
OpenAI wants Chrome, Cluely, Oscars accept AI, the cost of please, and more.
Loading summary
Jason Howell
This episode is brought to you by Chevy Silverado.
Jeff Jarvis
When it's time for you to ditch the blacktop and head off road, do it in a truck that says no to nothing.
Jason Howell
The Chevy Silverado Trail Boss.
Jeff Jarvis
Get the rugged capability of its Z71 suspension and 2 inch factory lift, plus impressive torque and towing capacity thanks to an available Duramax 3 liter turbo diesel engine. Where other trucks call it quits, you'll just be getting started. Visit chevy.com to learn more.
Jason Howell
You know that feeling when someone shows up for you just when you need it most? That's what Uber is all about. Not just a ride or dinner at your door. It's how Uber helps you show up for the moments that matter. Because showing up can turn a tough day around or make a good one even better. Whatever it is, big or small, Uber is on the way. So you can be on yours. Uber on Our way. This is AI Inside episode 65, recorded Wednesday, April 23rd, 2025. The era of experience. This episode of AI Inside is made possible by our wonderful patrons at patreon.com aiinsideshow if you like what you hear, head on over and support us directly. And thank you for making independent podcasting. Hello everybody and welcome to another episode of AI Inside the podcast where we take a look at the AI that is layered throughout so much of the world of technology and beyond. I'm one of your hosts, Jason Howell, joined as always by Jeff Jarvis. I can hear you now. How you doing Jeff?
Jeff Jarvis
Hello there. Sorry about that. Jason is like in five places in my life right now. I'm on YouTube, I'm watching stream on Twitter. There's Jason. Jason's everywhere.
Jason Howell
You're checking all the things. I know when I cut to you, I heard myself echoing in the background. I'm like oh, is. Is that me now or me then me in the past?
Jeff Jarvis
It's the Matrix.
Jason Howell
Yeah, it is the Matrix. And you know what? I'm getting comfortable living in the Matrix. This is just. This is the year 2025. 2025 is the year where we realize we are living in a simulation before we get started. Huge thank you to those of you who support us on our Patreon that's Patron of the Week. Dan Merchant is one of the the amazing patrons that support us each and every week. Just go to patreon.com AI insideshow if take part in that. We really do appreciate it. And again, just a quick call out. Leave us a review or a star rating on your podcatcher of choice. But the reviews are really helpful. We have gotten at least one that I've seen in the last week that is more current, and that's kind of what I'm hoping for, is to kind of freshen up the reviews and get some newer reviews. So even if you have an older one, renew it, refresh it, whatever. We really appreciate it. But let's jump in because this is a news episode. We got a lot of news to talk about, and I think maybe we start with Google, and I say Google ahead of OpenAI, even though OpenAI is at the start of this, this news story as well. But Google's antitrust trial is. Would you call it the punishments phase? I don't know. It's kind of like how can we punish you?
Jeff Jarvis
Google Shed phase. Yeah.
Jason Howell
Yes, exactly. The company was found guilty of its. Well, found guilty of unlawful practices in online search and advertising in the US and as a result, the Department of Justice has recommended. I kind of feel like it's not going to happen, but has recommended that Google, you know, if we get our way, Google's going to have to unload Chrome. They're going to have to find a new owner for Chrome. And it sounds like OpenAI has an executive that testified at the trial. Hey, we'd be interested. We, we'd be down. That's Nick Turley, head of product, who testified that the company would be interested in owning Chrome. Just saying, if you're selling it, we'd certainly be interested in owning it. What do you think about that, Jeff?
Jeff Jarvis
It's a stunt. It's just like perplexity saying they're going to buy TikTok. It's now the way to punish everybody is to make them sell something and then everybody jumps up and says, I'll buy it. And it's a kind of ridiculous story cycle that we're in now. And OpenAI. A, I trust Google with Chrome a lot more than I would trust OpenAI, period. B, I think it's a stunt. I think it's just for the. What? It worked, but we'll see all over. We're doing it right now, but it was in the. We're doing that because it was in the news all over and so that's where it is. And C, I wonder what the real value is to OpenAI. Sure, it could insert itself in that browser, but hello, antitrust. It's the same problem that it tries to fix. Then it's even worse because the company will use it for its own purposes and not allow others in. Google has always allowed others in, but.
Jason Howell
That'S a 20 year problem, Jeff. That's 20 years from now when they finally say, oh, actually that was a bad idea 20 years ago.
Jeff Jarvis
Just like Microsoft and the browser in the past, my browsers have been such a, a focal point because I think they're the main, they're our main entry into India. You know, I remember many, many years ago, at the beginning of the web, my son, we ran focus groups when I worked for Advanced Publications in Cleveland. And the people in the focus group said, you know, there's this amazing thing on this online. It just has everything. It has the weather, it has sports, it has news, it has fun. What's that? It's called Netscape. And you know, we've seen for a long time people don't understand the brands that underlie the web and everything else. I think that's probably less the case now. But I think there's this kind of naive view among both regulators and media that the browser is everything, whether it was Microsoft or whether it's now Chrome. So it seems like a kind of a silly moment in all of this. And it's serious stuff going on with Google. Absolutely, I see that, Chris. And the questions ask whether Google could make another browser. I don't know because there is no decision yet whether this is actually going to be the path. It was a recommendation. And I don't know what limitations there might be, but once again, I would trust Google more than I would trust OpenAI. And what does it do to all the rest of the services? The browser is key to everything we use to email and Docs and Drive and maybe not Maps Translate. All these services we use are out of this hub of the browser. And if you try to split that off, it's like saying the phone company can own the handset, but somebody else has to own the phone. For those of you who remember the old days where those were two different parts. I'm sorry, I just dated myself.
Jason Howell
And there was a wire that went into the wall.
Jeff Jarvis
Yeah, curly cord on it too. For you kids. I'll show you in Jeff's museum later.
Jason Howell
Well, I think what you were just talking about kind of illustrates both sides of. You were kind of saying the browser, like it's long been seen as this very, very important thing. But I don't know that that's really, you know, maybe the case anymore. Maybe I'm getting your words mixed up a little bit. But. But also it is really important. And we do channel and funnel so much through it. And so I can see why a company like OpenAI might love to have Chrome as their kind of anchor for, you know, especially when we're talking about the AI ambitions, you know, to have a browser that you can just then completely connect your AI service and all of those agentic qualities directly into, you know, Perplexity has its comet browser that it's doing this. I think Perplexity would be another, you know, kind of interesting party to, to want to own it. I don't think that they've explicitly said, hey, we'd be interested, but I wouldn't be surprised if they do. And nor do I think any of it matters because I think at the end of the day Google's not going to have to sell Chrome. That's my hunch. That would be a, like, you know, to your point, the, the kind of tangled web of everything that is connected that would interplay in that move just seems, that seems like a lot. And I know that the doj in a case like this, at least my, my recent understanding of this is that they shoot for the moon and often they end up somewhere in between where it was and where they're shooting for. And I don't think that that place in between necessarily means Google has to get rid of Chrome.
Jeff Jarvis
Yeah, it's an entirely different DOJ than it was when this case was brought. So who knows what that'll be in the long run. They're continuing on the theme that they had of Google bad, but we'll see. But you know, you raised two other points that I think are really interesting, Jason, and one argues with the other. The one is that there's finally competition in browsers. Perplexity is going to create a browser and they have reason to do so. And if OpenAI is hungry enough for browsers, it could make one. It wouldn't cost.
Jason Howell
They're going to.
Jeff Jarvis
It's trivial. So on the one hand, just as the argument is that this is anti competitive and we have to pull it away from Google, there's competition. The other contrary argument is that we talk a lot about how whether generative AI replaces search. What if it replaces the browser? What if agentic AI replaces the browser in essence, and it makes the browser far less important because your pathway to applications and to information and to functions is going to be otherwise. It's going to be through command structures, new command structures, voice and so on and so forth. Whether that happens or not, we can predict, you know, till the cows come home. Yeah, but the idea that the browser, it's exactly the same as the Microsoft fight, the browser was the key to everything. And then it wasn't for Microsoft. They lost the browser war. There was competition and there's competition still. So. Yeah, I want to agree with you that I don't think they'll be forced to do this these days. I can't predict anything.
Jason Howell
Yeah. Who knows?
Jeff Jarvis
But the other thing that obviously bothers me is I'm a Chromebook guy.
Jason Howell
Right.
Jeff Jarvis
And what does it mean to have sell Chrome apart from is that the browser alone, is that the os? What does that really mean?
Jason Howell
Yeah, good question.
Jeff Jarvis
Yeah, go ahead. Sorry.
Jason Howell
No, no, no, no. I want to hear what you had to say.
Jeff Jarvis
Just one other thing is the other day it just occurred to me when Google originally used to go to google.com and there was the blank on the page and you type that in. Right. And when Google went to the. What's it called, Jason? The one bar or the one Whatever.
Jason Howell
Yeah. What is.
Jeff Jarvis
Yes, when the address bar became everything. Right. It was actually, it was confusing. It confused me for about a week. Well, was that address bar, is that where I go to put things in? And when I would go to the course to the Google search page, it would go ahead and put it up into there to train me.
Jason Howell
Yeah.
Jeff Jarvis
And say Omnibox, where you're doing everything. Omnibox.
Jason Howell
Thank you, Omnibox.
Jeff Jarvis
Well, so the Omnibox, the browser is not just a browser. The Omnibox is the path to all kinds of functionality. So anyway, the browser is a fungible beast.
Jason Howell
Now when's the last time I went to a Google search page and clicked on the search area in the middle of the page and put my search in there? It happens very rarely and randomly and I couldn't even tell you what is the circumstance that takes me there. But everything that I do is in that omni box. I mean, and what you're saying also really reminds me of the conversation that I had at Mobile World Congress with Google's Android head Sameer Samat, when we were talking about a post app world where agentic AI becomes so prevalent, is there a need for apps? Is there a need for applications when Agentix AI can just kind of go to the places it needs to go to do the things. And I think Samir's point was, was also you know, appropriate and spot on, which is that even, even in that world there is still a need for companies, for brands, for destinations to have some sort of a kind of a place to go or a, you know, maybe they've got a brand that they want to convey and that's that's how you do it. Like the agentic AI can do those things, but it might not necessarily mean that we don't have those other things as well because they also serve other purposes too. So. So I don't know. I do see if, you know, obviously this was a publicity stunt on OpenAI's side to gain more of the oxygen from the PR room, which it's very good at doing. But no matter what, OpenAI guaranteed gonna do the browser thing. I guarantee you they're working on it behind the scenes already.
Jeff Jarvis
And it's gotta be trivial. In fact, in fact, tell Open A, tell the browser, tell the chat to make the browser software. It's fairly trivial, I think.
Jason Howell
Yeah. And I guess what's coming, what's coming to me right now also is that there's some, there's some overlap here between like, how necessary is, is it to AI to have its own browser in the same way that how necessary is it to AI to have its own piece of hardware like a rabbit R1 or what ever. You know, it's. Everybody's looking for these different ways to make it, I don't know, make it more of an immediate utility and seems to be doing all right in the form that it is right now. But, you know.
Jeff Jarvis
We should pay tribute to the browser. Was a paradigm, paradigmatic shift. I was working at Delphi for one horrible month, which was Delphi Internet, before I got the hell out.
Jason Howell
Remember that?
Jeff Jarvis
And they were going to have a GUI because everybody had the gui, right? There was AOL and Prodigy and so on. And you had to have your graphical user interface. And then along. I was, I remember the day when somebody came in and said, you got to see this thing. And it was the browser, the first crude blue browser. But it, it, it immediately said that changes everything. That's a pathway to things. And that's what a browser is. It's not a program in and of itself. It's just a way to get to stuff.
Jason Howell
Yep, interesting stuff there. And then we were talking a little bit, you know, I mentioned Perplexity. Perplexity comes up a lot these days on this show. I think this is interesting. Perplexity is working on some big deals right now. One of them we might actually hear more about tomorrow with Motorola. This was a deal that, I guess Motorola has an event tomorrow. It's expected that the event is going to be about their new Razer phone. And according to Bloomberg, probably going to get some information around a deal that Motorola has struck with perplexity to have its perplexity Agent preloaded on Motorola devices. Gemini, I'm guessing, would still be present on the device. I don't think that this is necessarily saying that Motorola would not have Gemini installed, you know, and Google's Gemini is out and Perplexity is in. But I think this brings the option of another AI assistant, or at least the app, onto the device. And then there's Samsung, which apparently is in early talks with Perplexity as well right now. I think according to the case that we were just talking about, it was revealed that Samsung has a two year kind of licensing deal with Google. Samsung's been working really closely with Google on a lot of things and one of them is bringing Gemini into Samsung phones. But it turns out Perplexity is talking and potentially making deals with Samsung to bring the Perplexity Agent onto Samsung devices, potentially in place of Gemini. And as all of these court cases happen and start to kind of strip apart the status quo of how Google does its business and strikes its deals, this could be something that we see more of in the next couple of years.
Jeff Jarvis
So back to the prior conversation. I didn't expect this to be tied in, but it, but it is. Let me ask you a question. Because you're a pro phone user, right? You study phones and how they operate and your use of them is critical to your research. Right. When you think of doing something on your phone, what proportion of the time do you. I think these are three choices. You go to directly to an app, do you go to the browser, do you go to the assistant?
Jason Howell
Oh, that's a really great question. Trying to break it down. I mean, I probably. That's a fantastic question. I don't. Okay, I'll start with Assistant. I don't use the kind of baked in shortcut for Assistant very often.
Jeff Jarvis
Right.
Jason Howell
Neither do I. I mean, and that's really slowed down. I think that was different when Google Assistant was more leveraged and kind of a little bit more dependable and new because it was the new thing. I wanted to get in the habit and I certainly had that habit for a while. I don't use it as much and I definitely don't use it as much with Gemini. It's not something that I go to regularly. Every once in a while I do. I'm almost more inclined to launch the app, honestly, to do that either with Gemini or with Perplexity, which I do launch Perplexity and it does have the voice assistant capability inside the app. And so I will sometimes launch that. Am I going to like. I could probably Open up my pixel and, you know, in the settings and assign the Perplexity Voice agent as my main agent. But I choose not to. I couldn't tell you why. I think it's part of the reason is because Gemini is kind of tied on a little bit deeper level to the Android operating system. Like Perplexity's Voice agent is great for search.
Jeff Jarvis
It could be called an antitrust trial just for that comment.
Jason Howell
But it's true. You know, this is. This has been Google's big, you know, strategy of for better or for worse. And I think, you know, it's. It's starting to kind of bite them in the butt. It is the intertangled web. At the same time, I'm a consumer and I kind of want those conveniences. So that's part of the reason why I don't kind of remove Gemini from that placement and put Perplexity in its place. And then how often do I go to the browser? I don't know. That's. That's almost always. If I go to the browser, it's because I used the Google search on this, you know, that's on the home screen to like ask a question of something or to, you know, if I, if I really need to go directly to a website, I guess I just put it in there. I don't know.
Jeff Jarvis
That's interesting.
Jason Howell
It's hard to. It's hard to give concrete answers on that. I don't know how to answer that other than when I need to.
Jeff Jarvis
So. So I'm an old fart. So I mainly go to the browser.
Jason Howell
Okay.
Jeff Jarvis
There's certain apps that I'll use, obviously. You know, there's a weather app or whatever. But I tend to. My reflex is what I'm saying. I want to, I want to get somewhere. I want to look up something I want to do. I go to the browser. If I'm using the app, it's really more voice search. You know, I'm sitting at the dinner table and how old is Brook Shields now? I don't know. I'll ask. You know, hey, gee, how old is Brook Shields now? That's really not AI. It's really not the agent. I don't think it's just voice. That's more of.
Jason Howell
Yeah, just voice search.
Jeff Jarvis
Yeah, it really is. So I'm not using the agents much at all. Which. The reason I ask that an answer is what's the bet? If you're trying to deal with Samsung, what's the bet of what the. What the user use basis is going to be? Of these things. Yeah, of, of a default agent versus a default browser. I don't know. But again, it goes back to our prior discussion. The browser may not be the. For me, it's still the hinge point of everything. For you it's not.
Jason Howell
No, no, definitely not. The browser is definitely not the hinge point for everything. But I've got my feet in all different pawns, I suppose. Yeah, you're weird. But I do think, and we talked about this a little bit last night on the Android Faithful podcast as well about like having a perplexity as a voice assistant tied into my phone by default does eliminate, like I said, some of those deeper kind of connections to say Android operations or settings or some of Google's kind of, you know, kind of the special sauce that Google has integrated with their services into Gemini and stuff. So having perplexity in that spot eliminates or reduces some of that functionality. But as we know in AI everybody kind of has their favorites or they have, they're AI models that they turn to for very specific certain things and their go tos. And so that might not matter as much. It might not matter to every user the fact that when they use their voice assistant, it can't know what to do with turn the lights on in the living room or whatever where Google's can. It might just matter that 90% of the time when I want to use voice AI search, it's because I'm researching or it's because I'm doing this thing and therefore I want that to be assigned to something that has a different skill set that's more tailored to what I actually need, not, you know, doing what Google thinks I need.
Jeff Jarvis
Yeah, yeah, yeah. Well that's a really good point is that the, the, the assistant, AKA agent will be charged with having more knowledge about you. It will be more personalized necessarily. That'll probably be killer, not killer app. What I'm trying to say, the characteristics that makes it win.
Jason Howell
Right. It also has a lot of stickiness to it. OpenAI is really working hard on this right now. You know, they're really kind of doubling down and opening things up as far as memory for users. And turns out that becomes really, really handy and helpful over time as the model learns what, what you're constantly looking for when you ask it to do this certain thing. That means as a user that's less hoops I have to jump through to get the answers that I'm looking for. And I think this is just kind of an interesting fact is that the more we give into these models and get that memory, the more sticky those models become. Because why would I want to take my toys and go over there when this one already knows so much about me? I'd have to start over again going over there. And I think our phones are going to get there too, too. And that'll be really interesting from, from that perspective.
Jeff Jarvis
Yeah. And I, I can already hear replaying some of the, the, the battles of yore. One is obviously privacy. It knows too much about you. How do you, you know, what control do you have? And the other is the filter bubble argument will resurface. And the filter bubble argument was made by Eli Periser and then Axel Bruns wrote a book called Our Filter Bubbles Real in which he had lots of research. No, they're not that Google was not in fact personalizing to the level that it was presumed was not putting you in a filter bubble. But an agent will. So what was worried about in a moral panic past will come back perhaps with some cause.
Jason Howell
Yeah. Yeah, Cool. Well, we've got a whole lot more to talk about. We're going to take a super quick break and then we'll talk a little bit about Demis Hassabis interview on Six60 Minutes. That's coming up in a second. Let's talk about something we don't talk about enough. What happens to all the data we share with AI platforms like ChatGPT or Claude? Every question we ask, every idea we brainstorm, it's all being collected and tied back to us as individuals. But then what does it get sold to advertisers, corporations, maybe even governments. We've also grown accustomed to social media companies selling our data over the last decade. And I'd like to think that maybe we've learned a thing or two so we don't make the same mistakes again. That's why I've been using Venice AI, who's sponsoring today's episode. Venice AI is private and permissionless, using leading open source models for text code and image generation. And it's all running directly in your browser. So there's no downloads, no installs. In fact, your chats and history live entirely inside your browser. They don't even get stored on Venice's servers. Their pro plan is where things get really interesting, though. You can upload PDFs to get insights and summaries. You get a user controllable safe mode for deactivating restrictions on image generation. You can customize how the AI interacts by modifying its system prompt directly. And finally you get unlimited text queries along with high image limits that I couldn't even hit if I tried. We talk often on the podcast about the benefits of open source AI and that's exactly what Venice AI is using. If you care about privacy like I do, or you just want an uncensored and truly open AI experience, Venice AI is worth checking out. Go to my Sponsor link Venice AI AIInside. Make sure to use the code AI Inside to enjoy private uncensored AI. Use my code and you'll get 20% off a pro plan. That's Venice AI AIInside with code AI Inside for 20% off the pro plan. And we thank Venice AI for sponsoring the AI Inside podcast. This episode of the AI Inside podcast is sponsored by BetterHelp. I've noticed a big shift in recent years towards taking mental health seriously and I welcome that change because I recognize firsthand the benefits of taking care of my own Mental health therapy can be a transformative experience and it definitely has been for me. But no question it can be pricey. Traditional in person therapy can run anywhere between 100 to $250 per session and that adds up and it really should not stand in the way of getting the help that's needed when it counts. BetterHelp is online therapy that can save you on average up to 50% per session. With BetterHelp you pay a flat fee for each weekly session and that adds up to big cost savings over time. And not only that, BetterHelp is much easier to access than traditional therapy because it's an online experience that meets where you are at with quality care from more than 30,000 therapists at a price that makes sense. You just click a button to join. Your therapist is there from wherever you happen to be. You can get support with anything from anxiety to relationships to everyday stress. And if you just aren't feeling it with your current therapist, you can easily switch to another at any time. It's mental health within reach and it's totally worth it. I know firsthand. I used BetterHelp a few years ago myself. It was incredibly convenient and more importantly, impactful to my life. I felt heard and supported and that's what I really needed. Your well being is worth it. Visit betterhelp.com aiinside today to get 10% off your first month. That's BetterHelp. H E L p.com aiinside and we thank BetterHelp for their support of the AI Inside podcast. Did you did you get a chance to see Demis's appearance on 60 Minutes?
Jeff Jarvis
No, I've Been writing too much about the head of 60 Minutes quitting.
Jason Howell
Oh, different 60 Minute story entirely. And I was not aware of that, actually.
Jeff Jarvis
Oh, yeah, I was on CNN last night talking about it, actually.
Jason Howell
Oh, no kidding. Wow. I'll have to look that up. Well, yes, Google's DeepMind CEO did make an appearance on 60 Minutes over the weekend. And I think it's interesting because there were a lot of, you know, fresh off of our conversation with Yann Lecun from Meta. There were definitely a couple of points throughout the interview where it was like, okay, I've heard that before. Jan was talking about this too. Particularly the fact that Demis was, you know, basically saying, you know, AGI, as people define it very differently or don't, is at least 10 years down the line. So Demis and Yan appear to be on the same kind of timetable as far as when they think this randomly defined concept of artificial general intelligence will actually happen. And it's not immediate, it's definitely somewhere down the line. But he did make the prediction that AI could potentially cure all diseases within this next decade. All diseases. That's kind of crazy to think. I feel like anytime you put all or nothing, there's at least a small amount of invalidation in my mind to something like that. Because really, all of them, maybe that's just an easy way to say most, but just say most would be probably where I would go with that. But yeah, I don't know. What do you think about that?
Jeff Jarvis
I think even most is too far. It's a turbocharged view of technological solutionism, and the argument is that the Internet really brought out solutionism, thinking it's going to solve everything and it's going to bring peace, and certainly has not. Not.
Jason Howell
That's not.
Jeff Jarvis
But this is a whole other level where it just, it's part of the AGI ASI presumption that we're going to get there and it's going to be so amazing. It can do these things and it's being ascribed to it with no reason, no basis. Will it help with medicine? Yes. Will it help find different uses of molecules? Yes. Will it do things behind the scenes like protein folding? Yes, all that's. Yes, all that's amazing enough, but this just goes overboard in a ridiculous way, in my view. And I think it's harmful in the long run on two sides of puts a target from his perspective, it's dangerous because I think it puts a target on the back of the technology he's building. Well, it's going to fail. It's not going to reach the heights that it's been predicted to do. And on the other hand, it makes it more fearsome. Oh, it's all powerful. It's gonna, you know, it's. God, it's not. Let's take the wonder that we can have with this on its level. Why does it have to be everything? It's, it's irritating. And I respect him and I respect his work and he's a genius at this stuff. I'm not taking any of that away. Just don't oversell it, man.
Jason Howell
Yeah, hard, hard not to, I suppose, when you're that close to it and you know, there. Yeah, yeah, maybe, maybe they, well, they believe that they know something that the rest of the world does not.
Jeff Jarvis
But that's, that's part of the problem. Yeah. Is that I think, I think that sets up a distance. That's that they haven't learned what happened in this period of, of the arc of Internet hype. And this is this, this, you know, the Internet hype was hypey enough. This is 10 times hyper. Wouldn't you agree?
Jason Howell
Yeah, yeah. I mean, I was a lot younger when the Internet was first coming around and I certainly wasn't as analytical at that time. I was probably caught up in the hype more than anything because I was very excited by it. But it feels that way from my point of view now. At the same time, it's really impressive some of the accomplishments that have happened here. Right. Like he discusses, DeepMind's AlphaFold mapped more than 200 million protein structures in a single year. If that was equated to the amount of time it takes traditional researchers to do their work prior to this, that would have been 1 billion years of traditional research time and taking a sale. That's just amazing. Right? That's absolutely amazing. And that gives the confidence to say, well, if we're doing that now, then what are we going to accomplish in the next 10 years? It's going to be a million fold where we are right now.
Jeff Jarvis
Yeah. Presuming the hockey stick is applicable to everything in life because it presumes the basis of, if we do this much now, then you've given a definition of what this is and then you multiply it by 100, you say, well, that's everything. No, there's a lot of challenges in life and I'm glad the technology is. And we're both boosters of this to the extent that it does the amazing things. But the booster, the high end boosterism just drives me nuts.
Jason Howell
While perplexity CEO Aravind Srinivas agrees with Demis, called him a genius after this interview and says he should be given all the resources he needs to realize this goal. So that's, by the way, perplexity entering the conversation. It seems like more and more right.
Jeff Jarvis
Now they're brilliant at peer are. They are brilliant where OpenAI obviously was brilliant because it, you know, it took over the world and it's gotten all this money and so on and so forth. But just in terms of, of, of and, and, and perplexity is not as hypey, oddly enough. Right. It doesn't. I don't hear the AGI stuff quite as much from them. What I see is we can do this. Oh, we're going to enter this conversation letter in that conversation. We're going to buy browsers, we're going to buy tick tock. We're going to agree with our competitors. They just sneak in the stories. Just brilliant.
Jason Howell
Seems, seems to be the case. Yeah, this next one you put in there, and I did not have this on my radar and I thought this would be a really interesting conversation. The Trump administration considering a draft executive order that would direct federal agencies to integrate AI into K through 12 education here in the U.S. of course, it's in a very early form at this point. According to this article in the Washington Post, it would integrate AI into teaching, also administration tasks, create programs using AI technologies with partnerships with private companies and nonprofits and schools to create and promote foundational AI literacy. And yeah, interesting. I mean, this just seems to go deep. And obviously I have not read the draft executive order in its entirety. I've just read this article to kind of get a general sense of what's going on here. I find myself a little conflicted because on one hand I think it's really important to recognize this inflection point that we're in right now with technology and to in many ways embrace it, get ahead, if not ride that wave. On the other hand, it feels so sudden and drastic to commit so quickly to the level at which this article seems to illustrate.
Jeff Jarvis
Well, I love the word they used in the Washington Post story. It is a pre decisional, which is a word I hadn't heard before.
Jason Howell
And I'm sorry, the concept of a plan. It's a concept.
Jeff Jarvis
I have to do the joke here because the joke is obvious, but they're having instructs Education Secretary Linda McMahon to prioritize federal grant funding for trained teachers, blah, blah, blah. So she's going to put A1 sauce in our schools. You see the story last week that she confused. She kept on calling AI A1. And so A1 Sauce had a bonanza with that. And so we're all going to pour A1 sauce over our students. Yes. Having done the obvious joke, but this is. It's the problem with all these executive orders, is it with the stroke of my Sharpie, I can change the world. And Lord knows, in some ways he's doing it. But this is not that easy to just say we're going to put AI into everything. And the irony here is while. And I'm trying not to get overly political, though my views are fairly known, while they're cutting into education in every other way possible.
Jason Howell
Right. Well, that's part of what feels so drastic. Right. It's like, on one hand, take an axe to all this stuff. On the other hand, let's replace it with AI. And so deeply, like, just based on reading through this, it feels like such a deeply embedded kind of solution. Obviously, you know, they're chasing down, you know, countries like China who are pursuing, you know, integrating AI into their efforts in education. And there's a big sentiment right now in U.S. leadership that, like, well, we can't let China win the AI game. We've got to win. And so let's do, do it by every means necessary. And it's just. Yeah, it's such a response. It would be such a response if it actually passed.
Jeff Jarvis
Yeah. The fear, I think, is that if you're a teacher, they're going to come and say, well, yeah, okay, we just. We just gave you 20 more students, but no problem, you got AI, right? Or, yeah, preparation's hard, curriculum's hard, but you got AI now. So this makes your job easy. And of course it doesn't. Not at all. I. This morning I watched something still going on right now. William and Mary College, they. They did something about education and AI. And my friend Matthew Kirschenbaum, University of Maryland, and Rita Raley from UC Santa Barbara had done a piece in the Chronicle of Higher Education about whether AI will kind of ruin universities. And the joke today was, well, AI doesn't need to. It's happening elsewhere. But not a joke. But there's concern in the academe at that level, at the university level about the relationship to these big centralized companies, about the resources that are provided or not provided, about the freedom that academics will have to do things, and whether they were talking about whether they could run a model under the desk, which in a way maybe you can do with some of the stuff we're seeing. And so there's big concerns at an educational level about AI all around. Nobody is saying it's not amazing. Nobody's saying it's not a tool that we should use. Nobody's saying we shouldn't teach our students. Students. But this presumption that, okay, I can pour the A1 sauce into a syllabus and I'm done is kind of ridiculous. But there is a demand out there. So at Stony Brook, I wrote a syllabus for a course in AI and creativity. And last I knew, a week ago, it already had 91 students signed up. And so there's a popular demand and desire for this stuff. And so I think that's great all around. Just do it smartly. Don't do it as if you think. Think one signature and it's done. That's all.
Jason Howell
Yeah, reactively and swiftly. Although that's, you know, that's proving to be kind of a hallmark of where we are right now is reactively and swiftly, for better or for worse. So, yeah, like I said, I'm a little conflicted on this because I do. What I don't want is for the US education to only see the bad potential of AI. You know, well, students are going to learn to cheat, blah, blah. Like, I do believe that AI and what it, you know, the current state of LLM and everything that it's developing into through agentic and beyond. Like, I don't think this goes away and I don't think that wishing it or pretending like it doesn't exist does any good. And I don't think that the younger generations coming up necessarily see it or will see it that way either. They're going to embrace it in a way that we older people are not going to have as easy a time doing because it's not our normal, but it's their normal. And so, you know, so there is a need to kind of embrace and kind of lean into that education piece. Just please do it in a responsible way that doesn't throw and that involves.
Jeff Jarvis
A lot of other good and involves the community and how it's done. Yeah, and not just say, you're not doing enough AI.
Jason Howell
Right.
Jeff Jarvis
Do more AI.
Jason Howell
We need more. Everybody needs an open AI subscription. There we go. We've done it. Now do all your work on open AI. Okay, perfect. We've done it. We've done the AI thing. You know, that's one way to do it. We'll see. Okay, let's talk a little bit about AI generations because I thought this article, another one that you put in here actually was. That was. I don't know, I appreciated reading through it. I'm having a hard time pulling it up here.
Jeff Jarvis
But if you go to Archive today, you can get it.
Jason Howell
Yeah.
Jeff Jarvis
Not subscribing to Business Insider.
Jason Howell
Yeah, well, the problem is I try and pull up the archive links on Chrome and for whatever reason it never works for me.
Jeff Jarvis
Really.
Jason Howell
I had to load it in an entirely different browser. Browser in order for it to work. Anyway, that's a little behind the scenes. But you had put in this, this article that talks a little bit about AI eras. The fact that like, you know, not too long ago we were in the simulation era, which is kind of the AlphaGo era, where models were learning through repeated and digital simulations and reinforcement learning. And there was all the Alpha AlphaGo, you know, and playing the game and, and whoa, can you believe that the game is capable of playing this so quickly and dominating and everything. That was the beginning. Then there was the human, or rather is the human data era where we are right now, dominated by Internet scale data transformer models, of course, and where we reside right now. And then Google researchers David Silver and Richard Sutton have proposed, according to this Business Insider article, a major shift in AI development with a concept called the era of experience. And yeah, like tell me a little bit about the era of experience. Experience and what they say.
Jeff Jarvis
So yeah, I thought this was interesting. And by the way, this paper is going to be part of a book called Designing and Intelligence from MIT Press. So it's a preprint from Silver and Sutton and I agree with where this goes. The funny thing was it repeats what what Yann Le code told us. Yeah, right. So it's credit is given to Google and that's nice because they don't get much credit in the AI world as much as they want. But this isn't just Google saying this. Jensen, Huang, Yann LeCun, Google are all saying that the next phase has to be experience to teach AI reality. And that's where you're really headed.
Jason Howell
And it's going to reality world models.
Jeff Jarvis
Yeah, it's going to happen through robotics and it's going to happen through digital twins and it's going to happen through data gathering, through glasses and all that kind of stuff. But it's got to have some sense of cause and effect and it doesn't have that yet. It doesn't know that. So that's going to be really interesting. So I think that the point of the paper is good. Business Insider does kind of a simplistic view that Google's telling the world what for. No, yeah, right. This is where everybody's going and I think we're waiting for that. I don't want to say leap. I think it's just a. I'm going to use the word paradigm again. You know, when I worked at Delphi way back when they had a five dollar paradigm jar, if you use the word paradigm, you had to put $5, not just $5 in it. It was that much.
Jason Howell
It's an easy word to lean into. I, I am so guilty of that. In the new paradigm, I've had to try and back off of that word.
Jeff Jarvis
So, so there will be, I think a paradigm sh. Bucks already for this, this experiential layer. But I don't think we've seen it yet. Apart from robots obviously learning some things but in ways we can't touch because we ain't the robot or digital twin factories. But we don't touch it because we're not seeing what those alternative futures are or anything like that. I don't think we've seen the consumer level version of experience yet where, oh, it understands that the egg drops, it cracks. Right, right, right, right. And so I think that's what I'm kind of waiting for is the application layer of experience learning.
Jason Howell
And it could be a ways away.
Jeff Jarvis
And it's not going to be like generative AI because I don't think a token based world. This, I'm getting way out of my depth here, way out, folks. But I think this is part of what Yann Lecun told us in the wonderful interview, which if you haven't seen it yet, Jason will give you the link in a second. Second, is that, is that when you're just dealing with this abstraction of tokens, as I keep on saying, there's no meaning. Well, reality has meaning insofar as that's an egg and this is what its properties are and this is what can happen to it. And it has to associate it with that concept of egg. That, that's not the case in generative AI. It's not the case in machine learning as it stands now. Now it will be in robotics. Right hand has to shave, it's an egg. Don't punch too hard because it'll break.
Jason Howell
You push too hard, it breaks. Right, right.
Jeff Jarvis
So I won't do that again. I've just learned that about the egg or whatever, however it abstracts that notion of egg, you know, spheroid weight thing. And so this is all fascinating to me. I just love this next Part of it, but I don't know when it's going to get to our actual attention. Past theory.
Jason Howell
Yeah, yeah, well, yeah. And I think one, one thing that was kind of interesting to me that, I mean is, is probably just a different way of explaining what you were just talking about is that the current era that we are in, you know, we often talk about data scarcity, about the fact that these models are so hungry and they just need so much information to get smarter and smarter, but yet at the same time we've almost fed it almost everything we can at this point. The only way that they get, get better be, you know, leaps and bounds better into a kind of a new paradigm as you put it, is by learning these, these skills and these limitations themselves beyond just the information that they've been fed. Being able to kind of, you know, interact with the world and learn with, learn by doing in the real world and encountering things that, you know, the, the written word that travels over the Internet doesn't even, doesn't even describe properly for a system like this to truly understand it. Maybe, maybe it understands it conceptually, but it doesn't understand it from a lived felt sense, let's say, which is probably the wrong way to put it for a machine, you know, lived sense, but.
Jeff Jarvis
Well. Right, right, exactly. Yeah. Well, even learning is a troublesome word. But this paper at the end of it emphasizes mainly not robotics, but agents.
Jason Howell
Right.
Jeff Jarvis
And says that that's what in everyday life personalized assistance will leverage consistent, continuous, rather streams of experience to adapt to individuals, health, education, professional needs and long term goals. Perhaps most transformative will be the acceleration of scientific discovery. AI agents will autonomously design and conduct experiments. That's an interesting word there. It's constantly experimenting. What if I do this? What if I do that? That? Well, then it has to fail.
Jason Howell
Let me try again.
Jeff Jarvis
Right. That kind of, that's, that's learning in fields like material science, medicine, hardware design and so on. So agents, what I hadn't seen before is now I get a better understanding, I think, of why we see this rush to agentic AI. It's not just because it's the next thing. It's not just because it gives us cool things, it's the next training. Yes, that the agents are the way it learns.
Jason Howell
Totally.
Jeff Jarvis
And so it's a business model that every agent you have will add value to their machines, to their larger models.
Jason Howell
Yeah, it's probably a small representation of this, but like the example that pops into my head is, you know, we're thinking about Agents that go online, like, I need to buy those plane tickets and agent go do that. And so it goes onto the site and it goes through the standard methods and learns the website and everything. But it encounters an issue that it can't work around. Is the agent, does the agent stop there or does the agent, like humans often do, stop, you know, stop, pause and think like, okay, what, what is a way around this hurdle? How could I possibly, you know, get to this from a different perspective and work myself around it? And maybe that's a good, maybe that's a legitimate way and maybe it's an illegitimate way. But, you know, so from a human perspective, we don't just like, start, oh, face a hurdle, stop. We kind of think around these things and in so doing we teach ourselves alternative pathways and alternative ways to see and understand the world when those pathways work or when they don't. And yeah, I don't know how that connects with this exactly. It's just kind of what popped in my head is like, I think when I think of agents that do things in the current paradigm, it's like, did you buy the plane ticket? And I think maybe the agents down the road, a very easy challenge would be, no, I didn't. But here's how I figured out how to get around it and do it in this alternative method or way.
Jeff Jarvis
So two things on that. I love our discussions. This is fun because we go off into other things. So I found a paper yesterday that I didn't put up on the rundown because it was relevant to us though. Funny that now I see I'm trying to see if I can find it in my history that explained why it's so difficult to search plane, can't find it. Plane fares. And through all of these charts, it saw the level of how many flights there are possibilities there are from one city to another. And then all of these code variants and fare variances and then trying to compare them all. So to then tell an agent, it sounds like, oh, agent makes plane reservation. Well, it's incredibly complex. Right. And I think we're shorthanding all these tasks in life right now as if, well, yeah, the agent will do it, but we go with our judgment into what it is. The next thing you raised, which is interesting, is when you hit the barrier. And this happened when we talked to folks about art in this is when it does the wrong thing, is that good or bad? Is that a lesson learned, as we were just saying?
Jason Howell
Right, right.
Jeff Jarvis
And I think it was Rita Rayleigh from UCSB at this event this morning at William Mary, I think it was she who said that the, the creativity is leached out of the models because they've put, they've, they've modified it down so there's, there's no unpredictability because unpredictability is where you get to problems, hallucinations, all that kind of stuff, right? So you've got to leave in mistakes to learn. Right? So you've got to tell it to go off and buy the plane ticket and it doesn't find the plane ticket. And then it has to. That's, that's, that's part of the process. Learning, Learning is failing.
Jason Howell
And it's really absolutely, I mean, absolutely in the human experience. So much is, so much is learned through failure, even though it's incredibly uncomfortable. But that's part of the reason why you learn so much from it. It's profound.
Jeff Jarvis
Right?
Jason Howell
And yeah, so that's, that's necessary. And do we, do we as, as humans who have created this thing, do we have the patience for failure with these systems? And, and it's, you know, largely, it seems like people express that they don't because they continue to harp on AI systems that aren't 100% information accurate 100% of the time. They're just not going to be that way. Same as humans. Humans aren't either. We're patient with humans because we realize it's part of the human condition to be imperfect. But we aren't with the machine. And maybe we need to, maybe we need to give the machine a little bit more grace than we do right now.
Jeff Jarvis
Well, again, if it can't fail, it can't learn. If it can't fail, it can't get that experience. And so do we have that tolerance for that, for that failure? How do we build that in? Because I think we have this, this, this idea of the machine is a machine, so it can't make mistakes.
Jason Howell
Interesting stuff. Now this next one. Oh, you. And you put in another link here. Did you want to talk about it?
Jeff Jarvis
Only parenthetically is that, is that as a business insider gave Google credit for this thing that we just spent last time I was talking about. Similarly, ieee interestingly came in because Google often is said to be behind, behind OpenAI, behind others. IEEE came in and said Google succeeds with LLMs while Meta and OpenAI stumble. That's the first time I've really seen major credit being given by somebody of as much stature as IEEE saying that. Just talking about the model, just talking about the performance, I don't really want to go into any depth here, but it was interesting to see a slight vibe shift there. Google's getting some, some good juice here.
Jason Howell
There you go. You get what you go Google, you go Google this next one. Oh boy. Got thoughts on this one? A 21 year old former Columbia University student has raised $5.3 million in seed funding for his startup called Cluly. It's an AI tool designed to help users secretly, quote, cheat on everything. So exams, interviews, sales calls, first dates as shown by the verifiably creepy promotional video that they shared on X that I'm pretty sure only incels will find appealing. The app concept was born out of founder Chungin Li and co founder Neil Shanmgums. I'm sorry if I mispronounce your name. Their tool called Interview Coder that they developed while studying at Columbia University. Did they develop this for their work at Columbia University or was this on the side? Because they were. They were ultimately suspended from the university. And I couldn't figure out if.
Jeff Jarvis
I'm guessing that's a connection, but it's not clear.
Jason Howell
Yeah, it's not clear. But anyways, the app was designed to allow users to.
Jeff Jarvis
They were embroiled in disciplinary proceedings at Columbia over the AI tool tool, right? Then they both dropped evidence, dropped out. So.
Jason Howell
So did they create the tool on their own outside of the university or was it something that they created?
Jeff Jarvis
It began as a tool for developers to cheat on knowledge of leet code, a platform for coding questions that some in software engineering circles consider outdated and a waste of time. So maybe it was their way to just say yeah, you know, but, but this goes to the definition. What is cheating? Well, yeah, is it cheating or is a calculator. Right.
Jason Howell
And that's kind of part of what they're saying, right?
Jeff Jarvis
Story I tell in my, in my book that no one bought called Public Parts is that Mark Zuckerberg, when he was still in Harvard, there he had an art class and at the end the final on the class would have to be writing things about all of these pieces of art and renew that that. And so they would do study groups and so he organized a study group so that everybody was sharing the best of this. And the argument in the book that Zuckerberg made was that at the end everybody did better by using social, by, by not seeing as competitive by collaborating they all learned more and he had to study less. But, but he said that the grades for everyone in the class was up. So was that cheating or was that a smart use of social collaboration? Collaborative thinking. Right. Is it cheating to use the technology or is it a smart use of the. Of. Of technology? As an aid to you, I think we have to re examine the notion of cheat. What's. What, what does cheating mean? Is that merely is. Okay, this is interesting question I just asked myself, but I'll ask you too. Is cheating being unfair? Is cheating being.
Jason Howell
Yeah.
Jeff Jarvis
Opaque. What, what constitutes cheating?
Jason Howell
Yeah. Is cheating. Yeah. Because I mean, I think when I think of cheating in my older kind of school time paradigm, I think of this is a question that wants to know my knowledge of something. And instead of sharing my knowledge of something, I'm sharing what I've written down or what I, what I'm reciting or regurgitating from this thing in a, in a moment where I was expected to know it instead. But now, instead of knowing it.
Jeff Jarvis
Right, right, but now, now leave school. You have a similar task.
Jason Howell
Right?
Jeff Jarvis
Is it. If you get, if you get the answer you need, is that by any means?
Jason Howell
And does it, does it matter? Right. If you're tasked with a job and you're able to do the job, does it matter if you knew the answer or if you sought the answer?
Jeff Jarvis
Right.
Jason Howell
Yeah.
Jeff Jarvis
Now when it gets to dating, that is creepy because it's Cyrano de Bergerac. Am I really dating you or am I dating the app?
Jason Howell
I mean, that just felt like incredibly deceptive, that promo. The promo video is the guy is sitting at a table with, I don't know if it's a blind date or a first date with an attractive woman, of course, and she's asking him questions. And then you see his kind of like Terminator view coming up of the AI kind of coming up with the answers that he can feed to her. So he's essentially cheating on the questions that she's asking, lying about, you know, being. Being untruthful or dishonest about his age. When she asks and says, well, you look kind of young. Are you sure you're 29? And he's, you know, he's being fed all this information and then when she decides to walk out, then, then like the AI kicks in to like win her back. And so he rec that from a very heartfelt place and almost gets her to the point to where she finally realizes, I just need to get out of here and leaves. And it was just kind of like, I don't know. I don't think that does anything to endear me to what you're talking about because I do agree with what you're saying. There was a time when calculators were probably seen in the same.
Jeff Jarvis
Oh, they were.
Jason Howell
Same perspective. Spell check. I mean, for my preparation for these shows, often I'm using AI tools to research, which I would have had to do manually and by hand earlier. I would have to like do a Google search and find the stories and collect them, open them in many windows, read through, pull information. Instead of taking 20 minutes to do that, I can take five minutes or maybe even less and have it pull back those things. And so you could see that as cheating for these shows, but it doesn't mean that I don't synthesize the information and do something that. I mean, these shows are a prime example. Hopefully you get benefit and value out of it. And if you do, then it's just an example that it's kind of.
Jeff Jarvis
For those of you listening or watching, I hope you think that. Oh, good. Jason and Jeff read some stuff that I don't, I don't need to read now. Of course that pisses off media, hearing it said that way, but it's true. You don't have time to read everything. And maybe in some cases you say, oh, that's interesting to me, I'm gonna look it up. I want to learn more. But that's your choice. It's the same exact problem we get to with search and media right now and social media right now. Now is everything need not be the destination. So anyway, yeah, so these students are out. I say more power to them.
Jason Howell
I mean, yeah, I bet they've got a pathway here. I think this will be interesting to watch. Just drop the manipulative kind of aspect with dating and stuff, because.
Jeff Jarvis
All right, how about the other example they give? The main example they give is sales calls. Is that bad? Only. Only if you get lied to.
Jason Howell
Yeah, it's. Yeah, I suppose it's bad if it's dishonest, but if it's not and it.
Jeff Jarvis
And it. And it's targeted to what my needs are and it sells me what I want.
Jason Howell
Yeah, totally. If, if I'm a sales agent, I'm going to go through training in order to effectively sell and effectively say the right things and effectively not say the wrong things and recognize cues and all this kind of stuff. If there's a tool that enables me to do that part of my job better, I don't see anything wrong with it.
Jeff Jarvis
I mean, the key to all sales things, there's a guy named Jeffrey Gittamer who writes sales books, like my first book. And so I watched how this operates, and same as in what I teach in journalism. It's listening. It's listening to people understanding what their needs are, empathizing with those needs and trying to come up with solutions for those needs. And if your solution is in fact legitimate and good, you make a sale.
Jason Howell
There you go. Right?
Jeff Jarvis
That's okay. In fact, so, so we hear a lot about how this is going to come to customer service and phone, mail, jail and all the hell we go through. Right? So the, the agent is reading the script, you know, get off the damn script. And the fear is that AI will be even worse than that. But it may be far better than that. It may understand my, my need better. It may be more responsive to that need. It may be able to get to a solution faster.
Jason Howell
I was going to say maybe faster.
Jeff Jarvis
Only if the AI is giving, is given the true agentic power, if it has agency to do so.
Jason Howell
Yes. Yeah, yeah. Very interesting. Let's take a super quick break. Then we got a few more stories to round things out, including Oscars. Kind of becoming a little bit more welcoming to AI. Trust isn't just earned, it's demanded. And whether you're a startup founder navigating your first audit or a seasoned security professional scaling your GRC program, proving your commitment to security has never been more critical or more complex. That's where Vanta comes in. Businesses use Vanta to establish trust by automating compliance needs across over 35 frameworks like SoC2 and ISO 27001, centralize security workflows, complete questionnaires up to five times faster, and proactively manage vendor risk. Vanta not only saves you time, it can also save you Money. A new IDC white paper found that Vanta customers achieve $535,000 per year in benefits, and the platform pays for itself in just three months. Join over 9,000 global companies like Atlassian, Cora and Factory who use Vanta to manage risk and prove security in real time. For a limited time, Our audience gets $1,000 off vanta@vanta.comaiinside that's V A N T A.comai inside for $1,000 off.
Jeff Jarvis
This episode is brought to you by State Farm. You might say all kinds of stuff when things go wrong, but these are the words you to remember. Like a good neighbor, State Farm is there. They've got options to fit your unique insurance needs, meaning you can talk to your agent to choose the coverage you need. Have coverage options to protect the things you value most, File a claim right on the State Farm mobile app, and even reach a real person when you need to talk to someone like a good neighbor, State Farm is there.
Jason Howell
All right. The Academy of Motion Picture Arts and Sciences officially updated its rules to allow films that are using generative AI to compete for Oscars. So basically coming out with an official stance to say, hey, you know what? Just because AI tools were used, which, by the way, it's. I mean, it's. It's. If it hasn't, you know, just overtaken or at least, you know, highly influenced how these movies are made, it's going to. In a very swift, swift move. But this just ensures that. But they're basically saying, like, it's okay as long as there's a human involved. They say they do emphasize that the films where human creativity and human involvement are central will be more heavily considered. Heavily considered. So not like requirement. But the filmmakers do not have to disclose the use of AI that had been considered as one thing, and that's not the case here. So basically, they're saying, at the end of the day, what we've been talking about, AI is just another tool, and yes, you can use it. Just be responsible, and hopefully you've got humans also doing things on these things, too.
Jeff Jarvis
You can still win a Pulitzer if you use a typewriter.
Jason Howell
Yeah, right. Hey, that's a great example. Yeah.
Jeff Jarvis
Yeah. When word processing came in, it wasn't to the level of moral panic, but it was some fear that somehow this was too easy. Somehow this was going to change things. And it does. It changed the way I wrote because I wrote the old typewriter did days. So it did. It changed immensely. It made it easier. It made it faster. It made. It gave me more power.
Jason Howell
Lowers the. The barrier. It levels the playing. It lets too many people in.
Jeff Jarvis
Yeah, right.
Jason Howell
It doesn't quite gatekeep the way we used to have it.
Jeff Jarvis
Bingo. Bingo. Yeah. Yep.
Jason Howell
Yeah.
Jeff Jarvis
I'm a Howard Stern fan, and he complained when podcasts started, I think. I think I had this argument with him once on. On the air. Oh, podcast is nothing. You got to learn radio. You got to work your way up. Just full old fart right now. You see, the Joe Rogan's of the world are huge, and even he has to admit that. Okay, well, yeah, they. They're there.
Jason Howell
Yeah. Yeah. Howard Stern. Is he. Is he still rocking? I haven't listened to his show.
Jeff Jarvis
He's on serious. He's got to pay.
Jason Howell
Yeah.
Jeff Jarvis
But, yeah, he's. He's a. He's become an amazing interviewer. Speaking of.
Jason Howell
Yeah, yeah. Cool. Yeah. I used to really be into his show. Used to love it. I need to check it out again. And then finally, yeah, okay, we're, we're back to OpenAI. But I thought this was a good.
Jeff Jarvis
Way to leave back.
Jason Howell
They begin and end with OpenAI these days, but I thought this was a good way to kind of round out the show. CEO Sam Altman shared that users say please and thank you to chatgpt as we know we've talked about it before, and this results in tens of millions of dollars in operational costs. Says there are significant energy costs to processing every word that is typed into a chatbot. Of course, please and thank you are also words that enter in. Help himself in saying it's still a good idea to be nice because you just never know. Someday the robot might have mercy on your soul. Couldn't help but kind of get that.
Jeff Jarvis
That how many useless words? All. You know, there's a paradox of text as I'm writing this book about the Linotype. And if you go back to the difficulty of writing in the past, whether it was by. By Scribal Quill, right. Or by setting type one letter at a time, all that was really laborious. Yet people were very long winded then. And we get to this age of the Internet and especially things like Twitter, where we could go on as long as we want, and suddenly we come up with new ways to be as economical with our language as we can be. It's just kind of interesting to me. So on the one hand, I think that we were used to using the least words possible for both Twitter and Google search. And now AI comes along and says, no, say more. But every time you say more, it costs money, it costs energy.
Jason Howell
I mean, it all costs money. I think people are dumping incredible quantities of data into their LLMs, you know, and a short one or two letter nicety is, is. Is not moving the needle here. I mean, I guess, you know, in the, in the sense that everything at this scale adds up to some large number, but large number by comparison to the actual large number. That is the overall cost of all words and everything. It's just a. I mean, it's a spec. It's a grain of sand.
Jeff Jarvis
Yeah. And there's new efficiencies to be made. I remember when search and web came up with caching. That was a big deal. Saved effort. Speaking of which, there was a story that didn't make it the rundown. I just want to mention real quickly because we talked about this a few weeks ago, where sites are being driven mad by AI Scrapes, scrapers coming in and costing them a huge amount of bandwidth. And so Wikipedia, the Wikimedia foundation, finally said, oh, to heck with this. And so they've put up a 461,000 freely accessible data sets here. Don't scrape us. Go there.
Jason Howell
Take it. Okay. Yeah, we talked about this not too long ago on the show.
Jeff Jarvis
I was talking about how if news and other sites did this and said, here, just take it. It's okay.
Jason Howell
Here it is.
Jeff Jarvis
But stop scraping me every day because it's costing me money. And I think this is, as so often the case. Wikimedia foundation is ahead of the rest in thinking smart about this technology.
Jason Howell
Don't scrape me, bro.
Jeff Jarvis
So you too can go get that data on Kaggle. Is it Kaggle or Kaggle? I guess two G's. I'd guess Kaggle.
Jason Howell
I would say probably Kaggle. It appears to me that it's Kaggle, but who the heck knows? Interesting. Cool. Well, we have reached the end of this episode of AI Inside. Jeff Jarvis, thank you so much for being with me for another hour of getting smarter on artificial intelligence and everything in between. The Web We Weave is a wonderful book that everybody should read. You can go to jeffjarvis.com to find that. That. The Gutenberg Parenthesis magazine. Yes. And magazine. You cannot find Public Parts here, though. You. You said that. Was that. Nobody read that. I didn't.
Jeff Jarvis
Yeah, it's. You can probably find it on ebay. I don't know. Let's see if you go to Amazon. His Public Parts was, of course, my Howard Stern joke because he wrote Private Parts.
Jason Howell
Yes. Yes, indeed. Okay.
Jeff Jarvis
You can still get the audiobook.
Jason Howell
Yeah, let's see. I got it.
Jeff Jarvis
You got it.
Jason Howell
There you go. Yeah, yeah. Hardcover, $6. Paperback, $12. Yeah, these are used. You can get it on audiobook, of course.
Jeff Jarvis
Prime. No.
Jason Howell
Maybe.
Jeff Jarvis
Yeah, no, it's.
Jason Howell
There you go. If you want to go deep into the catacombs of Jeff's work, you can.
Jeff Jarvis
Cool.
Jason Howell
And this is from 2011, so, yeah. Hey, you've been writing a lot of books for a long time. It's worth mentioning your whole catalog vlog.
Jeff Jarvis
From time to time.
Jason Howell
Thank you, Jeff. So much fun.
Jeff Jarvis
Jason.
Jason Howell
Always big time. Thank you to. To everybody for visiting the site, of course. Where you can go to, you know, find all the ways to subscribe to this show, AI Inside Show. And then, of course, there is the Patreon. Patreon.com AI inside show. And I will just go ahead and throw that up on the screen along with our amazing executive producers, Dr. D.O. jeffrey Maracini, WPVM 103.7 in Asheville, North Carolina, Dante St. James, Bono Derick and Jason Niefer. By the way, he corrected me on how to say his name and Jason Brady are amazing, amazing patrons that support us on a deeper level as executive producers. So patreon.com AI inside show. But I think that's about it, y'all. Thank you so much. Thank you again, Jeff. It's a lot of fun. We'll see everybody next time on another episode of AI Inside. Bye, everybody.
AI Inside – Episode 65: The Era of Experience
Released April 23, 2025
Hosts: Jason Howell & Jeff Jarvis
Produced by: Yellowgold Studios
Introduction
In Episode 65 of AI Inside, titled "The Era of Experience," hosts Jason Howell and Jeff Jarvis delve into the evolving landscape of artificial intelligence, exploring its integration across various sectors, regulatory challenges, and ethical considerations. The episode navigates through pivotal news stories, insightful discussions, and expert opinions, providing listeners with a comprehensive understanding of AI's current trajectory and future prospects.
1. Google's Antitrust Trial and Its Implications
The episode opens with a discussion on Google's ongoing antitrust trial. Jason Howell highlights the Department of Justice's recommendation for Google to divest its Chrome browser as a punitive measure for unlawful practices in online search and advertising.
Jeff Jarvis expresses skepticism about OpenAI's announced interest in acquiring Chrome, likening it to a publicity stunt.
The hosts debate the significance of the browser in today's AI-driven ecosystem, contemplating whether browsers remain the central hub for user interactions or if agentic AI could redefine this role.
2. The Future of Browsers in the AI Landscape
Jeff elaborates on the historical importance of browsers, referencing the early days of Netscape and Microsoft's browser wars. He questions whether the browser's role is diminishing as AI agents become more integrated.
Jason counters by suggesting that despite the potential for OpenAI to develop its own browser, the intricacies involved might prevent such a move.
3. AI Integration in Mobile Devices: Assistants vs. Browsers vs. AI Apps
The conversation shifts to how users interact with AI on their smartphones. Jason shares his personal usage patterns, indicating a preference for launching specific AI apps like Perplexity over integrated assistants like Google's Gemini.
Jeff contrasts this with his reliance on browsers for most tasks, highlighting generational differences in AI adoption.
They discuss the "stickiness" of AI models with user data, emphasizing how personalized experiences can lock users into specific ecosystems.
4. AI in Education: Government Integration and Ethical Concerns
A significant portion of the episode addresses the Trump administration's draft executive order proposing the integration of AI into K-12 education. Jason Howell expresses mixed feelings, recognizing the necessity of embracing AI while cautioning against hasty implementation.
Jeff shares insights from academia, noting concerns about the readiness of educational institutions to adopt AI responsibly.
They advocate for a balanced approach that leverages AI's benefits without undermining educational integrity.
5. Insights from Demis Hassabis on 60 Minutes
Jason and Jeff review Demis Hassabis's appearance on 60 Minutes, where the DeepMind CEO discusses the timeline for Artificial General Intelligence (AGI) and AI's potential to cure diseases.
Jeff criticizes the overly optimistic projections, likening them to past technological solutionism.
They acknowledge the impressive achievements of AI, such as DeepMind's AlphaFold, while urging for realistic expectations.
6. Google's Performance in LLMs vs. OpenAI: IEEE's Recognition
The hosts highlight a Business Insider article noting that Google is outperforming OpenAI in large language models (LLMs), with recognition from prestigious institutions like IEEE.
This marks a shift in the AI competitive landscape, acknowledging Google's advancements in the field.
7. The Ethics of AI in Cheating: Startup Cluly's $5.3M Funding
Jason introduces the story of Cluly, a startup aiming to help users "secretly cheat" in various scenarios, raising $5.3 million in seed funding. The episode delves into the ethical debate surrounding AI-assisted cheating.
The hosts discuss the evolving definition of cheating in the age of AI, questioning its implications in education and professional settings.
They explore whether using AI tools like calculators parallels the use of AI for information retrieval, emphasizing the need to redefine academic integrity.
8. AI and the Oscars: Generative AI in Filmmaking
The Academy of Motion Picture Arts and Sciences has updated its rules to allow films utilizing generative AI to compete for Oscars, provided human creativity remains central.
Jeff draws parallels to past technological shifts, like the transition from typewriters to word processing, suggesting AI will similarly transform filmmaking without diminishing human artistry.
9. OpenAI Operational Challenges: The Cost of Politeness
The episode concludes with a revelation from OpenAI's CEO, Sam Altman, who disclosed that users frequently saying "please" and "thank you" to ChatGPT contribute to substantial operational and energy costs.
Jeff comments on the linguistic shift prompted by AI interactions, noting the paradox between economical language use in the digital age and AI's demand for more verbose inputs.
Conclusion
In "The Era of Experience," Jason Howell and Jeff Jarvis offer a nuanced exploration of AI's multifaceted impact on technology, society, and ethics. From regulatory battles and educational reforms to ethical dilemmas and operational challenges, the episode underscores the complexity of integrating AI into various aspects of human life. The hosts advocate for responsible and informed adoption, balancing innovation with ethical considerations to navigate the rapidly evolving AI landscape.
Notable Quotes:
Jeff Jarvis [04:05]: "I trust Google with Chrome a lot more than I would trust OpenAI."
Jason Howell [17:04]: "I'm almost more inclined to launch the app, honestly, to do that either with Gemini or with Perplexity."
Jeff Jarvis [29:39]: "This is a turbocharged view of technological solutionism... this is 10 times hyper."
Jeff Jarvis [66:02]: "You can still win a Pulitzer if you use a typewriter."
Jeff Jarvis [68:03]: "Every time you say more, it costs money, it costs energy."
Further Resources:
Jeff Jarvis's Works: For those interested in exploring Jeff Jarvis's extensive contributions to media and technology, visit jeffjarvis.com.
Patreon Support: Support AI Inside by becoming a patron at patreon.com/aiinsideshow.
Thank you for tuning into AI Inside. Stay informed and engaged with the latest in artificial intelligence and its profound impact on our world.