Anthropic's AI Got Shut Down by the Government

00:00:00:05 - 00:00:26:14
Unknown
Coming up next, Jeff Jarvis and I make sense of how anthropic s most powerful AI models ended up offline after a 90 minute government deadline. Also, SpaceX hits $2 trillion with its record setting IPO and immediately buys cursor for $60 billion. Although we saw that one coming and AR glasses from snap, that is Snapchat snap and X real.

00:00:26:21 - 00:00:53:06
Unknown
The ones that I've seen before. You can finally by those this fall. We've got some new information there. Oh and all birds officially an AI company now. More details there. We have so much coming on this episode of the AI Inside podcast.

00:00:53:09 - 00:01:11:25
Unknown
Welcome to another episode of AI Inside the Show, where we take a look at the AI that is layered throughout the world of technology. The AI world keeps turning the the the globe keeps spinning. I suppose in the world of AI, I'm one of your host, Jason Howell, joined, as always on the other side of the country by Jeff Jarvis.

00:01:11:25 - 00:01:35:12
Unknown
Good to see you. Hey, boss, good to see you. Yeah. So the the globe is layered and spins. Yeah, I don't know. I don't know where I was obvious that. And then the I don't know you know, you start digging into the ground. You see all the different layers of AI right. The Precambrian layer. Alabama. Yeah. It's exactly the planet Earth.

00:01:35:15 - 00:02:13:23
Unknown
Good to see you. Good to see you. Lots, lots to talk about. As you since we last spoke. Yes, I know, kind of like what happened here. What? What the heck is going on? We got the anthropic mythos, fable debacle, saga, whatever you want to call it happening. And boy, did it happen. You know, and I have to say, because I was getting so used to having access to fable, and I was starting to really understand how I was going to get maximum benefit out of it before it likes the payment kind of thing switched and it went away, and then I'd have to pay a bunch for it.

00:02:13:23 - 00:02:32:12
Unknown
And so I had kind of mapped out like my, my approach, like what I wanted to do with it. And then as we know, poof. Was there an error message? Was there a oh, sorry, folks, we have to tell you, we've been banned. How did you. Or it just didn't work. So I didn't I didn't initially see the news.

00:02:32:12 - 00:02:54:13
Unknown
I went into the cloud app to use to to start using it, and my app still had fable in the toggle. It was still kind of selected by default because I had been working with it the last time I had used the app, and so I put in my prompt and everything, and then it just kind of gave me a quick error message when I tried to run it through, and I was like, oh, that's that's weird.

00:02:54:15 - 00:03:14:10
Unknown
I'll just try it again, you know? And then I tried it a couple more times, and then I was like, let's go to Tech meme and see what's going on here. And I went to Tech Meme and saw that everything was upside down. It had, you know, launched on the ninth, which we had talked about last week.

00:03:14:10 - 00:03:46:00
Unknown
We discussed this three days after it launched. This is Friday, June 12th, last Friday in the evening, 5:20 p.m. Pacific. Anyways, I think it was PM Pacific. Anthropic gets a letter apparently from the Commerce Department. The letter says that anthropic needs to suspend all access to fable five to mythos five. For those of those who had access to that by any foreign national, whether inside or outside the United States and the government.

00:03:46:05 - 00:04:05:18
Unknown
It turns out over time we've been getting like all this extra information. So feel free to correct me if I'm wrong on any of the sequencing or whatever. Jeff. But sounds like they had about 90 minutes to comply with this. They looked at that, basically said, well, we can't filter out foreign nationals from American users. We can't do that in real time anyway.

00:04:05:18 - 00:04:27:19
Unknown
So they effectively just pulled both of those models offline for everyone. Poof went away. Jason was sad because he couldn't do the work he wanted to do, and they've been gone ever since, you know? And yes, and ever since there have been updates to the app. Once I reboot, you know, it gives you a little a call out that says fable is not accessible right now or whatever.

00:04:27:21 - 00:04:51:01
Unknown
Coming back soon, we hope, we hope. It's like it's like those restaurants, you see that dust all over them. Clearly, the door hasn't been opened in ten years and it says closed for renovation. Yeah, right. Wow, that's quite a renovation. What do you. Yeah. What are you waiting for? The times just put up a story this morning about the the 90 chaotic minutes inside of.

00:04:51:04 - 00:05:10:11
Unknown
Oh, well, the Washington Post had one of the times that number one and they're, they're, they're going around on their slack saying, what the heck are they going after us? What's the problem? What do we do? Are we being bullied based on bad vibes? One employee asked in a chat viewed by The Times. I have bad vibes. No, you kind of true.

00:05:10:13 - 00:05:35:23
Unknown
Kind of true. Yeah. They've been they've been. You know, I got mad. I go back and forth on anthropic, you know, and I've said this on the show before where at one point I'll say, wow, they're doing good work. For the next minute they'll do something crazy, like the doom stuff they did a week ago. Totally. But here you got to have sympathy for them because they're getting just jerked around.

00:05:35:27 - 00:05:56:25
Unknown
However, there were a lot of people who said you did this to yourself. You had I was going to say, okay, the ones that were saying, you know, our model is so strong, our model is so impressive. We need a tighter kind of controls. And here you go. So, you know, and actually didn't, didn't Yann LeCun, friend of the show, mentioned something about you reap what you sow.

00:05:56:27 - 00:06:22:11
Unknown
It's kind of like, well, you asked for it, so here you go. Bruce Schneier went for the kind of the next step and basically said, you know, get used to this, folks. It's not just mythos. Every model is going to be able to do what mythos did. So Pandora's box is opened. That's where we are. Yeah. And you better understand that powerful AI is now here.

00:06:22:11 - 00:06:49:18
Unknown
And if you ban it in the US, which would be a really stupid thing to do. But, you know, China is going to have it, France is going to have it. Somebody is going to have it. Well, I mean that that that development continues while we don't have access to this. And so a lot of the critics of this move are saying, you know, all you're doing once again, is giving, you know, countries like China, the ability to catch up, the ability to surpass.

00:06:49:21 - 00:07:23:06
Unknown
Yeah, it's a it's a tricky position. Now what is the what is the involvement of of Andy Jassy. This is the Amazon CEO apparently according to the Wall Street Journal, Andy Jassy is the one that on Thursday, June 11th, last Thursday, basically said that that that changes needed to be made, that an Amazon researcher had tested feeble safeguards, found a way to jailbreak it, and that the jailbreak was as simple as something along the lines of fix this code.

00:07:23:09 - 00:07:48:10
Unknown
It was a really simple prompt that apparently got around the jailbreak, which is a whole other can of worms. But so is this Andy Jassy. I mean, basically he's calling, you know, bringing this to the attention of what? To anthropic to say, I think to the white House, to the white House to say, hey. Right. Okay. So then is it a matter of Amazon pointing at anthropic and saying, see, they're doing bad things.

00:07:48:12 - 00:07:53:05
Unknown
That's what I can't. Yeah. And you know, it's tattletale.

00:07:53:07 - 00:08:15:04
Unknown
Snitching ratting out. I mean, pick your pick your mobster reference here. Yeah. It seems on the one hand, I don't know what I think. On the one hand, I think it it just the way all these people are in technology in Silicon Valley, whether they're in Seattle or Silicon Valley, they all feel kind of ratty, mobster like, in a way.

00:08:15:06 - 00:08:33:17
Unknown
On the other hand, if it was in fact dangerous, was it the right thing to do? Andy? Jesse, did you know everybody said, hey, stop this right now? I don't know yet. Yet. The thing that it did, my understanding of the thing that it did is fix this code is it was more of like a defensive move, right?

00:08:33:18 - 00:08:54:16
Unknown
Like it wasn't a command to, like, execute an offensive cybersecurity thing. It was take a look at this code, tell me how to fix it, which is kind of what a lot of which is what to do. Anyways exactly what every model does. Yeah tons of them do. Yeah. You got a really good point there. Yeah. So I don't know whether any Jassy is a villain in this tale.

00:08:54:18 - 00:09:23:11
Unknown
Yeah I need to know more to decide. But it feels a little dark. And and once again it is. And reap what you sow. I think anthropic has been a fool all this time to be so duma oriented and I'm sorry. I'm going to open my laptop so I can find it, but I can't get the video of us to mute.

00:09:23:11 - 00:09:45:10
Unknown
So I'm hoping this mutes. Is that what. That's what I was doing? Yeah, sorry about that. All right. It was. There was a really good column. I forgot to put it in the rundown. There's a really, really good columnist warning in the New York Times. Let me find it that I linked to, profile notifications by Cal Newport.

00:09:45:12 - 00:10:12:29
Unknown
And it's the doom. The doom trolling needs to stop. Okay. He's a computer science professor at Georgetown, author of Deep Work. And it's really, really smart and really good. Saying that anthropic recently dropped a classic of the form of the doom trolling, a scary sounding report titled when AI Builds Itself, the company, blah blah blah blah. The public relations.

00:10:13:00 - 00:10:38:22
Unknown
The public reaction to this report focused on a section that seemed to call for a worldwide pause in AI development, but if you read more carefully, it becomes clear that a pause isn't actually what anthropic proposes. Its report says that if it were possible to slow down the technology, then we should. But so long as the least cautious actors were advancing in full speed, it suggests that anthropic will have no choice but to do the same anthropic BS, like a cat leaving a dead bird at your doorstep.

00:10:38:22 - 00:10:58:11
Unknown
Anthropic catalogs the grim futures that his products might produce, shrugs its shoulders, and then returns to his furious efforts to make these warnings a reality that's isn't alone in this nihilism. And so is it. That anthropic can't make up its mind, then? Is that kind of no is doing this to promote itself? This is how powerful we are.

00:10:58:11 - 00:11:26:22
Unknown
We can destroy the earth, right? Right. And, he compared the company's large language model efforts to the development of the atomic bomb. This is Sam Altman now, and posted an image of the Death Star from Star Wars. So he's not just going after anthropic. He's going after all of them. Let's call this strategy doom trolling. It's one of the defining and most arresting properties of our current AI moment, and I've come to believe that it's morally indefensible.

00:11:26:24 - 00:11:42:29
Unknown
And it's a really good column again by Cal Newport. I put it up for a gift link, by the way, folks. So if you want to go to my my feeds, you can find it. And I think that's the that's that's what, that's what we should be discussing. Now anthropic will find have to find a solution to this.

00:11:43:01 - 00:12:11:10
Unknown
They'll put up something that's crippled. Meanwhile this has hurt all of American AI because, you know, people are saying in countries around the world, we know our own native AI. We cannot depend on America. This is too dangerous to risky in all kinds of ways. It can be cut off with no notice. And so Mr.. All and France, I'm sure, is having a very good week right now, and the Chinese are having a good week right now.

00:12:11:10 - 00:12:44:21
Unknown
And so this this hurt everybody. I think it was certainly an overreaction on behalf of the white House. But it was also anthropic state. Set the stage for this. What do you think? Yeah. How much do you think anthropic at fault. Well, I mean, like exactly what you said. Anthropic is the the voice that continues to point at all the possible kind of negative aspects and kind of doomsday or doomsday outcomes.

00:12:44:27 - 00:13:05:10
Unknown
So I guess you shouldn't be surprised when something like this happens, whether you like it or not. I just don't understand how they don't see something like that coming. Yeah, I don't either. I want to read, if I may, one more paragraph from from this up end. As a computer scientist and a digital ethicist, this is him talking.

00:13:05:10 - 00:13:27:22
Unknown
I am both optimistic about the possibilities of AI, and confounded by the terrifying and grim way that current technology leaders insist on talking about it. This could have been a period of hopeful innovation, but instead, our emotions are being manipulated by Silicon Valley's self-serving and morally untenable addiction to doom trolling. This communication strategy has to stop the harm it's causing to the public's.

00:13:27:22 - 00:13:48:24
Unknown
The public's mental health has arguably outweighed the benefits that AI has so far delivered. And if you look at we talked about this in the show in the past, all the opinion polls that come out of people who are anti I now, they the Nimby going on about data centers, the hostility to it. People are running political campaigns anti-air.

00:13:48:27 - 00:14:24:19
Unknown
They bought this for themselves by trying to show how powerful they are. And it's male ego at its worst. Well, yeah, kind of along that line I saw I didn't read too much into this. I didn't have the time to read it, but I think it correlates with what you're talking about just from a sentiment perspective that WordPress basically had a had a survey that says 60% of US consumers are saying that AI in a brand message is a turnoff, is not necessarily going to get you a lot of excitement on the other side.

00:14:24:19 - 00:14:53:10
Unknown
And, I mean, I, I think that personally, I think that squares up with a lot of what I hear from different people. I don't know, it's really interesting to me because from a sentiment perspective, like we've talked a little bit about some of the things that I'm working on outside of the podcast world, and one thing that I'm getting a lot of interest in right now, both in Petaluma, where I live, and then also in Healdsburg, which is a town, you know, pretty a town with a lot of money.

00:14:53:11 - 00:15:24:28
Unknown
Not too far from here is I have two different opportunities to explore doing, like AI workshops. And so I'm kind of starting to put together these workshops. But in both cases, what I'm hearing from people and what I'm hearing from the person in Healdsburg that I'm going to be working with on this, is that it's such a mixture of like the same people that are saying, we want to use these, we want to know or not that we want to use these, but we want to know more about how to use these are the ones that are also equally like kind of turned off by it all.

00:15:24:29 - 00:16:10:10
Unknown
It's kind of people are begrudgingly relenting on on AI. And I think to your point, a lot of that, that messaging that gets put out does kind of feed the fear, which is very much a a symptom of media discourse in 2026. And where we've been the last, I'd say the last ten years, you know, things have really built up to this point where so, so many of the the marketing messages and things put out there do kind of tap into the fear element because it gets reaction, it gets involved, it gets people involved or integrated in some way, shape or form or interested and, you know, things like this.

00:16:10:10 - 00:16:31:17
Unknown
This is probably part of the downside of that is that then people don't see things through a clear mind when it comes to this stuff. And, you know, companies and I think there's phases shouldn't be surprised about that. And the venture era of a company. This doom stuff works. It gets them venture capital. They love this powerful male stuff.

00:16:31:18 - 00:16:51:06
Unknown
Right? Then it'll be interesting to see as all these companies go public. Well, that's different then. They've got to deal with a large number of shareholders who have a different view of this and could lose faith in it. Now, it's just it's not unlike politics, where all you need is a core number to do this. And the truth is we're all going to own we all by the end of this week.

00:16:51:06 - 00:17:08:12
Unknown
We're all going to own space if you have it for one, because it's going to be in the index funds. Yeah that's true. So it's not as if it's it's really in our choice. But you know what's interesting to me. So what I've learned I'll spare you the plugs right now. I'll do it later in the show. But I've learned in my books that I've written.

00:17:08:12 - 00:17:30:23
Unknown
Is that what I constantly say is the technology inevitably fades into the background. Once, once it's demystified. Once people learn how to use it, the technologists and their power fades in the background. People take it over with, happen with print, happen with electricity and steam, happen with. In a way, this is the the substantiation of that happening in the broadcast.

00:17:30:23 - 00:17:50:16
Unknown
I was just thinking before we got on today, it's amazing. Today you can make, hey, we're going to make a show and you can totally. So but AI is the one where it's going to happen faster than any of the others. I think because it is intended, it is designed to work in our own languages, to our commands, and it's going to be easier than any of the others.

00:17:50:16 - 00:18:14:03
Unknown
So right now it's this, this high pitched level of priestly doom. You can't understand the power of all this. This is beyond your Ken for mortal human. And that doesn't square with the way people are actually using it. So the people you're talking to, finally, to your point is, the one hand, they're hearing all that doom stuff saying, oh my God, it's more.

00:18:14:03 - 00:18:40:11
Unknown
But the other hand, when they sit down and use it, they say, oh, wow, that's cool. That's simple right? Yeah. It's such a mixed yeah signal. And yes, obviously making a large language model is complex and requires great computer science skills and math skills that most people, including me, do not have. I'm not I'm not diminishing the difficulty and the impressiveness of making a frontier model, but once it's out, we can all use it.

00:18:40:11 - 00:19:07:13
Unknown
So the problem is, they took a technology that was potentially the most democratized technology we've ever had and tried to keep it in the priesthood, try to rob it to people, try to depress them about it as as Calverton says. And so we got to get this stuff out of their hands. They're screwing up here, but they've got all the resources, they've got all the money they've got, and boy, do they have the resources.

00:19:07:16 - 00:19:24:25
Unknown
Yeah, I mean more in many ways. And we'll talk about space, you know, in a little bit. But yeah, it's it's ridiculous the, the numbers that we're talking here, the of these resources also just the interconnected web of those resources and how much they rely on each other, which we've talked about many times. What is this free fable thing.

00:19:24:25 - 00:20:06:15
Unknown
This is so I want to mention that. Yeah. So, so there was an open letter two days ago from dozens of security people led by Alex Demos, who I respect immensely, now chief product officer at corridor. And it's a free on transparent AI cyber protections. So it's to the to let Nick and the National Cyber Directory Cairncross, we, the undersigned executives, technical leaders across the US and his allies write to you to ask you to lift the export control directives on fable and mythos and commit to an open, scientific and transparent process of handling AI risk assessments in the future.

00:20:06:17 - 00:20:31:13
Unknown
We believe that AI is having a significant impacts on cyber security, including greatly reducing the difficulty of finding flaws and writing exploits for those flaws and mythos class models are quite good at finding flaws and weaponizing exploits. However, they are not uniquely good at these tasks, and many of the undersigned individuals regularly use other foundation models and open source models for security audits and red teaming every day.

00:20:31:13 - 00:20:55:25
Unknown
So they're saying, you know, it ain't alone and is built multiple protections into its fabled model. Well, that it is essential to provide AI coders and security teams so that they can work with it. The Chinese open weight models are only months behind American models, and so we're focused on determining whether a human prompt section of code was insecure and on and on.

00:20:55:25 - 00:21:23:04
Unknown
So let me see here. As a result, this action has taken the best models away from defenders, creating market uncertainty, and risked America's AI leadership without any risk to justify it. So they want a future structure that is one grounded in scientific evaluations, two created through a democratic rule making process, three enforced transparently and fairly, and for used only to the minimal extent necessary to ensure the safety of the American public.

00:21:23:05 - 00:21:54:16
Unknown
Very same letter. None of this six months stop us. I'm going through the names Brian Dorf, who's an open source pioneer I respect immensely. There's a lot of people here who I think are really important and really good. And yeah, I wish that sane heads would listen to them. Yeah. I think what that what that speaks to me about is this question that I hadn't considered, which is where where is the line drawn then?

00:21:54:16 - 00:22:15:24
Unknown
And how do you know that you've found it? Like, okay, we've talked a lot about mythos, amazing ability to do all these things as cyber, you know, cybersecurity and blah blah. And a lot of people have have had access to that and would agree. But like, how do you make that a tangible thing, comparatively speaking, to other models?

00:22:15:24 - 00:22:41:00
Unknown
I'm sure other models are also capable of finding insecurities and code bases. Maybe not to the capacity or to the the effective quality of mythos, but I'm sure that they can too, if given the right prompt or the right amount of time, or the right number of subjects, or whatever the case may be. So then how do you look at one model and determine that one's dangerous, and look at another model and determine, oh no, that one's okay.

00:22:41:00 - 00:23:01:10
Unknown
You can go ahead with that. And that's kind of what I learned from that is like what is the scientific basis for saying this one's too strong. This one people don't, you know, don't shouldn't have access to versus this one. It's okay. You know, whatever. It's the Pandora's box problem. Yeah. If this model today isn't too strong, the next one tomorrow will be.

00:23:01:16 - 00:23:20:03
Unknown
Yeah. You find is strong. The other thing you're highlighting is it's not a rule. It's a process. It's going to require smart people to look at it and say, how could bad actors use this? But this goes to the argument I make all the time about guardrails. You would have to anticipate every possible malign use I'm going to put to it.

00:23:20:03 - 00:23:41:22
Unknown
And and you can't. You just simply logically cannot. So you're going to and in this case the guardrail was it was strong apparently. And it was broken by fix this code which is something that all these models do. So yeah, I think it was an innocent thing. You know, it reminds me of Matt cuts. Our friend Matt cuts who?

00:23:41:24 - 00:23:59:12
Unknown
You folks may not know, but he was in charge of beating down spam. You should thank him for many years at Google. Brilliant at it. And it was always, you know, one inch behind because they would come up with a new trick and then you had to be aware of it, then you had to solve it. And that's going to be the case here.

00:23:59:12 - 00:24:23:06
Unknown
But this is going to be mightier and speedier, and it's the reality we're in now. We got to just accept it and figure out what a world looks like, where things have to be made secure enough for this world, and wait till you get to quantum and wait till you get to know GS. The ability to get through any encryption, how we're going to protect things in the future.

00:24:23:06 - 00:24:41:22
Unknown
I don't know, I just saw a story today that has nothing to do with AI, but beware sending checks through the mail. And then the safest thing. Well, no, there's now there's ways they can get easily. Get rid of the ink on the check and then put in their own amount and their own payee. So it's now less safe to send a paper check.

00:24:41:24 - 00:25:01:01
Unknown
Okay. That is online. Right. Always step ahead. All these chicken exact cat and mouse game and every method. Nothing. Nothing's fully safe. Nothing can be fully safe. This is the world we live in. It's always been the world we lived in. We fool ourselves to think it can be safe. And so thus, the doom talk doesn't help. Yes.

00:25:01:09 - 00:25:33:07
Unknown
Interesting. Fascinating. Well, this is also. Yeah, this is also the first time that, I believe that a government has forced to publicly deployed frontier AI model offline after launch. There's you know, of course, there is the precedent setting kind of, you know, concern that I think a lot of people have because because the facto de jure, the US government is now regulating AI because they did it and it did it.

00:25:33:10 - 00:25:59:01
Unknown
And so then what does that you know, how does that feed into. You mentioned sovereignty and AI sovereignty. This is the week of the G7 summit. So you know the timing is interesting here. You've got the UK, you got the EU, you got Canada. You know, all these countries using this model and there, you know, researchers and everything using the model.

00:25:59:01 - 00:26:25:09
Unknown
And suddenly they don't have access to it to. So it reaches outside of just in other words it's a US decision or US impact. But that that touches everyone. And this really played into an interview that I recorded with Joel Pinot, who was the former VP of the AI research at meta in the Fair Research department, which we talked about many times.

00:26:25:10 - 00:26:53:29
Unknown
Right. Joel now is chief AI officer at cohere, and I spoke with Joel on Monday. And so we've got a full interview coming to the feed later this week with Joel is a fascinating, excellent interview all about kind of her history, but also about the models that cohere. And we talk a little bit about sovereignty and AI. And so I pulled a little clip because it was kind of perfect opportunity.

00:26:54:00 - 00:27:20:05
Unknown
The sovereignty is something as a Canadian company. This is based in Toronto. Sovereignty is something that she talks about a lot. And this kind of this anthropic moment with mythos seemed like the perfect kind of data point for for that and the, the impacts of that. And so I want to play a clip. It's not a long clip, but a clip with me talking with Joel on Monday about this topic.

00:27:20:05 - 00:27:45:10
Unknown
So hopefully it plays as I expect it for for those who are listening and don't know Canadian Canadian bass or Canada bass. So Toronto right. That's Toronto headquarter where you're where your HQ is. But there are some kind of us like tendrils. I don't know what the right word is. You've got you know, you got US shareholders, US executives, capital routed through Delaware.

00:27:45:10 - 00:28:11:04
Unknown
If is there any concern here with what's going on with with cloud and anthropic that, that you might not be insulated from such an event if it were to go that route? Is there a concern? Absolutely. And there's a concern. And so much as we are all dependent on each other in terms of building this technology, we all have to adhere to the laws of the countries in which we operate.

00:28:11:05 - 00:28:32:27
Unknown
You know, the US government has the ability to set the laws within the country, as do Canada and the other countries in which we we operate. We we use a lot of technology from other companies. You know, in some cases we do deployment with other models and so on. And so that is definitely a dependency that that we have that we take.

00:28:32:29 - 00:28:54:06
Unknown
On the other hand, you know, we build as much as possible full stack solutions so that our customers have choice and agency. And there's amazing technology coming out of the US. And so, you know, we will continue to to build on that. But there's a lot of people around the world who also want to have more control over their technology stack.

00:28:54:06 - 00:28:58:16
Unknown
And we're going to develop the solutions that they need.

00:28:58:18 - 00:29:28:13
Unknown
So yeah, she goes into more detail on it, but I think, I think this is just one of those examples of the diversity of what what companies like cohere are offering is even more important because then you aren't fully reliant reliant upon one single thing or your clients anyways. You know, your customers aren't. Well, tell me if I'm wrong, but she's Canadian, so that means she's a polite and decent

00:29:28:16 - 00:29:49:12
Unknown
If she were of a different culture, it almost sounds like she should be saying this is an opportunity America just tripped over itself and people are going to have this need, and we're there to provide this need no matter what America does. Yeah, we're there to give you the, the menu. And oh, that one went down. Well that's okay.

00:29:49:12 - 00:30:11:19
Unknown
We got all this other stuff for you to fall back on and everything. Yeah. It's a great conversation. I'm sorry. I'm sorry you couldn't be there. Jeff. You had a. I was testifying in Trenton in our state capitol, wearing a tie. Wow. Okay. Is that something you want to talk about? So Montclair State University won the bid to take over public television in new Jersey.

00:30:11:19 - 00:30:32:12
Unknown
And I've been on the periphery of that effort and working with Montclair State on things. I think it's very exciting. Some saw losers were there trying to change the process, but I hope that that gets moved around. So I was there to say that I got I got to sit in front of the the committee and say that what I concentrate on now is the death of mass media.

00:30:32:15 - 00:30:36:16
Unknown
What will replace it? I found that fun.

00:30:36:18 - 00:30:58:09
Unknown
Awesome. Well, yeah, I'm happy you were. That's why it was. It was a good excuse, folks. I had to be in the state capitol. Solid. Excuse. Solid. Excuse me. I do want to throw a thank you to ozone nightmare. Once again throwing a super chat our way. Thank you. Appreciate it. Says Happy Wednesday gentlemen. I agree that anthropic brought this upon itself.

00:30:58:09 - 00:31:21:27
Unknown
It relates to the larger issues with these companies over hyping their tools. Histrionic hype leads to elevated response. Yeah, exactly. You can't you kind of can't have both. Right. So this is a good example of that. So thank you Joe. Thank you ozone nightmare. And then real quick before we take a break with just one more quick thing from anthropic.

00:31:21:27 - 00:31:41:16
Unknown
And then I'm not sure how much we talk about them through the rest of the show. So we'll just fit it in here right now. But they were planning a pretty big billing change for the Claude Agent SDK. It was supposed to go live on the 15th, which was just a couple of days ago, and they have decided to postpone that.

00:31:41:19 - 00:32:04:13
Unknown
I don't think they're deciding to not do that. I think they're just deciding to put it off. Now. As you can imagine, they got a lot of things happening right now. Maybe not the right time to continue to make their developer community mad by raising prices on how their. This was certainly, certainly a factor in it. I mean, with the government's action, but it was also OpenAI is out there hinting that they're going to start a price war for tokens.

00:32:04:15 - 00:32:32:16
Unknown
Yeah. And meanwhile you have Deep sea, which has already started a price war for tokens. Yeah. So anthropic I think was feeling chuffed and top of the world and we've won for now. And it was on that basis they were going to raise their prices. But then a lot of rugs got pulled out from under. Yeah indeed. Well so there is that you get a little bit of extra time, which is not a bad thing.

00:32:32:17 - 00:32:52:05
Unknown
I'm sure there were some people looking at that point on the calendar and gone, things about to get really pricey. Who knows when they're going to actually, you know, decide to have to make to make that a change officially. And what that will look like, considering, like you said, that OpenAI might kind of change the rules a little bit.

00:32:52:07 - 00:33:13:15
Unknown
Hey, we do have an amazing Patreon. If you want to support the show on a deeper level and encourage us to do even more episodes like this, and I want to throw a quick thank you to a few patrons Chris Spackman, Steven Kimbrough, and our newest patron, Robert Lemos, just a few hours ago. So, Robert, good to have you on board.

00:33:13:17 - 00:33:31:01
Unknown
Yes, indeed. Patreon.com AI inside show. You get access to every episode that we do here, including the interview episodes with no ads. So you get all those ads removed. If you if you listen to the the main feed, you'll know there's some ads in there. So it's nice when you don't have to listen to them because you're part of the Patreon.

00:33:31:07 - 00:33:53:05
Unknown
You get special discord access. You get every weekday an entirely different podcast called AI Inside Daily five days a week. I tell you a few things that are happening in the world of AI that's most important, and a lot of that we end up talking about a greater length on this show, but it's kind of a yeah, it's been really useful to me to get a better understanding and knowledge of everything that's going on.

00:33:53:05 - 00:34:13:06
Unknown
So Patreon.com sideshow, thank you, patrons, for your support and for being with us along the way. All right. We're going to take a break. Then we're going to get a little musky on the other side of the break. Talk a little bit about space and a major IPO coming up here in a moment.

00:34:13:08 - 00:34:38:04
Unknown
Get musky right. Let's see here. Space. Big week record breaking week right. SpaceX went public on June 12th. I could have swore we talked about it on the show, but that's impossible because it was after the show. This just goes to show how fast things move. But I was like, man, didn't we already talk about that? Nope. SpaceX going public last Friday.

00:34:38:04 - 00:35:07:13
Unknown
I think the ticker is SPC if you're curious. IPO price just looked it up. Go ahead. Yeah. You go. No, I was just going to set up 135 a share initially. IPO open at 150. Hit 175 midday. Closed at 161. What is it today? 199 and 2099. And the market cap which started below 2 trillion, went to I think 2.11 trillion at the end of the first day is now at 2.62 trillion.

00:35:07:17 - 00:35:28:10
Unknown
Wow. It's gained a half $1 trillion. What is the world's insane. Jason, what is going on here? Yes, yes. Fourth most valuable company in the US. And we're still in this period where the indexes have to have to buy this. So they're going to try to find a way. They're going to the fear is is going to go up.

00:35:28:10 - 00:36:03:22
Unknown
And they got to buy it to be part of the to to reflect the market. And so you know we'll see what happens. But the insanity has taken over. The world stopped with more than that wasn't. Yeah it did it surpass Amazon. Yes. Crazy. How how does space how is that. That's crazy. The so this was the largest IPO in history was is I guess is previous record was Saudi Aramco at around $30 billion.

00:36:03:22 - 00:36:32:18
Unknown
So space basically doubled that for the most part. Elon Musk who owns roughly 40% of the company, he himself crossed into the trillionaire territory, one person worth more than 11. $1 trillion. I guess maybe that's more if the value keeps going up, but, but that's pretty, pretty crazy. He's the first person to get to that point. So richest man on earth, which I don't.

00:36:32:19 - 00:37:01:18
Unknown
I know that I don't feel very comfortable about that. No, I don't either. There was a do this obnoxiously crazy sometimes on the show I was reading date from Germany. I read a really good column by a guy named Georg Diaz saying that we have to redefine progress, and he looks back at the Industrial Revolution and say that the progressive movement came out of the Industrial Revolution and the Gilded Age and the inequalities that were there.

00:37:01:18 - 00:37:27:23
Unknown
So it was the eight hour work day came from the printers I read about in Hot Type, the push for health at the workplace, the push for the welfare state all came out of that industrial revolution. And now we're in a second Gilded Age now. And what he said was to the point, finally, the we're allowing the technologist to define progress, and they have the capital and the power thus to do that.

00:37:27:23 - 00:37:56:12
Unknown
And Elon Musk is in the most powerful position as an individual on Earth to define progress. And what's he defined as progress? Putting chips in us, taking over Mars, impregnating the world, all this stuff that he does and we, the humans have to take over, take back our definition of progress. And so that's what bothers me most, is that he's he's the single richest man in the world, which gives him incredible power to do that.

00:37:56:13 - 00:38:21:10
Unknown
And that's wrong. Yeah. Well, and I mean, when you've got resources like that. How do you question someone in that position with those kinds of resources? I mean, literally, you know, from a from a financial standpoint, more resources than anyone. And so technically speaking, that kind of means you can probably get away. You know, you could probably advance any I was going to say, get away with anything you want.

00:38:21:10 - 00:38:43:18
Unknown
And I think with with Elon Musk, that's probably pretty true. I'm not a huge Musk fan, but you can advance any effort or agenda or purpose that you personally want, and it doesn't just apply to you when you are a trillionaire. It applies to everyone in some capacity. And yeah, that's just that feels wrong to me. That feels very wrong.

00:38:43:18 - 00:39:07:14
Unknown
And I don't know what the antidote to that is. No. And, you know, you think back when we when we laughed when you took over Twitter, it gave him what he wanted. It gave him a public presence and influence. And the price was seemed ridiculously high. But it means nothing. He can he, as we were going to say in a minute, he just bought a technology company, but he could buy anything.

00:39:07:22 - 00:39:32:07
Unknown
Yeah, he could buy CNN tomorrow. If he could buy a network, he could buy newspapers. He could buy whatever the hell he wants 100%. Well, what what he slashed, they chose to buy was not an entirely big surprise, necessarily, because we kind of have known about this for the last couple of months. But four days after the IPO, SpaceX announced that they're buying cursor.

00:39:32:08 - 00:39:59:08
Unknown
This is the AI code editor that you've certainly heard of, buying it for $60 billion. All stock deal. So no cash. It's just all stock but still 60 billion. You know, that's another record breaking, buyout I think. Right. Has there been. Yeah. And I think it illustrates it shows the paper Tiger that is this IPO. The truth is Musk doesn't have an AI strategy.

00:39:59:11 - 00:40:22:18
Unknown
Right. How is he making. He has to buy cursor. He built big data centers that are powered by gas turbines in black neighborhoods that are that the government says are critical to national security. So they can't sue against it. So people's people's homes are being ruined by this. And what does he do with them? He's not building his own powerful AI.

00:40:22:19 - 00:40:45:19
Unknown
Yeah, he has grok. Who cares? He's leasing that out to the other companies because there's such a shortage. So he's nothing but a but a computer host. He's a parking lot. That's what I just don't get about this. The rockets are not a sustainable business in and of themselves. It's a it was privatized government business. And if the government decides we don't care about space anymore, that business goes away like that.

00:40:45:20 - 00:41:08:20
Unknown
That in turn subsidized his internet from the skies business, which is successful. But eventually Bezos is going to come along and compete, and it's going to be other methods there. His cars were subsidized by government and there's tons of competition coming along. Then his AI doesn't really have a strategy worth a thing. Yeah, he's getting a lot of revenue, but he's getting it for hosting, for being a parking lot.

00:41:08:25 - 00:41:32:25
Unknown
That space I, I just don't get it. And yet we we don't get it. And yet it was the biggest IPO in history. It's just so we yeah we can not get it. And still it seems to be working for someone for for muskets certainly working. If it grows a little bit more he'll be bigger than Microsoft to.

00:41:32:28 - 00:41:54:25
Unknown
Yeah. Wow. That is just that is just remarkable. Yep. Crazy. Yeah. And like you said, you know, rocks. It's not like rock is an AI model. Like, I rarely meet someone who's like, let me tell you about the best model in the world. It's grok. And I'm not saying they don't exist. I'm not saying that some of some of you listening are already yelling at me, being like, oh God, here we go.

00:41:54:25 - 00:42:16:00
Unknown
Grok eight because it's attached to Musk. But I mean, you know, when you're talking about coding performance, you know, I'm not I'm not a designer, I'm not a coder. But my understanding is that it's pretty mediocre in comparison. So a company like cursor in your in your playbook can be very beneficial to that at a time. That Codex is getting a lot of real attention right now.

00:42:16:00 - 00:42:47:22
Unknown
Cloud code also. Yeah. So you buy your way into into success or relevance or whatever. Yeah. It's interesting. Yep. I wonder how cursor how do cursor stands feel about something like this? I guess it probably just depends on how they feel about, you know, Musk, and his kind of entities. I could be wrong here because I'm not a coder, but cursor soon be hot for about two weeks and some people were loving it.

00:42:47:22 - 00:43:10:22
Unknown
But then I know some people got overcharged by it. The prices went up. It's built on top of AI. Yeah, he needed it. He needed to have something more for AI because he doesn't have a strategy. Well, it makes me wonder what's next. You know, what are the when you've got that kind of capital and that that kind of access.

00:43:10:24 - 00:43:39:26
Unknown
What is the next thing that he's going to swoop in and decide to pull into the space XII umbrella? Yep. We'll find out. Augmented World Expo 2026 is an event that's happening, happened. I know it was happening yesterday. I don't know if it's happening today in Long Beach, California. And actually, sorry, I just need a whole huge conference of people who are bumping into each other because they're wearing things.

00:43:39:29 - 00:44:10:11
Unknown
Yeah. They're just they're just, Yeah, like like like a large pinball game bumping into the walls. Couple of AR glasses announced. Well, 1 in 1 announced that we didn't know of officially. And then another one. More details. We'll talk about snap. First snap unveiled the their newest glasses called specs. These are true augmented reality glasses with waveguide displays in both eyes 51 degree field of view.

00:44:10:13 - 00:44:38:10
Unknown
Fully standalone glasses. So these are not the kind that have like an attached compute and battery puck that you keep in your pocket like an Apple Vision Pro, like the glasses I'm going to talk about after this from X called aura. But these are totally standalone. Look a little bit like safety glasses to me. But you know, when you're hiding the battery and the compute and everything entirely inside the frames like that, stuff's got to go somewhere.

00:44:38:10 - 00:45:01:26
Unknown
And so I'm not surprised that they kind of look like safety glasses. But I did see some like, is it part of this article where I saw like, yes, where they show beautiful people wearing the glasses and, you know, I don't know, I guess maybe they look less weird, but it kind of reminds me of Google Glass on the runway, you know?

00:45:01:28 - 00:45:06:19
Unknown
Yeah, I'm getting such PTSD and all this talk.

00:45:06:22 - 00:45:27:17
Unknown
So anyways, this is this is, I think, a four hour charge for our charge, you know, for an early technology, I get it. But if you, if you're out for the day. Yeah. What do you do. Take them on. Put them off. Take them on. Put them off. You. I don't understand how you can really use it at that level.

00:45:27:19 - 00:46:07:10
Unknown
Yeah. I mean, these are that I have in my hands here. These are range X3 Pro. There we go. Radio X3 Pro. These are standalone. These are also binocular displays in the glasses. And when I've used these and actually used them the way they're supposed to be used I mean I get like maybe 45 minutes out of them, maybe an hour, you know, so it's different if what snap is saying here, that's a definite improvement for hours at least expands or spans through like an event, whereas an hour, you're lucky if it's working at the time that you actually needed to be working for whatever you're using it for.

00:46:07:12 - 00:46:34:29
Unknown
So maybe things are improving, but but I don't think that challenge is going to be very easily solved with this form factor. If you don't offload the compute somewhere, you know, if it's all embedded in the frames, then you're going to end up with big chunky frames you still might not get a full day was worth of use out of them, but they're going to look weirder than normal glasses unless we get to a point to where suddenly society finds big, chunky glasses like these to be stylish in and of themselves.

00:46:34:29 - 00:46:59:23
Unknown
Which, hey, that could happen too, I suppose. Yep. But yeah, these are $2,200 glasses, so they are not cheap. They've got the little blinking LED right above the nose bridge. What does this have to do with AI? Well, there are different cameras on board. Infrared, regular cameras meant for understanding what you're looking at, you know, meant for hand tracking and everything.

00:46:59:23 - 00:47:26:25
Unknown
So. So these do have, hooks into Gemini, hooks into ChatGPT and stuff like that. So there you go. Curious to see. I'm definitely curious to to get my hands or my eyes in them at some because you've been trying all of them. I've been trying a lot of them. The ones that I can. I've definitely tried X reels announcement from or which is the X real aura.

00:47:26:25 - 00:47:54:13
Unknown
And you know again these these do look kind of funky but but I don't see these. Whereas the snap ones are maybe more designed for being worn out in the world. The extra aura glasses I don't see these as being that because they are the kind of like if you see that there's the main glasses on the front, which are really kind of like the electrochromic dimming, like sunglass layer.

00:47:54:13 - 00:48:16:00
Unknown
And then on the inside is a separate kind of lens system that kind of projects into your eyes. And I don't know, it's just not the same. Wandering through the world with with these glasses doesn't seem like the right approach. These are more like you want, you know, you want an Apple Vision Pro experience in your living room, but you don't want to have to put on the huge headset.

00:48:16:00 - 00:48:37:11
Unknown
You put on something like this instead. And you do a lot of the same things that you do in that experience. But with this. But I don't see this as out in the real world necessarily. So I went to the X real site and it says Founder Priority Pass $299. I thought, wow, that's a hell of a sale for an expensive thing.

00:48:37:11 - 00:49:05:24
Unknown
It's not. No, no pain. You're paying 300 bucks just to reserve one of the first ones out of 2000 units, supposedly, and that's that. Then you can pay between $1,999 and get $199 credit if you reserve. And I don't know if these are refundable or not, the only thing they guarantee that the final retail price will be no more than $1,500.

00:49:05:26 - 00:49:27:24
Unknown
Yeah. The price of of of Google Glass, by the way. Oh that's right, it was 1500 ish right around there anyway, around there. And so you don't know what you're really going to pay, but they did give you a top at least. Okay. And that's less than what you get out of the the snap glasses. So there's that.

00:49:27:26 - 00:49:59:23
Unknown
Not so not display glasses. Spatial computing glasses. Yeah. I mean, like I said, this is this is this is essentially an Android XR unit, but in a glasses form factor. So, you know, there was the Samsung Galaxy, was it the Galaxy VR, Galaxy VR? I can't remember what they called it, which is like literally meant to kind of be in the same playing field as the Apple Vision Pro, kind of like the full VR experience locked away.

00:49:59:24 - 00:50:36:08
Unknown
Massive kind of view, field of view, really immersive. This is a step down from that. But the same software. So like when I went to Google last December, I think it was when they invited me out to check out these glasses. One of the demos that I did, I put on these glasses and they had a tabletop game called DeMeo DeMeo that suddenly snapped into view, and it was like where I was looking, there was the board, and I could reach out and I could grab the pieces and move them on the board, and it felt it was really cool, like it was like, that's the demo that you need to show everyone, because that

00:50:36:08 - 00:51:01:12
Unknown
will make people go, oh, I get it. You know, are they going to be super successful at at this price with this technology? That remains to be seen? I don't know, I'm kind of I'm intrigued, but I'm not entirely convinced the the longer I go on that I've, you know, and they're going there diving in the deep end because their underwear to buy it lists Best Buy and Micro Center stores.

00:51:01:13 - 00:51:20:16
Unknown
Yeah. So they're, they're going to have to stock retail places and, and that that these days is a marketing cost. Sure. You've got to buy floor space for this. That's really what Best Buy's their whole business now. And so this is going to cost a lot of money to market this thing. So it's like a show a showroom sort of experience.

00:51:20:18 - 00:51:40:20
Unknown
But you know hopefully that means that people that go into Best Buy can actually put this on to kind of see what it is, which is going to matter if you're going to sell any. Yeah, but I think it's going to be a hard, hard rule. Yeah. Yeah. But it is neat. You know, as someone who has worn at least this one, I will say it's a cool experience.

00:51:40:20 - 00:52:02:04
Unknown
Is it $1,500. Cool. I don't know, I think the price that they're talking about, less than $1,500 is around. What I was guessing these would probably cost maybe a little bit less than that, but, we still don't know what it's actually going to cost. Is it going to be 14.99 or is it going to be 1350 or whatever, just less than $1,500.

00:52:02:04 - 00:52:32:02
Unknown
That's all we know. So yeah, I'd be curious about that. Jeff Bezos did his first public interview about his AI startup, which we've talked about before. Prometheus no longer project Prometheus, by the way. It is just it's more permanent than that. Yes, it's more permanent now. It's just Prometheus. This is, let's see here. This is the do artificial general engineer.

00:52:32:04 - 00:53:02:29
Unknown
That's what they're building here. So these are AI tools to automate the design and manufacturing of physical systems. Very complicated and complex physical systems like jet engines, aerospace, medical devices, hardware, drug compounds, that sort of stuff. And, I don't know how to pronounce his last name. Is it Baja or Barge? Vic Bajaj, who's a co-CEO along with Bezos, he actually co-founded Verily at Google, the life sciences company inside of Google.

00:53:03:01 - 00:53:29:01
Unknown
So he's over here now. And he basically the example he gave is like designing a jet engine takes a team of engineers more than a decade, if, you know, a decade or more, let's say, and Prometheus is looking to compress that by ten x. So what, you would normally use a team of engineers to produce with a jet engine in ten, ten years, do that with one with Prometheus if it works.

00:53:29:04 - 00:53:42:28
Unknown
Yeah. And we've heard some of the things from beyond LeCun with real, real world models is that their target is industry for manufacturing, design for.

00:53:43:00 - 00:54:13:25
Unknown
Warehouse stuff, medicine. And what's interesting to me is that this could be the real heart of AI's impact on the economy. If you can design a jet engine in one tenth of time, thus one tenth the cost one presumes, plus testing, obviously, and do a better job of it. You know, just just as if AI can find faults in code, one presumes that can find faults in jet design, then that has a huge impact potentially on industries.

00:54:13:25 - 00:54:36:14
Unknown
And it's not sexy and it's not retail. No common person can use this and say, you know, can you write me a song about my kid? No, no, not going to do that. Can you create me a jet engine? Right, right. But it could be this could be the heart of what an matters. I think the more we get from this, the more this is.

00:54:36:16 - 00:54:56:14
Unknown
This is the rest can be a part of their trick and fun and has value. I'm not questioning all that, folks, but I think we've been waiting for the economic value of AI, and where people have been looking for that economic value is in eliminating jobs. We're not really seeing that yet. It's going to it's changes. Work is the argument.

00:54:56:16 - 00:55:20:00
Unknown
It doesn't really reduce it much yet. Some companies try to, but they're really using it as a veil for layoffs. This kind of efficiency is what's going to matter economically. Yeah. Bezos did say that. He's talking about this now because the story was leaking out. And he said, if you let that just be a complete void, they'll fill it with nonsense.

00:55:20:05 - 00:55:50:01
Unknown
So we're going to get ahead of that, tell you exactly what it's all about. And this is Bezos's first CEO role since he stepped down from Amazon. And was it gosh, was it already 2021. Jesse's been in 2021. Wow. Crazy interesting. And then I you know, I'm fascinated by this Allbirds thing. And we have a little bit more of of information here and then something.

00:55:50:04 - 00:56:09:12
Unknown
So so go ahead, tell the story first and then I'll, I'll do the next one. Yeah. And I mean, we did talk about this briefly when it happened a couple of months ago. But all birds this is the wool shoe company of which I have a shoe, an all birds pair of shoes. Apparently I'm not going to be able to buy any more because they are officially out of the shoe business today.

00:56:09:12 - 00:56:35:23
Unknown
But back then, a couple of months ago, they had sold to American Exchange Group. 39 million pivoted to new bird AI. Like I said, we talked about that the plan was to pivot from shoe AI infrastructure, and now they're announcing Smart Bird, which is the now the name of their new AI company, essentially renting out GPU clusters as a managed service, named CEO Nadya Carlson.

00:56:35:25 - 00:56:58:18
Unknown
She actually has experience in AI infrastructure at AWS, at sandbox, AQ and stock went, you know, pretty nuts. Surged 600% on the announcement. So people are like thumbs up on okay, going from a shoe company to an AI infrastructure company. So for those of you who are who are lamenting you can't get your Allbirds anymore, I have a recommendation for you.

00:56:58:19 - 00:57:22:22
Unknown
Yeah, very on air story. These are my geese. Find shoes. They're from Austria. They're wool. They're really nice and comfy and cool and good in water. So I recommend forget the Allbirds. I didn't find them that comfortable. Thanks. I found these better. So buy some real wool shoes from geese. Fine. There you go. This this episode. Not sponsored by geese.

00:57:22:22 - 00:57:41:02
Unknown
Fine. But maybe we should reach out. Hey, guys. Fine, if you'd like to. There's a lot of a lot of former Orban's customers here. Who who should be coming to you? Oh, my goodness, that would be awesome. It would be. Suddenly we haven't had them as a sponsor. We'll just have to make that happen. He's fine. Don't be surprised if I reach out to you with a pitch.

00:57:41:05 - 00:58:07:24
Unknown
Yeah. I mean, how do you how do you feel about a shoe company making the seemingly, at least at this early stage, somewhat successfully received transition slash pivot from shoe company to AI infrastructure company. See earlier renting out GPUs of idiot market. Yep. You know, I saw another story today. There's a company called. It wasn't relevant. It's called bending spoons.

00:58:08:00 - 00:58:35:26
Unknown
Okay. Yeah. And they bought a whole bunch of dead brands AOL, Eventbrite, Vimeo. We transfer Evernote, beat up StreamYard, which I didn't know was a dead brand. Hop in Splice and Bright Cove, to name just a few of them. And they take over these dying brands and cut the heck out of them and do something new. But they do something new that's in the turf where they were, in this case, Allbirds from a shoe company.

00:58:35:29 - 00:58:59:02
Unknown
Company. It's just insane. It's insane. Like, I can't help it. Yeah, you can't help but think that, like, there was some sort of, like, court, like dinner conversation or over cocktails or something like that. And one guy, one person, you know, involved with the company was just like, you know, we really need to do we really need to not be a shoe company anymore, get into this AI thing.

00:58:59:05 - 00:59:21:20
Unknown
Wait a minute. That's actually really great. Like, I'd be super curious to know the story of how that, that kind of, that idea even entered entered into the discussion because it's just wonder what else, what other crazy conversations it's inspiring out there. You know, a taco company says, how do we become an AI company? Yeah. Yeah, exactly. Well, and you know what?

00:59:21:20 - 00:59:51:14
Unknown
This has happened before. I kind of came across this as well. CNBC ran a piece that compared this to Long Blockchain Corp., which was Long Island Iced Tea, right before they pivoted to change in 2017. So it's a thing. It's just blockchain, maybe a little less sexy than than AI infrastructure in 2026. But ten years later, have we learned nothing or have we learned everything?

00:59:51:14 - 01:00:10:18
Unknown
I don't know. Very, very interesting to me though, and I will set this. You have mentioned vine before, and I'm happy that you mentioned it, because I meant to do something with this, because I do need to buy a new pair of shoes. So I will order my geese fine and let you know and see if maybe they want to sponsor us.

01:00:10:19 - 01:00:31:22
Unknown
In the meantime, be cool. Hi, guys. Fine. Hi. Oh, your shoes are amazing. They just. They fit. And I've got a coupon code. No, I don't have a coupon code. Real quick. If you're enjoying the show, leave us a review on Apple Podcast. We appreciate it. Lots of reviews hitting up there a few months ago, but we could, you know, we could always use more.

01:00:31:22 - 01:00:44:29
Unknown
So if you haven't done it yet, jump over there. Leave us a review. Appreciate it. All right. Going to take a break and then come back with a little speed. Round five quick stories and then we'll get you out of here. Hang tight. It's coming up in a moment.

01:00:45:01 - 01:01:16:21
Unknown
All right, so let's see here. Sundar Pichai, this is a Google alphabet. I always want to say Google but really it's technically it's alphabet CEO. Sundar Pichai was doing a commencement speech at Stanford on Saturday. And parent apparently a lot of students walked out for it. They were you know, they were holding signs or there were signs there about Ice and Google providing cloud services to Ice, about Google's Project Nimbus contract with Israel.

01:01:16:23 - 01:01:44:22
Unknown
And yeah, but but Pichai didn't actually talk about AI during his commencement speech. He didn't bring up the the no no word. It's Stanford. You guys went to Stanford. Stanford is the is the wellspring of Silicon Valley. It's all this stuff here. You're already schmutz with Silicon Valley and technology folks because you're at Stanford. You could have gone to Haverford, and then you would be nowhere near Silicon Valley.

01:01:44:22 - 01:02:12:21
Unknown
But you're in the heart of it. I'm sorry I don't take your protest very seriously from there. There's a lot of them, you know, maybe by 200, by one count. Yeah. By one account. Yeah. I'll just mention that my interview again with Joel Pino. It just so happened that last week, last Friday, she gave a speech as well, because she was also an honorary graduate degree recipient at the University of Ottawa.

01:02:12:21 - 01:02:43:23
Unknown
And she did talk about AI. And as far as I know, nobody walked out that that I was aware of. But we are nice. Yeah. Exactly right. We do spend some time talking about, about what she talked about in, in the commencement speech and specifically like the thing that I was really curious about is like, what is it like amongst the, the tenor, the temperature of those graduates kind of going out into the workforce after the work that they've done at this moment in time especially.

01:02:43:23 - 01:03:08:10
Unknown
So we talk a little bit about that as well. You have that to look forward to. Returning to Elon Musk for a quick moment, just to let you know that he lost again to OpenAI. In court, Judge Rita Lin dismissed XYZ trade secret lawsuit that had alleged OpenAI stole secrets about grok during a recruiting interview with a former AI engineer.

01:03:08:10 - 01:03:28:06
Unknown
And the judge actually said, you know, asking a job candidate about their previous work. It's kind of a normal part of hiring. I don't know if there's really much we can do there. So they dismiss the case? No. You know, no amendment possible. Can't refile. So that's it. I'm sure there are other cases. He doesn't need the courts.

01:03:28:06 - 01:04:01:24
Unknown
He can win with his money. Now. He just. Yeah. Now he's just got his money, which maybe that's the biggest, you know, thumb to the nose that, that he has. Let's see here Microsoft okay. So Microsoft reportedly starting or started using Amazon AWS to handle GitHub's capacity issues. Tons of commits happening per week, 275 million commits happening per week on pace for 14,000,000,000 in 2026.

01:04:01:24 - 01:04:25:22
Unknown
That's up from 1 billion in all of 2025. So that number has increased tremendously. And when you're talking about agents, you know, requests went from 4 million in September to 17 million in March. Because we are in the midst of an agent kind of moment right now, lots of that activity happening. And so Microsoft is going to its biggest cloud rival for help.

01:04:25:27 - 01:04:48:19
Unknown
Yeah. This is like this is like McDonald's asking Burger King for some ketchup. Yeah it is, but I think it's but it's a circular economy. We keep on talking about it because there's a scarcity of capacity in data centers right now. That's why Elon Musk once again the road leads back to Musk. He's running out the capacity that he doesn't have a strategy to use.

01:04:48:19 - 01:05:07:13
Unknown
And Amazon has tons of capacity because they're really good at that. So Microsoft needs to help. It's a good sign for Microsoft's business actually. You know, it means they have demand and they're trying to meet it and they're being smart and going to somebody else to get it. And Amazon, it's good for them too. Yeah. So it's a weird world we live in now.

01:05:07:15 - 01:05:31:10
Unknown
It is a weird, weird, weird world that we live in. DoorDash launched a new feature. I don't really use DoorDash much. Personally. I don't think it's even available where I am. Oh, really? I'm out in the boonies. Oh, okay. All right. Maybe. Maybe someday you'll be able to use Ask DoorDash. It's, as you can imagine, an AI chatbot for ordering food.

01:05:31:12 - 01:05:50:03
Unknown
It allows you to snap a photo of a cookbook page. Then it builds your cart and quantities and all that. And then wait wait wait wait wait wait wait so I can deliver it to you. Delivery from a grocery store or from a restaurant? I know DoorDash works from grocery stores. Oh, okay. Now it makes more sense to me.

01:05:50:04 - 01:06:14:07
Unknown
Okay, I didn't get that. Yeah. I mean, you can use DoorDash to to go to a restaurant and pick up your food. I would like this filet mignon. Well, all we have is, is McDonald's. We can get you a Quarter Pounder with cheese. You know, that's good enough. Yeah, right. Good enough. Almost the same thing. Totally. Yeah. Okay, so that's rolling out to iOS initially.

01:06:14:09 - 01:06:44:21
Unknown
Of course, us Android chumps have to wait. And then Google released diffusion. Gemma, this is an experimental open model that generates text completely differently. Apparently, instead of a single token at a time, it generates entire 256 token blocks simultaneously. Like drumroll please, the printing press might do. So. There you go. Yeah. So the trade off here. Did I hear Gutenberg?

01:06:44:22 - 01:07:04:05
Unknown
Yes, exactly. Yes. It's it's a it's a drinking game. Lower quality than regular. Gemma for Google says go ahead and use the standard Gemma for production. Use this for speedy stuff like inline code editing, that sort of stuff. So now we got to now we got to determine which one we're using for which task in the middle of it.

01:07:04:05 - 01:07:28:00
Unknown
But, apparently they fine tuned it to solve Sudoku as a proof of concept. So yeah, this one I sort of understand, but not entirely. So I'll take your word for it. Google. I think that's about it. That's what we got today. Thank you Jeff. Mr. Jarvis, thank you so much for being on each and every week and sharing your knowledge with me.

01:07:28:00 - 01:07:55:00
Unknown
And I learned so much talking with you about this stuff. You've got Jeff Jarvis Hot Type available for preorder, and I'll take for one second. Just got a review from Publishers Weekly that says Jarvis pens a colorful and enthralling portrait of Gilded Age industrial ferment. Readers will be wrapped. Yeah, there you go. So then, is that the sort of thing that you can, like, put on the, on the fold or or is it too late for that?

01:07:55:01 - 01:08:14:22
Unknown
It's too late for that because the book exists. Yeah, I guess it's it's all done. So yeah. Okay. All right. See a little book maybe put a put that on your web page. See it forced and foremost. That's cool. That's awesome. That's a that's a great review. Hot type. There it is. Parenthesis for more magazine for even more on the web.

01:08:14:22 - 01:08:42:25
Unknown
We for even even more all available at Jeff Jarvis. Com. More books coming up, including intelligence, AI and humanity from Bloomsbury. As for me, Pod Tune Up is my podcast consulting site. I'm doing some consulting projects right now. I'm working with not just one team. I'm working with multiples now, like it's growing. It's really amazing. So pod tune up if you have a podcast or you're considering one, hey, I can help you with that.

01:08:42:25 - 01:09:05:10
Unknown
I can help you think through it and strategize and do all the things. So there you go. AI. Inside that show is the show page on the web. For this show, you have everything links to everything related to AI inside the podcast episodes, audio, video, Patreon, subscribe information. Everything is there. You know ways to find all the feeds, all that kind of stuff.

01:09:05:12 - 01:09:25:27
Unknown
And then I think I mentioned last week very briefly that we have a fourth wall page. If you actually want to buy some, some gear, some like a hat or a mug or a t shirt or whatever, that says AI inside, it tells everybody that you've got a little AI inside of you. Go to AI inside Dash shops.

01:09:25:29 - 01:09:44:25
Unknown
Fourth wall. I'll work on on my little plug there. I'm sure I can come up on the way soon. Yay! Hey, I can't wait to see it. Once you. Once you get it. Where it on the show. You it here? Yes. Hopefully. Hopefully I'll get mine soon too. And we can wear it on the same episode. Patreon.com AI Inside Show.

01:09:44:25 - 01:10:05:23
Unknown
If you want to support this show on a deeper level. We have many amazing patrons who do this and we could not do this show without you. Want to give a quick shout out to our executive producers? Doctor Do Jeffrey Marikina Radio Asheville 103.7 Dante, Saint James Bono, Derek, Jason Knife or Jason Brady, Anthony Downs, Marc Starker and Karsten Samachar.

01:10:05:25 - 01:10:36:12
Unknown
This just some of our executive producers, but there are many different levels and you get things like the show ad free, you get the discord community, you get the daily, the AI Inside Daily podcast, and then at that executive level, you do get a t shirt thrown in for good measure. So thank you to all of you who support us on a deep level as you do, and a huge thank you to those who help us behind the scenes as well, Daniel Croft for doing our social video and Victor Bogart for doing the video edit.

01:10:36:15 - 01:10:50:04
Unknown
We appreciate you both. I appreciate you both because I just don't have time to do all that stuff. And you do. And thank you, thank you, thank you. All right, that's it. Thank you Jeff. Thank you everybody for watching and listening. We'll see you next time on AI inside. Peace out.