OpenAI Cleans House Ahead of IPO

00:00:00:06 - 00:00:09:05
Unknown
This episode of the AI insight podcast is brought to you by Airia. Get started for free today at airia.com.

00:00:09:07 - 00:00:42:28
Unknown
Coming up next on the AI insight podcast, Jeff Jarvis and I break down OpenAI's sudden shutdown of Sora and the billion dollar Disney deal that went down with it. Claude gaining the ability to control your computer, and a federal judge saying the Pentagon's anthropic ban looks like an attempt to cripple the company. You think? That's coming up next on AI inside.

00:00:43:01 - 00:01:02:12
Unknown
Hey everybody, welcome to another episode of the AI Insight Podcast, the show where we take a look at the AI that is layered throughout the world of technology. I'm one of your hosts, 50% of the show, Jason. How the other 50% is Jeff Jarvis? Actually, you're probably like 51% 8020. You're the 80 of the 20. Everything thing is that 8020 rule.

00:01:02:15 - 00:01:04:25
Unknown
You know that.

00:01:04:28 - 00:01:22:25
Unknown
Good to see you, man. Good to have you here. I hope you're doing well. Yep. And, yeah, we got we got some open AI to talk about. Apparently. I don't know that I saw this one coming. It's it's kind of an interesting, little story here about about the soar, app. Did you end up using that very much?

00:01:22:25 - 00:01:48:14
Unknown
Or if at all, the soar app? No, I think was very much, I don't think it even rose to the level of fad. Yeah, I think it was a short time gimmick. It was a very short term thing. Of course, Sora is the company's video generation app. Yeah, I mean app and service. Right before it was an app, it was just their video generation platform.

00:01:48:14 - 00:02:15:09
Unknown
And then they launched about six months ago. I gosh, is it already been almost six months? They launched an app that you could get on Android and on iOS and you could go to on the web, which kind of made it a little more like a social media platform for generating video. And I mean, by all accounts at launch, it was it was very popular, you know, it it hit the highest points, the app stores.

00:02:15:09 - 00:02:38:11
Unknown
Yep. You know, it did really well right out of the gate. And then now they've basically suddenly announced that they're shutting down the app. They're shutting down the API. Yeah. Sora. Sora, as a standalone thing going away, I don't know that that necessarily means that you can no longer use OpenAI's tools to generate video. Is it?

00:02:38:11 - 00:02:54:15
Unknown
Is it just, Well, I guess it maybe it would be. Or maybe you can still generate those things, but you can't tap into it with the API and you can't use the app. I'm not entirely sure what that means for video generation with it, with OpenAI tools, it's hard to believe they would get rid of video generation since it doesn't stand out.

00:02:54:15 - 00:03:19:13
Unknown
Apart from, that's robotic. I mean, we were warned about this to the extent that this that they're going to get rid of all these side things, they were going to concentrate on, on fighting anthropic. They're going to concentrate on business. And that is to say, on revenue and on ongoing value. In other words, I disagree with, but I think that there has to be a fair amount of whiplash inside.

00:03:19:15 - 00:03:45:02
Unknown
It's one matter if you're Google and they are. Yeah, we did this for a few years and never mind, it's gone. This was still pretty fresh. This is still pretty new. So it's, it's it's it's a it's a hairpin turn to get rid of something so young. Yeah, yeah. And I mean, video generation, it's it's an expensive endeavor when you're talking about these servers and GPU, very intensive GPU when you're using this service.

00:03:45:02 - 00:04:04:04
Unknown
I mean, granted, you know, most people were limited to a number of, a number of generations per day and everything, but it's still it had to add up to a lot. Or maybe it didn't, because, I mean, maybe people weren't using it or. Yeah. Was it that people didn't care and weren't using it? Or was it that it was too expensive or a combination of both?

00:04:04:04 - 00:04:31:00
Unknown
Probably. I guess a combination of both. Yeah, it was focus, focus, focus. I think, and and the big question for me now is whether open AI is a B2B or a boutique company. It started, I think, in great measure as a B2C company, it's all we're famous for. The most famous, we're the Kleenex of AI. People assume that it's awesome when it comes to AI.

00:04:31:00 - 00:04:59:01
Unknown
And we are we are chat, and so on. But that's a that's always been a hard business. You know, I know this from media, from training entrepreneurs and media. It's really, really difficult. It's really expensive. It's really risky to try to build a large open market of consumers, B2C, I mean, B2B rather, enterprise sales is a lot easier, more predictable.

00:04:59:03 - 00:05:21:07
Unknown
Yes. There is a time lag and trying to do the deals and do the lunches and all that. But if you get a big deal in enterprise, you get a contract. You know what's going to happen over the next 2 or 3 years. And it feels a lot safer. And so I think and, but, but, but B2C is always the stars before the eyes, where every famous everybody's going to know it's we're going to be on every phone, we're going to be on the lips of the nation.

00:05:21:09 - 00:05:40:18
Unknown
We're going to be everywhere. And and I understand that the seduction that occurs and is certainly occurred up in AI and they were yeah, they were B2C. They were the that's kind of that's kind of how they came about. It was, was the B2C. And yeah, that's interesting. But they never really had a shifting the matter that they they have a rise of revenue.

00:05:40:18 - 00:06:02:04
Unknown
That's all true. They raised bucket loads of bucket loads of money. But the strategy was never clear. I don't think beyond the you know, the general we're going to build AGI all that because I don't think there was a clear strategy saying here's where we're building value, here's how we're going to get recompense for that value, here's who our customers are.

00:06:02:08 - 00:06:25:17
Unknown
Here's what they could expect from us. Any company does that. And and, you know, it's, forgiveness of their sins. When Google started, they didn't know what their business model was. They didn't want ads. They knew, but they didn't have a clear mission to organize the world's knowledge and make it accessible. Right. I think I'm leaving out of Frasier.

00:06:25:20 - 00:06:50:06
Unknown
And whatever was there. So their mission was clear. Their business model wasn't clear. Open AI was, were a tool looking for a solution. And we're going to be the best at it, and we're going to go all the way to AGI, whatever that means. But they didn't have a business strategy. They had neither mission nor business strategy above you, which I'm trying to think in a sense.

00:06:50:06 - 00:07:06:14
Unknown
I would argue that there's a touch of meta about them, too. Is that Zuckerberg tomorrow? I've criticized them about this in my books and stuff. I don't think he ever had a North Star for Facebook that that better to say this is what we're building. This is what we are. We're the biggest social network. We have a social graph.

00:07:06:14 - 00:07:31:26
Unknown
Yet all of that. But to what end? Why? Google's mission was clear organize the world's knowledge and make it accessible. That's a mission. Anthropic has a whole constitution as a mission, which I think is a little over overdone, but I think it has more of a mission than OpenAI has. So I think OpenAI is really, flailing at the moment.

00:07:31:26 - 00:07:59:04
Unknown
What do you think? And and meanwhile, you've got the kind of push or the drive towards this IPO that is somewhere in the not too distant future, I think. And, or so it seems. Anyways. And so you're seeing right now especially this is what kind of makes this interesting, is that we're seeing a further kind of walk and walk down the path of narrow, of narrowing their focus.

00:07:59:04 - 00:08:24:26
Unknown
Right. I saw a video maybe that's, you know, a distraction from their core product so they get rid of it. You know, you've got, what else do they have here? They've got their shopping. That's another aspect of this, right? They're they're, scaling back on on the plans to turn ChatGPT into a shopping portal. There was that instant checkout feature that launched last September as well.

00:08:24:28 - 00:08:50:06
Unknown
Apparently that wasn't resonating with users. Wasn't resonating with merchants. Also, there were studies that were showing that users who were using it weren't really, you know, becoming great referrals for, for purchases to complete their checkouts or whatever. And so, you know, they're kind of trimming, trimming that off, starting to well, I think more than trimming that, Walmart is shutting down its engine of commerce with open AI.

00:08:50:08 - 00:09:17:21
Unknown
So that's another kick to the kidneys I think. Ouch. And the I'm trying to find the number here. I can't find it yet. Where the the sell through there was much lower than their own website and the presumption was going to be if you were an open AI and you were looking for something, you were just shipped, you're looking for something and you got to an endpoint that I wouldn't have that was going to be a very qualified customer, as they say in the biz.

00:09:17:24 - 00:09:39:07
Unknown
Yeah. Coming to you saying, I've done my research, I know what I want. I'm not going to a million places, okay, take me there and buy it. But, Uses the website Sparky through ChatGPT says the trip.com complete, purchases at about 70% of the rate of those who make purchases through Walmart.com directly, according to the Tech Buzz.

00:09:39:09 - 00:10:00:03
Unknown
So what does that say about the interaction? And and let me know in a chat bot situation then. Good question. A lot more transient or like low intention. It's like there is a lot of just kind of like, I think we're going to open this thing because I have this the spur of the moment, you know, curiosity or oh, I should check into that.

00:10:00:03 - 00:10:19:29
Unknown
But versus going to a website and saying, I'm going to look up that product, there's like a specific purchase intent when I open up Walmart to search a product or whatever, or and that's why it may be an unfair comparison. The better comparison might be search versus chat. Right. I'm in research mode and then I go oh what is it like.

00:10:20:01 - 00:10:41:27
Unknown
What's the conversion at Google versus open. I probably open that looks better. In that case if you go to the site you go to the site to do one thing you're going to buy. Well no I take it back, I take it back. I go to Amazon all the time for research. Right. I don't necessarily go there with an intention to buy immediately.

00:10:41:29 - 00:11:08:18
Unknown
So, so there's, so you're going there to, to learn more about a product because it. Yeah. Amazon does give you a wealth of information about products but which Walmart does to so. It's not a completely bogus comparison by any means if that's what matters to to Walmart. But, is Walmart willing to give a piece of its transaction to open AI?

00:11:08:20 - 00:11:36:07
Unknown
It's really a marketing cost, right? I was it content asked what was it. Advanced publications we started style.com and we started the first luxury store online where Neiman Marcus had the merchandise. We had the power to say what Neiman Marcus should buy. Believe me, I was not in on that because I'm not fashionable. The people who do were, and the crash hit and Neiman Marcus.

00:11:36:09 - 00:12:01:14
Unknown
And so we were sharing profits with the Marcus Neiman Marcus said, we can't we can't afford to share problems with anybody right now. But to the consumer, it didn't change at all. What change was the business model behind it that that you, Marcus, bought ads from got invest. They saw it then as advertising and then they could they could better, figure out the value of that.

00:12:01:16 - 00:12:20:21
Unknown
So in this case, if Walmart is going to deal with open AI in the store where I, I was going to expect a cut of any sales either directly on the per sale or the Amazon model, setting me that qualified customer. Correct. And if Walmart is saying they're not that qualified, they're not that good. It's not worth it for us to share right now.

00:12:20:24 - 00:12:43:29
Unknown
Never mind. So that's two, major kicks in the kidneys, I would say, for open AI. And there is a third that comes as the fallout from Sarah. They pissed off Mickie. That's true, that's true. Well, and this is this is also kind of an interesting aspect of this is the deal. You know, how how much was their deal worth?

00:12:43:29 - 00:13:09:08
Unknown
$1 billion, billion dollars billion dollar investment, deal with Disney, a three year deal, with Sara. And, you know, Disney's major, you know, characters and IP and all that stuff that was apparently happening. Apparently they were having, you know, the day before this, this news was made public. They were kind of in talks directly about this.

00:13:09:11 - 00:13:29:13
Unknown
Also, the day before the news, OpenAI had, published a safety blog about Sora called creating with Sora Safely. And then a day later, they killed the product. And so Disney is like, well, it didn't see that coming. You know, the probably the Sora team didn't see that coming. That was dealing with Disney the day before. Right.

00:13:29:20 - 00:14:06:12
Unknown
Too clearly they're releasing their, you know, their, their communications around the safety of the of the product and everything the day before, which just kind of paints this picture of, you know, was this decision hastily made or like, like how did the the mechanics of this happen behind the scenes is and I think the bigger question that it kind of points to that some people are asking anyways, is, is OpenAI in somewhat of a desperate position right now where they're kind of making these, these quick decisions and, you know, what does that say about kind of where they're headed right now?

00:14:06:12 - 00:14:27:05
Unknown
You know, it reminds me of a conversation that we have had, and you are the one to bring this up often, which is like, you know, maybe OpenAI has a short shelf life. You know, it's hard to believe that because they have been such a a major name in the beginnings of this kind of current moment. But maybe that's maybe that's what we're looking at.

00:14:27:05 - 00:14:49:16
Unknown
I don't know, I don't know that I believe, but yeah, well, I don't I'm not going to predict. I'm no futurist. That's the most ridiculous job title on earth. So no, I'm not going to predict anything, but I am going to raise the question of whether there's there's a you know, last week we did I forgot the other we did last week was, but this week there's kind of a death ometer for, for, open AI.

00:14:49:19 - 00:15:18:19
Unknown
And I wouldn't say it's, it's a doom and red doom yet, but there's a lot of questions. Yeah. Again, they don't have a clear strategy where they're going. They're they're waffling between B2C and B2B. That's a huge difference there. Scale. They're disappointing or outright pissing off the companies they're working with Microsoft, Disney, Walmart. Yeah. That's not smart.

00:15:18:22 - 00:15:45:22
Unknown
But the good of the show and everything else, believe it or not, Jason, I'm sacrificing here on watch. I can't stand Lex Friedman podcasts. I am watching a 2.5 hour interview. Lex Fridman and Jensen Wang. Yeah. No. Two. His interviews go long, but Jensen's Lex's. Oh my god, yeah, like three hour epics. You know, it's partly because he speaks so slowly.

00:15:45:24 - 00:16:15:00
Unknown
Yes. That's that. It stretches it out. Sorry. You don't know how hard that is for me to do. And so what's clear? One thing, though, and I'm kind of in the Jensen Wang fan club. At times it sounds like that. But I do admire his his strategic smarts. And I think he emphasizes his. You don't go to jets, you know, one goes to, Nvidia to buy a computer, like you can buy a computer, but in large terms, right?

00:16:15:03 - 00:16:33:03
Unknown
Yeah. He's embedded in the entire AI ecosystem. So he doesn't piss off anybody. In fact, he wants to be ahead of them or the way he the way he puts it is he wants to socialize the ideas he has so far in advance. The by the time he does it, people are saying, what took you so long? When in fact he's been thinking about it.

00:16:33:03 - 00:17:01:22
Unknown
So it's that's his conceit. But but he's in every competitor. He's in AOS, AWS, he's in Google, he's in Microsoft. He's in core. We've he's in all these companies. Well, Sam Altman and OpenAI do not have a friendly neighbor policy going on here. And I think that that if another company comes along with a potential big deal, what Nvidia was going to supposedly invest, 100 billion and now it's 30 billion, right.

00:17:01:22 - 00:17:32:10
Unknown
Or 20 billion, which is, add that to the list. And they'll talk friendly, but I don't think you trust OpenAI and and Sam Altman. Strategically, if you want to, if you want to try to do something together, his whole, I forget the name star. Star or whatever the we talked about this last week, the the huge infrastructure project, that he was doing to build data centers that's gone nowhere.

00:17:32:12 - 00:17:57:09
Unknown
So all of this is making things look a little difficult. Yes, indeed. Yeah. And you know, Edsa Tron on X had tweeted about kind of like this confluence of events and had said this is, you know, and, and specifically the fact that that OpenAI posted that creating with Sora safely the day before the news happens.

00:17:57:12 - 00:18:17:11
Unknown
And I think his response was something along the lines of like, this isn't what, company this is this is something the company does when things are going well. Yeah, I love British irony. It's the best. And now you add had all of that hope what I am saying. But we're going to be doing new things. Yeah they're going to go ahead.

00:18:17:13 - 00:18:42:24
Unknown
No no no. Are you are the super app. Is that kind of what you're alluding to. Go ahead. Yes. Yeah I mean I mean this is just an interesting kind of idea, I suppose that they're taking the products that I imagine they are deeming the products that are working, which would be ChatGPT the browser Codex, and they're putting it into a single streamlined experience, a super app to combine them all.

00:18:42:26 - 00:19:01:04
Unknown
And I don't know what the purpose of doing that is. Is it to tell a cleaner story, you know, ahead of the IPO or. I'm not really quite sure, but do you think that's a good idea that still strikes me is a B2C strategy. We're going to create an app that people can can rather download or use on the web, whichever.

00:19:01:11 - 00:19:20:16
Unknown
But we're going to create that, that's not a corporate B2B strategy. So they're still doing that. They're just trying to make it cleaner. At the same time, they're taking things away. They're taking Sora away, which I think is an issue. And at the same time of all of this, I'm all for says that they're going to double their headcount this year.

00:19:20:18 - 00:19:46:14
Unknown
Wow. That seems so counter. Right. In order to think about what are the possible the figures, the value has gotten so insane. There's no exit strategy. There's an IPO. That's the only exit strategy there is. Nobody can buy them the values to ridiculously investors. The valuations are just too high. And there's no merging. There's no, he's got to do too much.

00:19:46:14 - 00:20:10:10
Unknown
There's no, buying other companies to fill in the blanks. Really? So he's on a tightrope. Sam. Yeah. Meanwhile, he needs more compute, and he needs more power. So what's happening there? Well, here we go to the into the circular, economy. So open, I guess, because I didn't get a chance to to read this one. This is interesting.

00:20:10:10 - 00:20:41:06
Unknown
So open AI if it'll open for me. According to Axios. Is that a big deal with the company called Helion Energy? If that might sound familiar to the astute of you, that happens to be the company, the fusion startup that Sam Altman backed. Which means the Sam Altman had to resign has, stepped down as the board chair, and is no longer involved the company's board.

00:20:41:06 - 00:20:59:27
Unknown
So that if you believe if you're Altman believer, then that kind of devalues Helion a little bit. Yeah, it is fusion, so who knows where that goes. Is this open? I try to back up his bad investment as his, his bad as his investment, trying to buck up the energy needs of OpenAI. I can't tell because it's all too incestuous.

00:20:59:29 - 00:21:03:28
Unknown
It's all too circular. Yep.

00:21:04:01 - 00:21:24:20
Unknown
One giant circle that is the. Oh, and OpenAI, you, as all of this, I think there was some kind of story out there. I didn't put it in. I forgot to where we're coming to the Yann LeCun paradigm question two, where OpenAI AI is the ultimate super scalar scale. Scale, scale. We got us all there, right?

00:21:24:24 - 00:21:45:25
Unknown
DeepMind has already backed off that a bit. Yann LeCun and Feifei Lee are obviously are not, that view or the sense of scale levels. And now you have, Altman brothers real quickly, said,

00:21:45:27 - 00:22:07:03
Unknown
I can't find enough about it. But but he acknowledged that, we got more to do here. It's not just going to be the low levels. We get us there. We need paradigm shifts. Sure. So which which I think is true. But true. I don't know. I'm not. It's not in the red button and the green, but but.

00:22:07:05 - 00:22:31:06
Unknown
Yeah, but it's but it's not looking amazing either. And maybe it ends up being, you know, that that little things like. Or little things, maybe they're not little things, but things like this become a blip. But yeah, we don't really don't really know. There is just like a little whiff of I don't know if panic is too strong a word, but it you know, there's, there's a twitchy factor which going on.

00:22:31:09 - 00:22:52:03
Unknown
Let's go inside of the company right now. And that that seems to be what I think a lot of people are picking up on is like, okay, so what is the thought behind this? And what does it mean about the health of of open AI right now? Yeah, yeah. Interesting stuff. Well, that is, that is open AI in a nutshell is at least as far as this week is concerned.

00:22:52:03 - 00:23:11:25
Unknown
And my apologies. Suddenly I while you were talking, I had to mute the mic because, I don't know if, a massive pollen thing just blew through the window or whatever, but I just started sneezing like crazy. So if I sound stuffed up, it's because my allergies are suddenly the buds on the trees right outside my window are all there now.

00:23:11:27 - 00:23:27:20
Unknown
Oh, man, it's just been nuts for me. The last few weeks. And it's been so warm here. So I think it's just kicking things into bloom and. Yeah. So I'm going to try and not be all nasally through the rest of this episode. We'll see how I do want to throw a quick thank you to our patrons.

00:23:27:20 - 00:23:49:07
Unknown
Patreon.com slash AI inside show. We've got some amazing patrons, including Mark as Dorian and Michael O'Hara, just a couple of our amazing patrons who support the show week after week. We appreciate you and thank you so much for, for enabling us to, do the show and a regular, consistent basis. Couldn't do it without you.

00:23:49:14 - 00:23:49:27
Unknown
All right.

00:23:49:29 - 00:24:18:25
Unknown
I'm going to take a quick break and thank the sponsor of this episode. And it is area. Area. Been talking about these guys for quite a while now, and I'm super happy to have them on board because, what they provide is very interesting. If you're if you're wanting to kind of play around with these agents that we talk about on the show all the time and kind of build your workflows and not just fire up, you know, open up a lab and put in a query, but construct kind of more complex.

00:24:18:27 - 00:24:49:19
Unknown
Workflows that do multiple things with different, you know, LMS and different tools and MCP and all these things. Well, that's what area is all about. Bring in AI into your company is it can be really exciting, right? The potential, the benefits. But it can also feel risky in between data privacy concerns, security threats, this constantly evolving compliance, requirement landscape, it can be easy to then, you know, hesitate and not act on it.

00:24:49:19 - 00:25:15:12
Unknown
But that's where Aria comes in and can really save the day. Area provides a unified security layer with advanced threat detection and strong governance tools, and is designed to protect your AI ecosystem from day one, so you can actually safeguard your sensitive data. You can monitor the activity and stay aligned with compliance standards, all while continuing to innovate at speed.

00:25:15:14 - 00:25:42:17
Unknown
So instead of slowing down because of that uncertainty or not getting started all together, you can actually move forward with clear guardrails and real confidence. And you don't have to choose between innovation and security to do it. Aria helps you do both at the same time. Pretty great stuff. You should check it out if risk has been the thing that's actually holding you back from going bigger with AI, well, consider this your green light visit area.com to learn more.

00:25:42:19 - 00:26:08:29
Unknown
You can start a free trial and check it out. That's AI aria.com pronounced Aria. Visit aria.com and check it out for yourself and see what it's all about and see how it, you know, the whole process and the whole layout and the structure of everything just makes sense. Aria.com. And we think Aria for, the continued support of the AI inside podcast has been great, having you on board, the last few months.

00:26:08:29 - 00:26:24:29
Unknown
So thanks for supporting the show. All right. We're going to take a quick break. Come back and, talk about clod. A few changes to Claude that are making things a little more automatic and giving much deeper, control over your computer that's coming up here in a moment.

00:26:25:01 - 00:26:51:20
Unknown
All right, anthropic has a couple of features that seem related, and I guess they are to a certain degree, but they are actually different from each other. They're they're unique. Announcements. The first one is probably one that more people can can appreciate and enjoy. This is a change coming to Cloud Code and Co-work that will allow you to, give it access to your computer.

00:26:51:26 - 00:27:12:09
Unknown
So, you know, everybody's got, claw on the mind right now. So I think this is this is like Claude's claw. I don't know, you know, wants wants to have its own claw solution. So this is how they're doing it. It gives you, it gives Claude code and Claude Co-work access to your computer. So could open files.

00:27:12:09 - 00:27:32:20
Unknown
It can control your browser. It can run developer tools. That's all to start off with. And of course, it's not perfect when you sign up. If you're watching the video version, on YouTube, when you sign up, you get this, you know, definitely a big, a big, kind of call out to let you know, you know, some of the things that happen here can't be undone.

00:27:32:20 - 00:27:58:22
Unknown
So be careful with that. Some of the apps that you approve could open other apps that you haven't approved. So, you know, it might find its way into an app that you haven't explicitly given it, permission for. Of course, you know, websites and docs could contain malicious instructions like prompt injection, that sort of stuff. Although anthropic has said that they've built in prompt injection, protections here, but nothing's perfect.

00:27:58:25 - 00:28:22:10
Unknown
And, and also just a general warning to close anything that might be sensitive because clod can in this sense see your screen. And, you know, it's initially when it kicks off into a task that you've worked with it to do it. Well, at first tap into the connected services. So things, you know, like connected links underneath.

00:28:22:12 - 00:28:48:05
Unknown
So maybe Google Workspace or Slack if it has a connected service that it can connect directly to, it will do that. If that doesn't work, though, and it still wants to accomplish this task that you've given it, then it will operate the machine that you're using like a human. So with a keyboard and a mouse entry, the same that you would and you can actually my understanding is you can actually watch it controlling as a lot of these things do.

00:28:48:06 - 00:29:06:07
Unknown
So so it's kind of a nice fallback. It's like if we have the direct connection, we'll do it there. If not, then we'll go to the new, clearer route. We'll take over your machine and do the things, in front of you and get them done that way. And with your permission, of course. So. So there you go.

00:29:06:07 - 00:29:34:04
Unknown
Are you going to be, clogging your computer anytime soon? I got to keep it figuring out what it is I want to do with it. That's the thing. I don't have any things I'm trying to figure. So I've got another task of mind. I've got to think it through first before I ask what to do. As I was listening to the Johnson Wong interview, about an hour and 15, in Lex Fridman, halfway there, put it out, put it on, like, one and a half times speed.

00:29:34:09 - 00:30:13:23
Unknown
I need you to that. That's true. Especially when Friedman is. Yeah. Talking. A little slow. So when Nebo clock came. Yeah. When they wanted to develop that one. Because as we discussed last week, just one went not on or over the clock, and wanted to, to join in immediately. He said they called in all bunch of security people and the rule they came out too, which is interesting to me, is too, is that he says it's secure because they you get to choose two of three rights the ability to access sensitive information, the ability to execute code, and the ability to communicate externally.

00:30:13:25 - 00:30:40:26
Unknown
Never. All three. Never like all three simultaneously. There were all three, right? There were on the same function. Right. The on your clock can do two of three things. Like not all three. Right. That's why it keeps secure, which I kind of get though. It has access to sensitive information and access to communication. That sounds dangerous, but it can't execute code.

00:30:40:28 - 00:31:00:12
Unknown
If it connects you code but and it can communicate but it doesn't have access to sensitive information. It can still do some bad stuff. And I guess it's not as bad. So I think the struggle now for, for, for anthropic and company is to get your agent to functionality going. And how do you make it safe?

00:31:00:14 - 00:31:23:25
Unknown
And you can't now 100%. Yeah. I mean, you know, are we getting used to as as regular kind of, users of the last couple of years of these systems? Are we getting used to the lack of certainty with these systems and what they're capable of? What do you think? I don't know, like, I think there's a part of me that's kind of like, yeah, okay, I get it now.

00:31:23:25 - 00:31:48:14
Unknown
Like, if that's just the way it is, like, should we always strive for these things to be as close to 100% as possible? Absolutely. But I also get a lot of benefit out of these things, knowing full well that's not going to get everything right. And you know that there are other ways that I address that. And just kind of just kind of accepting that, like it's an imperfect system and there's lots of imperfect systems in life and it's just get, you know, hope it gets better.

00:31:48:14 - 00:32:11:26
Unknown
But I understand that it's this, you know. Yeah, I think I guess what I'm saying is, for me, I think I've kind of hit a point where I, I don't automatically see the imperfection as a reason not to, you know what I mean? So it's it's risk. It depends on what it could do. You know, the dog ate my homework is now replaced by the AI by email.

00:32:11:29 - 00:32:35:28
Unknown
Yes. Right. Or deleted my folder. You know exactly all my important documents in it. Exactly. I had a paper that that I read last week. That looked at it was very interesting about about how agents were operating together and around and, they kind of read team them. And one interesting example of what, where and what did really well, somebody did really badly.

00:32:36:00 - 00:32:56:11
Unknown
And they were arguing the researchers were arguing for the need to discuss just these risks. My favorite example of where it went off the rails was that what it was told to erase one email, it couldn't figure out how to do it. It just killed all the email. No. Got no. And so they all went away. It was like, I can't I don't know how to delete mission accomplished.

00:32:56:13 - 00:33:13:00
Unknown
You told me to kill it, I kill it. I did what I had to do, not kill it. Yeah. It's gone. Did I do something wrong? So where do we get. You know, it's not just when you have one agent doing things. It's when the agents become a society of agents. Yeah. It's true. It gets really, really interesting.

00:33:13:00 - 00:33:44:19
Unknown
Does it? Where does responsibility laws. Yeah, yeah. Oh. Can you, can you the channel. Yes. Can you program responsibility in accountability. Yeah. Can you regulate this. I don't know. Yeah. Yeah. That's a that's a real good point. I hadn't, I hadn't really considered the fact that, like, you know, just dealing with one system and being okay with the imperfection layer, let's say, of that system.

00:33:44:22 - 00:34:09:12
Unknown
But when you're talking about agents and agents talking to other agents and all negotiating with each other, boy, that that exponentially broadens the, the potential kind of impact of of those, those getting things wrong. Yeah. It does. Yeah. That's a really interesting. Well, this is a research preview. So it's, you know, it's something you can check in into if you choose.

00:34:09:12 - 00:34:42:18
Unknown
You have to actively turn it on. You only get access to it if you're a paid Cloud pro or Mac subscriber only on the Mac OS for right now. Yeah. I don't know what I would do with these either. I'm sure if I had them installed and just went all in, I'm sure I'd find some things because I've seen, you know, I've, I've certainly heard of some interesting, use cases of the Claw technology for people who are a lot braver than I am opening, you know, the gate, the floodgates, and just being like, all right, you know, I, I still firmly believe that if I did this, I know just enough to, like,

00:34:42:18 - 00:35:03:27
Unknown
totally be dangerous. You know, and not enough to, like, safeguard myself against the majority of things. And, I would lose stuff, but, like, I have, you know, a huge I mean, this is this is so baseline. This is so basic, but it does actually sound appealing to me. I have a huge folder of, I don't know, a couple of thousand files in like a downloads folder.

00:35:04:04 - 00:35:29:04
Unknown
And I have, you know, a screenshots folder that is absolutely filled and to go into that folder and to just say delete all like I, I'm sure that I'd be leaving some things that I don't want to delete. So having something like this go through and automatically just like sort it and put it into some sort of sensible sorting so that I could then decide what to delete and not would be useful.

00:35:29:04 - 00:35:47:01
Unknown
But am I willing to like, risk other things suddenly disappearing or what? I don't know what would happen, you know what I mean? Like the things I don't know that could possibly happen is a risk. I don't know that I am. I'm not quite there yet. Yeah, yeah, but that sounds useful. You know, these things have to have.

00:35:47:04 - 00:36:07:27
Unknown
There's no control, Z. Well, they just right now. Yeah. Well that's that's another thing. Right. Like if it, if it deletes your emails, like when I hear you know about that. Oh shoot. My entire inbox has gone. I mean, like, I know my Gmail. If I delete an email, it goes to the, the trash. Like, would that not be the case in that situation?

00:36:07:29 - 00:36:26:23
Unknown
But like files on my desktop, if I've got a solid state drive and, you know, I've noticed a lot that when I delete a file, it doesn't automatically go into trash anymore. It's just absolutely gone. And so, you know, maybe it would be on and yeah, if it raised it application, it all went with it. There's no there's no control Z.

00:36:26:23 - 00:36:53:14
Unknown
Yeah. What do you do. Yeah. You can't undo that right. Yeah. Well that's not the only thing that anthropic announced. They also announced, a new feature to Cloud Code, another new feature to cloud code called auto Mode that gives the model, well, autonomy, to run shell and coding actions without constant permissions prompts. And I think this is really targeting, developers who are constantly running up against those permissions.

00:36:53:14 - 00:37:13:16
Unknown
That kind of slows them down. And some people are going to feel comfortable to just be like, no, look, I just want to give this thing permission to do all these things and, you know, cross your fingers and Godspeed. But, you know, anthropic says, you know, it's probably a good idea to kind of keep your work in sandbox environments and all that kind of stuff.

00:37:13:16 - 00:37:35:14
Unknown
Maybe. Maybe it's a good idea to keep backups of this stuff when you're working in a tool like this. No, maybe about it. It's a good idea. So. So that's an interesting, also kind of kind of related but kind of different. What do you think about arm getting into, making its own chips? They've got, you know, they've been doing for 36 years now.

00:37:35:14 - 00:37:59:24
Unknown
They've been licensing their designs for other big players. So for Nvidia, for Apple, other big companies. Now the company has announced its first chip, the it's its first chip, the arm AGI CPU, which couldn't help but you know, have a little eyeroll with the AGI being thrown in there. You got to get that terminology and I suppose. But co-developed with meta.

00:37:59:25 - 00:38:27:29
Unknown
So meta is going to be getting these. I think others are going to be getting it to Cloudflare. I believe Cerebus launch partners. Yeah. What do you think about ARM releasing its own chip for the first time? So this is this is intended for, inference. Inference. Yes. And that comes off what we saw last week with, Justin Wong's keynote at, the Nvidia event.

00:38:28:01 - 00:38:55:20
Unknown
And so that's that's the heat right now is everybody get all the inference. And so it makes some sense. I don't I've never understood the chip world well enough. Yeah. Me to really get it. That do you did you say that they don't have a they don't have a fab, do they. Are they, are you factoring them or were they or are they just talking about the stat they used to just sell standards and they're going to sell off all designed.

00:38:55:23 - 00:39:18:00
Unknown
Yeah. They used to design for others and now they're producing their own chip and they're, they're you know doing that with somebody fat elaboration. Yeah. So yeah. And you know they're marketing it and selling it themselves instead of creating these designs and these chips for others essentially it's my understanding. But but I'm but I like you Jeff.

00:39:18:02 - 00:39:39:24
Unknown
The chip world is something that I like. I try and try to really understand on a deeper level. And and you get you get into terminology soup with these things. And I just don't have a direct, a direct knowledge of, of that stuff other than, you know, that I know that it's important. And I know that these are big players.

00:39:39:24 - 00:40:00:06
Unknown
I know that this is a big moment for arm. People are really, you know, very excited for ARM to kind of be at this point after the amount of time it's been in the industry, finally kind of doing doing its own chips for itself is a big moment for the company. And it's actually a solid time. Right. CPUs are kind of in shorter supply right now.

00:40:00:06 - 00:40:25:19
Unknown
There's a lot of demand. PC prices are rising this, moment with, you know, AI agents kind of increasing in demand. All this is, you know, pulling on the demand for chips like this. And so it's a good time for ARM to get into it. Yeah. I think my understanding, you know, especially during the claw era that we find ourselves in to.

00:40:25:20 - 00:40:54:00
Unknown
So, and the list of their on their press release, they list, the Cerberus, as you said, Cloudflare, meta, OpenAI, positron rebellions, SAP, SK Telecom, are the their partners for this. You wonder whether it is Samsung happy about this? Is Nvidia happy about this I don't know. Yeah. Yeah it's a good question. They're all they're all behind.

00:40:54:05 - 00:41:16:21
Unknown
So they're all looking for a piece of the pie. TechCrunch article also points out, you know, it's interesting that ARM is producing a CPU, not a GPU. You know, in this in this AI moment, but maybe they will have that. And, yeah, I wonder if we're going to see a lot more AGI CPUs and GPUs coming from ARM.

00:41:16:24 - 00:41:54:21
Unknown
Got to throw that terminology in there. The Trump administration in last week released a national legislative framework around policy framework around artificial intelligence. And, I mean, this has been an evolving story of course, but, this framework is all about centralizing regulation at the federal level. So they're basically saying, like, we don't want the states to get in there and muck up with anything we we want to be competitive with, with places with countries like, like China in AI.

00:41:54:21 - 00:42:18:18
Unknown
And that means that we want unfettered AI, development in the States. You can't get in there and enact your own AI rules and fragment the whole market place and all that kind of stuff. So that's a big part of this. So this I read, I read the framework, and it's written by grown ups. It's it's not insane.

00:42:18:20 - 00:42:42:29
Unknown
It has seven key pillars. The first is protecting children, empowering parents. Got that right. All the more focus on that right now because as we're speaking, as we're recording this show, the jury came back in Los Angeles, and meta and YouTube lost the, jury trial about, addicting children to social media. Oh. Oh, okay.

00:42:43:01 - 00:43:00:17
Unknown
So that's God knows what impact that's going to have on section 230. What has on those companies particularly, and it puts the, you know, along with social media, I mean, social media feels so decade ago, doesn't it? Yeah. Yeah. You know, does regulation the courts are catching up now. Or if we're past that, we're into the AI age.

00:43:00:19 - 00:43:22:24
Unknown
But this stuff will have a we'll have a blow over there. So, the pillar one also gives a plug to Melania Trump for protecting children. Okay. Pillar two is safeguarding and strengthening American communities. Saying that, Congress should ensure that residential taxpayers do not experience increase electricity costs because there's competition with electricity from the data centers.

00:43:22:26 - 00:43:56:27
Unknown
I'm not sure exactly how you do that. Streamline permitting for building the centers. They threw into here, federal enforcement against, AI enabled impersonation and fraud, national security, resources for small businesses. Okay, that's number two. Number three is protecting intellectual property rights. The this one really interests me because it said that, it talked about protecting them in regards to output, not input.

00:43:57:00 - 00:44:28:12
Unknown
It says and I quote, although the administration believes the training of AI models on copyrighted material does not violate copyright laws, it acknowledges arguments to the contrary exist and therefore supports allowing courts to resolve this issue. Congress should consider enabling licensing frameworks for collective rights systems for right holders to enable to, you know, make it more efficient, a federal framework for protecting individuals from unauthorized distribution of commercial use of AI generated replicas of voice.

00:44:28:16 - 00:44:59:02
Unknown
Like this kind of stuck that in there too. They follow all the places of the stuff. And monitor the copyrights. Okay. That's that's three for, preventing censorship and protecting free speech. Well, there's a lot of definitions about that these days, aren't there? There is. They cover the gamut, that's for sure. Yeah. So included in this is Congress should provide an effective means for Americans to seek redress from the federal government or agency, efforts to censor expression on AI platforms.

00:44:59:04 - 00:45:34:15
Unknown
That's loaded in ways that I think we can't fully predict. Number five, enabling innovation, ensuring American AI dominance. The pretty straightforward get rid of regulatory, problems, create sandboxes, provide resources for six educated Americans and developing an AI ready workforce. Oh this is well, if it's going to eliminate jobs, what do we do about it? Seven establish federal policy framework preempting cumbersome state AI laws so it doesn't outlaw state AI laws, but says that this is a principle anyway of federal government.

00:45:34:21 - 00:46:00:28
Unknown
The federal law should preempt state law. The federal should have a privacy, policy. And these are the rules, this section. But but things like that, that that should be set at a federal level, which actually the technology companies wouldn't disagree with to that extent, because having to do it 50 different, regimes is difficult, for sure.

00:46:01:00 - 00:46:22:18
Unknown
But it said that, preemption must ensure that state laws do not govern areas better suited to the federal government. So it's a it's a common statement of Trump saying no laws against no laws whatsoever. Yeah. Okay. That's so that's there's there's there's booby traps in here. There's a few land in here. But it depends on how whether it's implemented by Congress.

00:46:22:18 - 00:46:42:05
Unknown
But the Congress agrees to any of this, how it's then implemented and regulation. Who knows. But I think it's worth. It's worth it's a quick read. It's four pages or three pages, actually, with the cover, sounds, and it's worth looking at anyways. It sounds more reasonable than I think a lot of people are giving it. Yes.

00:46:42:08 - 00:47:15:10
Unknown
Okay. That's cool. I didn't see a whole lot of reporting about the, you know, I didn't either. And I guess this is just indicative of the way the stuff works. But so much reaction to the states the kind of the, the aspect of it around, you know, state laws and just basically drawing the line in the sand and saying up, this means no state AI regulation will be allowed or whatever, but that that seems to allude to there being some wiggle room and have, you know, having a reason and a rationale for that.

00:47:15:10 - 00:47:37:01
Unknown
So it's easier to get, a little nuance into a three page document than a, yeah, whatever they call the tweets. Yeah, yeah, indeed. Indeed. How do you feel about the whole, the whole copyright aspect of this, the fair use I've set off on, on the show that I think, I think training is fair use and transformative.

00:47:37:03 - 00:48:02:07
Unknown
And I think in terms of output, there do need to be, negotiations. And I've urged the news industry to come together and make that possible by creating APIs for news. And news isn't as important as we think it is. Trading is over as an issue. A great measure because I'm going to go back to quote, just long ago, he said most of the data that's going to be trading systems from now on is going to be synthetic data anyway.

00:48:02:09 - 00:48:24:07
Unknown
They're pretty much they've train their models. They know how to train them. They've used up the data that we have. Training data is not an issue anymore. They already got what they needed. Yep. So the question is, if you want to call upon current, information and be able to feed it back in different form,

00:48:24:10 - 00:48:40:23
Unknown
That isn't necessarily a violation either. Right? If it's only on the open web and you read it and you you summarize it, that's fine. That's transformative. It's a different use. But there's clearly going to be ongoing fights about this. So I think it's just trying to say, can we can we find some piece? Can we all get along?

00:48:40:25 - 00:48:59:18
Unknown
Why can't we all just get along? There you go. Okay, cool. Thank you for running through that. That's that's super helpful. That gives me a a better understanding of kind of the nuance of it. And, yeah, they, they say in the coming months we'll, we'll be learning more about how this would be implemented. And what that even looks like.

00:48:59:18 - 00:49:25:26
Unknown
So, definitely threw this in here for you because I'm super curious to get your take on this story, which is Google running a small and narrow test that uses AI to rewrite search results, news headlines. This is kind of reminded me of, you know, it's, clearly it's not the same thing, but of, like a tech name sort of thing where tech me regularly pulls in all of these news stories.

00:49:25:26 - 00:49:52:28
Unknown
But then they have editorial staff to rewrite the headlines so that it, you know, makes sense. On the tech meme website. They're not taking the headlines specifically and putting them on there. They're kind of doing their own summary thing. Really. Good point. Google. Google's kind of doing something similar here, although maybe because they're using automated systems. It's different, but they're shortening or rephrasing the headlines to ultimately to improve engagement.

00:49:52:28 - 00:50:29:13
Unknown
So for example, I used to cheat on everything AI tool and it didn't help me cheat on anything would be the the source headline that in this, in the case of this article and search engine land becomes cheat on everything AI tool. So it removes a lot of nuance from it. You literally have no idea. Yeah, fluff, but at the same time, like, it doesn't it doesn't tell me kind of like anything about perspective or, you know, is this just an informational thing about this AI tool, or is it something happened to this AI tool or anything like that?

00:50:29:15 - 00:50:49:04
Unknown
In this case? Anyways, gets completely ejected. And is that more. And I'm not sure whether that's more engaging. I think you're telling me more is excellent because take me as a brain a better, more, more informative headline. Yeah, I want to get the other out of the clicks and then it gets me all the clicks, right. Gives me a huge, sometimes a huge lump of of things to click on to reward.

00:50:49:04 - 00:51:05:00
Unknown
Right. It's terribly useful. In this case, I think one of Google's arguments is that they're probably going to try to increase engagement, as you said. Yes. The other argument is that this enables Google to personalize. If they know that you really care about this one angle of it, they can bring that angle of the story out and put it there.

00:51:05:00 - 00:51:38:10
Unknown
So I can't do it right. I know from a lot of data, the personalized, headlines in email newsletters, perform a lot better, when it comes to, open rates. It just stands to reason. So I understand editors getting pissy about this. I also understand the danger, because a headline that had legal nuance like including the word alleged right in front of it, right in front of a, accusation of crime could end up getting taken out.

00:51:38:13 - 00:52:02:22
Unknown
And, does does the fault for that then lie with Google or with the brand? It's going to fall to the brand whether it should or not. Just people can say, well, the Washington Post said the Washington Post didn't say it. So that's a risk that Google takes on. I think, but was also true is that so many of our headlines are clickbait, trying not to fully inform you totally.

00:52:02:22 - 00:52:21:09
Unknown
Oh my goodness, I see this every single week. For for prepping for Android Faithful. Actually, when I'm going through all the news for for Android faithful to to pick through for that podcast, I think like 90% of the headlines I see now are kind of like, I used this blah, blah blah, you know, never naming the thing.

00:52:21:10 - 00:52:39:06
Unknown
Right? Always alluding to, you know, it's super clickbaity and, yeah, like, I, I wouldn't mind those being a little bit more direct because it's hard for me to tell. Like, are you just telling me your opinion on something or is there actual news here? You know, it's really hard to determine with the headline as it gets ruined as a result.

00:52:39:06 - 00:53:03:24
Unknown
But also the lead is dead. You go to these stories and, you know, simple, stupid example of like when I go to the Google News stuff on the last thing I go to in my in my news diet is a Google News thing, and some of these headlines are just ridiculous. This is ridiculous. I will say this, this, this beloved food chain is closing and it will act like it's, you know, is this something one of the ones I was a Taco Bell or something I care about deeply.

00:53:03:24 - 00:53:24:17
Unknown
Yeah. Right. And a it's nowhere near me. It's in Texas. Why do I care? But be. I don't learn that. Or the name of the, of the food chain until the sixth, seventh paragraph. Right. The worst. It's horrible. It's just horrible. Horrible. So the old the the the lead was there as a as a production necessity in newspapers.

00:53:24:17 - 00:53:51:19
Unknown
I've written about this. Yeah. But it also was a consumer service and that's gone. So in a sense, you know, I'd love to see enough. Usage of this to see whether the consumer experience is better with Google or not. Yeah. Yeah. They, they, they could be providing a more direct thing out of this. And, you know, it could also just not go any further than the test that it is right now, which.

00:53:51:19 - 00:54:09:11
Unknown
Yeah, which they do often, you know, which they do all the time. Exactly. They're constantly, you know, playing with things. And so, you know, I know some people's immediate reaction on this was, oh God, can you believe Google that. That's horrible. That it's like, yeah, they've tested a million things that were probably bad ideas. All the companies do.

00:54:09:13 - 00:54:30:16
Unknown
And, and they test them with a very small percentage. And many of them never roll out. So. Okay, relax. We'll see. We'll see what happens. Cool stuff. Well, real quick, if you're enjoying this show week after week, please leave us a review on Apple Podcasts. We do appreciate it. It really does help, spread the word.

00:54:30:16 - 00:54:50:27
Unknown
And, you know, you get featured on our website when you do that. I inside dot show. So, thank you for doing that. We're going to take a quick break, get back, on the other end of that and talk about a few quick hits, some, speed round items to get you out of here, including Apple's, upcoming Wwdc and what that might mean for AI on iPhones.

00:54:50:27 - 00:54:53:05
Unknown
Coming up in a moment.

00:54:53:08 - 00:55:16:27
Unknown
All right. A federal judge said that the Pentagon's ban on anthropic tools, looks like punishment for the company's decision to. Yeah. Think, yeah. Fashionable security designation and contracting dispute with the DoD. Yeah, exactly. It's like, okay, well, you're you're telling us what we already knew, but at least someone said it.

00:55:16:27 - 00:55:44:08
Unknown
At least a U.S. district judge said it. Of course it wasn't San Francisco that woke judge in San Francisco. It, quote looks like an attempt to cripple anthropic. According to U.S. District Judge Rita F Lin in San Francisco. So. And that would violate the company's First Amendment rights. That's the important part. Lin was at a point here by sorry, Lin was a Biden appointee.

00:55:44:10 - 00:56:14:26
Unknown
Got it. Okay. All right. Oh, so so then totally woke. Very well. Is basically what you're saying. So just go woke out. Yeah, yeah, yeah. The judge will now decide whether to grant anthropic temporary relief from the ban. As this all moves forward, but. Okay, this is the latest development. There you go. Apple is set for the Worldwide Developers Conference Wwdc, taking place June 8th through 12th.

00:56:14:29 - 00:56:42:16
Unknown
And Apple did include the headline AI as part of the headline anyways. Or part of the announcement, that they're going to focus on AI advancements. Yes, you're going to get your platform updates for iOS, macOS, tvOS, watchOS, all the things, all the developer tools and software and everything. But a focus on AI advancements, which seems kind of like an acknowledgment of, you know, all that stuff we've been working on that we keep not not rolling out the we promised a ways way.

00:56:42:18 - 00:57:04:27
Unknown
And that didn't end up happening. Well, that's going to be coming at Wwdc. So, Bloomberg's Mark Gurman says, you know, the Apple whisperer, as I sometimes call, says that Apple's big Siri overhaul is going to come through this announcement. That makes a lot of sense. Or there's, apparently going to be a standalone app. Also an Ask Siri interface.

00:57:04:27 - 00:57:23:22
Unknown
So you can use your voice, you can use your text, and then it integrates all of the data from, opt in, of course, because Apple is all about privacy and stuff. But, integrating personal data from messages, email notes, apps, all that kind of stuff. And yes, Gemini would be underneath all a lot of this, if not all of it.

00:57:23:24 - 00:57:47:20
Unknown
So which may be just fine. Yeah, totally. And I wonder if it's just fine if Apple is just like, okay, then we don't need to do anything else. Let's just stick with Gemini running all this stuff. Don't be surprised. Hey, if it accomplishes the goal, great. Meta is secretly building an AI detector inside meta AI. This is according to some code sleuthing.

00:57:47:22 - 00:58:08:24
Unknown
That was done. Some feature flags that were found that seem to allude to a, yeah, some sort of a service on a meta side that users could upload their content, or share their content into to see if it was generated by. I. Of course, this is interesting because meta, you know, creates that AI content with its own tools.

00:58:08:24 - 00:58:29:16
Unknown
But I guess Google does this too, right? Like it has the detector and and stuff for its own tools as well. Yeah. I mean, my, my Facebook feed is just awful now. Yeah, yeah I agree. So I wonder if this is the sort of thing that you can apply to the feed or if it's a separate app. Yeah.

00:58:29:17 - 00:59:01:04
Unknown
Don't, don't really know. But it has been called out. What is this. This is the ex post that kind of points to it and says oh I detector. It's there but it's not usable. There's no server side switch that's been that's been flicked on to make it work quite yet. So but yes, Gemini has a similar detection mechanism that works on its own AI tools in that case, as well, Google has said that they want to bring broader, setup to support, coming somewhere down the line.

00:59:01:04 - 00:59:26:25
Unknown
So who knows, meta might do the same. This, I believe, happened while we were doing the show or right before the show, so I didn't have it marked in there. But, Google launches Elyria three Pro. It's its new music generation model. It had recently launched a, let's see here, a 32nd long support with the Elyria three.

00:59:26:28 - 00:59:48:07
Unknown
But now the Elyria three Pro model can do up to three minutes long, music tracks. So there you go. If you want to create even longer music tracks with, Gemini's AI, you can. I haven't played a huge amount with their their music generation, but the what I have I've found it produces some pretty generic sounding music.

00:59:48:11 - 01:00:06:15
Unknown
So yeah, it's not not winning any awards. But you know, once they can get things to sound a little more original, a little more cutting edge, if I could upload my own music to it, like have it remix it and stuff, that's that's the stuff that really gets me excited. So I think it's more about the background for commercials and stuff.

01:00:06:15 - 01:00:26:23
Unknown
Totally. Right. That's yeah, it's music beds. Like, you know, if you're if you're a YouTube creator, for example, you could create a music bed that's never meant to be featured, but just meant to kind of be way back in the early days, my son of of of the net and webcams and stuff, there were, there were programs I used and taught students with that had teleprompters.

01:00:26:23 - 01:00:48:01
Unknown
And you could make a video on your, on your, on your apple. And it was cool stuff. And they had a few, standing, rights free pieces of music. And they were really cheesy and awful. And I was listening to a commercial, when I'm driving and I listen to a mass now on Sirius, there's Sirius.

01:00:48:01 - 01:01:11:24
Unknown
All the ads are there. So there was some cheap local ad, and it was that same cheesy exact music I heard. Yes. Yeah, 15 years ago from from that app. So if we can improve on, on, on, the commercial beds, fine. No harm, no foul. No, I totally agree, I totally agree. And finally, I didn't have this on my radar.

01:01:11:24 - 01:01:43:09
Unknown
But you put it into the rundown, and, this is an upcoming AI documentary called the AI doc, or how I became a and what how do you process? And, the optimist, an apocalypse optimist. Is that an optimist of the apocalypse? Is that what that means? I guess so it's, it's moral panic, pure out. And it has, right now we're seeing, you know, Kowski who's a boomer, and, Tristan Harris, who's the ultimate moral entrepreneur.

01:01:43:09 - 01:02:07:16
Unknown
He was always against social media and the internet before, and now he's gone against AI. And, you would have to, drug me to watch this. I'm I'm amazed that this is going to theaters like, this. Seems like a Netflix documentary to me or something. I don't think it's gonna be there long, but I think it's it's it's tied to moral panic and and and there's a market for that now.

01:02:07:18 - 01:02:29:08
Unknown
Yeah. There for sure is. Yeah. Some of the, some of the quotes from here, we need to take the threat from AI as seriously as global nuclear war. Was one Tristan Harris said, I know people who work on AI risk who don't expect their children to make it to high school. I mean, come on. So.

01:02:29:13 - 01:02:52:16
Unknown
Okay. Yeah, the the AI doc coming to theaters in a couple of days. I think you can have my seat, folks. Well, Jeff won't be there, but he will be working on his book, so I, I linked to Jeff jarvis.com, but I'm realizing I need to work into my rundown to link to your new work here. I need to get that on to Jeff jarvis.com too.

01:02:52:16 - 01:03:11:11
Unknown
I got to get my son to help me. Oh yes. Okay. That would that would be good as well. So, Yeah. I know you spoke about it a little bit last week. So in addition to AI and humanity is a as a new book series from Bloomsbury Academic that I'm editing. We announced the first three books, but I'm also looking for other authors and ideas and books.

01:03:11:11 - 01:03:39:11
Unknown
And, the idea here behind the series is that it's about people from many disciplines who reflect on AI and how it reflects on society. It's not about the technology per se. It's about society on the technology. And so, you know, what is what is learning makes us reexamine things like that. What is creativity? So there's that in addition to that, then, hot tape, my book about the Linotype is now delayed.

01:03:39:13 - 01:03:56:18
Unknown
Did I tell you that, because there's a, there was a production thing and they're going to move it from June and July? I said, no, July is dead time. So now it's the end of August, beginning of September. And so it's a fall book, which is fine. So it's just been better sales time. Yeah. So you can preorder any time now.

01:03:56:20 - 01:04:18:29
Unknown
There you go. Get in there, preorder and then be very be very patient through have your wonderful summer. And on the other side of summer, you have something on the other side of the hot summer you'll get hot type type. Right. Exactly. Thank you. Jeff. Jeff Jarvis was I inside dot show for everything that we do, including yes, your reviews of the podcast.

01:04:18:29 - 01:04:43:00
Unknown
They are listed there. But, all our interviews, all our episodes, everything you need can be found there. And then, of course, patreon.com slash AI inside show. Oh. I'm logged in as a as a different Patreon account. So you're seeing a weird interface. You wouldn't normally see this if you went there, but Patreon.com slash I inside show, we've got some amazing executive producers as well.

01:04:43:03 - 01:05:04:11
Unknown
We've got Doctor Dew, Jeffrey McKinney, radio actual one of 3.7, Dante, Saint, James Bonnet, Eric, Jason, neither, Jason Brady, Anthony Downs, Mark starker and Carsten Smoky. Thank you so much for your support. Thank you everyone who supports us on the AI inside, Patreon and for, you know, watching, listening, reviewing, doing all the things we couldn't do it without you.

01:05:04:14 - 01:05:20:26
Unknown
And thank you, Jeff. So great hanging out with you. I don't forget, if you're a podcaster that wants to make it better, if you're a company that want to think about getting into podcasting, Jason Howell has a shingle that's out there where he can hire you because he knows his stuff. Thank you. I keep forgetting to mention it.

01:05:21:01 - 01:05:40:10
Unknown
Yeah, I do have a website pod tune up.com if you're, you know, you're thinking about getting into podcasting or you have one and you just want to make it better, hey, I've done it for a long time. See, that's me a long time ago. Doing podcasting. It's like Babyface Jason. Yeah, that was a long time ago, that's for sure.

01:05:40:13 - 01:05:55:17
Unknown
Anyways, I can help you with your podcast, so pod tune up.com. Thank you Jeff I always forget to do that. Got I got to be self get your horn out and blow it to truth. All right. Thank you everybody for watching and listening. We'll see you next time on AI inside. Take care. Bye.