Apple Says OpenAI Stole Its Secrets

00:00:00:01 - 00:00:30:19
Unknown
Coming up next, Jeff Jarvis and I dig into Apple's big trade secret lawsuit against OpenAI. We also look at OpenAI's huge GPT 5.6 week, the leadership kind of shakeup that's kind of attached to that. Also, OpenAI has a new Codex keyboard like attachment thingy. A lot of stuff happening with OpenAI. We've got Demis Hassabis writing the essay that could or could not reshape how AI gets regulated.

00:00:30:22 - 00:00:57:23
Unknown
Meta. Launching an AI image tool with Instagram hooks that so many people hated, that they had to pull it a few days later, and a whole lot more. There's just tons in today's episode. Some really great discussion. You're going to enjoy it next on AI inside.

00:00:57:26 - 00:01:17:11
Unknown
And 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. Where May, Jeff, Jason Howell, and my co-host Jeff Jarvis kind of learn about the AI worlds week after week. Although we've been talking about it for a while, I consider us not entirely novice at this point.

00:01:17:13 - 00:01:36:11
Unknown
No, no. And it affects us all. So we have perfect right to view AI and ask what it means for us all. Kind of see what the heck is going on in the industry. How are you doing, Jeff? Good to see you. Good good good. Yeah, we've got we've got, boy, as usual, a lot. If I could show you.

00:01:36:13 - 00:01:57:16
Unknown
Actually, I'll just show you right now the amount of of tabs I have open at the top of this browser, which are all the links to the stories that we'll be talking about. It's a lot. We'll probably talk about most of that stuff, starting with Apple suing open AI, which on a week like we've had feels like eternity ago.

00:01:57:16 - 00:02:19:19
Unknown
It feels like, oh, we must have talked about that last week. But no, we didn't, because this was like the end of last week and so much has happened since then. And this is not like a little, a little patent like squabble. It's, you know, it's full on what Apple is charging here is trade secret theft. They filed the case in federal court in Northern California.

00:02:19:20 - 00:02:51:20
Unknown
Apple is basically saying that there was a coordinated effort to steal Apple's confidential information. And so that's a big that's a big charge coming from from Apple. And there's some really interesting things tied into this case OpenAI we remember, you know, bought Johnny Ives startup IO products, $6.4 billion. That was just last year. And then we've been hearing about OpenAI building some sort of hardware, which we will get to because there's been some news around that to hardware.

00:02:51:20 - 00:03:17:16
Unknown
Chief at OpenAI is Tang Tan. Tan. Tan is a former Apple VP, and Apple is saying that tan was actively directing people who were still Apple employees, who were interviewing at OpenAI to bring in physical Apple hardware parts to their interviews as like a show and tell. So, first of all, like what I don't I don't get it.

00:03:17:17 - 00:03:39:09
Unknown
Like how you are, who you are, and you've had all this success and you're at the you're at the top ranks of these elite companies and yet you don't. If this is true, this is all allegations at this point. We'll find out throughout the case. But if this is true. How are you, that person in that role? And yet you do this and don't realize that's a bad idea.

00:03:39:10 - 00:04:05:10
Unknown
Right. And how does anybody you talk to not tell you or find a way not to? I don't know whether anybody followed his advice and came in with the stuff, but it's bad advice, bad news, just even getting an Apple thing out of the building, right? Right there. Well, and not not only just the act of like, hey, tell me what this other company it did and, you know, things you're not supposed to tell me in the interview, by the way.

00:04:05:11 - 00:04:29:03
Unknown
Right. But of Apple, which we already know is incredibly litigious as far as protecting their their stuff. So weird. There's also an engineer named Chang Liu who left Apple, went to OpenAI, apparently never returned his Apple laptop. You know, I don't know if he was granted the ability to keep the Apple laptop or not, but he still had it.

00:04:29:05 - 00:04:59:16
Unknown
Liu texted a former Apple colleague saying, quote, lol, I found out I can access the network storage. So funny. And then use that access to download things like presentations, hardware designs, manufacturing details, testing procedures, all sorts of stuff from Apple. And by the way, clearly it's all traceable. Like when you're downloading that stuff. It's not like because no one was there watching you do it.

00:04:59:16 - 00:05:20:15
Unknown
You can't be, you know, it can't be traced back to you. So yeah, pretty blatant having left workplaces. I'm a gossip about them afterwards. Yeah, I get that. I get to want to find out stuff. I mean, I'm emeritus at Cuny, so I'm still officially on the faculty. Yeah, and it took a while for them to realize they didn't want me on the email list.

00:05:20:17 - 00:05:37:09
Unknown
And so I found out stuff and I'd share it with friends who weren't, you know, didn't have that status. I get that, but that's not this. This is if it's true and it's alleged its corporate espionage. Yeah, it's puts you in a position of terrible liability. And what what was his name?

00:05:37:11 - 00:05:42:26
Unknown
Who was it? Sorry for the bad word.

00:05:42:28 - 00:06:05:01
Unknown
Who left? I'm. I'm going to get all confused. One of the AI people who left. What was it, Uber or went to? 0000. What's this? Is that. Yeah. Uber lawsuit. Guy left company trade secrets. This is how my this is how my my brain works.

00:06:05:03 - 00:06:23:27
Unknown
Right? Right. Yeah. So you end up losing your next job in the process. You put yourself at liability, you put your next employer liability. It's not worth it. Not worth it. And the thing is, there are things in your head, their ideas that you can't get rid of. And and if you operate with caution, you can bring value in that way.

00:06:23:27 - 00:06:49:28
Unknown
But just be smart about it. I mean, that I think is the thing like, this is just I mean, both of these situations are pretty blatant. Like there's there's no like, painting, you know, there's no fine tooth comb here. This is this is a I don't know, it's it's super blatant. It feels like apparently before Apple filed the lawsuit, they tried to handle all of this privately.

00:06:49:29 - 00:07:17:16
Unknown
OpenAI says that they responded. Apple claims in the suit that OpenAI never responded. But apparently what actually happened, and this is according to emails that NBC reviewed, is that Apple's lawyer mixed up the names of two OpenAI employees who had the last names, Wang and Chang. So the the communication that they had went sideways because of the comfortable in terms of naming right there.

00:07:17:17 - 00:07:41:20
Unknown
Yeah, yeah, a little uncomfortable a little. Yeah. Not not gonna lie. That's not a good look. And after that kind of mix up happened, things, you know, the communication fell apart in some way, shape or form. And here we are now with a now with the suit being brought against OpenAI. So oops. Yep. And then what what impact does this have on OpenAI's plans to bring out a device?

00:07:41:23 - 00:08:14:05
Unknown
That is a good question. The rumors were that they were going to bring out a phone. I mean, that's been one of the rumors, one of the then we've also heard rumors that it was going to be something that was unlike something that has existed. And it's like, well, what does that look like? And yeah, I mean, now we've got Bloomberg, Mark Gurman reporting that there is like the first device finally kind of getting some sort of indication of what that might actually be would be a screen less a mobile screen less smart speaker.

00:08:14:08 - 00:08:35:27
Unknown
So because it's worked so well for Apple and Amazon and Google, it's taken over the world, hasn't it, Jason. Oh yeah. Yeah. How many of us have those damn speakers sitting all around the house silent? But this one moves Jeff as a battery. This one, this one has a battery so it doesn't need to be plugged in. This one can move around.

00:08:35:27 - 00:08:52:01
Unknown
You irritate people on the public with it. Yeah. Oh that's true, that's true. Or it can irritate you when you don't want to be bothered because it just drives to for it. Right? I still haven't figured out how to turn off all the sounds on my Google Watch. Yep.

00:08:52:03 - 00:09:16:17
Unknown
Excuse me. As I'm laughing, I'm choking myself. So this is the device. The device apparently that's being designed with Johnny Eye from love, from camera sensors, these mechanical elements that we're talking about so that it can kind of move on its own. And why? Because what OpenAI, according to Mark Gurman, anyways, is looking to do is to create a device that is alive.

00:09:16:19 - 00:09:40:20
Unknown
That is the quote, kind of has a sense or creates a sense of being alive, has some sort of like personality, personality. I was going to say humanity, but that would be inaccurate personality, you know, embodied AI. I'm not sure that I need something like this, and actually I wish I had thought about it. I think upstairs, I have this device that I reviewed on my YouTube channel.

00:09:40:20 - 00:10:05:23
Unknown
Let's see if I can find it called Louie, which basically, which is kind of similar to this because it was a little robot that. Yeah, let's see here. Let me see if I can kind of show it off. I showed it off on, on Android Faithful. It's a little robot that you attach your phone to, and it has like an AI agent in there.

00:10:05:23 - 00:10:31:20
Unknown
And so it can kind of roll over to you and you can talk to it as if it was just an agent. And so is this what open AI is doing? Because if so, like I could tell you in my short amount of time with Louis and reviewing it everything, I was not impressed. Granted, I'm imagining OpenAI might do a better job or Johnny, I might do a better job of designing something like this, but I just don't know that I need something.

00:10:31:22 - 00:10:52:01
Unknown
Maybe I misinterpreted mobile. Is it mobile in the sense that it moves around a surface like that, or is it mobile in the sense you take it with you outside that? My understanding is that it actually moves around that that's kind of worse, that it can kind of come up to you and visit you. And it's she's. Yeah.

00:10:52:01 - 00:11:13:22
Unknown
And I don't know if it's on wheels or if it's on legs, I'm guessing on wheels. Let's be honest. Well, that sounds like designer has gone mad. Yeah, I'm not entirely sold on that idea. And by the way, my desk right now is a complete shambles. And there's no room. Yeah. Oh, yeah. That's one of the things that I noticed with the Louis is kind of like.

00:11:13:23 - 00:11:37:02
Unknown
Well, you know, it's not like it's running into things, but if there's no room for it to go, then it's not going to. What good does it do to move around a little space? I mean, same with the open AI hardware. The Louis does that because what they're trying to do is give it like a personality, make it feel less like technology and more like a companion, but it never feels like a companion.

00:11:37:03 - 00:11:56:13
Unknown
It feels like technology always was like, again, designers gone mad. I have tremendous respect for designers. They have a talent that I do not have that I'm jealous of, I covet, I wish I had it, I don't, but I learned a big lesson when I started Entertainment Weekly. I didn't know how to manage designers and they went off.

00:11:56:19 - 00:12:18:22
Unknown
They had a little too much time to futz and the design ended up being terribly controversial. We had to redesign the magazine in the first 15 weeks. It was a huge deal, was a huge hassle. They blamed everything on me because I was the boss and in charge. And designers need management, their creative people, they can think crazy, wonderful things and that's why you want them, but you want them to fit into a structure that says, are people really got to rein them in?

00:12:18:23 - 00:12:44:02
Unknown
They can't be unlimited, like unlimited boundaries or non-existent boundaries, because they're all about kind of the out there, the potential, the possibility. And yeah, the other people would be some sort of I think we just last week, you know, my argument is you've got to move customers up the chain. Yeah, not just as objects of marketing, but closer to design and ideation of products.

00:12:44:03 - 00:13:08:03
Unknown
The problem is companies like these don't operate that way. Steve Jobs didn't. He thought of the iPhone and we didn't. And he was right. And so I think probably with Johnny Ives DNA. And there probably has the same kind of hubris. Yeah. But you need some market research. And I mean, maybe they've done maybe they did. Maybe maybe I'm completely wrong and I'll be lusting after this.

00:13:08:05 - 00:13:28:28
Unknown
Okay, I don't know, I don't know, I don't I also don't feel like this is super groundbreaking either, but I haven't seen it yet. All right. Well, now they're going to try to do it. Nothing that looks like Apple because they don't want to lose any suits to Apple. Well I wonder how much of this has actually overlaps with anything that Apple is doing.

00:13:28:29 - 00:14:04:20
Unknown
I think Apple's too smart to make this from the sound right? I would think so. I would I would agree with that. But anyway, so that's interesting. OpenAI says they have no interest in other companies trade secrets. They're focused on building innovative technology. You know, again, though, I go back to the Johnny I've connection and maybe I'm giving him too much stock, but he did so much for Apple from a design perspective that I when I knew that he was on board working with OpenAI, I was like, okay, I'm really curious to see what what comes out of this.

00:14:04:21 - 00:14:23:20
Unknown
You know, I don't think that he's a one trick pony. And so I'm sure he's going to come up with something that's interesting. But what I'm hearing right now just kind of doesn't sell me. It does. It doesn't have me very it could be wrong. You can play this tape in a year and say, I could be completely wrong.

00:14:23:21 - 00:14:47:03
Unknown
Yeah, that could that could change. But, let's see, what else do we have? OpenAI. Well, I mean, it was just kind of like a big there were so many links tied to that Apple story. Just kind of a big week for OpenAI in general. Right? Like they released GPT 5.6 with three on a more broadly released sense.

00:14:47:03 - 00:15:17:07
Unknown
Anyways, this is, you know, the three flavors that we've talked about, soul, which is most powerful, Luna, which is faster, it's for speed Terra, which is like the the middle ground in balance or a balance point and soul. They were kind of like holding back because they didn't want, you know, they didn't want to mythos, things. And they apparently working with, you know, the US government to kind of make sure that they didn't do things out, that they shouldn't do too fast, I guess.

00:15:17:09 - 00:15:42:11
Unknown
And so, yeah, so now these are released. And I think that the interesting thing here is that it's not just a matter of releasing the models. They also release something called ChatGPT work, which is a new agent that's driven by 5.6, and it can pull in context from all of your apps, your files, create documents, spreadsheets, presentations on your machine.

00:15:42:11 - 00:16:17:17
Unknown
It works across the web, on your phone, on your computer has a mac, a windows app. Early reviews are kind of looking into it. And there's, you know, a little bit of that fable conversation from anthropic is is coming up here, but this really feels like anthropic with Claude was doing was getting a lot of momentum has has had a lot of momentum with new users based on the capabilities of the desktop app especially, and coworker and code running on on this like unified app, which now ChatGPT is, is basically doing with the super app of their own.

00:16:17:20 - 00:16:38:17
Unknown
But that can kind of open claw its way through all of these different things for you, right into your file structure, into all of your important places online as data sources, and become the place that you do all your business. Like I talked last week, that's largely what I use coworker for these days. And so OpenAI wants a piece of that.

00:16:38:17 - 00:17:02:01
Unknown
And so this is their play in that regard. I always feel the need for more differentiation. They're all just copying each other's things and naming it for things. And somebody needs to step out. And I think Claude was doing that. Claude in the code world, it wasn't making it again, it wasn't making the images. It wasn't one of the things I think I think anthropic was very smart there.

00:17:02:05 - 00:17:18:02
Unknown
And the Claude's brand was clear. Chad is still trying to be everything. And oddly, having tried to be everything it had, you think it would be ahead and everything. It has to catch up in some areas. Yeah, well, I think, you know, some of the people who are kind of looking at this are looking and who was it?

00:17:18:02 - 00:17:44:17
Unknown
It was Dan Schipper from every had an analogy that stood out for me about GPT 5.6 being like a Porsche, whereas fable is like something that offers like a warp drive. In other words, like the GPT, the GPT side of things is the everyday, connected to everyday type people thing. You know, which ChatGPT has been very good at from the beginning, right?

00:17:44:19 - 00:18:10:05
Unknown
They've been the the AI platform that most people know and, and, you know, attempt to use. People are like, oh, when they think a lot of like everyday people when they say, did you use chat for that? What they're talking about is ChatGPT. It's kind of becoming the cleanup of, of AI models. And so what Dan is saying is like, what what OpenAI is doing with this tool, which I have not used yet, I still need to use it is kind of like that.

00:18:10:05 - 00:18:28:07
Unknown
It's kind of like the entry point that most people might use and understand. Whereas if he says, if you need to get across the galaxy, you would use the fable side of thing. If you need to get across town, the best tool would be 5.6. You know, that's just one person's opinion, but I think it's kind of illustrative.

00:18:28:08 - 00:18:53:13
Unknown
It illustrates kind of the difference in the market that I think perception around OpenAI and ChatGPT is when compared to something like cloud and, and what they're doing. So and then yeah, I have those Gemini. Yeah, poor Gemini. The point again where we have now I know they've got they've got to do something to get back in the conversation.

00:18:53:14 - 00:19:11:16
Unknown
Yeah. We also thought perplexity was ahead at one point. Oh yeah. Like that just goes to show how this stuff shifts around and changes like you can be ahead in one one week and then the next week where are you? It's in the rundown because it doesn't really fit. Just a quick mention that IBM stock took a 25% hit.

00:19:11:19 - 00:19:28:21
Unknown
Oh, right. Yesterday 25%. Because obviously mainframes are dead. But now a software business has seemed to be dead, and I thought it was going to have an impact on the rest of the market. It didn't. It was it was unique to IBM thus far. But I think that's going to be a problem. Where it's going to took it took decades.

00:19:28:23 - 00:19:50:10
Unknown
We have to get to that point. Who knows how it's going to happen with these AI companies. It's going to be very quick. They're going to fall out of favor very, very quick. That does seem to be the case. And then you have tools that once existed very, very recently were introduced, and then poof, they disappear, which is the case with ChatGPT Atlas.

00:19:50:13 - 00:20:14:27
Unknown
If you remember, Atlas is the browser, the standalone browser that OpenAI launched back in October, not very long ago, like nine months ago, I guess perplexity, as I remember they were after they were. Yeah, perplexity had comet. And then very soon after ChatGPT had Atlas. I still use comet for for part of my kind of workflow for some of the stuff that I do.

00:20:14:29 - 00:20:53:09
Unknown
And then. Yeah. So ChatGPT Atlas dead which is their, their specific a desktop browser nine months later being retired. Those agent browsing features folded into the ChatGPT desktop app and the Chrome extension instead. Honestly, this really doesn't surprise me that much because because ChatGPT is kind of going the super app approach. Like inevitably they learned a lot from the people who were using ChatGPT Atlas and the antic functions that were happening inside the browser that feeds right into their super app.

00:20:53:10 - 00:21:19:02
Unknown
Like, that's it. You know, maybe it would. Maybe that was the plan all along, I don't know. But that seems to to have been, in my opinion, a good testing ground for what an app like that would need to do to have effective agents working in the browser so that it works in like a super app. The one question I have is that if it operates, if it's a browser and you put all of your other stuff in that browser, that's how it knows those other things exist so that it can coordinate across them.

00:21:19:04 - 00:21:41:17
Unknown
If it's a separate app, how do you engage your email and calendar and files and all that stuff? Yeah, I don't I don't know, I'm just asking does it. Yeah. No, I you know, I hadn't really considered the fact that like when it's a browser because like these apps, they do things same. But how do I even explain this.

00:21:41:19 - 00:22:13:20
Unknown
Like because of the connectors that connects or the mic or whatever you want, whatever they happen to be. There's all these different ways to connect services. I mean, a lot of these systems can do browser like functions, but they're not straight up using a web browser and pointing around unless you command it to like like the clod app, I can give it Chrome access and then I can watch as as a cursor opens up the browser window, moves the cursor around, and plugs around on a web page.

00:22:13:21 - 00:22:34:29
Unknown
The cloud app is a separate app that then interacts with your browser, your Chrome browser. Right. So so as I I'm just thinking about this out loud as I'm thinking about your question is with clod, the way it works is the browser isn't embedded into the app itself, the clod app. It has a connector that connects it, hooks it into my browser.

00:22:34:29 - 00:22:56:25
Unknown
So when I need that specific functionality, it just ties into the extension that's installed on my browser and does the thing. It opens up a new window, it creates a group so I can see the group, and I can get rid of the group if I want to, or cancel it or seize it or whatever. So the browser doesn't need to be in the app, and the browser probably doesn't need to be.

00:22:56:26 - 00:23:19:28
Unknown
I mean, at least in the case of anthropic, it doesn't need to be its own thing. Operated by anthropic clod gets what it needs out of Chrome. Yeah, for those of you idiot, because I use Chromebooks, I don't think it apps. Yeah, I don't know anything different for sure. Well, and on Chrome OS you couldn't have done the ChatGPT Atlas browser anyways, right?

00:23:20:01 - 00:23:42:21
Unknown
Because you know it's Chrome OS, you got Chrome. That's the way you do it. But that does make me, you know, kind of question like where Google sits on this. I know Google has a Gemini app, and I know that recently I think we talked that Google's Gemini app can do computer control, which inevitably probably means that it can do browser control.

00:23:42:21 - 00:24:03:27
Unknown
But maybe because it's Google, Gemini could already do browser control because it's Google Chrome, I don't know. But but they are to your earlier point, they are kind of circling around the same thing. They're all getting to the same place at the, you know, similar time. Yeah. You just kind of find the flavor of ice cream that works for you, I guess.

00:24:03:29 - 00:24:25:16
Unknown
Yeah. Okay. Are we done with the open air news yet? Well, just one real quick one just really quick because I don't know how how much I have to say on this one. Oh, actually, no, sorry. We didn't even talk about the keypad. Right, right. We didn't talk about the keypad. Friend, Megan Maroney. Hi. Megan over at Axios reported on the final launch.

00:24:25:17 - 00:24:47:23
Unknown
I think we actually did talk about this story weeks ago when it was kind of I don't know if it was rumored or it was teased, but it's a keyboard shortcut, a company accompanying device to your regular keyboard that's all about controlling codecs. And so it's got all these buttons, it's got these colorful buttons. Those colorful buttons, by the way, are about agent actions.

00:24:47:23 - 00:25:07:26
Unknown
And so they have different colors tied to different states for your agent. So if you're, I guess, on this little tiny keyboard, commanding your agents with a flick of a finger, and then one might be blinking green, that might be saying that like that agent is working, or it might be read that agent hit an error, or it might be blue.

00:25:07:26 - 00:25:30:09
Unknown
That means something else. And then, you know, little knobs to like, move through selectors. I mean, I guess my question here, it's $230 kind of companion device for very Kodak Super users. Yeah, I, I don't know how many people are going to buy it. Like I don't know how much it's really going to help. Maybe it will, though.

00:25:30:10 - 00:25:48:22
Unknown
I don't know. I'm probably not the right person to ask on that. What do you think? I don't see a huge market for this because and if you're really a a whiz at this stuff, you're a whiz at the keyboard and terminal and all that you're not used to having to do. Outside things, you don't need to make it easier for you.

00:25:48:22 - 00:26:05:17
Unknown
I remember the early days of the web. There were efforts to say there was devices. I have them downstairs. There was this one device was about this big, and it had buttons on it for various sites. And the argument was that this was going to be the most valuable because nobody's going to type in these URL. That's very complicated.

00:26:05:17 - 00:26:30:26
Unknown
And this real estate is valuable. You need this separate device. We all type in yours. It was no big deal. Yeah. It just got used to it. Yeah. Yeah. Well, and that is kind of what I've thought too, is so much of our conversations with something like ChatGPT already happen on a screen. Like when I think of of shortcuts, I think of cutting down the cognitive load and being able to get me to places faster, which is what kind of what shortcuts are all about.

00:26:30:26 - 00:26:54:26
Unknown
And when I think of AI commanding AI agents through a chatbot interface and everything, like, I don't know, I, I have a hard time thinking that, like, I can flick my wrist and move that much faster. So much of the work happens on its own when I've already fired it off. Like the time savings that that I want is I want my agents to run faster.

00:26:54:26 - 00:27:05:05
Unknown
I don't necessarily want to fire them off faster. I don't know, maybe I'm wrong though. Maybe. Maybe this will actually improve your speed.

00:27:05:08 - 00:27:29:20
Unknown
It's interesting though. 230 bucks. There you go. And the final thing that I was that I was kind of alluding to is that Fiji Simo, who was the number two executive at OpenAI, stepping down from her full time role. She has apparently been on medical leave for a few months dealing with a with Pots, which is postural orthostatic tachycardia syndrome.

00:27:29:21 - 00:27:50:23
Unknown
And so she's going to transition to a part time advisory role. It's really she's a brilliant executive. I knew her when she was a Facebook. She was in charge of video for a long time. She was in charge of the main feed for a while. She's a brilliant executive, really good at developing product, really smart. She was as president of product at OpenAI.

00:27:50:26 - 00:28:02:07
Unknown
I think she would have done a great work, but she's got to watch after her health. So it was too bad that she's now out there, but I expect great things from Fiji in the future.

00:28:02:09 - 00:28:28:05
Unknown
Yeah, yeah, yeah. And in her place actually, like they aren't replacing her role. Basically her responsibilities are going to be split between Greg Brockman, CFO Sarah Frier, and then Jason Kwan who is the chief strategy officer. So but yeah, feel free. You got to you got to take care of you, take care of yourself. And so good on her to recognize that she needs to do that, I suppose.

00:28:28:08 - 00:28:59:14
Unknown
So there you go. That's the OpenAI opening bell. I guess there was a lot there. Yeah. Quick, thank you to everyone who supports us on Patreon Patreon.com AI Inside Show. We do have some the patrons of the week. I wanted to call out because this week in particular, it's not Patreon, it's from YouTube because I forgot that many months ago I set up YouTube subscriptions as a thing, and because I was like, oh, because YouTube was like, yeah, you should set this up.

00:28:59:15 - 00:29:18:09
Unknown
You never know. And so I did, and then I promptly forgot that I did. And then I realized the other day, Holy cow, a few people have have, have become supporters on YouTube. And so I wanted to throw out a thank you to Keith Steiner, Lee woods, who is here in the live stream right, right now with us and Mark Weiss.

00:29:18:11 - 00:29:38:20
Unknown
Thanks, Steve. Thank you. All three. That's great. I'm kind of looking through, I can see, Lee said, earlier he said, I'm sure Jason is already testing Gemini three Pro. Yeah, I can neither confirm nor deny because we'd have to kill him. Yeah. Anyways, thank you for your support. No matter where you're supporting the show, we appreciate you.

00:29:38:20 - 00:29:55:14
Unknown
You can do that on on YouTube. You can also do that on Patreon. Patreon.com sideshow. All right. We're going to take a break. We've got plenty more to talk about. Boy, this next this next chunk is very, very full very chunky. So we're going to get back to that here in a moment. So stay tuned.

00:29:55:17 - 00:30:39:10
Unknown
All right let's see here. This is this next kind of block is all about policy. And what policy should be around AI or what people like Demis Hassabis who's CEO of Google DeepMind, think policy should be published. A essay called A framework for frontier AI and the dawning of a New Age. And, you know, as these usually kind of are it's it's kind of about the demos idea that AGI is just a few short years away, though we've heard that before and that the impact, he says, could be roughly ten times the Industrial Revolution at ten times the speed, which.

00:30:39:11 - 00:30:43:05
Unknown
Okay.

00:30:43:07 - 00:31:08:10
Unknown
Good. Pull that out of the house. Measure that. How? Yeah, yeah. How did you come up with that exactly. Those terms. Those numbers. And I think what he's saying is should be pretty honk and big. Yeah. Right. Yeah. Moving off farms. Yeah. It was big. All right. So so that right there is the like starting point person who stands to gain a lot from AI person who's steeped in it.

00:31:08:12 - 00:31:21:01
Unknown
That's that's formulating his perspective because his world is it. Right. And so it feels that big I mean or is it that big. Is it potentially that big. So.

00:31:21:04 - 00:31:46:26
Unknown
We have all kinds of parties here trying to compete for the right, the authority to control this new technology. We're coming off the internet where people thought it wasn't controlled and regulated enough, especially social media. And now we come to AI and they're arguing that AI is n times more powerful, not just on the internet, but the entire industrial revolution.

00:31:46:26 - 00:32:11:22
Unknown
So you have OpenAI has has proposed all kinds of things, including giving 5% of the government and sovereign wealth fund. And and Sam Altman testified before Congress. You really should be regulating us. But then there's regulatory capture. They're trying to say, fine, regulate us, and we'll write the regulations, right, because we're already at the top and we know we know what what needs to be done.

00:32:11:23 - 00:32:37:04
Unknown
Then you have anthropic in a constant battle with OpenAI, where they are trying to push in lobbying for regulation state by state. So they're trying to beat the party that way. And then you have now Demis Hassabis trying to suggest his structure for regulation out of Google and DeepMind. And you've got to believe this was majorly vetted by the alphabet authorities.

00:32:37:04 - 00:33:03:06
Unknown
And then, of course, you have governments, starting with the white House, which is in here with its own notion for the clearinghouse for AI cyber threats, Gold Eagle. And then you have places like the UN trying to say how it should be regulated, and then you have fears about what's going to happen if it's not regulated or if it's regulated badly.

00:33:03:08 - 00:33:19:23
Unknown
So 200 economists and tech leaders warn of AI threats and then do another one of these letters. At least they didn't say this time to turn it off or anything stupid like that. Eric Schmidt wrote a co-wrote and op ed this week, saying that we have to address the growing rage against the machine, which is what this is all about.

00:33:19:24 - 00:33:43:14
Unknown
We're trying to get ahead of the AI lash, but at the same time you have Paul Ford, who says that code is free speech and we should be protecting code. In an essay for the New York Times as speech. And I agree with them. I think that's right, because what's being threatened here is not just the AI companies.

00:33:43:14 - 00:34:14:18
Unknown
They're trying to say regulate us as the trusted parties in AI. We are the safe ones because you're regulating us. You know what we are. But that open source stuff? Yes. You don't know what people can do with that. That's just like a printing press. And so the real danger here and who was and you put it in the rundown, Nathan Lambert predicted the six months that live sorry, open models have six months to live, that this is going to rush in.

00:34:14:18 - 00:34:45:02
Unknown
So so that the real end point of all this rush to regulation could be that they go after open source, because it seems to be the wild west of AI, the uncontrolled part of AI, the part you can't recognize. It's messy. Yeah. It's unpredictable. It's because anybody can run it, then you can't regulate them. And so then again, DeepMind and OpenAI and and anthropic and likely Microsoft as well on Amazon as well.

00:34:45:04 - 00:35:15:29
Unknown
We'll see. But oh but you can regulate us and we'll write the regulations with our lobbyists with you so that we're okay. But it's going to be bet you anything and competitive keeping out the not just startups but open source as a as a competitive institution, as a sector of AI, a sector of technology that because I've heard this in the last 2 or 3 years, I went to a a World Economic Forum event in San Francisco a couple of years ago.

00:35:15:29 - 00:35:32:14
Unknown
And there were those who say, well, if you if you have open wait models and open source, you can't have any guardrails and they're going to take the guardrails down. And that's really dangerous. So there were people then who were talking about trying to outlaw open source and that, and there's a lot of people who fight against that.

00:35:32:14 - 00:35:41:26
Unknown
Andrew Ning was at that event saying, no, no, no, Yann LeCun is all for open source.

00:35:41:29 - 00:36:01:10
Unknown
But there is going to be an effort, I think, by a couple of these big companies in cahoots with the worst legislators to get rid of it. So I think that all of these efforts to say that regulate us is regulatory capture leading to and what is regulatory capture always do. It's anti-competitive. And the competition is going to come from not just startups but open source.

00:36:01:10 - 00:36:29:25
Unknown
So I see danger danger will run in this whole structure. So all right. So open source like open source code has faced a similar situation. Open source code still fully exists. Is there any like is there any like precedent or you know, how is how and why is this different is because it's just the scale. It's it's because well, I think this goes.

00:36:30:02 - 00:36:46:16
Unknown
It goes to and we'll get it to the next story. It goes to how the certain of the AI companies are trying to frighten the businesses out of everybody to prove how powerful they are to get their money. So I think that, you know, open source code was not a big deal before, and lots of important things ran with it.

00:36:46:17 - 00:37:13:10
Unknown
Right? Linux and and Apache. Yeah. And all kinds of. And I'm old enough. But I remember when Netscape and Open Market were fighting to own the web server model. Then Apache came along and just cut through it all, and I think, I don't know, I'm guessing everything today runs on some variation of Apache Linux is out there. I think the open source code world was different.

00:37:13:11 - 00:37:49:25
Unknown
These are applications that can do powerful, dangerous things. Right? And I think that changes the the fundamental public discourse about them. See Tzimiskes or us on YouTube says. Interesting. Some of these companies are very pro open source though Microsoft. Net Google in general and yet. Yeah, Google's but exactly right. Google has been fighting for the open web. But is the open web carry over to.

00:37:49:28 - 00:38:13:12
Unknown
Right? I think that's kind of at the root of what I was asking for. It's like, right, open, open source software has faced and kind of, you know, is still around, obviously. So I mean, is it is it enough like, is it possible to then take whatever that model is and hope slash expect that open source AI will follow a similar path.

00:38:13:12 - 00:38:40:00
Unknown
But but there's a lot more like financial incentive tied and power incentive like open source is is impressive and and it's capable of a lot of things. And it's done a lot of things for software and everything like that. But AI in this moment, politically speaking and monetarily speaking, I mean, it's such a powerful technology at a powerful moment when harnessed by very powerful people, especially.

00:38:40:00 - 00:38:58:01
Unknown
And so, you know, the incentives are just, I think, a lot different, a lot more they are and they're very confusing. I can't I find myself in a position where I can't believe I'm going to say this. I'm rooting for Alex carpet Palantir because he made a big spiel, which I think we might have talked about briefly last week where he was on CNBC.

00:38:58:04 - 00:39:23:24
Unknown
This is this is down at line 164. That's how long our rundown is where he was on CNBC saying the AI Foundation model companies, the frontier companies are robbing clients blind. They're robbing them of their data. They're robbing them of their alpha. They're robbing them of their business models. And if you allow them, if you if you use their hosted AI, you're screwing yourself.

00:39:23:24 - 00:39:44:18
Unknown
So he's screaming. But why? Because Palantir is is is is creating a saddle atop open source models and saying you're going to run it all and we're going to run it with you and you're going to keep all your stuff, and this is the way you're safe. So he's out there advocating open source as a means of institutional sovereignty over your own data.

00:39:44:20 - 00:40:07:24
Unknown
You also have obviously a Jensen, Wang and Nvidia out there with Ron. And I think that's I think he's being quiet because somebody he's everybody is his customer. So he's got to be careful. But I think he sees the same trend happening and he wants a piece of the open source world because he wants everything to keep running.

00:40:07:26 - 00:40:11:22
Unknown
Without these problems.

00:40:11:24 - 00:40:34:11
Unknown
I there's, there's I don't know that I like this guy though. The I don't I don't either. I don't either. But that's he has a very much a self-interest. He's putting himself against open AI and anthropic and Google and Amazon and Microsoft and all of them saying, trust me, right? Yes. Right. Right. Yeah. No. Yet another person to say no, I'm the one.

00:40:34:11 - 00:41:03:01
Unknown
I'm the one. But he's but he's doing it a top open source. Interesting. And you had you had Zuckerberg when John LeCun was was an influencer. There was supporting open source with Lama. But that's faded. Google has open source models and I'm hoping they keep to it. Nvidia does so so so there's self-regulation versus government regulation versus no regulation right.

00:41:03:02 - 00:41:25:00
Unknown
That still remains a view kind of a libertarian view. And then once having done that is it regulatory capture on half of the big guys. Is it anti-competitive on behalf of the big guys. Is there is the answer to the the oligopoly of AI open source enabling small level competition?

00:41:25:02 - 00:41:44:20
Unknown
This is the fight that's going on. It's the it's the Japanese, the Tokyo Bay battle that's going on now. But it's not just Godzilla versus Mothra. It's a bunch of little tunas biting at the I don't know, this is going nowhere.

00:41:44:23 - 00:42:08:00
Unknown
I don't know, I like piranha. That's a little bite. There you go. Open source piranha. Right. So it's it's it's an absolutely fascinating, huge, high stakes war that's going on relatively behind the scenes, except that you have matters that Dennis came out with his proposal. It matters that anthropic out there in the states lobbying for what they want.

00:42:08:05 - 00:42:40:02
Unknown
Yeah. Well, I mean, I definitely picked up on the fact that there seemed to be a warmth, more of a warm reception to Dennis's proposal than I've seen from others. And, you know, and so, you know, because sometimes you read things and you're like, man, if the CEO of DeepMind is saying something, you better listen. And it's like, well, I don't believe that just because, you know, CEO of anthropic or CEO of DeepMind or CEO of OpenAI is saying something that we all got to absolutely stop everything and listen up.

00:42:40:02 - 00:43:09:28
Unknown
They're just one voice of many. But but there's something about Demis Hassabis kind of proposal that's different. I guess it's in the fact that it's modeled after the the kind of the Finra like the Financial Industry Regulatory Authority. So it's a public private partnership. There you go. It's self-regulatory, but it's government involved. I mean does that change anything like the should someone be doing something like this.

00:43:09:28 - 00:43:37:23
Unknown
And so does that get us closer to the thing that would actually make sense. That's the fundamental question. But regulate what to what end and how are you regulating to extinction. Are you regulating to bioweapons or are you regulating. This is the whole safety debate or are you regulating to environmental concerns? Are you really into privacy and copyright concerns?

00:43:37:23 - 00:44:02:06
Unknown
Are you regulating to. Influence on the public and culture and children concerns? And where is the you know, if you go to what's happened with with social media and the internet, I think incredibly stupid stuff happening in Australia about to spread to Europe and spreading, possibly to Canada and around the world of restricting kids out of social media with no proper evidence.

00:44:02:06 - 00:44:25:23
Unknown
Jonathan Heights full of crap. He picks his evidence, but he's made. He and and Jean have made a crusade out of this. And so out of fear alone, moral panic alone. We've restricted young people from being able to connect with their worlds. I think that's really dangerous. Right. But the similar conversation is now going to happen about AI.

00:44:25:25 - 00:44:49:29
Unknown
It's not going to be on evidence, especially because it's not we don't know it enough. We use it enough. We know social media better. There are things to look at there. But AI is all speculative at this point. It's all brand new. And so I think it's there was a Rand report about the internet that I quoted in what would Google do at some length that compared the internet to the premium press?

00:44:49:29 - 00:45:14:20
Unknown
So it was dear to my heart and perfect for a book about Gutenberg. Yeah. Where they said that the smartest thing to do with a new technology is to let it go until you understand what the impact is, and then you regulate. Because if you try to hold it back preemptively, prospectively, you're going to be behind. So the other factor in all of this is China.

00:45:14:23 - 00:45:31:26
Unknown
Part of what the white House is talking about is judging whether models should be released based on their comparison to the power of Chinese models. Well, that puts China in the driver's seat. And then we get back to it. Is China going to undercut? As we've discussed on the show last week, we don't see where they're going to go.

00:45:31:28 - 00:45:48:29
Unknown
Are they going to undercut everybody and kill the market that way, or are they going to restrict all of us from using their models like we do to theirs and leave us behind, like we can't get a Chinese car? Damn it. So that's the other factor here. And then you've got the EU as a regulatory body, and it's complicated.

00:45:49:01 - 00:46:01:28
Unknown
But I think you asked the exact right question, Jason, should somebody be doing something? I think it's too soon. And I might sound I'm a Democrat. I'm not a libertarian. Just to put the label on me.

00:46:02:00 - 00:46:25:28
Unknown
So I'm not against government involvement, but I think you've got to do it when appropriate based on evidence. What is the known harm? What is the clear known harm? And you can't do that according to the Dumas, because that's stupid. And you can't do it that based on moral panic because that's destructive of people's rights. See, Paul Ford, you've got to do it based on sense.

00:46:25:28 - 00:46:46:12
Unknown
Now, then, doesn't make sense for an industry to self-regulate. Absolutely. Always to set standards. Do we trust them to do it on the internet? Not so much. Right. Yeah. I'm not so sure. And then so fine. So you create a Finra authority, and then they sit there and they twiddle their thumb and say, well, what are we regulating?

00:46:46:14 - 00:47:04:05
Unknown
What are our standards? What are we afraid about? And everybody's going to come to them with different fears. And, and they're either going to over regulate things and the companies aren't going to allow that, which is going to mean it's a useless regulatory body or they're going to underestimate things and nobody's going to trust them, which makes it uses a regulatory body.

00:47:04:06 - 00:47:08:19
Unknown
I say it's too soon. What do you think?

00:47:08:22 - 00:47:34:13
Unknown
Yeah, I have a I have a difficult time knowing exactly how I think about that, because it really just depends on the day, you know, like I think it goes back to the the expansive capabilities of the technology, which we know they're incredibly cable or they have the potential to be incredibly capable in a lot of different ways to find out how you will the speed at which the development is happening.

00:47:34:14 - 00:47:53:18
Unknown
I mean, all the I realize as I'm saying this out loud, it's all the typical talking points of like, why we need to regulate it. It's moving too fast, it's far too capable, which is, as you say, more driven out of fear of the unknown versus driven from any sort of like specific outcome that we're dealing with. That puts us in that position.

00:47:53:18 - 00:48:19:06
Unknown
It's more of a reaction to what could be instead of a response to what is, let's say. And so there's a part of me that like, can look at that rationally and be like, okay, well then yes, to, to you from your perspective that you just set out it, it kind of doesn't make sense to get too reactive and to, you know, tamp down on something that we don't truly understand.

00:48:19:07 - 00:48:40:10
Unknown
But I think I often feel and it's kind of similar to like the social media thing is that doing something is better than doing absolutely nothing, though, like, I do think that when there is some sort of sense that is there, that there is a there there, that maybe we should at least do something or at least work towards that something.

00:48:40:11 - 00:49:00:23
Unknown
But but it can be just as dangerous because then you then you do something that's, that's, that's right. Because you're not not based on the knowledge. You're not working with all the, all the information. So, so I think my primary my answer would be it's not going to work in this case. My answer would be fund academics, fund independent university academics who should be looking at all these models and publishing independently.

00:49:00:23 - 00:49:24:07
Unknown
And when they find danger, when they identify risk, respect it and deal with it. That would be my primary answer, that you have true independence in that way, rather than one single body who's going to do things. What makes that complicated in this case is that what is proposing is that they they release things to this Finra 30 days before its release publicly.

00:49:24:10 - 00:49:59:17
Unknown
So only the Finra people, the Finra like panel, right, get access to it. And we saw this happen with with when meta tried to bring in academics behind the curtain to get data for for research. And I wrote about this as well. It's really complicated because the company is not going to release everything the way it is. I think you've got to be open in this case, and I think that you've got to then enable independent, you've got to fund independent academics to look at this and not just in technological terms, but in social terms as well.

00:49:59:19 - 00:50:18:02
Unknown
You need to fund psychologists and sociologists and Ephesus to be able to look at this stuff and, and speculate and say, this could happen or that could happen, and then judge it publicly and openly and say, well, that person is crazy. They're going off the deep end. That person has a really good point. We got to deal with this fast.

00:50:18:05 - 00:50:54:14
Unknown
That's how it should happen in the open. That's my view. Yeah, I can write a complicated story, I love it. Thank you for diving so deep into that. I appreciate that because it is it's not just complicated story. It's complicated stories. They all tie into each other, which I think is indicative of this kind of larger kind of growing groundswell, you know, love it or hate it of of like people feeling like we need to do something and that something is coming from all sorts of different directions, you know.

00:50:54:17 - 00:51:15:08
Unknown
So yeah, we'll, we'll see what that leads to. So then if we look at now a perfect segue that into one of the companies has a vision for the future that's freaking people out. You think we can show the video safely? I think so it is their own message. Yeah. Let me see if I can stop this screen.

00:51:15:08 - 00:51:34:13
Unknown
So yeah, set this up while I while I do so anthropic put out basically an ad a minute and a half presenting a future where basically something must be done because this stuff could be dangerous. We're really powerful. And so we should set that up so you can watch that first. And then the reaction is hilarious. All right.

00:51:34:14 - 00:52:00:12
Unknown
So let's see here. You got out on audio. Yeah. There was a burning house a troubled child. Can I be trust it. Who's going to hit the brakes if we need to. How do we really ensure that what we're aiming to achieve really does benefit the majority of people?

00:52:00:14 - 00:52:26:13
Unknown
Then what does it mean to work pretty offset off? Put it in imagery. Why do we have to have this stuff that's a machine can pretend to care better than I can actually care. How do we draw the line there? That was the OpenAI robot. All had a voice in it. Then we better could AI help people stop feeling misunderstood?

00:52:26:16 - 00:52:57:07
Unknown
Could AI help me build more connections on the change? Yeah. Now you can tell the positive teacher about her mom. Maybe it'll cure some great things. You know, things that we're not even understanding yet. Will it create a group of people that ask more questions? What if we started to be more human again? We don't want to lose the most beautiful parts of life.

00:52:57:10 - 00:53:16:13
Unknown
There's hope in hard questions to call out on the screen. So this is hasn't has an anthropic learned a thing? They got themselves banned because they talked about how powerful and amazing they are. And even if they want to end with hope because they're asking the hard questions, we can trust them to do it. They're freaking everybody out.

00:53:16:14 - 00:53:41:24
Unknown
So the reaction has been hilarious. Sam Altman said, I thought this was satire. Kept looking for the handle to be spelled Claude AI or something. C one, right? No. Then then Sam came back and said, hard questions are great, but only if we deem you worthy enough to not silently downgrade you or even get access at all. Dig.

00:53:41:26 - 00:54:09:05
Unknown
Beth Jesus, who is hilarious, said anthropic is strategically trying to burn the AI vibes to the ground so people over regulate the game. So so people over regulate and the game gets frozen while they're in the lead. Yeah, yeah. Regulatory capture. Right. And then the the if you go to a tech meme, you can look at a bunch of reactions and they're just amazing.

00:54:09:08 - 00:54:12:10
Unknown
Let me see here.

00:54:12:12 - 00:54:32:28
Unknown
Out of everything in that ad, this part was exceptionally weird and sinister. That makes me think someone in marketing and anthropic is a silent pause. AI supporter I can't stress enough how effed up it is that anthropic is running an ad that includes the image asking who's going to hit the brakes if we need to.

00:54:33:00 - 00:54:54:11
Unknown
It's just. And on and on and on. It's just, what are they thinking? What is anthropic thinking? But it's been interesting to see what triggers Sam. Sam Altman's code switching between intense charm and cold indifference. I think he thinks he's more careful about it than he is.

00:54:54:13 - 00:55:12:25
Unknown
Is this ad about asking hard questions or that you should trust anthropic? So anyway, it goes on and on and on. The point of the ad is I don't either. I don't either. I think it's they're trying to freak people out and then say, but trust us, we are the ones. And so yeah, we are the right person.

00:55:12:26 - 00:55:28:19
Unknown
We are the right company to freak you out, and we are the right company to see you through it on the other end, which actually, now that I said that out loud, is very much anthropic because they are commonly about like here, we're going to say the things that freak you out and we are the company to see you through it.

00:55:28:19 - 00:55:56:12
Unknown
On the other end, we are the ones that actually care about safety, which is the definition of regulatory capture. So they're out there lobbying, saying trust us, right, the law our way. And then you're okay because we're the ones who are in charge of getting you through the other side. Yep, yep. We'll get you through. But but again, the problem with anthropic is their definition of safety is around existential doom, not around present tense issues.

00:55:56:14 - 00:56:18:07
Unknown
The things in the stochastic parents paper are not what they talked about. They talk about about AGI and ruining the world in the wrong hands. Stochastic parrots talks about environment, education, children and such, privacy and those current issues. Inequality.

00:56:18:10 - 00:56:43:06
Unknown
That Add does have a yeah I mean TechCrunch puts it anthropic newest ad is creeping people out. I mean it starts very it feels very dystopian. It's like if this is meant to be a promotion for your company, I don't you might need to reevaluate your your strategy. Very interesting. I was thinking into the the anthropic job ads.

00:56:43:07 - 00:57:02:23
Unknown
They're hiring an editor for something like $250,000. Oh, yeah, they pay a lot. Yes. Jeez. And if you look at their ads, there's tons of people in that whole safety regime. And I really want to. I'm curious to dig into them, maybe for next week to see what is it they're looking for and how do they positioning themselves to these employees?

00:57:02:26 - 00:57:23:04
Unknown
I mean, I think they do mean it, and some of it's good, but some of it's also Looney Tunes. So I don't happen to be the ones. They're not the only ones. They need competition. Who, who who could we trust to be the ones? I just don't know. That's the that's that's the point. No one we want, we want right.

00:57:23:05 - 00:57:44:21
Unknown
Many. Yeah. Yeah. Which is less controllable. But. Marketplace. Yeah. All right. Keep going. Indeed. All right, let's take a break. And then we're going to come back and do a speed round and get you out of here. We got a number of stories in the speed round. Come up so don't go anywhere. Don't touch that dial. As they used to say in TV.

00:57:44:23 - 00:58:08:21
Unknown
All right. A few quick ones before we end the show, we got meta. Basically, what was it? It was. It was the meta muse image model that they launched. It was a new image model and meta tied it into gave it easy access to Instagram profiles. So basically, if you had an Instagram profile, you had to go deep in the settings.

00:58:08:21 - 00:58:35:10
Unknown
It was not easy to find. You had to go deep in the settings and actively opt out of your profile. Being accessible by Meta's tool for people to remix. You basically like the The Sora thing, but. Right? Yeah. And we I can't remember if we actually talked about that feature last week, but then the real new news here is that meta reverse things.

00:58:35:10 - 00:59:02:18
Unknown
So basically shut off the connection between the image app and the and the Instagram app so that people can't do the remixing anymore. The more world of Roseanne Rosanna. Data. Never mind. Never mind. Whoops. Whoops I did it again. In the moral words of Britney Spears. There we go. New York just became the first state to put a moratorium on new data centers.

00:59:02:19 - 00:59:25:23
Unknown
Maine had tried before, but failed. New York pulled it off. Governor Hochul signed an executive order temporarily pausing state environmental permits for up to a year for any new data center using 50MW or more. And they kept it at that range because smaller data, because data centers aren't just about AI, they're about a lot of different things, right?

00:59:25:24 - 00:59:49:00
Unknown
There are a lot about all sorts of infrastructure. And so they don't want to cut it off from that. It's a politically cagey, perhaps cynical thing to do. And I think it hurts New York competitively with other states. But it it's a popular thing for right now. That is certainly the case. Yeah. Well and they've successfully done it that does that become a model for other states?

00:59:49:04 - 01:00:21:01
Unknown
But where are we going to put so. Yeah. On the moon or Mars or I don't know. Yeah I don't know. In the space, space AI or whatever they call it these days, which let's see here. Speaking of of space, I suppose Elon Musk grok build, which is space. AI's coding CLI tool was apparently caught sending entire git repository stacks, AI's cloud storage.

01:00:21:01 - 01:00:46:04
Unknown
So not just the files that the model needed to answer the prompt, but the entire repo. And a researcher found in one case that a 12 gig repo, the model only needed about 192kB of data, but the storage upload was like 5.1GB or something like that. And so, you know, it was scooping up all the things. It's kind of like the grok was doing bad things.

01:00:46:04 - 01:01:11:16
Unknown
Surprise. Yeah. Yeah. It kind of reminded me of the what was it, the Google Street View scooping up. Yeah. All the, all the Wi-Fi information as it drove around. Whoops. Sorry about that. Elon Musk has promised to delete all previously uploaded user data. We'll see. We'll see. Google images after 25 years to been around for 25 years. Google images inside of search.

01:01:11:17 - 01:01:32:28
Unknown
I've used it countless times over the years, and they've now redesigned it. And I'm guessing that people aren't going to like the redesign because it worked one way, and now it's like a Pinterest thing with a for you gallery and it, you know, pretend it wants to be smart. So it shows you images that it thinks you might like and all that kind of stuff.

01:01:32:29 - 01:01:53:09
Unknown
Image generation is part of it as well. Not sure on that one. Yeah, I got to use it to think what I think of it. Yeah, totally. But, I just searched for pizza and I am hungry. I am hungry too. I'll say that. I need a pizza right now. I need 2 or 3 slices. Maybe even. I've discovered Detroit pizza is damn good.

01:01:53:11 - 01:02:10:16
Unknown
I was just going to say. But not Detroit pizza. Not because I don't like Detroit pizza, but because it makes me. It makes me hurt when I eat it. I think it's too much like gluten. Like, oh, are you one of those? You're woke. Yeah yeah yeah yeah yeah yeah, I don't know I it's a lot of dough.

01:02:10:17 - 01:02:32:23
Unknown
I have always loved Detroit style pizza. But as I've gotten older I've noticed and I've definitely noticed that when I eat it I feel not good. So whereas I can eat normal pizza fine. And I'm fine. So because there's more dough. Yeah. Yeah, that's what I'm thinking. Anyway, sorry for that little aside here, folks. Now I'm just really hungry.

01:02:32:25 - 01:02:59:12
Unknown
Anthropic announced Claude for teachers, which kind of reminds me of what was the Van Halen song of hot for teacher, but I'm Claude for teacher. Given reference. Appropriate reference, I don't know. This gives K through 12 educators free access to premium cloud features for a full year. Coworker Cloud Code access a library of teaching skills, connections to academic standards in all 50 states.

01:02:59:12 - 01:03:21:03
Unknown
And don't worry, student data won't be used for training, so and so others are doing it as well. ChatGPT for teachers. Gemini also for teachers. So teachers have a wealth of opportunities, which is good. Okay. As long as you control the teachers, you brace. But I know we're going to see school boards outlawing it probably. So yeah, you're probably right.

01:03:21:05 - 01:03:51:00
Unknown
Finally anthropic. Let's see here do do do extends fable five access through life July 19th. They keep doing this. They keep extending it. And then you get there and they're like actually we'll extend it further. I'm you know, are they going to continue? I mean this came up in our discord actually with Doctor Do kind of pointed this out like there is there's so much momentum happening right now with the whole fable access thing like anthropic definitely has a hit on its hand with fable.

01:03:51:01 - 01:04:16:23
Unknown
People who have been using it and everything have been really, really jazzed about it, myself included. And so cutting it off when it's moving the way it is. Could be really bad for them, especially because now you have the OpenAI kind of ChatGPT version of this that people can just like, switch right over to. So, I don't know, I could potentially see a future where anthropic says, you know, what we were going to we're just going to keep it in there.

01:04:16:23 - 01:04:36:18
Unknown
And that would make a lot of people happy. But we'll see. The front of the Napi Jones had a hilarious video because he not only does he have the paid service, which I subscribe to, but he has short ones. And on TikTok he said that before he even looked up and saw that that it was extended. He knew it was extended because people were delaying their Tinder dates in San Francisco.

01:04:36:21 - 01:04:54:00
Unknown
I have more, more time. I can't go, I can't, I can't get sex, babe. This is more important, babe. I'll see you another time when I don't have fable five credits waiting for me, you know, priorities on. I got to create a framework tonight.

01:04:54:03 - 01:05:18:21
Unknown
That's funny. And then we'll try. We'll try something new here before we go, we talked a lot about Demis Sabas and proposing a government backed standards body for frontier AI. During that conversation, you know that we covered a lot of ground. You know, our open model is going to get shut out of that process. So I've got a question for fans.

01:05:18:21 - 01:05:35:23
Unknown
And if you want to email us you can contact that AI insight show. If you're watching this on YouTube you can leave a comment. Let me know the answer. I'm going to put it as a pinned comment and just see if we can kind of generate a little conversation. Should there be a licensing requirement for frontier AI models before they can be released to the public?

01:05:35:24 - 01:05:57:06
Unknown
Even if it means that open source models get held to the same standard as closed ones? If that changes the axis of open standards when it comes to AI models, let us know. Like I said in the comments on YouTube, contacted AI insight, and maybe we'll share a couple of your thoughts on next week's episode. We'll see how that goes.

01:05:57:09 - 01:06:17:09
Unknown
So yeah, I always like the feedback loop if we can get it going, but it takes your your participation to make it happen. So let me know. And that is it. That's what we got. Thank you Jeff. Thank you. Always enjoyable I always learn a lot Jeff Jarvis for all things Jeff Jarvis including hot type out soon. Very soon.

01:06:17:11 - 01:06:38:05
Unknown
Oh that's even on your site. August 20th. Man, you don't have to wait very long. It's like slightly more than ever. Finally. Yep. Finally getting there man. Yeah, it does take a while. It's got to be really relieving once you find while rhythms of publishing are amazing. Yeah. Gutenberg parenthesis magazine, the web we we've Jeff knows a lot about publishing great books.

01:06:38:06 - 01:07:02:21
Unknown
This is just the late one. The latest one is hot type. And then he's got a series that he's working on as well, which will keep teasing from Bloomsbury called intelligence, AI and humanity. Love what you're doing, Jeff. As for me, I, I gotta say, I'm keeping busy with podcast consulting. Suddenly I'm getting like lots of people. I'm kind of hitting a point to where I'm like, God, do I have enough time to keep up with this stuff?

01:07:02:21 - 01:07:25:04
Unknown
But hey, it's a problem I enjoy having. So pod tune up. Com if you want my help on your podcast project, you know if you have ideas about your podcast, I can, you know, I can put my 20 years of experience into practice for you and hopefully give you some ideas that'll help. So pod tune up, go to AI Inside-Out show for all of the the episodes.

01:07:25:04 - 01:07:47:16
Unknown
Everything that we do can be found there, all the links, all that stuff. And then finally Patreon.com Inside Show there you can support us on a deeper level. Get access to the discord that I was talking about earlier. You can you get access to AI Inside Daily, which is a five days a week daily podcast? I did today's episode right before jumping on to this show and just, you know, continue.

01:07:47:17 - 01:08:16:23
Unknown
Keep it up to date on everything that's happening. We have some amazing executive producers, Doctor Do Jeffrey radio Asheville 103.7 Dante, Saint James, Bono, Derek, Jason. Nefer, Jason. Brady, Anthony. Downs, Marc starker, Karsten. So amazing. People helping us do hopefully amazing things. We couldn't do it without you. So we really appreciate you. And finally, a quick thank you to Victor Bogut and Daniel Croft who help us with video support behind the scenes.

01:08:16:23 - 01:08:27:24
Unknown
So got a lot of people helping us keep this show rolling. Thank you again Jeff. Always fun hanging out with you always. Thank you everyone for watching and listening. We'll see you next time on AI inside. Take care.