Gemini 3 is Here

00:00:00:00 - 00:00:22:02
Unknown
Coming up next, Jeff Jarvis and I explore Google's Gemini three with its native multi-modality and advanced a genetic behavior, and I show a little example of what I was able to create with it Microsoft's AI copilot rolling out across windows and the enterprise. Jeff's chat with Yann LeCun in Brooklyn and more. Coming up next on the AI and side podcast.

00:00:31:18 - 00:00:32:21
Unknown
Hey, everybody.

00:00:32:21 - 00:00:49:11
Unknown
Welcome to another episode of the AI Inside Podcast, the show where we take a look at the AI that is layered throughout so much of the world of technology. Sometimes you get one episode, a week, sometimes you get two. We'll talk about that in a second. I'm one of your host, Jason Hale, joined, as always, by Jeff Jarvis, my co-host.

00:00:49:11 - 00:01:07:19
Unknown
How you doing, Jeff? Hello there. How are you? Good doing. Doing all right. So we'll talk a little bit about it a little bit later in the show. But yesterday's episode, we had it all lined up. I'm sure. You know, if you didn't realize there was an episode in your feed. Check it, because it's behind this one.

00:01:07:19 - 00:01:25:29
Unknown
Now, if you're getting this this episode, I had high hopes, high intentions for that episode that Google did their thing and they announced their big, big news, which we will also talk about. And it just basically sucked the oxygen out of the air for everything else. You think they maybe tried to do that while Microsoft was doing some?

00:01:26:01 - 00:01:43:02
Unknown
Yeah, I know, I'm sure it was innocent. I'm sure. Yeah, I'm sure it was going to end it all. Just total coincidence. So anyways, we'll talk about, kind of what that was all about and everything a little bit later. But if you did see that episode, sorry for the confusion. Yes, there was an episode yesterday and there's an episode today.

00:01:43:02 - 00:02:06:28
Unknown
See, we just give you so much AI goodness. We can't we can't help ourselves. We can't help ourselves. But first, let's, start things off, not talking necessarily about anything, particularly newsy, I think. Just think this is a really cool opportunity. You, Jeff, had the opportunity to attend the pioneer work scientific controversies, conversation friend of the show, Yann LeCun.

00:02:07:01 - 00:02:26:22
Unknown
From meta was there on stage DeepMind's Adam Brown. And they they seem to be, labeling it as sparring about AI, but I don't know if it was. That was a sparring. It was it was a debate. Yeah. It was, it turned into obviously cordial as could be an intellectual as could be. But I was so delighted.

00:02:26:22 - 00:02:51:28
Unknown
So this invitation came to Jason. Jason said, I can't quite get to Brooklyn. And so what is Jarvis, you want to go? This is that when I trudged out to, through the Brooklyn Battery tunnel to the hip part of Brooklyn. Of course, it's all hip. And went to this wonderful event. Jan 11, has, played host to 20 odd of these conversations of all kinds of fascinating topics.

00:02:52:01 - 00:03:12:27
Unknown
John was on once before. And so this was was a really interesting talk because obviously everybody wants to know what your on the court is up next. Yeah. We have it from the left. That is leaving matter. He said he would neither confirm nor deny with a with a sly grin. Did they do they ask. Oh stage like, hey, by the way, of course you've got to ask if you've got to make an announcement on our podcast.

00:03:12:27 - 00:03:35:18
Unknown
Yes. Yeah. And see how. Right. And then I had the great privilege of meeting you on afterwards, which was great because we oh, awesome. Spoken on this podcast. Yeah. And what was it actually a lot of what he said, Jason was, recast from what he did, told us when when we interviewed him here at episode 63, by the way.

00:03:35:19 - 00:03:54:02
Unknown
Thank you. And I recommend watching it. And, and he talked about his vision here. And underscored more we've talked about since, one of the PowerPoints that I should point out on the tour, on the show some time ago, which we talked about last week. He said the labs are basically a dead end all the way to AGI.

00:03:54:05 - 00:04:28:19
Unknown
And so he was asked about that. John presented what is to me a realistic, reasoned, but still, really exciting, real optimistic vision for AI. Yeah. And I think optimistic. All in all, he, you know, he was asked straight, Will this be doom or a new renaissance? So he said, a new renaissance. He does believe, as he told us back in the day, that, the machine will get smarter than human beings in every, most, every region of thought.

00:04:28:21 - 00:04:55:29
Unknown
But he has been very clear that labs are not the path. They're that they're a dead end. That and Adam Brown from DeepMind was the opposite. Labs. Just look at the look at the progression of what they've been able to do. If we just keep that progression going through scale, we will surely get there. And John says, no, there's limitation because all it does is predict the next word, that language is a finite universe to work in.

00:04:56:01 - 00:05:15:00
Unknown
And he sat there with a glass of water and said, for me to go through the process of describing what it means to move this glass of water and that my hand doesn't go through my body and it doesn't go over here, and I have to take the microphone and put it over here. That's a very complex view.

00:05:15:02 - 00:05:40:08
Unknown
Right. And so he was quite direct and firm versus Adam saying limbs ate it. And so they had they had a distinct disagreement there. Now, did Adam, did Adam create or propose a pretty compelling argument other than saying, well, I just think this is the case or reject from your view. It was trajectory. Look how far we've come with where we are.

00:05:40:08 - 00:05:55:28
Unknown
If we just keep going, we will be there, which is kind of the scale argument, which is that if we just keep throwing scale at it, keep throwing money and building and building, just think of all the great things it's going to do. Yeah, right. So that's that's the beauty and and so there was a clear disagreement here.

00:05:56:00 - 00:06:17:13
Unknown
And John is presenting an alternative version. He complained that the scale LM version, what he called when I met him afterwards, he called the people who are land pilled, that they, are sucking all the oxygen out of the room. And not only the oxygen, but all the venture capital and all the chips and all the data centers.

00:06:17:16 - 00:06:40:13
Unknown
So my hope is with his, the respect that he has and the prominence he has, that he's going to draw competitive investment into another area that he's argued for, world models. He, he and Fifi Lee and others at a paper we mentioned last week in which he talked about, understanding reality through videos, how would he chairs or in the room?

00:06:40:15 - 00:07:00:04
Unknown
And I think that's where, you know, afterwards I said, it's pretty clear where you're going to head, you know, is. Yeah. You know, it says he's been talking about this. He's been proposing these models. But we need to see investment in it. And, so it really two things, Jason. One is that he, asked about the dangers.

00:07:00:06 - 00:07:22:29
Unknown
Adam Brown said, yeah, there's there's good stuff and there's also dangers. And pretty much echoed what we hear about that. And John said, no, the next version of what we're going to see in AI is things that are directed to tasks that the present AI doesn't solve a problem. It has this ability to predict things, and you present it with problems and it's based on that.

00:07:22:29 - 00:07:42:00
Unknown
But he thinks there'll be a different version. I hope I'm not misquoting him here or mis paraphrasing him here, but it will be be given direct tasks that deal with this world model of reality and that that's controllable. You tell it what that task is, you tell it what it is not. Those are guardrails that will work.

00:07:42:03 - 00:08:07:02
Unknown
And so he was optimistic about that. He was optimistic about the reticence in the sense that we will have these aides helping us do things. All over and in certain areas. They already are, right. They're already playing chess better than we are. And so on, so forth. And then finally he gave a rousing, defense of, open source, and emphasized that strongly.

00:08:07:02 - 00:08:14:14
Unknown
And I think we have a little video clip if you want to play that for a minute. 22 yeah, indeed. Here we are. It's about a minute 20 but yeah, it's really is

00:08:14:14 - 00:08:26:02
Unknown
if we don't have those mature systems okay. In the future, every single one of our interaction with the digital world will be mediated by any old system.

00:08:26:04 - 00:09:03:23
Unknown
Right. We're not going to go to the website or search engine or whatever. We just talk to our AI system. Our didn't so our entire information diet will come from, you know, systems. Now, what does it mean to culture, language, democracy of everything. If those systems come from a handful of companies from the west coast of the US or China, I tell you, no country in the world outside the US and China likes the idea.

00:09:03:25 - 00:09:29:16
Unknown
So we need a high diversity of AI system for the same reason we need a high diversity of the press. We cannot afford to have just a handful of proprietary systems coming out of this. One of our companies. There's one thing I'm scared of, and that's it. Okay? If we don't have open platforms, we're going to have, you know, capture, post information flow.

00:09:29:18 - 00:09:33:26
Unknown
But is it for us companies, some of which we may not like?

00:09:33:26 - 00:09:48:09
Unknown
I apologize for the videography. My arm was getting tired. I didn't know that. Don't worry about it. You can find the entire event. John 11. Yeah, at her Substack, has has the entire event up with much better. Yeah, that's right than mine.

00:09:48:11 - 00:10:08:23
Unknown
That's right. But, yeah. So it was a it was a great event. I'm very grateful that they invited you when you managed to get me in, and that I got to meet Yann LeCun. He is the person I follow more than anyone else in this world. He's the one, the other. The other thing he did and I'll stop on this, was when Adam was making the trajectory remark.

00:10:08:26 - 00:10:26:13
Unknown
Yann is old enough and has been around long enough. He's younger than I am, but he's still old enough to have seen all of these cycles in AI, and he went through each one of them. If you watch the video, you get a history lesson. This was going to take over the world, and that was going to take over the world, and this was going to take over the world.

00:10:26:16 - 00:11:20:28
Unknown
And so he's seen it before. And that realism and experience, I think is critical here. Unlike, the Sam Altman's who were frankly new to the field, Yann is in fact a pioneer of the field. And that lived experience and intellectual experience matters. So I appreciate his kind of, his his balanced rationale around AI because I hear when I hear people get too riled up and to just imagine and to, you know, bombastic with the predictions and everything, I don't know, the hair on the back of my neck starts to pop up and I start to just feel like, okay, nothing is as amazing as you might be believing about the potential

00:11:20:28 - 00:11:40:02
Unknown
of something like this. It all has, you know, positive and negative impacts. And that's not to say that there aren't really great things that come out of it, but there's also, you know, a word that it can't do amazing things. But, you know, there is the reality that maybe the technology, the way it's being implemented right now isn't the right way.

00:11:40:02 - 00:12:03:12
Unknown
Or maybe there's just a bunch of different ways to do it. And one way, you know, actually takes, you know, safety and, security and all these things more into, consideration than others, all these factors. And when I look and listen to what Yann LeCun says about this stuff, I don't see it as a Yann hates AI, which I think some country.

00:12:03:14 - 00:12:21:08
Unknown
Yeah, some people see what he says. And because he's not rah rah rah rah like, oh, yeah. What the what the hell does he know? Like, doesn't he see all this stuff? And I just see him coming from a more balanced perspective. And that matches me. That matches with me. There's a there's a, there's a Lakota meter.

00:12:21:11 - 00:12:39:20
Unknown
Right. And so on the one hand on the, on the and the paradox of AI, of course, is that the most aggressive are also the most do. Marie. Right. You go to the anthropic and so on and you get regular warnings from Sam Altman is warning, I'm so powerful. You got to deal with me, right? Yeah. And the other end, well, the other, other end you have here is.

00:12:39:21 - 00:12:56:08
Unknown
But somewhere up here you have Gary. Mark, I got my hand in the for. For those of you who are watching, I'm not doing a good beta. So what end is, is, is the far end of, take over the world? Destroy it or not? The other end is I hate I up here somewhere is Gary, Marcus, and Gary.

00:12:56:08 - 00:13:21:06
Unknown
Marcus is, tries to be the realist of realists, and he tries to puncture every bit of hype. So he once in a while zaps. Yeah, And says yawns too optimistic. Yawns a little bit up on the on the meter here. But I think he's in the middle range where, he has the realism. You just described, but he also has the optimism of what this can accomplish.

00:13:21:12 - 00:13:39:05
Unknown
He's not anti-Islam either. There was a there was a headline out that said that, you know, everybody disagrees with you on the court. And he said, no. He said more and more agreed with me. I think that's true. And one thing I meant to ask him, but I didn't have the opportunity to was, I think you hear echoes this different.

00:13:39:05 - 00:14:03:09
Unknown
Different. You hear harmony from Johnson, one, talking about role models to say, well, Yan does Johnson's more from the robotics end and yarn is more from I think the video and right now in terms of how they're they're thinking about this, I'm talking about this. But but the world model need is the same and the optimism is the same.

00:14:03:12 - 00:14:29:11
Unknown
So it's interesting to to start to plot these folks out on where they stand. But yeah, yarn is the mature, realistic, optimistic, experienced and seen AI executive. So far I'm on Team Yarn. Yeah, yeah, I would agree. I'm really happy you got the opportunity to talk to him. You just see him to talk to him. That's that's really great.

00:14:29:14 - 00:14:44:17
Unknown
I really was when I first got that email, I was like, all right, let's go. I just thinking it was in San Francisco or whatever that was in Brooklyn. I know you think you're the center of the universe, don't you, Jason? You're not. I don't think it. But I hope things come here. And you know, some things do come in here, folks.

00:14:44:17 - 00:15:05:11
Unknown
I'm in New York, so we're close networks. We can. We can cover either coast. You got your stuff? Totally. Absolutely. So I'm really happy we were able to have kind of. If you have the good Korean spicy meatloaf sliders that were at the, the, the event, they were very good, very excellent. Well, I'm super hungry right now, so that, that sounds good to me, too.

00:15:05:14 - 00:15:29:23
Unknown
I don't know if ozone nightmare is hungry, but he's hungry for what we do here at Air Inside. Thank you. As I gave us a super thanks. Thank you. Ozone nightmare, says the quote. Limbs will be smarter than us. End quote statements drive me mad. Beyond being a currently baseless declaration that limits what smart and intelligence means via a flawed equivalence between humans and I.

00:15:29:26 - 00:15:52:13
Unknown
Yeah, I think I think that's a really good point and important. And John also raised that. I mean, he was asked toward the end about consciousness and he was kind of trying I don't know what how you define consciousness. What was interesting is he said that like the Nutella worms, but I will have tasks and goals that that's what you give it and thus it may have emotions.

00:15:52:15 - 00:16:20:27
Unknown
Emotions. Yeah. In the sense that when it can't meet that goal that it has been given, it will look to us like an emotion. But all of that is anthropomorphic in my view. It is. Well, I was going to say that. Does it have it or does it project it? Bingo. Well said. Very well, said Jason. I think that's that's the way you know, I yeah, I think there's a big difference between having an emotion and just projecting it or, you know, knowing to say it and knowing that in this circumstance, in this situation, this is when this emotion often comes up for humans.

00:16:20:27 - 00:16:44:04
Unknown
Therefore I will project it, you know. And the other thing I would ask, yeah, yeah. I, he was off the stage, so I didn't want to bug it with too many questions when I, when I met him. Sure. Is whether in his vision, and I don't know the answer to this, but I almost sense that the AGI crew is thinking we're going to have one machine that can do everything.

00:16:44:07 - 00:17:07:17
Unknown
We're going to have the ultimate machine. I sense, and I can be wrong about this with Yann, that he's talking about plurals that the the totality of what I can do in all kinds of areas will add up to between us in individual areas. So to your point, ozone, I don't know which to call you here.

00:17:07:17 - 00:17:27:23
Unknown
Your real name or your onscreen name, but, is that, this idea that you have one general intelligence that beats all? I agree, I think that's I still think that's B.S., but I can start to by what John says when he starts to split that up. We already see it. We've discussed this, right. It's better at us than chess.

00:17:27:23 - 00:18:07:12
Unknown
It can do things linguistically. It it can do. But he he. This is what he told us when we interviewed him. He said, come back to me when it can load the dishwasher. When it can learn to drive in ten hours of instruction on its own. Right. He as far as he concerned present efforts at what supposedly autonomous driving or cheating because they have all this other cultural when you when you could really do those things you can learn those things like a ten year old can or when you can learn how gravity works like a nine month old can, then yeah, you're starting to talk about general intelligence, but he says

00:18:07:12 - 00:18:29:27
Unknown
we're not there in any case. Yeah, yeah, we might be able to load a bunch of data in there and feel like it really understands these things. But there's a lot of information the world model that he often talks about that is completely lacking there. And so a true understanding in our kids doesn't actually exist. It seems incredibly smart because it's got all the data there.

00:18:29:27 - 00:18:51:20
Unknown
But connecting the dots is kind of impossible in to certain aspects of things that you and I as human beings on planet Earth, you know, know very well because it's a lived experience, not just a knowledge experience. And John 11 did a really good interview this whole discussion, she started with that question. She asked them whether they understood.

00:18:51:23 - 00:19:14:08
Unknown
And Adam was on the yes, it's understands. And John was on the not so much side define understand. Yep. What does understand really mean. You know does it understand because it knows that when it hears this it knows to query against a data set and find the answer does, you know, like, yeah, it's it feels like a little difference.

00:19:14:08 - 00:19:40:27
Unknown
But they're actually big differences when you're talking about these things becoming, you know, human like, you know, they're miles apart. I think this was also just a great one more second, a great day, York weekend, because Sunday night I went to Brooklyn, to that Saturday I spent all day at the New York Historical, with a symposium on tolerance in, New Amsterdam and Amsterdam and the New Netherlands.

00:19:40:29 - 00:19:58:19
Unknown
Russell Short, who's a brilliant author, brought that together. So it was just great. It's a place where you can go was a 400, 500 people in a room, and they're talking about ideas, and it's the whole thing. Totally. That's awesome. When I love New York. Yeah, I love New York. Do you have a t shirt? I heard New York.

00:19:58:20 - 00:20:23:20
Unknown
No, I wouldn't do that. You see, we're too cool to do things like that. Yes. It's like living in San Francisco and calling it Frisco, which. Yeah. Or going to Fisherman's Wharf or. Yeah, let's go to Fisherman's Wharf. Yeah. Or going to Girardelli Square. The brilliant. When I lived in San Francisco, the brilliant thing I always said about San Francisco is that it gets as the tourists, the San Francisco, the tourist is nothing like the San Francisco.

00:20:23:20 - 00:20:50:10
Unknown
The residency. Yeah. And it keeps them there. They're all down to Fisherman's Wharf in General Square. Good. Have a good time. Have your Irish coffee. We're not having it. Yeah, yeah. Totally true. By the way, you do have permission to use those zones. Name? Joe. Oh, okay. There you go. That's what it says. All right, so yesterday was a big deal for Google, as I alluded to earlier.

00:20:50:10 - 00:21:14:20
Unknown
And why is that? Well, I had heard and we had heard that possibly Gemini three was going to be releasing very soon. Then it was imminent. And sure enough, yesterday morning, an hour before my fun or our fun embargoed interview was set to go live, Gemini knows we're going to have big news. Oh well, it's a little oh, the oxygen just left the room anyways.

00:21:14:20 - 00:21:35:00
Unknown
Not too not to, you know, push down the news and the quality of news that is Gemini three, because everybody flipped their lid over this and, we can talk about it a little bit and I can actually show you a little bit of a demo. Some of the things that they've been touting, I haven't gained access to yet, even though I have gentlemen, I three in my Gemini instance, or on my account.

00:21:35:00 - 00:21:51:22
Unknown
But anyway, just for folks out there, just so you know, because I was cursing as I do, I don't care. I get this. If you I'm on Chrome and if I go to Gemini, you don't see it. Evidently. You've got to go to the to the deep reasoning. And under there you will see that it's Gemini three.

00:21:51:24 - 00:22:12:13
Unknown
Otherwise the default Gemini two. And on my mobile, I still haven't gotten three. Yeah. It's not like if you go to, ChatGPT, you'll see up in the left hand corner, it'll say dash five or whatever the version is right here. It's very unassuming on the Gemini, interface. But if you pull that off, I'm sorry. Right. Yeah.

00:22:12:15 - 00:22:35:08
Unknown
I mean, these are both using three is my engine. Oh, no. When mine fast is two five and thinking is three. Oh, okay. It's so interesting how these things they didn't really Google and branding. They just have a problem with branding. Yeah. But you really have to look for it to know oh there's three pro you know, that's when I stop for a second I found it.

00:22:35:10 - 00:22:54:20
Unknown
Anyways, let's, talk a little bit about what it can do. Native multi-modality, which when I heard that, I was like, well, wait a minute, aren't these systems multimodal already? And what it actually means is that it can take multiple types of input, and ingest them together. So it could be text, video, audio. It doesn't do them in, in sequence.

00:22:54:26 - 00:23:15:24
Unknown
It could actually contain it all at once and do it simultaneously. So, you know, it just removes a little bit of the barrier of what you throw at it. And when and what it's capable of understanding and pulling from. So that's cool. It can create magazine style layouts. Does this thing give you this article on The Verge?

00:23:15:24 - 00:23:48:25
Unknown
Give any sort of an instance of what that actually looks like? Basically, it's very visually rich. Let's say. And I don't know that that is the best indicator of what that is. But let me see here. Well, I'll just play the promo video while I talk. Yeah, it's probably the easiest. Yeah. Kind of creative style layouts, very visual interactive flashcards, web apps, games, tools all done through something called Dynamic View, which is this visually rich and interactive kind of, addendum to the Gemini three experience.

00:23:48:27 - 00:24:04:04
Unknown
That's the thing that no matter how hard I tried, I really wanted to, that's something different. I'll talk about that in a second, but I really wanted to get access to it, and I don't have access to it in my dropdown. It would appear down here in Tools Dynamic View, and I'm not getting it. So I don't know why I'm not getting it.

00:24:04:04 - 00:24:33:12
Unknown
Hopefully I do too. But it creates these really rich, interactive things throughout the experience that are just, a lot of, yeah, very, very, very appealing. It really pulls you in generative user interface, a lot of animate, a lot more animations, that sort of stuff. Google also says, stronger agent like behavior so it can handle more complex tasks, can also reliably, reliably plan ahead over longer horizons.

00:24:33:14 - 00:25:05:07
Unknown
And then they have a new thing called anti-gravity, which is a multi-agent. I'd, I'd for those who don't know, integrated development environment. So this is really about development using all of these different agents and models all together to create pretty complex code projects. With all of its first party and third party models, you know, integrated in and it also produces something, along the line of artifacts as it completes its work.

00:25:05:07 - 00:25:30:00
Unknown
And that's all meant for verification of its work. So that's done by way of screenshots, plans, tasks list, browser recordings. And so it's not just creating the thing, it's also creating more understandable ways to check the thing that it created, which I think is interesting. And Google stock went up. Yeah. But $304, almost three. The $4.

00:25:30:00 - 00:25:59:22
Unknown
It's down now to 294. But the market loved this. And Google you know, had had cooties when it came to AI. Oh they're behind OpenAI. That's Microsoft I think Google has well caught up in the market's view. I mean, there's yeah, there was a lot of, a lot of excitement around Gemini three and even not just yesterday because it launched, but like leading up to it, we started kind of seeing little hints of, of it on the horizon.

00:25:59:22 - 00:26:32:08
Unknown
And I could sense that people were getting really excited about this. I think another thing that's really interesting is that Google pointed out, made a, made a point to point out that they are making this model less about flattery and more about telling you what you need to hear versus, you know, OpenAI and ChatGPT, which with forero had that kind of flattery, personalized perspective that they then kind of jettisoned for five and then kind of brought back to 5.1.

00:26:32:10 - 00:26:56:11
Unknown
So they're trying to figure out the best way to do that. Google's like, no, we're about information. And that's how I use Gemini. I use this a lot for like the research stuff. I've yes, I find it's larger context makes it really good for that. And so I think that that plays to its strengths. And meanwhile, just by the way the meta stock keeps on going down.

00:26:56:14 - 00:27:25:11
Unknown
Interesting. So in the past month it's down 20%. Oh yeah. Whereas Google stock attempted goof stock, Goog stock, full Google stock is the past month is up 14%. Yeah. Was you know, there was a bit of a curve for down curve last few days for all tech stocks. But this this this really the market is impressed.

00:27:25:13 - 00:27:52:25
Unknown
Like I should be. Yeah. Well it's very it's a very visual approach. And so one of the examples that I wanted to try and you know was kind of trying to figure out like what's a good way to test this. And I haven't really worked a huge amount with canvas, but some of the changes that they made with canvas, I used a, a prompt, which I will fully admit, I helped me create, make a neo brutalist because I would never do brutalist persona.

00:27:52:25 - 00:28:14:16
Unknown
No, I make a neo brutalist personal site, make it chaotic, colorful, and use smooth scroll animations. Really push the limits. That was the prompt. That was it. And I put that into canvas and it came back with like, you know, I'm not I'm not a web designer. So things like this, I mean, some of the, the elements that are in interactive actually let me go full page on it.

00:28:14:18 - 00:28:35:22
Unknown
Yeah. This is so first started a web page and it's not just a static, you know, a couple of blue links on the page. It's not just images either. It's like these kind of animations. I certainly wouldn't know how to do this, but I'm not a web designer. Maybe this is easy stuff, but, you know, it did this in a matter of, I think, like a minute.

00:28:35:25 - 00:29:00:25
Unknown
It gave me this, this interactive web page. Not everything works like this. Little spin. Me, I can't, like, enter, you know, interact with that. Don't know how to go backwards. Once I got here, there's no way to click to go back or anything like that. But anyways, I find that pretty darn impressive. And, and kind of a, you know, almost certainly a sign of more things to come down that down that road.

00:29:00:25 - 00:29:22:13
Unknown
And definitely a good example of the visualization, the visual kind of, I don't know, is acuity the wrong word? Probably. But, capabilities of Gemini three, I've really got to I have not, I must confess that this is like. Hello folks. I have done my homework. I haven't played with vibe coding enough yet because it's really impressive. I just I just gave it a prompt.

00:29:22:13 - 00:29:41:29
Unknown
It's still it's still building it. And to see it build that code in front of you. Yeah. And realize that I'd be capable of doing it. This is a little weird, to the animation, over to the side. But, I mean, just the fact that, like, it's, you know, it's it's inverting the colors. It gives me this.

00:29:42:01 - 00:30:00:00
Unknown
I don't know, this. This is, it has a dynamic quality to it. I'm sure there's a ton of imperfection. If you got in the code, I'm sure there's plenty that people would look at it and be like, oh my God, this is garbage. But maybe not. I mean, it's rendering, you know, pretty reasonable as far as, like a good first stop, first step.

00:30:00:00 - 00:30:22:00
Unknown
I think that's pretty impressive. So I would I haven't done much of that either. I just asked it to make a site design appropriate for the 19th to 20th century typography that made a fake newspaper. That it's it's okay. It's not great. But yeah, you know, it did all that in no time flat. So yeah, pretty amazing stuff.

00:30:22:02 - 00:30:59:13
Unknown
So Gemini three out now, and I, you know, I know everybody's playing around with it to try and see, like, what is it really good for? I really want to test out the dynamic view aspect because, you know, some of the examples that I saw like give me a like a layout of a van goes art and you know, it had this, it just immediately spit out maybe not immediately, but after a couple of minutes, spit out this fully interactive way, which, you know, instead of just having a mountain of text, which we're really used to getting from Lem's right now, this is kind of taking that text and putting it into kind of

00:30:59:13 - 00:31:26:25
Unknown
a dynamic web page style interactive, presentation, let's say. And boy, if it's capable of this now, I'm, you know, once again, it's like, what's it going to be able to do in a year? You know, it's crazy, right? Really cool to see, how it's. Yeah, I think these things will conquer media world. They are world media world, that is to say the things that that that that mimic media.

00:31:26:27 - 00:31:48:05
Unknown
That we can see in those two dimensions. Yeah. Text, images, video. It's the third dimension where it gets in trouble. And that's what John's working on. Right. And that's what, what Jensen Huang is worried about. And that's where the real development will come. But as a media person, both of us media people, it's pretty amazing.

00:31:48:07 - 00:32:02:26
Unknown
You can do it. It's getting to the point where it can do pretty much anything in media. Now, that means a lot of what slop we got to deal with that issue. But as a technical ability, it's phenomenal. Before we get to the break, I just had to say I was out for a walk with my dog the other day.

00:32:02:26 - 00:32:23:08
Unknown
My dogs, and I was flipping through YouTube to try and find something. And, this video, this short comes up and it's like, make, make 100 Instagram Reels in one minute. And I was like, oh my God, okay, I had to see what the heck this thing is about. And sure enough, like the process, you know, it does that.

00:32:23:08 - 00:32:42:22
Unknown
It was it had to do with fitness and it was like going to ChatGPT and tell it, you know, come up with 100, 100 very, you know, popular fitness prompts or not prompts, but like motivational, you know, sentences or whatever, human prompts kind of yeah, yeah, human prompts sort of thing. So here's 100 motivational fitness related things.

00:32:42:24 - 00:33:07:05
Unknown
And then she takes that into like Canva. And, you know, with a through a CVS or whatever file and CVS, CSV, not the, not the drugstore and anyways, the process would totally work. You know, at the end of the day, in a, in a matter of a minute, she had like 100 Instagram Reels to share of this thing, but it was just like, oh my God, this is how this is how the internet's ruined.

00:33:07:07 - 00:33:39:22
Unknown
This is this is where it's going, isn't it? Yep. Just like churn it out. Churn, churn churn. Yeah. Anyways, interesting stuff. Let's, take a quick break to thank the patrons of this show. The supporters who support us on a deep level. Patreon.com slash I inside show. If you like the conversation that we have with ourselves and with other people like guess like yesterday, with Charles Le Mana from Microsoft, you can support us in doing all of that.

00:33:39:26 - 00:33:58:00
Unknown
Probably the best way to support us is to go to Patreon, Patreon.com, slash I inside show. Sign up for one of the tiers that comes directly to us to support. The production of the show allows us to continue doing what we are doing. And you know, every once in a while we like to throw out a little thank you to a few key people.

00:33:58:00 - 00:34:25:14
Unknown
This week, it's Tim Epperson. Cyprian, is it Cyprian? I've never known how to pronounce that name. Cyprian. Cyprian I hope I'm getting it right. Nicholas Gou and high five freak. So thank you for being amazing patrons for your support. And, yeah, we appreciate you going to take a quick break. Get on the other side of that and talk about more news, including Microsoft leaning even further into AI and windows coming up in a second.

00:34:26:25 - 00:34:52:10
Unknown
All right. Now it's time to get to the news that was pushed out of focus because of Gemini three yesterday. Not that I'm jaded or anything. We have at least one piece of the news from from build is just what we've we've been talking about it a little bit already, but that Microsoft is, you know, leaning further into a, the AI ification of windows.

00:34:52:12 - 00:35:13:25
Unknown
There's now going to be a new dedicated copilot taskbar for those of you using windows, a little taskbar that you can, you know, that serves the purpose of putting the AI front and center right in front of you so you can't ignore it. But that's going to be able enable you to summon different AI agents across all your apps, across, you know, the connected devices, that sort of stuff.

00:35:13:27 - 00:35:37:20
Unknown
So and you can imagine all, all different types of agents, right? Copilot specific troubleshooting, third party agents, all of them going to be kind of part of this, to do all the things that you might want to do from file search to document analysis, all that kind of stuff. And yes, you'll be able to say or maybe you already are able to say, hey, copilot and, it'll, it'll trigger it off if you prefer to use your voice.

00:35:37:23 - 00:36:02:09
Unknown
So AI layer is coming to the operating system, which. Yeah, not surprised about that. Yeah, I think it's going to it's in everything. And one more time back to on. He says it's already the case we're talking to. I just constantly during the day. That's true. And I think that it's the more it's apparent and transparent, the better.

00:36:02:12 - 00:36:27:00
Unknown
Yeah. Yeah, totally. Although yeah I mean yeah, that's a that's a really good point. I already layered in so many different places as we talk. Where does that compare throughout the world? Yeah, I wonder whether that sounds familiar. But there's such an emphasis on putting it into places now, whereas before maybe it just existed and we didn't emphasize the AI aspect of it.

00:36:27:02 - 00:36:45:22
Unknown
And, yeah, I think there's a little bit of yeah, it's a good point before it helps the company around. Right. It made translation immensely better. Yeah. And and if you asked oh it got better because of AI. But it wasn't, it wasn't kind of added in obviously. Right. Whereas now an operating system says here's where AI is going to help you.

00:36:45:26 - 00:37:12:10
Unknown
Oh home by by the way, AI is over here is over here. I yeah, that's all going to fade in the background. It's when I becomes accepted, except when it's coerced for letting people create slop. Yes. 100 videos in a minute. And they're also, by the way, leaning into further the use of voice as like a pre, like a premier, input method.

00:37:12:10 - 00:37:32:19
Unknown
So not just a keyboard, not just your mouse, but also your voice. I don't do that on an OS level, but I could see myself moving in that direction because I'm definitely using more of the AI models that I interact with through voice now than ever before. And so I could see myself kind of moving in that direction if I had it all set up.

00:37:32:22 - 00:37:52:09
Unknown
I reached out to shake Eleven's hand. In Brooklyn on Sunday, and she pulled back because she had a cast on. She broken her hand, and so I said, using voice more. She said, oh, yeah, it's amazing. One day I hurt my hand only for one day, and just to be able to have voice available was phenomenal.

00:37:52:11 - 00:38:11:23
Unknown
Yeah, totally. Click a button or just say, hey, whatever and do this. Start talking and. Yeah, yeah. And then of course, another big announcement, was agent 365, which, you know, we don't have to go entirely into detail because after all, we have a full episode on it. That said, why don't you summarize for this for our viewers?

00:38:11:29 - 00:38:47:11
Unknown
But, yeah, sure. You know, Charles Samana, who leads the effort, at Microsoft of Agent 365, came on to talk a little bit about what it is is basically it's a new control plane for managing agents across the enterprise. And I think the real kind of interesting aspect of this is that, you know, is the workplace moving in a direction where agents, actual AI agents are seen as kind of like another employee on your team's call or, you know, appears alongside human workers inside of Microsoft's apps.

00:38:47:18 - 00:39:10:09
Unknown
So you, as a human worker might see that there's a list of agents they're attending or editing or whatever the case may be all fully disclosed, but it's just yeah, it just kind of transforms what we think of when it comes to workplace collaboration, when, you know, this army of agents, army makes it seem like such a bad, you know, bad word.

00:39:10:09 - 00:39:37:08
Unknown
But better than Bob and Cordray. Cordray. Cordray. Is that a bad word? Depends on the context of whether you're in power or not. Yeah. Okay. All right. Anyways, those agents appearing right next to regular workers, how do workers feel about that? You know, because it's kind of like an in your face, I don't know, example of how this thing is doing the work that you're not doing, but maybe you're okay with that because it frees you up to do other work.

00:39:37:08 - 00:39:56:14
Unknown
I don't know, it's it's a complicated soup. So I it was a really interesting conversation. So I do urge you all to, to go back and watch the prior episode. And a couple things struck me about it. First is that agents will have identities. They'll have basically a, you know, an employee identity card. Yeah. And this is who they are, and this is their provenance.

00:39:56:14 - 00:40:22:28
Unknown
And this is what, authorities, they have and don't have from the security perspective. It enables a, security department to know what agents they have. And there's tens of thousands that they they're already finding that are at work in various of the companies they're working with and to, audit, their authority and limit that authority. So on the one hand, I thought, oh, this is the agent cops coming.

00:40:23:00 - 00:40:48:05
Unknown
But no, what what Charles said more than once is the reason they're doing this is to encourage the adoption of agents that they think this is a good thing, and it's not a scary thing. And, that by putting these controls on, people will have more confidence in them and know what to do. And then the other thing that struck me as very interesting is this is not just inside Microsoft.

00:40:48:07 - 00:41:18:04
Unknown
This is although they did mention they have 130,000 agents. Right. Internally running. Yeah. Right. So but this is for anybody anywhere. This is open and free. And it's intended to become they hope a, platform, a standard for identifying and, controlling the authority of and auditing the work of agents. And so what they hope is that others will use it, whatever platforms they use, because the agents are going to be operating out there on the internet and everywhere.

00:41:18:04 - 00:41:37:14
Unknown
So it was a really interesting conversation. And, and I'm glad, thanks for Microsoft for bringing it to us. Yeah. And I mean, you know, it also goes back to the fact that, like, these things are going to exist, whether Microsoft builds, you know, corrals these agents or not, people are still going to be utilizing them in these environments.

00:41:37:14 - 00:42:02:14
Unknown
And I think that was part of, you know, their impetus is like, would you rather have these agents in the, you know, in this environment in a controlled way, in a, in a way that gives you real clarity and understanding of what their roles are and what they're capable, what they are allowed to do. And if they do something that they're not allowed to do, you know, about it and everything, or that it's the wild, wild West and these things just exist and you never actually know that that's an agent on the other side.

00:42:02:14 - 00:42:25:01
Unknown
It could be a human, you know, all those things. So this is just a way of, of Microsoft is saying, hey, they're already here when you want to pretend like they aren't. They are. This is just a way to give you a little bit more clarity and power over that. So yeah. Interesting stuff. The future workplace is looking more and more interesting.

00:42:25:03 - 00:42:48:25
Unknown
You put in about another conference, a link about another conference, and I should have I should have told you not to read it. So I could have guessed. What your answer. Yes. If anybody in the chat live wants to tell us, go ahead, do the headline. I'm not. I'm not entirely sure I would have guessed correct, but a major a conference one startup got voted most likely to flop to everybody in the chat.

00:42:48:25 - 00:43:18:15
Unknown
Want to guess what that might be? Yeah, we'll give you five seconds. And then even more surprising is what came in number two. Yes. Totally. Totally. So this was at the Cerebral Valley AI, which is a great time for this conference to. Yes, totally, totally 300, ish founders and investors anonymously voted for their choice of least likely to succeed, essentially multibillion dollar AI startup.

00:43:18:18 - 00:43:44:14
Unknown
And if you guessed perplexity, you win the top prize of nothing. Perplexity essentially, I guess voted like. And would that just be because they're tapping into other models? Not not as much working on their own thing. You know, we want to we're we're reverse engineering this being this poll. Yeah. It's totally, totally. But I think it's a few factors.

00:43:44:14 - 00:44:06:01
Unknown
And I, and I used to have high praise for Leslie here on the show because they were there were doing creative things and they were staying ahead, but they were doing it through a certain amount of moxie, more than fundamental. I work number one, and number two, I think that they've been exposed for being more PR than reality.

00:44:06:03 - 00:44:23:18
Unknown
And I think it's a shame because they've done really interesting things where they came out with the first real AI browser. They've done other things like that, but they, they pretty much kluge it together to get out there and get the press release. So my guess is that that's a lot of it and that there isn't much business there.

00:44:23:24 - 00:44:48:13
Unknown
But number two in the like the 12 list ready for drumroll OpenAI and and OpenAI. Yeah. Very scary. Yeah. Right. Last week we last week, a week before last, we talked about an open AI bubble. Yep. Which I think is the case. There was a story that I put up at the last minute, if I can find it.

00:44:48:15 - 00:45:15:08
Unknown
No, I can't. Somebody else said, oh. It was hugging face. So. Yeah, I've got that in here. Oh. Is in there, right. Is is that an ad lib bubble? Yeah, I would echo what like I was saying earlier, I was going to say that that speaks to me. That's how I does feel. So. And if you believe that then open AI is obviously the most extended and the most, identified with pure lives.

00:45:15:08 - 00:45:35:12
Unknown
Anthropic two. Certainly. But but it has more revenue that open AI has an open AI has more obligations. And it has more, venture money that it's got to justify and open eyes on it, on a high wire act, itself. So there's a lot of almost wishful thinking right now for the for AI bubble to burst.

00:45:35:12 - 00:45:51:11
Unknown
And it's not good for anybody. I mean, there's a lot yeah, there's a lot of attention and energy going in that direction right now. I feel like, yeah, you know, even people who aren't like, who I end up in conversations with and, you know, the topic of AI comes up. That's one of the first questions they ask me.

00:45:51:11 - 00:46:17:03
Unknown
You know, they don't work in these in this industry that you and I do. And, you know, live and breathe this stuff. And yet that's the first thing they think. Are we in the middle of it. Is there an AI bubble. It's going to pop that. How's it going to affect me? Which is a proper question. And I hope there is it because if there is, we got trouble, and even the, so as Sundar Pichai.

00:46:17:04 - 00:46:36:13
Unknown
Thank you. Yeah. Warned that no company is going to be immune if an AI bubble bursts. Right. The tentacles out there are all over in terms of we talked about this last week in terms of who owes money and has business with whom. So I think it's bad to be wishing for it. I think there's a little bit of a comeuppance.

00:46:36:15 - 00:47:03:12
Unknown
A little reality check is a fine thing. Deflate it a bit is a fine thing. Raise the expectations for actual revenue and real business models. That's okay. Get more competition and models. That's okay. All that's fine. But to just say they hope the whole thing collapses. Could have disastrous effect. And I don't think it's I don't think it needs to I don't I'm not one of those who believes that this is not NFTs.

00:47:03:12 - 00:47:32:03
Unknown
As we've said in the show before, it's not Bitcoin. There is enough reality here and enough benefit here. It's not as much as people saying but and again like that that really speaks to me. Back to what Huggingface CEO Clem DeLong said, which, you know, really what I really connect with in what he says and that that perspective is that AI as a technology is not going anywhere.

00:47:32:03 - 00:47:53:25
Unknown
Like it's not like a bug, a bubble pops and suddenly AI goes away. We've talked about that before. The, the kind of massive energy and attention on Lem might tamper down, might, you know, minimize, let's say. But I as a technology, it's not like the bubble pops and suddenly. Well, I guess that was the AI phase.

00:47:53:25 - 00:48:18:29
Unknown
Okay, AI's done. What next? You know, like, right there, there is so much already there, you know? And so then when you look at a company like OpenAI, like do, do you believe that OpenAI could fail so spectacularly that they go away or. I don't know that I believe that would be the case. Like they they might have a real big moment and it might be very impactful.

00:48:18:29 - 00:48:39:13
Unknown
But I couldn't see suddenly OpenAI going AOL or whatever. Yeah, let me go ask AltaVista what they think is your, that's true. I mean, you're right. There's a lot of examples to, to, to point to, I suppose, in that regard. I think the problem is they put every egg in that and that whole basket in that scale basket.

00:48:39:15 - 00:49:00:14
Unknown
Yeah, that's an issue. So that is you know, there goes back. So so Herbert Simon I have I have a couple of books about him here. Subir is, is considered one of the inventors of the notion of AI back in 1956. It's been around a long time. It's gone through lots of iterations. So no, he has not gone away, I agree.

00:49:00:20 - 00:49:29:06
Unknown
Yeah, no. And the the ad industry is benefiting a lot from AI. So I'm sure the ad industry, you know, stands to benefit a lot from AI. I'm sure they don't want it to go anywhere. Meta. Google and Amazon all launched pretty major AI platforms. A lot of, are I ad platforms. In other words, which, you know, I guess there's a lot of similarities between these meta's generative ads model.

00:49:29:08 - 00:49:55:04
Unknown
GM not to be confused with Google's Gems or gems, whatever it was here, I let's see here. Where is this? This is meta first. There we go. Driving ad recommendations on Instagram of Facebook, personalized, efficient targeting. A lot of the goals of these are going to be very, very similar. This models complex user ad interactions. So, you know it it knows what you want before maybe you do that.

00:49:55:05 - 00:50:15:18
Unknown
That's that's the goal in advertising. Right. So it's all the all all of this is a genetic, what Neta says is they created a new foundation model for advertisers, but it's pretty much the same thing. Amazon has a new, set of agents that create and optimize campaigns at scale. The target campaigns accelerate time to insight in Amazon Marketing Cloud.

00:50:15:18 - 00:50:48:23
Unknown
A lot of a lot of marketing speak. But they're trying to do that. And then, Google Google's has as AI advisors a genetic tools to drive impact and insights. Your ads advisor, your partner for AI optimization and action maximize performance with minimal effort. I mean, it's so funny to see advertising language aimed at advertisers. I guess they speak their language, create new campaigns, assets and keywords, take advantage of personalized performance analysis and so on and so forth.

00:50:48:25 - 00:51:19:06
Unknown
So, yeah, we've I've said this 100 times on the show that we're going to see the most uptake in, in generative AI first from advertisers. We see it in, the creative side, we see it in the media side and the analytics side of the business. And then beyond that, I talked about a study some weeks ago where it's evident to me that while media companies are trying to block the door to AI, brands obviously are not.

00:51:19:08 - 00:51:39:29
Unknown
Brands are saying we want to be there. We want to make it as easy as we can for for these models to find us and mention us when you ask for the best bike rack. And so I think we're going to find as media and media brands are going to be left way behind here, because advertisers are going to say, well, you guys are nowhere with AI.

00:51:39:29 - 00:52:00:06
Unknown
You're you're acting as if it's the enemy and you're the victim. We want to be there and you're not helping us get there, so bye bye. And even more so, I think the media would be very wise to look at what advertisers are doing, because this is where the business is going to go. And meanwhile, Fuji CMO gave an interview to I think wired this week.

00:52:00:09 - 00:52:21:23
Unknown
And because of her experience at meta and at Instacart, she's going to figure out how to bring not only paid products to OpenAI. Apps but also advertise it. And so you're going to see the AI companies are going to start competing more and more with media for ad dollars, because they're going to have a friendly relationship.

00:52:21:26 - 00:52:42:21
Unknown
So it's going to be very interesting to keep watching the relationship of advertising and AI at about a dozen different fronts. Yes. Yeah. And all the major companies have a lot invested in in that aspect too, and a lot to gain from it. So yes, indeed. All right. Real quick, if you are enjoying the show, we would appreciate if you would leave a review.

00:52:42:21 - 00:53:05:14
Unknown
It's a small thing that you can do that makes a big difference for getting the word out on the, kind of awareness around AI inside. So go to Apple Podcasts, leave a review. It even appears on the AI inside website AI inside Dot show. I've got an area there that features your reviews so you know you'll be AI inside famous if you do that.

00:53:05:21 - 00:53:17:19
Unknown
And and us grateful. Absolutely grateful. If you would do that, it would mean a lot. All right. Super quick break that we got a little speed round and we'll get you out of here one moment.

00:53:18:16 - 00:53:43:20
Unknown
Weather fans unite. Google's DeepMind. Whether next to it's an advanced AI model that is for global weather forecasting. Weather forecasting, you have to imagine, is a pretty complicated thing. There's a lot that we know about weather patterns and everything, but there's also a lot of unpredictability. And this model can predict hundreds of possible outcomes. I don't know how that compares to previous models.

00:53:43:25 - 00:54:06:22
Unknown
I imagine it could also. But they say it dramatically outperforms current models in speed and accuracy, and it can also handle joint weather events. And that's where things get really unpredictable, because like, you might have a single weather event that's easier to predict, but when they combine with other weather events, you know, it becomes a lot more difficult to know exactly what might happen or to predict what might happen.

00:54:06:22 - 00:54:33:10
Unknown
And these systems are better, more resilient, more effective for all that stuff. So what struck me here is this is this is Mother Nature's digital twin. This is in the thread that Google DeepMind put out. They explore, as Jason just said, hundreds of possibilities in less than a minute from a single starting point. You know, that one butterfly wiggle flapping, and it starts from there and says, okay, what are all the various futures that could occur?

00:54:33:12 - 00:54:59:02
Unknown
And they said this would require hours on a supercomputer using physics based models. Wow, wow. So it's a new approach called Functional generative Network, which generates the full range of possible forecasts in a single step. And then they add targeted randomness. This is where it gets beyond me. I don't understand what that does into the architecture, allowing you to explore a wide range of sensible weather scenarios.

00:54:59:04 - 00:55:13:16
Unknown
So this is I think it's going to save lives. It's going to make a difference to people, and it's also going to make a lot of convenience. Do you take an umbrella today or not? When is it going to start raining? Exactly. But it really is, an impressive use of this kind of technology. Yeah.

00:55:13:17 - 00:55:34:15
Unknown
I mean, time and time again and I think maybe this comes up in a few minutes, but, the specialization of these systems versus the just the way to go best, which is the way to go, that's where the really big, amazing advancements come from. I get the draw of generalizability in the one hand is the AGI, IBS, but on the other hand it's say we created a machine.

00:55:34:15 - 00:55:59:25
Unknown
This is what Adam Brown said at the event on Sunday. Once we create a machine that can program itself, that's it. Yeah, right. That it can it can generalize its ability to any task that and that's why generalizability is such a goal for these folks. But it doesn't all have to come from that source. You can create an amazing weather machine and wow us and have huge value and take the win, take the sample totally.

00:55:59:25 - 00:56:30:13
Unknown
And you can do that now versus somewhere down the line when AGI might possibly happen. Yes. Yep OpenAI announced new group chats coming to ChatGPT and more collaborative I guess environment. It's first kind of in the midst of a pilot happening with, subset of users. But basically this is a group chat experience inside of ChatGPT. I don't know how it'll be presented, at least inside of the app with a chat GPT bot on standby when you want to bring it in.

00:56:30:13 - 00:56:47:04
Unknown
When I first heard of this, I was like, oh my God, I like what I envisioned was a bunch of people in a single query window like, oh, what about this? What about this? And just the noise that that would create. But this is really just like personal chat with the ability to ever once in a while ago.

00:56:47:04 - 00:57:10:16
Unknown
Well, oh, that's a great idea for dinner. Hey, ChatGPT, tell us what you know. This thing you and I are disagreeing about something, and, you know, you can say who's right. Tell Jeff why he's wrong. Yes. Exactly. I guess it makes sense because we're both lone wolf operators. Yeah. We don't operate in slack anymore. No. True.

00:57:10:18 - 00:57:32:13
Unknown
And if we did operate in slack, we would have access to AI in slack chats. Google has AI in in. I think Google Chat. You know what I mean that. So ChatGPT is is following up. Yeah. Yeah. Following is as far as this is good. But I haven't heard from anybody I'd love to know from folks out there who who are in are social beings in real companies who are using slack and and Google and such.

00:57:32:20 - 00:57:55:20
Unknown
Whether you all find that useful you that's that's a big question. I have to how often does it. Hey, we could just ask the AI right here and it can give us the answer or the yeah, mistake, whichever the case may be. Right. It reminds me of when I was on the city desk at the midnight shift in the Chicago Tribune, and drugs would call for bars, settling bets and asking, the city desk.

00:57:55:22 - 00:58:13:26
Unknown
And my editor, Bill was his name said, always give him an answer. Preferably the wrong one, because he envisioned them getting into a fist fight. And the next morning when they went to the library where they find out that the Tribune was wrong. Yeah. They're like, that was a horrible idea. Yeah, but I can imagine the same thing happened with ChatGPT.

00:58:14:00 - 00:58:39:02
Unknown
Well, it's judgment. Is that one who trust that I don't know, yeah. Will it really settle arguments or will create more arguments? I don't know, right. Oh, great. You just invited the slop into the chat. Why do you do that? You know. Yeah, yeah. Microsoft partnered with anthropic to bring cloud AI to Azure. Anthropic is committing to $30 billion in Azure compute up to one gigawatt capacity.

00:58:39:02 - 00:58:57:28
Unknown
Nvidia is on board. They're throwing in 10 billion. Microsoft's invested 5 billion. And yeah, I think a lot of this is just about, anthropic kind of shoring up its own enterprise offerings. There's a there's a large industry reliance, as we talked about earlier on open AI products. And so a bubble there would not be very good.

00:58:58:00 - 00:59:13:02
Unknown
But, I think anthropic wants a piece of that as well. So I think I need to get a new filing cabinet to file away all these circular deals. Oh, I mean, it's so confused. Just they're never ending. It's hard to keep it all straight. We give you money to give us money to give you some stuff, to give us some stuff.

00:59:13:02 - 00:59:34:27
Unknown
And where the real value is created. We'll find out in five years. Yeah. Or we'll find out when. When. Suppose a bubble does pop and all and all of those complicated stacking of cards kind of fall. Yeah. Amazon and Microsoft who appear to be very popular today. We've talked about them a lot, supporting legislation in Washington DC to restrict.

00:59:34:27 - 00:59:54:25
Unknown
Oops, sorry, that's the wrong article. Restrict Nvidia's AI chip exports to China. Both companies want easier access to those advanced chips for their own data centers. Of course, Nvidia on the flip side, would love to keep things open to the Chinese market. Of course, you know, it's it's like when Microsoft goes after Google and I get it, they're companies.

00:59:54:25 - 01:00:15:02
Unknown
They operate in their self-interest. But at some point is their honor among thieves and AI people? What's best for Nvidia in some ways is better for them, too, because it provides free rates in the long run. It creates more capacity. And but they also don't what the hidden agenda here, it's not just they want the chips, they don't want China to have the chips because China is going to compete with them.

01:00:15:04 - 01:00:37:10
Unknown
Yeah that's true. And that's and so they're trying to block progress in the overall AI ecosystem for their self-interest. Again I get it. They're companies. That's what they do. But I think here I'm on Team Jensen. Yeah yeah. And by the way Nvidia earnings I think happening. They come out today after the close. Yes. After the close close hanging on their fingernails to here.

01:00:37:11 - 01:01:01:05
Unknown
Yes I know it was a lot of a lot of hype around it. So I'm sure we'll talk about that next week. And then finally Jeff Bezos. Amazon. Well once once Amazon CEO I guess he's not he's he's Hubert chair and yachtsman. Right. He has a new AI startup called Project Prometheus. What a what a name.

01:01:01:08 - 01:01:31:04
Unknown
He's co-chief along with Vik. But Bajaj Bajaj, who's a former, Google X scientist, $6.2 billion in funding, one of the best funded early stage tech companies. I wonder why Bezos, uses AI for engineering and manufacturing. So things like cars, computers, spacecraft, spacecraft, of course, because Bezos is a space nut and yeah, are you interested to see what they come up with, like with these kinds of resources?

01:01:31:04 - 01:01:54:14
Unknown
I like what do we expect to see? Yeah, I don't think it's going to be stuff that we're going to see. So for automobiles I don't know. We're going to see it on a consumer level. Okay. I think it'll be at the high end level. Yeah. The investment Prometheus by the way was the, according to Wikipedia, the Olympian god that took fire from the gods and brought it to humanity in the form of technology, knowledge and civilization.

01:01:54:16 - 01:02:18:08
Unknown
Okay. The Jeff Bezos. Yeah, sure. Prometheus. Sure, sure. Whatever. Bezos. We've reached the end. Jeff Jarvis, thank you so much for being with me and talking about all these wonderful, interesting, insightful topics. Go to Jeff jarvis.com to find all of Jeff's written work of, you know, not even all of your your books are here, just only four of them.

01:02:18:11 - 01:02:40:13
Unknown
But the web we weave magazine. There's three others professor and hot type. Yeah. Your your pro. You know what you're doing when it comes to right next one. Maybe the hot typos in the copy editor's hands as of today. Now. All right. Now I'm on twitching because he found another stupid mistake and another stupid way to do a footnote and all kinds of harassing things.

01:02:40:13 - 01:03:04:06
Unknown
Copy editor. Save us from. Yes. Oh, did you really mean this? Thank you. Copy editors though. Yeah, they've. It's a luxury to have them still. Absolutely. Absolutely Jeff Jarvis Scott. Com for all your work I inside show for all of our work with this show. Everything that you need to know including our most recent episode, the interview with Charles Amana from Microsoft can be found there.

01:03:04:08 - 01:03:29:07
Unknown
And then finally, I inside on Patreon. Patreon.com slash AI inside show. If you want to support us on a deeper level, that's the place to do it and we appreciate it. We also have some amazing executive producers who continue to support us on a deep, deep level. Doctor Du Jeffrey Berrettini, Radio Asheville one of 3.7 Dante Saint, James Barnard, Eric Jason Knight for Jason Brady, Anthony Downs and Marc Starker.

01:03:29:09 - 01:03:52:02
Unknown
Y'all are awesome. Thank you, thank you, thank you and thank you for supporting the show like that. All right. I think we've reached the end. Who knows, maybe next week we'll be surprised by Gemini for I don't was made by Gemini itself. Yes. Right. Oh, like, could you imagine? That would be crazy. That's the announcement that. Yeah, it's a big announcement.

01:03:52:03 - 01:04:06:29
Unknown
No human being made. This model. Oh, geez. I'm not sure how I feel about that potential future, but if it's happens, you know, we'll talk about it next week on the AI inside podcast. Take care everybody. Have a wonderful week. We'll, see you soon.