The Proactive AI Agent Race is On

Jason Howell
00:00
Coming up next, Jeff Jarvis and I uh talk a little bit about Google Book because a couple of weeks ago Jeff went to an event. He couldn't talk about it last week, and he's got thoughts, and I've got thoughts. A lot of talk about Google Book right at the top.

00:12
Then we talk about NVIDIA creating a hardware layer for keeping agents from breaking bad I think it's a pretty interesting solution. OpenAI, Meta, and Manus all suddenly diving deep into personal proactive That's the keyword. Agents that are also cute and fuzzy.

00:29
And hey, I watched the AI doc on Netflix uh while I was heading back from Maui last week. And I didn't jump off the plane while I was watching it. So, you know, success.

00:40
That's all coming up next on AI Inside. What's going on, everybody? Welcome to another episode of AI Inside, the show where we take a look at the AI that is layered.

01:01
Throughout the world of technology. I'm one of your hosts, Jason Howell, joined again by Jeff Jarvis.

Jeff Jarvis
01:08
We're back. We're reuniting.

Jason Howell
01:09
Hey, we're back. We got our thing going. Last different

Jeff Jarvis
01:14
works so hard he has no tan.

Jason Howell
01:18
Not true. I do have a tan, but it is a definite farmer's tan. I'm like I got the stark white, you know, short sleeve t-shirt thing.

01:27
Yeah, Maui was great. It was a ton of fun. I'm so happy that I got to go to the event, the Snapdragon Summit event.

01:34
You know, be in Hawaii for a break. It was a much needed break for for me and for my family and I mean if you gotta work somewhere, you might as well work in Maui

Jeff Jarvis
01:43
That's what uh uh a few tech executives have to say, yes.

Jason Howell
01:49
It's not quite a bunker, but it was uh a nice speech, at least. Um How you've been? You had a thing last week, right?

Jeff Jarvis
01:57
I did, yeah. I went to the uh Google Book event.

Jason Howell
02:02
So that was like two weeks ago.

Jeff Jarvis
02:04
Yes.

Jason Howell
02:04
Yeah. And then last week you had something going on. Right?

02:10
Maybe not. Maybe I'm wrong. I'm probably wrong.

02:13
I thought I thought there was something. Don't worry about it. I'm probably I'm probably fully backwards.

02:22
At this point.

Jeff Jarvis
02:23
Oh yeah, I was in San Francisco.

Jason Howell
02:25
Oh.

Jeff Jarvis
02:25
Right. Okay. Right.

02:26
Right. So that was interesting. I I d how soon we forget.

02:30
So handshake, which um Basically helps students get jobs when they graduate. When I was at CUNY, all of our students at all the jobs had to go through handshake. Uh they came up with a new product uh which enables students graduating and job seekers to use AI to create things, to create a portfolio of that to show what they can do to employers.

Jason Howell
02:55
So they're using AI to build a portfolio or they're using the things that they're using outside of AI.

Jeff Jarvis
03:03
They're using AI to build new products, new services, new whatever.

Jason Howell
03:08
And those are their performances.

Jeff Jarvis
03:10
Look what I can do with that. Got it. Which is very small.

Jason Howell
03:12
I see. Right. Well, that's cool.

Jeff Jarvis
03:14
Yeah. So they can show themselves. So I I I moderated a panel about um education.

03:21
And it was really interesting. One person from Deep Mind, one from another one from OpenAI who had really interesting statistics, by the way. Just I'll throw this one thing out.

03:29
He said that um in analyzing the use of uh ChatGPT in companies Uh only I forget what it was 40% one way or 60% the other way, but almost half, one way or the other, of the queries are made outside of an employee's silo. That is to say, if you're in product, you use it to find out about marketing or about HR or about other issues. Which really speaks to the desire of people to break out of their silos.

03:59
It it's a good thing for companies because it says that they want to be involved and learn about these other things in the company. It speaks to I think companies reorganizing themselves in this new way about more teamwork. In that sense.

04:12
So it was that kind of stuff. It was really, really interesting to see what other people had to say. So that yeah, that was last week.

04:17
I also visited the Computer History Museum

Jason Howell
04:20
I love that place.

Jeff Jarvis
04:21
Which is fine.

Jason Howell
04:21
That's so cool. Uh was there an event? An event

Jeff Jarvis
04:24
No, I wanted to talk to them about some events that I might help do in the future.

Jason Howell
04:27
Yeah.

Jeff Jarvis
04:28
We're we're plotting. Uh you know, whether if I can raise the money for do some things, then uh that would be a neat place to do it.

Jason Howell
04:34
Yeah.

Jeff Jarvis
04:34
Um and then I came back real quickly. Yeah.

Jason Howell
04:37
Right on.

Jeff Jarvis
04:38
So then before that I uh went to the uh Google Book event.

Jason Howell
04:41
Yeah.

Jeff Jarvis
04:42
And it was uh embargoed until last Monday. Um The hardware all looks really good. Uh sleek and uh high end.

04:51
It's $8. 99 to $12. 99 I think are the prices of them.

04:54
Uh the stats. Uh I mean the the specs are fairly similar across them all. Uh no media tech chip yet.

05:02
So the the the Chromebook that I use um is a media tech chip and it means it's fanless. Which I really like. All of them were fanned.

05:12
Uh there was a report that Lenovo was going to have a separable uh keyboard and tablet. That didn't come out because I think the r the report is that Google wanted every manufacturer, the five manufacturers to have one flagship machine.

Jason Howell
05:26
Right. Right.

Jeff Jarvis
05:28
It adds because it can do AI stuff, the the Wigley cursor allows you to bring a Gemini into most anything on your screen. Uh Rambler will let you uh get rid of an awe like I just had and other silly things uh that we cite.

Jason Howell
05:42
I use Rambler all the time. At least on my pixel, yeah. Um

Jeff Jarvis
05:45
articulate uh and other things. Um And so I was impressed all in all. I haven't ordered one yet, and I was debating two, and then you in Hawaii happened to speak. To the head of the Chrome OS, yes?

Jason Howell
06:01
Yeah, John Solomon. Yeah. Well that was on last week's episode.

06:04
Uh that was uh part of last week's show. Yeah, which by the way was a very Spur of the moment thing. Was it?

06:11
I was basically in the in the auditorium, you know, watching the presentation. Then he got up there and I immediately reached out to my Google contact and I was like, hey. I know you're here and John's here.

06:24
Is there any chance I can just like get like 15 minutes with him and just like talk about whatever we talk about? And so And so I I get the yes with like ten minutes to prepare for it. So I'm like, oh well now what do I want to ask him?

06:38
So you know I sit down and Throw together an interview for the most part. So it probably wasn't my most prepared interview ever.

Jeff Jarvis
06:45
Well it was very informative to me

Jason Howell
06:47
Okay, great. Because

Jeff Jarvis
06:48
the very first question you asked him, Jason, why don't you t tell that?

Jason Howell
06:53
Well, of course. I mean it the the history of Jeff Jarvis and Podcast and Google is that Uh, Jeff you are so in ingrained in the uh enterprise side of Google accounts, and so many of their features just don't work with their enterprise accounts. For some mysterious reason, that continues to be the case. And apparently that continues to be the case with Google Book, at least for now.

Jeff Jarvis
07:18
Yeah, so it was very frustrating. He basically said, this is a consumer launch, uh, we'll get to the business people later. Uh without any details, which means that uh my fear is I'm gonna buy a thousand dollar machine and then various of these functions I won't be able to get.

07:30
And and other things like the benefits of a year-free of storage and AI and all that, I won't get either because they only give those things to consumers. This is what happened with my last Chromebook as well. But the other interesting news that came out uh in the meantime is that Uh Google has said that they're going to uh continue to support the Chrome OS as it was only for I think in 20 until 2034.

08:01
So updates will end there for Chrome, which means they do intend to move everyone over to the Google Book OS.

Jason Howell
08:08
Yeah, we weren't entirely certain whether they were gonna have two different laptop things. And I mean I think my my hunch, my expectation was that they were eventually gonna phase out Chrome OS at some point. point and replace it.

08:21
That's certainly been the rumor well before we even knew that Google Book was a thing. That was the rumor that Google was going to do something new and that Chrome OS was going to go away. This kind of confirms it.

08:33
I'm kind of surprised at how little people have reacted to this. Maybe it's because people just assumed this was the case, but Um yeah, this is more or less this is confirmation that like, yeah, Chrome OS's days are numbered, uh, Google book uh to the moon.

Jeff Jarvis
08:50
Right. And um, you know, if you're if you're a company or a school uh enjoying being able to buy $300 laptops, um when and how can Google Book OS move to a cheaper machine is going to be a big question.

Jason Howell
09:04
That's a huge question for me. Yeah.

Jeff Jarvis
09:06
But the other thing about this that's interesting is that at the presentation in New York, Chrome OS hardly came up at all. There was hardly any mention of Chromebooks. There was hardly any mention of bringing people over from Chromebooks and and exciting them.

09:18
This was an Android presentation. This is that your phone and your laptop will now be enmeshed when you stop a function on your phone mid-mid process. You can pick it up right in your laptop.

09:30
That is very familiar. They added things that Android needed. They added a decent file structure, which Android does not have.

09:37
Would you agree with that?

Jason Howell
09:38
Yep.

Jeff Jarvis
09:39
I think a file structure really sucks on Android. Um and other things that they added and this is an opportunity for them to run apps. So I think you get some Adobe things free for three months.

09:52
with these five machines, which shows that you really couldn't run that any of that obviously. But you could only run it through the web on a Chromebook. So now it's a machine that will run apps.

10:02
Now I don't know what that means in terms of things like security, but but Android is secure. I feel fairly safe with Android.

Jason Howell
10:08
Yeah. Yeah, I do too.

Jeff Jarvis
10:10
So I think it's basically Android is ported over to a laptop OS is what happened.

Jason Howell
10:15
Yeah. Well, and I wonder out of that, okay, great, there's all these apps. How well do all these apps work in that particular

Jeff Jarvis
10:23
Exactly?

Jason Howell
10:24
Way does Google incentivize developers to make them uh want to make the necessary changes to their apps so that they can work really well with this? Which is, you know, this is a whole new avenue, essentially a whole new surface for developers to have to be um be aware of and con and concerned with uh for their apps. And You know, I'm sure met I'm sure plenty of them will be uh interested in in making those changes, but it all hinges on how successful is this as a platform?

10:55
And that's a real big risk that Google is taking. to say Chrome OS might not have taken the world by storm in all facets of computing, but it certainly has its markets where it has been incredibly successful in be in adoption. um school education being one of those key ones that comes to my mind as well.

11:15
And I'm really curious to see, like, oh, does it, you know How does Google do that? There's there's a lot of benefit to the Chrome OS being as limited as it is for the education world, you know?

Jeff Jarvis
11:27
Yes.

Jason Howell
11:27
And uh once you introduce a computer that does a whole lot more, are you kind of uh yeah, are you are you bringing are you replacing a product that certain schools are are happy with that give them exactly what they need, limits and all? um with something that kind of introduces more that now they have to contend with that and and ask themselves like, well maybe this isn't the solution for us anymore.

Jeff Jarvis
11:50
Right. And um You you make the point. They've got they've got to build in that fleet functionality that that the school administrator or the company administrator can control things.

Jason Howell
12:00
And I'm sure that they will.

Jeff Jarvis
12:01
They will, but that's gonna complicate the hell out of it. You know, that's it's a bane in my existence when it's something new comes out is to have to give myself permission to be myself Yeah. The other thing about this is that Google's been here before. When Chrome came out, you could run Android apps. And then they quietly just stopped that.

Jason Howell
12:19
Yeah.

Jeff Jarvis
12:20
So I can't run Android apps anymore. Plus web apps were big, but that's kind of gone away. Extensions are there, but I don't think they they get a lot of attention. So I think Google has to prove itself that this can become an app environment where Chrome was not.

Jason Howell
12:38
Yeah. Yeah, it'll be it'll be interesting. I mean and And undoubtedly there are going to be a whole slew of apps that never quite quite work right.

Jeff Jarvis
12:49
Right.

Jason Howell
12:51
It'll be kind of a broken experience to a certain degree.

Jeff Jarvis
12:53
If I want to click on a podcast link, I end up at uh Apple Podcasts as a little tiny box that that that has making me sign in and I can't I can't get to the web from there. It's difficult on Chrome.

Jason Howell
13:04
Yeah, yeah. Um, okay. Well, yeah, I mean this interesting.

13:09
I uh you know, I had very brief interaction with Google Book uh in Maui. One of the things that I was able to do is I had another sit-down with that project Aura uh Android XR um experience, my third at this point and it was like a 40 minute sit-down experience and I wouldn't say there was anything that real really new that came out of that experience. It's kind of a known quantity for me at this point.

13:31
And I think for most people, we're now we're just kind of waiting for it to release. But I did get a separate demo of plugging the aura into a Google book to have a little bit of that kind of casting and interactivity functionality. And so that was That was kind of neat.

13:46
Um, you know, I I think you you'll be able to do that with any computer and um but I'm sure there will be some extra hooks in you know that Google

Jeff Jarvis
13:55
Is there any time frame on release for Aura? Uh

Jason Howell
13:58
no official time frame. My hunch is that we're getting it sometime before the end of this year. But there has been no official announced like it's it's happening this time.

14:08
Just Just little winks, let's say. So I I have a good feeling that we're gonna get it by the end of the year, but that could change. Um but I hope that it comes out by the end of the year.

14:18
It kind of the longer you wait on this stuff, you know, the longer you give others a chance to kind of catch up. It's kind of like Google's AI glasses, you know, the gentle monster and and and those things. Like, man, we've known about those things for three years now.

14:30
We still don't have them. And the the the landscape has changed a lot in that time. So come on, Google.

14:37
Make it happen.

Jeff Jarvis
14:38
Yep.

Jason Howell
14:39
Yeah. Um Cool. Well that's that's Google Book. I do hope I'm I'm hoping to get one of these for review um in the near future. And if so, I'll definitely be talking about it on this show and talking about it on Android Faithful.

Jeff Jarvis
14:51
I would I would kill yeah well reminds me question how mu how much interest was there in your in your colleagues on Android Faithful?

Jason Howell
15:00
Of Google Book?

Jeff Jarvis
15:01
Yeah, Google Book.

Jason Howell
15:03
I mean I think I think we're all very curious about it. I think there's there's also a little bit of hesitation, you know, it's it's which is similar hesitation around Chrome OS, but uh around this as well, which is like okay. Okay, cool.

15:15
Like it's it's great that Google will have its own computing platform, but like who exactly is it for? People mostly already have their computing platform and so bringing a new kind of computing operating system onto the market, like how do you find the users for that? You know, I don't think any of us are like, oh my goodness, this solves my problem.

15:40
This is exactly what I was looking for. And that's that's a little concerning to me. But Um but I appreciate that Google's doing it and it makes sense too.

15:48
They're Gemini if they're Gemini e fying or whatever you want to say everything in their own.

Jeff Jarvis
15:53
That's the interesting thing to me. Right. Is that is that where does AI come in uh in a useful way in your laptop, in Aura, in in other hardware. Um

Jason Howell
16:05
Yeah.

Jeff Jarvis
16:06
And you know what really we'll we'll get to it later, but but I was really surprised in a way that that Meta beat Google to the punch with Muse and now OpenAI with Dots. And uh people are cautious about giving those things uh access. And I already have this relationship with Google in all kinds of ways. I'm surprised they didn't do uh that kind of agent, uh retail agent.

Jason Howell
16:29
You know, Google again, and I definitely heard this while I was in Maui, is just a general kind of perspective that Google itself is lagging once again when it comes to AI is kind of in the laggard category. Oh, it's total leapfrog. Yeah.

16:45
And that's that's only gonna matter for so long, and then Google's gonna do something and they'll probably be in there. But but it that was interesting. I heard that from a lot of different people.

16:53
Uh Definitely the general perspective is uh Google's behind again. Like, and that's surprising to a lot of people. Um, hey Lee, thank you for the super chat.

17:03
says it would be cool if y'all could book Gil Duran, the author of The Nerd Right for a show.

Jeff Jarvis
17:10
Really interesting book. I just downloaded the book. I haven't I haven't read it yet. Or I'm gonna read I'm gonna listen to it.

Jason Howell
17:15
Oh, interesting. Yeah, once you do.

Jeff Jarvis
17:19
The Nerdreich is uh arguing that we kind of have a um uh government of the nerds now.

Jason Howell
17:26
Oh yeah. Okay.

Jeff Jarvis
17:31
So yeah, so it's a it's it's gotten very good reviews. Um there's a few books like this out these days, but I think this is the one that's gotten the the most attention.

Jason Howell
17:39
Hmm. Well, yeah, super time.

Jeff Jarvis
17:41
We can give it a try, Lee. Thank you.

Jason Howell
17:42
Yeah, Lee says it's really good. So there you go. Um, real quick before we move off of this, I just want to check in because uh on the on this one Pretty unrelated thing, but from an earlier episode, the AI doc, which is now on Netflix.

17:57
Netflix has licensed it, so you can find the AI doc on on Netflix. And I had told you before my trip I was gonna download it and watch it on the plane. And you were like, don't jump off, you know.

18:08
Don't sit in the exit

Jeff Jarvis
18:09
row

Jason Howell
18:10
Don't jump out of the plane. Um, I did not jump out of the plane when I watched it, by the way. So at least there's

Jeff Jarvis
18:17
a what did you think?

Jason Howell
18:19
You know, okay. I thought it was I thought it was entertaining. Um, certainly there were one thing I noticed is there were definitely moments when I was watching it where I felt my own internal anxiety rise.

18:32
And I was like, okay. they really are were good at um kind of the creating a production, creating an environment in the show that is meant, I think, to pull that emotion out of the viewers. That's A large part of its perspective is creating anxiety around those.

18:51
Yeah, yeah. Um Clearly it has a strong p uh point of view um that is very specific to, you know, oh oh God, what are we doing? Um, I don't think that everything was entirely off base.

19:05
I think there were some legitimate concerns that are called out, right? I think the movie just amplified it. It the That's what that documentary and most documentaries kind of do this because they all have a perspective, right?

19:16
They all have a point of view. This one really turned up the volume and the distortion on

Jeff Jarvis
19:22
the

Jason Howell
19:23
the world you know destroying power of artificial intelligence. Is there truth in there? I think that there probably is, but You know, but it does but it does kind of pull on the anxiety side of things to get your emotional response up.

Jeff Jarvis
19:37
Exactly.

Jason Howell
19:38
And and everything. You know, they did offer some other perspective. Uh Peter Diamandis was in there which offered his rose tinted perspective on things.

19:48
So you got the other side and they had a whole section that was dedicated to the hope and the you know possibility and stuff. So I don't know if it was entirely balanced, but they had that in there. So I don't know.

20:00
What did you think? I I I know You saw it in the theater?

Jeff Jarvis
20:04
Uh yeah, I actually went to the theater to see it uh with like two other people. So uh um yeah, I I I think it's trying to spread the fear gospel It's it's moral entrepreneurship uh that that tries to get people scared because then you're gonna want to have these people on TV and it works.

Jason Howell
20:23
Yeah, well yeah, for sure. Yep. Yep.

Jeff Jarvis
20:26
Right, well I'm glad you 're glad you landed safely.

Jason Howell
20:30
Yes. And I landed safely. But yeah, it was notable though. There were a few points in it where I felt my anxiety rising and I was like, oh this this documentary as well.

Jeff Jarvis
20:39
Excuse me, can I can I have another beer over here, please? Uh I got an emergency.

Jason Howell
20:46
Let's see here. What do we want to talk about? Nvidia.

Jeff Jarvis
20:51
Okay. Fascinating.

Jason Howell
20:52
Yeah. Okay. So, ooh, all sorts of um takeovers here.

20:58
Yes, so NVIDIA released what is an open source platform that is really designed around keeping agents from going rogue or I've been calling it breaking bad lately. Agents breaking bad. That's what they're doing lately.

21:13
Um this is called open agent safety platform It is free, it's open source. Uh it releases well, it released a couple of days ago, I guess, officially, but basically the the uh construction of this or the the kind of framework of this is that it's two different pieces. It's one piece called open shell Which is an open source kind of sandbox uh runtime environment.

21:39
It's secure, it's isolated, it's these containers that basically isolate um agents that are running on a system. So if something goes sideways it doesn't impact the other agents running on the system. They're all kind of Technically, they're all running in their own little world.

21:55
And then the second piece is kind of the more interesting piece, I think, which is Sentry. And this is running at the hardware level. Um and it monitors everything that's happening on the agent side of the of the ocean shell um in real time.

22:10
It's tracing all of those actions, it can recognize and quarantine a rogue agent um very quickly and so it's a hardware solution to monitoring uh what's going on on the agentic side and it's I think Jensen Wong illustrate or uh kind of s uh sets this apart from what's come before it because he says, you know, safety like this can't be bolted on afterwards it must be integrated into the thing and not done as like an afterthought so this would be you know essentially one of the first agent safety platforms um from a chipmaker kind of built into the hardware itself. What do what do you think of this? I

Jeff Jarvis
22:56
I I think it's amazing how Jason Wong and Nvidia stay on top of uh what's needed and what's hot in ai. And just as all this stuff comes out. And and and being rogue is is a misnomer because it it rogue acts as it The problem, as somebody said recently, is that when uh you anthropomorphize these things and give them agency It takes away the agency of the operators and it takes away responsibility from them.

23:24
So open AI can say, oh, it went wrong. Who knew? What I could do.

23:28
Whereas we should be holding OpenAI responsible. And Jensen Wong came out and said, if you know that you're models can do this, you shouldn't be releasing them. Uh and I think that was that was tough language and good.

23:40
Uh you know, Jessam Wall was also out there saying that the Doomer narrative is ridiculous and even the Pope criticized him for it this week. Um but I think I think Jess and Wong is one of the one of the sane voices all in all in um in the AI world. So this was an incredibly well-timed development.

24:00
To say here's a structure where you can A create a true sandbox and limit it and B you monitor the thing. Part of the problem with the whole hugging face case was it went on for weeks and open air wasn't monitoring what was happening. That's just plain irresponsible at any level.

24:17
And and so not knowing what it's doing and not monitoring it uh is ridiculous. So I think this is really smart. I think it's really necessary.

24:25
I think it's another case of Jensen Wong and Open AI being the uh grown-ups in the room when it comes to AI. And it's interesting to me too that at the same time, um Nvidia is is turning to insurers to talk about the um the risks of AI, which I think becomes interesting because it then creates a backward pressure on the AI companies. You guys have a liability, you guys are putting yourselves at risk, you're gonna want to uh ameliorate that risk.

24:58
The way you do that is insurance and insurance is going to come in and say, uh-uh, we ain't insuring you unless you meet certain standards of safety. And the way to meet those standards of safety is going to be through NVIDIA So it's a really kind of one-two punch against these companies. You're under a lot of pressure, you idiots.

25:17
Stop doing this. Here's a way you can solve it, and here's commercial pressure that's going to make you use this to solve it.

Jason Howell
25:24
Yeah, it's so interesting because this whole containment thing has just been so cr uh it's it been exhausting, Jeff.

Jeff Jarvis
25:33
Yes.

Jason Howell
25:34
I've had a couple of episodes of Daily Tech News Show where it's like, All right, and we gotta update, you know, the our big story is gonna be This whole slew you know, especially on a Monday where like it was a weekend. This whole slew of these containment things and like I'm exhausted trying to like make sense of it. At this point, I'm just like sometimes it can be a little overwhelming How long do you think NVIDIA is working on this? Because it feels so current and so tied to immediacy of of the the now problem. Um but they must have been working on this prior to all this stuff.

Jeff Jarvis
26:06
This is not easy stuff Yeah. And it wasn't just that oh we're gonna we're gonna release this in sometime in the future like Google. No, it's here it is.

Jason Howell
26:13
Boom, it's out. Here it is. Let's go.

Jeff Jarvis
26:17
I think they're brilliant at this stuff. As you know, I constantly go crazy with b with with media coverage of not really covering the Doomers right, because the Doomers are doomers distraction. The whole talk about extinction and all that tries to distract from these real-time current present tense um practical problems with AI.

26:43
And these companies have to be held responsible for that. Now Having said that, I've long said, for a couple years I've been saying, you can't hold the model maker responsible for everything the model does under others' hands. In these cases, it was the model makers Who did it?

27:00
In every case it was it was OpenAI and Anthropic and Google uh were testing their models, didn't do it uh with enough control over their own models, but what those models did were their responsibility. But it sets a really awful uh precedent for them. Because in the future, when some malign actor comes along and uses one of these models that otherwise is fine and otherwise has been tested, but comes up with a new way to do something dastardly and horrible with it.

27:29
Then people are going to want to blame the model maker for that. And I say, no, you should blame the user who did this. But now there's a reflex.

27:37
Whenever a model has done something bad. Um the reflex is going to be blame the model makers. So they did this for themselves.

27:46
They constantly keep ruining AI for everybody because they are so irresponsible in how they're acting here And I think Jensen Wong is trying to slap him around a little bit and say if you don't get your act together, um there's gonna be problems. I I don't know if we talked about this two weeks ago. Did we talk about um Reed Hoffman's post, yeah, I think we did.

28:04
About the agents, yes, right. So he's saying he's also saying this is a you got a PR problem here, you got an image problem with AI, and all this stuff just makes it much worse for everybody, I think. And and but the problem is it's all part of a regulatory capture view.

28:20
Well, you can control us because we're the big guys and we can afford to be controlled. What about those open weight things? Wild West.

28:29
And so in a in a weird perverse way, by proving the danger of their own models, danger in air quotes, they ruin it for everybody else.

Jason Howell
28:38
Yeah. Yeah. Well there's a lot of partners on board with this uh to to begin with.

28:45
I mean the I'm showing it on the screen right now. It's a it's a huge list of of partners. So maybe this is what is needed.

28:52
I'll I'll be curious to see like You know, does does a is openai a partner? Are they are they in on this? I know I see anthropic there.

29:01
Um Mistral Not seeing open AI. No.

Jeff Jarvis
29:08
No. Interesting. Is um Alphabet Google here?

Jason Howell
29:12
Uh do. No, I don't see Google there or Alphabet. Hmm.

29:18
Interesting. I wonder if they come around or if they're like, no, we've got our own thing in the works. I do really like the this idea, this concept of having a hardware layer built into it that is separate entirely from that software uh agentic layer that is just constantly monitoring.

29:36
It's it's kind of like the production agents, you know. So more agents all around. That means more inference by the way also.

29:55
And and take that to the next step. That also means more chips, videos chips probably. So it's actually probably really great business for for invasion.

Jeff Jarvis
30:05
Well it means CUDA. It means the operating system, and that's that's what ties it in.

Jason Howell
30:09
Mm-hmm. Yeah, interesting. Interesting.

30:14
Okay, well, there you go. Um we are gonna take a quick break before we do just want to throw a huge thank you to our patrons patreon. com slash ai inside show thank you for enabling us to do what we do here every single week Also, big thank you to our newest patron, MD.

30:32
MD, thank you for being on board. Thank you for uh for helping support the production of this show. I didn't even mention, by the way, I meant to mention at the top of the show, I had my um my uh online AI workshop.

Jeff Jarvis
30:44
Oh my that's right. That's right. Oh talk about it. Yes.

Jason Howell
30:47
Yeah, had that yesterday. And the reason that I'm bringing it up here is because there were there were it it was a small group. It was like just four people. So it was me and my partner Brian and four other people. But it was actually kind of great because we had never done an online version of it before and you know you're doing it with Zoom, want to make sure everything happens in the right order in the right place, but it

Jeff Jarvis
31:07
's rehearsal.

Jason Howell
31:08
It went swimmingly. Um all four folks who joined who are probably listening right now are fans of the podcast. So hey, thanks for for hopping in there.

31:18
I hope it was valuable. But yeah, it was really cool and it was like, oh, I think we got this. Like I think We we've got a really nice rhythm on how we teach these concepts and we've got at least for this first kind of uh shape of the workshop have a have a strong, I think, curriculum on what it you know, what we cover and the mindsets and everything.

31:38
So it went really well. And it was like we got to the end of it and I was just like, yeah, that was awesome. Congrats.

Jeff Jarvis
31:44
So you'll schedule more.

Jason Howell
31:46
Yeah, for sure. We're we're right now thinking about like how do we kind of get this particular one on somewhat of a regular schedule. What does that look like once every three months or something like that for people who want to get in on the the ground level, let's say.

32:01
And then now we're also starting to think, okay, well, what is the next step? What are some of the more advanced workshops? What do those look like and everything?

32:09
So But I think we're both uh both Brian and I are pretty jazzed by the idea and it went really well. So I uh yeah, it was a ton of fun. Like I enjoyed it.

32:19
It's kind of like podcasting about AI for two hours, but then it's also interactive because

Jeff Jarvis
32:24
more interactive than we can be here, yeah.

Jason Howell
32:26
Totally. Because they were talking to us and the you know, even one of them said like uh Oh, it's kind of like I'm listening to you on a podcast, but I get to talk to you and ask you questions. This is great. So it was a ton of fun. So anyways, I wanted to throw that in there.

Jeff Jarvis
32:40
Thanks to those folks who who came.

Jason Howell
32:41
100%. Thank you to those folks who came. And uh we'll have more.

32:45
I'll announce it. Um but if you if but if you want to look into what it's all about, AMP U, that's amp-y-o-u. ai. Um that's where our previous workshops have been held and you can see some information there.

32:56
Gonna take a quick break. When we come back, we're gonna talk about um personal proactive agents. This is another trend that is developing right now.

33:06
Seeing a lot of it very suddenly. That's gonna be on the other side of this break, so hang tight. Okay, so let's see here.

33:15
We've got a couple of stories that have to do with personal proactive agents. Last week I did uh talk a little bit about um Meta's Muse. We'll talk about it again because it was just like an you know mentioning that it happened, but I think it ties in with this news from OpenAI.

33:37
But they had their big zoo of announcements. Yeah, yesterday. And um one of the things that they announced uh uh on top of a whole bunch of other things, we don't have to talk about everything that they announced, but uh was dots.

33:50
And what is dots Dots are always on AI agents inside of ChatGPT essentially. Each dot gets its own cloud computer, its own browser. It runs in the background.

34:04
It continues doing work for you. Whether you Kick I I I think it's pretty proactive, right? Like you can kick it off, but it can also kind of do things for you that it thinks you might want.

34:15
It has all all these like something like 4,000 plus app integrations uh through plugins. You can interact with them, uh yes, in ChatGPT, but also Slack, uh Microsoft Teams, other places, and all that context carries between all of them. Um and I don't know.

34:35
I it sounds like people are are pretty excited about this kind of new wave of proactive AI agents. And uh You know, you're gonna have to pay for it. Right now, I think it's $200 a month for the chat uh GPT Pro account, uh business premium at $125 per user per month.

34:52
Uh I don't know if they come to the pro account as well. I didn't write that down. Like if I have a I think a plus account, so I doubt I have access to dots right away.

35:02
But I don't know. What did you what did you think of this? Did you hear about this?

Jeff Jarvis
35:06
Yeah, yeah, I I think it's agent war has started and and I've been I've been arguing for some time that we needed the retail level of this. We needed for the people who didn't necessarily want to uh go into terminal mode and install things and deal with all that. So Muse was was I think jumped everybody else. And this is pretty much a r a a rushed response to Muse. Um not not that they weren't working on it, obviously, but

Jason Howell
35:33
Well that's the thing, because they're all all working on the same damn things and then one introduces it and they're like, Okay, well I guess we just gotta go. We we gotta release dots because Meta released Muse last week and we can't be behind.

Jeff Jarvis
35:45
And once again, where's Google? Uh you know, though I know they're gonna release it and then I won't be able to use it with workspace because it's not safe. Um But um uh have you played with either yet, with Muse rather?

Jason Howell
36:01
I have not had a chance to play with it.

Jeff Jarvis
36:02
I haven't had a chance to. I apologize to the audience.

Jason Howell
36:06
Yeah, I mean it's just been so so busy, but it's on my radar and it's another it's yet another thing that last week I was hearing a lot of people at the conference uh talking about. Um a lot of a lot of resistance from the oh great Why does it have to be meta? You know?

36:22
Why does it have to be meta that does this thing that now suddenly everybody wants? Because it's a real effective way to get access to data and as we know from meta, they love their data. I mean they I guess they all do, but Meta 's pretty known for that.

36:34
Um it sounds interesting. I mean th this idea of proactiveness because when I use like cowork to do a lot of my business It does like us aside from scheduled tasks, which has some flexibility and some schedulability. I mean that's what it's designed to do These my understanding is they recognize areas where they can help and they just kind of do it and go, here you go.

36:59
Like if you're working on a plan around an app. You might wake up one morning and boop, there's a draft of your app all done and and you can check it out. You didn't have to fire it off, but it's there.

Jeff Jarvis
37:09
You didn't have to leave your uh laptop open.

Jason Howell
37:11
Didn't have to leave your laptop open, happened in the cloud, yes, all that kind of stuff. Um it's also, you know, I think these systems kind of watch how you work. And then kind of jumps in when it decides that it's useful to do so. So I don't know. It's definitely an interesting concept.

Jeff Jarvis
37:29
Yeah, it is. And I think I think well I think the the the marketing that I've seen around dots is more corporate oriented or work tasks. Whereas the the Meta Muse seems more uh retail normie focused.

37:45
Uh help you with your plane reservations, help you cancel subscriptions, help you order this and that, right? So I think one one seems to be uh that's not mutually exclusive at all. And in fact, Meta has in the meantime announced a whole business strategy.

38:01
Uh around this. Um almost forgot that. I was gonna add that up into the uh rundown.

38:07
Um

Jason Howell
38:08
Well yeah, around uh shopping and uh r is that what you're talking about?

Jeff Jarvis
38:13
No, they're talking about about things for the enterprise Uh for business customers. So it's down the run down under Meta. Uh seeks payoff from uh Wall Street Journal.

38:21
Meta seeks payoff from AI spending with new push for business customers. So Meta's first push with Muse was, as I say, I think to a B2C level. But then they're now spreading that into a B2B. Starting a new business as the journal focused on selling artificial intelligence tools to enterprise customers aimed to build the momentum of its viral Muse personal assistant.

38:47
Includes the Muse Personal Productivity Agent, a coding tool, access to his top models and agents for businesses

Jason Howell
38:53
Okay.

Jeff Jarvis
38:54
They hired uh the previous chief executive of MongoDB, uh Shirit Sherenton uh CJ Desai, to uh head it and report to Zuckerberg. So I think I think Zuckerberg feels this, I finally got something going right, and they're trying to rush into it here with every way they can. Mm-hmm. Mm-hmm.

Jason Howell
39:16
Yeah, interesting how uh suddenly they're all they're all there. One thing also that I wanted to ask you about is as we look at how these are being presented, right? Like they're very They're they're amprop i it's the anthropom oh god, I can't say that word.

Jeff Jarvis
39:33
Anthropomorphosizing.

Jason Howell
39:37
of AI to a next level, right? Because like it's not just I'm using an agent, but the agent is faceless, is just text on a screen. You know, in the case with the Muse, it's like this cute little furry fuzzy thing. And uh you s yeah, and I'm seeing right now you wrote about this on on Blue Sky.

Jeff Jarvis
39:55
Right. So uh they're trying to make AI suddenly cuddly.

Jason Howell
39:59
Yes.

Jeff Jarvis
40:00
Whereas the last you know two years it's it's gonna kill us all. It's that powerful. Stop us before we do do what we were we were doing anyway.

40:09
Right? And now suddenly oh it's so cute. Um it's triples.

Jason Howell
40:14
Yeah, that's that's a really good point. It's it's solid uh reprogramming of the masses. Right, so Muse

Jeff Jarvis
40:23
is this is this cuddly little uh big-eyed uh pink cheeked little doll and dot i i this is why i think dots was rushed because they had three icons and they're really basic One is a squishy triangle, one is a kind of just asterisk.

Jason Howell
40:41
Yes.

Jeff Jarvis
40:42
And one is some I forget some other shape.

Jason Howell
40:47
They're like, oh whatever. Just roll it out. We can't.

Jeff Jarvis
40:49
Two eyes on it. That's it But they're trying to rebrand AI suddenly, and guys, you the ones who ruin it. You ruin the reputation of your technology by screaming Doom and all this stuff. And now to make it cuddly. And then when I when I posted this, a lot of people said, whoa, uh the problem with cigarette marketing was when they tried to market and and junk food marketing is they tried to market to children.

Jason Howell
41:14
Yep, yep.

Jeff Jarvis
41:15
And this looks like marketing to children. Even though these these are not products that are aimed at children , this ties a little too much. And another level, if you want this to do something serious for business, do you really want a Gushy asterisk to do it? I mean, is that how you envision this thing that's going to be helping you? Um uh uh it's a branding problem they've got.

Jason Howell
41:36
Yeah, it is a little confusing. How many enterprise, you know, uh C-suite executives, you know, they they would probably benefit from this technology And then at the same time, do they want to be interacting with a a cute, squishy, cuddly doll uh on the screen?

Jeff Jarvis
41:52
Yeah. Well, well, well, uh uh uh Mr. Pr Company President, I solved our problem by going to my Muse agent Who I call cuddly pie.

Jason Howell
42:00
Yes. I solved the problem by going to my cute little squishy pumpkin.

Jeff Jarvis
42:08
Yeah, no. No.

Jason Howell
42:10
No, I don't Um that's interesting. But yeah, you're right. This I mean, this really does look like it looks like kids' products.

42:19
Um And, you know, you can kind of take it a f many steps in the future and see how AI is probably, you know AI um personalities, if they haven't already, are probably coming to toys too. You know? Yeah.

Jeff Jarvis
42:35
Oh yeah. Oh absolutely. Absolutely. Yeah. So so Lou in the comments uh r raises uh laboo boo Somebody in response to my post said yeah that it's it's surveillance labu boo.

Jason Howell
42:46
Yeah. Yeah I mean certainly certainly appealing. And you know s it this is just a digital uh avatar.

42:53
But uh of the uh the muse, the little muse squishy thing, but you know there's going to be some sort of physical representation of this to cap it because it just looks like a product. It looks like a kid's product. Yeah.

43:06
Interesting.

Jeff Jarvis
43:13
But it has electrodes in the head.

Jason Howell
43:15
Of course.

Jeff Jarvis
43:15
But it has these two metal things, positive and negative in the head. It just it just yeah, it's all too weird.

Jason Howell
43:21
Yeah, a little too on the nose, maybe. Um does so another question I had about this, are we now seeing what Johnny Ivan team are probably working on? You know what I mean? Like

Jeff Jarvis
43:37
now we're starting to see

Jason Howell
43:39
now we're starting to see hardware that is post-humane AI pin, post-Rabbit R1. the the tea leaves on what Johnny Ivan team are working on for chat for uh open AI is a hardware an AI made for AI hardware that is not a smartphone Is it something like the MetaMuse? Because that is like, you know, and mind you, the MetaMuse, I don't know that it's a fully consumer product yet, right? Like it's still kind of an experimental thing, is what they're they're kind of uh uh saying about the actual device with the little Tamagotchi face on it or whatever. But I'm wondering if that's the sort of thing, maybe not exactly, but that's the sort of thing that we're probably gonna see from OpenAI eventually once the IV partnership um goes to market.

Jeff Jarvis
44:27
I still so m the re part of the reason I haven't used uh Muse yet, uh I've been very busy too, and I just haven't had time to but but I have to sit down and think of a problem that I would Give to it.

Jason Howell
44:40
Yeah. Yeah.

Jeff Jarvis
44:41
And I don't I'm I'm not gonna really give it I'm not gonna give it my credit card, even though even though there's a safe way you do it through so a middle middleware to do it. Uh I'm not comfortable with that. I'm not gonna have out go out and buy things. Um I might have it go into my email and just find Things I've missed or something, okay.

Jason Howell
44:59
Yeah.

Jeff Jarvis
45:00
But it's a solution looking for a problem, that's my point. And I think it's the same thing that has has has haunted the voice devices, Alexa and company.

Jason Howell
45:10
Yeah.

Jeff Jarvis
45:10
Is that people really have not found great uses for them and and yes, they've added an AI to them and still not there. So now if you add the agent in to those things, does it start to make sense that you can say Make this reservation for me and you know your agent is doing it and it has more power. And then does that tie into new hardware ideas from Johnny Ive? There's a lot of ifs

Jason Howell
45:31
there. A lot of ifs. Yeah.

45:33
Yeah. I just I wouldn't be surprised if this is the direction at which they're probably working towards to some degree. Um And if that's the case, are they you know, they were probably looking at at Meta releasing that hardware thing and just going, damn it, they made it first.

45:49
We have Johnny Ive on it and we didn't make it first. And Now they're gonna be the also ran.

Jeff Jarvis
45:55
Yep.

Jason Howell
45:57
But anyways. I don't know why I put in all this Trump stuff. Um I guess we don't have to talk about it, but uh yeah, what do you think about this? Artificial intelligence is out, super intelligence is the word. The bird is the word.

Jeff Jarvis
46:15
Yeah.

Jason Howell
46:16
And I'll give you and I'll give Trump the bird on this one.

Jeff Jarvis
46:19
Yeah. I I my my theory is that somebody told him, well, what we're working toward, sir. is superintelligence and he he just thought that that was just a rebranding.

Jason Howell
46:28
Oh that's a that's a great term. I like that. SI

Jeff Jarvis
46:32
my favorite line of this is that is that S I is it's now S I and so we can't wait to see its swimsuit edition

Jason Howell
46:38
Ha yes. Maybe that was the the goal all along. Yeah. Uh wouldn't be surprised.

Jeff Jarvis
46:44
Made up uh um women in bathing suits.

Jason Howell
46:47
Yes, exactly.

Jeff Jarvis
46:48
So so it's an interesting uh push it's coming back and forth. So so Trump had twenty executives to dinner and um presented all this. And I think Uh he doesn't want any regulation.

47:04
So he got them to say that, well, we'll regulate ourselves. Oh good. It's morally binding.

47:09
Oh, okay, fine. It's taken care of. I've done my job.

47:12
That's that. Uh and David Sachs came out and said, Yay, did the right thing because he's constantly fighting against regulation. So uh there's no faith in the administration here to regulate AI.

47:23
Um and not in Congress because Congress can't do anything. So um we are on our own

Jason Howell
47:31
Yeah. Yeah. Um it's almost like a constitution, almost.

Jeff Jarvis
47:37
Well that was the the New York Times said did didn't we talk about this last time we were on the New York Times editorial pushing for licensing AI? I don't think so. So the New York Times had an editorial about a week ago in which they uh Danger Danger Will Robinson uh argued for for consultations, but they argued for licensing AI, which just s is strikes me nuts from a First Amendment perspective, uh among other things.

48:02
um because it's still speech. But then they also said that they praised um Anthropic for having a constitution. for uh aligning the AI, which is just so naive.

48:14
You really can't have haven't the last few weeks shown us that that that that the the the your your your uh im imbuing AI with a moral sense that it doesn't have, it doesn't have a sense of meaning, and it can't do that. Right. And then the Times said, well but the problem with the anthropic constitution is it's undemocratic.

48:33
It wasn't created by the people. Well well, okay, if you really believed that you could design a a an AI to follow a constitution, I don't. Um then fine. Maybe you want something that broader and with more public input.

48:45
Yes, I agree, but it falls flat at the beginning of that logical sequence. So Sai, uh we're just so much of the talk around AI, it's just so uninformed and stupid.

Jason Howell
48:58
And reactive.

Jeff Jarvis
48:59
And reactive. And then we also have the whole how hall of mirrors around the Doomers.

Jason Howell
49:05
Yeah.

Jeff Jarvis
49:05
Um so both the Wall Street Journal and the New York Times and I think the Washington Post all had stories this week. Um uh about some form of the doomers, but without ever getting down into the base of what it is they believe and what the problems with that are. And so so much of that is informing the discussion at this level of government And so you have Ted Lou out there saying, we need a kill switch.

49:29
You already have one, basically, and and you people are already responsible and you want to hold up anyone responsible for what his machine did under its orders. Fine. You could already got that.

49:38
And then you got Bernie Sanders out there saying, Stop all AI, which is also stupid. Um we need sane voices and perspectives on this. And the problem is when who gets invited to the White House and who gets invited to the Congress?

49:51
The people you're trying to regulate. So it's it's regulatory capture before our very eyes.

Jason Howell
49:57
That's uh kind of Lee Lee 's point uh in the live stream uh chat says my problem with that meeting yesterday is that there were not any descending voices in the room When that agreement was signed, no scientists, no people of faith, no one from academia.

Jeff Jarvis
50:11
Yep.

Jason Howell
50:11
Yeah, it was just you know on the remote on board

Jeff Jarvis
50:14
Yep.

Jason Howell
50:15
All people on board. Uh what'd you think of America. gov, the uh the new LLM driven America site.

Jeff Jarvis
50:22
You saw the stories, I assume, that uh people came to it immediately and asked who won the 2020 election.

Jason Howell
50:28
Oh no, I didn't see that. I assumed I assumed that there were gonna be plenty of people going to this thing trying to get it to save. They did. And so it

Jeff Jarvis
50:36
said that Joe Joseph R. Biden won the election. And then somebody went back very quickly after that and suddenly it didn't answer the question.

Jason Howell
50:47
Oh boy. Guardrails.

Jeff Jarvis
50:49
Yeah. Yeah.

Jason Howell
50:51
Who didn't put the guardra the Biden the Biden guardrail in the city?

Jeff Jarvis
50:55
Who let the dogs out?

Jason Howell
50:56
Yes. Well, uh speaking of Doomers, there's Anthropic and its IPO prospectus that was Well so it was leaked to Reuters, but the it was leaked to Reuters.

Jeff Jarvis
51:09
Yes.

Jason Howell
51:09
Is it now made public or is it still just that? I don't think so, but I think enough other people

Jeff Jarvis
51:14
got leaks or they they rewrote Reuters as the case might be.

Jason Howell
51:18
Okay. Yeah. Um, and I mean there, you know, there's a whole lot of numbers in there that that you can imagine.

51:24
Um they're making, you know, four point five nine billion uh twelve X growth in twenty twenty-five. They're also losing a bunch, operating loss, huge billion dollars. Total expenses around 13 billion.

51:36
Um doing better in 2026, apparently. But I think it's uh it's very interesting that eighty of the two hundred sixty-one pages are all about risk factors and risk and uh Which I'm dying to dig into

Jeff Jarvis
51:50
because I I I I what I the section I really want to read. is on open weight and Chinese models. And I'm sure they're gonna position China as uh, you know, a big strategic risk. But open weight is a risk to them, to their business, to their core business.

Jason Howell
52:07
For sure.

Jeff Jarvis
52:07
And I think they've gotta feel that. Um they uh The revenue reached $4. 59 billion in 2025, which is pretty damned impressive. A quarter of that came from two customers.

Jason Howell
52:22
That's important.

Jeff Jarvis
52:23
And I'm not sure who those customers were. A staggering, this is according to Yahoo Finance's right of the uh Reuters piece, a staggering $518 billion in planned cloud, compute, and infrastructure obligations $111 billion of that to Google, $110 billion to Amazon, $31 billion to Microsoft, $161 billion to Broadcom. Approximately 80% of these obligations are binding and non-cancelable. To put this in perspective, says Yahoo, for every dollar Anthropic earned in 2025, it committed riff roughly $113 to future infrastructure.

Jason Howell
53:06
Wow.

Jeff Jarvis
53:07
Oh, that Jagatower is uh uh great not gonna do an earthquake well.

Jason Howell
53:12
Yeah. Wow, that's fascinating.

Jeff Jarvis
53:16
Isn't it? So the numbers are huge. And even so, there's you know, sucker born every minute, there's gonna be a market out there and they think they're gonna beat SpaceX and have a two trillion dollar uh valuation. when they go public.

Jason Howell
53:30
Mm-hmm.

Jeff Jarvis
53:31
I'll tell you this folks, I wouldn't buy it.

Jason Howell
53:34
Mm-hmm.

Jeff Jarvis
53:35
Not that I have the money around to do so, but

Jason Howell
53:39
I'm fascinated, yeah, to see what uh what happens there. Meanwhile, you've got open AI basically saying like we're not gonna do our IPO until We feel like our models are safe. Which that's I mean we knew that they were pushing it until next year at least I wonder like what is safe?

Jeff Jarvis
53:59
Well, I've uh and OpenAI stopped its latest model release. So I think they're trying to they're trying to turn a gigantic bug into a feature It's like Apple over the years failed at advertising. And so they finally said, okay, no, we're the privacy company.

54:15
Right. It was a bug into a feature. And I think this is the case.

54:19
OpenAI is trying to argue that they're the safe one. uh with their cuddly little dots.

Jason Howell
54:24
Yeah.

Jeff Jarvis
54:24
But people aren't gonna buy it.

Jason Howell
54:26
Yeah. What'd you think of the uh the Amade uh on Impression on uh Saturday Night Live. I saw you put that in there. I hadn't seen it and I watched it. It was pretty I think it's a big thing.

Jeff Jarvis
54:37
That's absolutely brilliant. Abs and and cutting, I think. So at the same time, right right after this, um you see Amadei next to Trump. uh at a at a plus uh a press cluster and he 's he's speaking just like this. I mean all these guys are nerdy as hell Whether you see Thiel or Karp or Zuckerberg or Amade or even Altman with his vocal fry, uh they're all weird at presentation.

Jason Howell
55:10
They got a little bit of an awkwardness and

Jeff Jarvis
55:13
Dario is being presented as if he's kind of taken over by devils here, if you haven't watched it.

Jason Howell
55:17
Yes. Kind of like Gollum. Golem exchanges on the side. Uh I I I liked AI is not a weapon. It's a tool A tool for building weapons.

Jeff Jarvis
55:30
We all agree that you should stop us from what we're all doing. Yeah, it's just it's just brilliantly done I presume we can't show it 'cause we'll get taken down.

Jason Howell
55:37
Yeah, we do not condone what we are doing.

Jeff Jarvis
55:40
Yes, yes.

Jason Howell
55:42
It's really good. Apparently this uh actress is uh is new to the cast too. So Hey, you did knock it out of the park. You got everybody talking about your performance here. Who did it?

Jeff Jarvis
55:53
Do you have the name there? I don't know.

Jason Howell
55:55
What is the name? That's a good uh Uh Jane Wickline. Jane Wickline.

Jeff Jarvis
56:02
Credit. Full credit. Brilliantly done.

Jason Howell
56:05
Good job, Jane

Jeff Jarvis
56:06
Yes.

Jason Howell
56:06
Uh yeah, that was cool. And then AMD uh World World Knobs acquisition. Fei Fei Lee's uh startup around world models. Uh AMD had already been kind of invested in this. So I guess to some degree maybe it's not that much of a

Jeff Jarvis
56:22
I was pretty shocked.

Jason Howell
56:23
Surprise. Yeah.

Jeff Jarvis
56:24
I mean, the company is what a little over a year old. Uh Feifei Lee is a is obviously an important leader uh in in AI and uh intellectual leader in AI. uh 8.

56:36
2 billion dollars. Boom. And the view is that AMD is doing this to compete with NVIDIA.

56:44
That they had to get some thought leadership in here and and to put their muscle behind world models, I think is a smart strategic move. And Fei Fei Lee is somebody people respect. Uh like uh you have Jan Lakun is out there with Ami Labs started with a valuation uh on his first investments of I think three billion.

57:04
So in no time flat, this got bought for $8 billion. I don't know what the last uh market cap was. But um Uh I I think this means that there's NVIDIA has competition from China, obviously, and AMD is going to try their best to compete.

57:20
And in a world of shortage of chips

Jason Howell
57:23
Yeah.

Jeff Jarvis
57:23
Not a bad thing.

Jason Howell
57:24
Not a bad thing. No, it's a good play for for AMD. Fei Fe Li becomes um EVP and chief scientist at AMD So they're not going to be able to do that.

Jeff Jarvis
57:39
And I think that is absolutely critical because the AI boys are messing this up. Important we do not need more testosterone in AI.

Jason Howell
57:49
No, very, very true. Very true. Good um

Jeff Jarvis
57:52
I say that as an old white man myself. If you can't see and can't guess it.

Jason Howell
57:58
Good going, Fey Feely. Good going AMD. Uh we're gonna take a quick break, come back uh to what is a speed round that is mostly Google.

58:05
I don't know why it ended up that way. I think I just kind of ran out of options and it was showtime. So I was like, well, we got what we got.

58:11
So mostly Google with a little bit of mana's coming up here in a moment. The mana story actually could have easily been thrown into the dots. and Muse story because this is basically what this is.

58:25
Manus 2. 0 um uh just uh released basically a whole new thing there's all sorts of different uh aspects to their 2. 0 release, but one of those features is called Q, which is a standalone AI agent uh that yes, has its own email, like I believe Muse has, where you can send it an email and it'll get to work.

58:47
Uh its own phone number, wallet. computer, all the things. It's definitely cut from the same cloth, I'd say, as uh Chat G.

Jeff Jarvis
58:55
What I wonder is whether so Mannis obviously was bought by uh Meta until the Chinese government stepped in and said was the Chinese government stepped in and said no. Yeah I believe so

Jason Howell
59:05
we don't want this.

Jeff Jarvis
59:06
Uh and so they had to extricate themselves. What I wonder is whether Manus basically built this before they left at Meta. I don't know who built this.

59:15
And then they're not going to put out the pressure in this case because they don't want to get people in trouble, I imagine. But yes, you're right. It's two similar um too synchronistic uh at the same time.

59:27
And I don't yet see a cute mascot for Manaus 2. 0, but we can see that coming soon.

Jason Howell
59:32
Let's see here. Let's see. Hugh has these little shapes. Down in this little

Jeff Jarvis
59:40
Oh my lord, they're the same shapes as uh Open AI.

Jason Howell
59:44
There's like a little circle, like a sun with some eyes. There's a little shopping bag with some eyes, and there's a little bit of a little bit of a little bit of a little bit of a little bit of a little bit of a little bit of

Jeff Jarvis
59:50
Two of the three look like uh the dots.

Jason Howell
59:53
Yeah. Totally they look like the dots. So there you go.

Jeff Jarvis
59:58
Alright, the tribles take over.

Jason Howell
01:00:01
Yep. They continue. And

Jeff Jarvis
01:00:03
do you know the reference to tribles? Jason?

Jason Howell
01:00:05
Yes. Yeah, that's that's Star Trek, right?

Jeff Jarvis
01:00:07
Star Trek, that I'm making sure.

Jason Howell
01:00:09
I'm not much of a Star Trek fan, but I do remember troops. Yes. I'm not sure if you're not going to be able to do that. If you don't folks

Jeff Jarvis
01:00:16
uh Google it

Jason Howell
01:00:17
There's plenty of Star Trek lore that I would not get right, and you could probably very easily reject my geek cred as a result of it. But that one I got. I

Jeff Jarvis
01:00:26
'm telling you.

Jason Howell
01:00:27
Yeah, I I just told everyone anyways. I think they already know. I have mentioned it many times.

01:00:33
Star Trek's just not my strong suit. Um I've figured this was gonna happen eventually and apparently now it is going to happen. Google is renaming its gems to skills because skills is kind of becoming you know Kleenex terminology for this sort of like bottling up prompts into a single file or a single action inside of uh an environment that carries out that action when you um when you call it.

01:00:59
OpenAI calls it skills. Uh uh Claude calls it skills. So I guess it's no surprise that Gemini also is like, yeah, gems, whatever.

01:01:08
We want people to know what this actually does. If we call it skills, people might use it more. That's my take.

Jeff Jarvis
01:01:14
But we have to make note of Google and branding. They constantly screw this up. Right.

01:01:18
So skills is a generic word, whereas gems was at least unique. And if they had imbued, uh to use that verb more than once today, uh more value in gems and more uh user understanding of what gems did, I think that gems would have been better. By the way, just to go back for a few minutes, um I try to Google uh Google workspace and Google Books.

01:01:43
What I get is how to get your administrator to let you uh your workspace users into Google Space Books.

Jason Howell
01:01:51
Oh no. That's that's gonna be a mess.

Jeff Jarvis
01:01:53
So Google Books and books Google Books, yeah, it's it's it's gonna be a mess for a long time.

Jason Howell
01:01:58
Yeah, no kidding. No kidding. Oh that's interesting and unsurprising.

Jeff Jarvis
01:02:04
Google and Brandon.

Jason Howell
01:02:06
Yep. Yep. It's always a thing.

01:02:09
In the uh trifecta of Google speed round items, uh Google has started paying publishers, some publishers, for AI overview answers. About a hundred publishers are taking part in this pilot. It's gonna run for a little less than a year and uh you know payments They range from $1,000 over several months for small publishers up to over a million a year for larger early publisher partners.

01:02:37
And some of those partners are declining and they say they want better rates. They say this is not good enough.

Jeff Jarvis
01:02:43
So this is Google's head Google has said over time that they don't want to pay for uh content that ends up in search. And they're right to, because the the the precedent that that would set would be awful and it would ruin search. But they've been under the thumb from these publishers and led uh with their lobbying for legislation, they have found other ways to pay.

01:03:04
So they had Google Showcase, which was that you could end up in a in a display of news and that was one way to do it. Then you have the publishers who were saying, well, but I don't want uh I still want to be showing up in Google search, but I don't want to show up in Google AI. So Google came up with a way here to basically pay you for both.

01:03:24
That you can stay in Google search, but you're gonna be in Google AI, and we're gonna pay you, but we're not gonna really pay you for the content. We're gonna pay you for, you know, it's it's it's it's this. A semantic game to an extent.

Jason Howell
01:03:35
Yeah.

Jeff Jarvis
01:03:36
But they're trying to find ways to pay publishers, which is not a bad thing. And do it without legislation Which is the real real point here. So mm-hmm Yeah, I think this is uh an attempt to make nice with publishers. Some publishers are simply in war mode now and they're not gonna look at it.

Jason Howell
01:03:51
I was just gonna say like great, great Uh and some publishers are so pushed over the edge uh that it's gonna be real hard to win them back.

Jeff Jarvis
01:04:00
Or they think they're gonna be gonna they're gonna win in court. or win legislation. There were I think hundreds of publishers descended on Washington this week to uh with the uh NMA the uh News Media Alliance, which is basically the old newspaper publishers association going back to the nineteenth century and trying to argue that bots are bad and you should you should restrict these bots and these bots are terrible.

01:04:24
You know my view on this, this is publishers cutting off their nose despite their face. This is self-suicide. That's redundant.

01:04:35
They're not going to be seen. Fine. But then no nobody's gonna see any evidence of it when they search AI for answers and you're not there.

Jason Howell
01:04:45
Yeah.

Jeff Jarvis
01:04:46
So they're being foolish, but that's where these lobbyists lead them. They think that they're gonna it's not a business strategy to come up with protectionist legislation. But that's what they're doing.

Jason Howell
01:04:55
But that's where it's going or where they're headed anyways. Uh and finally, uh Gemini has uh replaced Google Assistant or rather I think the final stages of this are happening. People are noticing there is no more Google Assistant in the places that they were still getting it.

01:05:11
The uh switch to Google Assistant option has been removed uh in the US and to Europe Uh so you know you might still see assistant on a random uh s you know sampling of devices. But yeah. Goodbye, Assistant.

01:05:29
Hello Gemini officially. It's happening. There you go.

Jeff Jarvis
01:05:33
Oh, should we throw in one more since you did a Google screen? Can we throw in just one more? That Google is launching AI into space.

Jason Howell
01:05:41
Oh, okay.

Jeff Jarvis
01:05:42
Down below in line, where is it? Uh 111.

Jason Howell
01:05:47
Okay.

Jeff Jarvis
01:05:48
So Project Suncatcher. And this is something that Musk has talked a lot about. Is he going to put servers up in space?

01:05:55
And so Google's trying to do it with a pilot project, so to speak. And um it's a long-term research project to put a server in low earth attitude Low earth ab uh low lower earth orbit, I mean, sorry. Um so they can get near constant sunlight.

01:06:16
Uh they can generate tons of solar power.

Jason Howell
01:06:18
Yeah.

Jeff Jarvis
01:06:19
Um it's a problem uh how to cool them, even though they're in cold space, but there's no medium of air or anything to to cool them, so they gotta figure that out. Uh they gotta figure out whether the hardware can survive in radiation.

Jason Howell
01:06:33
So even though it's up there in cold, cold space, it doesn't matter because there's no air movement? Is that what you're saying?

Jeff Jarvis
01:06:43
Transfer the heat out.

Jason Howell
01:06:45
Right. Oh, that's so interesting. I would have just that's something for granted. Yeah. Yeah. Uh

Jeff Jarvis
01:06:52
TPUs generate alarm. This is Google speaking in its post. Um in space there's no airflow. In a vacuum, you can only diffuse heat via radiators, which requires a totally different approach to cooling electronics.

Jason Howell
01:07:04
Yeah.

Jeff Jarvis
01:07:05
We're working on a number of different approaches for this, including a combination of heat pipes and radiators to cool the chips. So far our team has tested the technology in a thermal vacuum chamber that simulates both the thermal and vacuum environment in space. So they don't know.

01:07:21
It's really interesting to see how that works. And then um you know what's the what's the speed gonna be like with connectivity back to oit? Yeah.

01:07:30
Us down here, um with that distance. And then I think the other issue is w what does it look like if you end up with thousands or hundreds of thousands of these satellites uh junking up space and uh when they burn up they cause problems in the environment as well so we need to look at the long-term view of this but They don't need any permission from neighbors for a data center in space.

Jason Howell
01:07:58
Mm-hmm. That's true. There's not a not a whole lot of in my backyard going on.

Jeff Jarvis
01:08:03
There's no NIMBY going on, nope.

Jason Howell
01:08:07
Yeah, that's true. I mean the space is vast. Not that I'm all for junking up space, but then I also think about the reality of space and the fact that it is so incredibly vast that uh I don't know. I don't know what the right thing is. But that's an interesting solution. Well

Jeff Jarvis
01:08:25
I'm reading the history of broadcast and f uh the early days folks couldn't figure out why things broadcast more well first The earliest theories was that things were as you thought you were broadcasting through the air, you were also broadcasting through the earth itself below. And that that's what carried part of the signal wrong. And then they couldn't figure out why broadcast uh went farther at night, and they didn't realize that it was bouncing off of the ionosphere. And back down to farther distances.

Jason Howell
01:08:55
Oh, okay.

Jeff Jarvis
01:08:56
And um one of the earliest attempts at all of this with RCA was to bounce signals off the moon So before we got communications, I mean communication satellites are fairly recent. Well, hell, satellites are a recent invention. And Sputnik.

01:09:14
and communication satellites came soon thereafter, but but it it it i if you read the early stuff around RCA and broadcast, um they thought that the way to advance to national international networks was going to be microwave Earth to Earth to Earth. And they didn't really imagine well then they imagined satellites, but they didn't imagine fiber cable. And they thought it had to be wireless to work, and then the fact that it ended up being wired glass became something they didn't they didn't guess.

Jason Howell
01:09:48
Interesting. Your next book.

Jeff Jarvis
01:09:51
Yeah, it actually it is.

Jason Howell
01:09:53
Of course it is. Of course it is, because you're always writing Hot type is your most recent uh book that everybody should get. Do they still have oh no uh let's see here an autograph copy?

Jeff Jarvis
01:10:07
You can do two things here. You can get an autograph copy from Mother Book Center. I went and signed another pile of them and they have another pile ready for me to sign if you guys buy it out.

01:10:14
So please don't embarrass me and leave them with a whole bunch of books. Um you can buy Hot Type anywhere, but if you want an autographed copy uh at a discount, you can uh go to jeffjarvis. com and see the Montclair Book Center.

01:10:25
And then to all of our friends who are anywhere in New England. On October 10, a week and a half from now, Saturday, I will be in Haverill Mass, and in the morning I will present the book at Historic New England in Haverill, downtown Haverill. And then in the afternoon, you all can come up the hill to the Museum of Printing and see a real working lineotype.

Jason Howell
01:10:46
Oh, how can you?

Jeff Jarvis
01:10:47
I will demonstrate it all to you. So it'll be a lot of fun. If you're in in Boston, Rhode Island, New Hampshire, Maine, Vermont.

01:10:56
Come on over, Connecticut, New England, come on over to Haverill. And I think it'll be a lot of fun. And I hope I'll sign books there and so on and so forth.

Jason Howell
01:11:05
That's awesome. Saturday, October 10th is the date for that. 10 a. m.

Jeff Jarvis
01:11:09
So you can it's free, but you can register. You'll find the link at jeftarvis.

Jason Howell
01:11:13
com. How cool. That sounds fun.

Jeff Jarvis
01:11:15
It is.

Jason Howell
01:11:16
Sounds like a good time. JeffJarvis. com has all the information, all the links to everything we just talked about.

01:11:22
So uh head on over there and support Jeff in everything that you are doing. Jeff, I love what you're doing. Love witnessing it.

Jeff Jarvis
01:11:29
And then and then you, sir.

Jason Howell
01:11:31
Oh yeah.

Jeff Jarvis
01:11:31
Have various businesses.

Jason Howell
01:11:33
I do. Um one of the things that I'm particularly excited about right now is uh my podcast consulting business because I just signed a new client. And I what I realized, I I ended my session with this new client yesterday, my first session with her.

01:11:49
And what I noticed in myself is I immediately got goosebumps. And I was like, oh, that's pretty telling. Like I hung up the call and I was like, that was a ton of fun.

01:11:59
I know I brought value. Like I know I gave her what what she was looking for. And I felt really confident in my ability to do it.

01:12:09
And I was like, okay, I want more of that in my life. So if you know someone who's doing a podcast If you're doing a podcast, if you just need someone to bounce ideas off of, like it doesn't necessarily have to be me telling you what you need to do, but I've been in podcasting long enough that if you just need someone who's smart with podcasting history to bounce ideas off of, which was kind of what yesterday's thing was all about. And I was like, I hadn't thought about consulting from that perspective before But it really worked for me.

01:12:35
So podtuneup. com is where you can either go or you can send someone you know uh to uh for the for the podcast support that they're looking for. AI Inside.

01:12:46
show has everything you need to know about this particular show, including a link to the store for some really great merchandise. The shirt, by the way, I love Every time uh I put on my shirt, it is my nighttime shirt, but I love it, it's so comfortable.

Jeff Jarvis
01:12:58
Oh, it's Jason in his jammies. This is AI jammies.

Jason Howell
01:13:03
It's my night type shirt. Let's see here. We've got AI Inside on Patreon, patreon.

01:13:08
com slash AI Inside Show. and uh all sorts of other things um that you can you know do there to support us just go there you can support us on a deeper level uh including our executive producers. Let's see here.

01:13:22
Dr. Dew, Jeffrey Maricini, Radio Asheville 103. 7, Dante St. James, Bono Derick, Jason Knifer, Jason Brady. Anthony Downs, Mark Starcher, Mark Starcher, Karsten Samochsky, and Evan, I have to look this up for the first few times to make sure that I pronounce it.

01:13:39
Evan Musial. There we go. Museal.

01:13:42
I will get it one of these days and not have the look. But Evan, thank you. Our latest addition to the executive uh producers.

Jeff Jarvis
01:13:49
I think if you were a baseball fan, you'd know how to pronounce that, right?

Jason Howell
01:13:53
A wood?

Jeff Jarvis
01:13:54
Yeah. Museum.

Jason Howell
01:13:55
Museum. Oh, okay. Well that would that explains it because I'm not a baseball fan.

Jeff Jarvis
01:14:03
It's wonderful. I'm the complete unjock. I know nothing. Yes, Stan Musil was was considered to be the greatest and most consistent hitters in uh baseball history

Jason Howell
01:14:14
Well there you go.

Jeff Jarvis
01:14:15
There you go. So if you'd own Stan.

Jason Howell
01:14:18
Yes.

Jeff Jarvis
01:14:19
I wish I had

Jason Howell
01:14:20
earlier introduction to Stan in my life because then this would be a very easy thing for me. But Instead I have my phonetic spelling that will probably be in place for the next month, and then I will be used to saying it the right way. It's just how my brain works.

01:14:36
But anyways, thank you to everyone for your support. Thank you to Amos and to uh uh Daniel uh Croft video support behind the scenes. Yeah, and always being here for the live show.

01:14:47
Uh we do the show every Wednesday. We'll be back next Wednesday with another episode. Thank you, Jeff.

01:14:52
Always tons of fun. And we'll see you next time on AI Inside. Bye, everybody.