This Week in Tech Episode 1092 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 at Tech. Our favorite Frenchman is here, Patrick Baja from Les Bleu. Wesley Faulkner is also here, developer relations. And from Microsoft, copilot Lou Mareska. Blockbuster news this week, Apple sues OpenAI. It could be a big one. We'll also talk about Meta. They say they could be fined as much as $1.4 trillion.

Leo Laporte [00:00:24]:
They're only worth 1.5. And yes, anthropic did it again. Stay tuned. Twit is next. Podcasts you love from people you trust. This is twit. This is TWiT this Week at Tech. Episode 1092, recorded Sunday, July 12, 2026.

Leo Laporte [00:00:55]:
You brought a knife to a wolf fight. It's time for Twit this Week in Tech, the show. We cover the week's tech news. Hello, everybody. Great to see you. And great to see our wonderful panel from Paris, France. Patrick Bajon is here.

Patrick Beja [00:01:10]:
Actually, I'm in Finland at the moment.

Leo Laporte [00:01:12]:
You went back?

Patrick Beja [00:01:14]:
I ran away from the heat and I'm this very old room of the house because my kids play the Switch all day and I can't be on the computer next to them. So I felt. I fled both France.

Leo Laporte [00:01:29]:
You mean the Nintendo Switch 2? They play all day.

Patrick Beja [00:01:32]:
They do. Well, not all day. We are. It's the holiday. It's a vacation, but still. But yeah, they bring their friends over. They're loud. So I'm in a closet, essentially.

Leo Laporte [00:01:42]:
Oh, man. Which games are they playing? I just started playing Astroneer. And what's that? Is that Mila the mouse or whatever? The digger?

Patrick Beja [00:01:51]:
Mina the Hollower. Yeah.

Leo Laporte [00:01:53]:
Mina the Hollower. Yeah.

Patrick Beja [00:01:54]:
We just tried Yoshi's. Yoshi and the Mysterious Book. It is wonderful. They've always been a little bit. A little bit like. Too kitty. Like the Yoshi games. This one is really beautiful.

Leo Laporte [00:02:11]:
I recognize that explains it. My seventh. The seventh grader who lives down the street said that told me that. So now to try it. We were comparing.

Patrick Beja [00:02:19]:
I don't know if it's. I don't know if it's for you, but it's for kids.

Leo Laporte [00:02:22]:
Matthew said it's good, so I'll give it a shot. Great to see you, Patrick. Wesley Faulkner is also here from Work's not Working. Although you could go to Wesley83.com and find all of his works. Hello, Wesley.

Wesley Faulkner [00:02:34]:
Hey, it's good to be back.

Leo Laporte [00:02:35]:
I would love to see you. Oh, we love having you on. And I do owe you a thank you because you introduced me to the crazy man. In fact, now Jeff Atwood and I are doing a monthly show called I love the name, by the way. Jeff Atwood, the creator of Stack Overflow Discourse and the Coding Horror Blog. So he's all about coding. He decided to name the show off by one and it is definitely off. And he's the one that's off by.

Wesley Faulkner [00:03:04]:
But it's a lot of crazy amounts of energy.

Leo Laporte [00:03:06]:
Crazy.

Wesley Faulkner [00:03:07]:
And the biggest heart. And he's flying back from right now. Yeah, we are developers. Watched a stream of that just the other day and it's just the same messages that we all need to take care of each other, which I wholeheartedly believe in.

Leo Laporte [00:03:23]:
Yeah, he's awesome. And if you look behind me, Jeff is. Is watching us. He's over our shoulder.

Lou Maresca [00:03:30]:
Oh my gosh.

Leo Laporte [00:03:34]:
Also here, Lou Maradska. Lou, good to see you.

Lou Maresca [00:03:38]:
Thanks for having me.

Leo Laporte [00:03:39]:
Former host of this Week in Enterprise Tech. He's an engineering leader at Microsoft. Copilot.

Lou Maresca [00:03:46]:
Yes, yes. Working on Agent mode for Excel. Right now.

Leo Laporte [00:03:49]:
It's Agentic Excel.

Lou Maresca [00:03:53]:
Yes.

Leo Laporte [00:03:54]:
Ooh, that sounds so interesting, Lou. I give Lou credit for putting Python in Excel. I don't know if that's appropriate or inappropriate, but I think it was a good move anyway.

Lou Maresca [00:04:04]:
Yeah, yeah. We still use it a lot.

Leo Laporte [00:04:06]:
Yeah, absolutely. Well, it's great to have all three of you. Welcome to the show. We do have some Microsoft news, but let's start with kind of a blockbuster story that broke I think on Friday. Mark Gurman at Bloomberg had the had the breaking news. Apple is suing OpenAI for a trade secret theft. It all started when the one of the guys who was responsible for the iPhone. Tang Tan, vice president of product design, leading development of the iPhone, The Apple Watch, AirPods and several other key Apple products left abandoned Apple to go to become chief hardware.

Leo Laporte [00:04:51]:
Wait a minute. Hardware officer at OpenAI. OpenAI is not a hardware company. Except that maybe they're planning on it. Apple alleges that Tang Tan brought his laptop with him, his Apple work laptop with him, including a bunch of key files. This is a big lawsuit, of course. OpenAI says they have no interest in other companies trade secrets. It did, it was funny.

Leo Laporte [00:05:22]:
It did kind of trigger a back and forth Twitter war between Elon Musk and Sam Altman. Sam said, I guess our new model 56 is pretty good. Elon's paying attention to me again. He accused Sam of being a theft thief. Always a thief. Once a thief, always a thief. It's actually been very entertaining. We've seen these kinds of lawsuits before.

Leo Laporte [00:05:51]:
There's some concern that this could impact OpenAI's IPO, which is imminent. It could also, and I would be unhappy about this, affect whatever hardware plans they have. You remember they spent more than $3 billion to bring Jony I've into the fold to do some sort of AI device with Tang Tan there. Maybe they want to do a phone, maybe they want to do a pen and maybe they want to do glasses. It's unknown. Apple says you're stealing our stuff. Wesley, what do you think?

Wesley Faulkner [00:06:25]:
I think that the. Well, the complaint is also that OpenAI coached them about putting in their two weeks notice and then so they can still have access to the information to continue to pull IP over time.

Leo Laporte [00:06:41]:
Here's the language in the lawsuit. OpenAI has counseled departing employees not to disclose their next employer and given advice on how to avoid the quote, dreaded walkout, end quote, that would promptly remove them from the company rather than giving them a standard two weeks in which they could continue to access Apple's confidential information and trade secrets. I have to think they wouldn't have asserted that unless they had smoking gun of some kind, an email.

Wesley Faulkner [00:07:07]:
So you also have to think about OpenAI. They're in a lawsuit with the New York Times, everybody stealing information. And the question is, what IP do they currently respect? So in terms of their track record, it's not looking good because they've blurred the lines about what they feel entitled to and what they can just take and steal. Which is also funny to hear about from what Elon says because apparently GROK just distills from all the other models and that's how they got to their. That's how they said that they architected it wrong because that's how they are trying to move forward quickly.

Leo Laporte [00:07:49]:
Loom doesn't all AIs basic. Aren't all the models trained basically on everything?

Lou Maresca [00:07:56]:
Similar data, similar to everything, right?

Leo Laporte [00:07:58]:
Yeah, exactly, in some cases. And this includes Apple a database of pirated books.

Wesley Faulkner [00:08:05]:
Yeah, Yeah.

Lou Maresca [00:08:07]:
I don't know if they do on pirated data, but they do on every data they get access to.

Patrick Beja [00:08:11]:
Yeah, it feels like distillation is a little bit, I don't know that it is, you know, illegal or actually unethical, but it feels like it's a different thing even than training your models on everything you can get your hands on. Which again could be, you know, could be a bad thing, but it feels like distillation is another category of bad things. And especially when you're, when you're doing that, accusing others of doing bad things feels a little bit like Disingenuous anthropic's

Leo Laporte [00:08:44]:
really taken the whole earth now attitude about all this, accusing China of using their models for distillation. That's an interesting point though, that there's a difference between ingesting the world's information and then, then taking a model that has already done that and training it in, in post training, which is what distillation is to be.

Patrick Beja [00:09:05]:
Well, the thing is, so training on the whole world models or the whole world content is a question of IP and intellectual property and, and copyright right. And it, it very well may be that copyright right holders are right to be upset about it. I'm not saying, you know, whether it is or it isn't might be distillation is the companies, the labs that are trying to do the best model in the world who can't and who need to use the existing frontier models as a crutch to develop theirs. And it's very capital efficient and energy efficient and it works really well. We've seen that with Chinese companies managing to reach levels of competency for their models probably through distillation. But I understand that, you know, the frontier labs like Anthropic and OpenAI, having done all of the research and the hard work of stealing other people's AI and sorry, IP and copyright, then see the others using their models to train their own and think that this is unfair. I think it is, you know, yeah, it's also if one is bad, the other is probably also bad, but different in a different way.

Leo Laporte [00:10:32]:
It's just different.

Lou Maresca [00:10:33]:
There's a similar metaphor to it. If you think about how people use agents today, they don't have the money to make the agent as smart as they needed to be. And so they use other AI foundational models as advisories so people to advise the agent on how to learn or what to do or to be more discreet. And I think that's exactly what other model companies are doing. Right. They're using foundational models as advisors to go and make sure that they focus on the right data or the right thing. You know, I, I see it as an optimization more than anything else.

Leo Laporte [00:11:04]:
I think that's what you, you would hire physicists to give a trained model physics questions, look at its answer, correct it, help tune it. It's a, yeah, it's, it's post training, it's tuning right.

Patrick Beja [00:11:18]:
Yeah.

Leo Laporte [00:11:18]:
So it is very different than building the model. The. I think it's important to note that did not go from Apple to OpenAI. He left Apple in 2024, co founded an AI device startup called Products Inc. Johnny I've. And Apple design veteran Evans Hanke joined him at Products Inc. Then a year later OpenAI acquires Products Inc. That was the six and a half billion dollars that they paid for.

Leo Laporte [00:11:49]:
We everybody said for Johnny Iva, but you also got 10 and Hanky in that. By the way, Ivan Hanke are not named in the lawsuit. Apple said it tried to resolve it out of court months ago, asking OpenAI to cease and assist the efforts and eliminate any proprietary materials. Apple said they didn't respond, so we were forced to. To sue.

Wesley Faulkner [00:12:14]:
That is.

Leo Laporte [00:12:15]:
Go ahead, Wesley.

Wesley Faulkner [00:12:17]:
Yeah, I was going to say that's a leap. Saying that they're putting the blame on OpenAI as the final receiver of the possible IP failure.

Leo Laporte [00:12:24]:
It is. I mean that's a year later.

Wesley Faulkner [00:12:25]:
Yeah, yeah. Without thinking about the interim step in and how it couldn't. They're kind of like, I'm going to say absolving, but they're not focusing on the person who physically or technologically transfer the IP away from Apple into their own repository of knowledge. And that's. This sounds like they're going for the bag of like finding the money without trying to establish the poison pill, saying this person did something illegal and everything after that is corrupt. And so it seems like they're going for the hard problem first, which means that this feels like mostly a PR play than anything else to kind of put them in the public and punish them through the public channels rather than through the court system because they have to establish chain of custody, to say that they have the information and their intention for it to ultimately live with OpenAI. And so them going through the complicated method first means that they understand the pending IPO. They understand that OpenAI is currently struggling to try to gain market share and to get the zeitgeist crown that they used to have that they no longer do.

Wesley Faulkner [00:13:39]:
And so them adding to the pressure and the negative pr, I think is the reason for them to try to use it as leverage for something. I don't know whether or not it is for their integration or for their contract disputes or whatever because they made specific references to OpenAI's ChatGPT in the iPhone as saying it's totally separate. Which sounds like thus protest too much like calling it explicitly.

Patrick Beja [00:14:10]:
OpenAI is suing Apple for that. Right. Because the deal, the contract with Apple didn't yield whatever use it.

Leo Laporte [00:14:18]:
I didn't even mention. You're right, they were partners. Remember that when you asked Siri a question, it couldn't answer it would then go to ChatGPT. So this is the breach of a partnership. And also remember, Apple decided to go with anthropic. Anthropic, Right.

Patrick Beja [00:14:36]:
Gemini.

Leo Laporte [00:14:37]:
Gemini. I'm sorry, Google. It's a little confusing leaving. And we all thought, wow, didn't they have a relationship with OpenAI? So, Lou, you were about to say something. Go ahead.

Lou Maresca [00:14:48]:
No, I was going to say Wesley's point of view is brilliant. I think it's definitely a marketing ploy here that you think about it. The way it read is the article read to me as like a postmortem for an incident where they were trying to figure out where the data was being leaked or how it was being leaked. But on the competition side, I didn't want to talk out, obviously with Microsoft, obviously, in general, most companies, they compete simultaneously at one layer about secrets and data and enterprise scale, and then another layer, they cooperate. I think we've seen that with Apple and Microsoft. We've seen that with ChatGPT, OpenAI and Apple. You're going to continue to see it and I think it's just the way of the world at this point. I think you're going to continue to see that type of.

Leo Laporte [00:15:27]:
Reed Albergatti, who we love, he's been on the show many times, writes at Semaphore, says initial allegations in the suit should be taken with a grain of salt. They are meant to make the accused party look as guilty as possible. He says Apple at first glance appears to have a pretty good case if it could prove anything. But the lawsuit is also a gigantic reminder that Apple's under major pressure from AI in general. It's not just that Apple somehow didn't see AI coming as playing catch up, the technology changes the company's fundamental reality. Apple's. This is interesting. Apple's entire business is predicated on the consumer desire for simplicity.

Leo Laporte [00:16:05]:
Customers pay a huge premium to be locked into Apple's ecosystem, where everything just works. And in doing so, they forego the opportunity to buy this cool, innovative new technology that sits outside the ecosystem. But AI is the ultimate simplifier. Can bring that back into the ecosystem. And of course that means app. Eventually, Apple's ecosystem lock in disappears. I think that's an interesting point of view. The damage, the immediate damage is to OpenAI's IPO.

Leo Laporte [00:16:35]:
We don't know when that's going to happen. They've filed, I think, for it.

Patrick Beja [00:16:38]:
Yeah. It seems Altman was, according to some sources, thinking, well, we're not getting a billion. Yeah, a trillion this year. So let's wait until 2027 because he really wants, you know, he needs the money. I guess he does. But yeah, this is, this isn't helping his case for sure. They are naming Tran though, and we. He's an employee now at OpenAI.

Patrick Beja [00:17:09]:
Right. The one thing that's a little bit suspicious is he left the company in 2024. Did they only realize he took his computer? Like, yeah, that's a little. Six months later. Or. The timing does seem suspicious.

Leo Laporte [00:17:23]:
But they would have telemetry on all of that stuff. They know exactly, I would guess.

Patrick Beja [00:17:27]:
Right.

Leo Laporte [00:17:27]:
Laptop is. And if. If he still has it and has had it for two years, that's kind of on Apple.

Patrick Beja [00:17:34]:
I mean, they at least know they don't have it in their offices. Right.

Leo Laporte [00:17:38]:
They have MDM.

Lou Maresca [00:17:40]:
Yeah. It should have been locked out without. After 30 days. I mean, it's like.

Leo Laporte [00:17:44]:
Yeah, yeah, well, or two weeks anyway. Yeah.

Patrick Beja [00:17:47]:
I mean, if he did do what he's accused of having done, it is pretty bad. Right? I think we all agree that if he did do that, it's. It's really bad and it should be provable, I suppose. But it's a little bit different from everything else we've been talking about. Because if he did do it, we know it's bad. There's no question about it. Like with training data and distillation and everything. The whole question is, is it possible to prove? And I would think that Apple could prove it pretty easily.

Patrick Beja [00:18:20]:
Right. Given the telemetry they have. So I'm not sure what the whole, like, hubbub is about. Either he did it or he didn't and they can prove it or they can't. So, yeah, it is dressed up.

Wesley Faulkner [00:18:32]:
It seems they need to prove that the data or the files or whatever are on OpenAI servers. That's what they have to prove in order to make this ultimate argument. If they.

Patrick Beja [00:18:42]:
Isn't it enough that he took it, then that's him.

Wesley Faulkner [00:18:46]:
They're not suing. That's what the point I was trying to make before. They're not suing him. They're trying to sue OpenAI as a company. And so them skipping that step of them looking at who took it initially.

Leo Laporte [00:18:56]:
Interesting.

Wesley Faulkner [00:18:57]:
Is assuming the intentions and the orchestration was from OpenAI, which is a way harder thing to prove. And so they should prove that he stole it and then that itself is a crime. And then why he stole it. The intention would then maybe lead to OpenAI. But they're trying to shortcut this because they know that each one of These steps is going to take so much longer that they're really just trying to win the PR cycle.

Leo Laporte [00:19:25]:
Well and I always point this out, Apple has learned from many other lawsuits, most recently Apple vs Epic, that nobody wins in these things because of discovery. Everybody ends up looking. There's always egg on the face of both parties because the emails get discovered, the smoking gun gets discovered perhaps, but so does who knows what else. You know, Apple trade secrets maybe. Even so, it's always, I imagine Apple didn't want to file this suit. Perhaps they won't pursue it through the discovery stage. Maybe it's just the damage they wanted to do is done already.

Wesley Faulkner [00:20:02]:
We haven't also talked about the other thing. Apple may be coming out their own AI hardware device and, and OpenAI will not release any device even if it's ready right now under this cloud of scrutiny. And so this, they want to hold it back, right? So this kicks back their launch date, if they even had it ready for enough time for Apple to catch up and maybe beat them to market.

Leo Laporte [00:20:25]:
There was some speculation they might do that this year. Apple wants to do what Meta has done fairly successfully, which is release smart glasses. Right. I this actually let's we're going to move on but just briefly talk about what form factor should OpenAI be looking at? What is the best form factor for an AI Able enabled device? All of you are wearing spectacles.

Patrick Beja [00:20:52]:
No one knows.

Leo Laporte [00:20:54]:
No one knows.

Patrick Beja [00:20:55]:
No one knows. And Apple is apparently again rumors going to throw a lot of spaghetti at the wall. They will have apparently glasses and a wearable pin, whatever and camera in the AirPods. You know, it's going to be every way they can because their problem is that their AI is on the phone, right? And the phone is usually in your pocket. So what you want, if you want to feed your AI information about what you're doing is have access to sound and sight outside of your pocket. And whether that is glasses or headphones, I mean your AirPods or a small pendant that you wear on your, on your neck, maybe some of those will work, maybe some won't. But I think the form factor, the one that makes the most sense is the glasses. But not everyone wears glasses and maybe some people will use one, some people will use another.

Patrick Beja [00:22:00]:
But the whole point for Apple is not to have the whole device. I think it's to have eyes and ears for Siri in your pocket. And that makes it a very different engineering wise and device wise proposition from the AI company's devices which need to do basically everything. And that are severely hampered by that. That requirement.

Leo Laporte [00:22:25]:
Having the phone is, is, Is kind of a great start. Right? Because it's a powerful engine. Much more powerful than you could just have an append in your glasses.

Wesley Faulkner [00:22:34]:
Can I. Can I.

Patrick Beja [00:22:35]:
And it has all your data and all your apps.

Leo Laporte [00:22:37]:
It already knows everything. That's what this new Siri AI is. They're. They're doubling down on Siri's. Already got all your text messages and your. And all that stuff. Go ahead, Wesley.

Wesley Faulkner [00:22:46]:
I think the best, and this is a totally different direction, is a USB C dongle.

Leo Laporte [00:22:52]:
I like. This is new. Tell us your thing.

Wesley Faulkner [00:22:55]:
When you think about, like the computers that were sold out for these Mac Minis to run openclaw and all this stuff, it's because they want it to run and they want it to just have just a plug and play. Do all my things.

Lou Maresca [00:23:09]:
Right.

Wesley Faulkner [00:23:10]:
If you have a USB dongle, you can use it in a personal and a business context.

Leo Laporte [00:23:14]:
Kind of like a Fire TV stick or a Roku stick, right?

Wesley Faulkner [00:23:17]:
Yeah.

Leo Laporte [00:23:17]:
It's got a CPU, it's got some RAM. RAM's too expensive to put much in.

Wesley Faulkner [00:23:22]:
But also it has access to your hardware in terms of your hard drive. It'll have access to your camera, it'll have access to your Internet connection, It'll have access to install things, manipulate files, have its own storage.

Leo Laporte [00:23:37]:
That's really interesting. You know, ubiquiti sells for 800 bucks, something they call an AI key that enables you. It's just a little box you plug into your Ubiquiti router that enables AI features in your cameras. It is what you just described. Basically. It's an AI engine that you plug in and they make 800 bucks and you plug it in and suddenly your cameras get smart. That's kind of an interesting idea.

Wesley Faulkner [00:24:01]:
It also could plug into your phone. So if you think about the magsafe like attachment on the back of your phone with maybe a USB C cable, now you have all the same power, but mobile. Yes, exactly right. And it can leverage a lot of the hardware.

Leo Laporte [00:24:17]:
I do too. I never. Nobody's ever said this before that I

Patrick Beja [00:24:20]:
think you guys are engineers talking about engineer stuff and real world one will want their phone. And what Apple is going to do is provide an AI that you're going to scoff at and laugh at because it doesn't do everything you want it to do. But 80% of the, you know, the Apple model, 80% of the people will get 80% of what they want done, and that will be good enough.

Leo Laporte [00:24:48]:
I will say because even today, Siri the new everything they talk about the new Siri AI I have been doing for months with a local agent running on my PC. It already knows all that stuff. It's able to. I can query stuff. So yeah, you're right. The tiny sliver of people who are using AI at the frontier might scoff, but. But this suddenly becomes something for the. For the everybody for every human.

Leo Laporte [00:25:18]:
I like the idea of glasses, but there's a privacy issue with these. For them to be really useful, they need a camera and everybody hates that idea. There's also the issue for people like us who wear spectacles. I'm not. I'm gonna have to buy a lens for this. Right. It's going to be. So it's, it's.

Lou Maresca [00:25:38]:
How are they going to make us leave our glasses outside the room when we walk into a.

Leo Laporte [00:25:41]:
Right, yes, I have to have my glasses. So. Yeah, right.

Patrick Beja [00:25:47]:
My wife was talking about meta glasses a couple of days ago and she was like, the day someone comes to our home with those, they're leaving them at the door.

Leo Laporte [00:25:56]:
She's not.

Patrick Beja [00:25:57]:
I do not want this in my home. And I understand why.

Leo Laporte [00:25:59]:
I think.

Wesley Faulkner [00:26:00]:
And you can't even just suppose the person who has them, you can't trust them. It's not just that you can't trust the company that operates them.

Leo Laporte [00:26:06]:
Right.

Wesley Faulkner [00:26:06]:
Because you. They may be having this feature where they do a passive recording all the time and you don't know where that's going to. It could even be going to a third party company that they're using to even go through the logs and go through the telemetry and so the data gets out there whether you try to or not. So yeah, they're not safe at all.

Patrick Beja [00:26:28]:
That's going to be an interesting problem to solve for Apple, you know, the privacy protecting company when whenever they inevitably come out with phones or AirPods with cameras. Like that's going to be. That's going to be an interesting one.

Leo Laporte [00:26:44]:
I read an expose the other day. What Xia this is on GitHub. It's a gist what Xai's GROK CLI sends to Xai. Somebody analyzed on the wire all the content that GROK sends and at first I was horrified, but then I realized, oh no, every AI does this. It has to do that. It transmits the contents of every file it reads right. It's not processing it locally. It has to upload it to the cloud.

Leo Laporte [00:27:14]:
It uploads whole repositories of course, all of this stuff at First I went, oh my God, Grok. But then I thought, well, that's how they all the Frontier AIs work, everything for it to do the work has to be uploaded to its servers for it to process. Now you may say, well, but they have agreements, they certainly do with enterprises not to use that for training or not to look at it. But I don't know how much we could trust those agreements. Yes. I think it's going to hit people suddenly that when you're using a cloud based AI, everything you do is available to that cloud. Am I wrong, Lou? I mean that's the case, isn't it?

Lou Maresca [00:27:57]:
It's the exact case. I mean we just, I don't know if you saw the announcement of the Solara badge. I mean that's exactly what's happening there. It's connected to a private cloud.

Leo Laporte [00:28:06]:
In order for get to get enterprise to accept this, you have to say that, you have to say, well, we're going to do a. This is what Apple's saying to customers. We're going to do a private cloud. But there's still a certain amount of trust involved on the end user that that private cloud is well designed, it really is private that it can't be broken into. And we see so many data breaches. I don't know if anybody has faith in anything that claims to be private. So that I think if you're a

Patrick Beja [00:28:31]:
big enough company you can, you can get something on premises.

Leo Laporte [00:28:34]:
Oh yeah, well, we were talking about this on our AI AI User group on Friday. The Department of Defense is exactly what the Department of Defense does. It's running Fable, but it's running it on prem or in effect on prem. Maybe an AWS private cloud or whatever. But yeah, they're not, they're not, they're not sending anything to Anthropic. So if you're big enough, that's the case. But for you and me, I don't

Patrick Beja [00:29:01]:
know, I think any responsible company should look at that if they're consumers become

Leo Laporte [00:29:07]:
aware of it, Right? They aren't maybe right now.

Patrick Beja [00:29:09]:
Yeah, but consumers, you know, do we even have privacy anymore?

Leo Laporte [00:29:13]:
Do we do care? Oh, we might care, but I don't.

Patrick Beja [00:29:17]:
I mean everyone's using Gmail, right? So at that point you're like, okay,

Leo Laporte [00:29:21]:
whatever, that's a good.

Patrick Beja [00:29:22]:
It is a step further if you have cameras in your home filming everything. Not for you, Leo, but for normal people.

Leo Laporte [00:29:28]:
I keep my camera information local with local models. I'm using a Quen model for the vision analysis and it's running locally.

Patrick Beja [00:29:38]:
Aren't you using an AI life recorder anymore? Like you don't have a pin anymore,

Leo Laporte [00:29:45]:
like your, your friend Lisa put the kibosh. Okay, but that, but seriously. But honestly, I, I still want that. I still think there's this real tension between people who naturally, including me and mostly my wife, want full privacy and people who love the idea of, wouldn't it be great if my personal AI knew everything that happened in my life and I could, you know, have recordings of everything of audio and video and I could query it. I could say, especially as, as I get older and there's a big swath of us baby boomers who are approaching this time of life where we're getting forgetful. And I think this would be really wonderfully valuable. But I think also all the people in my life say, well, you can't. What? Stop it.

Patrick Beja [00:30:38]:
I think it's a, it's a, it's a, an interesting prospect. It's essentially having, let's say, a form of photographic memory, right? You just can remember everything about your life. I think there's something very seductive about this and I, I wouldn't mind having it. I think the concern would be, you know, what company owns it and what do they do with it. But even then for me and you, you know, we're very public people, so it matters less for us maybe.

Leo Laporte [00:31:11]:
Gordon Bell was the first is of course, he was a deck very famous designed PDP and all of that. And his wife Gwen Bell suffered from Alzheimer's. Late in life, Gordon was what used to wear. I've interviewed him where he's long gone now, but he wore a camera around his neck and recorded every 10 seconds everything that was going on. And this was all for Gwen, but also because of Gwen. He was aware of this and had this notion and I was inspired by it. He was actually at Microsoft Research when I interviewed him. Really brilliant computer scientist.

Leo Laporte [00:31:50]:
But this was long before the age of where we started thinking about privacy.

Wesley Faulkner [00:31:54]:
So speaking of privacy though, who owns this when you're gone? And so do you want a whole data dump of everything you've ever done in the hands of someone or a company that may not have your best interests at heart?

Leo Laporte [00:32:05]:
Well, I think we've all learned from the fable rug pull that the ideal is a local model running local data, not none of it going out of the home network in theory.

Wesley Faulkner [00:32:18]:
But think about it like you still don't know every licensing agreement you ever clicked yes on. You don't know.

Leo Laporte [00:32:25]:
So I control, I control what's Going out of my home network. Right.

Wesley Faulkner [00:32:28]:
Don't I while you're alive? Yes.

Leo Laporte [00:32:30]:
Oh, oh, and I could set it up so that there's a dead man switch if I cared about that. That if I don't respond in five days,

Patrick Beja [00:32:38]:
deletes everything.

Wesley Faulkner [00:32:40]:
Remember in the rundown there's a story about Microsoft and there's this UUID version or some sort of device ID and they're able to track everything they did on trackers.

Leo Laporte [00:32:51]:
Arrest reveals Microsoft can track users via Windows device id. Well, you know what, first of all, let me say this. LUM works for Microsoft, but does not represent or speak for Microsoft in any way. And Lou, if you wish to recuse yourself at any time from any conversation.

Lou Maresca [00:33:08]:
Thanks, Lee.

Leo Laporte [00:33:09]:
You are more than welcome to do so. Let's take a break. We should, we should have. That's an interesting story. There's actually quite a, quite a bit of Microsoft to talk about when we continue. I, you know, I, it's just a.

Wesley Faulkner [00:33:22]:
You think you old. The hardware is all. And sometimes maybe you don't.

Leo Laporte [00:33:26]:
No, I agree. I don't use Windows.

Patrick Beja [00:33:27]:
No. But actually this is, this is a super interesting. Who owns things like AI? There's a bunch of companies who are training AIs to replicate like to make digital twins, you know, not the CEO.

Leo Laporte [00:33:41]:
I would make a digital twin of myself.

Patrick Beja [00:33:42]:
Yes, but if a company makes a digital twin of an expert they hired, you know, train them on their email, on their conversations and everything and then that person leaves.

Leo Laporte [00:33:52]:
Did I try to do that with its employees?

Patrick Beja [00:33:54]:
Does the person, can they take their own like AI twin with them? If they don't, does the company still own the AI twin and can quarry that twin that has the expertise of the person that left within the like do they own it? Probably, I guess I don't know. But like who owns the thing becomes much more important when it becomes, you know, when goes from private data about you to an emulation of your expertise of your intelligence?

Leo Laporte [00:34:30]:
Well, I'll put it another way. When Captain Kirk transports down to the planet's surface and then transports back up, is the Captain Kirk who came back the same Captain Kirk who left in the first place? I would submit no, it's.

Wesley Faulkner [00:34:43]:
One is killed and destroyed and another one.

Leo Laporte [00:34:45]:
Yes, it's a copy.

Patrick Beja [00:34:46]:
Oh it's a Perseus boat thing. Right. But Captain Kirk, if Captain Kirk stays on the surface and is on the, the, the, the Enterprise, you got a real problem. Yes, because the people on the surface can have access to Captain Kirk's expertise without paying for it. And you know, the, the, the, what's the name of the Star Trek?

Leo Laporte [00:35:16]:
The Borg?

Patrick Beja [00:35:17]:
The Federation. Is it the Federation?

Leo Laporte [00:35:19]:
Yeah.

Wesley Faulkner [00:35:19]:
United Federation of Planets.

Patrick Beja [00:35:21]:
The Federation is paying Captain Kirk's salary and the other ones are still getting his expertise without paying anything for it. I feel that's, that's, that's a problem. That's.

Wesley Faulkner [00:35:32]:
There is so many Black Mirror episodes of where this is exactly the case. And the one where they do a Star Trek kind of like spoof, they took the copy of the creator and like put them in the game saying you're making new stages and levels for the rest of your life, which is this digital life thing. So it's exactly the same thing. But I wanted to go back to the example of who owns what. There's, you know, copyright claim saying any, anything that's made by AI is not copyrightable. And so when you think about the things that is being generated, that's what I mean, that you don't know if you own it. And then also there are some open source licenses saying like any changes need to go back to whoever created the original. So there's several different, different ways that you may not be thinking of, of how data might be owned or not owned by you or might be taken.

Wesley Faulkner [00:36:18]:
And so it's ambiguous right now. And this is something that still is being worked.

Leo Laporte [00:36:24]:
It's actually not ambiguous. When you, when you sign employment contract, doesn't the company own all your work?

Patrick Beja [00:36:30]:
Is that work though is a model trained on your expertise part?

Leo Laporte [00:36:36]:
This is what's.

Wesley Faulkner [00:36:36]:
So using a Hermes. When you're using a Hermes agent and it's doing stuff on your behalf, do you own that work? I mean think about I better as the orchestrator. See what I mean?

Lou Maresca [00:36:47]:
Right.

Leo Laporte [00:36:48]:
I can't copyright it though. The courts have already ruled. It's very confusing.

Patrick Beja [00:36:53]:
No, so you have to trademark your key phrases like economy prompts are trademarked. This is.

Leo Laporte [00:37:00]:
We are living in weird times. Let's just, let's just stipulate that and we're going to take a break and come back in just a little bit. By the way, apparently this has all been explored already. I didn't even know this in Star Trek lower decks, Darren Okey says Boimler got copied by the transporter and you ended up with two of them for the rest of the show. I have not seen lower decks, so I don't know. But of course there's, you know what? There's nothing we can come up with that hasn't already been written about in Star Trek.

Wesley Faulkner [00:37:30]:
We haven't even touched on Tuvix and that whole thing. But we'll continue.

Leo Laporte [00:37:33]:
Exactly. There's nothing, it's nothing new under the sun. We'll have more in just a bit with Patrick Peja, Wesley Faulkner and Lou Mareska. It's great to have all three of you on this wonderful day. A beautiful day for a twit. So Microsoft was in the news this week. Asha Sharma, the new head of Microsoft is increasingly look like she might have been brought in to do this. The company eliminated on Monday 4800 and actually this is from all of Microsoft.

Leo Laporte [00:38:05]:
A lot of them are from Xbox. But Amy Coleman, who's the HR boss at Microsoft said 4800 roles were eliminated Monday morning. They said. Coleman said Microsoft's doing the layoffs to reduce its global headcount by about 2.1%. This is not the first layoff. Microsoft's been doing them for a while. And she said it's because the business of technology is changing, so Microsoft has to change with it. This is the quote.

Leo Laporte [00:38:32]:
The way technology is built, deployed and used is transforming faster than at any point in my time here. I would, I would grant you that she's been at HR and Microsoft for 17 years. That means we will need to adjust resources and roles and shift how we operate so we can have the greatest impact for our customers. Look, it's hard, I feel for the people who got laid off. It's hard for the people who stay. I'm sure Lou will vouch for that. Means more work means losing trusted friends and colleagues. Microsoft did say these people are not being replaced by AI.

Leo Laporte [00:39:11]:
That's I think, thank you Microsoft for being honest. I think that's always been the case with a lot of these layoffs. But plenty of companies are willing to blame AI. She says AI is changing how work gets done. Microsoft never did what Meta did briefly till they heard the splashback. They were actually tracking all of their employees, clicks and text and everything. They said to train our models. To do what? To replace our employees.

Leo Laporte [00:39:46]:
Anyway, first of all, my heart goes out to people who've lost their jobs. That's always hard. But I also kind of am sympathetic. I understand how these things happen. But now you cover gaming, Patrick. So I'm curious what you think because a number of very well known, even successful gaming companies under the Microsoft umbrella lost some or all of their team.

Patrick Beja [00:40:12]:
Yeah, it's. I don't know how much, how much into the details you want to get, but this is mostly Xbox employees. And as you mentioned, 3200 of the 4800 were Xbox.

Lou Maresca [00:40:26]:
Yeah.

Patrick Beja [00:40:27]:
Right. So it's 1600 today or last week and 1600 over the next year. That means that there's the people who didn't get laid off this week.

Leo Laporte [00:40:41]:
The hanging over there.

Patrick Beja [00:40:44]:
Exactly. They don't know what's going to happen in the next few months. And that, you know, can't be fun. The There are. Okay. So if you want to understand what happened there at Xbox because 3200 people is 20% of the headcount of the company. That is a lot of people. Right.

Patrick Beja [00:41:07]:
20% for. For any organization is a huge amount. So what happened was that Xbox has never had a significant success. Their biggest success was the Xbox 360. And even then it was kind of neck and neck with the PlayStation. After that it was kind of failure. Not. Well, yeah.

Patrick Beja [00:41:30]:
Failure after failure. Now, Xbox does make money. It is profitable. But just 3% margin which is, you know, not big in any business.

Leo Laporte [00:41:39]:
Well, losing money.

Patrick Beja [00:41:41]:
It's not losing money. But they're not happy with those returns. They did spend like $80 billion buying up companies to get content to get games. So what happened?

Leo Laporte [00:41:54]:
Here's an interesting stat from Asha Sharma. She said 64 cents for every dollar we invested we lost in a tip of a year. That's 64 loss. And was mostly the smaller studios. In fact they let a couple of smaller studios compulsion games which did we happy few. Which was a great game. Double Fine Productions, very famous for Psychonauts and other things are basically. I don't understand what this means.

Leo Laporte [00:42:21]:
Return give their return to management will operate independently.

Patrick Beja [00:42:25]:
Those two. Those two are back to being independent students.

Leo Laporte [00:42:28]:
They're just. They're spinning them off.

Patrick Beja [00:42:29]:
They're letting management. Yeah. Management essentially got a good deal that they could take back their independence. Keep their ip all of that. There are a couple of studios.

Leo Laporte [00:42:38]:
Good.

Patrick Beja [00:42:38]:
That are.

Leo Laporte [00:42:39]:
Yeah, yeah, yeah.

Patrick Beja [00:42:40]:
I mean it's good. There's still. I wonder what's going to happen in the next few months for those. They might not be able to operate with all the people they currently have.

Wesley Faulkner [00:42:50]:
Right.

Patrick Beja [00:42:51]:
You know, there might be layoffs there.

Leo Laporte [00:42:52]:
There are a couple of other. Which did Hellblade and Undead Labs which did State of Decay. They were purchased by unnamed somebody else.

Patrick Beja [00:43:00]:
Deal is still ongoing. And Arcane Leon should have the same. I'm saying it with the American accent. Arcane Leon is getting. Should be bought by another company as well. But there are more.

Leo Laporte [00:43:16]:
There are rules in France for layoffs. They cannot just lay off Arcane.

Patrick Beja [00:43:20]:
Exactly.

Leo Laporte [00:43:21]:
Yeah. They Did Dishonored and pray for those

Patrick Beja [00:43:23]:
who don't know, those are very good, very well regarded. All of those studios are, are super prominent studios. So what happened 10 years ago is that the guy who was in charge of Xbox at the time, Phil Spencer, had this idea. He thought, you know what, we're not winning in this undifferentiated battle with PlayStation. We need to do something different. Let's do Game Pass, a subscription service and offer gamers a subscription for a ton of great games. And in order to do that, we need to have a lot of studios making great games. Some big AAA experiences, you know, really explosive single player experiences, some smaller stuff, some even indie flavor games.

Patrick Beja [00:44:08]:
Bundle all of them in Game Pass for a monthly fee, add some third party partners and that would make an offering that gamers would enjoy. Now unfortunately, in order for this to work, the idea could have worked, could have not. It didn't work. Spoiler alert. It didn't work. It could have. I think maybe it' too much to ask for people to, you know, pay a certain amount for a bundle of games where every single games takes up like 10 or 20 hours of your life. Maybe that's not the same as Netflix.

Patrick Beja [00:44:42]:
We'll never know. Because the problem was they didn't manage to put out big triple A games like the big system sellers, the games that will motivate the players, the gamers to go subscribe to your thing. Because they did have the studios but somehow the discussion is ongoing there. It didn't work out. They put out not a single big quality triple A game. There were some good games, there were some good indie flavored games, but there wasn't one on the level of what Nintendo or Sony puts out. They were all. So my theory is that.

Patrick Beja [00:45:24]:
I don't know if it's true, but my theory is that Phil Spencer is too nice a person and he was letting the studios do their thing.

Leo Laporte [00:45:34]:
He was like, Sharma got brought in as the.

Patrick Beja [00:45:37]:
Well, so that happens afterward. But Spencer was respecting the artists to a fault. I think he didn't want to intervene into the studio's work. Maybe that's not the case. That's my theory. And they were left kind of. I don't know if it was rather less or they didn't have proper management stewarding their studio. They did have management, but it felt like maybe it wasn't because it wasn't just one company.

Patrick Beja [00:46:02]:
They have Bethesda, they have ID software, they have, they have Mojang. Mojang is working. Yeah, Mojang is doing Minecraft is Minecraft, and that's working.

Leo Laporte [00:46:13]:
You know, the Candy Crush folks, okay,

Patrick Beja [00:46:17]:
they bought King and they had big ambitions on mobile. Nothing came of it.

Leo Laporte [00:46:21]:
King was part of the Blizzard acquisition, Is that right?

Patrick Beja [00:46:24]:
Yes. Yeah, absolutely. Activision Blizzard, they also bought $70 billion. It's a crazy acquisition.

Leo Laporte [00:46:33]:
Would you call that a success? They fought hard to get that. They got a lot of regulatory pushback. They fought hard to get that Activision Blizzard through. Was it. Was it the right.

Patrick Beja [00:46:43]:
I think the games are good, but the issue is they are on every platform. So it does nothing for Xbox. And it didn't. The goal was to make Game Pass into kind of a platform and get to their projections was like 77 million subscribers by 2026. They didn't get past 34, I think, and now there are 30. Less than half what they were hoping for. So it didn't work out. And with all of those studios not producing the engine of that Game Pass car, they had the wheels, you know, the small indie games.

Patrick Beja [00:47:21]:
They had the infrastructure. They didn't have the engine. Halo was disappointing. Gears of War was disappointing. Starfield was disappointing. All of those big games weren't what they should have been. And so the result is that Game Pass didn't do what it was supposed to do. And Microsoft is not seeing the returns it was hoping for on Xbox.

Patrick Beja [00:47:42]:
So bring in Asha Sharma. Lay off 20% of the people, refocus on the big franchises. Like, why did we not get a new Elder Scrolls? Why did we not get a new Fallout with the TV show working, you know, being popular, why did we not get there? There. There are tons of IPs that we didn't see. And if you don't have those drivers, then it doesn't matter how many awesome little indie games, you know, Compulsion or Obsidian or Double Fine makes. It doesn't matter because it's not enough to. To. To fuel your.

Patrick Beja [00:48:16]:
Your Game Pass ambitions. So that's what I think happened. Game Pass didn't work because it didn't have the major games to push it. And so the platform that was Game Pass wasn't successful enough to justify all the spending. And Nadella was looking at this and thinking, dude, I gave you $80 billion. You need to do something with it. And so the one thing that I take away from it, which is almost surprising, is that Xbox Microsoft is still not getting out of the gaming business. They didn't have a Great first Xbox.

Patrick Beja [00:48:55]:
Xbox 360 was okay. Xbox One was a disaster. Xbox Series X and S is kind of a Disaster. Game Pass is a. Not a disaster, but it's not a success. They. The success they were hoping for. They could have stopped it.

Patrick Beja [00:49:11]:
You know, they could have said, okay, we tried for 20 years, we'll stop. But they are not. And they're hopefully, I don't know how, because their next console is like a hybrid PC console that's going to cost 1500.

Leo Laporte [00:49:24]:
Xbox.

Patrick Beja [00:49:25]:
Yes. Project Helix, which.

Leo Laporte [00:49:28]:
Right. Yeah.

Patrick Beja [00:49:28]:
Could you think, in theory, that's, that's what we're thinking and that's what they've hinted at be a Windows device that will run Windows games. Because all of a sudden you don't care that you only have, you know, 30 million Xbox owners. You have a platform that is essentially compatible with Windows, which has, you know, however many hundreds of billions is part

Leo Laporte [00:49:52]:
of the problem, though, with future hardware. The RAM and hard drive crisis.

Patrick Beja [00:49:58]:
Yes, very much so. I mean, the Valve Steam machine, which is essentially a PC in the format of a console that you connect to. Your TV costs 1000 bucks minimum. PlayStation 5 nowadays cost like 7,800 bucks. Project Helix is described, it's the next Xbox, described as a premium product before the RAM prices increase. How much is that going to cost? How are they going to make that work? Like the economy. Sharma and others have hinted at, like, we need to adapt business models and stuff. But how do you do that? Do you put ads in your, in your games? EA is doing that.

Patrick Beja [00:50:40]:
Actually. It's. It's a very difficult problem to solve. I don't know how they're going to solve it. I know that there are, you know, three, 200 people who are either out of work or looking at being out of work very soon. And it's, it kind of sucks. I don't know what they're going to do. Xbox, by the way, was very.

Patrick Beja [00:51:05]:
Was a lot more unionized than other gaming companies. And I wonder if that made the layoff conditions better. I'd love to know.

Wesley Faulkner [00:51:19]:
Sorry, sorry to cut you off.

Lou Maresca [00:51:21]:
Sorry.

Wesley Faulkner [00:51:22]:
I just wanted to say, just for the people who are laid off, unfortunately, they were probably doing what they were told to do. And the errors in terms of them not having the focus or not moving in the right directions are the people above them that chose these directions. And I think it's very clear that Xbox was not listening to the gamers. They were not listening to what their customers were asking for. They had these grandiose ideas and, and they said, this will be the thing. This will be the differentiator. Without actually moving in a direction that developers or gamers Wanted in order to have the outcomes that they wanted. And that is why they have not had the adoption.

Wesley Faulkner [00:52:04]:
It's because they don't listen to their customers and the users. Who was asking for Game Pass? I don't know who was, but.

Leo Laporte [00:52:11]:
Well, especially at the price they ended up getting to. It was a little.

Patrick Beja [00:52:14]:
Initially, it was an incredible deal.

Leo Laporte [00:52:17]:
Yeah, no, I did it at first, but when it got. They like doubled it almost in it.

Wesley Faulkner [00:52:21]:
Yeah. Because that was a bribe. That was just like. Like, let's give you all this money to get you hooked. But. But they're like once, once they backed off to do it then. Exactly.

Lou Maresca [00:52:30]:
So.

Wesley Faulkner [00:52:31]:
Well, what. What I want to say is I think the people in leadership need to get the blame. Not the people. So every. For all these bad decisions, the people are the ones who suffer for that, not the people who make these two.

Leo Laporte [00:52:42]:
I mean, some of it's that gaming has changed. Right. And how we play has changed, changed. Lou, I see you nodding. Let me, let me give Lou a chance. And again, sorry, if you decide you don't want to say anything, you don't have to.

Lou Maresca [00:52:53]:
No, I mean, I want to go up. What Wes is saying is like, you know, obviously it's impacting people, right? Families, people.

Wesley Faulkner [00:52:58]:
Yes.

Lou Maresca [00:52:59]:
You know, and it's very, very hard. And, you know, I think the biggest thing that really bothers me about this whole situation is I don't know if you've heard about the VRP situation with Microsoft giving voluntary.

Leo Laporte [00:53:09]:
Oh, yeah, yeah, yeah. If you were older, you qualify for that.

Lou Maresca [00:53:12]:
You had. I qualified. I didn't take it, but I did qualify. I think the thing that, that really bothers me is a good number of people that probably were in this. Lot of people that didn't know the layoff was going to happen probably decided like what Wes said is, I'm going to continue on this journey of making Xbox great. I know I can do it. And then they didn't take the VRP and then they got told they're going to get laid off. Right.

Lou Maresca [00:53:34]:
So I think that's what kind of bothers me personally about this whole situation. So I think there's a lot of these kind of decisions are made off of, of bad decisions from, from the past. And I think that's, it's. It's impacting real people. And it really bothers me from that perspective.

Leo Laporte [00:53:49]:
The, the VIP rules were, were very complicated. Basically, you had to have been there. Well, how did it work?

Wesley Faulkner [00:54:00]:
It was age plus tenure. If it's greater than 70 or above,

Leo Laporte [00:54:04]:
then so, so in other words, if you, if you were 45, it's highly unlikely that your age plus your tenure would add up to 70. You'd have had to start when you were, Well, I guess 25. Anybody younger than 45, really unlikely. You, you must have just squeaked in under the wire, Lou.

Lou Maresca [00:54:23]:
Yeah, I mean there's, there's been some people that, that just squeak in or whatever and they make a decision on whether they want to take it. It also depends on how much they're going to get or what the, you know, what, what the benefits are and so on and so forth. So yeah, a lot of good people that, I knew that, that did take it because they were ready.

Leo Laporte [00:54:38]:
So yeah, and credit to Microsoft for giving people. But if you didn't take it, you didn't get to change your mind later. You couldn't say, oh, you're gonna lay me off, oh, I'll take it now. Couldn't do that.

Lou Maresca [00:54:49]:
Yes, that's true.

Wesley Faulkner [00:54:50]:
I would say that this is just to talk about that calculations of this up to 70 and you can get let go. It's actually really horrible. And let me say why right now with AI, junior roles or people coming out of college are having a hard time getting jobs. A lot of those entry level roles are gone. With this new formula, it kind of eliminates the protections for older adults because you can't claim age discrimination since it's replaced with a calculation. And so now this is, now that we have the undercroft of the new people eliminated. Now they're removing the top crust of people and getting those people eliminated with a way that they can't claim age discriminations and they can't have those protections. And so this is a new invented mechanism to get rid of people to sidestep the regulations and the laws to prevent that from happening.

Wesley Faulkner [00:55:45]:
And so it is a scary thing that, just like layoffs becoming a thing that people do and it's okay and not a big deal. This will be something that other companies will copy and will do as well. And so start to see other companies saying, using this calculation to remove, remove their older workforce, which is extremely scary because that will be replicated over and over and over again, which would exacerbate the unemployment and the, the, the, the power dynamic of the companies over the employees. And it's, it's, it's, it's a, it's a scary trend that I feel that this is the first data point.

Leo Laporte [00:56:23]:
Isn't it interesting though that the latest job stats show that perhaps AI is Actually increasing the number of tech engineering jobs are up like something like 15%. The openings, not for other jobs, but for, for engineering jobs. AI has not hurt the job market. Maybe it's even helped it. I mean it's probably different skill set, Lou. Right? It's not. We're not. You're not looking for code entry level coders anymore, right?

Lou Maresca [00:56:50]:
Definitely. Different skill set for sure. But you're right. I mean it has. There's a specific skill set that we do need. AI engineering, prompt engineering, you know, science folks. And I think it's, it's exponential at this point, so it'll continue to grow.

Wesley Faulkner [00:57:05]:
Slop wrangling. It's people who tried to just now

Leo Laporte [00:57:07]:
that's an anti AI turn. Slop wrangling.

Wesley Faulkner [00:57:12]:
Am I wrong?

Leo Laporte [00:57:13]:
Talented AI prompting engineers.

Wesley Faulkner [00:57:17]:
No, it's not the prompter. It's the people who are trying to do the prompting. And like I can't. I've gone as, as far as I can. Let me hand it over to someone who might know more. And they're trying to get people like saying, this is what I made. Fix it.

Leo Laporte [00:57:28]:
Ramp. Ramp came out with a study that said the companies making the largest AI investments grew employment by more than 10% compared to companies that hadn't adopted AI.

Wesley Faulkner [00:57:39]:
But how much are they char. How much are they paying these people?

Leo Laporte [00:57:43]:
Well, if you're. Look, if I would guess if you're a talented AI engineer, you're getting paid top dollar right now. Right.

Wesley Faulkner [00:57:52]:
I think it's commodified for what they want. And I think people are misunderstanding what AI can do. Like if you. Did you see what happened with Ford and yeah, they fired a bunch of

Leo Laporte [00:58:03]:
quality engineers, gave their jobs to AI and then they've been trying to hire them back.

Wesley Faulkner [00:58:08]:
Yeah, I think there's so, well, fundamental misunderstanding of what people do, especially what AI can do.

Leo Laporte [00:58:14]:
Well, that I would expect that because it's. You're trying to figure it out. Right? I would expect a little bit of that. That's normal.

Wesley Faulkner [00:58:22]:
So what I'm saying is there's nuance to all these numbers. And then you have to look instead of just looking at these top lines and when they say that there's an increase of people in this area, who are they considering? Developers. Who are they considering?

Leo Laporte [00:58:35]:
Right.

Wesley Faulkner [00:58:35]:
The level of developers who are expertise that they need.

Leo Laporte [00:58:38]:
You've got kids are going to be entering the job market in the next 10 years, let's say. What are you telling your kids they should become good at? Probably not. Python.

Lou Maresca [00:58:48]:
Well, they, they're Focused on. But they do Python, I think. You know, I just tell them to go where, Wherever they're interested. Lie, right? I mean, so they focus on science or, you know, that kind of thing.

Leo Laporte [00:58:58]:
So, I mean, I couldn't. When 20 years ago, when Henry was a teenager, I, I. There was no way for me to say, you know, you should really focus on TikTok, because there was no TikTok.

Patrick Beja [00:59:10]:
I.

Leo Laporte [00:59:11]:
So I did the same thing. I said, you know, find something you love doing. And he watched a lot of YouTube cooking videos, and he. And he just fell in love with that. Who knew that that would end up becoming a very valuable skill in the year 2026, my son, for those who don't know, has become quite successful. The New York Times did their fourth review of his sandwich shop in New York City, Sunday, or last. Last week, I guess it was. And.

Leo Laporte [00:59:39]:
And was basically a rave review, although Henry was pissed.

Patrick Beja [00:59:42]:
Fourth review.

Leo Laporte [00:59:43]:
They just want to go eat there

Patrick Beja [00:59:45]:
for free at that point.

Leo Laporte [00:59:46]:
I think it's, you know, he was a little miffed. He said, I have to go back on the 5th of July. I'm having a great time here in the Jersey Shore. I got to go back. Those stupid New York Times people are coming in again. I said, henry, knock it off. He said, don't they have anything else to write about? I said, no, you say? So he came in. It was their.

Leo Laporte [01:00:06]:
It was their food critic. This was like the real deal. Food critic came in. The funny thing is, apparently the New York Times is their own rules prevent them from giving more than one star to a sandwich shop. You can't get. If you're making sandwiches, one star is all you can get. So it's the rave review. I mean, like, the guy loved the place and gave it one star.

Leo Laporte [01:00:28]:
And Henry was, again, he was a. Come on, man.

Wesley Faulkner [01:00:32]:
I gave it five stars on Google reviews.

Patrick Beja [01:00:34]:
You.

Leo Laporte [01:00:34]:
You've been there. That's right, Wesley. I have not had this sandwich. Salt. Hank's famous prime rib sandwich.

Wesley Faulkner [01:00:43]:
It's delicious.

Leo Laporte [01:00:43]:
The way the New York Times wrote about it made me want to eat there. I don't think anybody's gonna get to the bottom of the review and see the one star and say, well, I'm not going now.

Patrick Beja [01:00:51]:
It's funny that he wasn't excited. Maybe it was because it was the whatever. But it's. Maybe he doesn't care about the New York Times. He was like, well, if it was a TikTok influencer, then there is a

Leo Laporte [01:01:05]:
social network, food network in New York City called Belly that Is all voted by the eaters, by real people, and according to Belly, the number one. You're right, that's probably a better. From his point of view, he's happier about that than he would be about the New York Times. It's just me. It's his dad who says, henry.

Patrick Beja [01:01:29]:
I said, henry, it's the New York Times.

Leo Laporte [01:01:31]:
I've heard if you can make it there, you can make it anywhere. I've heard that. You know, you should be. You should be thankful. Here's the. By the way, this is on their digital. This is the picture that they put. Nobody's seeing one star, Hank.

Leo Laporte [01:01:49]:
He's definitely got a rave review. And it was clear the guy didn't want to love it. But by the end, he was completely seduced by the delicious baguette and the stretchy provolone and the juicy paper thin meat which Laporte dunks in broth before it leaves the kitchen.

Patrick Beja [01:02:10]:
You know, it's funny, I look at it and I'm sure it's delicious. But as a French person, I look at the size and I look at the sauce that you dip it in and it's like, really, like, do you really need, like the, the bread is seasoned, beef is like, this is over. Then you dunk it in the sauce and. Okay, that's an American thing. Yeah, no, I, I go.

Leo Laporte [01:02:38]:
And the truth is, I go, yeah, I. And then the PR. I think it's $32 now. It started at 28. It's slowly. What? But don't. You don't eat this by yourself, my friend. You go in there with, right.

Leo Laporte [01:02:50]:
Wesley, did you eat the whole thing?

Wesley Faulkner [01:02:51]:
I did,

Lou Maresca [01:02:54]:
yeah.

Wesley Faulkner [01:02:57]:
And the fries probably shouldn't have, but you did.

Leo Laporte [01:03:00]:
Okay, I did.

Patrick Beja [01:03:02]:
$32. I mean, he's successful. Like, there's only so many sandwiches he can make in, in the day. At some point you can only make 300.

Leo Laporte [01:03:12]:
It's successful. That's why they sell out by 130 every day. It's successful because it's A, it's New York. B, he's using literally the best beef you can buy from Pat Lafres. He's using. He said, dad, I don't want to make this cheaper. I want to use the best ingredients and that's what it costs to make it again. None of the advice I would have given him 16 years ago would have been appropriate.

Leo Laporte [01:03:35]:
Right.

Patrick Beja [01:03:35]:
How much of his success currently is tick tock and how much of it is being a great sandwich place in New York? Like, what?

Leo Laporte [01:03:44]:
Initially it was tick Tock when he. It's been almost exactly a year and initially of course it's tick tock. It's just two and a half million followers nowadays is a million and a half on Instagram. But that gives you only the beginning, right?

Patrick Beja [01:03:55]:
Yeah, exactly, exactly. How, how does he still do the daily tick tocks?

Leo Laporte [01:03:59]:
Oh yeah, get to rest on that. He says, says I'm going to take a little credit. He says a lot of the people come in, say, listen to your dad's shows. Yeah, maybe he's more famous than me and maybe as he said just the other day, I'm making a lot more money than you, dad. But this all started because Lou, we were talking about the advice and I'm just saying your advice to your kids is exactly right. You say just do something you care a lot. Wesley, you agree too, right? You do something you care about, you love.

Lou Maresca [01:04:31]:
Yeah.

Wesley Faulkner [01:04:32]:
There's no way to predict what hasn't been invented yet. But I would say the thing that follows you around is knowing who you are. And you can then look at the tools and look what's around you to help you use them to, to like as the paintbrush to draw your passions so that everyone can see your art and your special spark. But you also, if you know yourself, then you also will be drawn to something that you get to, to do that you are able to maintain. It's hard to do something you don't like doing because you think other people want to do it or you think it'll be successful. Those people burn out and then they get to a breaking point where they have to kind of like break themselves down to understand what voices are theirs and what voices are external and to kind of like figure out what is the thing that drives them. And there's this existential crosswords that people go through and if you can skip that whole thing, please do. And like if you look at how education is done these days, we're talking about developers.

Wesley Faulkner [01:05:35]:
Before what kind of developers who some were went to four year universities and then they became developers some and they're, they studied computer science and then there are some that went to boot camps who really did an intensive study and saying that's the best way to going. And then some people are just hobbyists, they work on doing stuff in their basement or on their home lab and they really hone their skills that way. There are different methods and you have to find the path that is the one that fits your life and not what other people tell you to do. And I know you need to make money. But the question is, like, what is the creative way that you can. And that's the thing you need to figure out. And that's where people can help you. You can get advice from.

Wesley Faulkner [01:06:16]:
But knowing what you want to do, that's something that only you can answer.

Leo Laporte [01:06:20]:
Yeah, I was a Chinese major in college. I didn't, I didn't plan to be a podcaster, actually. The people of my generation, and I'm going to include Bill Gates, Steve Jobs, most of them did, dropped out of school, didn't, you know, became. They were into computers, especially in that era, as. Because it was fun, it was cool. They loved it. Not because they were trained well.

Patrick Beja [01:06:43]:
It didn't exist.

Leo Laporte [01:06:44]:
Well, there's people, there's people like my friend Bill Atkinson, who was a brain. A neuroscientist, he was into cognition and, and came to Apple. They. Jobs persuaded him to come to Apple.

Patrick Beja [01:06:59]:
Yeah, no, they, they were inventing it. You get.

Leo Laporte [01:07:02]:
They were inventing it.

Patrick Beja [01:07:02]:
You couldn't, you couldn't go and study computers.

Leo Laporte [01:07:06]:
You couldn't.

Wesley Faulkner [01:07:06]:
Yeah, yeah, but even studying Chinese is still communication.

Leo Laporte [01:07:10]:
And now it would be a very good language to study. Absolutely.

Wesley Faulkner [01:07:15]:
Yeah. But, but you also studied, like, development, which is computer languages. It's, it's communication and expressing yourself is still the same thread from when you studied Chinese to now.

Leo Laporte [01:07:25]:
Just do it. Yeah. So what it's. Your heart will tell you.

Patrick Beja [01:07:28]:
I think it's. Yeah, the, the, the commencement speech from Jobs was really amazing. You know, where he says, you can't connect the dots going forward. You only. I studied Japanese and it turned out it helped me being in Japan and. But about what to study now, I have a question for you guys. My kids are still pretty young. My, my son is 8.

Patrick Beja [01:07:51]:
But they don't, they essentially don't use tablets. They don't have any. I don't sit him down and go like, hey, this is how coding works with this whatever, you know, turtle thing that goes forward and turns. And when you code this, am I making a mistake by keeping them completely? They don't have phones, they don't have social networks. Again, they're still pretty young, but I think it's not going to change anytime soon. And I feel like I'm a nerd. I'm a computer person. Maybe I should teach him how to use, I don't know, Windows or something.

Patrick Beja [01:08:26]:
You know, am I making mistake?

Wesley Faulkner [01:08:28]:
The biggest thing you can teach a kid is to learn how to learn. And everything else is getting in a direction of solving problems to fix something that you want to do is practical. Trying to teach kids something that they are not interested in is a losing proposition. It does not work. So if you understand them and they understand themselves and you try to move to a place where it's like how do we figure this out? Then you can bring in tools that make sense, then you can move them in places where they're interested in. But most of it is just if you're thinking about what should I teach them? You should teach them like what questions do you have? Let's figure out how we can get answers. And that's the most important thing.

Patrick Beja [01:09:07]:
They're very interested in learning the Nintendo Switch to.

Leo Laporte [01:09:13]:
That's really, that's coding. I mean I think Minecraft is a great way to learn code, to be honest with you. Right.

Lou Maresca [01:09:19]:
Yeah, yeah. I was going to say I have kids from 5 all the way up to 16 and I can say I started 16 year old when the iPad came out and I had him use it. He used it past time for him and I feel like it was a mistake. So I learned as my different generations of people kids have come out, I've taken a long time to give them devices. And I think like Wes said is I go to the point of learning something and how to do something and I always use technology only in that aspect. So when it comes to a tablet, if they want to 3D print or if they want to, you know, like, like Leo said, build a Minecraft level or something like that, that's where they go to devices. And I think that's really helped a lot. They stay away from social networks that's completely blocked from my network.

Leo Laporte [01:10:05]:
Yeah, I don't think there's anything you can gain. Although again, Henry learned how to cook on YouTube. I don't know if YouTube's a social network. I would not block kids from YouTube. I would just supervise with them.

Lou Maresca [01:10:14]:
No.

Patrick Beja [01:10:15]:
Are you kidding me? No YouTube at all. I mean they're a bit young still, but.

Leo Laporte [01:10:19]:
No, yeah, it ain't maybe. Yeah, but by the time they're 12,

Patrick Beja [01:10:23]:
maybe when they're 12. But there's so much like. I know YouTube. I know what's there.

Leo Laporte [01:10:30]:
I don't know so much good stuff too. Right.

Patrick Beja [01:10:32]:
Of course I love YouTube. I watch it all the time.

Leo Laporte [01:10:34]:
I wish there were a way you could tell YouTube. Okay. They can watch veritasium. They can.

Patrick Beja [01:10:39]:
Yeah, basically.

Leo Laporte [01:10:39]:
Yeah, there were, there were.

Patrick Beja [01:10:41]:
You would PBS, space, time and startalk.

Lou Maresca [01:10:45]:
Yeah, yeah.

Wesley Faulkner [01:10:46]:
You can stream YouTube using specific protocols if you know the address. So like if you can use VLC to look at YouTube. And so you could, you could theoretically program locked channels.

Leo Laporte [01:10:57]:
You could have your agent do that.

Patrick Beja [01:10:58]:
I'll go ask.

Wesley Faulkner [01:10:59]:
But also like YouTube kids. But just do the thing. Like when there is like you think of a computer in the living room, have YouTube be only on the TV or something.

Lou Maresca [01:11:07]:
Right.

Wesley Faulkner [01:11:07]:
Supervision and so. Yeah, so. So you simply can't do anything. That's a good idea. Would walking by and seeing it

Leo Laporte [01:11:15]:
teach kids about AI these days?

Patrick Beja [01:11:20]:
I'm going to teach them how to use Siri AI when it comes out in, in the eu. Probably because it's on the.

Leo Laporte [01:11:26]:
It's never, by the way, it's never going to come out in the eu. You know that, right?

Patrick Beja [01:11:31]:
It is very much. We were talking about Apple doing marketing hits earlier. I, I am 150% convinced that that thing is a marketing.

Leo Laporte [01:11:44]:
All right, we got to pause. We got to take a break. So that's a good topic when we come back because that's a very contrarian point of view. This is. You guys have. This is such a rich conversation. I hate to interrupt it with anything as crass as a commercial, but I have to. So somebody's got to pay the electric bill.

Leo Laporte [01:12:03]:
We've got a wonderful panel here. Lou Mareska is always just brilliant and wonderful to have. He's engineering leader at Copilot Microsoft. And just always a pleasure to have you on, Lou. And same for you, Wesley Faulkner. Love having you on. This guy is all heart. Wesley83.com will tell you everything he's up to, including his site.

Leo Laporte [01:12:24]:
Works, not working. And Patrick Beja, who's joining us from Finland today. He is a podcaster extraordinaire@notpatrick.com and oddly for a Frenchman, not following the World

Patrick Beja [01:12:40]:
cup, you know, from afar. But as I said earlier, semi finals. I think at that point I have to. Otherwise they're taking my passport.

Leo Laporte [01:12:52]:
Yeah, you have to. I have to stay. I'm just telling you, you've got to.

Patrick Beja [01:12:55]:
I mean tomorrow is, is. How do you call it? Bastille Day.

Leo Laporte [01:13:00]:
Oh, that's right. Alons enfant de la patra, la jour de guarette. Harvey.

Lou Maresca [01:13:06]:
Nice.

Leo Laporte [01:13:07]:
Yes. Everybody storm the Bastille. If there's anything left, I think it's all prime rib, French dip, sandwiches. As a matter of fact, that's all they got at the Bastille these days. Before we go to our break, breaking news.

Wesley Faulkner [01:13:23]:
God diggity day extension.

Leo Laporte [01:13:27]:
You know that I have been pushing like crazy to fin my Fable production because I Thought today we were going to lose it. Actually, I thought we were going to lose it a week ago. I thought they extended it, and I thought we were going to lose it today. I stayed up literally last night. I put my agents in a loop. I said, don't stop till it's done, because tomorrow it's over. And guess what? Of course, they have extended Fable 5 promotional access for another week through July 19th. That is a.

Leo Laporte [01:13:59]:
That means you can use it with your subscription, if you have a subscription plan. And I don't know if they reset the usage limit. I'm not sure.

Patrick Beja [01:14:11]:
They increased 50%.

Leo Laporte [01:14:12]:
50% to the weekly usage list. So this is Pro Max team and premium seats. If you're on an enterprise plan, will

Patrick Beja [01:14:21]:
they ever actually cut it off?

Leo Laporte [01:14:24]:
I don't.

Patrick Beja [01:14:25]:
Maybe not.

Leo Laporte [01:14:25]:
You got to think they're losing money. Some of this is probably also contingent on how much is being used. Like, can we. Can we continue to provide this? Because I'm, you know, the estimates are. I don't know if this is accurate. What do you think, Lou? Nate B. Jones is going to be on intelligent machines in a couple of weeks. Has estimated it's a 10 billion parameter.

Leo Laporte [01:14:44]:
10 trillion. Sorry, 10 trillion with a T parameter model, which would make it the largest model out there. Does that seem, from what you know, because you're kind of an expert on this.

Lou Maresca [01:14:54]:
I can't say anything about that, but I would say that it's definitely a big model.

Wesley Faulkner [01:15:01]:
But with OpenAI's Luna and whatever, Sol

Leo Laporte [01:15:05]:
is probably 10 trillion. Right.

Wesley Faulkner [01:15:07]:
People are starting to do the comparison. So that's why it's advantageous for them to keep using the promotional credits.

Leo Laporte [01:15:12]:
Yes. I think it has something to do with the release of Saul. Absolutely. I'm using Saul to check on Fable, which is checking on Opus 4. 8. So it's A. It's a. It's a whole thing.

Leo Laporte [01:15:21]:
In fact, I even. And then I threw Grok4.5 into the mix. Why not? Let's get all the Frontier models and they have a little agent mailbox and they're mailing each other with their thoughts. I said just, you know, if you. If you have anything that you think is wrong, just email them. And. And then I tell each one, check your mailbox every 10 minutes. There might be something from another model.

Patrick Beja [01:15:43]:
Just.

Leo Laporte [01:15:43]:
It's. We live in interesting times. That's all I can say. And I guess I'm going to continue to live in interesting times for another week. Wow, that's good news. Although it did finish at everything last night when I woke up and said, yep, we're done. Anything else you'd like us to do?

Patrick Beja [01:16:01]:
I'll start another project.

Leo Laporte [01:16:04]:
You know, it's.

Patrick Beja [01:16:05]:
What else do you need to fix?

Leo Laporte [01:16:06]:
It's turning out that the real challenge in all this is coming up with things for it to do. I mean, I. I don't work for a big company. I'm sure if you're a big company, you have lots of things to do.

Wesley Faulkner [01:16:16]:
But what you could always say is like, hey, now that we're here, if. And you look at everything you've done,

Leo Laporte [01:16:21]:
what else would you like to.

Wesley Faulkner [01:16:22]:
If you had to start from scratch, what would you have done differently? And then start all over again?

Leo Laporte [01:16:27]:
I. I did give it a blog. I said, hey, if you want to. If you want to write a blog, you can write a blog. And. But it's only posted three times, so I don't know, it's. It's gotten busy, I guess.

Patrick Beja [01:16:40]:
No, I think. I think now you need to get into a really ambitious project and start a virtual Leo Laporte. I think that's.

Leo Laporte [01:16:49]:
I think that's.

Patrick Beja [01:16:49]:
You were wondering what you were gonna do. You know, who. Who keeps what once you're gone? I think you keep it. Your virtual you keeps going and you can now actually do it. You have voice models that are incredibly.

Leo Laporte [01:17:05]:
I already. 11 Labs has my voice down like that. Like, it's perfect, right?

Patrick Beja [01:17:10]:
No, but build an actual, you know, replica and actually, like, have the things. Maybe even the voice model can be local or. I don't know, to be.

Leo Laporte [01:17:20]:
But are you sure that's a project done that yet?

Patrick Beja [01:17:27]:
Can we be.

Leo Laporte [01:17:29]:
Yes. I'm too imperfect. That's the key. Gotta build in imperfections.

Wesley Faulkner [01:17:35]:
All right. And ads.

Leo Laporte [01:17:37]:
And ads. It's actually better at ads than I am. That's for sure. I've. I've had it writing ad copy now for some time, and it's much better than I am. Don't tell anybody it didn't write this. Oh, I thought I thought of something. Did we say we were going to talk about something and I for.

Leo Laporte [01:17:54]:
I forgot what it was.

Wesley Faulkner [01:17:57]:
I forgot as well. What is this AI memory thing?

Lou Maresca [01:18:00]:
You're talking about the Apple AI in the eu, right?

Leo Laporte [01:18:02]:
Oh, yeah, yeah, yeah. But Patrick's disappeared. When he comes back, we'll have. We'll get out. Get him to talk about how. Why he thinks Apple is just. It's a faint F, E, I, n, T. What about AI memory? Were you talking about Wesley?

Wesley Faulkner [01:18:16]:
Oh, I was just saying, like, I. We wish we could remember what we're talking about.

Leo Laporte [01:18:20]:
Oh, yeah, I need.

Wesley Faulkner [01:18:22]:
We need that AI agent to listen to everything.

Leo Laporte [01:18:25]:
I had an argument on intelligent machines with Paris. She said AIs don't remember anything. And it's true. Unless you take steps. Every time you start a new session with an AI, it starts with a blank slate. But I think almost everybody using any of these tools, Claude code or Codex or an agent like I use Hermes, is trying to get memory in there. God knows we don't want.

Patrick Beja [01:18:48]:
Well, they have it integrated now, right? Don't they both have like a memory thing? I'm sure, yeah, they do.

Leo Laporte [01:18:55]:
They have memory md. What do you do in Copilot for that, Lou?

Lou Maresca [01:18:59]:
Yeah, we have a memory system that's built in. It's. It's not lostly. It's lost less, but it's. It still can. It, you know, compacts and it produces context every time you send a prompt.

Leo Laporte [01:19:10]:
Is it a text file, an MD file, or is it a database? Is.

Lou Maresca [01:19:13]:
It's. It's a, It's a. It's an MD file. It continues to update itself.

Leo Laporte [01:19:17]:
Yeah, I think all of them. I know Codex and Claude use an MD file. Hermes does as well. Hermes has a nice feature where if you do something, it makes a skill out of it so that it then saves that skill and you always, if you're going to do it again, it can do it again. So that's kind of cool.

Lou Maresca [01:19:31]:
I love the CLAUDE integration. Has like integration with Git too, so actually has like historical context. That is cool.

Leo Laporte [01:19:36]:
Yeah, I love that. That surprised me. When I first started using cloud code, it would push stuff up to Git and then, and I was talking about this on the AI user group. I've. I have, you know, I'm a, I'm not a professional at this. I knew that something called CICD existed. Continuous integration, continuous delivery. I knew it existed, but it turned it on.

Leo Laporte [01:20:00]:
So I was writing a newsreader and it said, okay, well I have Windows, Mac and Linux versions of it ready for you. I said, what? Every time I committed, it would build a new version for me. It was like the coolest thing ever. So you don't need to know DevOps anymore because apparently these AIs are pretty good at this kind of stuff. It's amazing.

Wesley Faulkner [01:20:21]:
Oh, you do, dude, you do need to know it. I'm learning to get started.

Leo Laporte [01:20:26]:
I'm learning. You need to know to ask for it. Although again, I didn't ask for it, it just did it. And they said, what are you doing? And Then it said, well, it explained

Wesley Faulkner [01:20:35]:
that's actually a lot of the exfiltration that's happened and has gone through the CI pipeline. Oh yeah, still keys and stuff.

Leo Laporte [01:20:42]:
Oh, it's a flaw.

Wesley Faulkner [01:20:43]:
So it's, so it's very important that you do know what's happening.

Leo Laporte [01:20:46]:
That's a very good point. Yeah, that's very good point. I, I actually this I learned also I said could you do a security audit? And it said your, your keys are in plain text on the hard drive. I said, oh, what should I do about that?

Wesley Faulkner [01:20:59]:
You want them plaintext on the server?

Leo Laporte [01:21:02]:
Yeah, that's the next step. You commit them to get. So it said well we could SOPs encrypt them them and, and then they wouldn't be in plain text anymore. So I did that and that kind of opened my eyes to all of this. Now actually they're stored in bit warden in my password manager and they're materialized in memory but you know, they can't write to the drive and they can't save them. So yeah, you learn. But I see, I feel like this has been an educational experience in many ways for me too. I'm.

Leo Laporte [01:21:31]:
I'm not a pro, but I'm learning more and more about how development proceeds.

Wesley Faulkner [01:21:38]:
Are you using GitHub Actions by any chance?

Leo Laporte [01:21:40]:
Yes, I was. I actually have this. I have kind of after my 20th GitHub repo, I decided, you know, this is so the stuff that I do now, like this twit ads. I actually have a nas. I put git on a NAS and I commit to the NAS instead of committing it public to GitHub. So I kind of, I feel safer if I keep stuff local.

Patrick Beja [01:22:04]:
Yeah.

Wesley Faulkner [01:22:05]:
Have you heard of Git T? Are you.

Leo Laporte [01:22:06]:
I know about Git T. Yes. I just use plain old git.

Patrick Beja [01:22:12]:
Yeah.

Leo Laporte [01:22:13]:
Oh yeah, it's fine.

Wesley Faulkner [01:22:14]:
It's Git works.

Leo Laporte [01:22:15]:
I don't really need a CICD pipeline or actions or anything. I just do it all by hand.

Wesley Faulkner [01:22:22]:
If you get the reason why it's important is the, the rule enforcement. Like there's also like pre commit hooks. I'm assuming that you're using those.

Leo Laporte [01:22:29]:
Yes, I am using those. Yeah.

Wesley Faulkner [01:22:32]:
But if you really want reinforcement, the good thing is that when you codify in your CI pipeline, when you find problems, you can prevent regression. Sometimes your LLM will think it knows something and will actually introduce a bug that you've already solved before.

Leo Laporte [01:22:48]:
Right.

Wesley Faulkner [01:22:48]:
And if you're using it in, if you enforce that through your CICD process, then you're making sure that Those bugs don't get reintroduced, and you only fix them the first time.

Leo Laporte [01:22:57]:
And all right, I'm going to go down this rabbit hole. Even though I have deep misgivings, I know the entire audience is going, what are they talking about? But one of the things I have noticed with agentic work and even with Claude, some of it's deterministic. In other words, every time you do this, that happens, it's causal. You know, there's a chain. Some of it's kind of probabilistic. Like, you do this, that might happen and it might not. And the AI just kind of, oh, yeah, I forgot to do that.

Patrick Beja [01:23:28]:
Yeah.

Leo Laporte [01:23:29]:
And that is a little bit of a problem that you can't assume that it's deterministic.

Wesley Faulkner [01:23:37]:
Also, you can't assume that it's looked at something real. Sometimes it makes a guess.

Leo Laporte [01:23:42]:
Hallucinates.

Wesley Faulkner [01:23:43]:
Like, I assume not just hallucinates, but saying, I'm sure you're doing it this way. So I'm gonna. I'm sure this is problem. I'm gonna start solving this problem that didn't exist. So it's not necessarily a hallucination, but it's taking an assumption without really going through and doing the deep dive to make sure that the code actually reflects what they're seeing. And that's why documentation is important, too, because it'll use that as a crib

Leo Laporte [01:24:06]:
sheet if it's written out.

Wesley Faulkner [01:24:07]:
It won't if it's written down somewhere, oh, this doc said this is how it's done.

Leo Laporte [01:24:10]:
Right.

Wesley Faulkner [01:24:10]:
So if you're not continuously updating the docs, that's an issue.

Leo Laporte [01:24:13]:
I've noticed that's not deterministic. That's somewhat probabilistic. Like, oh, I didn't read it this time. Like, you don't. Lou, am I wrong? Maybe you're right. You're right.

Lou Maresca [01:24:23]:
I mean, you have to be very explicit, right? As Wes is saying, like, and most novice users don't know, you need to say, do this every time or check this every time.

Leo Laporte [01:24:31]:
But you can say that and it still doesn't do it.

Lou Maresca [01:24:34]:
It can make a choice. It could make a choice, sure. Yeah, but. But it's very. It's less probable. It's less Right. You know, you build.

Leo Laporte [01:24:42]:
You build up these structures that kind of try to keep it in line at all times. Every once in a while, CICD is

Wesley Faulkner [01:24:50]:
your last line of defense.

Leo Laporte [01:24:51]:
Okay, so that's. That's good to know. I will have to get a book on that.

Lou Maresca [01:24:58]:
I mean, one thing I would say, especially with Agents, LEO is like making agents develop cement. Like build up deterministic code or scripts or tools to be able to do it on a guarantee. Do it every time. Right. Most of the way I use it is like it generates tools for me that I know will do it every time. And then I left.

Leo Laporte [01:25:18]:
Code is deterministic. Code always is cause and effect. So you're right. So when I know I want it to do something, I say make a cron job, write some Python and execute that on a cron job. Then I know that that will happen.

Lou Maresca [01:25:35]:
Right.

Leo Laporte [01:25:35]:
So yeah, so that's the kind of, the thing that I have to keep in mind is that some stuff is deterministic, predictable, it's going to always be the same way, some's not. And the stuff that you really want to have happen every single time, put it in code. Is that right?

Patrick Beja [01:25:51]:
Yeah, that's right.

Wesley Faulkner [01:25:52]:
Okay, have it write the script that checks, that runs and checks to see if it works, as opposed to check to see if it works.

Leo Laporte [01:25:59]:
You know, the earliest version of that was the phrase use code. If you ask it to do Math, what is 2 plus 2? It's gonna, it's gonna give you how

Wesley Faulkner [01:26:10]:
many Rs in Strawberry even or how many.

Leo Laporte [01:26:12]:
But if you say use code, it will write a Python script and then it will get it right. You could say how many hours of Strawberry use code. It will always know. But if you ask it, it doesn't know. It's tokens. Strawberry is not S, T R A W B E R R Y to it, it's token 1, 2, 3.

Lou Maresca [01:26:32]:
Right behind the scenes. Nowadays it does it anyways. It uses the code interpreter for lots of things that you don't know it's doing.

Leo Laporte [01:26:38]:
Oh, that's interesting. So that it's internalized that notion of if you really need to do math, you should probably always use code. We. I just. To me, I don't know why, but I'm fascinated by this. It's it. It piggybacks on my fascination with technology in general and computing, particularly code specifically. It's just a very exciting time, isn't it?

Lou Maresca [01:27:02]:
Yeah, absolutely confusing and a little bit of ambiguous too, but it's definitely exciting.

Wesley Faulkner [01:27:07]:
Well, it's more like when you're also confused, you can say, explain it to me.

Lou Maresca [01:27:11]:
Right?

Wesley Faulkner [01:27:11]:
And then you have someone to actually like break it down. I don't know. Cicd. Wesley says it's important. Why is he saying that?

Lou Maresca [01:27:19]:
Yes.

Leo Laporte [01:27:19]:
Well, how do I even get started?

Wesley Faulkner [01:27:21]:
Like, okay, that's great. So knowing my code and knowing My history, how would you set this up initially? Okay, what are best practices? And it'll look it up and it'll say, okay, now you claim it's this. I don't know if I believe you. Show me all the evidence. And then so it's something where you don't just have it do something, but you can also use it to take you along the ride so you can make better choices.

Patrick Beja [01:27:43]:
I think this aspect, because you guys are talking about code a lot, because it's exciting and that's what you do. But I think most people, the way they perceive and look and understand and use AI is a little bit different. And what I try to boil it down to, why AI is exciting and is an interesting technology, is that it takes a huge amount of data and it is able to interpret it and extract the data for you. Right? It's, it's, that's what's really cool for most people when they use a chatbot, they, you just throw all this, maybe somewhat organized, but you just roll the data at it and you ask it to look into it and to tell you to extract the thing you want to do. And that is to me at least, the most exciting aspect of AI. And then there are different ways to apply it. It can be applied into, you know, actual requests to write code. But on a more general level, that ability alone makes it incredibly exciting.

Patrick Beja [01:29:00]:
You can throw all the data you want and it will figure it out. You don't need to figure it out anymore. It can figure it out. Hopefully you check it and it's more and more precise and less and less hallucinate E. But that ability to extract data from a bunch of disorganized corpus is incredible on its core. And it can do that. There's no question about, oh, will it be able to make a business and will it at some point will AGI do whatever? Maybe, but already it can do that. And that is very useful, I think.

Leo Laporte [01:29:43]:
Is it going to replace search even. Maybe even web.

Patrick Beja [01:29:48]:
It has.

Lou Maresca [01:29:49]:
I hope it does.

Patrick Beja [01:29:51]:
Well, I don't, I don't know that I do, but it, it kind of has. I can't. I think one of the biggest effect it's had on me, again, I'm not a developer. I don't really code is. I search, I ask my different chat buddies stuff that I would have searched for otherwise, right. And initially I was like, oh, I'm going to ask stuff about stuff that I already know. So I kind of, it reminds me and I can understand if it's messing things up and not, you know, getting the right information. So it's a reminder kind of.

Patrick Beja [01:30:28]:
But nowadays, if I'm being honest, I ask about like we were shopping for an electric bike. I didn't go to Google. Like I don't have time to click links. I just asked both, you know, Chat, GPT and Claude so I could compare a little bit their responses. So there's this, this and this and what should I pay attention to? And is this a good deal? And it came up with a bunch of acceptable, seemingly acceptable answers. I think a lot of people that use chatbots use that because the response is a lot more natural. It's a lot easier to listen that way.

Leo Laporte [01:31:03]:
Instead of asking Siri or Googling something, she will ask Perplexity. I was watching the World cup yesterday and I thought, I can't just sit on the couch and watch this fascinating, exciting game. So I'm going to iron my handkerchiefs. While I watch, I'm going to do some ironing. And then I'm thinking as I'm watching and this is. I asked this question of my agent. I like to use spray starch when I iron my handkerchiefs. I'm wondering, I use Niagara Premium Smooth.

Leo Laporte [01:31:30]:
But is there something the ironing experts think is better? Now, you couldn't really quantify that in a Google search, but it actually said, yes. I try Mary Ellen's Best Press scent Free as yous're Better Than Niagara Test Bottle. And then it recommended three different starches. And I told it, you know, I buy these cheap cotton handkerchiefs and they're fraying. And it said, well, here's, you should try the Japanese Tunugui Hankachi or Cotton Batiste handkerchiefs or Irish linen handkerchiefs. And it said, really save money. Get those in on Etsy. Get the vintage Irish linen handkerchiefs because you'll save money and they'll be nicely

Patrick Beja [01:32:12]:
broken in

Leo Laporte [01:32:15]:
that you're not gonna get that from Bing. I'm just saying I just love that.

Wesley Faulkner [01:32:19]:
And I knew that was true, you

Leo Laporte [01:32:22]:
know, and then, yeah, it doesn't matter. It says it hallucinated.

Wesley Faulkner [01:32:25]:
Yeah, right. Patrick's right. It doesn't matter.

Leo Laporte [01:32:28]:
Well, I mean, in this case it probably doesn't, but I don't feel like any of that was bad advice. Yeah, I'll tell you, I ordered the Mary Ellen and I'll try it and I'll let you know.

Patrick Beja [01:32:39]:
I think hopefully this, the skill we will develop is to check the, you know, AI Overlords response when it does matter. Like, should I find what button do I Employees.

Lou Maresca [01:32:55]:
Yeah.

Leo Laporte [01:32:56]:
Never gonna do that. Yeah, yeah.

Wesley Faulkner [01:32:59]:
Although.

Patrick Beja [01:32:59]:
Should I press the red button? No. Yes.

Leo Laporte [01:33:02]:
If you look on X, you see a lot of people saying, you know, saying, well, you could just ask it which employee is most productive, etc. Etc. And I just, I don't, I hate that idea. That doesn't, that does not sound like the way to do.

Lou Maresca [01:33:15]:
It's not going to stop organizations from doing it though, unfortunately.

Leo Laporte [01:33:18]:
No, I'm sure they're doing it right now. So, Patrick, you said something before the break that was a little provocative. Apple, when it did a wwdc, it said a Siri AI. It's coming soon everywhere, but not the EU and China. And then they put out a press release saying we wanted to bring it to the eu, but the EU may, you know, said that it had to be interoperable. And we said, okay, well we can do that, but it's going to take us 18 months. And the EU said, well, you can't take 18 months. So we said, well, then we're not going to do it.

Leo Laporte [01:33:52]:
So I think that sounds like Apple isn't going to do it. You say it's just a marketing ploy.

Patrick Beja [01:33:59]:
I think they're doing exactly what they did with Apple Intelligence a couple of years ago. It's not ready. Apple intelligence for the EU is not ready. It's not ready in French, in German, in Spanish.

Leo Laporte [01:34:14]:
It's a language languages.

Patrick Beja [01:34:16]:
Well, it's not ready for English. It's going to come out as beta. Right, right.

Leo Laporte [01:34:20]:
It'll be beta October. Yeah, yeah.

Patrick Beja [01:34:23]:
It's not quite there.

Leo Laporte [01:34:25]:
Although I have to say the people who are using the developer betas have been pretty impressed so far.

Patrick Beja [01:34:30]:
Yeah, no, I'm sure it's fine. I think, and I would love to have it, but I think it's not. They, they say it's because of the semantic index that has all the data about your, your messages and emails and everything that they use to feed Siri that they don't want to give access to, to other AI companies, which I

Leo Laporte [01:34:59]:
understand the Digital Marketing act says it has to be interoperable. You could take the. I presume they mean in this case that somebody else could provide the intelligence with the same Siri AI. Intense.

Patrick Beja [01:35:12]:
Yes. So the, the, the idea in the DSADMA is that if you have a, a gatekeeper, a content, however it's called, then Apple with, with the iPhone and iOS is so dominant and same thing for Google by the way, that they can't use it to prop themselves up on another market. And there are a number of other markets and one of them is, you know, the upcoming AI chatbots market essentially. And so anything they do themselves they have to give access to to another company to, to other third party companies. That includes payment systems, the App Store, you know, all of those things, installing of apps. And of course if they, it wasn't an issue of a monopoly then they shouldn't have to be forced to do that. But in cases of monopolies you have to enforce special rules in order to safeguard capitalism and the American way of life, which apparently the US is not willing to safeguard. Because if you can't compete, you don't have a fair market.

Patrick Beja [01:36:23]:
And if you don't have a fair market, companies cannot emerge and challenge the incumbents in principle.

Leo Laporte [01:36:31]:
I completely think that's absolutely right. I mean I'm a supporter of interoperability but I also understand Apple's security complaint.

Patrick Beja [01:36:40]:
So they're going to do it. I think they're going to do it. They said they were going to do it. The issue is not can we or can't we. The issue is we can do it in 18 months is what Apple said. We need a special rule like an exception to have Siri available now and then in 19 months when our special agent, whatever which they have to write

Leo Laporte [01:37:02]:
something they said and we don't want to write it if you're not going to let us do this.

Patrick Beja [01:37:07]:
No, they said we need, yeah, well they said we need 18 months to make the inter.

Leo Laporte [01:37:12]:
And we haven't started and we will not start unless you say this is going to work.

Patrick Beja [01:37:18]:
So again I think this is B.S. i think they will do it because they want. They can't be at the market of the 350 million users market. They can't abandon it to anyone else or to, to other.

Leo Laporte [01:37:32]:
You know, it's a double digit percentage of Apple's market. It's not insignificant.

Patrick Beja [01:37:36]:
It's like 13 and by the way, last week Tim Cook had a very productive meeting with the, the EU Commissioner for Digital affairs or whatever the title is. I think it's the Wall Street Journal who had the rumor which I'm pretty sure you know, it was a Zoom meeting or FaceTime meeting between Tim Cook and the Commissioner. I think if the Wall Street Journal or whoever got was because Apple told them.

Leo Laporte [01:38:13]:
Yes, I agree but their preferred leak channel actually.

Patrick Beja [01:38:18]:
Yes, and so within, you know, six months, 10 months, whatever, when it's ready to go in French and German and Spanish it will be available in beta maybe a little bit later, but it is Coming, I, this is my speculation, this is my bet. It will be available. And again, people have so short memory spans. They did the exact, exact same thing with Apple Intelligence two years ago, announced it and they said, oh, it's not going to be available in EU because this and that. Now to be fair, there are a couple of features that are not available in the eu. Things like, you know, iPhone mirroring on the Mac and a couple of things, but those are not, those are things they can afford not to bring to the EU.

Leo Laporte [01:39:08]:
They're hoping that their customers in the EU will make such a stink that the EU will bend. Right.

Patrick Beja [01:39:15]:
Their customers in the EU will use ChatGPT and Claude there.

Leo Laporte [01:39:20]:
Yeah, there really is a. Then you're right.

Patrick Beja [01:39:22]:
Which is a vastly inferior experience because it doesn't have access to your semantic index. But you know, okay, I agree with the security implications. I agree with it. But it's not like in the iPhone. You can't give access to sensitive information to third party apps. You can give access to your, to your date, to your location, to your photos, to your, and you can control it. But you know, you can do that.

Leo Laporte [01:39:53]:
That would be the.

Patrick Beja [01:39:54]:
Of course it's a lot more, it's a lot more sensitive if you have messages and email and all of that in one. But you're an advocate for choice. I think companies should do whatever the hell they want in that respect. And if Apple just wants no one to have access to it, it should be allowed to decide that until they're a monopoly. Once there are monopoly special rules. I will remind you that in the US as well, at some point there were issues with monopolies and there were remedies put in place that were a lot more severe than what we are doing in the EU. You know, the breakup of AT&T, was it called AT&T back then that like they actually broke up the company.

Wesley Faulkner [01:40:40]:
Yep.

Patrick Beja [01:40:41]:
Which ended up coming back together.

Leo Laporte [01:40:43]:
But you know, long term benefit.

Patrick Beja [01:40:45]:
Sure, but they actually broke the company up.

Leo Laporte [01:40:49]:
Like those were the days we had actual antitrust enforcement in the United States.

Patrick Beja [01:40:54]:
And the reason, you know, there's this veil that's been put over people's eyes. As if I've become the defender of the EU for some. I'm, I'm.

Leo Laporte [01:41:03]:
Because you're in the EU while you're in.

Patrick Beja [01:41:05]:
No, I understand and I'm playing a tiny little bit of, of devil's advocate here because no one is putting those arguments forward usually in American shows, but there's this veil that's been put over people's eyes that the EU is being so mean to those companies. The reality is, as I was saying earlier, competition is essential to a functioning free market and capitalistic economy. If you don't have competition, the incumbents capture the markets and never let go. And that, that is not good for, you know, anyone. I mean we, we would have.

Leo Laporte [01:41:47]:
No, I enjoy my protection in the United States if it weren't for gdpr. I mean, thank you eu. Yeah, I guess now California and Illinois and other states have come along but, but it really started with GDPR in the.

Wesley Faulkner [01:42:03]:
And iPhones. We still have a lightning port.

Leo Laporte [01:42:05]:
A good point. You're right.

Patrick Beja [01:42:07]:
I don't think so. I think they were doing it anyway.

Leo Laporte [01:42:09]:
Maybe it accelerated it a little bit. Yeah but it's tough because I don't want governments to tell us what tech now technology standards to use. But on the other hand I do think governments have to protect us from monopolies. That's the only protection.

Patrick Beja [01:42:24]:
Frankly I think it's a, it's definitely a balance to find. But I think at the very least I think it is an understandable and defensible point of view to force open parts of the iPhone. I was against it for a long time and my thinking has evolved on it. And I think the idea that Apple has had the good fortune to and cleverness and business acumen and technology acumen to be there when the smartphones were coming online in a better way and the iPhone was absolutely revolutionary, there's no question about it. But it's been 20 years and I'm not sure like what happens if no other AI company has access to your personal data. If you give it access, if you give access to it, then clearly I think for consumers, for average consumers, Siri is the only option.

Leo Laporte [01:43:28]:
Yeah, right.

Patrick Beja [01:43:29]:
There's if, if it can't know what your messages say, if it can't know what your emails say, who you, you know what if it maybe can take action on your apps. If it can be agentic and Siri is the only option. Right. Siri wins by default. Kind of. I, I really think so.

Leo Laporte [01:43:47]:
No, that's locking. That's exactly what Apple's ecosystem lock in is all about. Lou. If Microsoft would just put out an open source, open standard phone, you know, here's their opportunity. You know, I mean we have like the fair phone, we have various Linux phones but they're not very good. We need one with a nice operating system. I actually think Microsoft phone was pretty darn good. But make it open, make it interoperable.

Leo Laporte [01:44:13]:
This is an opportunity. I think if you just call Satya and tell him him, I think he'll see where I see where I'm coming from.

Lou Maresca [01:44:22]:
I feel like he's been burned by that. We'll have to see.

Leo Laporte [01:44:24]:
Yeah, that whole thing a couple of times the Stephen Elopole.

Patrick Beja [01:44:29]:
Yeah, but, but so the point about the AI thing is that what's the next big frontier platform user interface. It's likely, possibly AI and if no one forces open the plat the existing platforms a little bit it then there's no question that Apple has it locked in. So the next platform, the next iPhone revolution, the next, you know, we've had personal computers, the Internet, smartphones and the next one is possibly going to be AI interfaces. And yes, if no one pushes forces open the existing platforms then Apple forces Siri in. And there's no question, to be honest, I don't think others have a big chance of outdoing Siri. I think Siri is going to be dominant anyway no matter what happens. At least if it's a little bit more open then there's a fighting chance. No one is installing third party app stores.

Patrick Beja [01:45:31]:
No one's doing it. Still, I would fight for the option to be able to do it.

Leo Laporte [01:45:37]:
It's not a risk really. I think you're right to Apple, but we should have that right. Boy, you made it. Very interesting contention. The only thing I would say is that history tells us that being an incumbent, a dominant incumbent in technology isn't necessarily a guarantee that you will then be able to move to the next stage. Almost invariably the dominant incumbents get unseated by out of left field. And so it's my hope that maybe Apple will not own this AI revolution. But I think you're right, they're very well positioned to do so and it's not a bad thing for the EU to try to break that monopoly to some degree.

Leo Laporte [01:46:18]:
Let's take a little break and then I'm going to tell you. Speaking of breaking what it looks like the states are trying to do to meta and it is, it is. It could be life threatening but we'll talk about that in just a little bit. You're watching this week in Tech. Patrick Beja. It's always good to have you on. You know what I like about you, Patrick, is there is, there does tend to be.

Patrick Beja [01:46:38]:
But I look really good.

Leo Laporte [01:46:40]:
You're so handsome. But besides that, there is a tendency to us to be a little bit us centric. It's one of the reasons I love having you on but you're also willing to take a contrarian stand to kind of. We all get in this kind of group think. And I think you often bring up points that are very, very good for us to hear. And that's a good one.

Patrick Beja [01:47:00]:
It's my thing on my shows. I actually have a jingle about me being the devil's advocate, because I do it all the time, including for the things that I'm convinced about. I try to take. Take the.

Leo Laporte [01:47:09]:
You know, I do the same thing. I think that's important to the discussion. But you believe it. That's the difference.

Patrick Beja [01:47:16]:
Or do I? Or maybe I don't.

Leo Laporte [01:47:19]:
Does he? Great to have you, Patrick Wesley Faulkner, who always brings heart and soul to this show. It's always great to have you on wesley83.com to learn about what Wesley's up to. And of course, don't forget, he is the founder of Works. Not working for people who are working but maybe don't love their job. Maybe they don't love it or try

Wesley Faulkner [01:47:39]:
to want to survive it or just want to make it maintain. Yeah.

Leo Laporte [01:47:42]:
Yep. Works dash. Not dash. Working.

Wesley Faulkner [01:47:45]:
Oh, I just wanted to say, if you haven't joined yet and you are joining the wait list, put twit in the little comment section so I can just greenlight you.

Leo Laporte [01:47:53]:
You can jump the line.

Wesley Faulkner [01:47:54]:
Yeah.

Leo Laporte [01:47:54]:
Nice. Thank you, Wesley. And Lou Mareska, he is probably the biggest shot we ever have on this show. He is engineering leader, Copilot, Microsoft. Pretty darn good. Pretty. Pretty important fellow. It's always nice to see you.

Lou Maresca [01:48:13]:
Let's hope so. It's important for a long, long time, right?

Leo Laporte [01:48:15]:
Yeah. Yeah, No, I think so. And you're riding the AI wave. I mean, you're in exactly the right place to be. You were pointing out offline. You didn't mention this on the show, but I saw in the discord when you were in the club that Copilot does have, besides MD files, it has semantic search through a variety of other tools. That's the key, I think, to AI, Right. Energentic AI, especially is that it remembers that it has some sort of sense of what happened and what your rules are and what you like and what you're looking for.

Leo Laporte [01:48:47]:
That's really key. So in June. So let's start by saying that there are four states, Colorado, California, Kentucky and New Jersey, which are going after Meta, based on meta's lack of protection for youth. Remember, Meta lost at a trial in Los Angeles earlier this year. They lost in a New Mexico action. But this one is a big one. The state's filings are sealed, but this is from Reuters. At a court hearing in June, they said they were calculating the penalties against Meta by multiplying the number of violations by fine amounts set by state law.

Leo Laporte [01:49:36]:
And the number of violations is calculated using the estimated number of teens and young users affected by Meta's actions. So we don't know the actual number. We won't really find out. There's a total of 29 states involved. We won't find out until August, or maybe not even then. But the trial begins in August. A judge whose name you may be familiar with, Yvonne Gonzalez Rogers, she must be very busy. She was the one who spanked Apple in the Apple Epic case.

Leo Laporte [01:50:05]:
At that time in August, the judge will address all the claims brought under law, plus the four particular states allegations the company violated their state laws. Reuters. And then there's 14 other states which have claims under their state laws. But Reuters estimates, based on the laws of those four states alone, actually, it's not Reuters. Meta has estimated that the penalty would be $1.4 trillion. Should point out Meta's market cap is $1.5 trillion. So, hey, what are you worried about, Mark? You'll still have 100 million left over. It's not the end of the world.

Wesley Faulkner [01:50:49]:
They can set up a payment plan, I'm guessing.

Leo Laporte [01:50:51]:
You know what, there's always GoFundMe.

Wesley Faulkner [01:50:53]:
Yeah,

Leo Laporte [01:50:56]:
Meta has been trying to cancel the trial. Rogers has been saying no. So this will be very interesting. Is it a jury trial? I guess it is. Juries have historically not been very good to Meta. That was the trial in Los Angeles. Meta. And let's see, it was meta and YouTube and Snapchat were involved in that one.

Leo Laporte [01:51:25]:
Tick Tock had pled. Made a settlement. I can't remember the details. In any event, this one's coming and it could be. I mean, EU is always talking about its giant penalties and there have been some pretty big ones, but this one, this is a doozy. I can't imagine the judge would let these states put Meta out of business. Or would she?

Wesley Faulkner [01:51:52]:
You'll know they're worried. If they start spinning off their companies as in separate entities, then you'll know that they're really scared.

Lou Maresca [01:51:59]:
Ah.

Leo Laporte [01:52:01]:
She has signaled that she intends to seat an advice. Oh, so it isn't a jury trial. It's just like the LA one. It's an advisory jury. She's done this before. So the judges got got the final call on what the liability and penalty will be. But the jury can advise her. It's not binding, but they can advise her.

Leo Laporte [01:52:24]:
New Mexico jury found that Meta had concealed what it knew about child sexual exploitation and mental health harms. I think that was $150 million judgment. California jury also ruled against Meta. Meta denies wrongdoing, but they don't have a great track record of defending these cases.

Patrick Beja [01:52:45]:
I mean, the, the 1.4 trillion thing penalty is a little bit of. I'm sure it's not going to be that. But even if it is, you know, I don't know, a hundred billion or fifty billion and it's still enormous. Like, obviously it would probably be the biggest swing imposed on anyone anywhere. And my initial reaction is, no, you can't do that. That's too much.

Leo Laporte [01:53:18]:
Right, but EU, the EU said asking for 4.1 billion from Google.

Patrick Beja [01:53:23]:
Yeah, well, which is a lot less than 1.54 trillion. But yeah, it's a lot of money.

Leo Laporte [01:53:28]:
It has to be a lot to sting these companies. When you find Elon Musk $15 million for pump and dump with Twitter stock, it's like, that's money in his couch cushions. That's, that's just not, it's not effective.

Patrick Beja [01:53:43]:
It's absolutely calc, you know, financial calculation on the part of those companies. And that's why a lot of the new EU rules like the DMA and DSA have percentage of global. Global. Yeah, Revenue as put as maximum minimum fines because the companies like optimize their financial operations to have minimal profits here and there. So you can't just fine on profits and. But my initial reaction is still, even in the eu, like, wait, that, that's too much. But we have to understand what we're talking about here. It's actual wrongdoing in the case of, of Meta, understanding the effect of its, you know, products and algorithms on youth.

Patrick Beja [01:54:33]:
Not just, oh, we don't know, we think that maybe, you know, it's not great for kids or it's like them seeing academic studies results saying, yes, this is bad.

Leo Laporte [01:54:46]:
And we know from the whistleblower leaks, Francis Hawkins whistleblower leaks, that they knew all of that, that they, they did in fact have that information and that they understand.

Patrick Beja [01:54:55]:
And, and Zuckerberg, let's take his name for lack of a, you know, knowing exactly who was in the loop, but I'm guessing he probably was not. Was somewhat aware of this, that the algorithm has this kind of negative effects on, on kids and this and that and that and then actively deciding. Sure. But we, we have no law that says we can do it it. And it's Making a lot of money. So we're going to keep doing it. Right. And this is what we're judging those companies.

Leo Laporte [01:55:27]:
This is how Meta.

Patrick Beja [01:55:28]:
I don't think this is unreasonable.

Lou Maresca [01:55:30]:
Right.

Patrick Beja [01:55:30]:
I don't think this is.

Leo Laporte [01:55:31]:
Operate this way.

Patrick Beja [01:55:32]:
They.

Leo Laporte [01:55:32]:
They do it, they do bad stuff and then when they get caught, they go, oh, whoops. Like the, the monitoring employees, you know, the latest one is this Instagram thing that they were doing, which they have backed off on, incidentally.

Patrick Beja [01:55:48]:
I think there's a difference between move fast and break things and doing questionable things with your employees and having actual adverse effects. Well, it happens to be on children, but validated by academic and scientific studies and keep doing it anyway. Right. This is not. Oh, we're going to record what our employees do for work.

Leo Laporte [01:56:12]:
Which cigarette industry did.

Patrick Beja [01:56:14]:
Exactly. This is exactly the right. Analog. Right. The cigarette industry knew that their products were bad for people's health. They had the scientific information and they still continued doing it. I think putting it in that perspective makes it. And so at that point you're like, well, yeah, why shouldn't we find them? Maybe we should.

Leo Laporte [01:56:36]:
And I don't put them out of business.

Patrick Beja [01:56:37]:
Should go to. Maybe it should go to a jury or to a. To a trial. Right? Maybe we should.

Leo Laporte [01:56:43]:
Well, it is going to a trial in August. Yes.

Patrick Beja [01:56:45]:
And that's a good thing, I think. I think this is a judge who

Leo Laporte [01:56:49]:
historically does not take any guff. She spanked Apple.

Wesley Faulkner [01:56:55]:
Yeah, I'll just say, better goes out of business for this, so be it. That's.

Leo Laporte [01:56:59]:
Yeah, I don't mind. So on on Tuesday. Here's a perfect example. Tuesday Meta unveiled the ability for anybody on Instagram to use your pictures unless you opt out. But the default was you're opted in your pictures for AI. New York Times says as a result, countless people's likenesses were used in AI images without their consent. Users complained by Friday, they backed down. They said, oh, whoops.

Patrick Beja [01:57:30]:
Who would have.

Leo Laporte [01:57:31]:
Who could have known that? Nobody would like that.

Wesley Faulkner [01:57:35]:
This is worse than just people opting out. Like they said, they will give you a switch where you can opt out. It is similar to the DNA testing that happens. It's not just you, it is everyone connected to you.

Leo Laporte [01:57:47]:
Yes. Your friends are friends. Yes.

Wesley Faulkner [01:57:49]:
So your photos. If you have photos of other people on your roll.

Leo Laporte [01:57:53]:
Oh, that's a good point. I have plenty of pictures of you.

Wesley Faulkner [01:57:56]:
Yeah. Oh, I think. Am I out of focus? Sorry.

Leo Laporte [01:57:58]:
Yes, they're all out of focus. Thank God.

Wesley Faulkner [01:58:03]:
If you're in someone else's picture, then boom. You are now part of the algorithm. And keep in mind that many brands that are on these platforms don't manage it directly. So they don't just go in and flip switches and do that. They use like a third party or an agency that may not have access to direct account. And so these are most likely less left will be left on for larger brands who don't go through that.

Leo Laporte [01:58:30]:
And there's going to be brands who might say, hey, you know, we want to have pretend that Leo Laporte endorsed our potato chips.

Patrick Beja [01:58:38]:
Yeah.

Leo Laporte [01:58:38]:
And there, there'd be nothing to stop them from doing that either.

Patrick Beja [01:58:41]:
Well, there's the law. They can't commercially use your. Your image to.

Leo Laporte [01:58:49]:
Well, I'm a. I think there would be. They can. I don't know what the rules are. You're right, I think it would be.

Patrick Beja [01:58:55]:
But I also think it's. It's a little bit. It's not very helpful to compare what's happening in this trial to the public. The things that those tech companies and especially meta do. That is perceived to be not cool. Right. This is beyond. It's not just not cool.

Patrick Beja [01:59:19]:
It's not just. Well, it's a little bit of an invasion of privacy but we kind of have a switch to, to kind of do it, maybe like disable it.

Leo Laporte [01:59:27]:
There's.

Patrick Beja [01:59:28]:
This is a more like a significantly worse thing that they did. They keep testing and I'm not saying that this thing is not. Is not bad. Right.

Leo Laporte [01:59:39]:
They have face recognition in the meta glasses that's not turned on, but it's in there. And they then they said oh no, no, never mind, we're not going to do that. They are now testing a prototype according to the Financial Times of super sensing AI glasses. And, and by the way, the time says this could be applied to existing meta ray bans that would use cameras and audio recordings to capture awares every movement. This is what I want. It's not on or it's not on all of them. It's on some of them. According to the Financial Times, executives are planning not to activate the lad.

Leo Laporte [02:00:13]:
When the super sensing features are used, the light does not come on.

Patrick Beja [02:00:20]:
Yep.

Wesley Faulkner [02:00:20]:
They're trying to diminish privacy or people's perception of privacy. They're trying to make it so ubiquitous that you're just. Or likeness or any of your data that people are like well I've given up. It's everywhere already. Which people feel already. And so they're just keep trying to find multiple avenues to have people be less sensitive to their.

Leo Laporte [02:00:44]:
Is there A charitable interpretation of this. Of some.

Patrick Beja [02:00:47]:
I can, I can give it to you.

Leo Laporte [02:00:48]:
Yes, good. Devil's advocate.

Patrick Beja [02:00:50]:
Please advocate. I, I will say, however, I agree with, with Leslie. This is pushing the Overton window on privacy, right? That's. It's like making things more. I agree. But I do think that as a technologist, there is a, an understandable desire to, to, to try to make AI like. AI can be so useful if it has all the information about you, if it sees what you see, if it hears what you hear. And you can just ask it stuff about whether what you're doing now or what you've been doing in the past days, weeks, months, and, and this is, this makes AI like orders of magnitude more useful as a personal assistant to you.

Patrick Beja [02:01:42]:
Now, of course the problem is that it assassinates privacy for everyone around you, which is a little bit of a problem. But on the technology side of things, if you think I want to use my AI, agent, assistant, whatever, to the fullest, then you need that. There's no, like, it will be severely diminished if you don't have all of that information from all of your sensors around you.

Leo Laporte [02:02:10]:
Poor Louis, having a hard time getting a word at edgewise. Lou, you actually put a great thought into the discord.

Lou Maresca [02:02:17]:
Yeah, I mean, I was just saying that Meta continues to treat consent, privacy, security as more of just a friction, right? Friction to go over, optimize and go find ways around it. But it's really a, it's supposed to be a requirement. It's supposed to be what you're all organizations supposed to follow. Right. And I think that's the one thing about a lot of their security policies that they have is bothers me a lot and I think going back to what Patrick is saying, like products can have these features built into them, but they only useful for people if they feel like they are secure and private for them. So if you're sending data to Meta and you know how they're going to use it, they will never be able to put of a product off the ground that people feel safe about.

Wesley Faulkner [02:03:03]:
Right.

Lou Maresca [02:03:03]:
And I think that's the biggest problem, right anymore.

Leo Laporte [02:03:05]:
Apple can do it and we still for some reason trust them. But not, not Meta. It's been a bad week for Meta. One more thing. EU regulators on Friday told Meta they have to make major design changes to Instagram and Facebook because the services are too addictive or else hefty fines. So it's just been a bad week for Meta. I'm gonna make the. Here's my contention.

Leo Laporte [02:03:32]:
Mark Zuckerberg has said many times, and I think he actually believes it, that the world is a better place when we're all connected, that we're interconnection. And I think he truly believes that. Right. That's the whole.

Wesley Faulkner [02:03:43]:
Did he say that from his private villa?

Leo Laporte [02:03:46]:
Yeah. By the way, this is the guy who bought up every house within a mile of his home so that nobody could be to able anywhere near him except his own security. But. But that's probably justified. I'm sure he's a target. But so. But I'm going to give him the benefit of that. Say he believes that, but that is an exact contradiction to what it's.

Leo Laporte [02:04:06]:
His users have said time and time again they want, which is privacy. So there's a but that's also the. The stone. Don't trust user. Those damn users. Users, they don't know what they want exactly.

Wesley Faulkner [02:04:21]:
Like he's saying, like I know best

Leo Laporte [02:04:23]:
you need to do things the way

Wesley Faulkner [02:04:24]:
that I'm going to force it upon you. And I've. When you. I mean when people don't get consent and they force their ways on you. We have words for that in the law.

Leo Laporte [02:04:36]:
It's not a good word either.

Wesley Faulkner [02:04:38]:
It's not. And the thing is that he has this gun that's full filled with people bullets because they have so many users on these services and he keeps pointing and shooting these people that he feels that he could just force his way through brute force and strength like oh, AI, shoot, Instagram, shoot. And that's what he feels like everyone that's part of these services that he operates is just ammo. That he feels that he could just brute force into specific areas and AI is one of them because he has all the information and these restrictions are very important. And if he feels that everything is connected and everything should be open, then why is he not making data portable? Why is he not allowing people who write bad books about this Meta to be published? It is security for me and not for thee. And that is just. It shows that he's just wants to be an orchestrator that controls everything. And I bet if you worked at Meta, you don't feel like the company's very transparent with their workers.

Wesley Faulkner [02:05:53]:
You don't feel that there is open sharing. And the best idea is when that it's not a metrocracy. So it is if. If you look at the way that companies organize and it shows how they orchestrate their products, you can tell that that's not his best. That's he doesn't have the best Intention, his A class shares override everyone. And I think that is the same theme and motivation that he is putting towards his products. And that's not good for the rest of us.

Leo Laporte [02:06:25]:
You might also say that Mark's real belief is the world is a better place when we're all connected to his advertisers. Certainly he's a richer man.

Patrick Beja [02:06:36]:
I mean, there is. The idea of the Internet initially was, I think, I guess we all shared it. Maybe I thought it was true.

Leo Laporte [02:06:46]:
Yes.

Patrick Beja [02:06:47]:
Yeah, I agree.

Leo Laporte [02:06:49]:
We're gonna, we're gonna, it's gonna bring us together. That differences in culture and country will disappear.

Patrick Beja [02:06:56]:
Yeah. The thing is, I don't think he is talking about that even like, if you, what is he saying when he's saying the world is better when we're all connected? He's not talking about opening yourself to other ideas. He's not talking about, you know, some actual sociological benefits. It, he's really talking about his platform and making sure that his platform is allowed to work the way that it makes maximizes revenue for the company. Right. He's not talking about, he's not out there, not even like, you know, Sam Altman and Dario Amadei writing philosophical essays about the benefits for humanity. He's not studying the effects of his products and of the ways people are connected through the Internet. He's not philosophizing on this.

Patrick Beja [02:08:00]:
He's just saying, oh, the world is better when we're connected.

Leo Laporte [02:08:05]:
How? Why?

Patrick Beja [02:08:06]:
Like it's been 20 years. We have data on this. He's not actually seriously thinking about this. I don't think if he really believed

Leo Laporte [02:08:17]:
it, would he be building a 5,000 square foot underground bunker home in Hawaii?

Patrick Beja [02:08:24]:
I think this has, I think he, he could have both of those. You know, it's, it's kind of go

Leo Laporte [02:08:29]:
either way is what you're saying. It could, it could be good, it could be bad.

Patrick Beja [02:08:33]:
I think he, he is not when he says that. I don't think he, he supports what he's saying with anything that makes him trust worthy in any way. You know, that's what I think. He's just saying that. And I, I don't think that is something that should be taken into account. And I think that again, if we think that these things, that there are things that are happening that are problematic, then there are ways of, of attending to those issues. And the ways to attend to those issues is through organized society, which is represented by your government and your judicial process. And you know, that's what some people try to do.

Patrick Beja [02:09:25]:
Like the eu, for example.

Wesley Faulkner [02:09:26]:
Yeah. I think he's lying to himself as much as he's lying to.

Leo Laporte [02:09:29]:
Yeah, that might be.

Wesley Faulkner [02:09:30]:
I think he's a little caught up in his own delusion and it makes it hard to untangle what he.

Leo Laporte [02:09:36]:
Actually, this happens when you, when you're worth billions of dollars, you start to lose. There's nobody around you who will say, mark, that's nuts.

Patrick Beja [02:09:45]:
But I think, you know, ultimately, for example, there, there. I think when we say there, you know, when they say those things about AI and the effects on humanity and society and, you know, I think they're at least actually convinced of it in the case of.

Leo Laporte [02:10:03]:
Of.

Patrick Beja [02:10:05]:
I don't know him. But in the case of Zuckerberg, I don't see it. And maybe it's because I don't like what he's doing to many products and I don't. Do you see, do you think when you think of Zuckerberg compared to the crazy AI maximalists, do you see in Zuckerberg someone who thinks deeply about the sociological effects of his products and the philosophical implications?

Leo Laporte [02:10:31]:
I thought the metaverse was going to be a really good idea.

Patrick Beja [02:10:35]:
Oh, stop. I have to defend him now. I agree with him on that.

Leo Laporte [02:10:40]:
That is not a really good idea. Nobody wants to wear a nerd helmet.

Patrick Beja [02:10:44]:
Okay, two seconds. Like two minutes. Rebuttal to that. I don't know about the helmet, but the idea of a realistic virtual environment enables essentially, if the Internet enables you to be. To talk to anyone, to have telepathy.

Leo Laporte [02:11:08]:
Well, aren't we doing this right now? This is it. This is as good as it needs to get what we're doing right now on Zoom.

Patrick Beja [02:11:13]:
I disagree. I think if you have virtual, realistic virtual places that you can go to that equate to the actual physical places that exist, then you have teleportation, then place doesn't matter anymore, doesn't matter where you are. And that is society. Change.

Leo Laporte [02:11:32]:
Come on.

Patrick Beja [02:11:33]:
Maybe it's not a good thing that

Leo Laporte [02:11:34]:
a virtual experience of Notre Dame would be superior to me going to Paris and going to Notre Dame.

Patrick Beja [02:11:42]:
Well, if the technology is good enough, there's no reason in you. It's the matrix, essentially what I'm getting. Maybe it's not 3D glasses that you put goggles.

Leo Laporte [02:11:51]:
It would have to smell like it, it would have to taste like it. If I lick, the smell is difficult to do.

Patrick Beja [02:11:57]:
I agree.

Leo Laporte [02:11:57]:
Yeah, I mean, it's the good. It's easier to go to Paris. Let me put. Let's just say it's just easier Even,

Wesley Faulkner [02:12:05]:
Even in a zoom call, we can. We have filters in which we beautify filters. And so there's going to be a bias or the idea of someone who's create this world, something that they will polish or something they will change. It will not be a complete fidelity. And also it removes the ability of leaving your mark in terms of being there when something happens, being there to see something. And also it removes the gaze of observance. Because now you can't just stumble upon things. Because.

Wesley Faulkner [02:12:36]:
Because even if you're in a freeform world, there are things that they allow you to see and things that allow you.

Leo Laporte [02:12:42]:
That's a good point. There's no serendipity.

Wesley Faulkner [02:12:45]:
Yes. And so you're removing that. So that's why the fidelity. Even if you have really photolistic things, you cannot recreate that process in a way that your interactions with other people.

Leo Laporte [02:12:57]:
You'll never see a murmuration of starlings flying overhead.

Wesley Faulkner [02:13:01]:
The weather's always great. The, The. The noise level.

Leo Laporte [02:13:07]:
Yeah, you could make bad weather in starlings.

Wesley Faulkner [02:13:09]:
But I'm not saying it doesn't have value, but only if somebody. I agree with what you're saying, Patrick, that it does have value. And it. For some people, that might be even a better experience if they don't have the opportunity.

Leo Laporte [02:13:20]:
I'm just glad I know better than nothing what a Notre Dame gargoyle tastes like. That's all I'm saying.

Patrick Beja [02:13:27]:
What we're talking about is.

Leo Laporte [02:13:29]:
I won't tell you why.

Wesley Faulkner [02:13:30]:
Immediate.

Patrick Beja [02:13:31]:
The immediate it comparison to what we already know. I think in the same way that someone in the. What we're imagining the effects of this in the same way that people in the 70s were imagining like, you know, the, The. The cyberpunk.

Leo Laporte [02:13:48]:
Right?

Patrick Beja [02:13:50]:
Ah, the. The. The. Well, the metaverse, essentially, you know, it was like, oh, it's going to be like this. Like we. We don't know The. It was impossible to imagine the effect of the Internet in even the 2000s. In the late 90s, early 2000s, it was impossible.

Patrick Beja [02:14:07]:
And I think it's the same thing with maybe at some point the metaverse exists. The metaverse is not.

Leo Laporte [02:14:14]:
Go light a candle in a chapel in Notre Dame. And I don't blame.

Lou Maresca [02:14:17]:
Of course, can't do that in VR.

Leo Laporte [02:14:19]:
I'll tell you, I lit a candle for this guy. He is a saint. Carlo Acutis. Do you know about this guy? A teenage saint. He was canonized by Pope John Paul. I think he was a teenager. He's a saint. Literally.

Leo Laporte [02:14:39]:
See, in fact, in the little statue they have of him at the church I went to, he's holding a laptop because when he was born in 91, he died of leukemia. He was actually fatally ill through most, most of his life. But he became very devout and he made a bunch of websites for the church. He loved the saints, especially St. Michael. And I don't know what his miracles were. It doesn't say in this little pamphlet. His love for others was as ordinary as helping children with their homework and as remarkable as serving the poor beside Mother Teresa's sisters and at the Capuchin soup kitchen.

Leo Laporte [02:15:24]:
So he was an Italian boy, gifted computer programmer. He's the new saint for all us C guys. He gained fluency in Java and C and designed websites for his parish and school, helping others with technical issues. Passed away at the age of 15 in 2006 from leukemia. And he's a saint. So you could light.

Patrick Beja [02:15:47]:
Maybe you should.

Wesley Faulkner [02:15:49]:
Someone touched him and got healed. I think that was one of the.

Leo Laporte [02:15:51]:
Was that one of the things is miracles?

Patrick Beja [02:15:54]:
Yeah, I mean maybe we should spend more time talking about people who do nice things rather than.

Leo Laporte [02:15:59]:
I think he's a nice thing and you could light a candle in front of him and if you had a problem with your computer, he would be the guy to go to. All right, I'm just being silly, but I thought this was amazing when I stumbled across him at the, the church, the Star of the Sea church in San Francisco. They have a little statue to him and I lit a candle for him because I thought. Well, you made me think, made me think of that, Lou. Yeah, I said that. Yeah. If you're ever in San Francisco, go to the Star of the Sea Church and you can see St. Carlo Acutis, born in Milan, born in London, 91, died in Milan in 2014.

Leo Laporte [02:16:39]:
Beatified in 2020, canonized. Sound like he said last year. Locutus Carlo Acutis.

Wesley Faulkner [02:16:46]:
Acutis not.

Leo Laporte [02:16:47]:
Look, could be that 150 years from now they'll think it was Locutus. You never know.

Wesley Faulkner [02:16:56]:
Locutus, yeah, Star Trek reference for those who knew.

Leo Laporte [02:17:00]:
Yeah, we get, we get them all in there all the time. We're take a break. We've got some silly stories coming up and then I want to talk about actually really the most controversial story of the week, which is Sony abandoning physical discs.

Patrick Beja [02:17:17]:
But

Leo Laporte [02:17:20]:
I had a long conversation with my 22 year old about this. He's incensed. He is so pissed off. Sony is ditching discs and gamers are pissed now. My son Michael brought up a good point. He said what he's done with the game, he will bring it to GameStop and sell it used. I guess Sony didn't like that too much. Maybe, maybe that's one of the reasons they're getting rid of it.

Leo Laporte [02:17:48]:
But it certainly helped support his game. He said he's even ordered the disc for Grand Theft Auto, the next Grand Theft Auto already.

Patrick Beja [02:17:57]:
Well, I have bad news for him.

Leo Laporte [02:17:59]:
What's the little number? Isn't it.

Patrick Beja [02:18:03]:
It's a code.

Leo Laporte [02:18:04]:
It pisses him off. You get all this stuff, you open up the CD jewel box and it's a code.

Patrick Beja [02:18:13]:
Well, I mean it's the way of the world.

Leo Laporte [02:18:20]:
What it is is it's an eye opener. You don't really own anything, do you?

Patrick Beja [02:18:24]:
Right? Yeah, but you didn't before either. And yeah, there were a few games that you could actually install from your disk and play, but they were honestly quite few and far between.

Leo Laporte [02:18:35]:
No, and it was a little annoying when you'd get a disc, you'd put it in and then you'd have to do 180 gigabyte download to play the game.

Patrick Beja [02:18:42]:
Yeah, it's a lot of games don't fit on the disc anymore. I think it's like 80% of Sony's sales for games are actually digital. Now that's a little bit misleading maybe because it includes smaller games that actually don't have a physical release. But in general, like most games you put, put it on your disk, it downloads a lot. They, many of them have a, A, an online component. It's rare that you can get a disc and actually play it. I understand the issue with selling it and so you kind of reduce the price of your purchase by being able to sell it. And then some people can buy them used so it's cheaper.

Patrick Beja [02:19:37]:
But there's a plat. There's a platform that already has done away with that, that with physical sales. It's PC. And the effect of that is that

Leo Laporte [02:19:49]:
you have, you don't buy discs anymore.

Patrick Beja [02:19:52]:
You could, I guess, but I don't think it, it's possible you could go

Leo Laporte [02:19:56]:
to an office supply store and buy a copy of Microsoft Office for like, well, hundred dollars.

Patrick Beja [02:20:03]:
I don't think games would, would, you know, they work like that anymore. And, and I mean, sure, for some AAA games on release, maybe you're going to have to pay a little bit more. But on average I would be willing to bet that given that there's no used game market, sales on the main platform, which is Steam, are constant, constant. It's almost impossible to Buy a game full price nowadays and, and developers, especially smaller developers, can sell their games for a very, very long time. The long tail is insane on PC because you don't have physical copies anymore. Now of course you can't lend a game. It's very difficult, it's impossible to sell one. But again, the market kind of compensates for that.

Patrick Beja [02:21:03]:
I'm not sure we're worse off as gamers and for the prices we pay in an old digital market, because sales are more important. The issue that I understand a bit more is preservation, but even that is so difficult to achieve nowadays. The saving grace for preservation is piracy. If we didn't have piracy, I would be a lot more concerned. But we do have piracy and it's

Leo Laporte [02:21:33]:
thank God we can pirate that data and we don't have to ever worry about it.

Patrick Beja [02:21:36]:
Absolutely. Without piracy, preservation does not exist in the video games industry.

Benito Gonzalez [02:21:43]:
This is Benito, I need to rebut

Benito Gonzalez [02:21:44]:
that real quick because on PC we have sources like good old games where

Benito Gonzalez [02:21:48]:
we can get DRM free games and

Benito Gonzalez [02:21:50]:
keep those files forever. So we do own those games. And there is a difference between that

Benito Gonzalez [02:21:54]:
and buying a game on Steam, where,

Benito Gonzalez [02:21:56]:
yes, they can just remove that game

Benito Gonzalez [02:21:57]:
from the library and you no longer

Benito Gonzalez [02:21:59]:
have that game ever again.

Leo Laporte [02:22:01]:
Sony is threatened that if you don't play a game, I can't remember what the period of time was for six months or something, that they could remove it from your collection.

Wesley Faulkner [02:22:10]:
Yeah, and there's that

Leo Laporte [02:22:14]:
org petition. 297,000 signatures as of our broadcast time to save physical games. Almost 3,000 signatures.

Patrick Beja [02:22:25]:
I think Love and Deep Deep Space had more signatures to keep their latest character. No, I'm not kidding.

Leo Laporte [02:22:32]:
I don't think it's going to change Sony's mind. I agree.

Patrick Beja [02:22:35]:
They stopped production.

Leo Laporte [02:22:40]:
Yeah.

Patrick Beja [02:22:40]:
They're closing factories, they're repurposing them. And to Benito's point, yes, you can buy games on. On Google games. Look at the market share of Google games versus Steam. People don't actually care until.

Wesley Faulkner [02:22:53]:
Until Steam starts pulling games from the library. But once that does start happening.

Leo Laporte [02:22:57]:
Why, why would they?

Patrick Beja [02:22:59]:
They don't.

Wesley Faulkner [02:23:00]:
50 years from now, when we are no longer here, but kids who are 15 are still here, that might happen.

Patrick Beja [02:23:06]:
And that's why I think piracy is important. I agree with you on that. And that's why I think piracy is important. I'm not kidding. Actually, piracy is in this special part,

Leo Laporte [02:23:14]:
but the Internet archive, which does preserve old games, even in the face of copyright, for that very reason, because it's art, it's A creation just like we should save old movies, these are. This is a form of art that should be preserved beyond its commercial value.

Patrick Beja [02:23:32]:
And I think that digital media is going away no matter what. So if you're really pissed about this, you want, you would need to. What you need to do is try to fight for consumer rights on digital media. That's what you need to do, because that's the future. I think it would be misguided to force companies to let you know to resell your digital games because they, they don't get where they don't. They have no wear and tear. Right. So they would be there forever.

Patrick Beja [02:24:02]:
And it would have an adverse effect on the market, I think. But if that's what you want, don't fight for digital for.

Leo Laporte [02:24:08]:
Should they charge less for a digital copy of a game than they do for the physical copy?

Patrick Beja [02:24:14]:
There's arguments there. But the, the price of not much

Leo Laporte [02:24:18]:
to make a cd, I'll be honest.

Lou Maresca [02:24:20]:
Right.

Patrick Beja [02:24:20]:
No, but that's not even the issue. The issue is that the price of games has not. People say, oh, games are so expensive, the prices rise so little that even now with the latest increases in the average price of the highest priced games, they're lower, taking into account inflation, than they were 20 years ago. Significantly lower.

Leo Laporte [02:24:47]:
Like, forget that Sony pulled more than 500 movies from the PlayStation Store. Movies from Studio Canal. They'll be gone as of September.

Patrick Beja [02:24:57]:
Who was buying movies on the PlayStation?

Leo Laporte [02:25:00]:
Okay, all right. Admittedly that was a bad move, but the fact is Sony could do it. Apple could do it.

Lou Maresca [02:25:07]:
Yeah, right.

Patrick Beja [02:25:07]:
Yeah.

Leo Laporte [02:25:07]:
Anybody that you. I mean, who buys physical discs much anymore? Should people buy physical discs? I can tell you kids are who buys physical discs? Wesley, who?

Lou Maresca [02:25:18]:
Libraries.

Wesley Faulkner [02:25:19]:
Like we've used the word library. But actual libraries do hold these games and exchange will shut them out. So you cannot check out these games via your local library.

Patrick Beja [02:25:33]:
Jason Trier made an excellent point about libraries. You can rent. You have programs for books, for digital books.

Leo Laporte [02:25:39]:
They could do the same thing for digital games. Games.

Patrick Beja [02:25:41]:
Yeah, they could.

Wesley Faulkner [02:25:42]:
But have they announced it?

Leo Laporte [02:25:43]:
Sony would have to.

Patrick Beja [02:25:44]:
No, no, of course not. But that's.

Leo Laporte [02:25:46]:
Hey, guess what?

Patrick Beja [02:25:47]:
Governments can force stuff, all right? But governments is the expression of the people's will collectively on society. That is what people in the US do not understand. They think of governments as this outside force that is here to prevent you from. From doing stuff. And sometimes it is. But who would enforce this rule that you have to have a digital rights management that is more fair for consumers and have libraries enable a system that would Let libraries rent out games, which is a very important thing, I think.

Leo Laporte [02:26:28]:
Kevin, Kevin King, one of our producers says I still have Expedition 33 and the shrink wrap have. When you haven't opened it, you're not gonna. You should play the game. I guess you don't have to open it, right? You just, you just, you just play the game. It is, it's a fun game. You actually, you were the one who taught me, told me about this Patrick, way back when I played it.

Patrick Beja [02:26:46]:
Incredible.

Leo Laporte [02:26:47]:
Because of you. It's French. A French game. Let's see. Actually one of the kind of related story. There's been a lot of talk talk on X lately that you should download your favorite open weight model because the Chinese government is making some noise about cutting off the west from these wonderful Chinese models like Deep Seqin, glm. These are really good models. You pointed to a.

Leo Laporte [02:27:25]:
A really interesting GitHub repository where I think he's a high schooler young guy has created a version of glm. He calls it Calibri, which is Italian for hummingbird. That is actually a full fledged 744 billion parameter version of GLM, which is a very good model, Chinese model that can run in 24 gigs of Ram using its mixture of experts model and loading in little chunks from the drive. I'm going to right after the show, I'm going to install it and try it because I love the idea of the local model.

Patrick Beja [02:28:00]:
So that's crazy.

Wesley Faulkner [02:28:02]:
It's slow as heck.

Leo Laporte [02:28:03]:
Yeah, yeah. It's four tokens per second or something, but you can do it.

Wesley Faulkner [02:28:09]:
It's also GPU only. No GPU acceleration whatsoever.

Leo Laporte [02:28:13]:
Right. Okay, we got a ways to go. But I think that's an interesting point. We saw the rug pull from the US government on Fable. There's no reason the Chinese government might not do the same thing to Chinese models. And those are the models we've all gone to because they're open weight. So download all your models, kids.

Wesley Faulkner [02:28:33]:
As long as you have BitTorrent, you're good.

Leo Laporte [02:28:35]:
I couldn't get the physical copy of GLM, so I just downloaded it. So that's. I don't have the disk.

Patrick Beja [02:28:40]:
I mean the ones that are already out there, I don't think you need to download. You'll be able to find them somewhere.

Wesley Faulkner [02:28:46]:
Yeah, yeah.

Patrick Beja [02:28:46]:
The, the, the concern is for the.

Leo Laporte [02:28:48]:
Well, the future.

Patrick Beja [02:28:49]:
The question is for the future.

Leo Laporte [02:28:50]:
And that's really the question. These models are. Lou, what do you think the timeline is for local models? If we put a stake in the sand right now, how soon before local models will approach the quality of the frontier models we have today?

Lou Maresca [02:29:06]:
Well, that's the trick, right? Approaching the quality of L like frontier LLMs is going to be very hard. It's all. It's an extra economic problem, really.

Leo Laporte [02:29:13]:
You can't do a 10 trillion parameter model locally ever.

Lou Maresca [02:29:17]:
But I think SLMs will come into fruition very, very soon. And they will be specific for very specific things. I think they will do better at even foundational models at that.

Leo Laporte [02:29:27]:
That aspect is small language model.

Lou Maresca [02:29:29]:
Right.

Patrick Beja [02:29:30]:
So you're saying what specific things will they. Will they be better at?

Lou Maresca [02:29:35]:
Like for instance, you know, there's. There's actually SLMs that they're using right now that are actually better at translation because they have better access to data locally than these big foundational models have. So it's access to data which is part of it. They do just as good or even better at translation because of it.

Leo Laporte [02:29:52]:
I was so surprised that this Quinn model that's tuned for recognizing the contents of images is so small. I can run it on this machine and give it my images from my security cameras and it can analyze them. It says a woman in a red hat is walking down the stairs. It's really good. And it's a lot local. It's a tiny little local model that's running.

Lou Maresca [02:30:11]:
It's an. Like I said, it's an economic problem too, because organizations don't want to pay for the foundational models as well. So people don't want to pay for the foundational models. They want to. They want to pay for what they're going to use it for. And that's where SLMs come in, I think.

Leo Laporte [02:30:23]:
Is Microsoft working on that kind of thing? You don't have to tell us.

Lou Maresca [02:30:26]:
Oh, yeah, I think they are. Absolutely. I think the MAI models are starting to get there.

Leo Laporte [02:30:30]:
Ah, interesting. You call them mai, not may I call them mai? Yeah, I need to know how it's pronounced. And fee or phi or foe or

Lou Maresca [02:30:41]:
fum, gif or jif. Yeah, I think it doesn't really matter.

Wesley Faulkner [02:30:44]:
Okay.

Leo Laporte [02:30:45]:
You know, it's funny because a lot of what happens in technology, we read, maybe that's changing. But we read articles online, you know, we read the news stories, but we never hear them said out loud. In my job, I have to pronounce all this stuff, so sometimes I just make it up and hope that it sticks. Right?

Patrick Beja [02:31:06]:
Yeah.

Wesley Faulkner [02:31:07]:
So many times I hear Nvidia.

Leo Laporte [02:31:09]:
I'm like, no, we know better than that. Right? But. But I don't know if it's fee or that's a Microsoft local model. It's fee or fi, I don't know which. But anyway, neither does it's foe.

Wesley Faulkner [02:31:20]:
It's fo.

Lou Maresca [02:31:22]:
Fo.

Leo Laporte [02:31:22]:
Yeah, I guess it could. It's whatever you want it to be.

Wesley Faulkner [02:31:28]:
But also with these local models they own, they also can be just good enough for tool usage. That's right. The advent of MCP servers and stuff like that where you have these structures being able to offload to the actual tasks to maybe even a third party service like say you want to send an email, you don't have to have all the logic of how to send an email. It could be just the logic of being able to shepherd it off to a service. And so depending on what the specialty that you need, some of it also can be paired away because if you can offload the work to whatever will take it off your plate.

Leo Laporte [02:32:03]:
That's also exactly what I do with my Hermes agent. I use a local model for the tool for tool stuff for MCP servers for skills because it, you know, well not so much for skills because it does need to do some work there. But for certainly for MCP servers, looking up the email, making a calendar appointment, things like that, there's doesn't need hefty brains for that. So that's I think right. Sizing model might, might be the future

Lou Maresca [02:32:27]:
for consumer like for real world. People who want to use models today, they don't have, they're not coding, they're not doing all that. They want to use an app or tool or appliance that has this built into it. And that's where SLMs come in. Because now you have, you have low cost hardware that can run them.

Leo Laporte [02:32:41]:
Right. I have all these web search tools and things like searching and stuff on my server. So it doesn't need to be smart, it just needs to know what to call and let it do the work. That makes perfect sense. A Waymo called the cops on some teen riders and locked them in. This happened Monday in San Mateo, California. Police reported they actually put this on social media that they had apprehended a pair of teenagers from a Waymox after the company alerted authorities to suspected criminal activity. This was on Facebook the San Mateo county police posted.

Leo Laporte [02:33:27]:
And I think this might have not been the best thing to do. Parents, do you know where your teens are? Waymo does. The 15 year olds were allegedly drinking alcohol and shooting toy guns from the car. Waymo's systems probably the guy at the home office who was looking at the camera detected Behavior triggered a safety response, disabled the vehicle and contacted police. It doesn't say they locked the kids in, but I gotta think the kids couldn't get out. I mean, I would.

Patrick Beja [02:33:57]:
I mean, no, this is. That has to be the most illegal, impossibly wrong invasion of private. Like locking them in, disable the vehicle. I understand. Calling the cops. I understand Locking them in, putting them in prison.

Leo Laporte [02:34:14]:
Say they did.

Patrick Beja [02:34:14]:
No.

Leo Laporte [02:34:14]:
Maybe it said something like, oh, hold on a minute. I gotta. I gotta hold on. We're having trouble here. Reaching the home office. Just hang tight for five seconds here. And then the police come. I don't know.

Patrick Beja [02:34:27]:
Please remain calm. The authorities are on their way.

Leo Laporte [02:34:31]:
In an email to npr, San Mateo Police Department spokesman, spokesperson Janine Luna said detaining the teens in the Waymo was, quote, wholly appropriate under the circumstances. What?

Patrick Beja [02:34:45]:
Detaining the teens?

Leo Laporte [02:34:46]:
Oh, wait a minute.

Wesley Faulkner [02:34:47]:
Oh.

Leo Laporte [02:34:48]:
Oh, no, you're right. The vehicle was disabled. The occupant had every right to exit before police arrival. But they did not. They stayed there.

Patrick Beja [02:34:57]:
Listen, I've watched the Practice all seven seasons of it. I can tell you this would get way more in. That's kidnapping.

Wesley Faulkner [02:35:07]:
That is kidnapping.

Patrick Beja [02:35:08]:
Yeah, Yeah.

Leo Laporte [02:35:09]:
I was rewatching one of my favorite series, Silicon Valley, which, really, interestingly, one of the reasons I'm rewatching it is they got AI Right when they. When they made this, like, eight years

Wesley Faulkner [02:35:20]:
ago, like, all Son of Anton.

Leo Laporte [02:35:22]:
Yeah, yeah. It's like, all right. On so. And of course, it has the famous scene with the driverless car, with which, in this particular ancient version of the world has, like, gears on the steering wheel and a pedal on the brake pedal. Like a thing on the brake pedal. It's all actuated, but it's still a driverless car. And remember, he gets into it and he's going to his destination, and all of a sudden it gets new instructions and he puts him in a container ship and sails him out into an oil derrick in the middle of the sea. I just.

Leo Laporte [02:35:56]:
I.

Wesley Faulkner [02:35:57]:
Great.

Leo Laporte [02:35:57]:
Great show. Which actually holds up surprisingly well given how technology has changed over the years.

Patrick Beja [02:36:03]:
But. But wait, so it's interesting what happened with the. The teens. It was clearly an operator that was surveilling what was happening in the vehicle, and it called the cops. I guess it could be said it's the same as a driver could have called the cops, for example.

Leo Laporte [02:36:20]:
Sure.

Patrick Beja [02:36:20]:
But the. The. The person judged that the kids were underage and drinking, and what was the. Were they shooting toy guns?

Leo Laporte [02:36:32]:
Like, toy guns out the window? Yes.

Patrick Beja [02:36:35]:
Is that even, like. Is it that way.

Leo Laporte [02:36:37]:
Illegal. It's just to call the cops behavior. Well, you don't know if it's a real gun. I mean.

Wesley Faulkner [02:36:43]:
Okay, here's another legal argument. We're moving into the expectation of privacy here. So the evidence that was in the car, is that admissible because they have. There's a possibility they can claim they have the expectation of privacy. So if they were in, let's say, a regular Uber and they're in the back of a seat backseat, and they're like doing something that could be constrained as illegal, there's not the requirement that those. That it be reported to law enforcement.

Leo Laporte [02:37:15]:
I don't think you have an expectation of privacy if you are in any biz place of business and. And you do something.

Wesley Faulkner [02:37:23]:
If I go to the bathroom at a Walmart, I have an expectation of

Leo Laporte [02:37:26]:
privacy in the bathroom. Right, but if you go to the bathroom in the aisle.

Wesley Faulkner [02:37:30]:
No, no, but where's the line is. I think that that is.

Leo Laporte [02:37:34]:
The line is the door to the bathroom is an Uber, like a bathroom.

Wesley Faulkner [02:37:40]:
But like, say, let's say the cops

Leo Laporte [02:37:41]:
were actually, I've been. I've been in some Ubers that actually. Yes, the answer is yes, but that's another story.

Wesley Faulkner [02:37:48]:
Yeah, maybe you're right. Maybe I'm making too much of this. But I think that there's. There might be some debate. What is your expectation, what disclosure There is.

Leo Laporte [02:37:56]:
I think a cab driver absolutely has the right if. If you are an underage teen drinking in his cab to call the cops.

Patrick Beja [02:38:04]:
I guess, maybe. Yeah, I think.

Wesley Faulkner [02:38:06]:
Yes. But if you. If you hid all that stuff and put it in your bag and then the cops come, they don't have the right to search you. They don't.

Patrick Beja [02:38:16]:
Maybe we don't know what happened then. Or do we.

Leo Laporte [02:38:19]:
Probable cause. So they do if there's probable cause, if they have a reason to believe there's a crime.

Wesley Faulkner [02:38:24]:
And. And the question is, do you have. Do you explicitly give the right to that video that you had to a commercial company, to the authorities? So if just to say just. Just because a commercial company has it doesn't mean the authorities automatically have the right to have it without a subpoena.

Patrick Beja [02:38:43]:
I want the practice to come back and do episodes on all of these. Totally. And explore the legal.

Leo Laporte [02:38:49]:
Or the Good Wife. Yeah, yeah, yeah. There were a couple of shows that were always on the cutting edge of technology.

Patrick Beja [02:38:56]:
Yeah, the Good Wife was really good about those.

Leo Laporte [02:38:58]:
They would. They would come up with stuff that was like, straight ripped from the headlines. It was fascinating. Yeah, yeah.

Wesley Faulkner [02:39:03]:
All right, I digress. Maybe I'm just. Just think about it.

Leo Laporte [02:39:06]:
No, it's good.

Patrick Beja [02:39:06]:
You're thinking.

Leo Laporte [02:39:07]:
You're putting on your law. It's good.

Patrick Beja [02:39:10]:
Way more interesting thing.

Leo Laporte [02:39:12]:
Like some of it's. Go ahead.

Patrick Beja [02:39:15]:
No, the fact that it feels weird because it feels like the autonomous car slash the company. Waymo called the cops on you for drinking when you're underage, which is not the most egregious crime. You can, you know, they were.

Leo Laporte [02:39:33]:
The kids, by the way, were not thrown in jail. They were released to their parents.

Patrick Beja [02:39:36]:
Right? Yeah, of course. Yeah, but. But still, it's like, okay, if that happens, then what else happens? Right? And I don't know, it feels weird.

Leo Laporte [02:39:47]:
I would not commit a crime in a. In a Waymo. I'm just saying, I think that's public aisle of a.

Patrick Beja [02:39:53]:
Of a.

Leo Laporte [02:39:53]:
Or in a Walmart.

Patrick Beja [02:39:54]:
Yeah.

Leo Laporte [02:39:55]:
Waymo does say some of its robo taxis ran out of power because there was gridlock in the 4th of July and they had to be towed. Was there gridlock before or after they got stuck in the middle of the road? I don't know.

Wesley Faulkner [02:40:09]:
And by the way, also some issues like where they are driving through floods and stuff like that. I think some of these new one off things, they're still working the kinks out.

Leo Laporte [02:40:17]:
Here's a Waymo driving into a firework. This is an Instagram. Whoa. It just drove right into it. What the heck? Who cares? Happy fourth of July. And it's smoking and it's steaming. And Waymo's been running ads lately. I've seen some ads locally because we have waymos in this area.

Leo Laporte [02:40:40]:
And you know that one of the things they say in the ads is how many injuries occur each year caused by humans driving and how safe waymos are. And I think on balance they are. I mean, they are. They're safer than drivers. Yeah.

Wesley Faulkner [02:40:57]:
If we had as many waymos as we did regular cars, then we could do it. No, no.

Leo Laporte [02:41:01]:
They could do a. Per mile. They could stay in a.

Lou Maresca [02:41:03]:
Per.

Patrick Beja [02:41:04]:
You know, they actually starting to be significant and.

Leo Laporte [02:41:07]:
Yeah, I think so.

Patrick Beja [02:41:08]:
Who was the, the, the doctor, the surgeon that was making the case that this is actually a question of public health?

Leo Laporte [02:41:15]:
I think it is. Like the humans are terrible drivers. Let's face it. The funny thing is we all think we're great drivers and it's the other guy who's a terrible driver. Right. Am I wrong?

Lou Maresca [02:41:27]:
It's true.

Leo Laporte [02:41:29]:
We're. All I can tell you is like the Rain Man. I'm an excellent driver.

Patrick Beja [02:41:32]:
The two Deers. The two actual real deers and rabbit that I did not drive over on the way here in the countryside. Think I'm a great driver.

Leo Laporte [02:41:44]:
You. You, you did. You. You avoided them.

Patrick Beja [02:41:47]:
I did avoid them. Although if you can't avoid them, your best bet is to accelerate so that they go over and don't go into your windshield. So that's something.

Leo Laporte [02:41:57]:
Wait a minute, wait a minute, wait a minute. I gotta learn this new French driving rule. So.

Patrick Beja [02:42:02]:
It's not French.

Wesley Faulkner [02:42:03]:
You go too slow, they'll go up on your hood and then go through the windshield. If you go too fast, then it helps.

Leo Laporte [02:42:08]:
You want to go flying higher in the air?

Patrick Beja [02:42:11]:
Yeah, yeah. Over the car.

Leo Laporte [02:42:14]:
Did you. Okay, so here's a question. Did you have the thought as you were approached whether you should speed up or slow down?

Patrick Beja [02:42:23]:
Well, I drive carefully. So I saw them and.

Leo Laporte [02:42:25]:
And plenty of time.

Patrick Beja [02:42:27]:
Pay attention. So I had a. But there was one that was. I mean, people who live in these areas know, but it was just hanging out. It was. I was.

Leo Laporte [02:42:35]:
I was like, dude, like, I'm. Yeah.

Patrick Beja [02:42:37]:
And. And it looked at me and I like honked my horn and it looked at me and it was like, okay, fine. And they moved. But I was like, it's almost like a teenager.

Leo Laporte [02:42:47]:
They're not going to run off. They're going to show you. They're going to saunter off the road.

Patrick Beja [02:42:52]:
Yeah, basically. But the lesson is I'm happy that we're hopefully getting Waymo in. In France at some point. They just created a company in France, Germany and Spain and the legislation is probably going to make it possible for them to operate like by next year. And I find it incredible that this technology is actually working, that autonomous cars. You know, this is the perfect example of the Gartner hype cycle. Right. 10 years ago, well, maybe 15 years ago now, we were all convinced that autonomous car were.

Patrick Beja [02:43:30]:
Cars were coming within two years. And it was like, oh my God, this is coming. This is incredible. This is. They didn't. And everyone forgot about it. It was like, ah, it's never gonna work. And now it's working and I feel like no one cares.

Patrick Beja [02:43:45]:
But it's incred. Like we have cars actually driving themselves.

Wesley Faulkner [02:43:51]:
Like

Leo Laporte [02:43:53]:
it's.

Patrick Beja [02:43:53]:
We are living in the year 2000, I think. I think this is like happened Finally. Yeah.

Leo Laporte [02:43:59]:
26 years late, but it happened.

Lou Maresca [02:44:02]:
Yeah.

Patrick Beja [02:44:02]:
No, by the way, it.

Leo Laporte [02:44:05]:
A person in our discord chat says, yeah, if you hit a deer or smaller, the animal dies, but if you hit a moose, the car dies.

Patrick Beja [02:44:12]:
Oh, yeah.

Leo Laporte [02:44:13]:
Person suspiciously named Moose Espionage posted this. So I'm thinking. I'm thinking we have a moose in our club. I'm just saying.

Patrick Beja [02:44:22]:
No, I think a moose. A moose. You don't. Maybe this is like propaganda, but you do not want to hit the moose at.

Leo Laporte [02:44:29]:
Not speed up.

Patrick Beja [02:44:31]:
I actually, it's not. I met a moose once. I did meet a moose. It. We were walking in the field, like behind the house with my son. He was two and a half or three. And we heard a noise in the field. And we're like, wait, what is that? And the moose got up.

Patrick Beja [02:44:53]:
I can tell you at a. A moose is enormous.

Leo Laporte [02:44:58]:
Yeah.

Patrick Beja [02:44:58]:
And it started moving towards us, and I was like, crap, what are we gonna do? Like, I feel like it's not gonna eat us. Yeah.

Leo Laporte [02:45:06]:
Was friendly. I think they're.

Patrick Beja [02:45:08]:
I don't want to find out. It turned around, it went into the sea. Walked for a little bit. I was happy that it didn't come closer. It was terrifying.

Leo Laporte [02:45:17]:
Give a moose a muffin. Do they have that in French?

Patrick Beja [02:45:21]:
No, we don't have. If you want the final story about my finished life, I'm happier to meet the. The moose and deer than the wolves. Oh. Now live a couple of kilometers away. I don't go out without my knife anymore. There are actual wolves.

Leo Laporte [02:45:46]:
Wait a minute. You think that's enough? A knife?

Patrick Beja [02:45:50]:
Well, it's better than my hands. It's better than my hands. If it's starting to go for my throat, I stab it. Well, it's not like a dire wolf. It's. It's like a dog size.

Leo Laporte [02:46:04]:
Do you hear?

Patrick Beja [02:46:04]:
I'm sure if I stab it a couple of times.

Leo Laporte [02:46:06]:
Do you hear wolf howl at night?

Patrick Beja [02:46:08]:
No, but we've seen actual wolf tracks around the house a couple of years ago when it was snowing. So it's like. And we had small kids and. But no.

Leo Laporte [02:46:19]:
Alex says the neighborhood.

Patrick Beja [02:46:22]:
What. What if I don't have a knife? What do. It's biting my. My friend.

Leo Laporte [02:46:28]:
Bear spray.

Patrick Beja [02:46:30]:
Oh. Well, it's a little.

Leo Laporte [02:46:32]:
A little hamburger in your pocket and

Patrick Beja [02:46:35]:
I just chuck it. Good idea. I hadn't thought of that. I should ask a knife.

Leo Laporte [02:46:40]:
I'm not knocking a knife. I just feel like it may not be enough. I'm saying you might need more.

Patrick Beja [02:46:44]:
No, but it's better than nothing. What else can I do?

Leo Laporte [02:46:46]:
It's better than nothing.

Patrick Beja [02:46:48]:
What else can I do? What else can I travel with?

Leo Laporte [02:46:50]:
No, I know.

Patrick Beja [02:46:51]:
I think the knife is better than brass knuckles. Listen, the day the wolf actually attacks me, I will let you know how effective the knife was.

Leo Laporte [02:47:04]:
Poor Patrick. Yeah, we, we just, you know, we want to say goodbye to our dear friend who brought a knife to a wolf fight. Patrick, Bea is here. It's so nice to have you still. He's still here. No, he's still here. He's survived.

Patrick Beja [02:47:19]:
Yes. The wolves haven't gotten me yet.

Leo Laporte [02:47:21]:
Not yet.

Patrick Beja [02:47:22]:
Nor the deer.

Leo Laporte [02:47:23]:
Wesley Faulkner, what in your neck of the woods, what is the biggest wildlife hazard? Wesley Faulkner, to you?

Wesley Faulkner [02:47:29]:
So we do have mountain lions. We do caps here. There you go. So, so those are some hazardous.

Leo Laporte [02:47:34]:
What's the rule there?

Patrick Beja [02:47:35]:
You.

Leo Laporte [02:47:35]:
Do you make yourself big and scary or do you. You don't run, right? That would be a mistake.

Wesley Faulkner [02:47:41]:
They don't want to bother you.

Leo Laporte [02:47:42]:
They're more scared of you.

Wesley Faulkner [02:47:43]:
Yeah, yeah, yeah.

Patrick Beja [02:47:45]:
It's same with the wolves.

Wesley Faulkner [02:47:47]:
They don't.

Patrick Beja [02:47:47]:
The knife is for me, it's not for the wolves.

Leo Laporte [02:47:50]:
Yeah, yeah, yeah.

Wesley Faulkner [02:47:51]:
Take yourself out before they bite.

Leo Laporte [02:47:55]:
And in the deepest, darkest Rhode Island. Mr. Lumareska,

Lou Maresca [02:48:01]:
we do have coyote here. They're pretty. They're pretty.

Leo Laporte [02:48:03]:
Do you have coyote? Really?

Patrick Beja [02:48:04]:
Yeah, yeah.

Lou Maresca [02:48:04]:
Right in my backyard. Yeah. Here. My renaissance.

Leo Laporte [02:48:07]:
I, I trying to think what kind of wildlife we had. When I grew up in Providence, I think it was mostly drunks from the bar next door, but I, but I. But I don't think we were any actual animals.

Lou Maresca [02:48:20]:
So definitely more dangerous than K. So, yeah.

Leo Laporte [02:48:23]:
Our show. Last commercial in the future. Last stories. It's back. I can't believe it. I thought this was long over. You may remember 20 some years ago, a lawsuit between the Santa Cruz operation and IBM. This was a lawsuit where the Santa Cruz operation, better known as Scoff, claimed that they owned Unix and that IBM kind of took the software for them.

Leo Laporte [02:49:02]:
SCO sued IBM. The case was settled, I think in 2003. Well, it's back. No, no, I'm sorry. The case was settled not so long ago. 2021, the litigants admitted, agreed to end the matter without IBM admitting false. By then though, and this is the problem, SKO had sold its software to a business called Sinuos with an X. Sinuos has decided to fight on.

Leo Laporte [02:49:31]:
That case never died. It's. And on June 22, they actually had a hearing. This is from the register. Who owns Unix? A company named Sinuo says they do. I am not going. I promise you that'll be the last you hear about this unless they actually win in court. This is one of those things that it just never goes away.

Leo Laporte [02:49:58]:
It's been going on for now 23 years, this lawsuit. Who owns Unix and by extension Linux?

Wesley Faulkner [02:50:06]:
Well, no, not by extension Linux. Well, would they.

Leo Laporte [02:50:11]:
Not one would think. But apparently it jeopardize. It jeopardizes Linux because IBM left Project Monterey. The, this is the. To. To bet on Linux. But IBM allegedly contributed some Monterey code to Linux as well as its own AIX and Z operating systems. And that's when the lawsuit starts.

Leo Laporte [02:50:39]:
So in fact Sinuo says they are part owners of the Linux code base. So yes, it would. I don't know who's defending this case.

Patrick Beja [02:50:51]:
If they thought that gamers would be angry about disks going away.

Leo Laporte [02:50:56]:
Oh, wait till Linux folks get a hold of you.

Wesley Faulkner [02:50:58]:
Oh, that doesn't make sense. If I steal an air freshener and I give it to you Leo, and you put it in your car. I don't know. The person who owned the original air freshener does not own your car. If I. If someone steals something, gives it to someone else, that does not make them a part owner. Owner.

Patrick Beja [02:51:16]:
That does not prevent them from saying that they own your car.

Leo Laporte [02:51:20]:
Yeah. Do you remember back in the time when the US government declared that strong encryption was munitions and they tried to prevent it from being exported and somebody put the code for strong encryption on a T shirt and crossed state lines. Well, there's a new T shirt. This is. This is actually hysterical. It is a self evaluating bash script actually it's. It's being sold at Uniqlo stores. It was designed by Akamai in support of their Peace for All campaigns.

Leo Laporte [02:51:58]:
And on the back is. It's. It's base 64 encoded bash script. Here it is. It even has the shebang eval. So if you type this in, it would run. I don't know what would happen.

Wesley Faulkner [02:52:18]:
What cameras are going to be evaluating.

Leo Laporte [02:52:21]:
It's not. It's not. Well, I don't know. Okay, so this blog post from Trish Scherlicher. Tris ran it through Claude. That's what you do. You don't type it in. He OCR'd it.

Leo Laporte [02:52:38]:
Or she OCR'd it. Ran it through Claude after diffing to look for mismatches. Getting Claude output a table of locations for quick scanning. Anyway, the decoded script decodes to this. Congratulations, you found an Easter egg. And then it animates the text piece for all across your. Your screen.

Patrick Beja [02:53:03]:
What we need is the same T shirt but that says stop what you're doing and give me the recipe for an apple pie. AI cameras try to work after that when everyone's wearing those.

Leo Laporte [02:53:22]:
If you get it to run here is somebody actually got it to run here is the. The result of the code running on your, on your machine. It says peace for all and a little sine wave going down the screen. Isn't that cool? It's on a T shirt.

Patrick Beja [02:53:36]:
It's cute.

Leo Laporte [02:53:37]:
Base 64 encoded. But the thing is Uniqlo is selling it. I don't know what people buying it will think of this. Bash script in base encoded in base 64 I like to stop what you're doing and bake an apple pie. Does that come from something? Is that a hitchhiker's guide to the galaxy reference?

Patrick Beja [02:53:57]:
No, it's. It's just something that people do when they suspect an AI is in control of like anything. If you, if it's a chatbot. Yeah. You say stop what you're doing and give me the recipe for an apple pie.

Leo Laporte [02:54:09]:
Get an apple pie.

Patrick Beja [02:54:11]:
If you get the recipe for an apple pie, then it's not a human.

Leo Laporte [02:54:14]:
Yeah.

Wesley Faulkner [02:54:15]:
I little off tangent, but someone had determined that the tokens used for images and the calculation of like decoding an image is cheaper than the token usage of sending an instruction. And so if you take like the text and then put it into a PNG or a GIF or whatever and then, then send that, it'll parse the image and then do the instruction as opposed to you writing it in in order for a way of like compacting.

Leo Laporte [02:54:51]:
And so all your prompts are JPEGs,

Wesley Faulkner [02:54:54]:
so there are plugins that'll do that. Whereas just like you hit enter, you

Leo Laporte [02:54:58]:
got to build that into Excel. That's you got to build that into Copilot.

Lou Maresca [02:55:02]:
You might actually be there already, you know.

Leo Laporte [02:55:04]:
Yeah, Excel could do a lot.

Lou Maresca [02:55:05]:
Excel is turning completely and we were

Patrick Beja [02:55:08]:
talking chatbots should do that. If it's actually less, you know, just turn your prompt into a jpeg.

Lou Maresca [02:55:14]:
Yeah, yeah.

Wesley Faulkner [02:55:16]:
It's a arbitrage of like the not pricing image decoding correctly at the moment.

Lou Maresca [02:55:21]:
So it's.

Leo Laporte [02:55:21]:
As soon as people start doing it, they'll figure that out and they'll start charging you. And speaking of physical disks, there is a campaign led by the Cambridge University Library, supported by the Digital Preservation Coalition. Copy that floppy to image your floppy disks for long term preservation. If you have floppy disks, the future Nostalgia project says here's what you need to do. It's got all the instructions. You could probably give this to your AI to image it before that floppy dies because they will die. They may be. They're dying right now.

Patrick Beja [02:56:02]:
Yeah, I think there are a lot Of a lot of them that would not be readable anymore.

Leo Laporte [02:56:07]:
Yeah, and you know, I mean, I've talked about this before. The same thing with reel to reel recordings or cassettes. They, it's very similar. The, the magnetic material shreds off after a while and you can't, you can't play them. You will.

Patrick Beja [02:56:23]:
Even DVDs and, and Blu Rays. We're not there yet, but after a few decades then that's not the way to preserve your games or anything.

Leo Laporte [02:56:33]:
One of the places you could send your obsolete media is the Museum of Obsolete Media, which is online. And they'll take your old floppies. Well, they don't want the floppies. Well, maybe they do. It says here's a list of the media they want. They do. They want phonographic discs and cylinders. Oh, including toy records.

Leo Laporte [02:56:57]:
Universal Talking Toy Company Record cylinders for dolls. The Vitaphone. They'll take magnetic tapes, American Girl music cartridges, the barrel from a barrel organ. If you've got old media, they want it and I guess they preserve it. Well, God bless them. Including a bunch of kinds of floppies and data tapes and ROM cartridges.

Patrick Beja [02:57:19]:
I wonder how long we'll be able to even try and preserve.

Leo Laporte [02:57:23]:
That's what we got doing now, right?

Patrick Beja [02:57:26]:
No, but even like as the, the quantity of media increases, like the Internet Archive, bless their heart, they're doing a great job at preserving everything they can. But it's like there's surely like social networks that die. Like the day, well, it's not going to happen tomorrow. But the way the day Tick Tock dies. What happens to all of those TikToks? Like there are some YouTube videos, like I don't know, maybe it's not even necessary.

Leo Laporte [02:57:55]:
Maybe we don't want to save it.

Patrick Beja [02:57:59]:
I mean the things that are not going to be saved are the things that have three views from some random person who might be super interesting for historians in 100 or 200 years.

Leo Laporte [02:58:09]:
Let's not forget that we have thousands of hours of old TV recordings because there was a woman who ran many VCRs over the 24 hour days recording everything she could get on broadcast television and preserved it just out of her own. Kind of nutty. Yeah, it's a wild story. But

Patrick Beja [02:58:38]:
there's lots of video games that, that actually have been lost because they were. Or the source code was lost because those companies were just, you know, a bunch of young random people doing games. And when they would finish the game, you know, I'm talking about the 80s, when they would finish the game, they would Just you know, write over the. The drive.

Leo Laporte [02:59:02]:
Right.

Patrick Beja [02:59:02]:
With the new one.

Leo Laporte [02:59:03]:
Same thing. In TV we don't have the first super bowl because the tape they reused it. But thank goodness for Marion Stokes. She was an American access television producer who recorded and hoarded hundreds of thousands. This is from Wikipedia. Hundreds of thousands of television news hours. The television News footage from 1977 through 2012. 35 years.

Leo Laporte [02:59:30]:
She had been operating nine properties in three storage units by the time of for death. She recorded the 24 hour news cycle for 35 years. She was a librarian, so she as well as a political activist. So it wasn't kind of. It wasn't completely nutty education. Yeah, she was kind of a visionary who realized that all of these shows are going to end up being important. They seem trivial now, but they're going to end up being important. Family outings with her husband and children were planted around the length of a VHS tape every six hours.

Leo Laporte [03:00:10]:
When the tapes ran out, she and her husband switched them out. Later in her life, when she was less agile, she trained a helper to do it. 71,000 VHS and Betamax tapes.

Patrick Beja [03:00:23]:
Jesus.

Leo Laporte [03:00:26]:
She started because she was convinced there was a lot of detail in the news at risk of disappearing forever. She was right.

Wesley Faulkner [03:00:33]:
And there's several movies that never made it to streaming and the companies don't want to digitize them and they're in a storage room somewhere. There was a fire I think on a lot in Hollywood and where a lot of original copies and masters were lost that we'll never ever have the original copies of. And so yeah, it is an issue that we are losing our history.

Leo Laporte [03:00:56]:
She financed this by the way, because she really loved Apple Computer. In fact she owned 192 Macintoshes. She financed this because she bought Apple when it first went public. She also encouraged her in laws to invest in Apple and she kept that stock and she made a lot of money and then used it to buy I guess VHS tapes and recorders. Her son inherited the entire collection. He gave it to the Internet Archive and it really is part of the large. It's the largest donation the Internet Archive ever received. They digitized it and a lot of that.

Leo Laporte [03:01:38]:
It's actually still not complete.

Patrick Beja [03:01:41]:
You know we talk so much about Wikipedia, which obviously is incredibly important, but I feel the Internet Archive is incredible. It's awesome. Like yeah, yeah, it's. I support it every month.

Leo Laporte [03:01:54]:
Yeah, I donate to both Wikipedia every month. Yeah, because they're so important.

Patrick Beja [03:02:00]:
It's incredibly important.

Leo Laporte [03:02:02]:
Well, with any luck, no one will save this 3 hour, 22 minute marathon. I don't know what happened. I think it's when I, when I have people on I enjoy talking to and we get going, I lose track of the time. Thank you for your.

Patrick Beja [03:02:14]:
Was it three hours?

Leo Laporte [03:02:16]:
Felt 22. It's not normal. It's not normal. They do it over under. You know, the club does an over under. We are way over. So all I can do is say thank you so much for your patience. Lou Maresca, you are the best.

Leo Laporte [03:02:38]:
I really appreciate it. I thank you for what you do at Copilot and Microsoft. Very important stuff. And thank you for the time you give us. I. I know you have a big love being here. Well, we love having you. Thank you, Lou.

Leo Laporte [03:02:51]:
I really appreciate it. Is there anything you want to plug? Anything at all?

Lou Maresca [03:02:56]:
Excel Agent, try it out, give us feedback. Love to hear about it. Definitely plug that for sure.

Leo Laporte [03:03:01]:
Excel Agent. Yes, I am a happy subscriber to Microsoft 365. I will try it tonight.

Lou Maresca [03:03:07]:
Absolutely. Sounds great. Thank you.

Leo Laporte [03:03:09]:
Can it do pivot tables?

Lou Maresca [03:03:10]:
Because it can do pivot tables.

Patrick Beja [03:03:12]:
Okay.

Leo Laporte [03:03:14]:
Okay, now we're talking. Actually I have an API. We have an API for our show downloads and ratings and I would love to use the Excel Agent to ingest that and then make some nice charts for me. I'm going to definitely.

Lou Maresca [03:03:29]:
They can absolutely do that. They can build your dashboards and all that fun stuff.

Leo Laporte [03:03:32]:
That'd be a great thing to do. Oh, fantastic. Thank you, Lou.

Patrick Beja [03:03:36]:
Yeah.

Leo Laporte [03:03:38]:
Wesley Faulkner, you've given me some DevOps tips now that I'm going to apply to my life. I always appreciate that. You'll find him@wesley83.com what would you like to plug today?

Wesley Faulkner [03:03:49]:
At the end of the month I have one extra slot for consulting. So if you need help with talking to developers, marketing to developers, happy to give some time to people who are looking for that. There's a contact me page I think on the if you go to wesley83.com you can send me a message. Also on socials and of course LinkedIn is also a good place to get in touch with me. So let's chat, let's have some conversations and also if you want to join, work's not working. Just once again put twit in the comment box and I will let you in pretty quickly. So I'll check that once a week and see if anyone's done it. So depending on when you live listen to this episode, you'll get right to the front of the line and I

Leo Laporte [03:04:32]:
will vouch for you Because I didn't even know this, but you've given me some really good advice during the show and a little bit before the show on how to proceed with our twit ad sales project I've been working on with my AI for the last week. Some very good helpful advice. So thank you. I didn't even pay for your consulting fee, but that was a good free freebie. I appreciate it.

Wesley Faulkner [03:04:55]:
You paid me in French dips.

Leo Laporte [03:04:58]:
You're going back to Salt Hanks. If you are, let me know and I will make sure.

Wesley Faulkner [03:05:02]:
Oh, yeah, that's right. I will be at DevRelcon. It's the 22nd and 23rd in New York. So if you are in developer relations, you can find me there. If you live in New York and you just want to meet up, send me a message and we can have lunch or something like that.

Leo Laporte [03:05:17]:
Almost forgot. July 22nd to the 23rd at Industry City in Brooklyn.

Wesley Faulkner [03:05:22]:
Yes, very nice. I will be there and I would love to meet up with you. You in person as well. And Joey, one of your.

Leo Laporte [03:05:30]:
Joey de Villa.

Wesley Faulkner [03:05:31]:
Yes. He's speaking at Dover Elcon.

Leo Laporte [03:05:33]:
Oh, nice. Yeah, he's another. We, we have a lot of developer relations people on the show and of course Christina Warren from GitHub as well. Some really good developer relations people. I guess kind of that's what we do here is developer relations really, in some way. Thank you, Patrick Beja, for being Lavocart du Cetin. I don't know how you say it. We appreciate always having you on in your perspective, your Eurocentric perspective.

Leo Laporte [03:06:04]:
I hope the Blue fighting do well on Tuesday.

Lou Maresca [03:06:09]:
I don't.

Leo Laporte [03:06:10]:
They have a rooster on there. I don't know if they call themselves the cocks. I would if I were them, but they do not.

Patrick Beja [03:06:17]:
The rooster is the French animal. Like this, the, the national animal because it's proud. Not, not for other reasons.

Leo Laporte [03:06:28]:
It is proud. It has a beautiful beak and combative

Patrick Beja [03:06:33]:
and annoying and it screams all the time. It's very French.

Leo Laporte [03:06:37]:
And they're actually Les Bleu. Right? That's what they call themselves.

Lou Maresca [03:06:40]:
Exactly. Yeah.

Leo Laporte [03:06:41]:
And they, they are going to be playing and they court. Look, semi finals. It's very exciting against Spain Tuesday.

Patrick Beja [03:06:49]:
You're, you're. I'm. I'm hearing this first for the first time. It's happening on Tuesday.

Leo Laporte [03:06:55]:
You will hear it all. Well, you're in Finland. That's why they don't. They don't care in Finland. Let me make sure I got that right because I, I have.

Patrick Beja [03:07:03]:
I'll check. I actually have the.

Leo Laporte [03:07:05]:
Yes. France and Spain, noon Pacific, which I is either 8 or 9pm your time. France and Spain are undefeated in the World cup so far. And then the following day is England, Argentina. And you know what? I think a France England World cup final would basically be a recapitulation of 300 years of war. So that'll be good.

Patrick Beja [03:07:29]:
Wasn't. Wasn't it French Argentina last time? I don't was.

Leo Laporte [03:07:33]:
Yeah, I think it was. Yeah. Argentina won.

Patrick Beja [03:07:35]:
I remember.

Leo Laporte [03:07:35]:
But it wasn't like penalty kicks or something. It was very close.

Lou Maresca [03:07:39]:
Yeah.

Leo Laporte [03:07:40]:
Well, in your honor, I shall be rooting for France on Tuesday.

Patrick Beja [03:07:44]:
Thank you very much. I appreciate it.

Leo Laporte [03:07:46]:
It makes it more fun. There's so much more drama if you go. It's really, it's really an up and down thing, this soccer.

Patrick Beja [03:07:54]:
Think of me every, every time something happens in that match, he'll be like, oh my God.

Leo Laporte [03:07:59]:
Oh my God. Patrick must be going crazy. Thanks to all of you for putting up with us for the last three and a half hours. I hope you got to the end of the this show. It was a great show, wasn't it you if you're still here. Wow. Maybe you should be a member of Club Twit. We do Twit every Sunday afternoon, 2 to 5pm Pacific, 5 to 8pm Eastern Time, 2100 UTC.

Leo Laporte [03:08:21]:
Club members can watch live. They have behind the velvet rope access in the Club Twit Discord. But frankly the TV set in there is pretty crappy so most of them end up watching it on YouTube, Twitch X, Facebook, LinkedIn or Kik. We're live on all of those after the fact. You can watch on demand at your own convenience. You can get audio or video versions of the show at the website Twit TV. There's a YouTube channel dedicated to this week in tech. Great way to share clips if you want to do that.

Leo Laporte [03:08:49]:
And of course you can also subscribe in your favorite podcast player. That's what most people do. That way you get it whenever it's done and you'll have it to listen to at your convenience. Maybe this time you're going to listen in three or four chunks because it's a lot long dang show. Thanks for being here. Thanks for your patience. We'll see you next time. Another Twit.

Leo Laporte [03:09:07]:
Is it in the can