What About... An OpenAI Bubble?

00:00:00:00 - 00:00:27:28
Unknown
Coming up next, Jeff Jarvis and I talk John Larkin's rumored exit from meta to launch his own AI startup focused on world models, SoftBank's billion dollar Nvidia sell off to fund new AI bets, and why history might actually be repeating itself. Anthropic contrasting path to profitability while OpenAI spends really big betting on its future, and the first AI generated country music hit to top the Billboard charts.

00:00:27:28 - 00:00:33:01
Unknown
Yep. It's happening. That's all ahead on the AI inside podcast.

00:00:41:14 - 00:01:07:02
Unknown
Good afternoon, everybody, and welcome to another episode of the AI Inside Podcast, the show where we take a look at the AI that is layered throughout so much of the world of technology. Our weekly show where we really look more closely at, like the news and kind of the top news stories that Jeff and I feel are kind of bubbling up to the surface and shaping the direction of, the development of AI and everything.

00:01:07:03 - 00:01:32:17
Unknown
I'm one of your host, Jason Howell, joined, of course, by Jeff Jarvis. Good to see Jeff. Hey, boss. How are you doing? Well, it was kind of funny right before showtime, as we were getting everything set up, I looked down at my desk, and I. And I made the realization, that I have one, two, 3456789 1011 1213, 14, 15 smartphones right now, you know, oh, which 1 a.m. I going to use today?

00:01:32:19 - 00:01:52:11
Unknown
This is this is crazy. Like this is it's crazy. And yet for my life it's normal. Like I'm so used to being surrounded by people change shoes, ages, phones. Apparently I, you know, I just have different reasons for needing different phones at a, you know, a hold up. The one you mind holding up, the one you showed me before the preview for video.

00:01:52:12 - 00:02:14:04
Unknown
Oh, we're on USB. Yeah. This is the, sooner the the very first Android device before it ever went, public, before they ever released of an actual device. This was their, like, earliest prototype, at least that we know of. And, it's kind of a legendary device because it looks like, a BlackBerry or it looks like a trio.

00:02:14:04 - 00:02:36:26
Unknown
I keep telling you what trio. Trio? I was a touchscreen has a really early version of Android. Very early version. And, yeah, I'm kind of taking a very close look at this. Was the phone that existed before Steve Jobs did his initial iPhone, keynote and changed everything. And my understanding is that, Oh, God, why am I suddenly blanking?

00:02:36:26 - 00:03:01:26
Unknown
And Andy Rubin, who developed Android initially and then came over to Google to work it out, my understanding is that right as that keynote happened, he's he's on record or at least written about saying, oh crap, that changes everything. And he immediately they immediately changed course and started working on the T-Mobile on on the, you know, like a touch screen version of Android, with the launch device being the T-Mobile G1 or the.

00:03:01:26 - 00:03:22:20
Unknown
I think the HTC desire is what it's called. So yeah. So anyway, he said he's it has various lives, of course, a god of Android and and phones and stuff. So do watch his other stuff as well. Android faithful my channel. I just search Jason Howell on YouTube. I talk about a lot of this stuff just because I've been on Android since the beginning, man.

00:03:22:20 - 00:03:48:12
Unknown
So that's why I'm surrounded by phones. It's like they they're my they're my gang like triples. Yes. Totally. So yeah. So that's that's a different life on this show we talk about our, our artificial intelligence life, I guess. Why don't we talk about friend of the show, Yann LeCun. I say friend of the show because Yann LeCun was a guest a number of months ago on this very show.

00:03:48:12 - 00:04:05:21
Unknown
If you have not seen the episode, definitely check out. You watch the Jeff and I had. It is fantastic. We'd love to have him back some time and maybe we'll have reason to have him back, because not that they're not that we need a reason. We'd have him any time, but apparently he is reportedly getting ready to leave.

00:04:05:21 - 00:04:37:16
Unknown
Meta where he has worked and where he was working when we had him on for our interview with him, preparing to launch his own AI startup. Honestly, I'm kind of surprised that we hadn't heard a little bit about this sooner because he's such a, you know, such a predominant mind in the world of AI. And, you know, his focus and much of the conversation that we had with him was about world models and the need for world models in AI, so that those systems can simulate real world environments and predict real world outcomes, that sort of stuff.

00:04:37:18 - 00:05:02:03
Unknown
And, reportedly, he's been fundraising at least this is according to anonymous sources to the Financial Times. And we could see an exit coming from Yann LeCun, out of meta and into his own thing sometime in the coming months. Who knows when that will actually be? But there we go. So I yeah, like you, I just jumped on this news and it's a Financial Times report and it doesn't say much except that he might leave to start something new.

00:05:02:03 - 00:05:30:16
Unknown
And I'm I'm dying to know I would kill to know, what it is that he's working on. We can guess in a few minutes. But I often quote Yann LeCun on the show. I think that he is the smartest and sanest person I know. Leading AI. There's a lot of different people, you know, there's the there's the kind of doomsayer AGI people and the young looking isn't doesn't believe that AGI is possible.

00:05:30:18 - 00:05:56:18
Unknown
But he's just he's a lot more measured and sane about it and tied to reality. And unlike some really like, Sam Altman is not an engineer. Yann LeCun is a professor and, an expert in computer science and understands this deeply. And because he's an educator, I think that he can express it more clearly than some of the others who were involved in their companies.

00:05:56:20 - 00:06:21:06
Unknown
So I'm, I'm just I just can't wait to hear what he's working on. And tie into this, surely. What have been the politics at MIT? A because, yeah. You know, Zuckerberg hired the the boy genius to had I, Yann was always off in a, strategic laboratory for AI. He was looking at foreign strategy about AI.

00:06:21:08 - 00:06:34:18
Unknown
Right. And he wasn't doing the the, I think so much the the close and the product work. So people thought, well, what happened to you on the horn when hired? I'm sure some people will get this and say, well, he had to leave because other people were hired and maybe there were internal politics and maybe all that.

00:06:34:18 - 00:07:00:01
Unknown
But I've anticipated for some time that he was going to go off on his own. It's kind of felt that way because it has seemed, at least from the outside, looking in and different reports here and there, that there have been some disagreements within meta about it. You know, the direction that it has really been investing in and focusing in and increased, you know, focus on its on how on how it's developing its lead products.

00:07:00:03 - 00:07:25:23
Unknown
And, you know, kind of kind of has seemed like Yann has gone public with I don't know, perspectives that seemed to go a little counter to the company line within meta, which maybe, maybe isn't always necessarily a bad thing. Right. You want someone, hopefully you want someone on the inside challenging your assumptions so that you get to a strong, you know, grounded place on the other end of that.

00:07:25:23 - 00:07:47:14
Unknown
But it has felt at times that he's his perspective, and his opinion is maybe going in a different direction than what meta seems to be developing towards. Yeah, a few important things about, you know, Lincoln's worldview, number one, that he's been a very strong backer of open source. And, and, and I think that's why, llama had open source models.

00:07:47:14 - 00:08:23:16
Unknown
I hope that they stick with that. But but we've talked about which is that's still a strength like a commitment. WOB seems to wobble a bit on that. So we'll see. But I think of open source is critical because it enables development. Entrepreneurially. And educationally. Second, he has argued, I shared, I think I was probably about two months ago, a very long PowerPoint he gave to students and alums at NYU, where he outright said that alums have hit the wall, that if you really want to work on the future of AGI, all of us around the place have gotten about as far as they're going to get.

00:08:23:19 - 00:08:48:00
Unknown
And I think that's really important as as others are throwing just in huge capital expenditure, trying to just throw more power at labs as they stand, he's looking at the next level. And he has argued strenuously, as he did on in our interview with him, that that next level is real world models. And in that he is in sync, in harmony with Jensen Wang, and others.

00:08:48:02 - 00:09:10:16
Unknown
Versus the scale for scale sake argument. So I've got to believe that whatever he's going to do is probably going to be some paradigm well past Liam's, potentially into, real world models. That's just a guess, a mod. And there's, there's maybe a few clues here. I'm going to sneak in another paper. Jason. Generally, yeah.

00:09:10:16 - 00:09:37:23
Unknown
I make my papers in later, but I'll sneak in one. And now I love a paper by a dozen authors, including Yann LeCun and Lee FFA. Cambrian asked towards spatial super sensing and video. And it's a long, detailed paper that's trying to use, video to see what, I can sense within that can understand within that.

00:09:37:23 - 00:09:55:00
Unknown
So one video, just for example, I didn't get to read up on this, but they had a video going around the room and it's how many chairs are there? And as it goes in circles around is it going to recognize when the video circles around and see the same thing again? Is it going to know what a chair is is going to be able to to count them and separate them?

00:09:55:06 - 00:10:31:00
Unknown
Those are the kind of real world things that he's working on. And note again that that leaf A is coauthor of this paper around with Rob Fergus, also at NYU and others. And then there's another announcement that just came out that, DFA is doing often in, in American media is called Feifei Lee, is doing another her own startup, which is, which is related, marble or she's already done the startup, but this is the first product, the first model race with marble.

00:10:31:00 - 00:10:53:09
Unknown
It's a commercial product that where she says that you're going to create, you're going to take a photo, a video or anything else and then create a world around that, a 3D world from a simple prompt. So this all goes to the real world bottle question can you figure out what's actually happening in real life? Because we live in 3D, thank goodness.

00:10:53:11 - 00:11:26:25
Unknown
And the matrix and, how to reason it is a form of reasoning. I think what's going on in that world to understand the, the rules of that to, to, apply those. So that's where the two of them have gone in one paper together and in one separate announcement. And so, if I'm just looking for tea leaves here because I'm really interested in what Yann LeCun is going to do next, but I've got to believe in something in a 3D real world view.

00:11:27:01 - 00:11:43:28
Unknown
So that's as much as I know. I don't really know anything but a little speculation for you. It's so easy to see things like this. And I haven't read through the Feifei Lee, World Labs article that you put in here, but it's so easy to look at this and just immediately think video game. But this really goes beyond that, right?

00:11:43:28 - 00:12:20:05
Unknown
It's certainly like that's certainly an application. You can see how this would be really useful in that. But that's I feel like in the long term that's really scratching the surface. Well, this is where I always also call up my my obsession with Jensen Wong's keynotes, because he goes on a great length about, digital twins and, to be able to recreate a real world environment through factory, warehouse, car, on street and so on, and understand the laws of that real world so that you can then act on it appropriately.

00:12:20:08 - 00:12:42:15
Unknown
That's where there's going to be a tremendous amount of work, in AI. And I think that is the next level. And if such a thing as AGI or something approximating it becomes, it's going to require that kind of diligent work, rather than the kind of 1 to 2 dimensional world of text an idea.

00:12:42:17 - 00:13:05:20
Unknown
So does does a story like this about Yann LeCun, being made public through, you know, anonymous sources or whatever. Does that move the transition. Like I'm just curious your opinion on this. Does something like that move the transition faster, or do you think this is something that meta probably already knew about? Yann LeCun I can't imagine he's been such a good executive at meta.

00:13:05:20 - 00:13:28:13
Unknown
And of course, don't forget he also has another life as an NYU professor. Yeah, yeah, he's a mature executive. He just got awarded by the King of England. He's been awarded by the president of France. He's a world renowned AI expert, and he's not going to screw with his reputation in those worlds. He's a very mature, academic and executive.

00:13:28:13 - 00:13:47:11
Unknown
So, no, this had to be known and had to be negotiated. And I think it's a little bit of a leak, you know, and meta has been, let me just see what their stock is doing lately. It's down 2% today. Looking over a month. It, you know, it's taking a bit of a dive from 751 down to slow.

00:13:47:12 - 00:14:08:11
Unknown
That's a big dive. 751 but the six year team, for sure. Yeah. And what was the drop related to the drop has been that at the same time that the market said, whoa, gee, we just woke up and realized Google's doing a lot of good stuff with AI, right? Right. And we woke up and said, oh, Amazon's doing neat stuff with AI.

00:14:08:19 - 00:14:31:24
Unknown
They looked back at meta and said, as we've been saying on the show for quite some time, they look desperate. And, and they're flailing a bit. And so Mark's throwing huge money after things. But the strategy is not clear. The, competitive support is not clear. By that, I mean, Google knows all about you, and you're already using Google in all ways.

00:14:31:24 - 00:14:51:28
Unknown
And that's where they're going to fit in AI. Amazon has AWS, and Amazon is working with all the commerce you to do, and that's how they're going to fit in AI. Apple, if they ever get their act together, they know you because you're a prisoner to their ecosystem. And that's where they'll fit in. I what is benefiting AI into weird glasses and party pictures?

00:14:52:00 - 00:15:17:07
Unknown
That's an issue for meta, I think even though because of Yann LeCun. Meta. Great measure because of you on the current meta was a real leader, I think in the early days of the dissemination of generative models, llama was really important. Llamas being every university uses llama, tons of things are built on top of llama. In a way, meta was positioned to almost be.

00:15:17:10 - 00:15:42:27
Unknown
It was almost an Android like model. Yeah. To play a part. And but then I think Mark showed some level of desperation. I don't think he's shown a clear strategy there. I never count out Mark Zuckerberg. He may well have it in there. But, I think the market looked at just said, we don't know what you're doing.

00:15:42:29 - 00:16:04:10
Unknown
And I don't think Yann LeCun departure, plays to that necessarily. But it can't help. Meanwhile world models would be incredibly useful when you're talking about things like, you know, their glasses and their, their headsets and, and all that kind of stuff too, like, you know, that stands to benefit a lot from that, whether that is the next big thing.

00:16:04:10 - 00:16:33:29
Unknown
You know, the company still called meta. It still seems to be dedicated in the long term to the metaverse. And the metaverse sure stands to benefit from all of this. If it's to come to pass the way. Yep. You know, they might they might be working towards some. Yeah. Interesting. Well, so, somewhat related, I guess when we're talking about kind of the, the, the worldly development of AI, you know, between companies and the investments that it requires in order to do this.

00:16:34:02 - 00:16:56:14
Unknown
This is one of those news stories that I'll be completely honest, yesterday when I was on my deathbed fighting, you know, a pretty nasty, cold, I was pretty much in bed all day, but still kind of do a laptop work and stuff and definitely setting up this show. I was putting in the news that SoftBank sold its entire $32.1 million share of Nvidia, last October for $5.83 billion.

00:16:56:14 - 00:17:14:07
Unknown
So getting out of its investment with Nvidia, essentially, I was putting this in there. I was like, I know this is a big deal, but wow, I don't have the brain cells to really think it through. So hopefully you can help me with that. But, one thing I did find interesting in this, as we talk about it, is this is not the first time that SoftBank has done this with Nvidia.

00:17:14:07 - 00:17:42:04
Unknown
Back in 2019, sold its entire 4.9% stake in Nvidia around $3 billion and actually ended up being a total like wrong, horribly timed sale. And they ended up losing a ton. Like if they had stayed in, they would have 30 or 50 axed their investment. But instead they got out and it's kind of one of those, one of those examples of like, oh, when you time the market and it just doesn't work in your favor, that's that.

00:17:42:04 - 00:18:11:13
Unknown
So will that happen again here? That had SoftBank's and they held out until today. Nvidia's stake would have been worth $210 billion that they sold earlier. That was for a 5% stake. Now, now, the holding that they have, an Nvidia is only 0.13% of Nvidia because it's grown so huge. In the meantime, they need this SoftBank needs the cash to be able to meet its commitment to invest $30 billion in OpenAI.

00:18:11:16 - 00:18:35:21
Unknown
I see, and that Necess, I think pretty much necessitated the sale. So I don't think it's a huge commentary on Nvidia. 0.10.2. I don't know if you had your bucket of billions, which I know you're going to have Jason very, very soon and you could only invest it in one company today. Would you invest in OpenAI or Nvidia?

00:18:35:23 - 00:18:51:17
Unknown
Again, we are not giving any financial advice and I would any time I've ever made any of these decisions in a fake or even a real sense, I almost always lose. But I would probably invest more in Nvidia than I would, I would too. I mean, one could argue that Nvidia is at a high, but I think it just keeps going up.

00:18:51:17 - 00:19:28:04
Unknown
And I think they benefit from everything done. And I hope and I has $1.3 trillion in commitments supposedly. Yeah. Oh boy. Right. Yeah. Super extended. We've had this story before. Anthropic is is on a far better trajectory to to profit. OpenAI is not and you know, the worry for everybody right now is open. I think we know about this last week is too big to fail because the, the, the, the tertiary, failures around it would be, hurtful, wouldn't it?

00:19:28:04 - 00:19:53:05
Unknown
Wouldn't fold, Amazon or Microsoft or Google, obviously. But, they all now have pieces of open AI business and Nvidia too. And it could hurt if they're wrong. So I think the whole AI world has to hope that OpenAI gets it, keeps its act together and does well. But who knows. And this is a vote of endorsement by SoftBank.

00:19:53:05 - 00:20:14:15
Unknown
But SoftBank is not always right. Right. Well and like you said and actually folks in our chat night Sky is is probably forced by their debt. Yeah. They're just kind of put put into a position where they need that cash to, to bolster their, their commitments essentially. And you mentioned the anthropic story. We might as well just talk about.

00:20:14:15 - 00:20:36:28
Unknown
That's right. I didn't see it. Was it here? I forgot it was in here. Yes. Yeah, yeah. No, no, it's all good. But it's, you know, it's it's good to couple these together. Anthropic. Like you said, taking a very different approach to OpenAI clod getting a lot of uptake actually with business users right now. So what anthropic is doing is essentially my understanding, seeing a lot of revenue growth at this point.

00:20:37:00 - 00:21:05:17
Unknown
And possibly or rather, they expect to reach profitability sometime by or around 2028, contrasting with open AI and its kind of strategy, which is heavily, you know, delaying its profitability timeline, essentially pushing, you know, making lots and lots of investments, basically bets on future success versus anthropic saying, hey, we've got the business right now and we could be profitable here very soon.

00:21:05:17 - 00:21:24:15
Unknown
By comparison, very different strategies. Yeah, I think that the, as you just said, they're very different strategies. And, you know, it's something that when I was advising startups and student startups and stuff, you always see this. Are we B2B or B2C? And the desire is to be B2C because that's when you're big. That's when you're Google.

00:21:24:15 - 00:21:49:16
Unknown
That's when the whole world knows you. That's when the stock market buys you. That's when, the revenue is larger. But B2B is a safer path. You have a more limited customer base. You know, the value you're bringing to them. Your brand may not be as big, though. Claude's a big brand in certain quarters. And anthropic has taken the B2B and open.

00:21:49:18 - 00:22:16:06
Unknown
I would like to think they're both, but I think you're never really both. They're trying to be a consumer brand. So I think anthropic is a very good strategy and I think it's working and they're smart. So okay sort of sort of related but unrelated side sidebar on this, I realized, just yesterday actually, as I was in my death bed, trying to get over this yucky cold, and we, all of us here resurrected today.

00:22:16:06 - 00:22:37:00
Unknown
Thank you. Oh my goodness. I woke up this morning. I was like, what happened? I feel like a million bucks. Like I'm so full of energy right now. I'm amped for this show. But what I realized yesterday is of all the mainstream, say, mainstream services, ChatGPT perplexity, which is kind of a wrapper, you know, of all the different services.

00:22:37:00 - 00:22:57:19
Unknown
But, Gemini, you know, and then, of course, you know, clod and all these mainstream things clouds, the one that I have the least familiarity with, or at least one of the ones that I realize I've used very little by comparison, at least explicitly. I might use it through perplexity here and there for certain things to like test around, but actually going to the cloud site and using it.

00:22:57:21 - 00:23:20:24
Unknown
And so yesterday I used it for something, and I got most of the way through the process, and then it totally blanked on me. And when I tried to refresh the chat, it would let me back into it anymore and all my work was caught. I was like the worst time. So I, you know that I it doesn't really relate to this conversation except for the fact that like that was my personal experience with cloud.

00:23:20:24 - 00:23:40:01
Unknown
I realize I need to use it more because I hear a lot of people say like, it's a damn good product for very specific, certain things. It's the one they go to. And yet I realize my kind of, deficiency in understanding around cloud for my own uses. So I'm going to commit to using it more. Yeah, I need to.

00:23:40:04 - 00:23:58:15
Unknown
Yeah. And this is an interesting question. Daniel. What's up? Daniel, in our live chat, Daniel Croft helps out with some of our social stuff on the back end. So thank you, Daniel, for that. Says I think we have an open AI bubble rather than an AI bubble. That's curious to me. Go ahead. What do you think of that?

00:23:58:17 - 00:24:20:17
Unknown
Well, I would just like there's a lot of talk about bubble. And then when you talk about, you know, the companies and the real electricity and movement in the industry, I mean, it goes a million different directions, but always open AI is a big part of that conversation. OpenAI has all of these relationships and guaranteed to have more relationships with all of these other companies.

00:24:20:17 - 00:24:41:21
Unknown
So if OpenAI fails, then that has a dramatic impact on on the health of other companies due to their their business with it. Like I don't know that AI is at risk of going away. I don't believe that like a bubble would happen and suddenly, oh, no one uses AI anymore. I think AI's here for the long haul, but that is an interesting question to me.

00:24:41:23 - 00:25:00:28
Unknown
That kind of bridges those two things. Maybe the bubble we're looking at is, you know, really comes down to a company that is so big, like we talked about, I think last week, so big that it's too big to fail, that if it does, you know, it's an open AI problem. It's not an AI problem. Yeah, I know I agree.

00:25:00:28 - 00:25:28:02
Unknown
I think well or put it this way, my fear is that it's an open AI bubble and they're going to ruin it for everybody else. Right. There's one there's 1.3 trillion or thereabouts dollars worth of orders for things out there based on open AI magically multiplying its revenue. And if open AI folded tomorrow, that's $103 trillion of bookings that people thought they were going to get in other companies.

00:25:28:02 - 00:25:48:10
Unknown
It goes away in a year. So you're absolutely right Daniel. I think that alone qualifies as a as a mini bubble, a little tiny bubble. Not the whole of AI but it would have huge market impact as a result because it would take away confidence. It would take away $1.3 trillion in orders and then it would take away confidence.

00:25:48:13 - 00:26:11:18
Unknown
Yeah. And, I would we had to another AI winner. Probably not. But it would have tremendous impact. So a lot a lot, a lot, a lot is on Sam Altman shoulders. Yeah. How's that feel? How's that feel to have the world on your shoulders? Oh, yeah. I think he likes it. Likes it? Yeah. I was going to say he likes it.

00:26:11:20 - 00:26:31:13
Unknown
Completely agree. And after the break, we'll have more on Sam Altman shoulders and the weight. Yes, totally. It's a it's actually a really great segway into our next topic, which we'll talk about in a moment. But first Patreon.com slash AI Inside Show, we like to mention it because this is your opportunity to support this show on a deeper level.

00:26:31:14 - 00:26:51:03
Unknown
To be involved at a certain tier, you get an AI inside t shirt, which I don't know if anybody has noticed. I made the correction silently, or the change a couple of months ago, but I kind of slightly altered the logo so that the it just always felt imbalanced. And so, you know, this kind of integrates it more into the AI.

00:26:51:03 - 00:27:10:23
Unknown
And I need to update the shirt so that it has the new version of the logo, but I like it a lot better. If you do contribute to a certain level on our Patreon, obviously you're supporting the health of this show and you get that t shirt, you know, you can wear it out, it can be your pajama shirt, I don't care, but you get one Patreon.com MRI inside show.

00:27:10:23 - 00:27:36:25
Unknown
And we think, we think so many of our fans for contributing, each and every month Steve Linthicum, Neil Wood, Brandon Kester, just a few of our amazing, patrons. So thank you for supporting us and for having faith in what Jeff and I do with the AI inside. After this break, we will indeed talk about open AI and, what all this talks with the US government.

00:27:36:27 - 00:27:39:20
Unknown
What is that all about? That's coming up here in a moment.

00:27:41:11 - 00:28:06:23
Unknown
Open AI held talks with the US government recently. And, you know, obviously we have its for profit structure kind of firmly in place. Kind of feels like that was the, the kind of segue to chapter two. And now a whole lot of things are moving in, in and fast pace, even though they say they're not, you know, they're not planning actively for an IPO.

00:28:06:25 - 00:28:31:08
Unknown
The business is expanding. They're really looking in a lot of different directions, things the pace at OpenAI seems to have really picked up, and now they're facing a ton of pressure, around expansion of its infrastructure like we were talking about before the break. Particularly in the semiconductor chip manufacturing side of it's business. And they're seeking these funding strategies to hold talks.

00:28:31:10 - 00:28:54:14
Unknown
I imagine from this report. Anyways, preliminary talks with the US government about federal loan guarantees for new chip plants and, Sam Altman is insisting this is not about the company asking for taxpayer backed bailouts. He says the company wants private capital for those data center build outs. Government support would be for the chip production itself. Am I getting this right?

00:28:54:14 - 00:29:28:22
Unknown
This is another one of those stories that I was putting together on my deathbed. Yes. So, word had it that that OpenAI was, going for, guarantees from the federal government, favors from Trump. And this led to what we were just discussing led to a fear that, oh, are they too big to fail? Are they trying to put themselves in that category that they're going to get government handouts and they're sucking up to the administration to get that, and so Altman had to come on and, put up a post saying, I would like to clarify a few things.

00:29:28:22 - 00:29:52:05
Unknown
Oh my goodness, this thing's a novel. So, first he says that the open AI does not have or want government guarantees for open AI data centers. We believe government should, and should not pick winners or losers, and taxpayers should not bail out companies that make bad business decisions. Okay. After that. But then he goes on to say a couple of things.

00:29:52:05 - 00:30:15:15
Unknown
The first is that he thinks it makes it might make sense for governments to build and own their own AI infrastructure. Okay. Which is pretty weird to me that that does that mean that, you know, in one sense is AI infrastructure than a utility? He's not saying that, but that's that's not a big step from one to the next.

00:30:15:17 - 00:30:34:04
Unknown
Is this like, you know, the post office for AI? I think that's a dangerous, dangerous world because this is a clearly a for profit business. There's a lot of independent capital going in. And so I don't think governments need to do that. Governments step in when there are market failures and this ain't a market failure yet.

00:30:34:07 - 00:30:55:10
Unknown
Maybe that's saying it could be. But he's saying the upside of owning their own infrastructure would flow to government as well. Well, that's where Trump is going in terms of investing in air quotes. Taking, taking, taking ownership of chunks of companies. The right wing and especially the libertarians and the Wall Street Journal go mad at this.

00:30:55:12 - 00:31:29:10
Unknown
They don't want government involvement in business to that level. So here's, Sam saying, and maybe it's not just to the government, to the American government, it's probably more likely to the Saudi government and to other governments around the world. Yeah. Why don't you buy a whole bunch of stuff and host us in your centers? Then he says that the loan guarantees he did discuss was not about his own data centers, but was instead about semiconductor fabs, factories, in the U.S and this this plays to Trump's argument that he's bringing manufacturing back to the U.S.

00:31:29:10 - 00:31:59:00
Unknown
Of course, this is not putting screws in phones manufacturing. This is extremely high end, very expensive, long runway manufacturing. So should the US government give loan guarantees to bring that into the US? Maybe. Maybe that's a better strategy, by the way, than tariffs on things. But it is government involvement in business. And when are we in America complaining that China puts its thumb on scales with government support of industries and businesses?

00:31:59:02 - 00:32:19:12
Unknown
Then we're just doing the same thing. So there, he says. The next question that comes out naturally is how, you know, he's asked all the time, how is opening? I'm going to pay for all this infrastructure. It is signing up for the one plus trillion dollars. So he says, we expect to end the year about $20 billion in annualized revenue run rate.

00:32:19:12 - 00:32:47:16
Unknown
That's revenue, not profit, to grow to hundreds of billion a year. Sam. Typo. You bet billions by 2030. We are looking at commitments of about 1.4 trillion over the next eight years. Obviously. Obviously this requires continued revenue growth. No poop. Sam. And each doubling is a lot of work. Yeah, but we are feeling good about our prospects there.

00:32:47:16 - 00:33:09:06
Unknown
We are quite excited about our upcoming enterprise offering, for example. And there are categories like new consumer devices and robotics. We also expect to be very significant. And there are also new categories we have a hard time putting specifics on like a high the cost of scientific discovery. So smoke, smoke smoke smoke smoke lots of money man. You put it that way.

00:33:09:07 - 00:33:29:07
Unknown
And so, you know, that's a lot of that's a lot of kind of word salad and talking around very obscure things that you. And it goes what we discuss any way shape or form you want to really. Exactly. Sorry. And it goes exactly what we discussed before the break is there's 1.4 I was wrong, not 1.3. Well, what's point $1 trillion left?

00:33:29:09 - 00:33:56:11
Unknown
$1.4 trillion hanging on open commitments, and they're at $20 billion gross revenue with lots of losses this year. RCF wow. And so all that has to change, all that has to ramp up to kind of meet what Sam is talking about in a matter of what four years is. He says he's talking. Well, he said 2030, but then he said, oh, okay.

00:33:56:12 - 00:34:18:13
Unknown
He has commitments over eight years, so he expects to have, hundreds of billions of dollars in revenue by 2030. And the commitments go over, eight years. So that he, he that's the elephant in the room on the rear end and says, is OpenAI trying to become too big to fail? And should the government pick winners and losers?

00:34:18:15 - 00:34:39:25
Unknown
Our answer to this is an unequivocal no. If we screw up and can't fix it, we should fail. Yeah, but easy for you to say. And now it's just you in one sense. And your venture investors. The hell with you. But it's not just that anymore. It's all these other companies whose stock, skyrocketed because of these deals.

00:34:39:25 - 00:35:03:07
Unknown
You made this $1.4 trillion. So, then third is, what do you need to spend so much now instead of growing more slowly? And he says, we're trying to build the infrastructure for a future economy powered by. But this all rests on the AGI, IBS. This all rests on this idea of not only exponential growth, but exponential functionality.

00:35:03:09 - 00:35:28:27
Unknown
So this all makes me a little nervous all around. Sam is trying to do some things down here, but in fact instead it says to me, can I see the mirror behind the smoke? Yeah. He's. Instead, it seems like he's drawing pictures on the wall to say, there's this there's this thing out there that I promise you is going to make all this worth it, and it's going to deliver.

00:35:28:29 - 00:35:45:13
Unknown
You can trust me. And, Wow. Yeah. So it really excuse me. It really seems like the the theme of the first half of this show is an open AI bubble. Not not an AI bubble, but an open AI bubble. Yeah, yeah. Question mark. Of course. Question mark. Yeah, of course. And I'm not. There's a lot I'm not hoping for that.

00:35:45:13 - 00:36:05:16
Unknown
That would be a bad thing. But and that's why we need open source. So that's why we need competition and that's why we need a lot of smaller ventures out there. And that's why this whole. And that's why I'm hoping that Yann LeCun comes along and says, I have a different model than scale. For scale sake, so that the winner isn't.

00:36:05:24 - 00:36:27:15
Unknown
I saw a story today. I didn't put the rundown that said that the that the or was a blog post or something, that just speculated that we're not in a race for chips, that instead we're in a race for electricity. And, and I think that's, that's right. Now that's true. The grids are there, the mechanisms of generating the power.

00:36:27:15 - 00:36:59:11
Unknown
This whole race for scale is pushing the grid. To its limit. And we're all paying for that and higher electricity costs and so on and so forth. Right. If we see new, paradigms from the Chinese and deep seeking sort of much smaller models, and if the likes of Yann LeCun and Feifei Lee come along and say, scale is not the way to win this, we have a different paradigm, and that's where we're going to put our investment, and it doesn't require quite as much.

00:36:59:11 - 00:37:26:25
Unknown
And I'm guessing here maybe it require more, but I'm just guessing that a new paradigm will be a new paradigm because it's smaller. One of the papers I looked at this week, which I didn't put in the rundown, was involve Google and, trying to exponentially increase the efficiency of generative AI by not dealing word by word in tokens, but instead by dealing with vectors, which it does somewhat now.

00:37:26:25 - 00:37:46:05
Unknown
But more of this where you take a phrase together, in essence, think of it that way and that that becomes the building block, and so it can deal with more at once. It gets more computing done, at less power, less resource, at greater efficiency. So I think we're going to see a lot of work going on there.

00:37:46:08 - 00:38:04:09
Unknown
So, you know, the way I look at this, Jason, is what if, you know, Deep Sik, is right, and they can do things that are just as amazing and far, far smaller. The other thing that can happen is that doesn't mean OpenAI fails. It just means they bought a lot of capacity. They're not going to need that.

00:38:04:09 - 00:38:26:03
Unknown
They don't need that. They don't need. And we talked about this last week in the case of the internet, it was an overbuild of fiber that left a lot of people hanging. Hold it. Holding bags of of of glass fiber. But other users came a lot. So maybe that's not a bad thing. Maybe it is a bad thing that paves away everybody's game.

00:38:26:06 - 00:38:48:19
Unknown
Yep. Yeah. Totally. So I'm guessing that could also be a title. But I think it's. Yeah it's very true. Interesting. Thank you for breaking that down. Because yeah like I said, I was I was scratching my head like, I need to understand this more than I do. And, that makes a lot of sense. I love that the first half of our show has that really strong, solid theme.

00:38:48:22 - 00:39:05:10
Unknown
Now we're going to undo it because, wow, this yeah, this next story is really shifting gears, but it was one that I saw that I was like, okay, we got to talk about this. I knew you'd love this. Wilcox. Yeah, you put this in there. And I hadn't seen it. I hadn't caught it. And I'm really happy that you did.

00:39:05:13 - 00:39:28:19
Unknown
Breaking rust is a hit number one song. The Billboard Country Digital Song Sales chart with a track called Walk the Walk. And apparently that artist could be the first AI generated artist to hit the chart. Or one of the first, I don't know, maybe there's one before it, but it's still as far as country music, certainly. It's certainly be number one.

00:39:28:21 - 00:39:55:09
Unknown
Yeah. Everyone's trying to crack the code right now, apparently trying to figure out, like, who the heck is this? Who? What is the identity behind this song? Abby. Yeah, Rivaldo Taylor is a name that comes up as people have been poking around. It's attributable, to, IBM or through other AI powered music projects online. If you go to, what is it breaking?

00:39:55:09 - 00:40:15:16
Unknown
Rust's Instagram. You know, it's a bunch of posts that all you know, are AI generated. Let's see here. Do I dare play one of the songs? If I do? Agent sorry, I hope, I don't know, it's all good. I hope I don't get my get this stream taken down if I, if I play this. But I think we got to play just a little.

00:40:15:18 - 00:40:38:17
Unknown
No, don't open that. Open something else. Dang it. My windows are not cooperating. There we go. Okay, let's try and see if we can hear just a little bit of break and rust. Let's see. Unmute. I don't like how I talk. I'm gonna keep on talking and walk my. Whoa. Ain't changing my tone ain't changing my song I was born this way being loud.

00:40:38:24 - 00:41:01:23
Unknown
Okay, so that's pretty good. Here's here's another song. You can get rocks if you don't like how I'm on. Okay, that sounds like maybe that's the same song, but there are different ones, but they all sound. There ain't got no damn apologies left in me. I'm saying my sorry. Okay? I'm only playing little bits because I don't want to to mess with anything and get the episode taken down.

00:41:01:26 - 00:41:20:16
Unknown
But one of the thing that people are, you know, even if it is I generated, which I mean, I fully believe that it is, is, one of the thing that people are noticing is there's a consistency across all the songs, like the vocalist as we, you know, put in air quotes. Because if it's an AI generated vocalist, it's not an actual person who's generated.

00:41:20:21 - 00:41:44:05
Unknown
But there's a total consistency among the style, among the, the vocal, qualities and all that kind of stuff. Instagram account around 35,000 followers. The lyrics, you know, as you read through the lyrics kind of have that I like kind of perfect quality or maybe that's just knowing what I think I know about it, and then I look at it through that lens.

00:41:44:05 - 00:42:04:23
Unknown
But I, know it's, it's interesting nonetheless to have kind of a, the watermark of, oh, here's an AI generated song. It's the first time that it's cracked the Billboard Country Digital Song Sales chart and reached number one, by the way. But I don't know what it takes to do that. Is that 3000 sales? Is it? You know what I mean?

00:42:05:00 - 00:42:22:07
Unknown
That's a really good point. What's a sale? What's the sell when you have what is a sale today especially. Yeah, this is one thing I'm working on. I've got a book proposal out there now on a mass media. And it's really interesting as I dig into the numbers and I've got to get more of these. Mass was never mass.

00:42:22:10 - 00:42:43:08
Unknown
It was never huge. Right. So the Super Bowl is the only thing that really gathers a whole mass of people. It it carries a lot of Americans basically. And that's the biggest you can get. The top movie, last year, was a kiddy, sequel. I forget which one it was. And it got 50 million tickets sold.

00:42:43:08 - 00:43:13:20
Unknown
But that's family stuff. So, you know, a lot of those repeats sold number of people. Yeah. The parents were dragged along and they wouldn't have watched it. They didn't want to. That's 5% of the country, a best selling book. Now. Would sell 100,000 copies. That's 0.03% of the country. And that's big old mass media. Now, if you go to, slices of media, country music is a slice of music.

00:43:13:22 - 00:43:35:15
Unknown
Right then that's going to be a fairly small slice. And so nothing speaks to the zeitgeist of the whole country. That's always been a lie. It's always been a presumption that, oh, we speak to everybody. We were the whole nation. This is what everybody's talking about is what everybody is bopping to. No, it's not true. So I don't think you actually need that much to be considered a bestseller.

00:43:35:17 - 00:43:56:28
Unknown
Yeah. I mean, in cases of books where certain Republicans, went and just bought a whole mess of their books to get themselves on to the bestseller list. Sure. We I've seen that. I saw that when I worked at Twit because, Justin Robert Young and, and, Brian Brushwood with, you know, with, why am I suddenly blanking Diamond Club?

00:43:57:00 - 00:44:19:02
Unknown
They did a book that was inspired by the 50 Shades of Gray. It was a it's a book full of gobbledygook. And they were able to mobilize their fan base to buy it on day one through Amazon and push it to the top of the Amazon like romantic bestsellers list. It doesn't take a lot. It just takes a certain number in timing.

00:44:19:04 - 00:44:38:01
Unknown
At the same time, like if it is 3000 bot and I don't yeah, again, I don't understand the whole buying nowadays. Like, you know, are people actually buying digital copies of songs? Yeah, that's a smaller subsection right? Yeah. Right. You're talking about country B being a smaller subsection. I mean, the people who actually decide to buy, is it an entirely different one?

00:44:38:01 - 00:44:59:19
Unknown
But, you know, cracking the, the, the list or the, the story for the first time, it's still a very interesting cultural story. Yes, yes. Yeah. Interesting to see that happen. And I would say it's probably a harbinger of other things to come, even though a lot of people really don't want to see that. Right. So I yesterday I spoke super curious.

00:44:59:19 - 00:45:27:25
Unknown
I was on a panel yesterday at, Gotham Ghost Writers, a conference of ghost writers. And yeah, everyone, everyone was there. They were they were identifying material spirits. They were, they embody. And so. So they're hired, you know, to write celebrities books and executives books and, and so on, and to do some topical books that they get hired to do.

00:45:27:27 - 00:45:49:11
Unknown
You know, I talked to one guy. I know he's written 80 books. You know, I've written seven. And they're nervous as hell because if I'm a celebrity and I want my story told, I don't have to hire a person to do it. Now, I can put a whole bunch of interviews into notebook hell, and it'll write my book.

00:45:49:13 - 00:46:12:25
Unknown
And so they're scared. What's going to happen to the argument is you still need the human touch, and the humans are going to do a better job and all that to be true. And and a lot of these books, a lot of this comes to people want their stories told. They don't know. I mean, I, I remember knowing writers who got hired by rich families, the gentleman's baking family hired someone to write the family story, the history.

00:46:13:01 - 00:46:32:04
Unknown
And I was a big fan of the chocolate chip cookies. I wanted to read it. They only wrote it for the family. It was internal. But they hired a journalist to write this whole thing. Right. And, there are a lot. They're rich. There are a lot of people who want to tell their stories who aren't. You see, these products now where you can make your own biography online.

00:46:32:06 - 00:46:50:12
Unknown
I was just going to say, I've got plenty of those emails like, hey, you know, I will help you write your own thing. Meanwhile, I could probably just open up Gemini like, know, you know, question like ask me questions that would fill out my bot biography and just spend an afternoon doing that and now organize it in a, chronological order.

00:46:50:12 - 00:47:04:22
Unknown
Okay, Jason, all right. You now you need to do that and put it on Amazon. See how does oh, boy. Jason I don't I don't I probably would want to put it on Amazon. I just be curious to see like do I even care about what it what it is about your own life. Yeah. It's a good enough.

00:47:04:24 - 00:47:27:25
Unknown
Yeah. So. Yeah. Interesting. Yeah. I think we are going to see more for this. You're right. Yeah, yeah, I think so. What time is it, Jason? Oh, you know what time it is. It's. Jeff. So I catch you down. Probably one of the only opportunities we have to play stupid little jingles created by I on this show.

00:47:27:25 - 00:47:50:15
Unknown
Speaking. Yes. Yes, exactly. Just go through, pour through 500 to 700 papers mentioning I on archive.org. I also look at, screen and the social sciences, archive version. They don't have as much rich stuff. And then I come through and I end up with about a dozen papers, and then I subject Jason to, three of them.

00:47:50:17 - 00:48:14:20
Unknown
And you love it. So the first one is really interesting. It was a comparison. Let me see what. Pages seven a comparison. How similar are rocket, PVA and Wikipedia by Taha. Yeah. Sorry. Who is, the Technological University of Dublin? So the conclusion is great. It found that,

00:48:14:23 - 00:48:45:26
Unknown
Let me see here, there's scores about about stylistic and content similarity. The platforms diverge substantially in form and informational scaffolding. Rocket PDA articles are on average several times longer with higher Flesch Kincaid grade levels. That's just a grade level to judge. You know, the how complex languages, but lower lexical diversity and reference density, that is to say, fewer sources.

00:48:45:29 - 00:49:14:20
Unknown
This pattern suggests that rocket PD is generation process elaborates on existing material, expanding text length and rhetorical flow. A nice way to say blather, rather than rather than producing substantially new or more rigorously sourced knowledge. A very fancy, nice, polite way to say that rocket pedia is f o s u fil a. It's jets. Yeah, when we talk about slop generation, I mean, that's what we're talking about here.

00:49:14:20 - 00:49:40:15
Unknown
It is taking a small amount of information let's say, and making it verbose. Yeah. And then and then adding in a certain amount of political worldview. The next one I find absolutely fascinating is brain organoid computing organoid or organoid? The New York Times had a feature this week about brain organoids. And they're not full brains, but they're taking stem cells and they grow.

00:49:40:18 - 00:50:06:05
Unknown
What would be a part of a brain and what one lab that they're times talks about has been growing them. The nuggets as they're called, are now seven years old. So that they they they grow. And then their chemical cocktail, they become like a fetal brain, but. Or part of it. Oh. So now the effort is to say, could this become the new chip?

00:50:06:07 - 00:50:46:04
Unknown
What would it look like to as, as, as AI is trying to imitate generative AI and machine learning and neural networks, neural networks are trying to imitate the brain. What have you used? Pieces of brain. And you can imagine all the sci fi and all the fears and all that will come up. But it's fascinating using. So this is a paper by Yannick Talavera, at all from, Germany thought like, it's just that's all it is, is Yannick and no bench Ullman at the from University of Applied Sciences at Frankford mine.

00:50:46:06 - 00:51:17:21
Unknown
So they say that this opens up new perspectives on how intelligence systems might be designed in the future. Using brain organoids for computing presents a possible pathway towards more adaptive, energy efficient and biologically inspired forms of AI. However, challenges persist, particularly regarding lifespan interfacing reported ability to reproduce, and ethical concerns regarding the use of human derived tissue. So that was an interesting one.

00:51:17:24 - 00:51:44:01
Unknown
Interesting. And then next, speaking of well, I guess that's kind of fake humans in a way. I'm just joking there and making a segue. I saw this fascinating little bunch of papers around. The first one is from Manley Choo at Clark University in Worcester, mass. ET al. LMDh based multi-agent system for simulating and analyzing marketing and consumer behavior.

00:51:44:04 - 00:52:11:27
Unknown
So this is an effort to say that we can come up with fake consumers and then test, marketing against them. Okay. And then there was a next paper by, sage Philippe at all from MIT and University of Cologne and Yale shareholder democracy with AI representatives. So this is an effort to, simulate, shareholders interests by having fake humans.

00:52:11:29 - 00:52:38:02
Unknown
And so I find this all to be a bit disturbing, because it's bad enough. I hate opinion polling. I think it's the ruin of democracy. I think it preempts the public conversation. And it's meant to measure. To paraphrase James Kerry, I think it it it robs people of their agency and their identity and their nuance, and it makes us into just statistical, cold statistical, artifacts.

00:52:38:05 - 00:52:54:11
Unknown
And so now what we see is a next step of that. It's not just that it's a poll where we are represented falsely. It's they don't even bother to look at us. They make up versions of humans and set them against each other and then make decisions based on that. You might say, okay, Jeff, that's fine. You're just panicking.

00:52:54:13 - 00:53:33:28
Unknown
That's just a, sci fi fear of yours. But at the same time, a story in London's Telegraph says that, Prime Minister Starmer's synthetic voters show Downing Street's lost the plot and how, rather than using focus groups of real voters, number ten Downing Street is experimenting with synthetic voter voters, according to a spectator boy, a piece, fascinating, but worrisome to me because we want to be there, standing there yelling at these companies and, these governments.

00:53:34:01 - 00:54:04:18
Unknown
We're here will real listen to us. Don't make up fake versions of us. That's the wrong way to go with this technology. Yeah, so that's my paper. Fun for the week. That is a little worrisome. I hadn't considered that. Excellent. Well thank you for giving us your showdown or rundown or whatever you want to call whatever this is just so I got you down and everything must be clear.

00:54:04:20 - 00:54:26:05
Unknown
Night screen in the chat says this is a common simulation technique. Yes, absolutely. It's been going on for some time, but I think I brings it to an entirely new, manufactured level, and I think we have to call it out, so. Now to the break. All right. Yes. Let's take a quick break before we do.

00:54:26:06 - 00:54:43:01
Unknown
There are a lot of you who catch this show on Apple Podcasts and I was actually just looking at the numbers the other day. I hadn't checked in on in a while. I'm like, wow, there are quite a few of you. So if you are listening via Apple Podcasts one thank you to have you left a review.

00:54:43:01 - 00:55:00:29
Unknown
If you haven't, leave one whatever you want to say about the show, but it really helps us out. If you can drop that quick review on Apple Podcasts so more people can discover what Jeff and I are doing, and we appreciate you. Thank you so much for sticking with us. All right, quick break. When we come back, we got a couple of quick stories, to round the show out.

00:55:00:29 - 00:55:02:26
Unknown
We'll be back in a moment.

00:55:03:05 - 00:55:34:04
Unknown
Perplexity CEO Arvind Srinivas is warning that AI chat bots, as companions are a dangerous combo, and that perplexity ain't going to do that. So it's on the record. Arvind says you will not get chat bots like, like, companion style chat bots from perplexity. Remember this in case, this the seas change in the future.

00:55:34:06 - 00:56:00:03
Unknown
But he says it leads to users preferring synthetic relationships over real life, or can lead to that over real life relationships, and that young minds are easily manipulated by these, by these chat bots. Doesn't like them at all. So the funny thing about AI is that the the the the techno panic, moral panic calls come from inside the house.

00:56:00:06 - 00:56:45:26
Unknown
It's the AI, people who are the ones to first warn about how dangerous we are. Yes. And in the end, it strikes me as condescending and patronizing. To say that we're all manipulable by these tools they create is to give them more power, and us less agency. And so that bothers me at a level I just also put up a piece from a friend of mine, out of a couple of Chevys, at the Chamber of Progress, which is, which represents tech companies saying that Tim Wu's new book, The Age of Extraction and Cory Doctorow is now, very famous bestseller in should ification are, in essence, patronizing views of the

00:56:45:26 - 00:57:04:00
Unknown
public. They act as if we have no agency and we don't make our own choices for our own good reasons. And I think we have to enter that idea back into this discussion of AI. People may do things that you think are stupid, like falling in love with AI, but they're making that choice to do so. The AI isn't making them do it.

00:57:04:03 - 00:57:31:08
Unknown
And the issue then is humanity more than technology. So I just wanted to add that it that is true. There is also the case though of, of you know like children. Yeah. You know younger users who use children. Yes. That's a, that's definitely a different, a different aspect. And I'm sure I'm, I'm pretty certain I feel pretty confident that there are certain dangers there that we want to, you know, be sure that we don't expose them or introduce new habits or, you know, who knows what I mean?

00:57:31:08 - 00:58:05:13
Unknown
It is a new world. It's it's kind of like learning. Okay. What are the what are the training wheels? What are the walls and the boundaries and the guardrails? I hate to say that word, but you know, that in which we work when it comes to these things. So, interesting stuff there. We've got Apple also, finalizing apparently a deal with Google around $1 billion a year to use its Gemini AI model, to push a major update to Siri, which we've been talking about for quite a while, and Apple's been kind of hinting at for quite a while.

00:58:05:15 - 00:58:39:12
Unknown
Possible release next spring. This would be a bridge to Apple's own model somewhere down the line, so maybe not permanently reliant upon other companies, AI models, but wanting to. Yeah. Create a system that maintains privacy because that's something that Apple has been really publicly committed to with their devices and their platform. By processing a lot of that device, the data on the device, and then using the Gemini and the cloud for more advanced computation needs.

00:58:39:14 - 00:58:54:28
Unknown
I guess as needed. So and it's also saying that Apple doesn't really have its own AI strategy. Yeah. And that's not necessarily a criticism. It may not be. They maybe they shouldn't maybe they should bring that. But these tools are changing constantly. They're leapfrogging. Maybe they should just total you bring the best of it to their to their users.

00:58:54:28 - 00:59:24:18
Unknown
That's okay. Apple often plays the wait and see game and jumps when the time is right for them. And you know this this could be seen through that lens definitely as well. Amazon launched Kindle Translate, which is an AI powered service for e-book authors on Kindle Direct Publishing. So currently it translates between English, Spanish and German. More languages are to come and authors can, you know, preview their translations before it's published.

00:59:24:18 - 00:59:48:18
Unknown
In that way, if they're able to read those translations, I guess, it's a free service. And those titles that are translated will be labeled for readers so that they know it was done by I. Would you trust Amazon's AI powered service to translate your books? No. But I might still want it because, you know, I still haven't gotten a German ray out there, knows a German publisher.

00:59:48:18 - 01:00:11:03
Unknown
I still haven't gotten a German publisher for the Gutenberg parentheses. Hello, folks. Wow. Yeah, it's just pisses me off. The publisher ghosted me. And there's other languages that I would never get to write. How am I going to get to, Estonian? Ain't going to happen, though. Estonians read English. But, I think there's an opportunity now.

01:00:11:03 - 01:00:32:10
Unknown
Interesting to hear this puts it in the author's hands. Which right is necessary given certain rights, but there's lots of books that I really want to read in other languages that I can't unless I play around. And I would end up doing the same thing, you know, breaking the, protection and find a way to use it for my purposes as a reader.

01:00:32:12 - 01:00:49:18
Unknown
So on both ends, I think there's a lot of power to be had here. Yeah. It's gonna lose nuance and it's probably gonna make some mistakes. But either one of my books opens up a whole door. Yeah it does, it does. I remember people complaining about the human translation of what am I, earlier books. So, yeah.

01:00:49:21 - 01:01:13:23
Unknown
By the way, you can by Gutenberg parenthesis in Spanish. That's a good human translation. Beautifully produced book. There you go. Yep. Right. Always. Yeah. I mean, yeah, I would agree. It's kind of good to have the option, for those who want it being a free service. I mean, we and we've talked about so many times about how these systems are just this is one of the things that they're, you know, pretty solid at is this whole translation aspect to things.

01:01:13:23 - 01:01:34:15
Unknown
So it might not get everything 100% perfect, but something is better than nothing potentially would be my opinion anyway. Yep. Having never written a single book, probably never will. Oh, you know, I, I use, as you know, because I quote them obnoxiously. I try to read German publications. And, I do get help from, from, Google Translate.

01:01:34:15 - 01:01:53:26
Unknown
I was at the Munich media in Tiger a few weeks ago. I just this afternoon put a video for that conversation, if you want to see it on my socials. And at one point in the middle of it, I said in German, the Aiden talk. Lazy ish. Süddeutsche Zeitung on the site, on Der Spiegel. And people started applauding when I said that.

01:01:53:27 - 01:02:18:22
Unknown
Feel helpful for Google Translate with much help from Google Translate, I was like one sentence, a German, and I and sometimes when I, when I share something, I do go to the trouble of looking at the translation. And I do change things. I think this is a sense there's, there's, you know, one week or so ago, use the word anger when I think what was meant was fury, right.

01:02:18:22 - 01:02:34:09
Unknown
It's things like that that there are is absolutely nuance. It gets missed. But do you get the gist of it? Absolutely. And you get more than you could have had. So, yeah. And what would you have had on the other side of it? Absolutely not. Nothing. So, you know, it's it's kind of becomes an accessibility, thing.

01:02:34:12 - 01:03:03:23
Unknown
Exactly. Good to have. You know, and as imperfect as it may be, good to have something versus nothing. Yeah. And then finally, the nano banana, apparently very strong this weekend, Google has been announcing that, Google Photos gets deeper integration with nano banana. So you can do things like if you have a picture of a few of you, you go in there and you can remove Riley's sunglasses, open the middle person's eyes and put a smile on the on the right person's face.

01:03:03:26 - 01:03:22:17
Unknown
So not that picture is absolutely perfect, even though it didn't get taken that way. But again, this goes back to like when we take a picture of something or are we capturing a memory, or are we capturing something that is specific to that exact moment in time? I think everybody has a different opinion on that. But definitely bending reality when you do things like that.

01:03:22:17 - 01:03:44:12
Unknown
But there's like people to do that just with an English sentence is amazing. What to do. Yeah. Yeah. Totally. Right. Like it's oh okay I know. Oh well. And especially like in that example what they're also showing off is in Google Photos. You've taught Google Photos. This is Kylie and this is this is Riley and this is Engle in this case.

01:03:44:12 - 01:04:12:26
Unknown
And so it knows. Oh well I put the smile on Engle. Who is this person that I've already identified. So it understands this deeper context. Bunch of other AI features and nano banana features coming to Google Photos. My mouse just stopped working. Okay, there we go. But also, and I thought, this is kind of weird. They announced that, natto banana coming into Google Messages because I think Google has this, this toy now that it wants to play with all of its products.

01:04:12:26 - 01:04:31:04
Unknown
And so they're like, well, let's put it into the messages app. It's kind of like Sora if you added in, right? It's not it's not video. It's not as elaborate. But yeah, it enlivens your social conversations. So if I'm having a text message thread with with you and you send me a picture, I could remix, it is what they call.

01:04:31:04 - 01:04:50:05
Unknown
So it's taking that picture that you sent me, remixing it and taking this picture of a dog in this example and turning it into a toy on a, on a desktop. So. Okay. Yeah. That's fun, I guess. Yeah. It's me. It's fine. Wait, if this is what people did with AI, I don't think apart from we don't know what's real.

01:04:50:07 - 01:05:08:13
Unknown
That aside, I think that we'll see it as a fun toy that can do amazingly creative things. And I think it would seem less threatening. Well, and I should also point out in this example you do see the little Gemini watermark. So you know, you know that this wasn't real because it was generated. So there you go.

01:05:08:15 - 01:05:27:00
Unknown
Nano banana coming to many different places, even places that you didn't expect it to happen. And then this last thing I threw in here, I just came across real briefly on Twitter, and I just thought it was kind of neat. Motion stream. This isn't really news as much as it is. Hey, if you're watching the video version, check this out.

01:05:27:02 - 01:06:03:00
Unknown
This is real time AI manipulation, in a video sense. So you can take an AI image of this dog and you have this focal point, and as you move it around the dog, all these frames are generated in real time to follow around your focal point. And I think this is just an interesting glimpse into, I think, where some of these tools for video generation are heading right now, where you can imagine somewhere down in the future, this is not going to be as latent as this example is showing, because it does take a, you know, like a half second for the dog to animate to where your cursor is.

01:06:03:00 - 01:06:27:23
Unknown
It needs to figure out those frames and generate them, and then and then insert and animate them. I'm sure we get to the point to where this happens in near real time, somewhere down the line, and this just becomes a really interesting avenue to explore. Yeah, I every year I get the activate, which is a consultancy company run by the son they forget his name.

01:06:27:26 - 01:06:52:20
Unknown
Executive I've known for. He used to be executive in charge of MTV. Michael Wolff and, not that one. The other one. And so, a huge long spreadsheet, I mean, PowerPoint of, predictions and media. I didn't put on the rundown because it's too, too deep to try to get into. But what struck me, Jason, is I will look at, my God, the number of applications, the number of companies, the number of startups that are in this space.

01:06:52:23 - 01:07:12:11
Unknown
Yeah. You know, I think we had a better sense of that in the early days, the internet, on our podcasting days because we'd see or we had we had no news site of the day. And, you know, yeah, I see new stuff. And now it's hard to it's, it's it's not as easy to grok exactly what one of these companies does, without seeing demonstrations and the kind of department.

01:07:12:11 - 01:07:32:01
Unknown
But, Lord, there's a lot of development going on out there. It's an insane amount of development. Yeah. And yeah, I love seeing examples like that because I'm like, whoa, okay, this this when I came across that the reason that I even wanted to show it off to begin with is because that felt like a glimpse into a near future where we're going to see more products like that.

01:07:32:01 - 01:07:51:07
Unknown
Like right now, people are going nutso about, you know, saw a two and Vo 3.1 and the taking an image and turning it into a video with sound. And now you can create video and then you get into the interactive element and, you know, get that close to near real time. And, and there's going to be some really fascinating developments around that in the coming.

01:07:51:07 - 01:08:10:19
Unknown
Yeah. And that this shows you said at the beginning we try to give you the, the latest news that we think is most important in the week and talk about it. But we also, in our research, have to stay ahead of what's happening. And so the paper, my beloved papers give you a sense that an a kind of fundamental research level of what people are working on to try to solve things like, like, yeah, tokens to vectors.

01:08:10:19 - 01:08:29:12
Unknown
Right. And then the application piece is really interesting because it takes what exists now and finds more and more new ways to improve it and use it. And there's a lot of investment and development going on in both those those tracks. Yeah, I'm pushing up against the edge and pushing it just a little bit further and a little bit further.

01:08:29:14 - 01:08:52:16
Unknown
Yeah. Fascinating world. Fascinating. Aspect of technology. And I'm happy we get to, talk about it each and every week on this show. Jeff jarvis.com. People should go there. And, preorder Hot Type. You can preorder now, right? You get links that you can. So yeah. Go ahead and preorder Jeff's new book, Hot Tape. You can also find a Gutenberg parenthesis magazine.

01:08:52:17 - 01:09:13:20
Unknown
The web we we've all presented in wonderful form on Jeff jarvis.com. Excellent work sir. You can go to AI inside dot show to check out all the information that you need to know about this show. You know, find our our RSS feed to subscribe the old fashioned way. Find us on socials, all of our episodes, audio, video.

01:09:13:20 - 01:09:37:10
Unknown
It's all there. AI inside dot show. And then of course, you can go to Patreon.com slash I inside show to support us on a deeper level, like so many patrons do, and especially our executive producers of the show. Doctor to Jeffrey Marrow, Chimney Radio Asheville, one of 3.7 Dante Saint, James Barnard, Derek Jason night for Jason Brady, Anthony Downs, Mark starker.

01:09:37:13 - 01:09:54:00
Unknown
We appreciate your support so much. And thank you. Yeah, thank you because we love being able to do the show each and every week. And with you that means that we can. So thank you everybody for watching and listening. We've got some pretty cool interviews lined up that we're working on getting booked before the end of the year.

01:09:54:00 - 01:10:08:22
Unknown
So you're going to see some special interview episodes appear in your feed. Don't freak out. Listen to them. And, so keep your eyes tuned for that. More to come in the coming weeks, and we'll talk to you next time on the AI inside podcast. Take everybody. Bye.