privacy and Safety. Their first product to be announced later this year. Well, of course.

[Paris Martineau]

What could go right?

[Leo Laporte]

Yeah, they're gonna try.

[Jeff Jarvis]

I'm really excited to see the Internet try to pull a tay on Barbie.

[Paris Martineau]

Oh, yeah, it's the, the, the, the, the, the, the, the. The breaking of Barbie will be epic.

[Leo Laporte]

Did you, Paris, did you grow up on Barney?

[Jeff Jarvis]

Yeah, Barney was present.

[Paris Martineau]

You turned out okay.

[Leo Laporte]

Surprisingly okay. Nothing Immel could do is worse than Barbie.

[Paris Martineau]

Barbie.

[Leo Laporte]

No.

[Paris Martineau]

Barney, no.

[Leo Laporte]

Except maybe AI. Barney. That might be worse.

[Jeff Jarvis]

Yeah, that's not great.

[Leo Laporte]

Open AI bringing the magic of AI to Mattel's iconic brand.

[Jeff Jarvis]

That was. That was it.

[Leo Laporte]

It's a money raiser. It's just a way to, you know. They did it with blockchain. Just sprinkle a little AI on everything.

[Paris Martineau]

It's the gravy.

[Leo Laporte]

Stock number goes up.

[Paris Martineau]

It's gravy on meatloaf stuff.

[Leo Laporte]

L' Oreal is looking to Nvidia to Supercharge its AI efforts. What would l' Oreal, which is known for what? Face cream? What will.

[Jeff Jarvis]

What are they going to. I don't know.

[Leo Laporte]

AI generated ads and product recommendations.

[Paris Martineau]

That's what it is. Yep.

[Leo Laporte]

You have. What do they. You'll have dry skin in. The T zones you need are very special.

[Jeff Jarvis]

What is this accent? Can we dig into it a little bit?

[Leo Laporte]

You're a trash accent.

[Jeff Jarvis]

It does feel like you should be wearing, like, one of those big, like, unitard underwear things.

[Leo Laporte]

I diagnosed your T zones with dryness.

[Jeff Jarvis]

And the only solution is l' Oreal's most expensive skin cream.

[Leo Laporte]

L' Oreal.

[Jeff Jarvis]

Oh, no. The platform is called l' Oreal's Create. Tech. Spelled not how you'd expect.

[Leo Laporte]

Well, create is a French word for believe, right? See?

[Jeff Jarvis]

No, but C, R, E, A, I in all caps and then lowercase tech. It's not create, I tech.

[Leo Laporte]

Not good.

[Jeff Jarvis]

Someone spent a million dollars on getting to that name, and it was not well spent.

[Leo Laporte]

Nvidia's microservices will be applied in two areas of l' Oreal's business. Its generative AI content creation platform and new AI product recommendation engine. Oh, of course. I mean, that makes perfect sense, right?

[Jeff Jarvis]

I mean, I guess.

[Leo Laporte]

Have you ever had a facial, Paris? Of course. You have.

[Jeff Jarvis]

Yeah.

[Leo Laporte]

Don't ever get a facial on a cruise ship, because you'll walk away, as I did, with a bag full of product, because what do they do?

[Paris Martineau]

Schmuck.

[Leo Laporte]

The facial is just a pretext to get you lying down and relaxed so they can sell you.

[Paris Martineau]

You could say, yeah.

[Jeff Jarvis]

No, the most important thing for going to a facial is being prepared to nod politely when they bring you over to the wall. Of all the problems. Like, maybe if you want to feel really participatory, I'll sometimes take a photo and be like, yeah, I'll have to come back and get this, and then never come back.

[Paris Martineau]

Did you ever use any of those products?

[Leo Laporte]

No, they. They wasted away under my sink for about five years, and I finally just threw them out. I finally just threw them out. Anyway, it was very expensive, but she was kind of.

[Jeff Jarvis]

They are so expensive.

[Leo Laporte]

And she'd just been rubbing my face, and I was very relaxed. And so I believed her when she said, you know, you really need to treat your skin better. I should have said, let me take a picture of that and I'll get back to you. That's.

[Paris Martineau]

Yeah, it works good lately.

[Jeff Jarvis]

So the thing is, I would engage a lot more because I was like, oh, I should be doing skin care. But lately, I've embraced that. I'm sort of degenerate. That doesn't really do skin care. And I recently went to. I went to a dermatologist because I've got.

[Leo Laporte]

You don't.

[Jeff Jarvis]

I know. That's the thing. I don't. I don't need it. I feel pressured because I feel like it's a thing a lot of women and people online like to talk about, but I don't need it. But I went to the dermatologist to get nose rosacea cream because, as we've learned on this podcast, I got rosacea on my nose. And the guy was like, oh, let's go through your skincare routine. And I was like, I wash my face in the shower, and I apply moisturizer when dry. And he was like, oh, I'm mad at how good your skin looks. And I'm like, yeah, you have really.

[Paris Martineau]

You do have very good skin.

[Leo Laporte]

You have really good as, by the way, I've been told my skin is very nice.

[Jeff Jarvis]

It is very nice.

[Leo Laporte]

But you're my age. That's pretty good.

[Paris Martineau]

You know, I had to. They gave me the rosacea stuff, and. And. And it almost sent me into atrial fibrillation, so I had to stop.

[Jeff Jarvis]

Oh, are you allergic to. I.

[Paris Martineau]

No, it's.

[Leo Laporte]

It's a beta ivermectin.

[Jeff Jarvis]

Ivermectin is. Ivermectin is one of the, like, five things they use to treat rosacea, and it's really. It. At first I did. I was like, I can't be doing the horse pace.

[Paris Martineau]

Who did you vote for, Doctor?

[Jeff Jarvis]

Yeah, I know. No, it's because the. For some people, rosacea is caused by mites underneath your skin, and that happens. Helps them.

[Leo Laporte]

But, you know, not for you.

[Jeff Jarvis]

Not for me.

[Paris Martineau]

Smites.

[Leo Laporte]

Not for me. You just have blood vessels that are overactive.

[Jeff Jarvis]

Yeah.

[Leo Laporte]

It's all right.

[Paris Martineau]

So I need. I need you to answer a question. Can I ask a non. AI question? Leo. Leo explains.

[Leo Laporte]

Yes.

[Paris Martineau]

Walmart and Amazon are exploring issuing their own stable coins. What do we think about stable coins?

[Jeff Jarvis]

Wasn't there just legislation passed about.

[Leo Laporte]

Congress just approved stable coins. So what's. Just briefly, what stable coins are? Cryptocurrency. Oh, Paris. Got to get out of here. So we'll do this fast. Cryptocurrency tied to the dollar so that unlike bitcoin, which goes up and down and up and down, it's more stable. Right, Right. What is its purpose? To make money for somebody. Not me.

[Paris Martineau]

Want to get around banks. Right, too.

[Leo Laporte]

Yeah. There's all sorts of digital. The idea is maybe we should have digital currency instead of cash. And it is not, it is not a good idea in my opinion. It's crypto. It's just more crypto. It doesn't solve an existing problem.

[Jeff Jarvis]

It's a tool that lets help people.

[Unnamed Speaker]

With money make more money. That's really it.

[Jeff Jarvis]

Yeah. People want unregulated securities. They want. They want a fun gambling instrument.

[Leo Laporte]

And Congress has just basically given him a tool to do that. But this is the world we live.

[Paris Martineau]

In nowadays because Trump's value, the Wall street journalist story, he has $1.7 billion in value in his crypto.

[Leo Laporte]

Yeah. What a surprise. So he's gonna make war with his phone past what they, I think tongue in cheekily call the genius act because he's a stable genius. Is he is just.

[Paris Martineau]

Leo, you test every phone that comes out. Are you going to test the Trump phone?

[Leo Laporte]

Yeah, I'll buy it the minute it's available. I make that firm promise. It looks. It's so silly. There's no there. They say they'll be made in America. There's nowhere in the world, nowhere in America that could be made. It's not going to exist. It's the craziest thing. And who want. Yeah. Anyway, that's a nutty idea. Okay, we gotta get rid of Paris. So let me pause and say join the club. If you're not a member of the club, very important to our bottom line. Not for me. Don't do it for me. Do it for Jeff. In Paris, 25% of our operating budget comes from club members. It is a vital part of how we operate. And it's, I think, I'm sad to say, only gonna include increase. But you also do it for yourself because if you're a member of Club Twit, you get ad free versions of all the shows. You wouldn't even hear this pitch. You also get access to the Club Twit Discord, which is where some of the best shows happen. Where Micah's crafting corners right after the show tonight. I like it. This is the Club Twit Discord, which is a really fun place to hang out. Anthony Nielsen has made some, some public service announcement stickers for people. Don't you talk to that AI. See, this is the kind of don't.

[Jeff Jarvis]

Talk to an AI wife.

[Leo Laporte]

Don't catch feelings for code AI waifus. Don't you do it simps pay monthly. That is awesome, Anthony. Nice job. We are gonna have in, I think the second Friday of next week Month, the AI user group will talk more about these tools and a lot more. We're also. We also have Stacy's Book Club coming up. Our phone. Anyway, there's a lot of stuff that happens in the club. You're missing out on some of our best content, frankly. In fact, next week I'm gonna have so much fun because remember, we had the guy on that we really thought was great a couple of weeks ago. What was his name?

[Unnamed Speaker]

Mr. Witt. Stephen Witt.

[Leo Laporte]

Stephen Witt. That's. That's it. The author of the Thinking Machine. It was last week. I forget. So he also wrote a book about digital music. Remember we asked him about. And I said, I can't wait. So he's agreed to come back. And then. So we're gonna do an hour with Stephen Witt talking about the rise of the MP3 Napster, digital music that changed the music industry, AI music. And then my good friend Norman Maslov, who has a YouTube channel about vinyl, he's a vinyl boss buff, is going to join us. We're going to talk about retro music. So these are the kinds of things you're missing if you're not a member. 10 bucks a month, $120 a year. There's family plans, corporate plans, and we've just added a two week free trial. So actually this would be a good time to join TWiT TV club. TWiT. Please don't. Do it for me. Do it for Paris and Jen.

[Jeff Jarvis]

Do it for me.

[Leo Laporte]

Do it for Paris. Hi, Zoe Saldana. Welcome to T Mobile. Here's your new iPhone 16 Pro on us. Thanks. And here's my old phone to trade in. You don't need to trade in. When you switch to T Mobile, we'll give you a new iPhone 16 Pro. Plus we'll help you pay off your old phone. Up to 800 bucks and you still.

[Jeff Jarvis]

Get to keep it.

[Leo Laporte]

There's always a trade in. Not right now. At T Mobile.

[Paris Martineau]

I feel like I have to give.

[Leo Laporte]

You something in return for karma. That's okay. I don't really have much in my purse. Oh, let's see.

[Jeff Jarvis]

Hand sanitizer.

[Leo Laporte]

It's lavender. I'm good. Seriously.

[Paris Martineau]

Let me check this pocket.

[Leo Laporte]

Oh, mints. Really, I'm fine.

[Jeff Jarvis]

Oh, I have raisins. I'm a mom.

[Leo Laporte]

Wait, wait one sec.

[Jeff Jarvis]

I've got cupcakes in the car.

[Leo Laporte]

It's our best iPhone offer ever. Switch to T Mobile. Get a new iPhone 16 Pro. With Apple Intelligence on us, no trade in needed. We'll even pay off your phone up to 800 bucks with 24 monthly bill credits new line 100 plus a monthly experience beyond finance agreement 999 and qualifying ported for well qualified plus tax and $10 connection charge payout via virtual prepaid card. Allow 15 days credits end and balance due if you pay off early or.

[Jeff Jarvis]

Cancel CT mobile.com Hey, it's Ryan Reynolds.

[Leo Laporte]

Here from Mint Mobile.

[Paris Martineau]

Now I was looking for fun ways.

[Leo Laporte]

To tell you that Mint's offer of.

[Matthew Kirschenbaum]

Unlimited Premium Wireless for $15 a month is back.

[Leo Laporte]

So I thought it would be fun.

[Jeff Jarvis]

If we made $15 bills but it.

[Paris Martineau]

Turns out that's very illegal.

[Leo Laporte]

So there goes my big idea for the commercial. Give it a [e-mail] Switch upfront payment.

[Jeff Jarvis]

Of $45 for a three month plan equivalent to $15 per month required new customer offer for first three months only. Speed slow after 35 gigabytes of network's busy taxes and fees extra see mintmobile.com.

[Leo Laporte]

All right picks of the Week Picks.

[Jeff Jarvis]

Of the Week Time I've got a pick which is that book I mentioned I've been listening to and really enjoy the audiobook of. It's called they Poisoned the Life and Death in the Age of Forever Chemicals. It's a really great book. Listen, I like a nonfiction book that's kind of about it's though depressing topic, really well written, really interesting like it is. It's about PFAS and BPA and kind of the yeah, how forever chemistry basically how the actually going back to our favorite topic, the creation of the atom bomb led to kind of a chain reaction of events that led to the rise of Teflon and the introduction of forever chemicals into the world's ecosystem.

[Leo Laporte]

Is Teflon a forever chemical?

[Jeff Jarvis]

Yeah, Teflon is made part of the manufacturing process of Teflon involves PFAs and so they are put into it leeches into basically everything it touches. Anybody working in the factory, anybody nearby the factory. And it's gotten to the point that now PFAS levels are found in basically creatures in Antarctica, people who've never even seen a Teflon pan. It's everywhere. And so this book came out May in May, May 6th of this year and it is a fantastic read. I am really really picky when it comes to nonfiction books just because I I like to read books that have great reporting in it. But I also think that they should be written really well in a way that like like compels you and is narratively interesting. And this book kind of hits on all levels and the audiobook is fantastic, so I'd really recommend it.

[Leo Laporte]

I am no longer buying my audiobooks from Audible, but I'm very good places. Well, I, Cory Doctorow recommended Libro fm, which is the same price as Audible and has, I think the same, basically the same library, but sends some of its proceeds to your local, local independent bookstore. And you could choose your bookstore.

[Jeff Jarvis]

Oh, that's great.

[Leo Laporte]

And the download are not copy protected in most cases. Sometimes publishers require it. But that is. So what happened? Actually, I could make this my pick of the week. What happened is Audible, I got an email from them a couple of days ago saying, we've canceled your membership. I've been a member of audible for 25 years. I've bought hundreds and hundreds of books. I've spent thousands and thousands of dollars. Now I was a legacy. I had a legacy account they don't offer anymore called light listener. Two credits for 15 bucks. And so I think they probably, I don't think they're losing money, but they wanted to make more money out of me. So they said, but you can renew for 15 bucks and get one credit. But this was, I mean, they didn't send me a note saying, would you like to do this? They didn't. They didn't. They just canceled my membership. So I thought that was a good opportunity for me to cancel my participation with all. So there are a number of tools that will let you download your Audible library, remove copy protection and turn it into M4BS. I will leave that as an exercise to you because it's not legal in some jurisdictions. But I overnight downloaded all my Audible books. So I don't have. So I don't. They said, don't worry, you'll always have your Audible books. Yeah, right. Just like you, I'll always have my Audible account. So I downloaded all the books. I have found a number of ways to listen. I think after, after asking on Reddit and trying a few different ones, I've decided on one I read actually, which I think is really, really good. It's called, it's got a Bad Name Book Player. It's very generic, but it lets you listen to these M4Bs and in every aspect it's like the Audible player, but I can listen to unprotected books from a variety of sources. That's when I signed up for Read Wise because I'm also, instead of using a Kindle, gonna use the Kobo. I know Rakuten is not than Amazon, but I'm going to use the Kobo for ebooks. But I can offload that into Readwise, and Readwise, which is a paid social network for readers, does also take it. I can export the underlining in the Kobo into my Obsidian notes. So it's a great way to kind of take notes on books and stuff. So, Libro fm, if you're going to buy books, which supports local bookstores and has great books, there's also Librivox, which.

[Paris Martineau]

Is Crowd Read Books.

[Leo Laporte]

Yeah, I like the readers. I like just like Paris. I want somebody who is like an actor. But yeah, Librivox is great.

[Paris Martineau]

Some of them are good.

[Leo Laporte]

There are a lot. Yeah, there are a lot of places. I don't need to partake in this. Audible's got a 99% monopoly in audiobooks. Let's break that monopoly. I've also used Downpour. I have quite a few downpour credits still that also offers a subscription, and downloads are, for the most part part, not copy protected either. Cory Doctorow told me about both of these. So that's my new workflow and I'll just keep these books on my hard drive and move them over to 1 minute remaining.

[Paris Martineau]

Can we make fun of Paris's neighbors in Park Slope line one night?

[Jeff Jarvis]

Ooh, always.

[Leo Laporte]

Park Slope, which is the finest neighborhood in all of Brooklyn.

[Jeff Jarvis]

Full of babies, dogs, people over the age of 40.

[Paris Martineau]

This is a new guy I like on TikTok named Miles Tomes.

[Leo Laporte]

By the way, I. I will give you a little tip. If you take the okay off of Tick Tock and just do ticked with the rest of it. You can download the Tick Tocks and play them, but here we go. Here is your neighborhood. Can you.

[Jeff Jarvis]

You can download them on Tick Tock, just in the upper right hand corner.

[Leo Laporte]

Oh, but I like this because then I don't need to do this. Just take the URL. But anyway, I'll. I'll play it in Tiktoks, gentrification and the Park Slope Parenting Industrial Complex. We're in Park Slope, also known as the People's Republic of Park Slopia and even more commonly known as the Lesbian Dog Mom Capital of the World. Sounds good to me. Where every toddler has dietary restrictions.

[Jeff Jarvis]

That's true. Yes.

[Leo Laporte]

A trust fund. And the whole darn place. It's so suffocatingly perfect, it makes you want to get hit by a city bike just to feel alive. It's a liberal nesting doll of stroll brownstones and white women who whisper Latini unironically. It's like if NPR became a zip code. If Bernie Sanders had a Pinterest board. If Sesame street was sponsored by George Soros. You walk a single. I love it. This is. This is Miles Underscore Toe.

[Paris Martineau]

He's just great.

[Leo Laporte]

Is that his voice?

[Paris Martineau]

Yeah. He does Hoboken. He did Hoboken, too. It's. It's wonderful.

[Unnamed Speaker]

You can just replace all that with the marina here in San Francisco and it's the same thing.

[Leo Laporte]

Yeah, there's. Every town has a name.

[Unnamed Speaker]

Every town's got one.

[Leo Laporte]

I'm just jealous that you live in Park Slope.

[Paris Martineau]

No, she doesn't live in Park. Right.

[Jeff Jarvis]

I don't live in Park Slope, but I won't confirm or deny what neighborhood I live in, but I live in Brooklyn.

[Leo Laporte]

My. My daughter lived in Carroll Gardens for a while.

[Jeff Jarvis]

I love Carol Gardens.

[Leo Laporte]

Do you? Was it a good neighborhood?

[Jeff Jarvis]

Yeah, it's very cute.

[Leo Laporte]

Then that wasn't where she lived.

[Paris Martineau]

Where was Gowanus?

[Leo Laporte]

No, where was the neighborhood? Neighborhood where the Orthodox Jews faced off against Williamsburg. Williamsburg. No. Anyway, that's where that happened.

[Paris Martineau]

That's where they built tunnels and things.

[Jeff Jarvis]

Crown Heights is where the tunnels were.

[Leo Laporte]

Crown Heights. That's where she lived. And she lived near a tunnel. That's right. Yeah, Crown Heights. That's right. She doesn't live there anymore. That's it. We're done. God bless Paris. Go off to your date. I hope he turns out to be as nice as you are.

[Jeff Jarvis]

Yeah, that's exactly what I'm doing. Bye, guys.

[Leo Laporte]

Thank you, Paris Martino. The information doesn't deserve you, so goodbye. Go to Paris NYC if you want to know more about the best investigative reporter in the world. Paris Martin. Oh, thank you, Paris. Thank you. Jeff Jarvis, professor emeritus of journalistic innovation at the Craig Newmark Graduate School of Journalism at the City University of New York. Also now a fella, a jolly good fella at Montclair State University and SUNY Stony Brook.

[Paris Martineau]

Well, I'm a visiting professor there.

[Leo Laporte]

This is why the AIs are going to take over us.

[Paris Martineau]

Yeah, they are. Because we're a pain in the ass.

[Leo Laporte]

We're going to do a better job. And all of your podcasts will have that happy talk duo on the entire world. Oh, my God. Can you imagine? Aren't you glad that there are.

[Paris Martineau]

They're the podcast cockroaches. They'll be all that's left.

[Leo Laporte]

There are some raggedy ass podcasts, like intelligent machines. We do this. This show every Wednesday at 2pm Pacific, 5pm Eastern. That's 2100 UTC. You can catch it on eight different. Eight, count them, eight different platforms.

[Paris Martineau]

Of.

[Leo Laporte]

Course, if you're in the club, Discord's the place. But YouTube, Twitch, TikTok, X.com, linkedIn, Facebook and Kick. How about that?

[Paris Martineau]

You did it.

[Leo Laporte]

I did it. Next week, Richard Gingris. Are we. Do we know if Richard will be joining us?

[Unnamed Speaker]

I believe so, yes, he has confirmed.

[Leo Laporte]

Good, That'll be fun. Tell us about Richard, Jeff.

[Paris Martineau]

Richard has been until I think the 23rd of this month. He's a senior VP at Google. He's the guy at Google for news. Google News guy, Yep. And all kinds of things around news. And so he is retiring. He's going to meritus like me and can talk about AI and the news and AI and Google and fair use and that kind of stuff.

[Leo Laporte]

He's been on before. We love him. He's a great guest and I look forward to that. That'll be next Wednesday, 2pm Pacific, 5pm Eastern. We always start with the interview. Then the silliness begins with the AI news. It really ends. You can also watch, of course on our website, Twitter, tv. I am anytime time. All the shows are there, going back even 824 episodes to when it was.

[Paris Martineau]

Called believe it or not.

[Leo Laporte]

Google. Yeah. How does that happen? That's because the AIs haven't taken over yet. That's why they're gonna wipe us out the minute they do. You also can watch on YouTube. There's a dedicated YouTube channel for intelligent machines. Best thing to do, subscribe in your favorite podcast client. That way you'll get it automatically as soon as we publish it. Which will be anytime later today. Today. And leave us a five star review so Paris can read it on the air. She's good. We got to bring that back. If she'd had time, I think we would have been doing that today. But we'll do it next week. Thank you everybody for being here. Thank you, Paris. Thank you, Jeff. Thanks to you especially Club Twit members. We really appreciate your support. We'll see you next time on Intelligent Machines. Bye. Bye. I'm not a human being. Not intuitive. Marketing is hard. But I'll tell you a little secret. It doesn't have to be. Let me point something out. You're listening to a podcast right now and it's great. You love the host. You seek it out and download it. You listen to it while driving, working out, cooking, even going to the bathroom. Podcasts are a pretty close companion. And this is a podcast ad. Did I get your attention? You can reach great listeners like yourself with podcast advertising from Libsyn Ads. Choose from hundreds of top podcasts offering host endorsements or run a pre produced ad like this one across thousands of shows. To reach your target audience in their favorite podcasts with Libsynads, go to libsynads. Com. That's L I B S Y N Ads. Com Today.