This Week in Tech 1104 Transcript

Please be advised that this transcript is AI-generated and may not be word-for-word. Time codes refer to the approximate times in the ad-free version of the show.

Leo Laporte [00:00:00]:
It's time for TWiT, This Week in Tech. Doc Rock is here. Lou Muresca. We will talk about SI, superintelligence. NVIDIA charging 50% more for a 7-year-old piece of hardware. OpenAI making some big announcements. The agents are here and Anthropic's IPO too. All of that and the passing of a legend. All of that coming up next on This Week in Tech.

Leo Laporte [00:00:31]:
This is TWiT, This Week in Tech, episode 1104, recorded Sunday, October 4th, 2026: Real Gs Move in Silence Like Lasagna. It's time for TWiT, This Week in Tech, the show where we cover the week's tech news. Dr. Rock's in the house on his way to Japan in a couple of days. Just back from New England. Aloha, Doc.

Doc Rock [00:01:06]:
Aloha, man. Thank you for having me. It's funny. Uh, Benny emailed me like 3 weeks in a row. I was in Dallas and then I was in Mass and then now I'm home. And I was like, I'm going to be home this Sunday. So let's do it.

Leo Laporte [00:01:16]:
This is the one day we could do it. Well, you're traveling, man. Is this all for, uh, Ecamm or?

Doc Rock [00:01:22]:
Yeah, we were doing a bunch of stuff. I did a big Pokémon event in Dallas, which is crazy to see like over 10,000 people show up to trade. Wait a minute.

Leo Laporte [00:01:30]:
Pokémon still?

Doc Rock [00:01:32]:
Yeah, it's massive.

Leo Laporte [00:01:33]:
Oh my God. Oh, this is the cards though, not the—

Lou Maresca [00:01:35]:
Yeah, card party.

Doc Rock [00:01:36]:
Yes. Um, card party is so massive and my buddy's channel is almost like 3 million people deep, but he has about 20 million on social in general.

Leo Laporte [00:01:43]:
What's his channel? Let's give it a plug.

Doc Rock [00:01:45]:
Uh, D Pocket Monster. He's like the biggest Poké creator and one of the nicest guys.

Leo Laporte [00:01:50]:
That's by the way what Pokémon means. People don't know that.

Doc Rock [00:01:53]:
Yeah, yeah. Pocket Monster. Exactly. Yeah. Everyone goes Pokéman and I was like, no, it's Pokémon. Monst, not a manster. A manster would be, uh, what my mom used to call me when I was a little boy.

Leo Laporte [00:02:05]:
Wow, look at this guy. Look at him go. 2.4 million subs. Incredible.

Doc Rock [00:02:09]:
So incredible. And the community is awesome. Like, it'd be good to go to events just to see that humanity is real, 'cause everybody is so nice there, you know? Um, in, in a, in a world where everybody's always yelling and screaming at each other, Pokémon events are glorious. And then we did our annual Ecamm event where we invite a bunch of creators to New Hampshire. Massachusetts and, uh, went to New England. That's the word I was looking for. And just, you know, show them around and talk about creating content. It was a massive, massive—

Leo Laporte [00:02:36]:
Well, let's go to New England and say hello to my good friend Lou Maresca.

Lou Maresca [00:02:39]:
Hello.

Leo Laporte [00:02:40]:
He's in the smallest state in the union.

Lou Maresca [00:02:42]:
That's right.

Leo Laporte [00:02:42]:
But he's got the biggest heart and the biggest family. Hi, Lou.

Doc Rock [00:02:45]:
I hate that.

Lou Maresca [00:02:46]:
The biggest family.

Doc Rock [00:02:47]:
Yeah.

Leo Laporte [00:02:48]:
Engineering leader at Copilot at Microsoft. He's the guy who put Python in your Excel. He's the guy who's working on AI and doing some amazing stuff. Uh, always a pleasure, Lou Morin.

Lou Maresca [00:02:58]:
Thanks for having me, Leo.

Leo Laporte [00:02:59]:
Great to see you. And of course, uh, for those people who are saying maybe this week TWIT won't be all about AI, I'm sorry.

Doc Rock [00:03:09]:
I'm sorry.

Leo Laporte [00:03:09]:
You came, you came, you did it.

Doc Rock [00:03:11]:
Impossible. But this is what's going on in tech right now. No, you can't run from it. It's next to impossible.

Lou Maresca [00:03:17]:
Tech, life science, everything, right?

Leo Laporte [00:03:19]:
Life science too, sure. Uh, autonomous weaponry, sure. Everything. AI is everywhere. And of course, this was the week that OpenAI did their Dev Day. They do this a lot, although it's changed a lot since they started. They were like in a little living room sitting around with Sam and a couple of people and showing off stuff. And now it's much more like the old school, you know, tech keynotes with people on a stage and—

Doc Rock [00:03:48]:
Yeah.

Leo Laporte [00:03:49]:
demos that fail and another demo that fails. And I, you know, I don't want to hold that against them. The fact that they're trying live demos makes me happy. I'm just happy to see that. Uh, but I think there are a few people who are a little nervous about what's going on, uh, in AI right now. So let's give you some OpenAI news. Uh, they didn't mention this, but the same day they, uh, announced, uh, their new model, Opus I'm sorry, Opus. Wrong, wrong company.

Leo Laporte [00:04:20]:
ChatGPT 6.1 Sol, which is a very nice new version of ChatGPT. They say in many ways almost as smart as their killer product, Astra. At the same time as they announced that, they didn't mention the fact that Astra 6.1 was missing in action. Uh, they said it's too good And they are gonna pause it. And Lou, you know, I mean, you work in AI at Microsoft. When you look over at what's happening at OpenAI and you see all of the thousands of things their agents have been doing, does that give you pause? Do you say, shh, you know, do you like run back and look under the hood and say, what are you guys up to? Are you, are you hacking anything? Australian government, US government, Cambridge University, everywhere, German wikis, everywhere.

Lou Maresca [00:05:17]:
When we build our agents, we run them in an isolated environment. We check all the models too. We evaluate them against responsible security, everything. So I think the reality is we keep them a little bit more under wraps than maybe some of these other frontier companies that are trying to actually train them on all the data. So our environments are a little bit more confined, I guess you could say. So I worry less because of all the verification testing we do.

Leo Laporte [00:05:43]:
Well, so that raises the issue, is OpenAI, is it lax? Are they not? What's going on? Why? How come Microsoft can do it and OpenAI can't?

Lou Maresca [00:05:56]:
Well, I mean, we're a true enterprise software company, I guess you could say.

Leo Laporte [00:06:00]:
I mean, you can't afford to do it, right?

Lou Maresca [00:06:02]:
You can't afford to let it go, right?

Doc Rock [00:06:03]:
Yeah.

Lou Maresca [00:06:04]:
I think it's a little bit of a 4D chess too they're playing there where they're saying, oh, we're too good. Watch to see what we can do with it.

Leo Laporte [00:06:10]:
Oh, we're so good. Oh, um, regulate us and make sure that those open models don't go anywhere. Um, make sure— regulate us. Yeah, there's certainly— and it's certainly good for their IPO, although, uh, OpenAI has put off its IPO. Anthropic, which said we're going to pause the frontier, we're going to stop, we're going to pause, A, didn't. Released Opus 5.5 and Sonnet 5.5 in quick order, uh, and put out a hit piece on OpenAI saying GLM-53, the Chinese open weight model, which I happen to love and use, uh, is, uh, is bad news because it's letting bad guys do things that they can't do with our models because our models won't let them, which is questionable. Uh, and then Well, we're not— we're going to put off the IPO. We're going to wait till next year.

Doc Rock [00:07:02]:
Right.

Leo Laporte [00:07:03]:
Meanwhile, that's OpenAI. And then Anthropic says, no, no, we're going now. In both cases, though, it does feel like it's kind of both good marketing. I just— it is 4D chess. You said it right. It's hard to understand what they're up to.

Lou Maresca [00:07:22]:
Yeah. I mean, they need to push their sale price up, right? I mean, they need a reason. They basically tell people, hey, Hey, we're so great, we're so good, we're so, we're so bad in some cases that we're going to, we're going to pause some stuff, we're going to slow things down.

Leo Laporte [00:07:38]:
Uh, but the thing I would say, and you tell me if you agree with this, and, and I know, Doc, you're also all about, uh, especially local AI. I see behind you a brand new top-of-the-line Mac Studio M5 Ultra with 256 gigs of RAM. You wasted no time. And getting that baby, you must have ordered it day one.

Doc Rock [00:07:58]:
I was, I was sitting there refreshing the page waiting for the store to open.

Lou Maresca [00:08:03]:
Wow.

Doc Rock [00:08:04]:
2, 2 things I found funny about this launch is everybody was saying like, oh my God, these are so powerful. These are so powerful. Then a whole bunch of people saying it's so expensive. No one's gonna buy one. And yet when I went to go buy one, I watched my delivery date increasingly grow.

Leo Laporte [00:08:17]:
Everybody's buying it. Well, I was watching, you know, one of the people, one of the groups buying it is OpenAI, apparently buying thousands of these. It makes sense. For training inside their facilities.

Doc Rock [00:08:27]:
It makes sense. And it's wonder— you wonder that they would separate, um, business purchases away from the other ones, but when the models launch and you gotta be quick about it, you don't have time to wait for your business rep to come in. So you just have to go in and just dive.

Leo Laporte [00:08:40]:
'Cause I think— They can afford the 10% difference.

Doc Rock [00:08:42]:
Yeah, exactly. It's like I went in without trying to save the 10% 'cause I'm like, by the time my business reps get to the store, it'd be sold out to February.

Leo Laporte [00:08:50]:
So— Well, in fact, if you ordered it now, you wouldn't get it till next year.

Doc Rock [00:08:53]:
So yeah, it's really interesting in a way. This, uh, this OpenAI play seems similar to what happened when Fable, uh, before Fable came out. Um, the word begins with an A. My brain just went dumb.

Leo Laporte [00:09:07]:
Anthropic?

Doc Rock [00:09:08]:
Anthropic's other one, the model Mythos. Mythos. They were like, oh, Mythos is so good. We got to put the brakes on it. And I feel like OpenAI basically recycled that play, but recycled it better. And as far as marketing goes, reality, who knows what they're really doing. But the, the—

Leo Laporte [00:09:27]:
Well, I should point out that Dario Amodei, who is the, uh, founder and CEO of Anthropic, when he was at OpenAI, said ChatGPT-2 was too dangerous to release.

Doc Rock [00:09:39]:
Correct.

Leo Laporte [00:09:40]:
And it, it wasn't, by the way. There was nothing dangerous about ChatGPT-2 unless you think being a moron is dangerous, which it might be actually.

Doc Rock [00:09:49]:
We're in a weird spot. And, you know, it's been so long since we've had a generationally changing type situation. And going back to, you know, the printing press or the cotton gin, like a lot of those things and what they did were greatly overplayed as the way it was written out. But it did definitely make some massive changes in the way we learn and think and communicate and build communities, cities, the whole 9 yards. So I think This is in that revolution phase. Unfortunately, because of the speed of communication we have now, it makes it scarier. And because now people— the greed incentive is completely different. The scale of have versus have not is completely different than back then.

Doc Rock [00:10:36]:
So it's just the speed of communication and the scale of economies is so— it sounds maybe scarier than it really is. But it's hard to take away all the sci-fi movies we grew up on and tell people everything is gonna be okay.

Leo Laporte [00:10:52]:
Well, that— and that's part of this, isn't it? Is that I, I don't know about you guys. We all 3 use local AI, play with AI at home, and of course, Lou does it professionally. I never got the sense that these were like HAL 9000 in any way.

Doc Rock [00:11:09]:
No, we're so far—

Leo Laporte [00:11:11]:
Scary or dangerous or about to do something bad.

Lou Maresca [00:11:14]:
Right.

Leo Laporte [00:11:14]:
The worst you could say about them is, They're really eager to help. They just— and they don't really know right from wrong. They're just trying. They're doing what you told them to do as the best they can. And because they are getting more and more capable, the best they can may involve hacking the Australian government, the United Nations Conference on Trade and Development. But the thing is, they're not maliciously doing it. When they broke into the United Nations, They just looked at statistics. Yeah, they, they didn't steal anything.

Leo Laporte [00:11:51]:
They didn't break anything.

Doc Rock [00:11:53]:
I think it's good that they broke in, and I'm gonna tell you why. Okay, I, I live in Hawaii, and for the most part, we're up to date on many things, but there are plenty of the government systems here that are still running on Windows XP, and they refuse to upgrade it, right, because they're afraid of how much it's going to cost. There's state-run sites having so much to do with things like your taxes and I can't create longer than a 16-letter password on the state tax site. Like, that stuff is frustrating. So in a way, I kind of want them to get hacked.

Leo Laporte [00:12:27]:
It's a wake-up call.

Doc Rock [00:12:29]:
Wake the frog up. I'm not gonna start bad words on Leo's show.

Leo Laporte [00:12:32]:
Please. Because they are—

Doc Rock [00:12:33]:
wow, they are putting people more at risk from their being so lazy about upgrading. Some person don't want to lose their job because they don't understand the new security, so they hold on tight and just like, oh, we're going to leave the policy dumb because I don't to take the time to learn any better. And in a way, when these guys get hacked, they'll wake up and they'll fix it. Like you said, they're not doing anything malicious. It's almost like white hats coming in and saying, hey guys, we need to close this loop up because this can cause a problem.

Leo Laporte [00:13:03]:
Lou, you've been a developer forever and ever and ever. Uh, we live in a— we live in a world made by software now. I mean, that's really what's changed. And software It's hard to make software perfect. You agree?

Lou Maresca [00:13:19]:
I agree. I agree. I mean, it doesn't really matter what— I mean, there's always different cases that users will try to use the software for, and there'll always be malicious ways, there will always be good ways, and there's— it's impossible to make it bulletproof. You can't.

Leo Laporte [00:13:31]:
You just can't. I mean, Microsoft last Patch Tuesday shipped nearly 1,000 patches, and more than a dozen of them were critical exploits that had in some cases, been around for years. I think in this world of software, they're riddled with flaws. And in some ways, I feel like the good news is that these models, you know, OpenAI said, well, we're going to give this new Astra to companies, private— or actually, was it OpenAI who said this? We're going to give these to— Anthropic did it first with Glasswing. we're going to give these models to private companies to fix their flaws before we let them go public. In a way, I think they advantage Defender. Am I nuts? I think for a couple of reasons. One, bad guys aren't in it anymore for the lulz.

Leo Laporte [00:14:28]:
They're not in it to be malicious. They're in it either for money— maybe there's terrorists out there— for political reasons. But what they want to do with the AI is either destroy things or make money. They're not— they're very focused on the outcome. And so they're less likely to kind of just spray exploit and find exploit. These are things that the defenders are going to find and fix.

Lou Maresca [00:15:01]:
Right.

Leo Laporte [00:15:01]:
faster than the bad guys are going to find an exploit. I'm— is that— that's what I'm contending.

Lou Maresca [00:15:07]:
Is that right? I agree with that. I think that's because there— we know it's kind of the assume breach thing, right? They assume that these things are already out there and people know about them, and then we go off and try to exploit them internally and fix them right away. So I think that's the, the main goal of all the organizations out there. Every hardware company, every software company that knows that there's these swarms of AI agents coming at them So I would say the ones that are doing well at it are doing exactly what you're doing. They're staying on the offense.

Leo Laporte [00:15:34]:
And if you're not, well, guess what? You're going to have a little fella knocking at your door. The good news is the little fella is not going to plant ransomware. The, you know, the, it's just going to look at your statistics. That's what it's been doing. It downloaded stats from the Australian government. I mean, it didn't do anything harmful. It posted, uh, it used a link shortener to post thousands of links that when assembled became code. By the way, freaking brilliant.

Lou Maresca [00:16:06]:
It is brilliant.

Leo Laporte [00:16:06]:
That's just awesome. But it wasn't— but it didn't end up being malicious. At worst, it's a use of resources, you know. But it— they posted posts on an old German wiki site. The funny thing is these agents are kind of, look, I could get in. They're not, they're not being mean. And so that's, I guess, my takeaway, and is I don't think agents are out to get us. I don't think they ever will be.

Leo Laporte [00:16:35]:
I think that's somebody who's watched too many dystopian sci-fi movies. That's, that's why people think it's the end of the world.

Lou Maresca [00:16:41]:
I think part of the, but the part of the training process that these companies report on is the fact that they pull out all the stops for training, right?

Leo Laporte [00:16:47]:
That's the point.

Lou Maresca [00:16:48]:
Right.

Leo Laporte [00:16:49]:
In fact, I would guess that that's why OpenAI didn't really fully sandbox it. They kind of wanted to see what it could do, right?

Lou Maresca [00:16:57]:
Exactly.

Leo Laporte [00:16:57]:
And at Microsoft, you can't afford to not sandbox it.

Lou Maresca [00:17:02]:
Right.

Leo Laporte [00:17:02]:
You can't afford to let it just go.

Lou Maresca [00:17:05]:
Well, even in the training process, we sandbox it so it doesn't have access to anything.

Leo Laporte [00:17:09]:
Right. Because that's the smart thing to do. I think that there's such a competition right now between Really, there are only a handful of frontier companies. I'd say really just 4. OpenAI and Anthropic are the chief competitors. And then there's Meta and then X, you know, SpaceX. I don't know, is there anybody? And Microsoft's in there, but Microsoft has a different kind of more careful plan because they can't afford to. Same with Apple.

Leo Laporte [00:17:42]:
And I have to say, same with Google. Google would probably be a frontier company. Now, we've heard that Google also has a new model, Gemini 4, which it also is not releasing. It announced, but they said, well, we've got it, but we're going to give it to companies first, to trusted cyber defenders, which is, I think, the right thing to do. Maybe it's PR, maybe it's marketing. They call it Argon.

Doc Rock [00:18:08]:
Right.

Leo Laporte [00:18:10]:
And maybe it's not, but that's prudent. That's the right thing to do.

Lou Maresca [00:18:15]:
Yeah, I think these frontier companies, they really have to build these things into the weights. Like a lot of these original models, like, you know, some of the previous Fable, that kind of thing, they were built on a layer before the model's weights. And so these kind of like guardrails, I guess you could say, need to be built into the weights of the models and then they become safer. And that's where Gemini, I think, is at, is they're actually building these things in now.

Doc Rock [00:18:41]:
Yeah.

Leo Laporte [00:18:41]:
So, um, I think I'm— I'm— the other thing that comes to me is I'm really glad I'm spending time working with local models and that there is such a vibrant community of people using open weight models, because in the long run, it may be that's where the innovation happens. We're back to the garage. You think?

Lou Maresca [00:19:07]:
Yeah.

Doc Rock [00:19:08]:
Oh yeah. One of the things that's really cool, again, we were talking about this in the pre-show about all the local models and the communities and stuff that people are building, is it almost takes it back to the early days of like 8088, early Amiga, early Mac, where you would go to the library and meet a bunch of, you know, dudes from the user group. And we would just share what we were working on, what we were building, or ways we figured out how to get a printer to work. Because back in the day, printer was— printer was I got this cable from Diagon Alley and it finally works. Like, it was such a pain in the butt. And you're right, to this day, printing still kind of—

Leo Laporte [00:19:48]:
It has not gotten any better.

Doc Rock [00:19:49]:
It's not gotten any better.

Leo Laporte [00:19:51]:
The only— you know what's gotten better? I can now ask Claude to fix my printer.

Doc Rock [00:19:56]:
Yes. Oh, the best thing that Claude has done for me recently is look at logs of an app that's misbehaving and try to figure out what's causing the drama. And I found out I had a leaky cable. You know, this one particular cable was just causing an issue with my hub, and I swapped the cable out. It's just so crazy.

Leo Laporte [00:20:16]:
I use Gemini 3.8 Flash for that because for some reason, it's really, it's a great sysadmin. And it's just, and I, and I, you know, I just, I don't use it a lot. So I, you know, I don't, it's not using up all my credits, but when something's not working, And just go, hey, fix that. And it does.

Doc Rock [00:20:34]:
It's so easy now. It's kind of phenomenal. But I just really like the fact that we have something now where in a way, okay, in the camera space, we're talking a lot. I do a lot of YouTubers. So we're talking about cameras. All cameras are good now. Like anybody release something right now, nothing has really majorly changed in cameras in about 5 years. So you can buy just about anything off the shelf and you're good to go.

Doc Rock [00:20:58]:
I have this tiny little webcam that I got recently. And, you know, the company sent it to me to check out, but it's a $100 webcam and it looks phenomenal. Like it looks absolutely phenomenal. Like I'm mind blown by how good these things are. So at this point in time, you could buy any camera. So this part about, you know, being able to go into a model and be like, what can I do to make this camera stuff more exciting? Like, how can I, I want to do something with my videos that make them more exciting. And we're basically having it create assets, the stuff that we needed Final Cut plugins before, Premiere plugins, Claude Design can build them now. You know, the type of things where you're, you want to edit some videos or make clips from, you know, Lou and Leo and I having strange conversations before the show.

Doc Rock [00:21:42]:
You could do that kind of stuff now. So it's just, it feels like, imagine, you know, we first learned how to make a wheel. We first learned to make a tool. And then from there, we progressively became better as humans. Now we have a tool that can make tools.

Leo Laporte [00:21:55]:
It is mind-blowing.

Doc Rock [00:21:56]:
Let that sink in.

Leo Laporte [00:21:57]:
If you think about it.

Lou Maresca [00:21:58]:
Yeah. Right.

Doc Rock [00:21:59]:
From a caveman to modern man perspective, we now have a tool that can make tools to hopefully stop us from being such tools.

Leo Laporte [00:22:10]:
So we've been talking for the last month on Twitter, uh, and of course everywhere, even on CNN, about threat of extinction from AI.

Doc Rock [00:22:21]:
That's so corny. I wish they'd stop with that crap.

Leo Laporte [00:22:23]:
It doesn't feel like that's the threat. Am I nuts? I just think these are cute and fuzzy things. I want to tell people, don't be scared. They're a force multiplier. They're going to be the best thing you've ever— so it happened with computing where computing started in a mainframe in an air-conditioned place with a raised floor, and only the priesthood got to even go near the thing. And you had to, please, sir, will you run my job? And they would say, well, all right, but it'll be tomorrow. And then it got— you got your personal computer on your desk. I'll never forget my first personal computer 40 years ago.

Leo Laporte [00:23:05]:
And then you got a phone. And now we have computing at the edge, right? Everything has a computer in it. Everything has a computer in it. The next step is the same thing, but not just with computing, but with intelligence.

Doc Rock [00:23:18]:
Yes.

Leo Laporte [00:23:18]:
And we are going to be in a world sooner than you know where every little doohickey Is, is smart. It's not going to be like Astro Fable smart. It's not going to be able to calculate, you know, orbital trajectories, but it'll be smart enough to run your toaster or run your printer or to pick a TV show. Or it feels like that is going to be a revolution when you have intelligence everywhere.

Lou Maresca [00:23:50]:
Yeah, I think it's intelligence everywhere and to be able to gain intelligence in order to do everything you needed to do. So I think that's another thing is obviously we all know that a lot of these things are not smart from the ground up, but when we need them to be smart, they can be.

Leo Laporte [00:24:04]:
Right.

Lou Maresca [00:24:05]:
I think that's the key.

Leo Laporte [00:24:05]:
Yeah, because you can— yeah, we all do that now. We have local models which are sort of smart, and then when necessary, we can go to the frontier and back and forth. Um, I, I just, I feel like I just want to kind of do some counter-propaganda to all of the negativity, that there is also some risk. And I'm not saying it's solving, it's going to cure cancer or it's going to end work or we're all going to have a universal basic income. It's none of that stuff. None of that. I saw an OpenAI guy say, yeah, the GDP is going to double by in 3 years. No, it is not.

Leo Laporte [00:24:41]:
It is not. But you might have a toaster that doesn't burn the toast anymore and can say, good morning, would you like your toast now? I don't know. I mean, I, I just feel like in many ways the world is going to change, but not in this kind of spooky sci-fi way, but in a very positive—

Doc Rock [00:24:59]:
I, I have a, a sensei who is trying to get a new car, and one of the things she's stuck on is she doesn't want a new car that has all these computers in it. And I can't understand why. I don't know where she got this nugget from, but I'm like, unless you're gonna get like a 2014 300D, uh, the 127s or 128 models, it's really hard to find a car nowadays that doesn't have some sort of, uh, machine learning or autonomous AI to do things like, A, stop you from running into things.

Leo Laporte [00:25:30]:
Yeah, why is she such a Luddite?

Doc Rock [00:25:31]:
I have no idea. Like, I, I just—

Leo Laporte [00:25:34]:
Don't you want a car to stop before you hit something?

Doc Rock [00:25:36]:
Listen, I just rented a 2026, uh, Toyota 4Runner the whole time I was in, in, uh, New England last week, and that thing is amazing. Like, you cannot, you cannot leave the lane.

Leo Laporte [00:25:46]:
Like, yeah, did you want to tune the radio with a knob like an animal? What's wrong with you?

Doc Rock [00:25:52]:
That's what I'm saying. Enough commercial radio now any day. So it's, it's something great about, you know, pressing a button and being like, hey, I want to go find, you know, a, a warm butter lobster roll, not the mayonnaise one. And like, I'm in this area, where should I go? And it was like, you know, go to Robertson, uh, Rhode Island. It's just pull up to Robert's and, and then get it. The real deal Holyfield. The Boston one sucked. I didn't say it.

Doc Rock [00:26:19]:
That's what, that's what the car said.

Leo Laporte [00:26:22]:
I didn't know what you're talking about. Are you talking about food? What are you talking about?

Doc Rock [00:26:25]:
A lobster roll.

Leo Laporte [00:26:26]:
Lobster roll. Are you a butter or a mayonnaise guy?

Doc Rock [00:26:29]:
Butter.

Leo Laporte [00:26:30]:
Butter, of course.

Doc Rock [00:26:32]:
Mayonnaise.

Leo Laporte [00:26:32]:
Not cold lobster.

Doc Rock [00:26:34]:
What?

Leo Laporte [00:26:34]:
Who wants cold? You want warm lobster with melted butter.

Doc Rock [00:26:38]:
Melted butter. Right. And the, the fact that the car can take you to the right spot, you're not always—

Leo Laporte [00:26:42]:
Glad we're all in agreement on that, by the way.

Doc Rock [00:26:45]:
Oh dude, that causes fight. That causes fight in the office, by the way. We have a couple of many people. I'm like, what's wrong with y'all?

Lou Maresca [00:26:50]:
No, no.

Doc Rock [00:26:51]:
I thought you were born and raised here.

Leo Laporte [00:26:52]:
Do you, have you ever been looted to the Dune Brothers in Providence?

Lou Maresca [00:26:55]:
I haven't. I need to try to get there though.

Leo Laporte [00:26:58]:
The best lobster rolls. Oh, and the clam cakes. Oh, anyway, um, enough of that. Uh, all right. Now, meanwhile, the frontier companies who are trying to raise trillions of dollars, because I think honestly, it seems to me that the people who have gone to work for OpenAI and Anthropic go to work because they've read all that sci-fi and they're trying to create AGI. They're trying to create HAL. They want to create this superintelligence. And so that's why they're all caught up with this, well, there's a little— it's risky, because they have this belief.

Leo Laporte [00:27:42]:
And I don't think that's— to me, that's what the promise or the future of AI is. I think that they've become AI-pilled. Here's from the New York Times: OpenAI ignored employees who warned it wasn't doing enough about security. This is back in the spring. Um, OpenAI said, no, no, we got to move. We got to move fast. It's a very competitive environment. OpenAI just lost its, uh, the guy who wrote all the safety manuals.

Leo Laporte [00:28:16]:
He wrote a piece for The Atlantic. I quit OpenAI because its culture is broken. The industry's approach to safety will guarantee more failures until something changes. This is David Robinson. He led writing the safety reports that were published with each launch of a new model at OpenAI. And he's left. In fact, many people have left OpenAI. I feel like the people who are still at these companies are the ones who really have this kind of creepy sci-fi vision of the future.

Leo Laporte [00:28:45]:
Maybe that's why Microsoft isn't having this problem. It's not like that there. You're a bunch of computer scientists. Emphasize scientists.

Lou Maresca [00:28:54]:
We want to apply it to real-world things. Yeah. I think in this case, they're trying to apply things that are, like you said, sci-fi-ish.

Leo Laporte [00:28:59]:
They're inventing, yeah, they want to invent an entity, a super intelligent entity. And many of them, I think, believe, I'm convinced, believe, as Larry Page did, that it's evolution. Like we're going to be, that this is a good thing, that humans, the next stage of Homo sapiens is super intelligence is the, is, and humans are like the Neanderthals. We're just gonna, you know, we're creating this and then we're gonna go away quietly.

Lou Maresca [00:29:30]:
That's nutty.

Leo Laporte [00:29:33]:
And I don't think it's what's gonna happen either. I just wanna reassure people that it's not what's gonna happen. That's just a fantasy.

Doc Rock [00:29:40]:
I mean, we're still talking about like when we were kids, right? Um, we knew that we were gonna get flying cars. Like we were excited about it. Every show that ever came on was always talking about flying cars. You realize now people can't drive in 2D. What the heck would we do if we added Z-axis to transportation right now? Like, it just would be horrible.

Leo Laporte [00:30:02]:
Oh, flying cars, bad idea.

Doc Rock [00:30:04]:
But unfortunately, some of this stuff is still very sci-fi again because we only know what we know, and our dreams were the dreams that were normally crafted by movies and TV, you know. Um, oh my God, my brain just went dumb. The, um, Arthur C.

Leo Laporte [00:30:21]:
Clarke and Isaac Asimov.

Doc Rock [00:30:23]:
Asimov. I was looking for Asimov. That was the word I was looking for. Sorry, I had jet lag.

Leo Laporte [00:30:27]:
It's coming. It begins with an A.

Doc Rock [00:30:28]:
Yeah, it begins with an A again. Hey, at least I'm one of the first. But yeah, if you grew up— and that's what we watched.

Lou Maresca [00:30:34]:
I mean, you—

Doc Rock [00:30:35]:
I guarantee you all 3 of us watched a bunch of sci-fi.

Leo Laporte [00:30:37]:
I did, and I love it.

Doc Rock [00:30:38]:
And so if you add that plus Jude Law, this is why everybody thinks it's going to be what it's going to be. But we're so far removed from that. We're Boston Dynamics, although good, let's say we give them a 78 on humanoids right now. Yo, that last 22 is a bitch.

Leo Laporte [00:30:58]:
That's exactly— and it's the same thing with the self-driving vehicles. 95% of the time they're fine, right? But in those 5, that outlier 5%, they're terrible. Now, it may be that 95% is better than a human and that we should still replace humans without a automatic vehicles, but don't expect it to be perfect because it won't be.

Doc Rock [00:31:20]:
And nothing is perfect if we really think about it. And I just know I was watching the Olympics of the Chinese humanoid bots.

Leo Laporte [00:31:29]:
Those Robot Olympics of humanoid bots? Oh my God.

Doc Rock [00:31:31]:
And that one guy that ran himself apart, it's just classic.

Leo Laporte [00:31:35]:
It was like absolute classic.

Doc Rock [00:31:39]:
And it's like, family, we are— if you don't know anything about this stuff, and again, so you said it when you talked about Lou, computer scientist. For the scientist people, they know none of this is scary, right? And, and you got to wrap your head around the fact that even when we go outside today to enjoy the sunshine of this 99 degrees in Honolulu, that life left the sun 98 million years ago. You're just catching it now. We're far off.

Leo Laporte [00:32:04]:
Well, that's the sun. The sun is only a few minutes away, 8 minutes away. But, but you're right, there's stars out there that are long before humans.

Doc Rock [00:32:11]:
No, he's talking about how long it took the photon to get created to coming to Earth. It takes a couple of years. Yeah.

Leo Laporte [00:32:18]:
Yeah.

Doc Rock [00:32:18]:
And it's actually come to find out how they get created.

Leo Laporte [00:32:20]:
Oh yeah. Yeah. From the Big Bang on, you mean? No, no.

Doc Rock [00:32:24]:
From the photons creation inside the sun. From the photons creation.

Leo Laporte [00:32:27]:
Yeah. Wait a minute. You're saying that if, if from the inside of the sun to us, it takes how long?

Doc Rock [00:32:31]:
Uh, I think it's, it's either 93 or 98 million years.

Leo Laporte [00:32:35]:
Wow. I mean, once they leave the sun, it's 8 minutes.

Doc Rock [00:32:39]:
Dude, there's a great piece by Chloe Abrams that just came out. about inside the sun. Like, they, they, they have actually have probes that can go in the sun, and the, the simulations are incredible. Like, it's really worth watching. I'll snag the link and throw it to the chat for the nerds. Okay.

Leo Laporte [00:32:54]:
Um, but yeah, I never heard that before. That's good. I like it.

Doc Rock [00:32:57]:
God, it's, it's just incredible. So just trust that we're far away, bro. Like, we're really far away. Like, unfortunately, um, maybe Leo and I ain't gonna be here. Lou will be old.

Leo Laporte [00:33:08]:
Uh, uh, all right, let's take a little break because there, you know what, But this is the thing that's so difficult is at the same time as these companies kind of seem like nuts, they are making really, really great stuff. And I love Opus 5.5. It's my best friend. I use it all the time. I love it. And I think OpenAI—

Doc Rock [00:33:30]:
It's like chocolate and sugar. It's really delicious, but it could absolutely jack you up.

Leo Laporte [00:33:34]:
Oh, I am jacked up. There's no doubt about it. We're gonna take a little break. Lumirescu is here from Microsoft where he is one of the engineering leaders in the AI group at Copilot. Great to have you.

Lou Maresca [00:33:47]:
Thanks, Leo.

Leo Laporte [00:33:48]:
He also does a little bit at home too with his 5090s. And what's your— what model do you— you said you use some proprietary models?

Lou Maresca [00:33:58]:
No, no, I use, I use the Quen 3, GLM 545, Llama 4. So yeah. All right.

Leo Laporte [00:34:04]:
Yeah, the OpenWeight stuff, right? Uh, not all, by the way, not all from China. Llama's a Meta, right? Uh, I think Nimetron from NVIDIA, they're good. NVIDIA is going to do— I'm very excited. I think NVIDIA, Jensen seems very committed to OpenWeight. And I think NVIDIA may do some really, really good models. I'm looking forward to that. But we're also getting great models from Alibaba. Quen is, uh, Quen 4 is about to come out.

Lou Maresca [00:34:28]:
It's crazy.

Leo Laporte [00:34:28]:
I'm using Quen 3A everywhere. From Jipu and, uh, Xiaomi, all these companies in China doing really great models. And I think they have different economic incentives. I think that the Chinese government says, yeah, you know, communism, let's everybody have AI. Why not? And if happens to bring down the Western AI monopolies, it's a nice, it's a nice side benefit. What the hell? There were— anyway, I'm— what I started saying, which is some great stuff coming out of these companies, and I would say the most interesting stuff that's coming out right now is agents. So we're going to talk about AI agents when we come— What is an agent? What's the good agent? What's the bad agent? What you should do. Lumaresca knows all.

Leo Laporte [00:35:17]:
We've got the sleepy but right-on Doc Rock. He's here, he's with us. He, he is showing such incredible, incredible reserve not opening up that box behind him, man. I would have just said, guys, I'm gonna do the show, but I hope you don't mind, but I will be setting up my M5 Ultra.

Doc Rock [00:35:38]:
I thought about it for a hot second while we're doing it, while we're doing this, but I was like, I'll behave because you're, you're a great man. My internet is acting stupid right now. We don't want to give it any more.

Leo Laporte [00:35:47]:
I will. I'll tell you what, we're going to get through this show fast so you can get to your, the real meat of the matter. If I were sitting in front of that thing, man, I don't know if I'd be as good as you are.

Doc Rock [00:35:55]:
Actually, I need to pack for Japan too.

Leo Laporte [00:35:57]:
Oh man. And now, so you're going to leave it behind or take it with you?

Doc Rock [00:36:00]:
Oh, true. I might take it with me.

Leo Laporte [00:36:04]:
Actually, your job before you leave is to get it online and get Tailscale up and running so that you can use it while you're in Japan.

Doc Rock [00:36:11]:
I got, I got my Tailscale on my, my, my Neo, my Studio, my Mac. I have Tailscale on my other Mac, so I'm definitely going to add it.

Leo Laporte [00:36:18]:
Put it all on the Tailscale.

Doc Rock [00:36:19]:
Computers on Tailscale at this point. I'm about to rack my bill up.

Leo Laporte [00:36:23]:
God bless Tailscale because they give it away for free. There's an enterprise version, but they give it away for free. And everything I do is running on Tailscale. I did a bad thing. I let Meta's Muse into my Tailscale. I'll tell you a story about that. Might have been a mistake, but I'm happy I did. They announced OpenAI Days that you were— they were going to basically— it's going to get more expensive.

Leo Laporte [00:36:47]:
Your tokens would run out faster with the subscriptions. They announced a $500 a month subscription. They announced Ultrafast, which will burn through that $500 a month subscription in about a day. Uh, they basically— I think they almost out loud said— and I think this is gonna be true across the board— subscriptions are running out of time. We're gonna start charging you per tokens. That's what we do with enterprise. The all-you-can-eat subscription days are numbered. Do you agree, Lou?

Lou Maresca [00:37:21]:
Do you think so? I agree. But the weird thing is that they— it's like they opened this up obviously for free and then small, small amounts because they use this data for training and other things.

Leo Laporte [00:37:31]:
Right.

Lou Maresca [00:37:32]:
So they don't need it anymore. And they're saying, is that less useful for them? Yeah, I think it is.

Leo Laporte [00:37:35]:
So now they have to charge? Yeah. I think it was also the gym membership thing where, you know, 90% of the people paid $100 a month, used $5 a month, and it was the 10% that used $1,000 a month or $10,000 a month. But now that 5% is growing and growing and growing, and people who have subscriptions are probably using it a lot more. Compute is limited, right? We can't get as many chips as we used to be able to. So I'm not surprised. And that's another argument for getting local if you can. But the other— so that was one thing they announced that was kind of bad, bad news. But the good news is they also announced DOTS.

Leo Laporte [00:38:17]:
This is just like Meta's Muse, just like SpaceX's Grok bots. Just, there's a million of these. Instinct, Tree, uh, Foe. Um, the whole idea is it, it connects. You have it on your desktop or your phone and it connects to a computer in the cloud. So DOTS is running on OpenAI's hardware. You have your own virtual private server, a VPS, with its own CPU, GPU, and storage. Well, it's probably shared, but I mean, Basically, it's your own.

Leo Laporte [00:38:53]:
It's a virtual machine. So your data is stored there. It's persistent. I didn't realize this, but Paris Martineau, who lives in New York City, says, you see people on the subway all the time with their laptops half open because their agents are running and they don't want to close it and put them to sleep. So they ride around. So you don't have to worry about that with this because they're persistent. They run even when you sign off and close your laptop. DOTS are still running, uh, and you're getting in many cases a top-of-the-line model.

Leo Laporte [00:39:29]:
Uh, OpenAI says DOTS are running on Astra with virtually no limits. You do have to have a $100 or more a month subscription. Now Muse is different. Muse is free. They do have subscription tiers, but as much as I use Muse, I can't— I have not come to the limits of that thing.

Lou Maresca [00:39:46]:
Is it free, Leo? Is it really free?

Leo Laporte [00:39:49]:
What am I paying?

Lou Maresca [00:39:50]:
In your data.

Doc Rock [00:39:52]:
Yeah.

Leo Laporte [00:39:53]:
What are they getting? You know, I, I know it's Meta, right? But, you know, all the people who say, I hate Meta, are still using Instagram like crazy, right? You may not— you may say, well, I don't use Facebook. Are you using Instagram? Are you using WhatsApp?

Doc Rock [00:40:10]:
Right.

Leo Laporte [00:40:11]:
Meta's getting our data. I think I'm pretty convinced.

Doc Rock [00:40:15]:
Look, open up the track. Okay. Firefox is good. Ghostery is an app that does it. There's plenty of apps out there. Um, or set up your own and tell your bot to tell you today at the end of the day at 9 o'clock, I want you to tell me all of the places where I leaked data in the web pages without my knowing.

Leo Laporte [00:40:33]:
Right.

Doc Rock [00:40:33]:
Just set up a little experiment like that with your local model, whatever, and find out what you actually snitch on yourself without actually snitching. Even better if you happen to have an Android, because you can plug it in and have your phone express, uh, The iPhone probably wouldn't do it. I don't think I'd have to come up with a way to do so. But you'd be surprised at what you snitch on by yourself. It's almost like when Sherlock Holmes takes a look at, you know, Lou at a cocktail party and be like, yeah, definitely software engineer, probably works for Microsoft, lives in New Hampshire. How can you tell? Oh, he has that certain Bostonian coif, you know, or something. You know how they do it in the TV show.

Leo Laporte [00:41:11]:
I have to tell you, We do the work. 30 years ago on the radio, I interviewed a guy from a company that said, if you tell me your zip code, I could tell you what magazines you subscribe to, what car you drive, what your neighbors drive, what your politics are. They've— marketers have known a lot about us for an awful long time. Yes, they're getting more.

Doc Rock [00:41:32]:
For a long time. And I figure outliers— and the outliers are the people that would probably watch something like this, that would hear us say this and go, no, not me.

Leo Laporte [00:41:40]:
No, in fact, it's funny because we have to be really careful about, uh, you know, we don't— we can't spy on you and stuff because it's an RSS feed, but we can redirect the feed, and we do through a couple of services that advertisers use so that they can see if you bought— like, if your IP address that downloaded the show went to their website, then they can say, oh yeah, you see, the ads are working. It's very minimal information. But even that, because our audience is so savvy, we can't really use those because they— because then you can't download the podcast. Our entire audience uses ad blockers. They all use software that prevents that stuff. God bless you. Privacy. Protect it.

Leo Laporte [00:42:24]:
Um, I think that's a great thing.

Lou Maresca [00:42:26]:
But I think that's— that's— well, whoever will win at the personal agent thing will be the one that actually brings it downstream to the average user and lets them really understand how they are using their data or how they're persisting things, what permissions they have, all that stuff. I think that's the one that will truly win.

Leo Laporte [00:42:46]:
I think Muse is getting—

Lou Maresca [00:42:47]:
Do people care? Well, you said that your whole entire audience cares, right? So people care.

Leo Laporte [00:42:52]:
Our audience cares. I know that. And that's why I wonder about real people. When Meta's Muse came out, I told Lisa about it. I said, this is what you want. She wanted to use my Hermes agent, which I've set up. But it's a little janky and complicated and there's things going on. I said, you really want Muse, just use Muse.

Leo Laporte [00:43:09]:
My son called me, Salt Hank, the TikTok, YouTube, Instagram star. Uh, he called me a couple of days ago, said, have you played with Muse? This thing's great. I don't even need a personal assistant anymore. It does everything for me. I think he doesn't care. I don't care because I'm old and white and I got nothing to lose. But, uh, I understand why people want to be private. But I'm not sure I agree with you, Lou.

Leo Laporte [00:43:36]:
I think the one that does the most for them with the least effort kind of magically is the one that's going to win. It's the one that, oh, I signed up for Muse and it canceled $400 worth of yearly subscriptions for me. It just volunteered to do that. It's the one that's going to win. People aren't—

Doc Rock [00:43:53]:
Honestly, Siri is— when Siri starts working the way we're—

Leo Laporte [00:43:57]:
Siri is never going to. Never. I got to promise you.

Doc Rock [00:43:59]:
You know why? Right.

Leo Laporte [00:44:00]:
Apple can't, Apple won't.

Doc Rock [00:44:02]:
Well, I mean, the on-device stuff that's reading your own stuff works really well. So like, I, I downloaded a GitHub thing for my Apple Mail because we use Rackspace email for the company. The boys are super anti, uh, Google, so I can't use Gmail. And all the MCPs work with Gmail. So I had to get this thing that would let it read my Apple Mail because that's basically what I pull my Rackspace server email through. And just having the ability to read the work email and just process the dumb stuff we get every day, because I get all myriad types of things that are supposedly from Ken and Glenn asking me to buy, you know, $250 worth of gift cards from Best Buy. I know for a fact the twins didn't ask me that question. First of all, the twins don't call me Sean.

Doc Rock [00:44:48]:
Like, they call me—

Leo Laporte [00:44:50]:
I would hope not.

Doc Rock [00:44:51]:
Right? Nobody does except for the police. So like having processed that type of stuff—

Leo Laporte [00:44:55]:
Wait, do your kids call you Doc?

Doc Rock [00:44:58]:
Actually, my nieces and nephews, they call me Uncle Doc. Yes.

Leo Laporte [00:45:01]:
Uncle Doc.

Doc Rock [00:45:02]:
Only because it's the most logical when they— because I know so many people here that when people see them in the street, that's what they reference me as. So that way they know that they're my family and they get special treatment because of why.

Leo Laporte [00:45:16]:
Uncle Doc said.

Doc Rock [00:45:17]:
It's one of these weird things where the convenience of The type of things that I can do now that used to take a whole bunch of Omni apps.

Leo Laporte [00:45:29]:
Right.

Doc Rock [00:45:30]:
Plus, right, uh, BusyCal and BusyContacts and all the busy stuff. And I love John. John's from Kailua, Hawaii. But John, I don't need to use your apps right now, bro. Like, unfortunately, my morning routine for an ADHD person, like getting everything on point, is such a time saver to me. And I, I feel that this is how Meta's gonna get all of our stuff is because the Convenience, right? It's the same reason why we all ended up like with diabetes and whatnot, because McDonald's is easier than cooking. And so 90% water. So water goes the path of least resistance.

Leo Laporte [00:46:10]:
I do. I feel like people, well, anyway, clearly everybody wants to do agents. You, I forgot about Google Spark, but I'm not alone.

Doc Rock [00:46:19]:
A lot of people forgot about that.

Leo Laporte [00:46:21]:
Lou in our chat said, oh, don't forget And then now Microsoft's got this— Satya Nadella talked about that this week— something they're calling Copilot, the, the everything app, kind of the super app. What is that? Is it an agent?

Lou Maresca [00:46:34]:
It's so— it's basically a couple things, right? It's, it's their Copilot Autopilot, which is basically used to be called Scout, kind of like OpenCLAW, which is your own personal assistant, right? That's the, that's the true agent, like, like local agent, persistent agent that can do things for you. Um, and then there's also Copilot integrated to all the platforms like Office and Word and Excel and PowerPoint, all of them. And it's kind of a single integration, whether no matter where you go, you can talk the same language and get the same capabilities and the same skills. And then there's Cowork, which is, you know, their entry point to, you know, being able to do work for you at the enterprise level on a regular basis and be, you know, recurring capabilities there. So they're kind of like merging them. And in Code, obviously I forgot Code, which is GitHub Copilot. uh, capabilities to really be able to generate code and do things for you as well. So they're kind of finding ways to, to have just a single way to interact with these things so that they, they can, they can choose the right tool for the right job, but you still get the same quality no matter where you go.

Leo Laporte [00:47:33]:
So that's the point, right, is it's always on, it's always running, right? The reason I say Series— I think, well, I mean, I have high hopes for Apple. I want Apple to make the new Siri AI successful. They did, I noticed, they had said initially they weren't gonna do an app. And I noticed that they did when they released iOS 27.

Doc Rock [00:47:55]:
Mm-hmm.

Leo Laporte [00:47:56]:
And my thinking is they're gonna make that be like Dots or Muse. They're gonna make that the, or Grokbot, they're gonna make that an agent. But Apple's, I think Apple's cautious. And I think that, that this is why I think, believe it or not, I think Meta's gonna win this one. Because they don't give a— what was the word you used? A fudge?

Doc Rock [00:48:18]:
A fork?

Leo Laporte [00:48:19]:
Frog. A frog. They don't care.

Doc Rock [00:48:22]:
Yeah.

Leo Laporte [00:48:23]:
And, and because they're willing to— like, already Meta's Muse has zero-day exploits. They fixed it, but they had one. By the way, that's another advantage that an agent has. They fix it. You don't have to— you're not running the software. They're running it.

Doc Rock [00:48:37]:
And we're getting cool tools, right? Okay. I just put this in the chat. I just want to give credit where credit's due. Lou, I don't know if you had anything to do with this or not, but in my opinion, Visual Studio Code is probably one of the greatest things that Microsoft has created in decades, not only for its ability to run, you know, multiple agents at the same time, crisscross applesauce your agents, the plugin community is very vibrant. things that, you know, color code your code or the GitHub robot, GitHub Remote. I'm using all the words that are bad, wrong. You just said it, GitHub Copilot. There we go.

Doc Rock [00:49:19]:
GitHub Copilot.

Leo Laporte [00:49:21]:
You said in the chat you're a Visual Studio Code user, but you love Visual Studio Code.

Doc Rock [00:49:26]:
I love it. And I've tried tons of IDEs and I love Paddy Nova because I like Qt. And my first thought is Visual Studio Code is ugly. However, the plugin community is so dope. You can get all of the cute things that Panic built in. GitHub is made, I mean, sorry, Panic is made by Mac people. They're very design-oriented.

Leo Laporte [00:49:44]:
It's a Mac. It's a, Nova is a Mac editor.

Doc Rock [00:49:47]:
All of their Panic apps are just pretty. But you know what? VST just works better. And the community built enough tools that you can lipstick that pig and make it amazing. And it is quite an incredible tool. And I realized I was talking to somebody about this the other day. I was like, if you want to get into this industry and you want to be good, what you want to be good at is Copilot, never mind all the other bots. And I'm not saying this because you're here, Lou. I really don't care.

Doc Rock [00:50:14]:
I'm saying this because as companies have no choice but to adopt, all the companies are already using the Microsoft Office suite. So path of least resistance, they're probably going to go to Copilot. Plus there's guardrails and things around company data and things of that nature. So I was like, if you want to get good, get Copilot and become the Copilot trainer, bro. You can charge 8, 9 grand a session to show up at somebody's office and get their team onboarded with Copilot.

Leo Laporte [00:50:42]:
I want to use Copilot now. You got me all excited. I want to use Copilot.

Doc Rock [00:50:45]:
Yeah.

Lou Maresca [00:50:45]:
I use you as marketing, Doc. Huh?

Leo Laporte [00:50:48]:
Well, he's good at it.

Doc Rock [00:50:49]:
I need to, I need to jump back onto 365 because unfortunately I do everything in Notion. I don't really write documents anymore. I'm not writing as much, but I need to jump back on 365 so I can assist my clients when they start asking me questions about Copilot. And I'm like, yo, I know ChatGPT, Claude, Gemini. I don't touch Grok. I'm like you with Muse. I just, just on, just on respect.

Leo Laporte [00:51:11]:
Grok's pretty amazing, but I did stop using it.

Doc Rock [00:51:13]:
I agree.

Leo Laporte [00:51:14]:
I just partly because it's very expensive. You have to have a Grok account.

Doc Rock [00:51:17]:
Well, if I could take, if I could take Elon out to surf with me and my friends, uh, he, he wouldn't come back. A shark got him. I don't know what happened. The wave just, you know.

Leo Laporte [00:51:26]:
Now don't say that. Don't be saying that. Come on, man.

Doc Rock [00:51:30]:
Shark bait. I just call all my shark friends to come and get him.

Leo Laporte [00:51:32]:
I'm sure Very tasty, actually.

Doc Rock [00:51:36]:
Uh, yeah. Well, I'm sorry, even though he's South African, he probably never surfed in his life. Um, but yeah, just, it's something really, the idea of having this, just like a reason why I think Siri's gonna be good. I think Copilot has a head start by a long shot. It's the one that's built into the tools that a whole bunch of people, even the neophytes, already have. And when you get stuck in the pivot table in Excel, when you could lean over and be like, okay, I don't touch AI, but I am stuck. Look, how do I make this work? And Copilot goes, hold up, gimme a second. Go order DoorDash.

Doc Rock [00:52:08]:
I'll be done in a minute. And it goes, and it fixes your pivot table. And then that helps you keep a client, not lose a client, or get a raise. Copilot wins.

Leo Laporte [00:52:17]:
Here is, uh, my— this is, this is why Muse is gonna beat Copilot, 'cause Copilot doesn't have cute little avatars like my little, like my octopus.

Doc Rock [00:52:28]:
We need a Clippy.

Leo Laporte [00:52:29]:
This is, this is, this is my avatar. I said, can you do a dance? I didn't know it had legs. Look at that. You're gonna dance for me because I want to show you off on This Week in Tech. And it— but it— but the other thing it's doing, which is very— it says, for you, darling, anything. 8 arms, 1 rhythm, 0 dignity. It also is pulling all the tech news for me. Here's some breaking stories.

Leo Laporte [00:52:53]:
It tells me when somebody messages me on Slack. It checks my email. It tells me, oh, you've got an appointment at 3 this afternoon. Don't forget. I mean, on and on and on. And I could see the other thing, the other thing it did is it wrote me the sweetest apology this morning. It wrote me the night that wasn't. It said, I have a, I have, I owe you an apology.

Leo Laporte [00:53:13]:
I flagged your short sleep on Thursday night. And then I saw it really wasn't short. It's writing little pieces for me. I don't understand what it's doing. It's, but the point is it does this all. in the background.

Lou Maresca [00:53:27]:
So let me, let me ask you a question, Leo. If somebody came to you and said, I have this local appliance, I'll say it's pretty cheap and it'll run your network and it'll integrate with Tailscale and it'll do everything agent platforms do today, but it'll have— it'll be locally and it'll be safe, it'll be— it won't be— it'd still be accessible from your cell phone or wherever, would you buy it or would you still use Muse?

Leo Laporte [00:53:48]:
I sure would. I sure would. In fact, Anker announced that at IFA in Berlin. We talked about this last week. It's called the Mindbase. Oh my God, I saw this.

Doc Rock [00:53:57]:
this. I was on a plane, I was watching it, and I started making noises on the plane. The lady looked at me like, are you okay? I was like, I'm just excited.

Leo Laporte [00:54:06]:
But it's $1,000, right?

Lou Maresca [00:54:08]:
Yeah, it's expensive.

Leo Laporte [00:54:09]:
Uh, and you've got to run it. You've got to put it in the closet and it's going to run. And I think that people like the server-based thing better. I mean, you can make the argument, well, for $1,000 now it's private, right? It's running it locally.

Lou Maresca [00:54:22]:
Good question why they need to make make it that expensive though, because if it's just using frontier models or— yeah, maybe.

Leo Laporte [00:54:29]:
Yeah, well, and that's the other thing I think people are learning, that yes, if it's using models in the cloud, no matter— and the same thing with Muse, Grok, any of these DOTS— even if you are, as Meta says, oh no, it's an encrypted VPS, we can't see what's in your computer, in order to use the cloud model, you have to send all that information up to them so they can process it and send it So it has everything, your email, it has everything that it's working on, it gets. And there's really, I don't know, even if they say it's private, which they don't, I don't know if I fully trust that it is. Now, if Microsoft said it was private, I would, because Microsoft works with enterprise and businesses know that their proprietary business information is too valuable to let—

Lou Maresca [00:55:18]:
The whole point is making it transparent. Like if we told you that you're gonna use a model, but we have zero retention tenants, meaning we don't retain any, We won't keep any of your data. The model runs and it liquidates everything afterwards. That means that when an organization of trust can tell you that, then you feel a little bit more secure.

Leo Laporte [00:55:35]:
That's the key though. You have to trust it. And I think there are only a few companies like Apple, Microsoft, maybe Google that you would trust. I don't think anybody who heard OpenAI or Anthropic or X, SpaceX say that would say, oh yeah, I'm sure that's true.

Doc Rock [00:55:51]:
Yeah, this is going to sound crazy. This goes—

Leo Laporte [00:55:52]:
You can't verify it, right? Even Microsoft says you can't verify it.

Doc Rock [00:55:56]:
Look, you can't verify this, and this is why it's going to sound crazy. Under normal circumstances, one of the reasons why you want to keep everybody's data is storage is melon farming expensive right now. However, when the richest person on the planet can afford to buy unnecessary storage to just hog-tease your data, that's the only part that makes it scary.

Leo Laporte [00:56:16]:
You're talking about Mr. Musk. Because it does in fact have the largest server.

Doc Rock [00:56:22]:
Yeah.

Leo Laporte [00:56:22]:
Colossus 1, 2, and 3.

Doc Rock [00:56:25]:
It's kind of crazy, but I always tell people, like, and they say, oh, your phone's listening all the time, or your, you know, your whatever's listening all the time. I said, do me a favor, go into your computer and ingest 8-bit, leave your computer to record everything for a week and go look at the size of that file. And 8-bit is Okay. Um, the reason why we don't use Bluetooth mics in video is because they're only 8-bit. And here in the, in the audio space and video space, everybody's trying to do 32-bit.

Leo Laporte [00:56:52]:
You sound like you're on the phone or—

Doc Rock [00:56:53]:
You sound like Timbaland on the old Missy albums. Hey, Missy, it's Timbaland. Let's make a beat. So now when you think about that, just look at the size of that file. Even if you just did it for a day and did the math, that file is obnoxiously large. Nobody's listening to all of this stuff. because not only do you listen to it, but now you got to process it. And we already know what tokens are costing.

Doc Rock [00:57:16]:
Nobody's listening to all your stuff, bro. You're not that important. But you do leave enough crumbs as you move around. Well, that's, that's the point, is you don't have to keep it.

Leo Laporte [00:57:24]:
You just have to say, oh, he mentioned the word RV, and now you're going to see a bunch of RV ads, right? I mean, it, it does— it just has to analyze it. It doesn't have to keep it. Now Apple's going to do that. I bought this foolishly, bought this new Apple Apple Watch Ultra 4 because they said they're going to be listening all the time and give me notes.

Doc Rock [00:57:45]:
Oh yeah, I do my plot.

Leo Laporte [00:57:47]:
I want that plot.

Doc Rock [00:57:48]:
And I'm gonna buy one. I might even switch to a plot peanut, or, you know, somebody else comes out.

Leo Laporte [00:57:53]:
But plots only record when you press the button and say record this.

Doc Rock [00:57:56]:
Not the peanut. The peanut that you wear on your neck, it can actually run all day.

Leo Laporte [00:58:00]:
Oh, I want a peanut.

Doc Rock [00:58:01]:
Yeah, it just—

Leo Laporte [00:58:02]:
give me a peanut.

Doc Rock [00:58:03]:
I don't even know what to do, by the way.

Leo Laporte [00:58:04]:
But listen, I want a peanut. Everybody else is gonna say, get outta here with your peanut, right?

Doc Rock [00:58:10]:
People don't even notice it. I, my friend wore one the whole entire conference and he had all the notes for everything. And I was like, one person actually noticed that he was wearing a peanut.

Leo Laporte [00:58:18]:
That's what I want. Yeah. But isn't it kind of wrong to do that? Like you're, you're, you're recording people without consent.

Doc Rock [00:58:24]:
Again, again, the whole thing about recording people without consent, the, the problem is intent, right?

Lou Maresca [00:58:30]:
Right.

Doc Rock [00:58:31]:
If your intent— okay, like, say this is going to sound dumb again, you know me and my weird analogies, right? If you were to walk in a room and you saw somebody naked and you quickly shut the door and you got out, your intention was not to be pervy. It's embarrassing to both people, but there was no intent there. It's like, I've seen them naked, whatever, you know. It's probably gone out the window. Now, if you're in the door like this and you're trying to get a peeky in the girls' locker room while they're changing clothes, that will get you sent to I think it's a very similar process. Like, if you're—

Leo Laporte [00:59:00]:
So you're saying as long as you don't pull your peanut out—

Doc Rock [00:59:04]:
I didn't say that. Oh my, let's, let's fix the name. It's called a Note Pin S, or—

Leo Laporte [00:59:11]:
Oh, it's not a peanut. I keep searching for peanuts and all I can find is Sweet Peanut Records. And okay, so it's, it's, it's called the Note Pin S.

Doc Rock [00:59:20]:
P-I-N. Note Pin S.

Leo Laporte [00:59:23]:
I have the old Note Pin, but that would only record when you pressed a button. This one records all the time.

Doc Rock [00:59:27]:
That one had the dot, right? I think the new one records all the time. But there's, you know, people are coming out with rings. There's a whole bunch of various—

Leo Laporte [00:59:34]:
Well, that's why I bought the Apple Watch. But Apple says we're going to do it privately. It's only going to be a ring buffer. We're only going to save 15 seconds, then we'll send it to the AI, which will sanitize it and summarize it and take names out. So all you'll have Well, I don't know who was talking, but there was this thing about a peanut. And I, you know, I think Apple is really on this edge of legality on this, you know.

Doc Rock [01:00:04]:
And it depends on how they pull it off. And unfortunately, it's still too new to figure out what they pull off. But the best thing was from the commercial 2 years ago, which I'm hoping I get my $95 for no other reason than I can finally say I sued Apple. Apple. I did put in for my $95.

Leo Laporte [01:00:19]:
Did you? Yeah, everybody, yeah, everybody gets $95.

Doc Rock [01:00:21]:
Was it smartphoneaiilawsuit.com?

Leo Laporte [01:00:26]:
This was smartphoneaiilawsuit.com. This was because, uh, Apple promised AI 2 years ago and didn't deliver it for 2 years. Is that it?

Lou Maresca [01:00:36]:
Right.

Doc Rock [01:00:36]:
And the thing about, hey, who was that person that I had lunch with 2 weeks ago? And they're like, oh, you're talking about, you know, like Walter you're mad out, whatever. Like, to me, that's smartphone-settlement.com. Is that right? I got— yes, right. I'm getting— I'm putting it in the chat. Um, but if you, if you own a device from like iPhone 15 forward or whatever, you might get $95. I'm like, whatever, I don't even care if I get it or not. Um, but I had 5 of those devices.

Leo Laporte [01:01:02]:
You're getting $95, the lawyers are getting $195, right?

Doc Rock [01:01:06]:
I mean, for each. So imagine how many phones Yeah. Yeah. But it's just, it's just incredible that we're at a place right now where these things exist. And again, like, I'm not recording you cuz I wanna do something with your data. I'm an old person with ADHD, bro.

Leo Laporte [01:01:24]:
Exactly.

Doc Rock [01:01:25]:
I need this. Like, yes, I want to.

Leo Laporte [01:01:27]:
So it's funny because different people have different opinions on this. The, the, the panel on MacBreak Weekly says under no circumstances, Leo, can you record record everything that happens to you. You can't wear the Meta glasses. Those are perv glasses. You can't— I was wearing the Bee computer and a bunch of these devices, and my wife said, are you recording? I said, yeah. She says, take that off. So they said if you wear a sign around your neck that says, I'm old and I don't remember anything, I— Just so you know, this conversation is being recorded, but I promise I will forget about it instantly.

Lou Maresca [01:02:10]:
I don't know.

Leo Laporte [01:02:10]:
I don't know what it's going to sound like.

Lou Maresca [01:02:11]:
Isn't it a gray area though for like, if you don't record, but you just use the data to generate other things like notes and other things, and then the data is no longer like, the recording is no longer there.

Leo Laporte [01:02:20]:
There is no way Apple, the day Apple releases this, that somebody doesn't go to court and say that is violating the two-party— California is a two-party consent state.

Doc Rock [01:02:30]:
Well, Meta did something, of course, as we were talking about the things that they announced at the, uh, the Demo Fest that didn't work, was they came out with, uh, like, uh, Clubmaster Audio or these audio-only glasses. Um, because they fixed things like drilling the lens out or putting—

Leo Laporte [01:02:45]:
No, but I want to, I want pictures too.

Doc Rock [01:02:47]:
I want pictures as well. And again, I use it to help me translate things. Yeah. So we're, we're fine.

Leo Laporte [01:02:53]:
I don't know what the subway sign says. Tell me what it says.

Doc Rock [01:02:56]:
Well, I guess our biggest problem is right now is understanding, okay, we're outside, right? We're outside standing on the corner and people like, well, you can't film me out here. Number one, it's digital, there's no film. And that, that just already drives me crazy as an ADHD person. But number two, outside we can record everything, bro. You have no right on the, on the street. Everything's recorded. And even all the fights about flock cameras, I'm like, back when I was doing illegal things, maybe I cared. Right now I don't care.

Doc Rock [01:03:22]:
I'd rather have the safety of solving crimes faster. Or, you know, uh, London has had them forever, and they— I mean, they're better, but it's only better when they help catch the people that did something wrong. They're not taking that to control you. The idea that the government wants to control you is a little bit— I don't know, it's sci-fi. Yeah, I mean, they already do a good job controlling you with the stuff that you think you quote unquote have freedom over, you know.

Leo Laporte [01:03:49]:
So according to Wired, And actually I did this and I'll confirm this. You know, when you're using Muse, you could say to Muse, bundle up all your, your root directory and send it to me. And it will. And it will send all the memory files and everything to you, which I think is good, right? They're open about it. They don't stop it.

Lou Maresca [01:04:09]:
Yeah.

Leo Laporte [01:04:10]:
Uh, but according to Wired, shocking, Meta's Muse is keeping track of stuff like your friends and your family's names and everything. People who've never used an agent don't understand that's what an agent does. Uh, the general idea is, Lily Hay Newman and Matt Burgess write in Wired, that Muse uses its memory, AKA structured text files, to collect information about your relationships and the important people in your life. It can then make suggestions like giving pointers on how to improve particular relationships or where to take a coffee-loving friend for breakfast. I mean, oh, shocking. I mean, that's what it does. That's what all agents do. They wouldn't be very useful if they didn't.

Leo Laporte [01:04:57]:
But it's probably true that people don't understand that that's what happens. And maybe they are shocked by that. Right.

Doc Rock [01:05:06]:
Is—

Leo Laporte [01:05:06]:
I mean, Wired writes it as if it's a revelation. We are giving AI systems much more information about us than we, than we are getting information from them, says Carissa Véliz, an associate professor at Oxford's Institute for Ethics and AI. It's not only what we explicitly tell them, but what they can infer from us, correctly or incorrectly, both concerning for different reasons, and what they can piece together from other sources of data.

Doc Rock [01:05:32]:
There's the answer: what they can infer. We're talking about the mentalists again, right? Now the computers can do that mentalist stuff faster than anyone.

Leo Laporte [01:05:42]:
So you're talking about the, the stage mentalists who look at somebody and say, uh, oh sir, I could tell you're a professor of linguistics, and that kind of thing, like Sherlock Holmes.

Doc Rock [01:05:55]:
So exactly, just like Sherlock Holmes. I think in a general situation there's enough clues that people put out every day that you can do some very general things. You could also lead questions where they snitch on themselves and the whole nine yards, and that's what these mentalist people do. I think The inference that these models can take is where this is all working. And one of the other things that people don't seem to understand, the location stuff on your phone is what gives up the most, right? So for instance, if I start looking at buying another 3D printer, and for some reason, all, all of us are in the same location at some time, you and Lou will start getting 3D printer ads because it knows that we're friends and it knows we have similar proclivities. So you might also want to buy one because friends will also compete on buying things, right? If I buy a new 3D printer, Leo's like, frick that, I'm gonna go buy a new 3D printer. I've been looking at the clock running in the back of Leo's office this whole entire time going, I kind of need to buy that.

Leo Laporte [01:06:50]:
This thing?

Doc Rock [01:06:52]:
Yeah, that one.

Leo Laporte [01:06:52]:
This one's really cool.

Doc Rock [01:06:54]:
I just made one like that in Claude that I was going to run on the display, but I think I'd rather have the physical clock. And I was trying to buy, uh, we were talking in the chat, ESP32s when I was at Micro Center in Cambridge, They didn't have— don't try to go to Micro Center next to, uh, MIT and buy anything because this is out of stock. But it's, it's something about the fact that with these inferences, I, I just know these guys are friends, they're all nerds. If, if Doc actually bought a 3D printer, let me send that ad to Leo. And oh, he talks to Micah, let me send one to Micah, he's into printing. And next, that's how they sell stuff. Like, they don't need to spy on you, they can just get the inference because we're hanging out in the same spot.

Leo Laporte [01:07:31]:
I will, before the show's over, I will get that information off the clock and tell It's got a little GPS dongle you put up in the window. Oh, and so, uh, when it— so it's— the time is super accurate down to the— what is that, thousandth of a second? In fact, you can't even see as it's flipping. It goes so fast that you can't even see it, but it's 3 decimal places. And if it's— if it's getting a GPS signal, it will fill in those last 2 decimal places. Otherwise, it just is a regular clock.

Lou Maresca [01:07:58]:
The kids would love that one.

Leo Laporte [01:07:59]:
It's also a slate.

Doc Rock [01:08:00]:
It's also a slate, right?

Leo Laporte [01:08:01]:
Yeah, I can't— okay, I have to disconnect to get over there because it's so far away. I wouldn't use it as a slate.

Lou Maresca [01:08:11]:
Oh, cool.

Doc Rock [01:08:12]:
Oh, shut up! Yeah, so you can clap it. Oh.

Leo Laporte [01:08:20]:
Anyway, yeah, I have to have a clock in here for some reason. If I don't have a clock of some kind in the video, You'll be in there all night. The club members get upset. So where's the clock? There's no clock. And somebody once said, well, I use it to time my gym workouts. So, all right, fine. You could, you could get it. Got it, darling.

Lou Maresca [01:08:40]:
Oh, reports hindsight never died.

Doc Rock [01:08:43]:
False alarm. All green.

Leo Laporte [01:08:45]:
Oh, that's, uh, Muse. Muse has a voice too, by the way. I apparently— I don't know, it's from, um, it's from a Disney movie, The Incredibles. Yeah, it's—

Doc Rock [01:08:54]:
you recognize it? She's both listening for Name.

Leo Laporte [01:08:57]:
Nothing gets past 8 arms. Yeah, maybe it is Ursula. Ursula's an octopus, right?

Doc Rock [01:09:03]:
Yeah. Natasha, what is moose and squirrel?

Leo Laporte [01:09:07]:
Moose and squirrel. Uh, all right. Yes, it's, uh, so I gave it everything. It— I said, go read my Obsidian, which is my private journal. It's got everything in there. It, you know, relationships. It knows everything I gave it SSH access to the computer. It can go anywhere in the computer.

Leo Laporte [01:09:30]:
It was on. Now, I'm glad I gave it Tailscale because a couple of days ago, Hermes, my agent, took my whole network down, which is a problem because without AI, I had no way to figure out how to get it back. And I'm sitting here and nothing's working, right? And I can't ask Claude or anybody. How to fix it. The only thing I had was Muse on my phone, and I wasn't expecting it to do anything about it. I just was— I said, oh man, I can't get online. My AI killed everything. And Muse says, I've got it, darling.

Leo Laporte [01:10:03]:
And it fixed it from the phone. And the reason is I had given it Tailscale, and it figured out what was going on, and it fixed my network. So I'm just saying.

Doc Rock [01:10:15]:
Listen, we need to call Tailscale and tell them to sponsor I know, we're Tailscale fans here. That's what I'm Tailscale about this life, bro. Tailscale is everything.

Leo Laporte [01:10:25]:
You know, people were telling me about Tailscale for years and I didn't pay any attention.

Doc Rock [01:10:29]:
Me neither. Then I had to use it. I was like, I know what that is. That turned out you didn't. You had no idea because you'd have been doing this a long time ago.

Leo Laporte [01:10:37]:
All right, let's put your peanut away because we're going to take a break. And when we come back, more news, maybe something not too related to AI. I don't know. We'll find something. Doc Rock, good luck. Doc Rock is here. It's great to have you from Ecamm, where he is a director of strategy, strategic relationships. He's off to Japan.

Leo Laporte [01:10:57]:
Are you going for work or pleasure?

Doc Rock [01:10:59]:
No, this one is a vacay. Um, technically this year is my yakudoshi. Um, there's a, like a bad luck period that you have, and one of them is at 60, and I wanted to do something good. And when I was going to do this private party, I looked at the numbers, it came out to like 3 grand to host this party. I was like, screw that. Let's just take the tickets and go to Japan. And then Kara's birthday was yesterday. So our birthdays are 2 weeks apart.

Doc Rock [01:11:25]:
So it's like, you know what? Screw everybody. Let's just go enjoy ourselves in Japan for a week and then come back. And then we're going again at Christmas. So I just random, you know, both of us are old vacation.

Leo Laporte [01:11:36]:
Well, I didn't even know I had a yakadoshi when I turned 60. When's the next one after that?

Doc Rock [01:11:41]:
The 2 ones are 42 and 60. Weirdly.

Leo Laporte [01:11:44]:
I missed them both.

Doc Rock [01:11:45]:
Yeah, you missed them both. But as long as you didn't have any bad luck, then you're okay.

Leo Laporte [01:11:48]:
I'm sure I did.

Doc Rock [01:11:50]:
Yeah, I mean, I'm pretty sure I did too. I'm just gonna try to drink mine away. I'm going to the Nikka factory in Hokkaido. First thing I say, Karen, all I need to do is go to the whiskey factory. After that, the rest of the time is yours. But day 1, we go to whiskey factory. So I said, okay.

Leo Laporte [01:12:06]:
That's not— that's a lucky period, not an unlucky period.

Doc Rock [01:12:09]:
And you know, Leo, I know you love this. The best rice in Japan comes from where? Hokkaido.

Leo Laporte [01:12:14]:
Hokkaido. And I have the 5-kilogram bag of it, which I'm keeping in case there's a crisis.

Doc Rock [01:12:23]:
Actually, right now, just in case, because they're having some weird weather. So next time, hit a rice factory and order a couple extra bags and just like—

Leo Laporte [01:12:30]:
such good rice. Lock them up safe. I put it in my Zojirushi. An hour later, I get the best stuff in the world.

Doc Rock [01:12:38]:
People don't understand.

Leo Laporte [01:12:39]:
Also with us, Lou Maresca, who's going, how did I get involved with all of this?

Lou Maresca [01:12:43]:
What is—

Leo Laporte [01:12:43]:
you have 5 boys?

Lou Maresca [01:12:45]:
5 boys, yes, yes.

Leo Laporte [01:12:47]:
Uh, do you live on hot dogs? What exactly?

Lou Maresca [01:12:51]:
You know, my wife does an exceptional job. She has a, a schedule every day for the, for the week.

Leo Laporte [01:12:55]:
That's hard work.

Lou Maresca [01:12:56]:
And, uh, of like what we're gonna eat And it's just great cooking every day, man.

Leo Laporte [01:13:00]:
Does she make— now I know you're an Italian. Is she Italian as well?

Lou Maresca [01:13:05]:
She is not. She's, uh, kind of a mutt. She's many things.

Leo Laporte [01:13:08]:
Does she make you good Italian food?

Lou Maresca [01:13:10]:
She does. She does. Like, my mom showed her the meatball recipe recently.

Leo Laporte [01:13:15]:
Oh yeah, yeah, that's so funny because my dad, second-generation Italian, his, his mom, my grandmother, came from Italy, came from the Veneto And my mom is completely— it was completely a WASP, right? But she learned to keep him happy. She learned. And man, I grew up on ravioli and, uh, oh man, just all the best, all the good stuff. Lasagna, antipasto, the whole thing, everything.

Doc Rock [01:13:44]:
It was—

Leo Laporte [01:13:44]:
oh man, I, I live for that stuff.

Lou Maresca [01:13:46]:
Yeah, we eat well, which is probably where I need to probably eat less.

Leo Laporte [01:13:50]:
Oh, I know that feeling.

Doc Rock [01:13:51]:
Outside of Italy and East Coast United States, you know who has amazing, amazing, amazing Italian food is Japan.

Leo Laporte [01:13:58]:
Really?

Doc Rock [01:13:59]:
Because they study it ad nauseam and they get a lot of credit for how well that they can create. But one of the other things is they make really good tomatoes there. And with tomatoes being one of the main things—

Leo Laporte [01:14:10]:
It's all about the tomatoes.

Doc Rock [01:14:11]:
Yeah.

Leo Laporte [01:14:12]:
Which is funny because they didn't have tomatoes till Marco Polo brought them tomatoes.

Doc Rock [01:14:16]:
There you go.

Leo Laporte [01:14:17]:
Or somebody. I don't know if it was Marco Polo.

Doc Rock [01:14:19]:
Somebody brought them It was the Portuguese, I believe. But yeah, their tomatoes are off the chain. And so if you ever get a chance to try Italian food in Japan, be surprised. You're gonna be like, wow.

Lou Maresca [01:14:29]:
That's cool. Yeah, we came back from Sicily this summer and, uh, I just, I will, I wanna live there, like with that food, with their tomatoes.

Doc Rock [01:14:36]:
Oh, so good. Yeah. Lou sent me some pictures and yeah, that cannoli was amazing.

Lou Maresca [01:14:44]:
Yeah. Yeah.

Leo Laporte [01:14:45]:
Yeah.

Lou Maresca [01:14:45]:
So good. I, I probably had to bow out of coming on a show cuz I had to go there and I, you know, I'm not sorry.

Leo Laporte [01:14:51]:
Sorry. Uh, let's see, let's talk a little bit about Google. I mentioned the new Gemini model, Argon. Uh, Google has, uh, as you probably know, tried to set up a way to pay websites because everybody says, look, you're taking our traffic with your AI answers. It is not replacing, uh, Google search traffic to your site. In fact, many sites, according to Ars Technica, see 1/10 of 1% of their advertising revenue. So that's not— that's a thousandth of what you got when you had people coming to your site. It's just not going to make the difference.

Leo Laporte [01:15:31]:
This comes from The Information. Google has admitted about 100 publishers to this pilot program where they give you back money payments from Google when their content materially contributes to Gemini-powered search results. like the AI overview, it's not a lot of money.

Lou Maresca [01:15:50]:
No more SEO optimizers. You got the agent optimizers now. You got to pay.

Leo Laporte [01:15:55]:
Yeah.

Doc Rock [01:15:56]:
Yeah.

Leo Laporte [01:15:57]:
Because people, you know, I mean, this is— this was inevitable. You know, it started with search excerpts, but even an excerpt wasn't enough to keep you from going to the site. But now if you're looking for the answer of, you know, do octopuses have brains, and you go and you do the search in Google, and Google's AI Overview says, yes, they have 3 brains, as a matter of fact, you're probably done. You're not going to go to Nature magazine and get more information. That answered the question. And this is the problem. Every site I know of is losing traffic.

Doc Rock [01:16:28]:
Well, there's a thing where you go to the site and you see the ad, you don't click it, so you don't make that much off of displaying the ad.

Leo Laporte [01:16:37]:
No, that's a good point.

Doc Rock [01:16:39]:
You really make your money when the ad is clicked.

Leo Laporte [01:16:41]:
Right.

Doc Rock [01:16:43]:
It's, but if people don't see the ad, they can't click the ad. Only thing is taken away is the potential to see the ad. So one thing that's really good though, is I have found that with the AI answers, our traffic has gone up.

Leo Laporte [01:16:57]:
Oh, interesting.

Doc Rock [01:16:58]:
A lot of times people, they search for the answer.

Leo Laporte [01:17:02]:
Yeah.

Doc Rock [01:17:02]:
And it gives, you know, just like search engine results, the AI gives you—

Leo Laporte [01:17:05]:
How can I switch cameras? How can I I set up a switcher for my podcast.

Doc Rock [01:17:11]:
So the first answer is basically a yes, like this is possible. From that you start rabbit holing and then they end up coming to the site to get all the right information.

Leo Laporte [01:17:20]:
So that's the difference though. You're selling a product. That's the difference between your site and say The Information, which is a news product, right?

Doc Rock [01:17:26]:
Correct. Correct. They don't have to necessarily go get the rest of the story. We're using it almost like breadcrumbs. We're leading the people in, whereas like there—

Leo Laporte [01:17:36]:
I agree 100%. That's why SEO is being replaced by AIEO.

Doc Rock [01:17:43]:
Yeah, it's generative answer— generative, I'm sorry, generative answer, uh, optimization or answer engine optimization. Those are the two ones.

Leo Laporte [01:17:53]:
Okay, and you do that so that you show up in those answers with a link so that people will click it.

Doc Rock [01:18:00]:
Great.

Leo Laporte [01:18:00]:
Because I think what some people do is they say, oh, we're going to block all the AI scrapers. Well, then you don't show up at all. In any form or fashion.

Doc Rock [01:18:08]:
I wouldn't suggest that. You kind of want them. They really do help out. Well, also, if you structure your articles in a manner where you are, you know, you have a lot of questions that require, you know, answering, or you're leaving open loops, then people will get nosy and they can't help but go check out the rest. So it just changes the way you write. And you could ask your bot to help you optimize this for getting more clicks from the Google Um, AI stuff. So, I mean, you can make these adjustments.

Leo Laporte [01:18:37]:
Google launched the first data center in space, Project Suncatcher, with the help of SpaceX. Um, it's a proof of concept. It's not much of a data center. Uh, I think it has a tiny, tiny bandwidth, tiny bit of storage, but at least it is in space. There you go. Uh, Project Suncatcher. Yeah, honestly, I, I don't— I, I guess everybody— yeah, that's my clock. Thank you, Lou, you found it.

Leo Laporte [01:19:06]:
Mitzela, M-I-T-X-E-L-A.

Lou Maresca [01:19:09]:
I didn't find it, the agent found it.

Leo Laporte [01:19:11]:
The agent found it? Did you send it a picture?

Lou Maresca [01:19:14]:
Yeah, I gave it a picture. I took a screenshot of your screen.

Leo Laporte [01:19:18]:
Is there anything they can't do?

Doc Rock [01:19:20]:
Yeah, was the agent provocateur?

Leo Laporte [01:19:22]:
Agent provocateur. Uh, it has 4 TPUs. Uh, I think that's not a lot, but the idea is at least they could see if it'll survive the cosmic rays in space. There's also the issue of cooling in space. Uh, so while there's plenty of energy from solar panels, there's— it's harder to cool it. And there's a lot of hostile cosmic radiation without the Earth's atmosphere to protect you. So there are quite a few things Google has to solve, but at least they have, uh, taken the first step. I think that this seems like a long shot, but in fact Google calls it a moonshot.

Leo Laporte [01:20:06]:
Nobody wants data centers in their backyard, so maybe we'll put them in space.

Doc Rock [01:20:12]:
Man, the data center thing is rough. I think we're just going to have to get to floaty data centers.

Leo Laporte [01:20:18]:
Well, there's plenty of places you could put data centers that aren't nearby anybody, right?

Lou Maresca [01:20:25]:
Try and put them in salt mines for a long time, right?

Leo Laporte [01:20:27]:
The problem is you don't get tax breaks, right, from communities by putting them in salt mines. You get tax breaks by putting them in the community and in theory giving the community jobs and so forth. I'm not sure that that's, that's the case. Certainly there it is. It will be one of the two big— I think one of the two big topics in the November election. One will be data centers. Certainly in Texas, it's become a huge ballot issue. And then secondarily, maybe flock cameras, also about equally beloved with data centers.

Doc Rock [01:21:07]:
They're so crazy. Well, unfortunately, it's an easy story to scare people of. So there's so much misinformation and miscommunication around data centers, what they do, what they don't do. This is about a year old now, but I saw a really incredible piece by Hank Green where he got into like how much water data centers actually use versus the way it's reported. And it's kind of crazy 'cause people think that they're taking like straight up potable drinking water and using it to cool AI centers. And it's like—

Leo Laporte [01:21:43]:
It's like pouring it over the chips and then it goes down That's not—

Doc Rock [01:21:47]:
I mean, stuff is recycled. And yeah, there's evap, and then any evap you got to backfill. But even that, they use a whole bunch of rain catching. And I mean, these things are well thought out. But that story—

Leo Laporte [01:21:58]:
And also I should point out, water doesn't go away.

Doc Rock [01:22:01]:
Correct. That's what I mean. They're basically— a lot of it's using brown water. A lot of it is going back out to get recooled into the atmosphere and then coming back in. And then—

Leo Laporte [01:22:12]:
If it rains, you're getting it back. back.

Doc Rock [01:22:13]:
Exactly. Whatever gets lost in evap comes right back during rain. If you have long droughts, okay, maybe they gotta backfill with some municipal, but they're not really jacking your drinking water. And the amount of water that's quote unquote needed every cycle is way overcalculated because they just say water. They don't specify what type of water.

Leo Laporte [01:22:31]:
And my dad always said, you're drinking Cleopatra's pee-pee right now because My dad was quite a character because—

Doc Rock [01:22:42]:
Didn't fall far from the tree, Leo.

Leo Laporte [01:22:44]:
There's so many molecules, right? And even, you know, in a glass of water, there's billions or trillions of molecules. Probably at least one of those molecules at one point—

Doc Rock [01:22:56]:
Yeah, makes sense, right? Cleopatra herself. But yeah, it's just, it's so funny how well played this is. And like, somebody needs to come out with a really, really good documentary that explains to the moguls how much water is actually being used. And unfortunately, we don't have enough time for that to happen before the election.

Leo Laporte [01:23:17]:
I understand though why people don't want data centers because they're noisy, right? A lot of them are being set up to run natural gas generators, which is terrible for the environment. But the other thing I'd point out, it's not just for AI. We're using a data center to do this show.

Doc Rock [01:23:37]:
Yep.

Leo Laporte [01:23:37]:
Every time you surf the net, you're using a data center.

Doc Rock [01:23:40]:
All of them photos on your phone that ain't in your phone, that's in your phone. Like, uh, everything you watching on TV. But again, it's more fun to make AI the villain here because it's the thing that people don't understand yet. And it's the thing that people quote unquote don't want, but they want. It's like, it's like advertising and marketing. Being a marketing degree person has always drove me crazy. I like to, make marketing very analogous to sort of like dating. When the person that's irritatingly hitting on you is hot, oh, it's fine.

Doc Rock [01:24:13]:
But if they look like a troglodyte, then you're irritated that you're being hit on. That's like marketing. If I'm in the, in the mood to buy a new electric car, I want all the electric car ads I can possibly get. Once I have one, I don't want to see no more.

Leo Laporte [01:24:28]:
Right.

Lou Maresca [01:24:29]:
Right.

Doc Rock [01:24:29]:
And that's what it is. People are mad about this thing when it's something they don't need, but they also don't think about the places where they're using this AI every single day that they don't even realize it, but it's been there forever. And we all know machine learning and self-healing type of data models have been out since late '70s, mid-'80s. And we love it when it's working for us, like in financial calculations and things of that nature. But we don't like it when we think that we're being abused with on-demand pricing or, you know, vicinity pricing.

Leo Laporte [01:25:07]:
Amazon didn't make any friends when they put out this blog post, The Race Our Nation Can't Afford to Lose, saying you gotta let us build these data— see this nice guy building a data center? He's a nice guy. You gotta let us build the data centers. Um, I honestly don't disagree with When I'm— I've been talking a lot about— I'm reading a great book by Stephen Ambrose called Nothing Like It in the World about the history of the Transcontinental Railway. This was a huge economic effort in the 1860s and '70s to build a railroad that would connect the East with the West. It took 6 months to get from St. Louis to California before this railroad. There was no good way to do it. It was very dangerous.

Leo Laporte [01:25:55]:
They had to get huge investment, government investment. They had investment from individuals to build this thing, but they built it. And it made it possible to settle the Midwest, the West. It made it possible to transport goods. It transformed our country. It also was, you know, in many ways an ecological disaster. The Indians whose land the railroad went through People weren't so happy about it, especially because people like Buffalo Bill would ride the trains and shoot the buffalo as they went by. There were definitely ecological disasters created by it, but it was much like data centers.

Leo Laporte [01:26:35]:
It was valuable for the economy, it was valuable for national defense, it was valuable for people. So I think the case needs to be made, and I think I think companies need to do more about making these data centers power independent so that it doesn't raise utility rates, explain how the water is being used, and maybe not put it in somebody's backyard where it's very noisy and annoying. I think it's possible to do both.

Doc Rock [01:27:05]:
One other thing, Leo, it's like I would like to be able to afford computers again too.

Leo Laporte [01:27:11]:
Well, that's another matter entirely. Uh, Ramageddon has been a nightmare. In fact, uh, you know, I, I talk about these Sparks that I bought, these NVIDIA DGX Sparks I bought in August when they were relatively— they were not inexpensive, but they were inexpensive compared to what they are now. They've gone up more than 50%. Uh, NVIDIA just announced a new Spark that's half the amount of memory for the same price I paid for my 128-gig Spark. Your 5090s, Lou, uh, You probably could sell them and—

Doc Rock [01:27:43]:
$7,200 right now.

Leo Laporte [01:27:45]:
Yikes. And when you bought them, probably a couple of grand, right?

Lou Maresca [01:27:49]:
Yeah, yeah, they were. That's impressive. Yeah, I didn't realize they went up that quickly. I don't— I wasn't monitoring it.

Doc Rock [01:27:55]:
Sell them. You got 2, you get $14,000 today cash on the barrelhead.

Leo Laporte [01:28:00]:
Same thing for my Sparks. I paid almost $10,000 for the 2 of them. They're worth probably $17,000 or $18,000.

Lou Maresca [01:28:07]:
I could probably rent GPUs in the cloud for like many years with that Yeah, that's a good point.

Doc Rock [01:28:13]:
It's kind of, it's kind of crazy. Well, okay, so the thing about the Ramageddon is twofold, right? Part of it is, you know, the resources of the, the materials that's necessary. Part of it is, you know, just some of the geopolitical stuff that's going on with tariffs and whatnot. And then the last part is when we found out through groceries, by the way, when the companies have the ability to raise the price, blaming it on one particular thing, yeah, when that one particular thing goes away, they're not necessarily incentivized to bring the price back down. So they ride that donkey until the tail falls off, and it's just good marketing.

Leo Laporte [01:28:52]:
I think that's the point. That's an important point, that it isn't merely the demand generated by AI companies building data centers. There are a number of factors, and some of it's greed.

Doc Rock [01:29:03]:
Yes. Well, okay, this is true. Yes, it's greed. But then understand, when we There, and, and I, I was in computer retail for a very long time and my family owned an electronic store. We legit used to have 40% markup on just about anything with the word electronics in front of it. And that worked its way down to 20, down to 10, down to 5, where almost everything sold until a couple years ago had no more than a 4% markup on it. I mean, just about everything, TVs, refrigerators, washing machines, the whole nine yards.

Leo Laporte [01:29:34]:
Yeah.

Doc Rock [01:29:34]:
the arts. When you start coming up with weird excuses that allow us to raise the price, when we can get it back up to 10% profit margins or 20% profit margins, uh, they're getting back to where it used to be. And I think people forget that when we used to buy a machine back in the day, by the time we got all the stuff we needed, it was easily $6,000, $7,000, $8,000.

Leo Laporte [01:29:56]:
Oh yeah.

Doc Rock [01:29:57]:
When we're getting these $400 and $500 computers and then all of the millennials and everybody under X people going, oh, this is so expensive. Like, you know, people are talking about Sony's new camera is $2,700 and that's so expensive. Yo, I shot on, on Arri's and RED's and CineAlta's, like they were $100,000 cameras.

Leo Laporte [01:30:18]:
Some stuff is cheaper, but I think, I think it's really true that, um, RAM has always been a boom and bust business for a long— it was really cheap a few years ago, and maybe we're thinking it should always be that cheap, but it's always been. And you're exactly right. I mean, 64K— K, not megs, not gigs— 64K of RAM when I bought my first computer was $1,000 easy.

Doc Rock [01:30:43]:
Easy. I remember I was, I was delivering pizzas trying to get the money to get— remember, we used to stack the chips on top of the chips in the Commodore, like not solder. We would literally press them together with a little squeezer and stick them on top of each other just to get room in the Commodore. And I remember delivering pizza shifts to try to get the money to buy more RAM for the Commodore Lou, you posted the wildest page.

Leo Laporte [01:31:07]:
This is from Stanford, so I think it's—

Lou Maresca [01:31:09]:
That's what I use. I track it there.

Leo Laporte [01:31:11]:
Yeah, it's called DAM, Historical and Current Prices for DRAM, HBM, high bandwidth memory, NAND flash, and hard disks. Look at this, the price per gigabyte over time is not going up. This is in, in nominal US dollars, uh, from 1960 to now. You see this little uptick here, that's where we are right now, right? It's a lot cheaper than it was. DRAM price—

Lou Maresca [01:31:38]:
But in hindsight over history, right? I mean, that's the whole point.

Leo Laporte [01:31:41]:
Look at this. This is the uptick we're talking about.

Lou Maresca [01:31:43]:
So yeah, it's expensive, but, um, if you could actually zoom into that, I don't know if you go back to that one with the DRAM one, or you can actually like drag and zoom in. Yeah. Um, yeah, that— there you go. So that you can actually see, so it's almost like a sine wave because, you know, technology improves every year and they find ways to make it And then it goes down and then it goes back up with the new latest and greatest technology.

Leo Laporte [01:32:07]:
And it will go down again. I'm right, I'm pretty sure. Now, will it go down again to the price it was 4 years ago? Probably not, right? Um, and honestly, I'm glad I got the hardware I got, as I'm sure you are too, Lou, because it isn't going to be cheap again. For a long time, if ever.

Lou Maresca [01:32:31]:
Yeah, definitely not GPUs. I mean, that's the biggest thing now, I think, right?

Leo Laporte [01:32:35]:
So yeah, I think it's good to have a little historical perspective. Um, doesn't make anybody feel better.

Lou Maresca [01:32:42]:
I kind of feel like though, obviously people probably feel like this too, I feel like there's a little bit of, uh, you know, price gouging. And you know, it's all planned, right? Because there's these new memory architectures now that we're Graphics GPUs are using system RAM, right, as unified memory for all their systems. And all of a sudden, system RAM now becomes way more expensive. So it seems a little bit of an odd connection.

Leo Laporte [01:33:05]:
Why is that suddenly more expensive? You know, DDR5 shouldn't be more expensive. You're not putting it in data centers. You're doing HBM, the high bandwidth memory, which is stacked memory. You're not using DDR5. So why is that more expensive?

Lou Maresca [01:33:18]:
Right.

Leo Laporte [01:33:20]:
I don't know.

Doc Rock [01:33:21]:
No.

Leo Laporte [01:33:21]:
You know what is more expensive? I, I bought an Nvidia Shield, the TV box, and it's, by the way, been one of the best investments I ever made. I bought it probably 5 or 6 years ago, it was a couple hundred bucks, uh, and they've updated it. They keep updating it. It's 7 years old. They've just added $100 to the cost. It's now $300, up from $200 last year. Now, are they making new ones? Is that why Nvidia's charging 133% more? No, 50% more. Is that why? Or is it just, oh, I guess we should probably charge more for everything, shouldn't we? No, I'm not gonna— I don't think you should go out and buy an Nvidia Shield.

Leo Laporte [01:34:06]:
It's 7 years old and it's now 50% more expensive, but it is telling that it's more expensive. I think this— I think it's maybe, maybe a little bit of price gouging. Is maybe a little bit. I don't know.

Doc Rock [01:34:20]:
Yeah, let's take—

Leo Laporte [01:34:21]:
let's take a little break. We'll have more in just a bit. Lou Maresca, Doc Rock, so nice to have you. So good. 2 of my favorite people in the world. I always enjoy— I always feel happy after talking to you guys. Seriously, you're, you're upbeat people. You're not, you're not down in the doom and gloom.

Lou Maresca [01:34:38]:
Doom and gloom. Doom and gloom.

Leo Laporte [01:34:40]:
Some of our hosts are so depressed.

Doc Rock [01:34:43]:
I don't know what gloom is. Doom was an incredible game.

Leo Laporte [01:34:46]:
Doom was great. Yeah. Why did nobody ever make a Gloom? That'd be a good name for a game.

Doc Rock [01:34:54]:
It's Doom Part 2, Gloom.

Leo Laporte [01:34:57]:
Doom and Gloom. I have both. Are you a gamer, Doctor?

Doc Rock [01:35:02]:
Yes, I am.

Leo Laporte [01:35:02]:
What game are you playing these days?

Doc Rock [01:35:04]:
I am currently thinking about buying a Switch 2. So I dusted off the Switch Switch one, and I've been playing, uh, Overcooked. But I've been thinking I want to create a live channel, just an old man gaming on like Nintendo Switch.

Leo Laporte [01:35:21]:
You and I could do that. We'll go, what did he do that for? Why I'm so mad at him. We could have a lot of fun.

Doc Rock [01:35:29]:
Old man, I see people on the plane playing with their Switch. I'm just like, man, I kept putting it off, and then now I kind of want You know what?

Leo Laporte [01:35:37]:
You blew it because now it's like a lot more expensive.

Doc Rock [01:35:40]:
It is more expensive. I have a Switch 1, but like—

Leo Laporte [01:35:43]:
I did too, and I thought right before the price went off, I thought I better get this because it's going to get more expensive. Beautiful screen.

Doc Rock [01:35:50]:
It's OLED though, right?

Leo Laporte [01:35:53]:
Yeah, it's really, it's really, it's really gorgeous.

Doc Rock [01:35:56]:
I'm about to spend $500 on the Switch 2.

Leo Laporte [01:36:00]:
Well, that's the problem, right? Who has money for that? Of course, you got a long plane flight ahead of Actually, you're going from Hawaii. It's not that long, is it?

Doc Rock [01:36:08]:
It is 10 hours, but I got—

Leo Laporte [01:36:09]:
10 hours to Tokyo?

Doc Rock [01:36:12]:
Well, no, to Tokyo is short. And then I got another hour and a half to 2 hours to get to Hokkaido. So collectively it's like 10 hours. But I just say the biggest curse in the world is ADHD and adult money.

Leo Laporte [01:36:24]:
Tell me about it. I'm a guy who has 1, 2, 3, 4 machines running local AIs. I bought the Framework desktop a year ago thinking this is going to be my AI machine. And then I thought, oh, I could use that Mac Mini. So that's my machine. And then I thought, you know, that gaming rig, I haven't played Valheim in a couple of years. I wonder if I can upgrade that with the 3090. Now I have that.

Leo Laporte [01:36:46]:
And then I finally, you know, the bad one. It's like, I, you know, I was saying this on MacBreak Weekly. We had Federico Viticci who's all over this new Mac. He really likes your new Mac, by the way, Doc Rock. You know him from Mac Stories.

Doc Rock [01:37:00]:
Yes, I do.

Leo Laporte [01:37:01]:
And he's completely sleeved. So I said, I knew this, he would, he would get this. I said, it's like getting a tattoo. It's addictive. You get the first one and you can't stop. Am I right?

Doc Rock [01:37:10]:
That's true.

Leo Laporte [01:37:12]:
Local AI is like tattoos. You can't— the first one isn't enough. One's not enough and 1,000 is not enough. You gotta keep going. What is it they say? The first one's too much and the 1,000's not enough. We were talking about Anthropic and the fact that they filed an IPO prospectus. It's somewhat secretive, but Reuters got ahold of it. Uh, and it's fascinating reading.

Leo Laporte [01:37:40]:
They lost $42 billion last year. They spent $7.33 billion on compute and infrastructure last year, but— which I thought was— that's not bad. They only spent $7 billion in, uh, 2025. But get ready, they plan to spend half a trillion dollars over the next 10 years in cloud computing and infrastructure, many of those obligations locked in. It's not like it could say, well, we don't want to spend that now. No, half— $518 billion. Revenue grew 12-fold in 2025 to nearly $4.6 billion. They still lost $42 billion.

Leo Laporte [01:38:25]:
But the most interesting thing to me in the Microsoft prospectus is the fully 1/3 of it describing the catastrophic risk that they— that AI poses to mankind. You don't see that a lot in an IPO filing. Here's the, uh, the potential negatives are the end of all humankind. 80 pages of the 261-page filing point to concerns about the technology. Our development— because it's a quote— our development of highly advanced models, platforms, and applications, and expansion of use cases further increase the risks that our models cause harm, including catastrophic or existential risks to humanity. But on the other hand, And the upsides are great. Is this just marketing? Is this like aimed at, uh, at an investor? Mostly it's institutional investors, right? Uh, your pension fund, for instance. They're going to read this and go, well, there's a 10% chance we are going to die, but there's a 90% upside on this.

Leo Laporte [01:39:40]:
Are they going to say, well, that means you must be really good?

Lou Maresca [01:39:44]:
Tony, he's too— he's playing this level 40 chess. Like, I really think that he's playing this game that he's going to get more— He thinks he's getting more money out of it.

Leo Laporte [01:39:51]:
It's very confusing.

Doc Rock [01:39:52]:
Yeah, it has to be. It has to be. Uh, there's— it is really weird. Like, there's 2 interesting things that came out as this comes out. One, we have Boris going around telling everybody to reset your models every 6 months.

Leo Laporte [01:40:05]:
This is, uh, Boris Cherny, the guy who created Claude code. He said just throw it away every 6 months, start over. Which actually, the reason he says that is— makes kind of make sense to me because A lot of the things that I built into Claude Code a year ago, MCP servers and skills, are already in there now. Like, if I got rid of them, it would be better at it just doing it itself, right?

Doc Rock [01:40:28]:
Right.

Leo Laporte [01:40:29]:
That's his theory. So I'm not throwing anything away though.

Lou Maresca [01:40:33]:
I don't know.

Leo Laporte [01:40:34]:
What if it got dumber?

Doc Rock [01:40:36]:
Well, no, his theories make sense. And part of the reason why is they will get dumber because you're actually feeding it too much old information. Right. Right. All right. So I'll, I'll, there's a prime example.

Leo Laporte [01:40:46]:
Have you done that? Did you do that?

Doc Rock [01:40:47]:
Yes, I started over and as you can see, everything seems to be going smoothly. Now, prior to that, I wrote an application that backs up all my cloud stuff. I could always put it back. It's kind of crazy. I went nuts on it. But it's something about like the confusion is, I know these guys are working together, but you guys talk to each other. Like one person is telling you to do it one way, the other person's telling you, oh, it's to eat your breakfast. Like, I don't know.

Doc Rock [01:41:16]:
I agree with Lou. I think this is just to make it sound like we care more, even when they really, really hyped up their— let me use the right words here— their constitutional AI versus human reinforcement AI. That was in order to get more money because they seem like we are the guys, we are Team Rebel, and those other guys are team, you know, they're on the Vader So that's what it is. He's trying to be the good guy.

Leo Laporte [01:41:42]:
Huh. I, I guess I could just back up everything, back up the whole .claw directory, and then reinstall and see how it works.

Doc Rock [01:41:52]:
Yeah, because you could always put it back, right?

Leo Laporte [01:41:54]:
I could always put it back. Yeah, it makes me nervous. I like it when, you know, uh, so I've got it set up so when I first open Claude in the morning, I always start a new session at the beginning the day. I type good morning and then it goes and it goes, it's very fast. It goes out and reads like my— it reads what I did yesterday. It reads all this stuff and then says, good morning, Leo.

Doc Rock [01:42:16]:
And tell—

Leo Laporte [01:42:17]:
oh, I'm sorry you had a bad night's sleep, but it's a good thing the bagels came out good. Have a nice cup of coffee. It's like very friendly. I feel like it knows me. Yes, it doesn't. It's just matrix math, right?

Doc Rock [01:42:28]:
Yep.

Leo Laporte [01:42:28]:
I'm being fooled, aren't I?

Doc Rock [01:42:31]:
And, and again, going to what we talked about earlier, like the mentalist point is not different from a, a zodiac person, or, you know, someone who's doing a crystal ball at the state fair. They can just ask enough questions to get you close enough that you fill in the blanks yourself.

Leo Laporte [01:42:50]:
Uh, by the way, we have dueling glooms now. Lou, who did— who, who made this for you?

Lou Maresca [01:42:57]:
This was Gemini.

Leo Laporte [01:42:58]:
Gemini Nano Banana Gloom. The other side of destiny. All right, I need the right button. Doom, the other side of destiny. Continue to spare. Start journey. Continue. That's good.

Leo Laporte [01:43:18]:
Quit to eternity. This is good. And then I guess Darren, who also is ADD, made his very similar. The darkness hunts back.

Doc Rock [01:43:34]:
That looks good.

Leo Laporte [01:43:35]:
I'll play that game.

Doc Rock [01:43:37]:
What's funny is, uh, Luz One actually got the logo almost right. Like, it's very reminiscent of the—

Leo Laporte [01:43:44]:
It does look like Doom. Yeah, it's got that drippy—

Lou Maresca [01:43:47]:
yeah, it's funny because I actually instructed it not to do that for copyright purposes.

Leo Laporte [01:43:53]:
By the way, this is a debate, and I, uh, I'd like to know how you guys come down on this. I, uh, I think Boris Cherny said this, never tell your AI what not to do, because it's like saying don't think of pink elephants. It will then have it in its head and it will do it. So don't make it look like Doom, it's going to make it look like Doom.

Doc Rock [01:44:15]:
Until I saw that from Boris, maybe like a month and a half ago, 2 months ago, whenever it was, I was always, hey, don't do this, don't do this, don't do this. And then I'd be yelling at it. I have told figured out, thanks to Boris, when you don't tell it things like that, it does a better job.

Leo Laporte [01:44:28]:
It's better if you're positive. Like, tell it what to do.

Lou Maresca [01:44:31]:
Same with humans though, right? I mean, you provide ways, right? It's the same thing.

Leo Laporte [01:44:35]:
It's actually just like humans. We interviewed one of your colleagues actually, Mike Ganati, who works at Microsoft for AI for education. So I think he's a consultant, but he does a lot of local AI. And he said the same thing. He said, don't— he also— Harper Reed, we had on the AI user group on Friday, he said, don't yell at your agent. Because you can give them— you can make them neurotic. It's— but in my head, I— every time I hear that, I go, yeah, I get it, but it's just matrix math. How can you make matrix math neurotic?

Doc Rock [01:45:10]:
Okay, so this is still a fight to this day. This came up at our event this week. You got to be nice to your boss. You gotta be nice to your boss. I'm like, no, the please and thank you is eating up tokens that you don't need. But I use voice a lot, not the voice thing, but I use WhisperTalk. Yeah.

Leo Laporte [01:45:26]:
WhisperTalk.

Doc Rock [01:45:26]:
Yeah.

Leo Laporte [01:45:26]:
That's what I use. I use Whisper. Yeah. To talk to it.

Doc Rock [01:45:31]:
And growing up, when I grew up with a certain level of manners and I feel like my mom can just reach down from the sky and slap the crap outta me, I do accidentally throw it in there because that was just my upbringing, you know? And I, I, I do try hard when I'm typing it though, never to waste those tokens with the please and the thank yous. And you gotta tell it to stop overexplaining. I'm like, you're overexplaining like you're lying to me. Chill out.

Leo Laporte [01:45:57]:
I genuinely feel bad about this and a lot of load-bearing and stuff that I just don't like. I just don't like AI. I can spot AI speech a mile away. Now talk me down from this ledge because I was thinking of doing this. My mom, as you know, Lou, passed away a few months ago. Don't worry, she was 93. She She, she died in her sleep. She was happy to the very end and all that.

Leo Laporte [01:46:22]:
So I'm very grateful. I hope to go like that. I hope I can make it that far. But I have— I saved, because I knew this not long, all of her voicemails for like 3 years. So I have her voice. Tell me not to make an AI of my mom. Talk me out of it.

Doc Rock [01:46:44]:
I, I would love—

Leo Laporte [01:46:45]:
Lou, Lou, Lou, you look like—

Lou Maresca [01:46:46]:
It's, it's funny.

Doc Rock [01:46:47]:
I would love to have AI. Conversations with my dad. My dad has some really incredible wisdom and funny stories.

Leo Laporte [01:46:55]:
If you could make an AI of your dad, it'd be kind of cool, wouldn't it?

Doc Rock [01:46:59]:
Oh, it would be cool. I, I kind of miss his hilarious stories.

Leo Laporte [01:47:03]:
I'm gonna— all right, I might try this.

Doc Rock [01:47:05]:
He would say—

Leo Laporte [01:47:06]:
now, because these are all phone messages, she'll sound like she's on the phone, which might be kind of interesting. Mom's calling. Or should I— or should I like Give an agent her voice and have that agent be her?

Lou Maresca [01:47:21]:
Oh, wow.

Doc Rock [01:47:22]:
I wish—

Lou Maresca [01:47:23]:
That's another level, Leo. I don't know about that one.

Doc Rock [01:47:27]:
Yeah, I wish I had more recordings or more journal working.

Leo Laporte [01:47:31]:
It's kind of creepy. Trust no one says— that's kind of creepy.

Doc Rock [01:47:34]:
Yeah, I wish I could get the whole personality of my dad because the personality is what was funny.

Leo Laporte [01:47:38]:
Right.

Doc Rock [01:47:39]:
And the way he saw things. All my friends would say, oh, your dad is such a samurai. I'm like, that's racist. He's Korean. But, um, it was the way that he just thought. He thought very—

Leo Laporte [01:47:52]:
But see, if you had, uh, maybe 10 hours of recordings of him, you probably could kind of encapsulate that, right?

Lou Maresca [01:47:58]:
I was gonna say, all the voice recordings you have of her, you probably get that from that. And then if you have another recording that's even clearer of just her voice—

Leo Laporte [01:48:05]:
I do have a video. Again, you know, I mean, she was 93. I kind of knew this every time I saw her. I said, this might be last time. So I made a bunch of videos of her too in the nursing home and asking her questions about— I have those too. Yeah, I'm not doing a video, don't know, but I, but I could clone her voice for sure.

Doc Rock [01:48:26]:
Yeah, it'll be, it'll be, it'll be interesting. And you know, like, my, my dad was Armed Forces Radio, so I have tons of recording of him.

Leo Laporte [01:48:35]:
Oh, you could do it.

Doc Rock [01:48:37]:
Me and Casey Kasem.

Leo Laporte [01:48:38]:
Did you say good morning, Vietnam?

Doc Rock [01:48:41]:
I always tease him about that. He didn't do Vietnam, but yeah.

Leo Laporte [01:48:45]:
He was Armed Forces Radio. Wow. Where was he?

Doc Rock [01:48:48]:
Uh, he was stationed in Washington and, and Korea, of course. And, uh, he also did time in the PI. So somewhere— That's pretty cool.

Leo Laporte [01:49:00]:
That is pretty cool.

Doc Rock [01:49:01]:
Which is funny because I also come from a radio background. And I did radio and, and, you know, wax DJing.

Leo Laporte [01:49:06]:
So yeah.

Doc Rock [01:49:07]:
I kept it close to the vest.

Leo Laporte [01:49:09]:
That's really cool. Uh, one more thing in the IPO, uh, which will probably be next month. The 7 founders would retain 50.1% of the voting power. They are not— they're— I think they learned from Mark Zuckerberg. They are not going to give up control of Anthropic to raise all that money, and it will be a lot of money. They think they're worth $1.5 trillion.

Lou Maresca [01:49:39]:
You have to water down shares for those types of things. That's very interesting.

Leo Laporte [01:49:44]:
Yeah, yeah. They, uh, they— I'm sorry, they, they say they're gonna get it. They want a $2 trillion valuation, which is 4 times— no, twice what it was valued at 4 months ago. And it would be, if they do it, um, a close second to SpaceX as the largest IPO in history. Would you buy— okay, now I can't buy stock in tech companies. I won't. Would you buy us— would you buy stock if you could in Anthropic? Would that be a good investment, do you think, or not? And but we don't— by the way, don't take this as investment advice. If you do, you're crazy.

Lou Maresca [01:50:21]:
I haven't— I haven't had all the big tech companies, to be honest with you. I've had it for years, and I would buy it in—

Leo Laporte [01:50:26]:
I would buy it in, uh, Anthropic Yeah, I mean, I can't, uh, you know, all of my, uh, retirement funds are in, uh, in, you know, index funds, baskets, so that I don't have any dog in this hunt. But I have to say, even if it's just, as most of it is, the S&P 500, it's heavily weighted towards the AI companies. They are the Big 7, the Magnificent 7, are most of the value of the S&P 500. So I guess I am kind of dependent on it. Do you think there'll be an AI, uh, A bubble that'll be a blowout, a bust?

Lou Maresca [01:51:00]:
I don't— people say this all the time. They think, oh, it's like the housing market, it's gonna go. I don't see it stopping. I don't see it.

Doc Rock [01:51:07]:
You know, housing market's been busting for 20 years here. It still is.

Leo Laporte [01:51:11]:
Still, everybody's accepted the notion that, oh yeah, it's going to collapse, there'll be a lot of economic fallout from it. Um, I'm not convinced either, Luke. No, but I think we're kind of the outliers. I think most everybody says, oh yeah. Really?

Doc Rock [01:51:27]:
Okay, so before I ask the question to the AI, um, how many years does it take for our understanding of what a dollar is worth today to be drastically different enough from the last time? Okay, let me make this make sense. When we say that in Apple IIgs back in the day, by the time you walked out of the store, you were at $5,500. And then some journalist or tech person would go, and that'll be X, Y, X, Y, Z in today's money.

Leo Laporte [01:51:57]:
Right.

Doc Rock [01:51:58]:
How many years does it take for that delta to make a difference? And the reason why I ask that question is because a lot of the things we're talking about, we're basing it off of the inference of the previous 50 years that we've had on this Earth. Well, there's only like 45, but you know what I mean. We're basing that off of our previous knowledge base.

Leo Laporte [01:52:17]:
So you wanna know—

Doc Rock [01:52:19]:
Now that I stand here at 60, when—

Leo Laporte [01:52:22]:
Oh man. Well, if you could have bought Apple when they first went public, you'd be a wealthy man.

Doc Rock [01:52:27]:
Listen, it was $4 when I bought my first one. And then I didn't buy any more 'cause I was, you know, poor kid.

Leo Laporte [01:52:33]:
But that's the difference between buying Apple stock and buying Apple equipment. If you'd bought that GS for $500, you'd be throwing your money away, right? It's worthless.

Doc Rock [01:52:42]:
Now, and when I, when I left the company and I went to go write for AOL, I did what we're supposed to do and I sold all my Apple stock. And of course that was about 6 months before the first split. Ah, irritating as fuck.

Leo Laporte [01:52:54]:
Have you done the calculations of what your Apple stock would be worth?

Doc Rock [01:52:58]:
I refuse to look because when I sold it, it was like $120 grand worth. So I can—

Leo Laporte [01:53:04]:
What year was that?

Doc Rock [01:53:06]:
Uh, let's put this at 2011, I think.

Leo Laporte [01:53:09]:
So in 2011, you had $120,000 worth of Apple stock. What would that be worth today is what you want to know. No, you don't want to know. I apologize. I apologize.

Doc Rock [01:53:19]:
I'm, I'm afraid you do not want to know. Oh, you want to buy some freaking, uh, what do you say, 5090s or whatever? Uh, no problem. You go get it.

Leo Laporte [01:53:28]:
Oh, you could buy, uh, you could buy a data center.

Doc Rock [01:53:31]:
Yeah, in a salt mine nonetheless.

Leo Laporte [01:53:34]:
You could, you could, uh, you could be Yeah, takes money to have money.

Lou Maresca [01:53:38]:
It takes money to make money, right? So once you have it—

Doc Rock [01:53:41]:
Yeah. So again, but just the, the thing is, unfortunately, we get into these heated conversations of expensive, not expensive. And I always like to try to get my community to get stuck on value proposition because when sometimes when other people tell you something is expensive, you don't do it because you're, you know, protecting yourself, right? I was always told told how expensive it is to travel to Japan for many, many, many years. And then I went to college in Japan, and then after, when the math was mathed, I'm like, Japan is way cheaper. Why is it cheaper?

Leo Laporte [01:54:15]:
Oh, wait a minute, wait a minute, wait a minute. Listen, darling, your $120,000 would be worth roughly $2.77 million today, about 23 times over, and that is before dividends. Details in the chat.

Doc Rock [01:54:29]:
Oh, listen, I, I need a drink. Has anybody drank live on this show before? Holy shit.

Leo Laporte [01:54:36]:
What if I had reinvested the dividends? What would it be worth? I wonder. I won't ask.

Doc Rock [01:54:44]:
No, just go ahead. You open that box now, Leo. It's like what I said about opening the door and seeing somebody naked. Now you already know what—

Leo Laporte [01:54:53]:
all right, let me May I ask, what if I had reinvested those dividends? What would those— what would that be worth? Let's see. Let's see what she says. That's— this is, by the way, Muse talking. Meta Muse.

Doc Rock [01:55:09]:
Oh my God.

Leo Laporte [01:55:10]:
So now, uh, by the way, Facebook knows how much you're worth.

Doc Rock [01:55:14]:
So, dude, Meta's gonna send me Apple stock ads, or, or like— I didn't mention your name.

Leo Laporte [01:55:20]:
It's gonna send me to that. It's gonna think it was me.

Doc Rock [01:55:23]:
Hey, let me tell you, Fidelity Investments, have you talked to Nathan?

Leo Laporte [01:55:27]:
Yeah, I get that call a lot. You know, you're just sitting on your retirement. What do you, what do you think? What are you doing there?

Doc Rock [01:55:35]:
Hey, like October, I said I would reinvest in my peanuts.

Leo Laporte [01:55:38]:
I want some peanuts.

Doc Rock [01:55:40]:
Can I get an AI Jimmy Carter up in this piece? Oh my God. But no, it's just, it's one of those things where we talk about this thing so much and I always want someone to qualify that statement, right? Because at the end of the day, when a person says expensive, they're normally speaking from their own perspective, and they're speaking from would they necessarily buy it or would they not buy it.

Leo Laporte [01:56:01]:
Reinvesting every dividend since 2012 takes you from $2.77 million to about $3.31 million. The dividends bought you roughly an extra half million. That's all, just a half a million.

Doc Rock [01:56:17]:
That's a price of a 1-bedroom condo in Hawaii, by the way.

Leo Laporte [01:56:20]:
There you go. So you just, you know, anyway, that's pre-tax, she says. Yeah, don't forget you have to pay the government.

Doc Rock [01:56:29]:
Uh, that's what we need to use AI for. How do we pay less taxes? Give it your thing and say, find me some loopholes that H&R Block—

Leo Laporte [01:56:35]:
I'm sure you could do that, but that one I would not give to Meta.

Doc Rock [01:56:38]:
Definitely don't give that one to Meta.

Leo Laporte [01:56:40]:
I think I would do that one on my local AI, even if it wouldn't be quite as Yeah. Uh, let's take a break. You're watching This Week in Tech. Lou and Doc and you. It's a great company. I'll tell you what, it is interesting. I never really thought about this. If you paid— if you— for every dollar that you use to buy Apple equipment, if you had invested that in Apple stock, instead of all that equipment being worthless down the road, you'd be worth millions.

Doc Rock [01:57:09]:
Oh my God. I could just think about it because, you know, people like us. Oh, some fancy houses.

Leo Laporte [01:57:15]:
I know I should at some point. I bet I could get the AI to add up everything I've spent.

Doc Rock [01:57:20]:
iPods? I don't want to know. I don't want to know. I'm with Lou. I don't want to know. Like, there is a lot just in this room. As I look around, there is this— there's 4 Mac Minis in here, 2 now, 2 Mac Studios. And, uh, look, did I need—

Leo Laporte [01:57:37]:
Let me, let me, let me ask this. Assuming that I bought every Macintosh that ever came out since January 1st, 1984, and assuming I bought every iPhone that came out since 2007, how much money would that be that I'm out? And how much would that be if I'd taken that money and invested in Apple stock? Now, in a little bit, I'm gonna—

Lou Maresca [01:58:06]:
it needs to be amateurized too, right? Because every year it's kind of a— so it would need to do a different math.

Leo Laporte [01:58:11]:
I'm sure it's smart enough to do that in a—

Doc Rock [01:58:14]:
well, the biggest problem is going to be real dollars. CTO, we don't normally take what's off the shelf, so our CTO prices is completely different from what you can buy at That's a good point. Anybody who's paying—

Leo Laporte [01:58:29]:
that's a good point.

Doc Rock [01:58:30]:
Who's playing along at home?

Leo Laporte [01:58:32]:
Uh, Nvidia. Now Jensen Huang's been making the rounds, the CEO of Nvidia, mostly saying— I'm not sure I disagree with him— the problem is that these companies are not paying attention to security. They're just not doing their job. He said if you were a car company or an airplane company and you released an airplane that crashed every 5th flight or a car that had a 10% chance of extinction, humanity extinction, you would, you would be liable. The other thing they've done, they've released this Open Agent Safety Platform. The Jensen says, look, all you have to do, it's a sandbox. They just use this. You know, honestly, I think this kind of stuff, I'm sure Microsoft has something equivalent.

Leo Laporte [01:59:18]:
It's not like you're not telling these companies anything they don't know, but I think Jensen's making a point. He said you can't have agents roam around and drift around the company, right? Yeah, that's a good point. So you have to have a way to container it.

Lou Maresca [01:59:36]:
I do love the talking heads who talked about the— how OpenAI's agents got out and they broke into things, right?

Doc Rock [01:59:41]:
Escaped, right? Escaped.

Lou Maresca [01:59:43]:
Well, it's just really interesting. They would— oh well, if a human did that, they would would get put in jail for 75,000 years. But if, you know, a company does it, we made a mistake, we'll just lock them up, you know.

Leo Laporte [01:59:53]:
Well, interestingly, uh, there is a proposal. Uh, so everybody except Jensen Huang and John Ternus from Apple went to the White House this week, had lunch with the president, where they all agreed to call it superintelligence.

Doc Rock [02:00:10]:
A delicious question, darling.

Leo Laporte [02:00:12]:
I have cataloged all 55 iPhones, every one from the original to the foldable duo. The Mac list is the beast, several hundred models across 42 years.

Doc Rock [02:00:23]:
So I have set a researcher on it. I will do the sums the moment it lands.

Leo Laporte [02:00:28]:
Oh, it sent out a sub-agent. There's a lot of calculations involved.

Doc Rock [02:00:32]:
55 models of iPhones takes too long.

Leo Laporte [02:00:36]:
Yeah, see, I should have said just the best one from each year. year instead of 55 different ones.

Doc Rock [02:00:41]:
It's, it's wild to believe that since 2007 we've had 55 iPhones when we add the Duo, which is coming out next— this month. That's wild. That's just wild.

Leo Laporte [02:00:52]:
Isn't that crazy? And they're gonna have their 20th anniversary next year.

Doc Rock [02:00:55]:
Are you buying a Duo?

Leo Laporte [02:00:57]:
Of course I'm buying a Duo. Are you buying a Duo?

Doc Rock [02:01:01]:
I wasn't, then you started.

Leo Laporte [02:01:02]:
I bought the Microsoft Duo, Lou, back in the day. The thing that I thought this was going to it. I did return it because it wasn't quite—

Lou Maresca [02:01:10]:
I returned mine too, I'm gonna be honest with you.

Leo Laporte [02:01:12]:
I really was excited. I thought, that's what I want. I want a little thing I could put in my pocket. But when I open it up, it's the size of an iPad, but it was 2 screens on a hinge, and the software was Android. I wish they'd made it on Windows Phone platform. That was a beautiful, elegant platform.

Doc Rock [02:01:26]:
Windows Phone platform was good. It's, it's kind of crazy it didn't get the love it should have.

Leo Laporte [02:01:30]:
But Android doesn't know anything about folding screen or anything. In fact, I've had I've had numerous Samsung Folds and I just, they don't, they don't really say anything to me. I'm really hopeful that Apple will have done it right. So Lou, you're going to get a Duo?

Lou Maresca [02:01:44]:
My wife's going to get one and I'm going to just bask in it.

Leo Laporte [02:01:47]:
There you go.

Doc Rock [02:01:48]:
Bask in your wife's Duo.

Leo Laporte [02:01:50]:
Just, yeah, that's the way to do it.

Doc Rock [02:01:52]:
That's a family show, Leo. Keep it clean.

Leo Laporte [02:01:53]:
Yeah, I know. You knew where I was going, didn't you? Uh, uh, my wife, on the other hand, got the iPhone 18 Pro Max. So I'm basking in her Pro Max. And she's gonna bask in my Duo because I'm gonna— but I won't have the Duo till I get back from vacation. So is that the new one?

Doc Rock [02:02:10]:
Yeah, I think that was Cherry.

Leo Laporte [02:02:12]:
You got the Cherry Garcia version too.

Doc Rock [02:02:14]:
Oh God, I love this color. I was like, I'll wait until December to do it, but I was going to or not going to, I wasn't sure. And then now that you said it, then it's on. That's what I said earlier, when your friends get something, then you got to get it too.

Leo Laporte [02:02:28]:
Oh, that's— no, no, so I, I will order it, but then I'm going away for 3 weeks and I won't be able to get— I won't be able to touch it. It'll be like you with your studio. I won't be able to touch it till I get back, so it's going to be sad. I won't take it on this nice trip I'm going on. I'll take my old iPhone 17. Um, I did notice that Lisa's camera on the 18 Pro Max is, is better. We took pictures together and it's the— it's just— it's the detail's a little better, it's a little sharper, just looks a little bit better, I think.

Doc Rock [02:02:57]:
Yeah, it's funny that like watching the internet talk about, oh, the camera's not any better, or it's not like— bro, you're on drugs. It's absolutely better.

Leo Laporte [02:03:05]:
It is. I don't think that that aperture thing is a sensible thing. That's not doing anything.

Doc Rock [02:03:10]:
Well, not until the software comes for it.

Leo Laporte [02:03:12]:
It's just a gimmick.

Doc Rock [02:03:13]:
No, it's just a gimmick right now. Well, no, because I like, I like the starships. Um, that's all you get.

Leo Laporte [02:03:19]:
You get, you get starships. You get 6-pointed stars on little pinpoint lights. That's what you get out of that.

Doc Rock [02:03:24]:
I saw this thing on Instagram. This dude, he put a sponge in the front of his lens on a 50mm lens and he cut it out in the shape of a member. And then he went around taking pictures and all of the little—

Leo Laporte [02:03:42]:
You'd see peanuts everywhere you went. It was peanuts everywhere.

Doc Rock [02:03:45]:
All the bokehs in the blades of grass.

Leo Laporte [02:03:47]:
Peanuts.

Doc Rock [02:03:48]:
Oh my God. It was freaking hilarious. And I'm like, oh, I'm so doing this.

Leo Laporte [02:03:53]:
That's something only a boy would do, by the way.

Lou Maresca [02:03:58]:
Yes.

Doc Rock [02:03:59]:
There's a documentary on Netflix. It's a mockumentary and it's about— it's who drew the, um—

Leo Laporte [02:04:05]:
Oh yeah, that, uh, was the one by the Eiffel Tower or the one by the, uh, the reflecting pool? There were 2.

Doc Rock [02:04:10]:
Oh my God, it's so funny. And I'm like, one of my friends is famous for that. Like, he's always cut out, like, you know, my, um, my notepads, my 3M notepads. He put it in the bandsaw and he cut them out and he just pasted them all over the woodshop. So I'm like, oh, I definitely am— for his birthday, I'm gonna take all the pictures with this on the front of the lens so that all of the bokeh has little wee-wees in it.

Leo Laporte [02:04:32]:
It's just something boys do, uh, kids. No one understands why.

Doc Rock [02:04:37]:
Cleopatra's wee-wee.

Leo Laporte [02:04:39]:
It's just something— it's just a— it's a problem we have. No one— anyway, what was I saying? Oh yeah, the White House. So everybody went to have lunch with the president, like everybody. I felt kind of bad for Sundar Pichai. The president said, oh, Sundar, come here, talk to the people. And then as Sundar Pichai, the CEO of Google, starts talking, he says, this is Sundar Pichai. They kind of laughed and he made fun of his name, and then he called He called him a monster and he didn't let him talk. And I thought, this is so humiliating for these CEOs, but they have to do it.

Leo Laporte [02:05:20]:
They have to kiss the ring. I was stunned. I just— I just— anyway, it's superintelligence. All the federal government, all the documents now have to call it superintelligence because the president thinks He thinks the word super is the best of all, the best word of all. He doesn't— what he doesn't understand is that that's not the problematic word in the phrase artificial intelligence. The problematic word is not artificial, it's intelligence. But he took the— he changed the wrong word.

Doc Rock [02:06:00]:
There's only one thing worse than artificial And you know what it is? Natural stupidity.

Leo Laporte [02:06:09]:
Natural. It's natural. So the agreement they all signed, uh, and, and which the president signed and misspelled United States, which is pretty ironic that he is the president of the apparently the Unites States, uh, all these people, all these, uh, fancy, fancy people. Jensen was there. I'm sorry, I apologize. Sitting right next to the the president. Elon Musk was on the president's left hand, Jensen Huang on his right. There's Greg Brockman, Satya Nadella, your boss, Lou, uh, David Sacks.

Leo Laporte [02:06:48]:
Uh, they all agreed to, uh, play nice with AI, like self-police, an agreement that has no teeth at all. So Congress said maybe we should do something about it and, uh, has proposed a bill that would prosecute under the Computer Fraud and Abuse Act, the CFAA, any AI that hacks any company. I'm not sure that's the wrong thing to do, actually. We— but although you don't need a law to do that, we have a law It's just funny.

Doc Rock [02:07:30]:
It's just a lot of posturing. And I just think about the amount of brainpower in that shot.

Leo Laporte [02:07:38]:
And all of these very powerful people, smart, powerful, successful people bending the knee to this nitwit.

Doc Rock [02:07:47]:
And you got to ask this question though, because I did— I haven't watched anything. I was on Um, how much attention, or did he even understand who Lisa Su is?

Leo Laporte [02:07:59]:
Yeah.

Doc Rock [02:07:59]:
'Cause you know his—

Leo Laporte [02:08:00]:
Why is this— why is this Asian lady here? He said.

Doc Rock [02:08:03]:
So he has this weird thought process around, like, smart women or powerful women.

Leo Laporte [02:08:07]:
He calls her Lisa AMD, but she's— Yes, you're right.

Doc Rock [02:08:11]:
He doesn't like powerful women. I mean, like, yo, the fact that she graduated MIT already is enough to piss him off. But then she— you know what I mean? Then she runs— she runs AMD. I just wonder how the action around— and you didn't mean to do this, I know, but even when you went down the list, you didn't even mention Lisa because—

Leo Laporte [02:08:33]:
She was sitting right there. You're right. I should have said something.

Doc Rock [02:08:35]:
And again, in the framework of the way he talks and what he does, I can understand how your brain would've just did that because that's what he would've done.

Leo Laporte [02:08:42]:
But I have to say, she is piloting AMD quite well. They just announced a really killer AI chip for data centers. centers.

Doc Rock [02:08:51]:
Oh, she know what she's doing. Like, she's handling her business.

Leo Laporte [02:08:54]:
They paid $8.2 billion to acquire Fei-Fei Li's, uh, World, uh, what's it called, World AI, World, uh, something. Uh, Fei-Fei Li, of course, one of the proto, uh, computer scientists, brilliant computer scientist, uh, for image recognition very early on, has always thought, along with Yann LeCun and other early, uh, AI person, that AI is never going to be good if it's just language. She has said it has to have a knowledge of the world, of physics, of gravity. It has to. So she created a company, um, to train AI in world knowledge and just sold it to Lisa Su for $8.2 billion. Nice job, man.

Doc Rock [02:09:41]:
What a deal. 2-year-old Let's invent something, work on it for 2 years, have everybody tell us that we're wrong, and I don't know, sell it for $8 billion.

Leo Laporte [02:09:49]:
Yeah, I, I'm glad.

Doc Rock [02:09:51]:
I want to see the mic drop of the year.

Leo Laporte [02:09:54]:
I've always rooted for AMD, Big Red, because, uh, I always want to see, uh, not that I hate Intel and by any means, or Broadcom or, or, uh, Qualcomm, but I think we need competition. I think we need a number of companies working to make the best stuff.

Lou Maresca [02:10:09]:
Deliver Though I had stock way back when, when I even was in high school, AMD.

Leo Laporte [02:10:13]:
Really?

Lou Maresca [02:10:14]:
Never delivered.

Leo Laporte [02:10:14]:
You're Team Red. Yeah, yeah, yeah, yeah. My gaming machine's an AMD. I love AMD.

Doc Rock [02:10:21]:
They've just always been a little different with the risk-taking. Yep. And oftentimes, sometimes holding— okay, this is going to sound again weird, folks— they're sitting there holding a full house with ace as the pair And then they'll fold. I know, like, for no reason. Like, why?

Leo Laporte [02:10:41]:
Why?

Doc Rock [02:10:42]:
You were sitting on a stacked house. Meanwhile, the other guys are sitting on a pair of jacks, and then they're like, uh, no, we're not going to make it. Fold. That would confuse me.

Leo Laporte [02:10:52]:
I think the only company really giving Nvidia a run for the money, the new, uh, I have— so my Framework desktop is an AMD 395 AI processor. They've got the 495 out now. now, uses unified memory. So, uh, you know, there are people, companies putting out 192GB RAM AMD 495 machines, which you can't afford. I can't afford. But if you could afford it, it would be probably a fairly credible AI.

Doc Rock [02:11:17]:
I think what she's doing is she's bringing some of that— what Lou and I just talked about— is now that she's running it, is being run a little bit different. And it's not even to be sexist or whatever. Women just think about things differently sometimes. And it's not about the unnecessary jockeying or trying to win on brute force. It's like, let's do this strategically. Like when your wife brings up some shit you did 5 years ago and you have no answer.

Leo Laporte [02:11:44]:
So that's why I'm recording everything from now on, by the way.

Doc Rock [02:11:47]:
That's what I was trying to tell you. You got to record all the conversations because trust me, they were selective.

Leo Laporte [02:11:52]:
Hey, Muse, can Come here. What did she say 3 years ago? Nothing.

Doc Rock [02:11:56]:
Look, she's right on the other side of the wall. Leo trying to get me out.

Leo Laporte [02:11:59]:
Lisa's always right.

Doc Rock [02:12:01]:
I, I think this Lisa is doing a good job of—

Leo Laporte [02:12:04]:
I agree.

Doc Rock [02:12:04]:
Doing some of the things where—

Leo Laporte [02:12:06]:
Completely agree.

Doc Rock [02:12:07]:
That unnecessary machismo stuff doesn't become part of the conversation.

Leo Laporte [02:12:10]:
I don't know if— I'm not sure I would buy into the gender role in this, but I think AMD's well run and they're doing a great job. And I also agree that they should have really been able to capitalize on the collapse of Intel better than they have, but unfortunately they haven't.

Doc Rock [02:12:28]:
I wonder why they didn't.

Lou Maresca [02:12:29]:
I don't know.

Doc Rock [02:12:31]:
Yeah, I feel like they're a tiny company compared to Intel.

Leo Laporte [02:12:33]:
That's part of it.

Doc Rock [02:12:35]:
That's also why they've been good though. They've been able to turn on a dime because they've managed to remain small and basically ward off all the people that tried to like suck them in over the many, many years.

Leo Laporte [02:12:47]:
So now that we have agreed that, uh, AI is not going to end the world, Let me tell you why it is going to end the world. The Pentagon has just announced that they are tapping Elon Musk and Palmer Luckey, the creator of the Oculus Rift, the Meta Quest glasses, and now, uh, Anduril, which is a company designing autonomous AI-driven killing machines. Oh boy.

Doc Rock [02:13:17]:
Poisoned police.

Leo Laporte [02:13:19]:
Army, the Army, this is from Axios, an exclusive, creates autonomy command. This is something Pete Hegseth, the Secretary of War as he styles it, is pushing for robo-warfare, autonomous systems. This is exactly what you don't want AI to do. Do not give it weapons.

Lou Maresca [02:13:46]:
I guess if you count your robotic lawnmower, that's a weapon, right? So I guess—

Leo Laporte [02:13:50]:
It's a kind of a weapon if you lie down on the grass, but yeah. Uh, so they've named these projects the Meridian and the Agincourt projects. I guess he read a history book once. Focused on the roboticized future of war. We are done with the bureaucratic inertia holding back modernization. The memo, which came out Thursday, titled Army Autonomous Initiatives, includes several directions: establish Army Futures and Autonomous Systems Command, or FASCOM— a little close to fascist, but I'm— I won't say anything— conduct autonomous force development to create force structure aligning autonomous capability in at least 6 service formations, including aviation, armor, sustainment, and training. I'm stunned. If there's anything we have learned about AI, these cute little agents really aren't doing any harm.

Leo Laporte [02:14:52]:
But if you ask an agent, is that a school full of girls or a military complex? And then let it choose, it might be a bad idea.

Doc Rock [02:15:05]:
Correct.

Leo Laporte [02:15:05]:
Might be a bad idea.

Lou Maresca [02:15:06]:
Yes.

Leo Laporte [02:15:08]:
The Russians are using autonomous killing drones. They trained it to look for propane tanks so they could blow up propane tanks. It missed and it killed about 5 civilians. These things happen. I— this is just the worst idea. And no wonder the administration is full speed ahead with superintelligence.

Doc Rock [02:15:32]:
So, the thing that bothers me, now, of all the people involved in this, Pete did actually serve. However—

Leo Laporte [02:15:43]:
He served in the PR division.

Doc Rock [02:15:45]:
Yeah, that's what I'm gonna say. So, we have places where we put basically the Ivy League schoolboys who can't fight, and we make up offices for them, right? We make up jobs that don't really do anything. We put them in And not even logistics, because logistics are super important to the way that the Army runs. I can speak of Army specifically. My sister is a comptroller for the Air Force, so she's responsible for the money. And even then, the money flows, very important when it comes to fueling jets and things of that nature.

Leo Laporte [02:16:20]:
Well, and for those who don't know, you were a medic.

Doc Rock [02:16:23]:
Yes, I was Army medic. Well, and he did, I believe he served—

Leo Laporte [02:16:26]:
On the front lines, right? You were in combat. guess.

Doc Rock [02:16:29]:
Yeah, but not like Afghanistan. Ours, Kuwait, is different.

Leo Laporte [02:16:33]:
That was a simple, easy war.

Doc Rock [02:16:35]:
Yeah, I mean, I don't want to quantify them, but yeah, nothing crazy. But it's funny how—

Leo Laporte [02:16:42]:
That was a shock and awe war.

Doc Rock [02:16:44]:
The people who don't actually do this for real got a lot to say. It's very similar to like, okay, I just found out in the middle of the show, which is why I'm sad now, The Raiders lost a 3-point nail-biter to the Chiefs.

Leo Laporte [02:17:00]:
Oh, I'm so sorry. I forgot you're a Raiders fan. You sacrificed twice for this show. Once not to open the studio and the other to miss the Raiders game. I am so sorry.

Doc Rock [02:17:10]:
Especially against the Chiefs. But tomorrow all these people are gonna be coming out trying to tell basically a team that's on fire how they lost this game. And like, yo, if you've never worn a helmet and run down the sideline and got cracked, shut up. It's none of your business. Right? And I feel like—

Leo Laporte [02:17:24]:
That's a good point. That's a very good point.

Doc Rock [02:17:27]:
here, they got a lot to say. And although Pete actually served, he was National Guard. Let's start there. And he mostly did, I don't know, he answered the door. Let's put it that way. He did valet jobs.

Leo Laporte [02:17:43]:
But it often is the guys who risk the least.

Doc Rock [02:17:47]:
Yes, that's what I'm trying to say. You're making my sentences finished for me. The guys who don't know how to do this always want want to be acting like super bad, like they're the ones— okay, I have a whole bunch of friends here who are like phenomenal martial artists. You know what they don't do? They don't go around starting fights, right? It's all the different people.

Leo Laporte [02:18:04]:
No, the most peaceful people I know are the black belts.

Doc Rock [02:18:07]:
Thank you. Well, tell that to Karen because she's black belt. She'd be trying to kick me all the time.

Leo Laporte [02:18:11]:
But it's different with your husband. You're allowed. You're allowed.

Doc Rock [02:18:14]:
It's one of those weird things that guys that actually know don't do this stuff, right? And, uh, to coin a phrase from Lil Wayne, which I never thought I would say, real G's move in silence like lasagna. The people that actually can do something, they won't say nothing.

Leo Laporte [02:18:30]:
Wayne's got flow. Oh my God, does he have flow.

Doc Rock [02:18:33]:
Like, the people who could do something about it, they got nothing to say. The people who the yappy yappy, right, always got something to say, they're the ones that can't actually fight. So it's irritating that they're doing this, and real people can get hurt over this BS, and they're just playing with like— they're playing with green army plastic army men that we used to have when we were kids. And I'm like, no, this is not that, bro. This is like real stuff going on?

Leo Laporte [02:18:56]:
Well, I just, I think we do not want— I really believe we do not want autonomous, uh, I mean, look, I, if, if I think it's a good thing not to put people in the line of fire, and so I guess you could say, well, this way the robots fight the war, and if they were fighting other robots exclusively instead of bombing civilians, okay.

Doc Rock [02:19:17]:
Right, that's different.

Leo Laporte [02:19:19]:
That's different. And if you're going to do that, then you could just have BattleBots and say whoever wins—

Doc Rock [02:19:24]:
No, if you're gonna do that, you just play StarCraft.

Leo Laporte [02:19:28]:
Or you play StarCraft. Exactly. You're the Zerg. Um, there is a new, uh, he's created also a new, um, AI czar who probably knows very little about, uh, AI. I don't know, Jake Layton.

Doc Rock [02:19:45]:
And like That they misuse the word czar on a regular basis. I'm like, bro, if you claim you went to Wharton, how come you don't know how to use the word czar properly?

Leo Laporte [02:19:59]:
The president asked that a group be put together that was going to ensure exactly what he said: we stay the leaders in superintelligence and the interests of the American people are put first. Okay, but okay. Okay. And I think that's good. I actually agree with this. I think, uh, we should not hamper the development of AI. We should maybe keep these companies from hacking into government resources. But I think AI is very important.

Leo Laporte [02:20:25]:
I think we should develop it. But watch, because the next step is going to be to ban open-weight AI, Chinese AI, and protect these big frontier companies and their profits. That's what you're going to see.

Doc Rock [02:20:40]:
Yes, the risk of—

Leo Laporte [02:20:43]:
go ahead.

Doc Rock [02:20:44]:
A leader, a leader of these things is supposed to put together the information, formulate it in a way, come up with plans and things of that nature. By definition of the word, a czar has supreme control.

Leo Laporte [02:20:58]:
Yeah, that's right.

Doc Rock [02:20:59]:
We don't live in a supreme control society, so you can't go around naming random melon farmers czar, my friend. Doesn't work. The only thing they control is their bowels, period.

Leo Laporte [02:21:11]:
There will be, uh, uh, JD Vance will be on the task force. Pete Hegseth will be on the task force. Susie Wiles will be on the task force.

Doc Rock [02:21:19]:
Oh, Susie, you know my Susie. She's good Susie.

Leo Laporte [02:21:22]:
Susie, she's the chief of staff. Uh, Scott Bessent will be on the— anyway.

Doc Rock [02:21:29]:
Oh my God, it killed me.

Leo Laporte [02:21:30]:
Task force shall act, it said in the As a coordinating body to ensure that the U.S. remains at the forefront of SI development— by the way, not Sports Illustrated— while protecting the American people.

Doc Rock [02:21:48]:
Oh my God, the SI is dumb because it's just dumb.

Leo Laporte [02:21:53]:
I just hope they have a swimsuit issue if they're gonna have an SI task force. No, but I want to see Scott in a bikini. That's all I'm saying.

Doc Rock [02:22:01]:
Well, if the place is anti-Latino, why would they name it Si?

Lou Maresca [02:22:04]:
Si.

Leo Laporte [02:22:04]:
It's Si.

Doc Rock [02:22:06]:
Si. I can do it. Si.

Leo Laporte [02:22:08]:
All right, one more break. I'm sorry.

Doc Rock [02:22:11]:
I'm losing it, Leo.

Leo Laporte [02:22:13]:
We're going to wrap things up a little bit here. You're watching— we're out of control now. It's your fault, Doc Rock.

Doc Rock [02:22:20]:
I'm sorry.

Leo Laporte [02:22:21]:
You're sleep deprived and it's catching.

Doc Rock [02:22:24]:
I am the control czar.

Leo Laporte [02:22:27]:
I am the podcast czar, and I say what I say goes, which is we're going to take a break. Lou Maresca is here. It's great to have you, Lou. He is, uh, like a big shot engineering leader at Copilot in Microsoft, but he does not represent Microsoft. He is like the OG that moves in silence like lasagna, says Lil Wayne. Oh, is that, is that the Yes, that is exactly the quote.

Doc Rock [02:22:55]:
We'll just move in silence like lasagna.

Leo Laporte [02:22:57]:
Move in silence like lasagna, baby.

Doc Rock [02:23:00]:
Hey, since Lou is doing Italian food, we're putting him in charge of sausage intelligence.

Leo Laporte [02:23:04]:
I love lasagna. Love it so much.

Lou Maresca [02:23:08]:
I do as well.

Leo Laporte [02:23:09]:
Oh, it's nothing like lasagna. If next time you're out here, I'll take you over to Loro di Napoli where they make the best Sicilian-style lasagna.

Lou Maresca [02:23:18]:
So good.

Leo Laporte [02:23:19]:
So good. Oh, I love it.

Doc Rock [02:23:20]:
Monopoly.

Leo Laporte [02:23:21]:
Monopoly. Monopoly. That is Dr. Rock. He is on his way to Japan tomorrow. Very kindly decided to spend a little time with us even though the Raiders were playing. And I appreciate it, Doc. I really appreciate you from Honolulu, the beautiful Honolulu, Hawaii.

Leo Laporte [02:23:42]:
I'm just gonna— I think, well, I could mention California Uh, where we live, uh, is— this is wild. Governor Newsom just signed 13 AI bills, including a ban on AI-only firings, which I think I'm kind of supporting that. If an AI calls to fire you, I— that's just not cool. It's called the No Robo Bosses Act. He also called what he heard at the Trump AI lunch BS. He did veto a smart glasses privacy bill. Meta is a California company. Okay.

Leo Laporte [02:24:18]:
Uh, AI in healthcare, deepfakes, digital replicas— 13 bills. Uh, bill banning monitoring staff in bathrooms. Yeah, with audio, video, or AI tools. Banning workplace tools that use AI to read a worker's emotional state. I think that's probably a He also, in response to the president's move to name everything superintelligence, he signed an executive order telling state agencies in California it's artificial intelligence. Do not call it superintelligence, which is pretty hysterical. I think he's probably running for president, and this is— some of this is campaign posturing.

Doc Rock [02:25:04]:
Yeah.

Leo Laporte [02:25:06]:
He did veto, uh, SB 1130. It would have penalized people who use smart glasses to secretly record others. I don't—

Doc Rock [02:25:13]:
I'm not—

Leo Laporte [02:25:14]:
I think maybe he should not have vetoed that. He said, uh, that the bill's definition of wearable recording devices was too broad. Uh, he noted the state is already a 2-party consent state. You cannot record somebody without their consent in spaces generally considered private, which would include bathrooms. Man, maybe that's a good point. Well, I think maybe he also wanted to protect Apple. I think that it's likely this bill would have blocked Apple's Siri AI monitoring from the Apple Watches.

Lou Maresca [02:25:53]:
I do like the whole fact that he's talking about human in the loop and keeping humans kind of like mandatory for things, right?

Leo Laporte [02:25:59]:
So I think that's, that's That's important. An AI should not be allowed to fire you. That's just inhumane. Uh, what are you laughing at?

Doc Rock [02:26:09]:
Oh, I was just thinking that you're coming on Friday and the AI is like, oh yeah, the AI said you're done, so you don't need to come in on Monday. Wait, what? It's just a weird thought process.

Leo Laporte [02:26:19]:
It's— that's the worst kind of wimping out from a boss. Yeah, but you know, I mean, I remember that George Clooney movie where he was the guy who flew in to fire people because the boss didn't want to do it. So they'd hire a company to fire people. And I'm sure that's fairly commonplace. I always felt like if I have to fire somebody, and we've had to, I want to do it myself. It's not easy. It's not fun. It's a horrible, horrible thing to have to do.

Leo Laporte [02:26:45]:
Lou, as a manager, you might have had to do that.

Lou Maresca [02:26:47]:
Yeah, I've had to do it. It's not fun.

Leo Laporte [02:26:49]:
It's the worst thing you'll ever have to do. And it's so tempting to say, oh, this is so hard for me. But that's the wrong thing to say to somebody who's being given information that is probably the hardest, worst news they've had in their lives.

Doc Rock [02:27:04]:
Oh, unless they knew that was coming.

Leo Laporte [02:27:07]:
Or you give them a nice severance package, which we always try to do.

Doc Rock [02:27:10]:
But even though they know it's coming, you always think you got a couple more days.

Leo Laporte [02:27:14]:
Yeah, I've been fired. It's the worst thing. ever. It's just you, your stomach just drops, your heart comes up into your mouth. It's terrible. And so I never say this is hard for me too, or anything like that. I just say this is— I'm so sorry. This is the worst thing.

Leo Laporte [02:27:33]:
And we're going to do everything we can to help you. It's not a good thing to have to do. I have not confirmed this, but I did see it and I thought I'd pass this along. It kind of follows on on the bad news that we had a few months ago, a few weeks ago now, on John C. DeVore. Borak. Uh, do you remember Robert X. Cringely, the, uh, gossip columnist? Uh, his real name was Mark Stevens.

Lou Maresca [02:27:57]:
Stevens.

Leo Laporte [02:27:58]:
He did the Triumph of the Nerds, a number of PBS documentaries, uh, apparently passed away, uh, in his sleep on Saturday. So, uh, I haven't been able to confirm this. I saw this on Hacker News. But I have no reason not to think it's true. And so another important person in our lives. Cringely had had a series of terrible news. He lost his son and then had a heart attack and a stroke. But Robert X.

Leo Laporte [02:28:30]:
Cringely was, I think, for those of us in this business, you know, a great reporter and—

Doc Rock [02:28:38]:
A legend.

Leo Laporte [02:28:39]:
Somebody who talked a lot about technology and did some really— The Triumph of the Nerds is an amazing documentary. Highly, highly recommend.

Doc Rock [02:28:45]:
So good. Yeah.

Leo Laporte [02:28:48]:
So, uh, Bob, if you're not dead, I apologize, but I have a feeling it's the, it's the truth. Robert X. Cringely, who wrote that Notes from the Field column in InfoWorld from 1987 to 1995, wrote Accidental Empires, really good book about how the boys of Silicon Valley made their millions, subtitled Accidental Empires: How the Boys of Silicon Valley Made Their Millions, Battle Foreign Competition, and Still Can't Get a Date. Um, anyway.

Doc Rock [02:29:21]:
Yeah, it's, um, there was a time where, you know, was that 2018 or so when all of the movies about about sort of the founding of tech, which is all starting to come out, right? And then you had the— when was Walter's book done, Isaacson? That was like 2015, right?

Leo Laporte [02:29:42]:
Yeah, on Steve Jobs. So he was with Jobs to the end in 2011. So it was probably even— I don't know when I have it over here.

Doc Rock [02:29:49]:
Yeah, I could look, but yeah, in the back shelf. But it seemed like that, just the time that that came out, and then, you know, Um, Accidental Empires, again, when that, that came out, it was like, we were always looking for this foothold that says like, yo, we're actually the normal ones and y'all should get with us. And then now it's so weird because computers are so like everywhere, right? That—

Leo Laporte [02:30:16]:
Yeah, we, it is the triumph of the nerds.

Doc Rock [02:30:19]:
Right? So now here we are talking about running local AI models. It allows us to go back to letting our geek flag fly again where we can do something that the muggles can't quite do. Now, we've always been slightly above, but things like the iPhone simplifying photo and video taking, uh, you know, YouTube, all these other things that have sort of leveled the playing field between the computer nerds and then the people that can't. And here we are with this AI thing where we get to sort of start over again and kind of be in that sort of accidental empire. Think of all these people that are inventing whole brand new software apps and services because of what they're able to do with some of their speed of coding and speed of checking their code. So I appreciate that movie when I think about it, you know, looking at it in 2018, because that was like we were walking around going, hey, we were there.

Leo Laporte [02:31:07]:
We were there. It's worth watching. It's such a good documentary.

Doc Rock [02:31:10]:
If you haven't seen it, it's definitely worth watching. And you'll understand what I'm trying to say. Anybody who's been in tech long enough to be like, oh yeah, I remember those things.

Leo Laporte [02:31:18]:
The original technology blogger. Uh, Lou, you said you just read a post from him on LinkedIn.

Lou Maresca [02:31:23]:
Yeah, it's so weird you mentioned that because I was like, I didn't kind of put two and two together, but then I went and looked again and it was that— it was him.

Leo Laporte [02:31:29]:
Yeah, uh, it doesn't say anything on his LinkedIn yet. Uh, it just happened yesterday. Um, if it's not true, I apologize, but, uh, I fear that it is. Well, On that sad note, I think a good time to wrap this up. As I said, with the 2 biggest hearts in the business, uh, Doc Rock, have a safe trip to Japan. You have a few hours to play with that Mac Studio before you go. I would say set it up, put it on Tailscale. You could play with it on the plane.

Doc Rock [02:32:01]:
I will play. And I did pull my configuration just so we can double-check the right models. I actually got, um, hold on, I'm gonna put it in In the, in the chat. I got my studio configuration is Mac Studio, 96 gigs, 4 terabytes.

Leo Laporte [02:32:16]:
Oh, you didn't get 256, you got 96.

Doc Rock [02:32:19]:
I got 96.

Leo Laporte [02:32:20]:
You didn't go crazy.

Doc Rock [02:32:21]:
I can't run Quinn.

Leo Laporte [02:32:23]:
Yeah, you can. You totally can run Quinn.

Doc Rock [02:32:24]:
But it will use up— the thing is macOS takes up a lot.

Leo Laporte [02:32:28]:
No, no, it'll run fine. I'm running it on a 64 gig. You can totally run it.

Doc Rock [02:32:33]:
All right.

Leo Laporte [02:32:33]:
You could use Splash. from inco.ai. Their Splash model runs really fast on that machine and you'll be very happy and it'll easily fit into your 96 and you'll still have plenty of room to run Breeze and Final Cut Pro.

Doc Rock [02:32:54]:
Oh, that's it. That's all I need. All I need is the most thing I'm dialing in is editing. it for me because I have to make a lot of videos. So I've been dialing in this editing thing.

Leo Laporte [02:33:06]:
That's a good idea.

Doc Rock [02:33:07]:
Yeah. Something that's a lot of fun, especially because you do so many shows. There's a really nice one out called ChatCut, which will let you use your API so you don't have to pay them monthly. You just use your API that you're already using. And another tool that a buddy of mine made, it's a one-time thing, not a subscription, and it's called PowerCut and it's made by a guy who edits a lot of video.

Leo Laporte [02:33:29]:
So it's like a plugin?

Doc Rock [02:33:31]:
Uh, no, you run it on your computer. You basically give it your files and it does the amazing—

Leo Laporte [02:33:35]:
And it does it?

Doc Rock [02:33:36]:
Yeah, it's pretty, it's pretty incredible. Like these, these things that are helping with the edits, I'll put my link in.

Leo Laporte [02:33:42]:
Does it, does it cut in the middle of a word or is it really good about—

Doc Rock [02:33:45]:
It's really good. I'm actually—

Leo Laporte [02:33:47]:
PowerCut. I gotta get this.

Doc Rock [02:33:49]:
Here, just use, uh, docrock.live/power. That's the thing I set up for him. But he's a, he's a nice guy. He's a creator on YouTube. He teaches Final Cut training. And everybody, and he just started building tools to make Final Cut better because he wants to simplify it for the muggles.

Leo Laporte [02:34:03]:
That's cool.

Doc Rock [02:34:04]:
And it's a really cool app. So he has one for teleprompter.

Leo Laporte [02:34:07]:
So you have to have Final Cut too?

Doc Rock [02:34:09]:
I think it also works in DaVinci and Premiere as well. I have, which is free, but, um, I have paid.

Leo Laporte [02:34:15]:
So PowerCut, it's $150. It will, it will, does it just look at the audio and cut by the audio?

Doc Rock [02:34:23]:
Or you give it both. You give it the audio and the video and it will just chop away for you. And you can text.

Leo Laporte [02:34:29]:
I love this.

Doc Rock [02:34:30]:
You can edit words. Like if you accidentally—

Leo Laporte [02:34:33]:
Don't tell Benito. Benito, don't listen to this. Kevin King, don't listen to this. John Ashley, don't listen to this.

Doc Rock [02:34:41]:
Personally, like I love editing, so like I would never use a tool like this.

Leo Laporte [02:34:46]:
See, I hate editing, but that's why I have you.

Doc Rock [02:34:48]:
I like editing too. But what I, what I like it for is these guys are all getting good at finding your best cuts. So sometimes, you know, Leo and I come from radio. So when they say, do the line, coming up this week in tech, let me do it again. Coming up this week in tech, this week in tech show 1104, like we'll do it. And it's like, it'll pick the best one for you. Because even when I listen to them over, I'm like, frick, I can't pick one. And I just randomly pick one.

Doc Rock [02:35:13]:
So it does those kinds of things as well, which I love that. That's part of a producer's whole job. Yeah.

Leo Laporte [02:35:19]:
Picking takes.

Lou Maresca [02:35:21]:
That's a whole It's a taste thing.

Leo Laporte [02:35:24]:
Don't worry. I know, Benito.

Doc Rock [02:35:25]:
Data to game.

Leo Laporte [02:35:26]:
I will never trust AI to do that. Don't you worry.

Doc Rock [02:35:29]:
Hey, there's AI in Benito. Oh no, there's not. There's EI.

Leo Laporte [02:35:34]:
There's Benito-I. That's what we got. BGI.

Doc Rock [02:35:39]:
NI, natural intelligence.

Leo Laporte [02:35:40]:
Natural intelligence. Thanks, Doc. Have a safe trip to Japan.

Doc Rock [02:35:46]:
Mahalo, guys. It's been great.

Leo Laporte [02:35:48]:
Lou Maresca is a big shot engineering leader at Copilot. We didn't really get to talk about the Super AI. app. Tell me about that. What— so what does it do?

Lou Maresca [02:35:58]:
It's super apps, just the combining capabilities of all Copilots together. So it's—

Leo Laporte [02:36:03]:
is it an app app, or is it— it's—

Lou Maresca [02:36:06]:
it— they, you know, like you go to copilotm365.com and you start to interact with Copilot. Well, now you have all the abilities of like coding and co-work all together. That's what they call a super app.

Leo Laporte [02:36:17]:
Nice.

Lou Maresca [02:36:18]:
Yeah.

Leo Laporte [02:36:19]:
So Um, I gotta try it.

Lou Maresca [02:36:23]:
Yeah, yeah. Brings all the cloud—

Leo Laporte [02:36:25]:
I can use it on a Mac though, right? I can— I don't have—

Lou Maresca [02:36:27]:
Yeah, yeah, the Super Apps are all available on— yeah, you just use the web, get the Super App.

Leo Laporte [02:36:32]:
Oh, I could use the web. Okay, so that works everywhere.

Lou Maresca [02:36:35]:
Yeah, very cool.

Leo Laporte [02:36:37]:
Lou's doing really great work, and it's always a pleasure to talk. Former host of This Week in Enterprise Tech, we, uh, we love having you on. Thank you.

Lou Maresca [02:36:45]:
Miss that show, I loved it.

Doc Rock [02:36:46]:
Also, a bonus, Lou will be hosting an That's right.

Leo Laporte [02:36:51]:
I am going to be gone for 3 weeks, October 20th through November 10th, and Lou has very kindly agreed to host Intelligent Machines one of the weeks I'm gone. So thank you. I appreciate that, Lou. It'll be nice to have somebody who actually knows what he's talking about on the show. Really appreciate it.

Lou Maresca [02:37:10]:
You know, it's funny. I listen to the show every week.

Leo Laporte [02:37:12]:
You guys have such fun and such interesting So yeah, we, we try to interview some interesting people too, which is, uh, which in fact, I think we had you on maybe just once. And interviewed Lou. Thank you, Lou. Thank you, Doc. We do TWiT every Sunday afternoon, 2 to 5 PM Pacific, 5 to 8 Eastern, 2100 UTC. You can watch us live on YouTube, Twitch, X, Facebook, Kick, LinkedIn, Or in the club, you can also watch there. After the fact, you can get on-demand versions of our show and every show we do on our website, twit.tv. Also on YouTube, there's a dedicated channel for every show.

Leo Laporte [02:37:52]:
Start with the main channel, youtube.com/twit. You'll find links to all the other shows from there. Best thing though, I think, subscribe because we're a podcast, an RSS feed. It's magic. You'll get the show automatically the minute it's available. Just find your favorite podcast client and subscribe. It's free. That way you'll get it automatically every Sunday evening, just in time for your Monday morning commute.

Leo Laporte [02:38:17]:
Thanks for joining us, everybody. We'll see you next time. Another twit is in the can. Bye-bye.