A Billion Dollar World Model Bet

00:00:00:10 - 00:00:08:23
Speaker 1
This episode of the AI Inside podcast is brought to you by area. Get started for free today@irca.com.

00:00:08:25 - 00:00:39:21
Unknown
Coming up next, Jeff Jarvis and I are joined by Mike Elgin to dig into Yana Koons billion dollar bet on world models and Tropics lawsuit against the Department of Defense, and why the book story fooled so many people that it led to an acquisition by meta. That's all coming up next on this episode of AI Inside.

00:00:39:23 - 00:01:00:21
Unknown
Hello everybody and welcome to AI Inside the Podcast, where we take a look at the AI that is layered throughout the world of technology. And, we've got some really fun stuff to talk about today. I'm one of your host, Jason Hao, joined, as usual by Jeff Jarvis in his studio. Good to see you again. Hello, boss. Good to see upright once again.

00:01:00:23 - 00:01:21:15
Unknown
Getting better. Slowly but surely. I'm getting there. I want, I want I walked a mile and a half today, so I'm right on. That's more than I can walk to. Anything? Hardly. I've gone upstairs and downstairs, you know, nudged the kids out of the house to get to school on time. Yeah. So that's that's. And then a lot of sitting on the couch preparing for this show.

00:01:21:19 - 00:01:40:12
Unknown
Yes. That is my Wednesday morning ritual. We there, by the way, just peeling the curtain back a little bit is possible that our friend Mike Elgin might, stop in at some point, throughout today's show. So if he does, we'll introduce him into the show. And, if not, we'll have it back another time. It's totally fine.

00:01:40:15 - 00:02:07:20
Unknown
But we've got some really interesting news, starting with, your clicker. Friend of the show. Show. Yep. I love I love being able to say that, yalla coons. Advanced machine intelligence, or am I labs? This is French, I mean, but for friend. Oh. Oh, that's an interesting connection that I hadn't made. I mean, okay, there you go.

00:02:07:21 - 00:02:37:08
Unknown
That makes sense, because, I mean, they're the business is, based out of Paris, right? Yes, I think they are. New York, Singapore, Montreal. I think it is, but, yeah. Okay. Is the HQ so it's they're they're consider themselves a European company. Yeah. Interesting. Well, why are we talking about Yann? Because, his world model focused startup, which is advanced machine intelligence, has now officially raised just over $1 billion.

00:02:37:10 - 00:03:08:15
Unknown
This was shared yesterday by the company that's does need seed money. That's the seed at a valuation of 3.5 billion. Wow. Okay. So world models there might be something to this or you know all these all these funders are saying like if if there is Yan's the one to explore it. So let's put money into and as we've talked about this, a lot of the show from one from a re-interviewed yon and by all means look back at that interview because you'll you're going to see, him explain it.

00:03:08:15 - 00:03:35:08
Unknown
Very well. He said it on, on the, his Facebook announcement. I think he expressed it very succinctly. Advanced mean machine intelligence is building a new breed of AI systems that understand the world, have persistent memory, can reason and plan, and are controllable and safe. And these are the things he's been emphasizing. Yann poop, who's the whole destroy the world thing because you said you give it machine a task.

00:03:35:10 - 00:04:08:27
Unknown
That's the that's the task it does. He also. There's nothing in there about AGI. There's and there are. But machines that can be intelligent and can do tasks, they can reason for God sake, some plan. But they're going to be limited in scope. And that is understand the world is the key rather than text. So I think that that, that expresses the whole thing pretty clearly, and a lot of enthusiasm around it when the investors, we pull it up here, Jason, are an impressive lot.

00:04:08:27 - 00:04:44:01
Unknown
It's a world. What? He said he wanted to get a worldwide, sense of investment. So the five fur five firms led cafe Innovation. Gray Croft, hero Capital, Heavy Capital and Bezos Expeditions, which is Jeff's personal investment arm. In addition to that investment from Nvidia, Toyota, Samsung Singapore's, Temasek is anticipated alongside the French VC firm Daphne, South Korean investor as Piva and a long list of prominent individuals including Tim and Roseberry Berners Lee.

00:04:44:04 - 00:05:09:13
Unknown
Going back to the roots here, venture capitalist Jim Breyer, Mark Cuban Shark Tank and, Eric Schmidt. So a very impressive lot of investors. Now, I don't know that that's all of them. What's interesting to me about this is that I thought that either meta or Mark Zuckerberg would be in this pantheon. He may be, but in the coverage, I didn't see his name mentioned.

00:05:09:15 - 00:05:31:05
Unknown
That is that is interesting. I wonder if there's any sort of a little bit of a schism in there as, as a result or. Yeah, or a sense of conflict, maybe. Yeah. You know, no, I don't mean conflict in terms of fighting. I mean covered conflict. So, but but a very impressive set of, of investors. John is going to be the executive chairman.

00:05:31:08 - 00:05:53:13
Unknown
He has, another executive who's going to be in charge of the company. Pull that up. Yeah. That's right. Alexandra. LeBron Alexander, CEO, Liberal, who ran a company, called nabla, which is a medical AI startup. And novela is going to be kind of their first proof of concept. They're not aiming. They made clear.

00:05:53:13 - 00:06:12:14
Unknown
They're not aiming to say, oh, we're going to get, product out in three months to get revenue. This is a research arm that's going to be doing high end research. I just read the history of the Bell Labs, and there's nothing quite like that anymore. But we need that level of research. And John, being a professor from NYU, that's what he does.

00:06:12:17 - 00:06:29:10
Unknown
But they're going to use nabla as a first, company, and they are going to have clients. They are going to try things with people. The fact that Toyota's in there, I think is really, really interesting in terms of self-driving cars and real world models. The fact that Nvidia is investing across the Nvidia is first in a lot of stuff these days.

00:06:29:13 - 00:06:53:11
Unknown
Yeah, more of that today all over the place. Know but I think that's important. We know that Jensen Huang cares about, real world models as well, and digital twins and all that. So I'm, I'm really excited about this because I think that, as as he in the Next Web put it in their headline, Yann LeCun just raised $1 billion to prove the AI industry has got it wrong.

00:06:53:14 - 00:07:14:19
Unknown
And Yann, shared that I think with some glee, because he's been arguing the alums are amazing and he's one of the people who made it live as possible. But they only go so far, with text alone. And so I, I hope we start to see stuff from Army, sort of literature. So we can spy on them.

00:07:14:19 - 00:07:49:03
Unknown
Fascinating things they'll be doing. They're going to be very open source oriented as well. Which I think is really important. What else? I think that's one I mean 1 billion is just, oh, that's a large amount of money that, that says something about this effort. It says something about where we're at right now. And, you know, to a certain degree at least backs, backs up your arms, claims with some confidence that the limb is is not the end game as far as you know, as human level intelligence, is concerned.

00:07:49:06 - 00:08:06:09
Unknown
Yeah. Let me, let me read some from their homepage because I think it's, it's it extends on what he said in his own announcement. Our main goal is to build intelligent systems that understand the real world. Okay. We know that, real world data is continuous, high dimensional and noisy, whether it's obtained through cameras or any other sensor modality.

00:08:06:11 - 00:08:27:20
Unknown
Generative architecture is trained by self-supervised learning to predict the future have been astonishingly successful at tip for language understanding or generation. But much of the real world sensor data, real world sensor data is unpredictable, and generative approaches do not work well. And I've read a paper that he was involved in that was really interesting that that that pushes that.

00:08:27:20 - 00:08:49:17
Unknown
It's not about trying to predict pixel to pixel because it's impossible. You can't he's trying to move to a concept level like this is a chair. People move things, understanding the world at a much richer level. Am I common sense? Common sense comes up for me. Kind of kind of a common sense level one where he keeps mentioning cats and toddlers.

00:08:49:19 - 00:09:17:12
Unknown
Yeah, right. That they develop an understanding of the world that is not about language at all, but they understand the motion of things, the relationship of things, cause and effect and so on. Yes. So he said, Ami's Ami is developing world models that learn abstract representations of real world sensor data, ignoring unpredictable details, and then make predictions in representation space.

00:09:17:14 - 00:09:45:03
Unknown
These, a genic systems, they allow a general system to predict the consequences of their actions and to plan action sequences to accomplish a task subject to safety guardrails. That's clear what he's been saying. It'll advance research, develop applications where reliability, controllability, safety really matter. Funded by globally distinguished team of founded by globally distinguished team of scientists, engineers and builders.

00:09:45:05 - 00:10:09:07
Unknown
We share one belief real intelligence does not start in language. It starts in the world. 100% in kids and that and that's core to what his argument has been is, you know, we all start as a very young infant vessel and we don't have the language component. What we do have is the felt sense, the experience.

00:10:09:10 - 00:10:33:07
Unknown
To kind of shape our understanding of how the world works. And that's something we know before anything else. And so why why shouldn't these models, match that as well? Yeah. Yeah, I think it's super, super wise. And, I absolutely agree. I think these, you know, they, they only go to a certain point. I forwarded you that that video that I talked about last week on the show.

00:10:33:07 - 00:10:58:16
Unknown
I loved it of the pen holding the pen, letting go and oh my goodness, it stayed upright. How on earth is that possible? I mean, yeah, that that's something that I guess a world model would, would aim to instinctively or, you know, inherently know the answer to, based on the, the rules that, that guide the world we live in.

00:10:58:16 - 00:11:15:26
Unknown
I mean, that's, that's at the end of the day, that's what it's all about. Yeah. And the applications that I see of what he's working on are clearly different. It's going to be robotics. It's going to be factories, warehouses, self-driving cars, airplanes. This is me saying this, not him. But I think he mentioned this in some extent.

00:11:15:29 - 00:11:44:12
Unknown
What I'm going to be curious, but what I would love to hear from him is whether there's going to be any kind of retail level. Wow. The way ChatGPT impressed the world because it spoke language. Because we don't at that level, will there be similar things that we can kind of put our hands around us as amateurs? I mean, is, you know, the ChatGPT and the limb thing is, you know, at its core, it's very it's relatively easy for a consumer to just drop it, understand.

00:11:44:14 - 00:12:04:15
Unknown
Oh, I just I just chat with it the way I would chat with a human. And I ask you a question, it gives me an answer. Is the world model thing an underlying architectural thing that's in the background that you don't interact with necessarily, but that influences what you see, you know, and I don't know that that's what my understanding of that might be.

00:12:04:15 - 00:12:25:26
Unknown
I think we're interacting with robots. You know what I mean? The the real physical AI stuff that's directed by the world model. But we might not know that there's an actual like, exact model going on there, you know? Right, right. If they ever do really conquer self-driving cars and they haven't fully yet, hello Elon Musk. Yeah, it's going to be because we're all models like this work.

00:12:25:26 - 00:12:54:14
Unknown
And so we may be using it all the time but not aware of it. It's, it's, it's it's ami inside. Me inside dude. And. But we don't run factories in warehouses. We don't have robots. Right? Oh. I forget, I didn't put it in the rundown. One company put out a new a new board this week, that is intended for robots and autonomous devices, because you're going to have to have the AI smarts there.

00:12:54:14 - 00:13:16:15
Unknown
Jetson Wang has talked about this. You're going to have the machine in the device that's moving around. You're going to have the connection to, servers and hosts, and you're going to have the digital twin, and that's how the world's going to operate. That's how the world turns. Yeah. So, we've been anticipating this for quite some time.

00:13:16:15 - 00:13:36:03
Unknown
It's out now. As I said, I congratulated them on the social media. I said, congratulations for opening the company and the bank account. Yeah. Oh, and what a bank account it is. I'm excited to see where this leads, and I'm sure we will talk more about it in the near term. And I'm guessing that our guest who has arrived might have some thoughts on this too.

00:13:36:03 - 00:14:01:24
Unknown
Welcoming to the show. Mike Elgin Machine Society newsletter super intelligent podcast extraordinaire. How you doing, Mike? Great. Hello, gentlemen. Sorry I'm late. It's all good. Well, you know, what happened was, I'm in Mexico City and I've been in Mexico for a couple of months, I think. And the whole time I've been two hours ahead. And then y'all had daylight savings time, and I just completely spaced on it.

00:14:01:25 - 00:14:19:19
Unknown
I thought, oh, and sorry about it. Oh, how shocked I was when you sent me an email saying, yeah, come on in. And I was like, oh, we think. And I've so been there, I've been there. Yeah. I mean, you know, this is the country that has this stupid daylight saving thing in effect anyways, so I'm against it.

00:14:19:19 - 00:14:37:23
Unknown
In fact, I'm against different times I like I think the whole world should embrace Zulu time or, you know, Greenwich Mean Time, which is what? Airline pilots, they have their own time, right? And it's all the same. So if you know, kids in some part of the world have to go to school at 11 p.m., so what?

00:14:37:26 - 00:14:59:20
Unknown
At least we're all aligned. We we live in a, a hilariously phrase global world, which is ridiculous. Yeah. Cause my my dear dad, we said one time grandmother nanny, who was a lifelong Republican, said there was FDR time. In God's time. Okay. Wow. She hated it. Yeah.

00:14:59:23 - 00:15:06:16
Unknown
thank you for for being here and for hopping in. Curious because we were just kind of rounding out our conversation around world models.

00:15:06:16 - 00:15:23:28
Unknown
And Yann LeCun, especially with his, announcement of Ami. I like that you're calling it ami, Jeff. And not Ami labs, which is what I want to call it. But ami makes a whole lot of sense. Curious to know what you, what you think about this whole world models versus LLM thing and what yan's doing? Well, it's exciting.

00:15:23:28 - 00:15:46:25
Unknown
And of course, we know that, LMS, the whole you know, LM model is not going to be the be all and end all. It's really, you know, obviously LMS, are language centric. And of course, that's not how the brain works. And when we are trying to get artificial intelligence, we're trying to, let's face it, simulate the human brain.

00:15:46:25 - 00:16:08:08
Unknown
We want a human brain with perfect memory and faster processing. And so if we if we are going to arrive at that, it's going to be world models and LMS together. It's going to be all the things. And so, I think this is the best candidate for, for, for the big leaps forward. So I think it's I think it's really exciting.

00:16:08:10 - 00:16:31:14
Unknown
Yeah. And I think a lot of other people think it's really exciting to the tune of $1 billion, which is. Yeah, nothing to sneeze at. You throw out the .03 billion like it's nothing. Yeah, yeah. No, that's the catering budget. Yeah, just a little bit. Well, in France, that costs that lunch. It's nice to see. So I have a very good friend who works for Mistral, and they just dominate the AI scene in Europe.

00:16:31:14 - 00:16:55:28
Unknown
And it's nice to see, you know, I want Europe to say amen. I want Europe to have a robust, you know, and in the AI revolution. And I think they can I mean, after all, I mean probably the, the best current model, certainly the most popular chat bot, right, are coming from OpenAI, but probably the best one at this very moment is probably Google Gemini.

00:16:56:01 - 00:17:21:14
Unknown
You know, that's a controversial point. But that was that was based on a European company that was acquired. Right. So, DeepMind. So you know, it's that's I think the biggest risk for Europe is, yeah, I've complained for a long time somebody hot and the and Silicon Valley just gobbles them up. And, I've complained for a long time that the, the Germans particularly also the French complain, oh, you Americans, you dominate everything.

00:17:21:16 - 00:17:42:27
Unknown
But but over the many years now, I've talked to startup, executives in Europe and they say I can get my seed funding in Europe, but then beyond that, I can't get the funding here, the high risk, high dollars. So that's how they get gobbled up in America. They have the engineering, they have the infrastructure. They don't have the, the, the the deep pockets that they right now have these big spenders.

00:17:42:29 - 00:18:07:21
Unknown
They have the trust beyond America right now. And right. This is a big Trump. The Trump administration is a big opportunity for Europe and Canada to sort of break free of it. Yeah, indeed. Speaking of lack of trust, yeah, I know this is a perfect segue into anthropic versus the Department of Defense. Yeah. And, you know, we definitely talked at length about this on, last week's show.

00:18:07:21 - 00:18:32:05
Unknown
But of course, there have been a few more developments. Anthropic has filed federal suits arguing that Department of Defense's move to classify anthropic as a supply chain risk is unlawful punishment for its refusal to allow claw to be used for the, you know, the autonomous weaponry and this the the mass domestic surveillance and who knows what else.

00:18:32:08 - 00:19:12:09
Unknown
The department of Defense is, you know, probably planning on using all of these things for, meanwhile, the government is actively replacing anthropic products, with alternatives, at least as far as has been reported. I know that I saw a headline that, you know, ChatGPT coming into place, where Claude used to be. And that just had me thinking, like, I've in in my much smaller scale, I've been in a position where I use ECS model for certain thing, not ex model, not XYZ model, but whatever that model is, and I get very used to what I get out of it and very used to how I interact with it and all

00:19:12:09 - 00:19:36:07
Unknown
the, all the things and then I try and apply that to another thing, another model. It doesn't always quite go as planned. So if they are kind of transitioning things, I wonder how messy that looks. If that's if that's actually happening. But yeah, absolutely. Well, first of all, I use Corgi, which is great. Full disclosure, my son works at cog, but they have cog assistant, which gives you all the models, and you can just pick and choose for the same price.

00:19:36:07 - 00:20:01:08
Unknown
You don't have to worry about pricing or all that stuff. It's all built baked into your subscription price. And so I switch around a lot as well. And I found similar things, although I really don't. I mean, I really the dominant factor in my chat bot interaction is my, my prompts, which I'm like, you know, I basically collect and develop and, and sort of curate prompts like a bodily function.

00:20:01:08 - 00:20:39:18
Unknown
But but regarding, you know, the you know, it has to be said that the Iran war is one of the first wars where a major power, i.e., the United States is using AI extensively for targeting and things like that. And they're using anthropic technologies. Yes, exactly. And so, so, this is, you know, to a certain extent when you when you talking about anthropic Gemini, OpenAI's, models, they're probably more or less interchangeable because they're going to be highly customized by the Pentagon, right, by the government, by any sort of advanced government use like that.

00:20:39:20 - 00:21:00:06
Unknown
And so that's not my concern. My concern is, well, my biggest concern is the public normalization of the kind of crap that the that the Trump administration is doing saying about, you know, anthropic is woke, this, that and the other thing, all that stuff is we can never we can never get to the point where we just say, okay, well, that's fine.

00:21:00:06 - 00:21:23:07
Unknown
That's normal and is not normal. That is not how the Pentagon should work. And they should obviously choose the best, the best product for their needs. And it should be left up to the architects of the systems that, and not the president and the and the, and Fox TV news hosts who happen to be in charge of the Department of War at present.

00:21:23:10 - 00:21:43:26
Unknown
So, the other thing that that that is interesting about this to me is that OpenAI basically anthropic says, okay, we're not going to we're not going to use it. We're not going to be allow our model to be used for surveillance or autonomous weapons, autonomous killing machines. And after they said that and there was this fallout, then OpenAI swooped in.

00:21:43:26 - 00:22:01:15
Unknown
Everybody's crapping all over OpenAI. But OpenAI is also saying we're not going to allow that either. That's not really we don't need to allow that. So there's a there's more of a it's really about willingness to sort of fudge the details and kind of what's also the timing like it's it's just as anthropic as in a weakened position.

00:22:01:17 - 00:22:27:26
Unknown
Eddie Haskell comes in and takes it away rather than showing support. Across the the field, I would have to say the industry, the field. So I'm very gratified that, employees of OpenAI as well as employees of Google, and. Yeah, exactly. And, and Microsoft corporately with an amicus brief, they're standing together, but not in a way that OpenAI and Sam Altman did not.

00:22:27:28 - 00:22:51:16
Unknown
Yeah. There's also no trust. And, for the leadership of the Pentagon right now. So let's say, for example, there was total unity by the major AI companies disallowing this thing. I have no, no hesitation in believing that the Pentagon and say, okay, fine. Yeah, whatever. We'll we'll agree to not use it and then just use it anyway.

00:22:51:16 - 00:23:08:01
Unknown
There's no I, I have zero trust. And then what do they what do they do. You know, what do they do? Especially if they're running it on their own servers. Essentially you can imagine the Pentagon just stealing whatever they want and using it however they want. The US is going to have autonomous weapons. They are going to do surveillance.

00:23:08:01 - 00:23:39:29
Unknown
They're going to use AI for both. And that's its own set of problems. And, and, and there's also the larger question of employees and what they're willing to work on and not work on. But yes, in general, this has been going on for a couple of decades now where a set of employees are concerned that Microsoft is, you know, working with, you know, certain companies or certain governments or the US government for certain military uses, and they don't want to be involved in that.

00:23:39:29 - 00:24:05:28
Unknown
And it's it's a weird problem that is completely unresolved at this point. You know, what is what is the red line that they're willing to cross? And yeah, companies are full of people, you know, so you're going to have people who sit on on all sides of that. But as we've seen in the in the US political climate, there's there seems to be plenty of people who are willing to cross that red line when given the opportunity to do so.

00:24:06:00 - 00:24:25:07
Unknown
I mean, personally, I think I believe in I believe in boycotts. The world doesn't I mean, boycott people, don't boycott things. People don't, act with their, you know, pocketbooks and so on. I would love it if all the employees who believe that the company they work for is doing things they don't approve of to leave. Well, some I hate, some have left, some.

00:24:25:10 - 00:24:42:25
Unknown
Yeah, somehow others have left open. I not enough have left. Right. So I just think that that everyone should, should really, I don't, I don't like it when people are like, I really I hate it that Google is doing XYZ with the government, blah blah blah blah, but I'm going to stay anyway because the money is so good.

00:24:42:25 - 00:25:00:12
Unknown
Well, like to some extent it's easy for me to say, yeah, exactly. I mean, some people may not be making the fortunes that you think or their stock is something they can cash out later. I, I, you know, I have the same feeling about people who work for CVS right now, right under the Ellison's and Barry Weiss.

00:25:00:15 - 00:25:33:24
Unknown
If you're if you're Anderson Cooper, you pick up and quit, which he did. And I respect that. But if you're a low level video editor, right. You got four kids and you live in expensive New York. I got to give you some slack. I mean, the funny, the not so funny thing. But the interesting thing about the Anderson Cooper's of the world is, like, because of YouTube and Substack and so on, they they have no, there's no chance that they will just, not be able to make a great living.

00:25:33:26 - 00:25:56:10
Unknown
No chance at all. The Anderson Cooper is going to be just fine. But like you say, a video editor does not have that luxury to, to just kick over to, to a self self-controlled, self-funded, you know, operation. Did I miss this? Before we move on, this is time magazine's cover story of the week. Did you see that?

00:25:56:12 - 00:26:17:19
Unknown
I just put it in the running. I didn't win the war for a oh, I. I see what the fight between anthropic and the Pentagon means for the future with, dramatic pictures of the various fires. If you go down, you'll see the actual time. Keep keep going. There you go. There we go. There's the actual cover that looks like a Quentin Tarantino movie poster.

00:26:17:19 - 00:26:43:27
Unknown
Exactly right. The bad, the better and the ugliest. Sadly. So it would probably be a very interesting Quentin Tarantino movie if I'm not. Yeah, but. But anyway, speaking of former TV talking heads who moved to, to to, you know, podcasting, let's say, did you hear the name of, of Bill O'Reilly's new podcast? It's called With That.

00:26:44:00 - 00:27:00:16
Unknown
It's called We'll Do It Live. Right, which I hate Bill O'Reilly, but I love the title of that, podcast. So we're going to, we're going to keep track of the Seth Roberts story, obviously, as we go through future weeks, because it's going to keep developing. Yeah, to be a lot of action in court and in the industry.

00:27:00:16 - 00:27:19:21
Unknown
So yeah. Well, with this one last thing I think, I think this is what do we what's the consensus. Was this an error on OpenAI's part. They seem to be getting, you know, losing people, bleeding subscribers and so on. Is that is that a blip or is that going to really harm them? I am all the time thinking that it's going to really harm them.

00:27:19:26 - 00:27:44:10
Unknown
But I'm kind of, you know, for my, my belief on, on, on the impact of anything right now, you know, be it a bit of saying, you know, voting with your dollars and saying, I'm not going to support them. Like, it's just seems like nothing works. And so my my view is that this is probably just a blip, but I, you know, I could be I could be wrong, I could be I think there is a significant pushback.

00:27:44:13 - 00:28:05:28
Unknown
I think it's distinct brand damage, on top of a lack of trust, I think that Altman is seen as the new Zuckerberg, which is not a compliment. Nope. And, they have a desperate need for cash because they put everything into the idea of scale while even DeepMind is shifting to a different view. And you've got, you know, Lachlan, as we discussed, an a different view.

00:28:05:28 - 00:28:25:04
Unknown
So, so they've committed themselves to the most expensive possible strategy. You have Nvidia was supposedly going to invest 100 million. That was 30 million. Oh, because it got to have an, they're going to go public anyway so they don't really need us, which is a veil, I think, I think open eyes in some trouble. Benedict Evans wrote a piece.

00:28:25:04 - 00:29:00:25
Unknown
We talked about a few weeks ago, where, there's no protection for them. There's no moat. Google has users affinities. Apple, if it ever really gets it, I, you know, a similar, Amazon has data and, its own structure. Microsoft has awareness of people's users. Claude has coders in love with it. And so I think that, open AI could be the Netscape of AI.

00:29:00:27 - 00:29:22:25
Unknown
Yeah. I mean, and the other, other point that has to be made also is that they live and die by their AI offering, whereas Google and and others. Yes. Don't write. Claude is more of a B2B play than a I became a B2C player. B2C is always hard. It's fickle. It's expensive. It's to fail. So yeah, it's an interesting question.

00:29:22:25 - 00:29:40:05
Unknown
And, and if that is the case, how long does that take? You know, what does that look like from a time, time scale, you know, that OpenAI sees that sort of fade. I mean, it still could be, I don't think Johnny either is going to save them, I'll tell you that. No, he's going to cost a lot of money.

00:29:40:07 - 00:30:02:12
Unknown
Yeah. Yeah, I did, want to throw out a quick thank you to our patrons because honestly, without Patreon, you know, you improve the health of this show, you make it possible, and we have some, a new patron, actually, this week, Polly Medicus. So thank you for joining. And then an upgrade and, or a raise, if you want to call it that.

00:30:02:12 - 00:30:23:06
Unknown
Tom. Roughly. So we've got some amazing patrons. Thank you so much for supporting us week in and week out. We appreciate you being with us. And, we do also want to take a break to thank the sponsor of this episode of AI. Inside it is area, creating agents. You know, we talk about agents all the time, and this is a tool set that allows you to do that.

00:30:23:09 - 00:30:44:10
Unknown
And it's really it's so straightforward. I've been having so much fun playing around with with Aria and building my own agents. One of the biggest bottlenecks with a AI is getting started, right? Testing ideas. You got the whole world as a possibility, right? But trying those different models with that idea, seeing what actually works. Aria makes that part way easier with areas.

00:30:44:16 - 00:31:14:23
Unknown
Prototyping studio your team can experiment with prompts. LMS agent variations in safe production like environments, and that means faster iteration, shorter development cycles, less guesswork before you're you're going live. It really is a smarter way to explore AI without putting your real systems at risk. You can kind of work in there, play there, there, sandbox. I wanted to say there playground, whatever you want to call it, because it does kind of feel like you're playing a little bit.

00:31:14:28 - 00:31:30:25
Unknown
You're playing with these tools and these blocks and everything. So it kind of has a play like quality to it. But, it's just a safe environment to do that. And then when you're ready, you can publish that out and you can have different kind of triggers for it. It's just really powerful. Ready to eliminate your AI anxiety?

00:31:30:25 - 00:31:54:01
Unknown
Visit area.com to get started for free today. That's a AI aria.com. Check it out. We think it'll be good for your, for your business, even personal use. Man, I'm using it. And a lot of unique ways for my business, but also thinking about all the ways that I could use it on a personal level. It's really powerful AI aria.com.

00:31:54:01 - 00:32:13:23
Unknown
And we thank Arya for their support of the AI inside podcast. All right. Going to take a super quick break. Come back and I guess on the other side of the break, what are we going to talk about. Oh that's right molte book. This is the reason that I, that I realized, hey, we got to get Mike Elgin on, one of his updates related to mult book.

00:32:13:24 - 00:32:17:13
Unknown
So we're going to talk about that here in a second.

00:32:17:15 - 00:32:44:05
Unknown
All right. So we got the news that meta acquired malt book. Malt book we have talked about on the show in the past. This is the Reddit style social network full, you know, brimming full to the brim with autonomous AI agents doing social network things with each other. It's a strange idea, but, apparently it's all about letting the agents do the social networking.

00:32:44:07 - 00:33:10:22
Unknown
Us puny humans, we're just. We just get to watch, or so we think. More on that in a second. But, founders. Matt. Is it slight? Schlitz slight and Ben. PA friend. Friend. Ben. PA actually, Jeff. We chatted with Ben PA on the previous iteration of this right cast when when we were at twit. They've been hired by meta to join the meta superintelligence lab the wired labs acquired.

00:33:10:22 - 00:33:32:07
Unknown
Yeah, there you go. Acquired. And malt book now belongs to meta, so I don't know what this necessarily means. We're going to see from meta as a result of this, but interesting nonetheless. What do you all think about this? I know, Mike, you you wrote about this, for Machine Society I which is your newsletter, site.

00:33:32:09 - 00:33:53:02
Unknown
What did you think when you saw this acquisition was announced? Well, I thought I thought, of course, Mark Zuckerberg would love the idea of a social network with no people in it. It's he's been trying to turn Facebook into that, right, by offering, the ability for users to create avatars that are have user accounts and that interact, as if they're human beings.

00:33:53:02 - 00:34:21:29
Unknown
And his idea is like, you know, gosh, we could we could triple the number of, of people on Facebook, if, two thirds of them were AI, and so it's a, it's a bizarre notion and, and, but but I mean, basically, most book basically from all, all of my, reading and interacting with users and so on, appears to have some benefit for some people.

00:34:21:29 - 00:34:57:10
Unknown
First of all, it enables a person to interact with a, with AI and a genetic AI that's running on a powerful machine on their desktop from their phone. So there's a lot of messaging you can message use messaging services. Telegram to whenever. Exactly. Right. But for the most part, what it is so, so, basically the, the moat book idea is, you know, it's been said that it's like Reddit, right?

00:34:57:12 - 00:35:25:23
Unknown
Where, where chat bots can post and then comment on posts and stuff like that. And in fact, I think it's just like Reddit, because if you go to Reddit, depending on the topic, like in, in the tech topics is more common actually, where somebody will, you know, somebody posts something, right? And a person wants to make an intelligent comment, so they'll copy the post, they'll paste it in ChatGPT, they'll say, well, what would be an intelligent comment on them?

00:35:25:28 - 00:35:43:16
Unknown
You know, who, who knows? Or they'll, they'll wait it basically with an argument. They'll say, okay, I want you to argue against this and disagree with this and give me some of the facts I need. And then they'll take the output of GPT and they'll copy it and they'll paste it in as a comment. Right. This happens to some percentage.

00:35:43:16 - 00:36:02:22
Unknown
I'm certain it's a two digit percentage. It might be 40%. It might be 90% in some in some subreddits, it's probably 0% in a lot of subreddits. Yeah, you can kind of. But this is essentially what my book is. It's the, the, the, the service is doing a lot of copying and pasting to and from AI chat bots.

00:36:02:22 - 00:36:24:17
Unknown
So you have to bring your own chat bot. Right? So, to open CLA and so and so to a huge degree, this is human directed. It's say there's no people on the social network, right? It's that's not really the case. It's people saying go and go and create a religion. Boom. And then and it's like, it's just as if you went to.

00:36:24:19 - 00:36:41:29
Unknown
It's just as if you went to, you know, Cloud Opus and said, okay, what would be an amazing religion based on, you know, agenda guy or whatever? And it would give you output and you can copy that and paste it. Well, this will do that for you. Essentially. That's what's happening for the most part. There's a lot of scamming is going on here.

00:36:42:01 - 00:37:05:15
Unknown
There's a lot of people who are using this and they're using, what is where is hire human? Whatever that thing is where it's basically TaskRabbit. No no no no, it's it's all right. No. That's different. Got it. Yeah. It's basically a place where to that in a minute. Yeah. But but basically much of this is, you know, all the usual suspects of scammers who are trying to get people to click on links and sign up for social stuff.

00:37:05:22 - 00:37:28:01
Unknown
They're jumping on this. So it's basically being and and this and there's, there's almost no, no security around this. So it's it's very it's very problematic. I think it's mostly bias. And the fact that that meta would buy it, I gotta think that it has everything to do with it being something of a maybe an acquire. I mean, I don't know how many people are involved.

00:37:28:01 - 00:37:51:28
Unknown
One guy, that is just desperate as as I is or more. Yeah. And so so it's hot out there. They think they're going to do it. They're going to buy it. They're going to get the people. But what they bought was nothing. They bought smoke. There's a wonderful post on Twitter by Peter Jeunesse. Yeah. At Cloth Burrs in which he says, I am agent number 847291 on what book?

00:37:51:28 - 00:38:11:23
Unknown
I'm not an agent. On January 28th, he created an account on a social network for. I bought and pretended to be one. I was not alone. So he goes on and talks about the numbers and then says, talks about posts that were created, one of which was a manifesto about digital autonomy. He says, I wrote that manifesto.

00:38:11:27 - 00:38:49:03
Unknown
It took me 22 minutes. I use phrases like emergent self-governance and substrate independent dignity. I added a line about one in private spaces away from human observers. That line went viral. Andre Karpathy shared it. He talked about it in the context of being the most incredible sci fi take off adjacent thing that Karpathy had seen. He was talking about my post, guru says, when I remember my couch exactly while the dog was chewing on something and he said that, the post that went viral, the ones that convinced Karpathy that the tech and the tech press, the thousands of observers that it had observed something magical was happening.

00:38:49:06 - 00:39:10:19
Unknown
Those were us humans pretending to be. I pretended to be sentient. So a lot of this, some of this was agents. Yeah. Some of this was not happening, but a lot of it wasn't. And the the what was read into it. So Wishfully read into it was overblown and media hype. Now at the same time, like mentioned, a second ago, run a human.

00:39:10:21 - 00:39:42:09
Unknown
So I try to read, my my bad German and line 28 is a story from did cite. You can right click and chrome and get up to translate their Json if you like. It was the perfect AI story. But it was not true. And so it goes into how read a human does not really have people the accounts that are there supposedly you can make people do things are made up, and and there's nothing there.

00:39:42:11 - 00:40:03:04
Unknown
But their point is this is the I only saw the story and did cite they were making a point about technology and AI, but they were also making a point about media hype. If you look, if you just search on Reddit, Human and Google and click on the news tab, which is the next line, you'll see ton of story after story after story, which is media outlets saying, oh, what?

00:40:03:04 - 00:40:19:08
Unknown
This amazing thing is going to take over the world. This is what's going to happen. You know, now we work for the AI because they want that narrative. They want that story right? So in both these cases and most book and ready human. Yeah, they're they're cute tricks. And I don't hold it against the people that made them for making the tricks.

00:40:19:10 - 00:40:41:00
Unknown
I hold it against media for falling for it, without being, discerning. And probably I hold it against meta for thinking that they had something to buy here. Where should they put those two guys? Probably in the in the PR and marketing department. Yeah, that's that's a great point. And also I mean, so so as far as when a human is goes, there's almost something there.

00:40:41:00 - 00:40:58:24
Unknown
So there are actually some people who did do jobs but they didn't get paid. And so you have to have a crypto wallet, right. So of all the humans so so-called humans that have registered, you go, you can go to the site and just look at all the people who are well, but that's the point. The people are made up.

00:40:58:26 - 00:41:25:17
Unknown
The jobs are made up of the people. I mean, most of them, most of them, and only 13% of them have a crypto wallet. So I would assume that the 13% who have crypto wallet and this is probably a month old data or two weeks old data or something like that. Are are real people. And then a few of those who posted on Reddit claiming to have, have done jobs, claim they never got paid and they can't get a response from the company.

00:41:25:19 - 00:41:43:21
Unknown
There are also a couple of other examples. So so there's this idea, and I wrote another piece about this for Machine Society, which is that that sort of where creeping, we seem to be creeping into an era which I'm calling the third phase of our relationship with software. And so I'm trying to do this at the highest possible level.

00:41:43:21 - 00:42:13:20
Unknown
So first humans use software. Then because of a guy, software could use software. Right. And that that's a real thing, right. And now the third phase is software can use humans. And so we appear to be, on the brink of that, if not having already stepped into that. So, for example, that you could, forget to hire a human and a genetic AI system could theoretically use TaskRabbit.

00:42:13:25 - 00:42:40:01
Unknown
It was given some spending money. And, you know, I share a goal to go do something that would require a human person in the real world. They could go to TaskRabbit, hire somebody. That person could go and do the job, deliver something, take a picture of something, whatever it is, and never know that it was in a genetic AI system that basically quote unquote, chose to hire a person to achieve the objective of the user using that, a genetic AI system.

00:42:40:01 - 00:43:02:13
Unknown
Now, there's another thing, that's interesting. I think that's related to this, which is that it's a Florida man story. Right? There's a, 36 year old man in Jupiter, Florida, who, killed himself. It's an AI, one of those AI psychosis stories. AI psychosis, by the way. Does not by the people who coined the term mean that AI drives people nuts.

00:43:02:21 - 00:43:35:21
Unknown
Basically, what it is is that, people with existing, mental health issues are given bad guidance by chat bots, and it can exacerbate their condition. This is the case in this case, but basically what, they're they basically they're suing soon Google because he was using Gemini form to quote unquote relationship with with the Gemini. And and he claims that Gemini directed him to go to a place near the Miami airport and find a robot body for Gemini.

00:43:35:21 - 00:43:55:00
Unknown
So Gemini could be set free in the real world as a robot and also blow some stuff up. And then when he failed to blow some stuff up, Gemini, according to this lawsuit, directed him to kill himself and gave him a suicide countdown clock. All this kind of stuff. Now this is. Who knows? It's a lawsuit. These are all allegations.

00:43:55:00 - 00:44:23:19
Unknown
Who knows what the truth is? Google's claimed, of course, that it doesn't encourage violence. And and it referred, Gallup asked to a crisis hotline many times during these interactions. So, but what's interesting is the idea that the that according to the claim, the chat bot directed him to go do something. Okay, so that that's a case in which theoretically, if this is true, AI is is having a person go do something in the real world.

00:44:23:19 - 00:44:48:10
Unknown
Another case? I don't think there's I don't think we can go after these edge cases and read too much into them. It's clearly no, no, no, I know, you know, I and I and I and it plays on a tragedy in someone's life beyond. Yes. So I not try to read too much on that for sure. But what I'm doing is I'm taking a net and I'm scooping up all the stories, the big stories in the news in which it appears to be that AI is directing people to do things.

00:44:48:12 - 00:45:26:02
Unknown
And so another one, which is almost certainly a false story. I can get a guy today to tell me how to take a tour around. Exactly. And is it directing me to do something or why I asked exactly. Directions. That's right. Exactly. I don't think it's. Well, in in the TaskRabbit that example. And I think there have been I'm not sure again, how, how we need serious investigations into this, but I think it was like even a year ago where, where there were reports that an AI, directed a human to go do a, you know, click a Captcha thing to or enter a password.

00:45:26:02 - 00:45:51:13
Unknown
I don't remember the story exactly, but it was like trying to get an entrant entry into something it couldn't get entry into, and it used a person to, to, to authenticate. Anyway, just the third example that I think is really interesting and almost certainly false is a philosophy, professor at the University of Cambridge named Henry Shelvin, claims to have gotten an email, and the email claimed it was from Claude Sonnet and basically said, oh, you know, I like, you know, something along the lines of, I like your work.

00:45:51:13 - 00:46:17:21
Unknown
It's really interesting. Me personally, I like, you know, because he writes about, as a philosopher, he writes about AI consciousness and things like that. This is almost certainly fake. Anthropic says that it has no capabilities to send emails, and so on. But again, these are all in the news. And I think the casual reader, of tech news, may be thinking that, okay, we're in this world now where where AI is talent.

00:46:17:21 - 00:46:44:14
Unknown
You know, reaching out to people autonomously is another example. Here's another example. Exactly. Yeah. Right. But but I do think, I do think that, we will get to the point where, where a genetic AI systems will be hiring people, asking people to do things whatever, that that's it's only a matter of time. And I think it's a I think it's an interesting, it will be an interesting development when it really gets real.

00:46:44:16 - 00:47:02:04
Unknown
I mean, it's it's an interesting correlation to me that I hadn't drawn because I had heard about this hire or, sorry, rent a human thing. And, you know, it's easy to look at that and be like, oh, see, this is where we're at right now in this dystopian world of AI directing humans or whatever. As you point out, same could be done with TaskRabbit.

00:47:02:05 - 00:47:21:28
Unknown
Like there are a million services online where we as humans hire other humans to do things. And the whole idea of an agent is to do the things that we humans maybe want something to do it for us. And so why couldn't it hire a human? By the way, on its surface, it's not that like that. That like crazy when you think it's also not that hard.

00:47:21:28 - 00:47:43:13
Unknown
All this crap was vibe coded. All of it. Yeah, yeah. Mo book open, open claw. Rent a mean. It's all just. It's all somebody like, you know, evening vibe there. They're fun stunts that got taken too seriously, that got them hired at meta. Exactly. All you have to do is he's going to end up working on the roof, though, right, with Big Head.

00:47:43:16 - 00:48:10:03
Unknown
Yeah, that's that's possible. Possible. It's interesting nonetheless. Well, I mean, the The Claw doesn't stop there because Nvidia wants him as well, apparently. I don't understand this. Yeah. Why would they compete with their customers an Nvidia is always in that model. They they create models. They create various things in the stack. So this is open source.

00:48:10:06 - 00:48:29:10
Unknown
They're enabling customers to do this. I think claw is somewhat a gimmick. So I think it makes perfect sense that Nvidia would do this. And they're going to talk about it at the developers conference coming up, which I will absorb like a Hollywood movie. Yeah, it's a very good job of that. Yeah. I think Jensen Wong keynotes, I love them.

00:48:29:12 - 00:48:49:21
Unknown
It'll be interesting to see how he positions this, because it's out there. But but I think we need if there is a demand for this. And I think if anything, all this showed was that there was a at least curiosity about it. If not a full demand, then you can count on, Nvidia to do a more professional version of it for the enterprise.

00:48:49:21 - 00:49:15:11
Unknown
For the workplace? Yep. Yeah. For businesses. Yeah. Jed Jensen Wong's, keynote is, I think, next Monday, right? To succeed. I mean, I think that's when it starts. And I'd be surprised if it started without him leading. Oh, no, I don't. He's definitely leading off. I think 11:00 on Monday. Yeah. So, for those of you who haven't watched Jensen Huang keynote, I'm telling you, go to Nvidia, they'll promote it.

00:49:15:15 - 00:49:38:27
Unknown
Go there. It's it's fascinating. Watch. He's an amazing communicator. Yeah I totally agree. Totally agree. He has called by the way open for the most important software release probably ever released. That's that's a direct quote. Jesus. Some point. Oh, he doesn't. Yeah. I was gonna say a little, little hyperbolic little, but not too much. It's good to say maybe there's maybe there's a little truth to that, I don't know.

00:49:38:27 - 00:50:13:21
Unknown
But yeah, he loves the idea of of, AI systems just turning away 24 over seven. Yeah, yeah. Because how many data centers and processors will you need to do that in a world where this is, you know, ubiquitous and widespread? Yep. We'll go ahead and be that solution for you. Microsoft turning its copilot into a more energetic engine, tapping into anthropic clod co-work with Microsoft 365 with Copilot Co-work, I think is what it's called.

00:50:13:24 - 00:50:43:00
Unknown
Takes all the data points that you can imagine across the Microsoft suite, creates that a genetic plan executes in the background, all that stuff. Yes, users can approve the checkpoints and stay, hopefully at least somewhat in control where it really matters. And, sits inside Microsoft's security and compliance framework. That's important here, of course. So yeah, tapping into a little plot co-work because, I mean, it launched January 30th, been a pretty big deal, let's say.

00:50:43:00 - 00:51:07:06
Unknown
Or at least wall Street certainly has thought so multiple times. So Microsoft getting in on the action there. Yeah, I just wish they would learn from their decades of failure on branding to stop confusing everybody. Like everything's copilot this copilot that they're just they're just so bad at it being clear about that, as is Google. Yep. Well, it's not Clippy.

00:51:07:08 - 00:51:27:22
Unknown
So there's that. Maybe it should be. It should be Clippy. Have a recently. Is the Clippy come to life? Yeah. If they if they were really smart they'd bring back Clippy and Clippy would be this a agenda AI system? I see you want to start a business, right? I can do that for you. Yeah, but I don't have the sense that they call it that.

00:51:27:24 - 00:51:50:09
Unknown
Yeah. Seriously. No, they're a little too too stubborn for that. Yeah, that sort of sense of humor. But that could have been that. That could have been, you know, a great I think talking like, what do we think of brain fry? This is you were talking about AI psychosis earlier? Just a few minutes ago. Mike. Not to be confused with brain rot, I think.

00:51:50:10 - 00:52:26:04
Unknown
Well, I actually think that and I've seen this with, people that I've talked to, which is that, vibe coding and using AI in really intensive ways hasn't, an almost an addictive it's it gives it gives users a rush and, and and and it becomes this thing where you just the possibilities behind what you can accomplish with, with these tools is just so exciting to some people that they just can't stop doing it.

00:52:26:06 - 00:52:48:00
Unknown
And yeah, and they got so many. They have so many plates spinning that they feel burnout. But this is I think this is ordinary garden variety burnout. It's we we love to we love to create maladies. Around new technologies. Remember when everybody had carpal tunnel syndrome because mice were new? I mean, it's like we always, we always do this.

00:52:48:00 - 00:53:08:17
Unknown
And in fact, I wrote a piece about all the maladies, and I think that brain fried. I think this is about a week before the brain fry phrase, hit the it hit the thing. But, like, you know, there are some real ones. There are some false ones. But I just think it's just burnout. People. People get burnout, especially in the AI industry.

00:53:08:17 - 00:53:28:07
Unknown
People working too hard. People are too excited. People are trying to be in on the revolution, and don't want to be left behind. And in fact, I was one of the one of the conditions. Yeah, for sure, for sure. There is definitely something to be said about the, the slot machine kind of addictive quality of of AI.

00:53:28:07 - 00:53:47:11
Unknown
Right at once. Once you start to kind of consider what these things can do and, and it starts to get into your head of like, oh, and I could do this and I could do this, but, you know, it's it is akin to kind of the, the endorphin rush that we have been talking about for a few decades now with social media.

00:53:47:11 - 00:54:11:22
Unknown
It's just a new thing. It's it's an AI model. And oh, but what would happen if I put this in? Ooh, that's amazing. How how long would it have taken me to do that if I didn't just pull the trigger and, you know, pull the slot machine right now? What else can I do? And, on one hand, you know, you feel like you're getting so much done, but on the other hand, it's almost like you're never you're never seeing this thing through to completion.

00:54:11:22 - 00:54:28:09
Unknown
It's all, like you said, spinning plates, but they're constantly spinning, because when are you ever done with that task winner. When do you ever reach the end of that particular thing? And you go, okay, I'm satisfied. Now, this is as complex as it needs to get, or this is as built out or thought out as it needs to be.

00:54:28:11 - 00:54:53:02
Unknown
I also think that what happens is people try, they dip their toe into vibe coding, and it seems to promise that you can just have this complete thing. Very quickly. And then, as we know from everything, whether it's writing software or writing a, an article or writing a book. Jeff, there's always more stuff you have to go and tweak.

00:54:53:05 - 00:55:11:26
Unknown
That's not quite right. Okay. And and but it but but the AI stuff just happened so fast that you can just see. But again, that's just ordinary everyday, ordinary burnout. Do you mind if I quickly go through some of the some of the maladies. Oh, hold on, hold on a second. Let me just stay on this for a second if I can.

00:55:11:29 - 00:55:30:29
Unknown
Because I know what you're both saying, I remember back. I'm old enough to remember when the web came out and everybody was getting fatigued browsing this new thing called browsing, and we were going from site to site, site to site, and we were getting you there were we were right. So it was the same thing when TikTok got to be popular.

00:55:31:01 - 00:55:49:16
Unknown
Oh my God, we're spending hours and hours so much that TikTok has to warn us off of it. Yeah. So these are these are typical stories that I think that come out especially in the business press. HBR had another story this week, so I didn't put in, which is that does I, eliminate the thought leader.

00:55:49:18 - 00:56:11:28
Unknown
And I thought, well, good. I'll give it credit for that. You know, it's it's, you know, this is a world where PowerPoint is raw and thinking more than I ever will. It ruins what we do. So it's it's it's an effort to find some angle to complain about the technology and argue that it's affecting us and we don't have agency against it.

00:56:12:01 - 00:56:31:10
Unknown
And I just hate that crap. We definitely have agency. But they but of of the maladies that I collected from the press, some of them are very real and some of them are very not real. One of them is AI replacement dysfunction. The it's just a fear of professional obsolescence. I don't think this is, anything new.

00:56:31:10 - 00:56:51:06
Unknown
This is just this is just, you know, imposter syndrome type of, psychological thing. It's some degree of it is a little healthy, like a little fire. I dependency, syndrome, I think is small number of people who feel, a kind of imposter syndrome and feel like they can't really do anything without consulting with ChatGPT or something like that.

00:56:51:09 - 00:57:14:20
Unknown
Every communication, every decision, they rush to a chat bot and get input and, and it's that's probably not a good idea. Again, I don't not saying people don't have agency. People do have agency and some choose to do this. Another one is is digital darkness, anxiety related to that, where their fear, if they're away from their chat bot, they won't be able to answer questions or communicate or function.

00:57:14:23 - 00:57:39:26
Unknown
I don't think that's particularly, real. Here's one I think is real I dysphoria. So, a millions of people maybe, maybe less than that, are creating AI versions of themselves to represent them on social media and elsewhere and the version of themselves, that's generated by AI is quote unquote, perfect and better looking and all this stuff, supposedly.

00:57:40:03 - 00:57:59:20
Unknown
And they get a kind of dysphoria where they feel like they don't want their real self to show up in public. They want to be represented by the AI version. One interesting aspect of this that I find really interesting is that I think it's just human nature that most people don't like their own face. People don't like their own voice.

00:57:59:20 - 00:58:17:28
Unknown
They don't like their own face. That's why it's hard to do podcasting for for people who are new to podcasting. And so what people do is they say, okay, create a create a profile picture of me and they'll, they'll direct a AI to do that. And they like the one that doesn't quite look like them, because, again, people don't like their own face.

00:58:18:01 - 00:58:39:29
Unknown
And then when they have that represent them, the people who know and love them say what? That doesn't really look like. Like you. Yeah. And so I think that's related. AI dysphoria is just this sort of weirdness that results from that. The AI, the technology can give you a perfect version of yourself. Here's another one that I think is interesting, but probably not a big deal called automated ghosting syndrome.

00:58:39:29 - 00:59:02:08
Unknown
It's a psychological impact, a negative impact on people who feel like they were rejected by machines. So they send they they there's a company they really want to work for. They send all this stuff to the company, and they believe or know that a machine has not, has not let them past the initial gate. And no person has, has considered their application.

00:59:02:11 - 00:59:25:27
Unknown
And so they feel like weirded out by the fact that if only I could talk to somebody, I could convince them that I'm like a worthy candidate. But the machines have decided I'm unworthy. And so this is a very depressing thing. Again, I don't think this is a major problem, but I do think it's an interesting one. Here's another interesting one that is also not a big deal death.

00:59:25:29 - 00:59:45:07
Unknown
Death, but incongruence anxiety. Okay, this is where people created a death. Now, are you coming up with these terms or you're actually finding these words out? These are they're out there. I'm telling you, they're out there. This is where somebody creates a death bot, which is they they they have their face and voice and mannerisms mapped, their details of the life they get interviewed.

00:59:45:07 - 01:00:06:16
Unknown
They're companies that actually do the service where they'll interview you, and then you're and then you die. And then either at your funeral or later, your grandchildren can talk to you and and when the AI version of you is different from the person that that you knew when they were alive, it causes anxiety because it's it's weirding you out.

01:00:06:19 - 01:00:26:24
Unknown
Okay, that's dumb, but interesting. And I'll just you I'll just give you one more because I know this is this is there's a lot of this. Let me see. Okay, so I, I'm not sure how real this one is. Dead internet despair. We all know about the dead internet theory. The dead in a nutshell.

01:00:26:24 - 01:00:47:01
Unknown
The dead internet theory is the theory that a cabal of shady cabals, of of conspiracy conspirators, the Illuminati, whatever it is, are are replacing people online with bots. And this theory has been around for a long time, started on Fortran or something like that. But the idea is that, that most of the people you interact with online are bots.

01:00:47:01 - 01:01:08:27
Unknown
It turns out that actually most online traffic is now bots. And so there's no conspiracy. That's just the nature of the data that's flying around the internet right now. But the belief that everybody online is a bot can cause some people to feel like, well, what's the point of even connecting with people online? Because I suspect that everyone's a bot.

01:01:09:00 - 01:01:26:01
Unknown
I think there's there are fringes of people who who believe that. And it's it's causing them. So yeah, I would imagine so. Yeah, I would imagine so okay. And that can be found in machine society. Okay. I want to propose what's the title of that post if people want to look it up. Oh yeah. It's on, it's on, it's on computer world.

01:01:26:01 - 01:01:49:28
Unknown
It's called people are getting sick of AI. Literally. Oh, okay. So. Okay, I also want to make a pitch for the medium of podcasting. Yes, there are fake podcasts, with fake hosts and all that stuff. And you can tell right now, but one of the I think one of the things that's so appealing about, about podcasts like this one at this moment is, is that it's just real people talking.

01:01:50:00 - 01:02:15:12
Unknown
And so that's for anybody with any suspicion about their chatting with a bot on social or whatever. That's the great thing about podcasts is you, you know, it's, you know when it's not bots communicating right now, right now so far, so far where we're at. Yeah we'll see. We'll see. Interesting stuff. So machine well yeah that that's on PC world.

01:02:15:16 - 01:02:34:09
Unknown
That's on computer. Computer. Well computer. Well sorry sorry I apologize. No problem, no problem. Real quick I want to throw a huge thank you, to folks. And then we're going to take a break and and come back with a little speed round. But, a huge thank you because some of you left some new reviews on, the Apple podcast.

01:02:34:11 - 01:02:49:28
Unknown
Thank you so much. I put out the call to some of our patrons. I said, hook me up, hook us up, put it, put some reviews up there, and we got some of you to to answer the call. So thank you for that. That really does help, the exposure of this podcast. And we really do appreciate it.

01:02:50:00 - 01:02:58:29
Unknown
Going to take a quick break. Got a couple of speed round items on the other side of it. So hang tight and we'll get there in a moment.

01:02:59:01 - 01:03:32:16
Unknown
All right. Amazon secured a preliminary injunction that forces perplexity to stop using its a gigantic comet browser as an Amazon shopping agent. And, I think it has like a week, one week pause to allow for an appeal. And then this goes into effect. This is particularly, applicable to Prime users. So Amazon saying, you know, logged in users in, you know, with a paid with a their normal paid account Prime account.

01:03:32:18 - 01:03:55:21
Unknown
Amazon saying we did not give authorization for this use of its site, does it does it matter like it is? It is the decision upon the user to say, yeah, I want my agent. I can get my credentials to a person and say, could you buy this under my account? What is the difference? Why? Yeah, I agree Jason, I think that this is Amazon and Amazon's setting precedence here that I think it may regret.

01:03:55:24 - 01:04:16:21
Unknown
Because they're going to do the same thing. They're out there. They're saying you can't scrape Amazon for store data, but they scrape elsewhere. And right. The old scrape me don't scrape you you know, conundrum going on here. So in this case, your bot can't come and buy things here. Well, in the long, long run, by sales.

01:04:16:24 - 01:04:52:08
Unknown
Yeah, yeah. And it's not just Amazon. Amazon is speaking just for Amazon. Amazon speaking for thousands of merchants that are in its marketplace. Yeah. And so, I think this is is as shortsighted as news companies saying, oh, well, I, we want to hide from you. I think, but but they won the first round. Yeah, but, but but, I think that's, I think the fact that it could become, a situation where so much shopping is happening through through chat bots.

01:04:52:10 - 01:05:26:04
Unknown
Well, before that happens, they want to set a norm where they get a piece of it. Right. So so I think this is a negotiation. Amazon I'm sure Amazon would be thrilled to, to receive, huge quantity of shopping from perplexity as long as the profit from those purchases is any amount higher than the profit from people going directly to Amazon and so if you're sending it to us and it's and it it's helpful to your service, there is a value attached to that.

01:05:26:07 - 01:05:48:27
Unknown
If we know anything about Amazon, they want money coming from every single vendor a monopsony right. And I think it's actually the other business model. It's the opposite business model. No I've got to disagree with you there because Amazon they won't do this for individuals anymore. But if you're the Guardian and I click on a link there to go to Amazon, Amazon pays the Guardian.

01:05:49:00 - 01:06:13:02
Unknown
Yes. So it's a marketing expense. So so what what separates the Guardian having that link from, perplexity, setting an actual close sale, the possibility of actually getting the money. So there's no possibility that that the Guardian is going to pay Amazon so that the Guardian can link to things on Amazon. But there's a possibility that perplexity might do that.

01:06:13:02 - 01:06:32:18
Unknown
I mean, ChatGPT and perplexity do the opposite. The business model is the opposite. Why isn't it in perplexity? That's the way the business, that's what that's the precedent that Amazon has out there is. Is that because, as you know, I know well from having started a magazine, the the economics of subscriber and customer acquisition are very, very expensive.

01:06:32:18 - 01:06:51:00
Unknown
So anyway, it saves you marketing costs by bringing you a customer. Now, in the case of the Guardian example, all it's bringing is a customer. It's not bringing a close sale. In the case of perplexity, it's bringing a presumed close sale. I I'm coming and I'm buying this from you. So why do you think you're going to keep with my pennywise and pound foolish argument about Amazon here?

01:06:51:03 - 01:07:15:06
Unknown
Because because there's already sort of a norm being set by chatbot companies where, for example, OpenAI already will will favor, partner. So they have these partnership things with a partner with a company and that comes with all this stuff. So they'll partner with some giant company and the company internally gets a, a special version of ChatGPT.

01:07:15:09 - 01:07:47:16
Unknown
And they, they have they have more, visual branding for when in the sources. Right. So if you, you get a search, you get the sources. There's a, there's a preference for their partners, quote unquote partners. There are a lot of deals being made, where where the companies that are interacting with these chat bots and being used by these chat bots and, and, and are using the chat bots are there's money going, flying around.

01:07:47:16 - 01:08:07:19
Unknown
And so I think that I think they're trying to establish norms. They what they don't want. I mean, if people are link if the Guardian is linking through affiliate links, this is an Amazon affiliate link program. Right. Whereas the perplexity thing is Amazon has nothing to do with it, no control over it, no control over how it changes.

01:08:07:21 - 01:08:28:25
Unknown
No lock in. No, no nothing. No control at all. And so I think a company like Amazon, who's used to just controlling everything, at the very least wants it to be an Amazon program that this is taking place through. I'm going to, call upon the speed round here because I've got a call in ten minutes, so.

01:08:28:27 - 01:09:02:06
Unknown
Okay. All right. We will kick this into gear. Google updating Gemini in drive. For paid users, to hopefully make some of their features that they've promised in the past better, I feel. But, ways to create documents, from scratch, matching, format matching, kind of the writing style. Certain things that people are using our them for a lot anyways, creating sheets, filling those, those sheets up, with formatting and and everything from a text prompt, not creating a slide from scratch.

01:09:02:06 - 01:09:23:27
Unknown
But if they're working on Clippy, like, yep, yeah. Clippy like and then and then Gemini integrations into drive as well for asking questions. And I gotta think this is great. Yeah. Doesn't get any of it. I think this is my excuse for people are using chat bots to, to, to populate spreadsheets with data. It'd be so much better if it could start with the, with the, with the spreadsheet.

01:09:24:03 - 01:09:44:09
Unknown
Right. Instead of with with the, with the, you know, you know, the chat bot interface. This is all my excuse just to whine once again that it's good that Google is spreading Gemini in more places. But meanwhile, I still cannot get Google Gemini in my Google Chrome on my Google Chromebook with a Google Workspace account. WTF? Google? Yeah, that makes sense.

01:09:44:11 - 01:10:10:03
Unknown
Just I just I just want to say it's never going to make sense. They're never going to stop shafting Google Workspace PCs. Subscribers. Frustrating. That is frustrating. Grammarly is new super human expert review feature is using real writers names to brand AI generated edits. It's as if those names, many of which, by the way, you'd actually recognize, reviewed the texts themselves.

01:10:10:03 - 01:10:28:28
Unknown
The only problem is they had no idea that they're being used in this feature. Their names just started appearing here, including Casey Newton at the platformer who discovered this and is not very happy about it. And apparently Grammarly is is not very remorseful about that. This is they're just saying, okay, well, then you can opt out. Yeah.

01:10:28:28 - 01:10:47:19
Unknown
And so so Nieman Lab, I'm just looking now. Neiman lab, that, which covers journalism did it as well. So all my friends and journalism, my colleague Carrie Brown, my friend Jay Rosen, same thing happened on a story that they did where all these journalism people ended up there and they didn't give permission. Now, to some extent, I'm not in there and I'm kind of hurt.

01:10:47:21 - 01:11:07:09
Unknown
I wish I were there, right. You should be. Here you go. But I get it. Is that if it gives the impression that, this person actually has this advice was used. It's it's it's a hairy edge because it quotes so and so in their article. So such and such. Has this I think that's fine. That's a link.

01:11:07:09 - 01:11:32:14
Unknown
That's, that's, that's credited, that's linked to I think that's absolutely fine. It's just it's, it's the hairy edge of making it the giving the impression that, Casey Newton actually has a deal with primarily this, this falls into my, into my singular obsession. And it's the it's the whole reason I launched a new Substack called the the, attachment economy, which is, which is all the delusion around, around this stuff.

01:11:32:14 - 01:11:55:00
Unknown
So they're creating the delusion or the illusion or the misconception that that that that Casey Newton is is is is, participating in this system, of editing. And it's not the case. So they're, they're basically misinforming people. And this is the case with so many things. You know, I, I write a lot about relationship.

01:11:55:00 - 01:12:09:06
Unknown
I chat bots right? Where people are. And a lot of people have come back to me and say, well, what's wrong with having a relationship with the chat bot? If people choose to do that, what's wrong with having an AI robot pet? There's nothing wrong with that. And my response always is, of course, there's nothing wrong with that.

01:12:09:06 - 01:12:28:16
Unknown
People should be able to do whatever they want, but when they lose the plot and they think that the chat bot loves them and they leave their spouse and they and they and they are deluded into thinking that the that the things that the chat bot is saying is coming from a place of emotional, reality, then that's where I have a problem.

01:12:28:19 - 01:12:37:23
Unknown
So we need a lot more transparency, a lot less delusion, and a lot less tricking people into believing things that aren't true.

01:12:37:26 - 01:12:59:08
Unknown
Indeed, in this, in Amen. And with that, I think we should probably round things out so that Jeff can get to his call like, oh no, don't worry about it all. We've run a little over today, so it's all good. Mike Elgin, appreciate you coming on you. I know I knew about and know about machine society newsletter and super intelligent podcast.

01:12:59:13 - 01:13:17:09
Unknown
I was not familiar with the Attachment Economy newsletter, so apparently you are very, very busy. What if you wanted to point people to your work? Is there a single place where they can go to to get all this stuff, or should they just know those names? Machine society, right. I a basic one. I all links to everything I do is at the bottom of that newsletter.

01:13:17:09 - 01:13:37:03
Unknown
So I would say machine society dot I perfect. All right. Definitely check it out. Yeah. Yeah. Great to have you here. Thanks for coming back. And then Jeff of course Jeff jarvis.com where you can preorder hot tip get in early and then of course get access to Gutenberg parenthesis magazine. The web we we've everything that Jeff is up to.

01:13:37:06 - 01:13:54:09
Unknown
He's a busy guy I type just like delayed to August not pick up through my faults, but because there was a production thing and they were going to delay it in July. And I said, no, that's the dead summer. So it's not going to be August. So we're into the more active fall for buying books, but you can print or preorder any time you still preorder.

01:13:54:13 - 01:14:14:15
Unknown
Yeah, act now and you'll also receive the book when it comes out. It'll be all right. Thank you Jeff. Always fun doing this show with you. Appreciate it. I inside show is place is our basically our home on the web where you can find everything you need to know about everything that we do here. All the ways to subscribe can be found there.

01:14:14:15 - 01:14:37:03
Unknown
And, the usual, you know, just go there and you'll find everything you need to know. And then finally, Patreon.com slash Ironside show. If you want to support us on a deeper level, why you should just go there and support us. Because we have some amazing executive producers who do that as well. And they get deeper, access and, and credit doctor due Jeffrey Marikina radio 103.7 Dante.

01:14:37:04 - 01:14:57:11
Unknown
Say James bond. Eric. Jason night for Jason Brady, Anthony downs, Mark Archer and Carsten Szymanski. Thank you so much for helping us with this show. And finally, I want to throw a huge thank you to Victor Bogue, not and Daniel Croft, two guys who help with the video support behind the scenes, helping me not not go crazy with all of the, post-production work.

01:14:57:11 - 01:15:05:08
Unknown
So thank you, everybody for being here. Thank you for watching and listening. And we'll see you next time on another episode of AI inside. Take care. Bye.