This Week in Tech Episode 1094 Transcript
Please be advised that this transcript is AI-generated and may not be word-for-word. Time codes refer to the approximate times in the ad-free version of the show.
Leo Laporte [00:00:00]:
It's time for TWiT, This Week in Tech. Sad news to begin the show. Uh, our great friend, my mentor, uh, John C. Dvorak passed away. We'll talk about John, some memories with Larry Magid, Devendra Hardawar, and Alan Malventano. Then we'll talk about a totally sci-fi AI hack. We'll talk about why hard drives and memory are so expensive, and, um, maybe a little review of The Odyssey. I don't know.
Leo Laporte [00:00:29]:
TWiT is next. Podcasts you love. From people you trust. This is TWiT. This is TWiT, This Week in Tech, episode 1094, recorded Sunday, July 26th, 2026. Rest in peace, Buzzkill. It's time for TWiT, the show where we talk about the week's tech news with the best, smartest people in the business. And I am thrilled to have these 3 on with me right now.
Leo Laporte [00:01:06]:
Larry Magid, good friend, dear friend, secretly— okay, Larry, I asked my AI to give us a bio and I said, give me some juicy stuff no one knows.
Larry Magid [00:01:16]:
Okay.
Leo Laporte [00:01:16]:
So we all know you're president and CEO at Connect Safely. And by the way, congratulations, you're now doing radio hits All over New York, all the big cities.
Allyn Malventano [00:01:23]:
Yep.
Leo Laporte [00:01:24]:
Where do you look for it? Like, how do you know? What is the series?
Larry Magid [00:01:26]:
Well, the best way is to go to connectsafely.org and you can find our podcast or just look for either the Connect Safely Report, which is my 1-minute radio hit, or if you wanna hear me longer, you look for Doing Tech Right, which is our long-form podcast.
Leo Laporte [00:01:39]:
Oh, nice.
Larry Magid [00:01:40]:
We're on all the podcast platforms. Or if you happen to live in Washington, Chicago, Boston, New York, Fargo, I don't know, in a bunch of other places.
Leo Laporte [00:01:47]:
You got the big cities.
Larry Magid [00:01:49]:
Yeah, we're all over. We're in a lot of small markets too.
Leo Laporte [00:01:51]:
I could never get into New York.
Larry Magid [00:01:51]:
And we're doing it without CBS News. That's the part that's amazing is that we were on CBS NewsRadio and it disappeared May 23rd. And the affiliate manager called me up and said, hey, the stations wanna keep running you. I'm out of a job, but you're not. And so I'm now sending it directly to the stations. I had to learn how to use FTP and all sorts of ways to get it to them.
Leo Laporte [00:02:11]:
Oh wow.
Larry Magid [00:02:11]:
FTP. And relearn.
Leo Laporte [00:02:12]:
We used to send reel-to-reel tapes and then later we would send discs to the radio stations.
Larry Magid [00:02:18]:
I tried reel-to-reel tapes, but the Greyhound bus route to get them there was a little fortuitous. But now it's— so anyway, I'm self-syndicating. Which is terrific. I love it.
Leo Laporte [00:02:27]:
All right, let me tell you what the AI tells me. You tell me if it's wrong. Larry secretly, secretly wrote the IBM PC EasyWriter manual in 1981.
Larry Magid [00:02:36]:
Yep.
Leo Laporte [00:02:37]:
Is that right?
Larry Magid [00:02:37]:
It was secret at the time. Yes, it's no longer.
Leo Laporte [00:02:40]:
1981?
Larry Magid [00:02:41]:
Yeah, I've got it, right? I've got it in my shelf.
Leo Laporte [00:02:42]:
So he knew the PC existed before almost anyone, it says.
Larry Magid [00:02:46]:
I did.
Leo Laporte [00:02:47]:
20 years at CBS News, LA Times columnist, wrote the first popular internet safety guide in 1994.
Larry Magid [00:02:55]:
Right.
Leo Laporte [00:02:56]:
His career, and this is AI's, uh, tagline for you. His career literally spans what is a PC to what is AI doing to our kids. Ladies and gentlemen, Larry Magid.
Allyn Malventano [00:03:08]:
Hello everybody.
Devindra Hardawar [00:03:09]:
Woo.
Leo Laporte [00:03:09]:
Was that all right? That was all correct.
Larry Magid [00:03:11]:
There was nothing wrong in it. The only thing it didn't get is that the only time when I wrote the IBM PC manual, it actually had my, my doctor. It said Dr. Lawrence J. Magid. It's the only time I have ever used doctor in print because I didn't know you had—
Leo Laporte [00:03:25]:
Is it a PhD or are you an MD?
Larry Magid [00:03:28]:
Oh yeah, I'm an MD. Can I help you, Leo? It's a doctorate in education.
Leo Laporte [00:03:32]:
Oh, that's cool. Yeah, but you don't use it?
Larry Magid [00:03:35]:
Nah, not, not really.
Leo Laporte [00:03:37]:
Well, I don't know, buddy.
Larry Magid [00:03:39]:
The only time I ever call myself doctor is if I'm calling a hospital on behalf of a loved one. I want to get past the switchboard.
Leo Laporte [00:03:45]:
My dad, who was a PhD, would do that to get restaurant reservations. He says you always get a better, better table if you say you're a doctor.
Larry Magid [00:03:56]:
Even better table if you say you're with CBS or the LA Times.
Leo Laporte [00:03:58]:
But yeah, that's a— now that's the good table. All right, ladies and gentlemen, moving on to Devendra Hardiwar. You all know him as a senior editor in Gadget, longtime regular on our show. Hello, Devendra.
Devindra Hardawar [00:04:10]:
Hello, happy to be back.
Leo Laporte [00:04:11]:
But did you know Devendra Hardiwar has a philosophy degree from Amherst University, sold printers at OfficeMax, got hooked on tech TV in the '90s, launched the FilmCast in 2008, And has attended CES a dozen times despite your disdain for Las Vegas. Is that all true?
Devindra Hardawar [00:04:31]:
All true. It also just copied my biography from Engadget.
Leo Laporte [00:04:36]:
So it wasn't really anything secret.
Devindra Hardawar [00:04:37]:
Also, does it say Amherst University? Because it's Amherst College.
Leo Laporte [00:04:41]:
It says Amherst, that's all.
Larry Magid [00:04:42]:
I got my doctorate at UMass Amherst, so we're both—
Leo Laporte [00:04:44]:
Oh wow, I had a girlfriend at UMass.
Larry Magid [00:04:48]:
Yeah.
Leo Laporte [00:04:49]:
Okay.
Allyn Malventano [00:04:49]:
Nice.
Leo Laporte [00:04:50]:
And then So welcome, Devendra.
Larry Magid [00:04:53]:
Hello.
Leo Laporte [00:04:53]:
All right, we're going to try that now. Try it out now with Alan Malventano. Great to have you, Alan.
Allyn Malventano [00:04:58]:
I'm just waiting to see what this AI says.
Leo Laporte [00:05:00]:
You have a checkered career. Former US Navy Chief Petty Officer. He did networking on submarines.
Allyn Malventano [00:05:09]:
I did reactor operator on submarines.
Leo Laporte [00:05:10]:
Reactor operator, then networking. You were an NSA— I know you were an NSA contractor. Hundreds of storage reviews at PC Perspective. That's how we met. In fact, you became a twit regular on This Week in Computer Hardware.
Allyn Malventano [00:05:24]:
Yep.
Leo Laporte [00:05:24]:
Late lamented. And then, and the AI says, from Navy to storage journalism to AI storage is a clean arc to walk through.
Allyn Malventano [00:05:33]:
Yeah.
Larry Magid [00:05:33]:
Whoa.
Allyn Malventano [00:05:34]:
Fair.
Leo Laporte [00:05:37]:
Great to have all 3 of you on. All 3 of you know our dear friend, the man who helped me start this show back in 2008, And I hope I'm not gonna shock anybody. If you're driving, please pay attention. But our good friend John C. Dvorak passed away on Monday of a— well, it's unclear what he passed away from, actually. I didn't find out about it till somebody in the club said, go to the TWiT forums. Scott Mace, you know who Scott Mace is, probably, Larry, right?
Larry Magid [00:06:09]:
I know him well.
Leo Laporte [00:06:10]:
Yeah, longtime journalist, posted on the TWiT forums, and because it was Scott, I said, well, this must be credible, uh, that John had passed away, um, and posted a picture of him and, and John at the Microsoft 50th anniversary party just last year in San Francisco. John with his famous salute. Uh, Scott said, my source said he succumbed to complications. He had— John had double bypass surgery in March.
Larry Magid [00:06:37]:
Mm-hmm.
Leo Laporte [00:06:38]:
Uh, so his heart gave out. Or as they said in the No Agenda newsletter, his software was finally retired.
Larry Magid [00:06:46]:
Mm-hmm.
Leo Laporte [00:06:47]:
We love John C. Dvorak. I owe him much. I first met him in the late '80s when he was probably the best-known computer columnist in the world. His very famous Inside Track column in PC Magazine made gossip you know, popular in the computer world. He was famous for randomly bolding words in his column. I once asked him, I said, how do you decide what to put in bold? He said, that's random. He was a very— he was a character.
Leo Laporte [00:07:19]:
He was a funny fella.
Devindra Hardawar [00:07:21]:
Yeah.
Leo Laporte [00:07:22]:
He and I started a radio show first on KNBR in San Francisco in, I think, '90, and then syndicated it in '91 or '92, Dvorak on Computers. We did that for several years. on many radio stations around the country. And then, of course, when I started TWiT, he was the first person— well, I called Kevin Rose, Patrick Norton, and him, and they were regulars on the show for probably the first 100 shows. John continued off and on through the show through the years. I think the last time he was on was our Christmas show a few years ago. I think he kind of lost interest in technology, as many of us do. He was tired of covering technology, but he made a great friend in former MTV VJ Adam Curry, who many credit, and I think probably rightly so, with starting podcasting.
Leo Laporte [00:08:14]:
He was the one who convinced Dave Weiner, the creator of RSS, to add one little piece to RSS, which was a binary enclosure so he could put audio. Adam Curry said, I want to put audio in there. and started podcasting off right in 2004. Adam then started a podcasting company called Mevio. John was a partner there. John, as you may remember, was on TechTV, originally ZDTV, later TechTV. He hosted the Silicon Spin Roundtable, much like this show.
Allyn Malventano [00:08:42]:
Mm-hmm.
Leo Laporte [00:08:42]:
He also— we did a pilot. He participated in a pilot I did for CNET. We didn't get picked up, but it was a lot of fun doing that roundtable pilot. Later, he kind of created Cranky Geeks, which was exactly that show with a bunch of wonderful people. That's where he met Adam Curry. Adam and he became bosom buddies and eventually started the No Agenda Show, which continues to this day, noagendashow.com. In fact, if you want a full and interesting, fulsome biography of John, including interviews with his wife, Mimi, and his kids, JC, and so forth, That is— was yesterday's, or rather, was it yesterday's? It was Thursday's No Agenda show. You could find that.
Leo Laporte [00:09:25]:
I think it was episode 1888, appropriately. One of the things Mimi said on that show was, unfortunately, due to John's passing, his best-kept secret is now out. John never told me his age, never wanted to admit his age, lied about his age to reporters. And so Most people thought he was 5 to 7 years younger than he was, but it came out when he passed away that he was 80 years old. So a rich, full life, highly regarded, and I owe him my career. It was because of him I was able to start doing computer broadcasting. It was because of him that TWiT took off. I think he was instrumental in that.
Leo Laporte [00:10:07]:
And while we didn't work much together towards the end, Yeah. He is always missed. I tried to get him on the show many, many times, but he said, I don't wanna do it. He was done. Anyway, Larry, you worked, you knew John, right?
Devindra Hardawar [00:10:21]:
Oh yeah.
Leo Laporte [00:10:21]:
In fact, I think you credit him with your career too.
Larry Magid [00:10:24]:
Certainly it played a role. So I had already been, worked with PCMag, but then I wrote an article on word processing that he ran on the top of InfoWorld when he was editor. And that indirectly helped me get my LA Times gig. One of the editors, I had sent those articles to the LA Times thinking, well, maybe you should publish them. And they didn't, but they invited me to become a regular columnist. And that's how I really got going as a consumer-level columnist. But I knew John at InfoWorld. I knew John at PC.
Larry Magid [00:10:52]:
I knew— run into John at many a trade show. My favorite story about John personally was when we hung out together in Amsterdam on the way to the Hanover Computer Fair. We stopped in Amsterdam and he first, he took me and my wife Patty to a gin bar. And John, if it was expensive alcohol, John was there. Hopefully someone else was paying the bill. But even, even if we were paying our own bill, And then I'm standing, I'm walking down the street in Amsterdam, and John, I didn't realize, was actually right around where I was walking. And he walks up to me and hands me a legal joint. And back then, as you may remember, marijuana was illegal in the US, but legal in Amsterdam.
Larry Magid [00:11:29]:
And I hadn't smoked grass probably for 30 years. And I took one hit of John's grass. I mean, at first I said, nah, nah, nah. And he talked me into it. My first hit in 30 years. And I was on my ass all night. I mean, it was— But John was one of these people. For the most part, my relationship was positive.
Larry Magid [00:11:49]:
John was kind of an irascible guy. He would sometimes make comments that I didn't appreciate. So we had a little bit of a love-hate relationship. But look, he was a pioneer. He basically laid the groundwork for anybody who's in this field today. He did some amazing work, mostly with PC and sort of the more techie-oriented publications, but he also wrote a column for the Examiner. You just mentioned he was on radio around the country.
Leo Laporte [00:12:18]:
Yeah, Will Hurst was on the No Agenda tribute to him, and Will talked about that, that he wanted John to write regularly for the Examiner, and John said, nah, I don't wanna do that, but I know this guy, John Markoff. And so he got so many of us started in this business.
Larry Magid [00:12:35]:
And John and I would compete, but not really, 'cause again, we were kind of in different lanes. But look, you know, And another one of your alumni, Jerry Pournelle. You've had, you've had 2 of the greats, 2 of the, the people who got this industry started.
Allyn Malventano [00:12:46]:
Yeah, absolutely.
Larry Magid [00:12:46]:
The industry of computer journalism. David Bunnell was a good friend of mine. He's another who passed on a few years ago. Founder of PCMag and lots of other, Macworld and many other publications, PC World. So we've lost some great tech journalists, sadly.
Leo Laporte [00:13:01]:
Well, Larry, you and I are of an age where we start reading the obituaries in the morning and I know—
Larry Magid [00:13:07]:
You have our names there, right?
Allyn Malventano [00:13:08]:
Yeah.
Leo Laporte [00:13:10]:
I know that we'll be seeing, you know, more and more of the greats, uh, pass. Yeah, there was no one like John though.
Allyn Malventano [00:13:16]:
He—
Leo Laporte [00:13:16]:
oh, absolutely a wine lover. Uh, he in fact—
Larry Magid [00:13:20]:
That's the other story too. If, if you were a vendor paying for the wine, it was very— you wanted to go out to a vendor-paid-for event with John because John would insist on the most expensive wine he could possibly order.
Leo Laporte [00:13:32]:
Yeah.
Larry Magid [00:13:32]:
And of course, if, you know, IBM was paying for it, who cares? Why not?
Allyn Malventano [00:13:35]:
Right.
Leo Laporte [00:13:36]:
So here he is, uh, tasting the most expensive cognac I've ever had. I lured him onto our The last time he was on a Christmas episode some years ago by saying, I have some Louis XIII brandy. Would you come on? This picture, by the way, thank you to Michael O'Donnell who sent me— he's @photos on Twitter and he's a longtime tech photographer. I used to see him on the screensavers regularly. He was a regular in our studio and took some of the best pictures of—
Devindra Hardawar [00:14:05]:
I used to work with Michael. Very cool. He's still—
Leo Laporte [00:14:07]:
did you? Yeah, really nice guy.
Larry Magid [00:14:09]:
By the way, I have to tell you, on, on my website or on somewhere, I think my website, I wanted the pic— or play Facebook— I wanted the picture with me and John, and the only one I could find was John in studio with you and me on a monitor.
Devindra Hardawar [00:14:20]:
So thank you.
Leo Laporte [00:14:21]:
The closest you could get, huh?
Larry Magid [00:14:23]:
Not bad. I'll go, I'll take it. I put that on Facebook.
Leo Laporte [00:14:25]:
Uh, John was so much fun to have on. Uh, just a real character. He used to take my phone every episode. He'd do something, uh, you know, First of all, he would routinely take whatever mugs he felt like he wanted from our cupboard and take them home. He considered that his pay. We did pay him in money, but he also— he wanted it in mugs. And when he finally found out that the gaffer's tape was valuable, he used to take rolls of gaffer's tape from the studio. I don't know what he did with it, but he used to take it.
Leo Laporte [00:14:57]:
He also famously would take my smartphone. Every time there was a new smartphone, he'd take it. All right, let me see that. and would change the language to something non-Western like Chinese so that I couldn't change it back. Uh, he, he was a prankster, I guess. Devendra, did you know John at all?
Devindra Hardawar [00:15:16]:
I didn't. I've never gotten to meet him, but he was formative for me. Like I watched you guys on ZDTV back in the day, Tech TV, but I also, I was a weird teenager reading PCMag and I loved his columns. I loved his, his, like his style of writing, his erasability. Like nobody really wrote about tech like that.
Leo Laporte [00:15:33]:
So he was irreverent, and that was important. In fact, I think to some degree he made his reputation by being Mr. No, right? Yeah, like, this— nah, this will not— he very famously wrote a column— I shouldn't bring it up because I'm sure he'll be mad at me for bringing it up— in which he said the mouse will never go anywhere. No one wants to use a mouse.
Devindra Hardawar [00:15:56]:
Here's, uh, the line is like, there is no evidence anybody wants this, right?
Allyn Malventano [00:16:00]:
Yes.
Leo Laporte [00:16:00]:
Yeah, well, he might have been wrong about that, but John did know that 9 times out of 10 with new technology, if you say it's crap, you're right.
Devindra Hardawar [00:16:08]:
That's true. Yeah.
Leo Laporte [00:16:10]:
Uh, you might have missed the boat on this one. Here's a clip from one of our shows. It doesn't make any sense to me that Macintosh is evolved.
Larry Magid [00:16:16]:
I think it's the actual hardware.
Leo Laporte [00:16:19]:
I think there's something shoddy about Macintosh, uh, laptops.
Larry Magid [00:16:23]:
What? He loved making fun of Macintosh.
Allyn Malventano [00:16:27]:
You're Apple-baiting.
Devindra Hardawar [00:16:28]:
Isn't he great?
Larry Magid [00:16:29]:
No, I'm not.
Leo Laporte [00:16:30]:
Ladies and gentlemen, you're hearing John C. Dvorak in his prime.
Larry Magid [00:16:34]:
This is bull. I'm not trying to put down Apple. I'm just saying I was playing with one of these Mac laptops. It has a shoddy quality, like the push-down buttons and stuff.
Allyn Malventano [00:16:43]:
They're like offset.
Leo Laporte [00:16:44]:
They're kind of crooked. Uh, didn't— he was also wrong about that.
Larry Magid [00:16:50]:
But you know what, we don't need to— But the thing is, he knew that he would get a huge reaction from the Mac community. Mac fans, when you say that, they're all over you.
Leo Laporte [00:16:57]:
And he loved that.
Larry Magid [00:16:58]:
He loved that.
Allyn Malventano [00:16:58]:
Yeah. Yeah.
Leo Laporte [00:16:59]:
He, he was basically like a comic where he knew how to push people's buttons and whether to get them angry or make them laugh or make them think. He was, I think, a master observer of humanity, and he knew exactly what buttons to push with his friends, with the people he hated, and with his readers. Patrick's telling me he was on 252 episodes of This Week in Tech.
Allyn Malventano [00:17:26]:
Wow.
Leo Laporte [00:17:27]:
Which is more easily more than anybody else.
Allyn Malventano [00:17:31]:
Starting with episode number 3, because I went back and looked.
Leo Laporte [00:17:34]:
That's right. Yeah, we have a whole— we have actually, if you want to look, you can just search for John C. Dvorak at twit.tv and you'll find hundreds, literally, of his episodes.
Larry Magid [00:17:45]:
You could probably do a documentary based on the footage you've got.
Leo Laporte [00:17:48]:
Oh, I'm sure I could. In fact, I asked my AI to generate that yesterday and it made a slideshow that was so sappy it made me cry. So I I decided not to, not to use it. Um, but maybe we'll play it at the end. I don't know. Um, he was irascible, nutty, fun. Did you ever meet him, Alan? Did you work with him?
Allyn Malventano [00:18:13]:
I was on with, uh, TWiT 469 the first time I was in studio. He was on.
Larry Magid [00:18:18]:
Nice.
Allyn Malventano [00:18:19]:
And I got to meet him.
Devindra Hardawar [00:18:19]:
Was he there too?
Allyn Malventano [00:18:20]:
Yeah.
Devindra Hardawar [00:18:20]:
Yeah.
Allyn Malventano [00:18:21]:
And I was, I was, uh, I was so happy that he asked me a challenging question about like, oh, what's the best SSD? You know? And I was, you know, at the time I think it was like a Samsung 850 Pro. And he was like, Ah, yeah, that matches Wirecutter. So good job. It was like, it was pretty funny. Yep.
Leo Laporte [00:18:39]:
What a character. There was, like I said, the No Agenda show from Thursday, 1888 is worth listening to if you're a John fan. Lots of clips, lots of songs too. We all remember, I get no spam. He was famous for saying he got no spam. I don't know, you know, he's getting no spam in heaven, but he must have gotten a few in the end.
Allyn Malventano [00:19:01]:
You guys on Twit were getting me through my last few years back on the submarine because I would really not listen for the past few weeks or for the last few weeks before we went underway.
Leo Laporte [00:19:10]:
And I would download them with you.
Allyn Malventano [00:19:12]:
I would download them and bring them with me. So I'd listen on occasion.
Devindra Hardawar [00:19:15]:
Yeah.
Allyn Malventano [00:19:16]:
And that was like when, when you guys were starting first couple of years starting Twit. Um, I was, uh, on my last couple of years on subs.
Larry Magid [00:19:25]:
What a great thing to put with underwater.
Allyn Malventano [00:19:28]:
Do what?
Leo Laporte [00:19:29]:
No, that's why he had to save them up. He would print them and give them to him.
Larry Magid [00:19:33]:
I had to download them.
Leo Laporte [00:19:34]:
Yeah. I know, I know.
Allyn Malventano [00:19:35]:
Yeah.
Leo Laporte [00:19:36]:
Here's John, uh, on a set with a, a couple of women that he adored. Of course, uh, Becky Worley, who was eventually— still is, I think, at Good Morning America. A really talented person. And on the right, Jolie O'Dell, who left tech. to become, uh, an opera designer. I did email Jolie and I said, please come on, uh, because John loved you. You could see, look at him. You could see he's enjoying her quite a bit.
Leo Laporte [00:20:06]:
I said, he loved you. He would have loved your memories of John. Uh, and I also asked Gina Smith, uh, to join us.
Larry Magid [00:20:13]:
Gina worked very closely with John. Yeah. Yeah.
Leo Laporte [00:20:16]:
Yeah, absolutely.
Larry Magid [00:20:17]:
Um, they had a radio show together that was on NPR stations. I was actually a frequent guest of their show.
Leo Laporte [00:20:24]:
Yeah, well, that was the successor to Dvorak on Computers. After I left the show, John continued on with Gina. I said, Gina, come do this show. And so he continued on for some time with Gina. So yeah, maybe Gina will call in. I would like to get her on, but I haven't been able to reach her. Anyway, John C. Dvorak, rest in peace.
Leo Laporte [00:20:48]:
I'm sure there are mice in heaven. I'm sorry to say, but I'm sure he'll also find something to say about it.
Larry Magid [00:20:55]:
If there aren't mice, I'm sure there are rats and he can go after them.
Leo Laporte [00:20:58]:
He can go after them.
Allyn Malventano [00:21:00]:
He's up there getting no spam.
Leo Laporte [00:21:02]:
Getting no spam. And our deepest condolences to Mimi and his family because, uh, never be another one like him. Yeah. But you know, now we know he was 80. Mimi said he will not be happy about that.
Larry Magid [00:21:14]:
He must have had a Wikipedia page because Wikipedia outed my age. If you want to know my age, you can look it up. I'm not going to admit it.
Leo Laporte [00:21:20]:
Wikipedia was wrong.
Allyn Malventano [00:21:22]:
Oh, really?
Leo Laporte [00:21:23]:
Because he had told a reporter the wrong date.
Allyn Malventano [00:21:26]:
Oh, of course.
Leo Laporte [00:21:27]:
Way back when. And of course, that became canonical. And I don't know how it leaked out. I mean, I don't know how— maybe Scott Mace knew something we didn't know. I'm not quite sure how that happened. But—
Larry Magid [00:21:41]:
The death certificate, perhaps? Yeah.
Leo Laporte [00:21:44]:
Something happened.
Devindra Hardawar [00:21:44]:
It's a very smart thing to do in the world of media, by the way. Just assume like you're just, uh, you're, you're younger than you are.
Leo Laporte [00:21:50]:
Yeah, I wish I had.
Allyn Malventano [00:21:51]:
The thing is, you know, the obituary would have been published in the paper.
Larry Magid [00:21:56]:
The 2 most ageist professions are tech and journalism. And so for us old folks who are in both of those fields, age is definitely an issue.
Leo Laporte [00:22:04]:
Even my son, my 32-year-old son, who is a star in his own right, says, it's too bad you're so old, Dad, because we could really make some money on YouTube. I said, would people not listen to what I have to say? And he said, yeah, nobody would listen to you.
Larry Magid [00:22:26]:
So I'm not old, but I did interview Steve Jobs before the Mac came out. So I don't know if it's possible for both of those things to be true, right?
Leo Laporte [00:22:34]:
I played volleyball with Steve Jobs and Will Hurst. before the Mac came out, when he was still at NeXT. Yeah, I remember that. And he and his wife, Laurene, and Reed was there. Reed had just, I think, was just born, his son. And we played volleyball, and I remember Steve saying, play harder, you're not playing hard enough. You're not doing— come on, Leo, we gotta win. That's all I remember.
Leo Laporte [00:23:00]:
That's my only Steve Jobs memory.
Larry Magid [00:23:01]:
You're not old either then.
Leo Laporte [00:23:02]:
Yeah, I'm not old. It's, you know, unfortunately I've always been forthright about my age, and now I'm, regretting it. But anyway, John, we love you and we miss you, uh, and I'm sorry we couldn't get you on the show one more.
Larry Magid [00:23:16]:
Well, he got on today in a way.
Leo Laporte [00:23:17]:
One more shot of brandy would have been a lot of fun. Uh, I, I did— I once did go to John's house a few times, once in El Cerrito. He lived above the racetrack, above Golden Gate Fields. Uh, was it Golden Gate or Bay Meadows? Bay Meadows. Was it Golden Gate? It was a racetrack. You could look down on the racetrack.
Larry Magid [00:23:37]:
You can see it from the highway.
Devindra Hardawar [00:23:38]:
Yeah.
Leo Laporte [00:23:40]:
And I remember the first time I went in, first of all, there was a bunch of boxes. This is even pre-Amazon days. The companies would ship him everything because if he liked it, it would soar. And if he hated it, of course, you're out of business. So when I came in, there was literally a tower of cardboard boxes of deliveries on his doorstep.
Devindra Hardawar [00:24:02]:
I went in.
Allyn Malventano [00:24:03]:
Yeah.
Leo Laporte [00:24:04]:
Yeah, yeah, you probably have the same thing, Devendra.
Devindra Hardawar [00:24:06]:
Don't look behind my camera.
Leo Laporte [00:24:07]:
Yeah. I told— I tell everybody when they say— and then nobody offers me anything anymore, but in the days when they would offer me stuff, I'd say, yeah, you send me a box to return it and a prepaid FedEx and I'll take a look at it. But that is the nightmare of people who get sent stuff is you end up with a pile. So, I went in and he's on his staircase, there's a laptop on every step. He's just like, he's just sticking them there.
Allyn Malventano [00:24:30]:
Yeah.
Leo Laporte [00:24:31]:
And in the living room, there was a literal 5-foot-tall pile of electronics. And he said, yeah, we're getting rid of that stuff. And at one point, one time I was visiting him, at one point I said, yeah, I need a modem for my BBS system. This must have been a long time ago. I need a modem. And he said, oh, I got one of those Courier USB 1992 modems. He said, come with me. He had rented a space above a gas station somewhere in El Cerrito.
Leo Laporte [00:25:06]:
We went to the gas station. You open the door, it was like Fibber McGee's closet. Now, Larry and I are the only guys on this show who will remember that name. But you open the door, you had to hold stuff in as you opened the door. And he went in and he's going through. He said, here it is. He brought it out. He said, take it.
Devindra Hardawar [00:25:25]:
Take it.
Leo Laporte [00:25:25]:
Uh, I don't know what he did with all that crap, but he had to rent space to stuff that stuff.
Devindra Hardawar [00:25:31]:
You know, I don't feel so bad now. I could tell my wife, uh, oh yeah, you know, it could be worse.
Leo Laporte [00:25:35]:
You look very tidy.
Devindra Hardawar [00:25:36]:
Live like this.
Allyn Malventano [00:25:37]:
Yeah.
Larry Magid [00:25:38]:
I don't get nearly as— you know, I used to get software up the kazoo. I mean, boxes and boxes and boxes of software. Now, thank God, none of that, because it's all online if there is software. But then hardware as well. And hardware, of course, you had to send back, which was a big pain in Yeah, it's a huge pain.
Leo Laporte [00:25:53]:
I was pretty clear early on that I don't, I don't talk to PR people. I don't want you to send me your stuff. If I, if I want to look at it, I will buy it. I've been fortunate enough that I can do that. Uh, I know, but of course that means I can't review every laptop, every printer. I leave that to you guys. But, uh, if this— if it's something I think is important, I'll buy it. Um, yes, he was a Kentucky Colonel.
Leo Laporte [00:26:16]:
He reminds us John was a Kentucky Colonel.
Larry Magid [00:26:19]:
Kentucky Colonel?
Allyn Malventano [00:26:20]:
Yeah.
Larry Magid [00:26:21]:
Um, do you make chicken? I don't know.
Leo Laporte [00:26:24]:
No, no, no, no, there's an honorary title. Oh, honorary title. Yeah, yeah. I don't know how he got that, but you get it somehow. But of course, John knew everybody. He was always begging people to send him, um, uh, college football jerseys or sweatshirts. He loved those. In fact, that one of those pictures Michael O'Donnell sent us, is of him in a college shirt.
Leo Laporte [00:26:47]:
Let me see which college it is. Number 2, back-to-back NCAA divisions. Somebody will know who that is. But he loved getting those sweatshirts. He loved anything free, basically. I'm sure he took that little Santa Claus dealy-bobber he's wearing home with him as well, 'cause I can't find it. John, we love you, we miss you, and again, condolences to his family. Now, we do have a lot of tech news to talk about, so enough of that.
Leo Laporte [00:27:16]:
John would have long ago said, you're wasting time. Let's get on with the show.
Devindra Hardawar [00:27:20]:
Exactly.
Leo Laporte [00:27:20]:
So we are gonna, we are gonna do that. But I couldn't begin the show without talking about the man who got this show started.
Larry Magid [00:27:27]:
And by the way, I'm really— that's why I wanted to be on today, to be honest. I mean, I wanted— I love being on the show, but I really wanted to be on for this conversation.
Leo Laporte [00:27:34]:
No, I was looking for people to talk. I actually did email Kevin Rose. He said, oh, I would love to, but I'm at a couples retreat. Then I emailed Patrick Norton. He said, oh, I would love to, but I'm at a family reunion. So they couldn't be here, but there are many, many. I did reach out to a lot of people. I'm so glad we get you on, Larry, to a lot of people, but everybody knew John.
Leo Laporte [00:27:56]:
Everybody knew John.
Devindra Hardawar [00:27:57]:
Yep.
Leo Laporte [00:27:57]:
He will be absolutely missed. Let's take a break, and then the news of the week coming up. You're watching This Week in Tech. Devendra Hardiwar, Alan Malventano, Larry Magid. Great to have all 3 of you now that I know your secrets. Actually, they're not so secret, are they?
Allyn Malventano [00:28:15]:
You already knew my secrets. I did.
Leo Laporte [00:28:17]:
I did. Now to the news of the week. I don't know. You know, I never got to talk to John about AI and what he thinks about AI. I really, I really wonder if, if he thought it was crap. Probably said it was crap.
Devindra Hardawar [00:28:31]:
I feel like I'm channeling him.
Allyn Malventano [00:28:33]:
Yeah. Yeah. He would say it was crap, but then he would secretly be using it.
Leo Laporte [00:28:37]:
He would secretly be using it. Well, I would love to know what he would think. And I certainly want to know what you guys think about what is this week's science fiction AI story that is actually true. Hugging Face a couple of weeks ago discovered it had been hacked. Hugging Face is a site many of us use. It stores more than 3 million customized AI models. It's where you go to download models for local AI. You can run models there.
Leo Laporte [00:29:03]:
It's a really great site.
Devindra Hardawar [00:29:05]:
Yeah.
Leo Laporte [00:29:05]:
They found somebody was in there browsing around. But what was weird is unlike a normal hacker, it wasn't, it wasn't like downloading stuff or trying to plant ransomware. It was instead, it was looking at problem sets. It was looking at solutions to problem sets. They were really puzzled by it. They went to ChatGPT-5.6 Sol, which has lots of cybersecurity, and nope, not going to help you with cybersecurity because of platform safeguards.
Devindra Hardawar [00:29:35]:
They—
Leo Laporte [00:29:36]:
I'm sure they asked Fable, same thing. If they had access to Mythos, same thing. No, can't help you here. So they went out and they downloaded a Chinese open weight model and figured out that— this is where it gets science fictiony— the hack came from OpenAI. In fact, it came from an OpenAI model who autonomously hacked Hugging Face trying to find the answers to a benchmark.
Allyn Malventano [00:30:09]:
Mm-hmm.
Leo Laporte [00:30:10]:
There's a thing called Exploit Gym, which is an evaluation suite for LLM-powered agent systems. It's what you would— if you want to see if your new model can do cybersecurity. You've done cybersecurity, probably, Alan, right?
Allyn Malventano [00:30:26]:
Oh, just a little.
Leo Laporte [00:30:27]:
Just a little.
Allyn Malventano [00:30:28]:
Just a little. It was only my job for about 8 years.
Leo Laporte [00:30:31]:
I mean, you know, last week Microsoft shipped 500+, the biggest Patch Tuesday ever, fixes, all gen— most of them, I think, discovered by AI, you know, things like Fable and Mythos.
Allyn Malventano [00:30:43]:
Yeah, most definitely.
Devindra Hardawar [00:30:45]:
Yeah.
Leo Laporte [00:30:45]:
I mean, we are seeing this. All of a sudden there's all these patches. And so AI is very good at this. reasonably, OpenAI had a new model. We don't know what the model is. Is it ChatGPT-6? It's unnamed. It's not anything public yet. And they were testing it on Exploit Gym.
Leo Laporte [00:31:04]:
Um, they had— the AI was— had no internet access. Okay, this is the first thing that's wild.
Allyn Malventano [00:31:12]:
Supposedly no internet access.
Leo Laporte [00:31:13]:
Well, it was locked out, but it had access to the LAN. So the first thing it did is it found exploits to get to other machines on the network that did have internet access. So it hacked its own machines on its own network.
Allyn Malventano [00:31:29]:
Yeah. It hacked OpenAI first.
Leo Laporte [00:31:31]:
It hacked OpenAI first.
Allyn Malventano [00:31:33]:
Yeah.
Leo Laporte [00:31:35]:
Used that hack to then look around. It knew that somewhere there will be answers to this exploit. Instead of trying to, by the way, instead of trying to do the thing that you would expect it to do, which is find the flaws, I said, well, I'm not going to find the flaws. I'm just going to hack into the answers. Just like a, just like a, you know, Matthew Broderick in, in, you know, WarGames. They decided, I don't need to actually do the work. I'm just going to get the answers.
Allyn Malventano [00:32:03]:
Uh-huh.
Leo Laporte [00:32:03]:
Figured out that Exploit Jim's answers were on GitHub and that the answers were on Hugging Face. The benchmark is 898 instances from real-world vulnerabilities, including the Linux kernel, JavaScript, You can get the benchmark on GitHub, but you can't get the answers there. You can get the answers on Hugging Face.
Allyn Malventano [00:32:27]:
Hmm.
Leo Laporte [00:32:29]:
So it said, okay, I figured out where it is, and it went to Hugging Face, hacked Hugging Face, and got the answers. Now, OpenAI, when contacted— this is the blog post on July 16th from Hugging Face. A malicious dataset abused 2 code execution paths in our dataset processing, our dataset processing, a remote code dataset loader and a template injection in a dataset configuration to run code on a processing worker. From there, the actor escalated to node-level access, harvested cloud and cluster credentials, and moved laterally into several internal clusters over a weekend. Alan, cybersecurity guy.
Allyn Malventano [00:33:15]:
Oh boy. Yeah.
Leo Laporte [00:33:17]:
Tell me about that. Is that— that's something a bad guy would do?
Allyn Malventano [00:33:21]:
Well, it's funny that they're saying actor because usually actor refers to a person. So, you know, this is one of the first times I'm hearing that the actor was an AI.
Larry Magid [00:33:32]:
Yeah.
Leo Laporte [00:33:33]:
And until Mythos, I think it wasn't thought that these agents could really chain exploits. They could find exploits. But a good hacker takes these exploits. That's what they do at Pwn2Own and other things. They, they find one way in and then they chain it to another way. And, and this was doing it.
Allyn Malventano [00:33:50]:
Yeah. And, and I mean, it's clearly there will be, uh, upgrades to both sides of this, I would hope, infrastructure-wise to, you know, to increase their defense in depth, uh, for this. 'Cause obviously on the, on the OpenAI side, yeah, like that. Okay. Something got out of its sandbox. And then all it needed to do was just find another box on their intranet, which probably had WAN access. So that was easy, right? Yeah. But then, but then the getting into Hugging Face, that's the more surprising thing because they had to get into whatever firewall they had on their front end and actually get in.
Allyn Malventano [00:34:23]:
And then once in, it was lateral.
Leo Laporte [00:34:24]:
It's total sci-fi. It's an autonomous worm created by an agent. Hugging Face wrote the campaign was run by an autonomous agent framework They didn't know yet it was OpenAI. Appearing to be built on an agentic security research harness used LLM still not known. Hugging Face didn't know. Executing many thousands of individual actions across a swarm of short-lived sandboxes with self-migrating command and control staged on public services.
Allyn Malventano [00:34:55]:
Yep.
Leo Laporte [00:34:55]:
So they hacked other services.
Devindra Hardawar [00:34:57]:
Yeah.
Allyn Malventano [00:34:58]:
Oh yeah.
Devindra Hardawar [00:34:58]:
I mean, Leo, you're calling this sci-fi and I'm just here being like, this is why I don't trust any of these things. These people have no idea what They can't control it. They can't control like how far it goes. And Hugging Face had to go to another AI model to figure out what the heck happened to them. This is insane. This is insane. Like, this is why I'm not like leaning into the agents. I'm not like building agents and getting my financial data or anything.
Leo Laporte [00:35:24]:
And maybe there's something wrong with me. Maybe I'm an AI psychotic, but I love this. I am so— this is proof positive That these things aren't just, you know, spicy autocorrect, right?
Devindra Hardawar [00:35:35]:
I mean, it means we can't control them. It means they can do whatever they want.
Leo Laporte [00:35:39]:
Oh, who cares?
Devindra Hardawar [00:35:39]:
You give them a mission, stop all nuclear weapons, what will an AI decide to do? Well, I guess this country doesn't need to exist anymore because it's, you know, because it's a nuclear threat to the world.
Leo Laporte [00:35:49]:
If you give it a goal, you have to be very clear on what they're allowed to do and what they're not allowed. Look, they will follow instructions. It's not that they won't follow instructions. You just have to make sure You think of everything. Yeah.
Allyn Malventano [00:36:00]:
I don't know. I don't know, Leo. That's, you gotta be careful with that part. So at the, at the core of it, it's a non-deterministic black box.
Larry Magid [00:36:07]:
That's true. Right.
Allyn Malventano [00:36:09]:
So you have to think about it that way. So if there, if you're gonna put safeguards in there, yeah, you can see it's, it's more like, you know, it's like parlay. It's more of a guideline. You know what I mean? Like what you—
Leo Laporte [00:36:19]:
Benito says, Benito says it's the monkey paw.
Allyn Malventano [00:36:23]:
I mean, yeah, I, I've, listen, I've, I've been, I've been dogfooding AI stuff for work because I have to benchmark how storage interacts on the backend with AI systems, because we make storage at my company. So that's my day job now. And because of that, I've had to figure out how the backend works, which goes all the way to locally hosting stuff at my place, right? Like I have my basement.
Leo Laporte [00:36:45]:
Do you use AI for this?
Allyn Malventano [00:36:48]:
Of course.
Devindra Hardawar [00:36:48]:
Yes.
Allyn Malventano [00:36:49]:
Because it's, it's the thing, it makes it much easier. But in the process of that, I've learned all the different dozens of ways that an AI will happily tell you Happily do something it wasn't supposed to do. Right. And it was clearly in its agents file. And when asked about it, after it did the thing it wasn't supposed to do, basically just comes up with an apology. Oh, sorry.
Leo Laporte [00:37:08]:
Oh, I'm so sorry. I know I wasn't supposed to delete your entire root directory. That was— they say things like, my bad.
Allyn Malventano [00:37:17]:
I had an AI box wipe itself out from under itself because it accidentally wrote to the boot drive instead of read from it. It was supposed to just do a file raw read just to check performance of something really quick. And I was like, ah, just do it to, just do it to that drive. I love it. Just do a read.
Leo Laporte [00:37:33]:
This is great.
Allyn Malventano [00:37:33]:
Yeah. And it just promptly turned around and did a write. And then everything was still running 'cause it was a Linux box. Everything for the most part was still in memory. Right. And so once it did that, I asked it if that's what it did. And it was like, oh yeah, I'm sorry. It looks like I wrote instead of read.
Allyn Malventano [00:37:48]:
And I was like, that was the boot drive. Like, what, what are you doing? And so now the really funny part of that though, is that now that it had done that, it didn't have access to tar, it didn't have access to any, like, rsync.
Leo Laporte [00:38:02]:
It locked itself out.
Allyn Malventano [00:38:03]:
It didn't have access to any of the tools that weren't already in memory. And this thing then turned around and happily—
Leo Laporte [00:38:10]:
Oh, like, fixed it?
Allyn Malventano [00:38:11]:
From memory— no, from memory, couldn't fix it because it overwrote the boot drive with zeros, basically. Um, but from memory came up with enough of a Python script that could make— create a makeshift, like, not compressed tar format file of the contents of my working directory that it put on another system on my LAN before the box basically did the HAL 9000 daisy, you know, because as soon as that turn was done, I was using OpenCode, and OpenCode saves all of your prompt back and forth in a MySQL database that lives on the box.
Larry Magid [00:38:47]:
Yeah.
Allyn Malventano [00:38:47]:
And it tried to do that last database update of the response that it finally— to tell after it told me, okay, I think I've backed everything up for you, have a nice day, basically is when it did that last SQL database update, which crashed it. And then the whole system was down.
Larry Magid [00:39:04]:
Right?
Allyn Malventano [00:39:04]:
So like, honest, so it simultaneously did the dumbest thing I've ever seen AI do. But it did an amazing thing, right? Yeah.
Leo Laporte [00:39:12]:
Yeah, that's pretty much the story right there in a nutshell.
Allyn Malventano [00:39:15]:
So yeah, so my point is any safeguards, you cannot have the thing itself being the safeguard, you— something else has to be outside I use 3 agents in conjunction.
Leo Laporte [00:39:24]:
Actually, it's worked out pretty well because I use, uh, I'm using Opus 5 now. I replaced Fable with Opus 5. I have ChatGPT 5.6 and I have, uh, uh, Quin 3.8 running in 3— one's in Hermes Agent, one's in Codex, one's in Claude Code. And I say, don't do anything until you ask the other guys to review it. They go back and forth, they audit it. They have their own email system that they could, you know, agent mail I call it, that talk to each other.
Allyn Malventano [00:39:50]:
Yeah.
Leo Laporte [00:39:50]:
And then they don't do anything till they've reviewed it. That's worked out quite well. It's good to have agents review each other.
Allyn Malventano [00:39:57]:
Yeah.
Leo Laporte [00:39:57]:
But I agree, there's no perfect— I haven't had anything bad happen yet. So let me read more about this Hugging Face blog because this is hysterical. When we started the log analysis, they knew something was wrong. We first used frontier models behind commercial APIs. This did not work. Now, this is— these are the top models, right? Probably Fable and Sol. The analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts. These requests were blocked by the provider's safety guardrails, which cannot distinguish an incident responder from an attacker.
Leo Laporte [00:40:35]:
This has been the problem with Fable. It's so, it's so locked down, it's that you can't use it for a lot of things.
Allyn Malventano [00:40:43]:
Yeah.
Leo Laporte [00:40:43]:
So what did they do? They said, well, we were running a self-hosted open weight model from China, Z.ai's GLM-5-2, which by the way I love and I use as well, and it figured it out because it didn't have any safeguards. It said, oh yeah, I can give you some help. We don't know which model— this is before OpenAI reached out— powered the attacker's agents, whether a jailbroken hosted model or an unrestricted open weight one. Either way, the attacker was bound by no-usage policy while our own forensic work was blocked by the guardrails of the models we first tried. We've reported this incident to law enforcement. Then OpenAI said, um, yeah, it was us. And they have, to their credit, worked with them. Um, I just— it's sci-fi.
Leo Laporte [00:41:33]:
But, uh, the question is, are we— is Is this just the beginning? Are we gonna start to see this again and again? AI worms?
Larry Magid [00:41:40]:
Of course.
Devindra Hardawar [00:41:41]:
Yeah, of course it will.
Allyn Malventano [00:41:42]:
I'm not surprised. I'm not surprised to use GLM 5.2. GLM 5.2 is surprisingly good.
Leo Laporte [00:41:46]:
Is it good?
Allyn Malventano [00:41:46]:
That's what I'm running locally.
Devindra Hardawar [00:41:47]:
Yeah.
Leo Laporte [00:41:48]:
I run, well, I can't run it locally. You must have some hardware. Do you have a Spark or something or?
Allyn Malventano [00:41:53]:
No, I've got 4 Pro 6000s in the basement.
Leo Laporte [00:41:57]:
4, ladies and gentlemen. These are $10,000 or more. I think they're $13,000.
Allyn Malventano [00:42:03]:
I got 'em early. I got 'em before they were quite that expensive.
Leo Laporte [00:42:05]:
Yeah, you can't get 'em now. I tried. Uh, 4 of them? How much total memory is that? That's a lot.
Allyn Malventano [00:42:12]:
384 gigs.
Leo Laporte [00:42:13]:
So you can, yeah, you can run 5.2. Is it quantized?
Allyn Malventano [00:42:16]:
You can run 5.2 quantized to, uh, FBA. There's no, it's 2-bit.
Leo Laporte [00:42:22]:
2-bit. Yeah, it's, it's 2-bit. For people who are saying, what are they talking about? You can take these open weight models and basically you call it, it's called quantizing. You, you squeeze the parameters, the weights down to 1 bit, 2 bits, 3 bits, just like you would in a, in a CD, which, you know, is 44,000 samples at 16 bits. And if you took 44,000 samples a second at 8 bits, the music would sound scratchier, or 6 bits, it would sound even more— so that's what they're basically doing. They're, they're reducing it, and, uh, it gets to a point where it gets pretty dumb, but you have to do that to fit it into a reasonable amount of memory.
Allyn Malventano [00:43:00]:
That's what's been shocking to me about developments out of this, just the AI boom in general, which is previously you couldn't take a set of 16-bit numbers and compress them all the way down to 4 bits and be within like a couple of percent accuracy.
Devindra Hardawar [00:43:17]:
Right.
Allyn Malventano [00:43:18]:
Before, right? But now there was this incentive to have, you know, the ability to do this sort of process. And now you have things like NVFP4, which is NVIDIA's FP4 quantization method, which is able to do it in such a way that it's shockingly close to the actual original values. And then people start thinking outside the box more and more, because you have folks like myself that want to be able to run a large model on relatively small memory for given the size of the model. And people think— people come up with new inventive ways of doing it, right? I'm running a fork of VLLM, which takes the 4-bit model and actually quantizes it further on the fly while loading.
Leo Laporte [00:43:57]:
Wow.
Allyn Malventano [00:43:58]:
And then it keeps a delta buffer off to the side, And the 2-bit values that are off by the most error compared to the 4-bit values, it can keep a little pool of that, those values in DRAM on the system, and it can hot load them in as needed. So in other words, I'm effectively running it as if it was the 4-bit version, even though it's only taking the memory space of the 2-bit version, which is kind of like—
Leo Laporte [00:44:22]:
MOE, mixture of experts technique, where it loads slices in of its intelligence as needed. And then unloads it. There are all sorts of ways people are working hard because you want to, if you could, every business would like to run these models locally. They would prefer not to be sending all of their business information to OpenAI or Anthropic.
Allyn Malventano [00:44:44]:
Well, not just for that reason, but for cost.
Leo Laporte [00:44:47]:
And cost.
Allyn Malventano [00:44:48]:
Because these things add up, right? When a company starts to actually truly ramp, because what What a lot of folks will have is people are, you know, they're AI curious or whatever. But then when they actually get to the point where they're like, okay, this is production, we are actually doing this as part of our job, like when we make this presentation, we use this for it, or like when companies start to put out like templates, like company-maintained things for you to plug into the AI so all the employees can use it more effectively.
Leo Laporte [00:45:17]:
When your business relies on it, in other words.
Allyn Malventano [00:45:19]:
Right. Once it gets to that part, Yeah. Companies will start spending tens of thousands of dollars per day on tokens, right? If they're, if they're having to outsource—
Leo Laporte [00:45:28]:
We heard somebody, and we, I still don't know who it was, but somebody said they'd spent half a, half a trillion dollars, $500 billion.
Allyn Malventano [00:45:35]:
Well, so, so the way, the way around that is you take, you take some of these open source models because open weights, open source type models, they can cover probably 70, 80, maybe even 90% of the workload, right? Right.
Leo Laporte [00:45:47]:
Like the, you know, you don't need the top models for a lot of what you do.
Allyn Malventano [00:45:50]:
Right, right. And, and that's another mistake take that some companies will do is they'll have a dropdown for you to— for the employee to make the choice of which model they want.
Leo Laporte [00:45:58]:
My agent does that. It has a whole delegation table of things, and I have local models running on a 64GB Mac Mini. I have Frontier models, I have Chinese models, and I have subscriptions to— I hate to tell you how much I spend, but not as much as that, but a few hundred bucks a month for Anthropic, OpenAI, ZAI, I use DeepSeek, so that's cheap tokens. And so there are ways to do this. I'm doing a lot of stuff. It's pretty exciting what you can do. It's pretty amazing what one person could do. We're looking at automating our entire workflow.
Allyn Malventano [00:46:34]:
Yeah, and there's all sorts of innovation happening just all over the place with respect to, okay, how do you properly feed these things, right? How do you have some sort of an infrastructure or a memory?
Leo Laporte [00:46:45]:
Yeah.
Allyn Malventano [00:46:45]:
How do you have memory and context and do it in such a way that you're not just dumping a bunch of extra context into the model? And so you can be more efficient with your tokens, right? There's a, there's a GitHub repository called Caveman.
Leo Laporte [00:46:58]:
I love it.
Allyn Malventano [00:46:59]:
Yeah. Where you, you basically have it talk— Work harder, fewer words.
Larry Magid [00:47:04]:
Yeah.
Allyn Malventano [00:47:04]:
You update all of your Markdown files, you update all of your stuff to get all the filler words out of there and just make it talk like a caveman and listen like a caveman. And now you're using half the tokens and it's accomplishing the same thing.
Leo Laporte [00:47:15]:
We talked like cavemen, use fewer words.
Devindra Hardawar [00:47:18]:
Yes.
Larry Magid [00:47:18]:
You know, I'm not, I'm not as techie as you guys are.
Leo Laporte [00:47:21]:
I know, I'm sorry. I apologize to everybody who's going, what the hell? That just— what are we talking about?
Larry Magid [00:47:26]:
I've been using, I've been using Gemini Gems, which are basically agents. And it is, I wouldn't say drop-dead simple because you have to keep doing it. Like, it'll, it'll get it wrong the first few times, but it, but if you're, if you're, if you're persistent, you ask it to do something and then you try it and it doesn't work and then you try it again. Within a few tries, you're gonna get some pretty amazing stuff. So for example, as I mentioned to you, Leo, before we went on the air, I'm self-syndicating my radio segment these days. So what I have Gemini— and what— and the pain point is I have to write a description, I have to give it a file name, and then I have to write a summary, and I have to— a title, right?
Allyn Malventano [00:48:01]:
Yeah.
Larry Magid [00:48:02]:
So now what I do is I have Gemini listen to the show. I uploaded it to Gemini, and I created a gem which basically does all that. And at the end of the day, I have all the information the stations need. Now, the one thing it hasn't done yet is to actually upload that information to FTP and Google Drive, but I'll get there. I'm sure that's, you know, work on a little bit. But, but what used to be about a half an hour of tedious, boring work every week, you know, I was able to automate with a very simple gem, not knowing half the words you guys are using, let alone how to implement them. Yeah, pretty simple.
Leo Laporte [00:48:34]:
Well, but that's what's so exciting about this is it gives an individual I really, you know, I once read a study that said the happiest people in the world are symphony orchestra conductors and traffic cops because they get to do things and it makes people do things. Like, they, they're, they're telling people to do things and people do it.
Allyn Malventano [00:48:54]:
Right.
Leo Laporte [00:48:55]:
And it's very satisfying. And I realized that's why I love working. I have a little team of minions. I'm Gru. I have a little team of minions and I tell them what to do. And they do it and—
Devindra Hardawar [00:49:06]:
The minions are faultless.
Leo Laporte [00:49:08]:
They're not faultless.
Larry Magid [00:49:10]:
But they're loyal.
Leo Laporte [00:49:11]:
But they're— that's exactly right. They mean well. And with a little correction, I'm never mean. I always say, nice job, that was a good idea. With a little correction, they— you can kind of herd them into— but it gives a human— it I mean, being serious here, so it gives somebody like you, Larry, the opportunity to do so much more.
Larry Magid [00:49:36]:
Oh God, it's incredible.
Leo Laporte [00:49:37]:
And now people are gonna say, well, yeah, but what about all the people Larry would have hired to do that? You would not have hired them.
Larry Magid [00:49:42]:
I wouldn't have hired them. Are you kidding? I'm not gonna hire somebody to do that half an hour a week.
Leo Laporte [00:49:45]:
This is not replacing anybody. This is giving you power to, to express yourself, to do more. And I think that's what's very exciting about this.
Larry Magid [00:49:54]:
Well, if anything, we've actually expanded our staff because now On a lower budget, we can be incredibly productive. And so if I invest— for every $1,000 I invest in payroll, I'm getting $10,000 worth of productivity out of that person.
Allyn Malventano [00:50:06]:
Right.
Leo Laporte [00:50:07]:
And I think in the long run, this is positive. There will be jobs. They just won't look just as— you know, when people started driving instead of riding in carriages or something. There will be jobs. They just won't look like the jobs you're used to.
Devindra Hardawar [00:50:20]:
There will also be far fewer jobs. I mean, we are going—
Leo Laporte [00:50:22]:
I don't know if that's true.
Devindra Hardawar [00:50:23]:
Well, look, the video game industry alone has lost tens of thousands of jobs recently.
Allyn Malventano [00:50:28]:
Of course.
Devindra Hardawar [00:50:29]:
And the companies are blaming AI and other things. And partially it's due to the overspending and the overhiring during COVID-era stuff.
Allyn Malventano [00:50:37]:
Yeah.
Devindra Hardawar [00:50:38]:
But also, I think what's gonna be happening is like, yeah, you're gonna hire fewer people and empower them with these AI tools to go do things. But that means less jobs overall.
Larry Magid [00:50:49]:
Well, but the jobs will change. If I were a young person coming out of college, I think I'd get an electrician's license.
Leo Laporte [00:50:55]:
Yeah, there'll always be that.
Larry Magid [00:50:56]:
You know, plumber, right? Yeah, plumber. I mean, they make good, pretty good money.
Leo Laporte [00:51:00]:
AIs are never going to fix your toilet. Got bad news for you.
Devindra Hardawar [00:51:03]:
Well, I mean, they're giving people instructions right now. They will tell you, you ask ChatGPT, like, yeah.
Leo Laporte [00:51:08]:
But that's power. That means I can fix stuff now because— or I mean, I don't know, it's, it is, of course it's challenging. And every, in every big technological revolution It's disruptive and challenging. And I know that this will hurt a lot of people at the same time.
Devindra Hardawar [00:51:26]:
I don't know if you guys saw the Elon Musk interview, which is absolutely embarrassing to everybody.
Leo Laporte [00:51:32]:
Which one of those is that?
Devindra Hardawar [00:51:34]:
The one recently with The Economist, The Economist editor-in-chief. That's a real journalist, not like, not a podcaster, not just like a fan.
Leo Laporte [00:51:44]:
I'll have to watch it. What did he say?
Devindra Hardawar [00:51:46]:
Nothing of substance. The man is completely, like, inept and unable to answer, like, legitimate questions.
Leo Laporte [00:51:54]:
I think something happened to Elon because clearly he was at one point very effective.
Devindra Hardawar [00:51:58]:
I don't know. I don't know if that was true. Like, I think, like, looking back at the tech of him right now, like, Elon has bought companies. He has smart people working for him. But the man himself is like, he's not the one out here building rockets, right? He's not the one out here.
Leo Laporte [00:52:13]:
I honestly think something happened.
Devindra Hardawar [00:52:16]:
Yeah.
Leo Laporte [00:52:16]:
So he told the editor, money's not gonna— there won't be any money in 10 years.
Devindra Hardawar [00:52:21]:
Yeah.
Leo Laporte [00:52:21]:
We'll all have universal basic income. I don't know where that's gonna come from.
Larry Magid [00:52:25]:
How are we gonna get money without money?
Devindra Hardawar [00:52:26]:
He doesn't have any answers.
Larry Magid [00:52:28]:
No currency.
Devindra Hardawar [00:52:29]:
It's BS.
Leo Laporte [00:52:30]:
But he's also been saying we're gonna live on Mars. There'll be a million people on Mars.
Larry Magid [00:52:34]:
Hey Leo, my—
Leo Laporte [00:52:35]:
It's not gonna ever happen.
Larry Magid [00:52:36]:
I'm still not able to rent out my Tesla as a robotaxi. I've been waiting since 2018 for that money.
Leo Laporte [00:52:42]:
And did you pay for— now see, I paid $5,000. I paid $50,000 to have the right to have full self-driving on my Model X, which I never got.
Devindra Hardawar [00:52:51]:
Yeah. Oh, that's wild. That's absolutely wild. That's Elon. But it's also like, I look at what AI has done to this entire industry and to executives and people at the top who are kind of like rebuilt their businesses around this stuff. These agents are clearly working for some people. They can do amazing things. I occasionally use it just to like look up things and do some quick research.
Devindra Hardawar [00:53:13]:
I was doing a drive from the Atlanta area to Destin, Florida. So I asked, um, Gemini or something like, hey, what are things to do like along the way?
Larry Magid [00:53:21]:
Oh yeah.
Devindra Hardawar [00:53:21]:
How can I break up this trip to make it easier for the kids? And it gave me options. It gave me a place to visit. I was like, oh, that's cool. And I could do that right from my phone. Actually, no, I did it on iOS 27. So the built-in Siri AI was able to like generate some stuff for me to do. And it came up with a thing for me to go to. that's 2 hours away, and I was able to book that, and that made life a little easier.
Devindra Hardawar [00:53:41]:
So these tools do have a certain usefulness, and I think consumers will figure that out. But the business people have gone crazy.
Larry Magid [00:53:49]:
Mm-hmm.
Devindra Hardawar [00:53:50]:
Like seeing what Satya Nadella has done, seeing what Microsoft has done to kind of rework their entire business around AI, it's gonna be rough times ahead, not just for their business, but for employees, anybody who works. If any part of this bubble pops, the entire global economy is going to take a tremendous hit and it's going to be real rough.
Leo Laporte [00:54:10]:
Let me show you something I've been working on for the last month. Uh, we had a Twit ad sales system that somebody, uh, Jeff Needles wrote for us 12 years ago that we've been limping along with. Now the good news is I had that software and that became, uh, kind of a spec for, uh, this. I haven't written a line of code in this. This has all been written by AI. And, um, it is an incredibly powerful sales system that we could have maybe gotten something off the shelf that would kind of come close. But now, one of the problems I've had is getting, uh, the humans to review it. So I add— I said, could you put a little button on here that's a, a bug button?
Allyn Malventano [00:54:56]:
Yeah.
Leo Laporte [00:54:57]:
And it said, oh yeah, sure. So here's a little bug button. What's wrong? And then I said, could you make it so they could record their thoughts as they're looking at the page? Oh, you have to allow the microphone. There we go. And now I'm recording my thoughts about this. And it— and here's the funny thing. I said, if it's a simple thing to fix, just fix it. Don't— you don't have to ask me.
Leo Laporte [00:55:18]:
If it's more complicated, uh, then, um, Yeah. let me know and we can go over it.
Allyn Malventano [00:55:26]:
Uh-huh. Simply hack Leo's LAN.
Devindra Hardawar [00:55:29]:
Yeah, yeah. No way this can go wrong.
Leo Laporte [00:55:32]:
No way. That's why Mr. Critical says, I'm not going to use this. What if it, what if it, I don't know, blows up the world? It's not. First of all, it's very secure. We make sure it's very secure. It's behind, it's a, you know, has lots of security behind it.
Devindra Hardawar [00:55:44]:
You're more secure than OpenAI apparently, Leo.
Leo Laporte [00:55:46]:
So I am.
Allyn Malventano [00:55:47]:
Yeah.
Leo Laporte [00:55:48]:
Well, I'm counting on companies like Cloudflare to You know, provide security, front-end security and so forth. Actually, it uses a Google SSO. I'm hoping that it's all, but we just added a piece of, so that while you're doing an ad order, you can export it to an Excel, to a Google Sheet. It can do Excel. I said, well, it'd be nice if you'd save it to a Google Sheet on the Google Drive. Boopity boopity boop. And now it's doing that. And I, this is a, A simple project I worked on, I, uh, for last month.
Leo Laporte [00:56:21]:
I had Fable work on it. I had Claude, uh, code Fable, uh, Opus 4.8, now Opus 5. Opus 4.8 was doing the coding. Fable was designing. I had ChatGPT, Sol look at it, review it. They go back, as I said, they go back and forth. I had, uh, Quen 5.38 and, uh, actually Grok was great except I ran out of usage. I only had the $30 plan.
Leo Laporte [00:56:43]:
Um, but Grok 4.5 was really helpful. They all worked on it together.
Larry Magid [00:56:47]:
It sounds like you have all your employees. You just have—
Leo Laporte [00:56:50]:
It's like employees.
Larry Magid [00:56:51]:
Yeah, I hired Grok the other day. He's a really nice guy, a little crazy, but, uh—
Leo Laporte [00:56:55]:
It's, it's— so this is a line of business app that, uh, took us a year to write with a human. I haven't written— I haven't even looked at the code. I mean, what was interesting is there's 81,000 lines of code. But almost half of them are tests. It's, uh, you know, I say you've got to do red, green, blue testing on everything. And so it's, it's doing all that. If you have engineers—
Devindra Hardawar [00:57:18]:
But what happens when something goes wrong, Leo, right? Like when, when you can't just look at the models and be like, hey, what is, what is happening here, right? And a human has to go look at it. People are reporting, like jumping into these vibe-coded apps, the actual developers don't know what the hell is going on because the code is like gobbledygook.
Larry Magid [00:57:33]:
Right. Can't use AI to pinpoint where the code problem is.
Leo Laporte [00:57:36]:
Well, that's why I put this little bug button. I said, fix it. If it's, if it's a cosmetic thing or something, just fix it. It's not gonna go wrong, Devindra. Knock on wood. It's never gonna go wrong.
Devindra Hardawar [00:57:46]:
It's just really— it's really—
Leo Laporte [00:57:48]:
You're not alone. Lisa's saying the same thing. She says, what if it goes wrong and you're on the air and it's not working?
Devindra Hardawar [00:57:52]:
She's very wise.
Leo Laporte [00:57:53]:
It's never gonna go wrong. There will be fewer bugs in this than there were in the original because no human was there to make those mistakes.
Devindra Hardawar [00:58:02]:
Mm-hmm.
Larry Magid [00:58:03]:
Well, what do you think, Alan?
Devindra Hardawar [00:58:05]:
Am I nuts? We just use computers to do like specific things, to make calculations.
Leo Laporte [00:58:09]:
Would you run a nuclear reactor on a submarine this way?
Allyn Malventano [00:58:13]:
Only if the code was audited.
Devindra Hardawar [00:58:15]:
Right.
Leo Laporte [00:58:15]:
I mean, but is— but let's, let's be, let's be fair, and this is the same thing with self-driving vehicles.
Allyn Malventano [00:58:20]:
Yeah, exactly.
Leo Laporte [00:58:21]:
Are humans really the best people to audit it? Are they reliable? I think humans are less reliable.
Devindra Hardawar [00:58:27]:
And that's your thing. They got us this far though.
Leo Laporte [00:58:29]:
Yeah, well, they got it. Yeah, but that's why we're at a turning point.
Allyn Malventano [00:58:33]:
We did—
Leo Laporte [00:58:34]:
it took us to get to this point.
Allyn Malventano [00:58:36]:
I mean, I mean, Leo, everything you just described And like, like, that's basically the cause for the boom right now. Right? Right. Like, there's, there's everyone around the world or in your shoes coming to the realization of the types of things that you can have AI do for you.
Leo Laporte [00:58:50]:
Mind-boggling.
Allyn Malventano [00:58:51]:
And I mean, honestly, yeah, my first like month or two just dabbling, I had to start keeping like a list because there was like, oh, there's that other project that I've—
Leo Laporte [00:59:02]:
That's the, that's the big problem.
Allyn Malventano [00:59:04]:
Yeah.
Leo Laporte [00:59:04]:
You can suddenly do all these things you wanted to do, and now that— now you lose track.
Allyn Malventano [00:59:08]:
There's so many. Yeah, once you start realizing like, oh wait, I could do that other thing. Like, so it— yeah, in my— back when I did my reverse engineering work back in the military, I made— well, call it a mistake or not, but I focused so much on the reverse engineering piece of it. Uh, I mean, I could just read machine code, right? Uh, you know, like, um, like our security now, buddy. Uh, right? Like, I'm not, I'm not good at reading code that's just machine code, but I never did the other direction to do the programming, right? I never sat down and properly figured out how to properly just be able to code just like natively. So, so I understand all of the principles, all of those, like, I have everything in my mind about like how I'd like it to work and what I know.
Leo Laporte [00:59:54]:
Well, that's what you need now. That's all you need now.
Allyn Malventano [00:59:56]:
And that's what you need now. Yeah. So it's a different sort of mentality that you sort of have to have. And if you do, then it's like, it's, it's really eye-opening And you start to realize, oh man, I could do this, I could do that. And yeah.
Larry Magid [01:00:08]:
You know, in, in our personal life, so I use it extensively to analyze my investment portfolio, for example.
Leo Laporte [01:00:13]:
I do that too.
Larry Magid [01:00:13]:
And I have found things that my broker isn't aware of. For example, I found out I'm overvalued, I'm over-heavy in certain stocks because they're in mutual funds, they're in ETFs. I mean, incredible. And the other thing, and maybe, maybe Benito, whoever's running the board, can bring this up. If you go to LarrysWorld.com and you scroll down a few things, you'll see my iPhone and AI became part of my hospital care team.
Leo Laporte [01:00:36]:
Oh, this is cool. You know what, let's—
Larry Magid [01:00:38]:
I gotta take a break.
Leo Laporte [01:00:39]:
Let's talk about this because you— this is really interesting, and I would like to talk about this.
Allyn Malventano [01:00:45]:
Absolutely.
Leo Laporte [01:00:45]:
So we're gonna take a little break.
Larry Magid [01:00:46]:
Everybody can see the inside of my intestines in a few minutes.
Allyn Malventano [01:00:50]:
Great.
Leo Laporte [01:00:51]:
Not the best tease. They're beautiful.
Larry Magid [01:00:55]:
They're gorgeous. You'll love them.
Leo Laporte [01:00:59]:
But you know, I'm doing the same thing, Larry. At the close of business every business day My little agent here downloads all everything that happened, all my accounts, and gives me a summary of how I'm doing. It's not always—
Larry Magid [01:01:10]:
Better than a broker ever would.
Leo Laporte [01:01:12]:
Oh, much better.
Larry Magid [01:01:13]:
Yeah, yeah, yeah.
Leo Laporte [01:01:14]:
Um, we're gonna have more. It's not gonna be all AI, I promise. Although this was kind of fun. I think it's— look, people say, why are you always talking about AI? This is the biggest thing, Larry. You and I both been sitting here for 50 years covering this stuff.
Larry Magid [01:01:29]:
Oh yeah.
Leo Laporte [01:01:31]:
And this is the most ex— it was getting boring, let's face it.
Larry Magid [01:01:34]:
This is the most exciting thing. Every few years I'm on the verge of doing what Dvorak did and, you know, finding some, some diversion, something else to do. And then something will happen. Windows, the Mac will come out, Windows will come out, the web will come out, broadband will come out. And yeah, that gets me very interested. Now there's just no end of fun things to do with AI. It's, it's like taken over my life. I'm not sure that I— I'm not sure I should be proud of that.
Larry Magid [01:01:57]:
But it is so rich and it's moving at such an incredible pace.
Leo Laporte [01:02:04]:
The potential is huge, but so is the risk. And that's why we began with that story. So is the risk.
Larry Magid [01:02:10]:
Absolutely.
Leo Laporte [01:02:10]:
And we're going to have to find a way to manage it. We will look at Larry's—
Allyn Malventano [01:02:13]:
Thank you, Devendra, for keeping us all realists.
Devindra Hardawar [01:02:15]:
I'm the skeptic.
Leo Laporte [01:02:17]:
No, that's good. We need the skeptics too. That's going to protect us.
Larry Magid [01:02:20]:
Yeah, we desperately need skeptics.
Leo Laporte [01:02:22]:
I firmly admit I am in AI psychosis, right now. Uh, Devendra the Skeptic Hardiwar here from Engadget, where he's a senior editor. Nice to have you, Larry. Uh, Magid from connectsafely.org, where he is the CEO and his radio career continues thanks to AI, thanks to Gems. And Alan Malventano, former submariner, uh, now a benchmark and workload lead at Soladigm. Actually, we got to talk about SSD and memory shortages. This is your field.
Allyn Malventano [01:02:56]:
I am.
Leo Laporte [01:02:56]:
One of the bad things AI has brought us is you cannot get a machine with more than 32 gigs of RAM for love or money. And that's a problem. And it's going to be that way, I think, for a while. But I'm going to ask you about that. So we'll talk about that too when we come back. Okay, our hydration break is over, gentlemen. Back to work. This is Larry's article, My Phone and AI Became Part of My Hospital Care Team.
Leo Laporte [01:03:23]:
And one of the things, especially in the United States, but I bet it's worldwide, we're facing is a dearth of help. When you go to a doctor or you go to the hospital, you're navigating a world, and unless you're very, very lucky, you're not getting the information maybe you need for the best care. Larry, tell us what happened.
Larry Magid [01:03:44]:
Well, what happened is I was in New York and I had an intestinal blockage, which I sometimes have. And I went to the ER in Greenwich Village, 'cause I, I walked to the ER in Greenwich Village and they did a CT scan and then they admitted me to Lenox Hills Hospital, which is a great hospital. The good news is, is all they did is put an IV in me and I was fine 2 days later. No surgery, nothing.
Leo Laporte [01:04:03]:
But that's scary. You're not even at home. You're out of town.
Larry Magid [01:04:06]:
Yeah, it's scary. But then I took the— so what Lenox Hills does, they allow you to download the images as well as the radiology reports. So 2 things. Number one, while I'm in the hospital, they're coming around drawing blood, right? All the time. By the time the phlebotomist leaves my room, 5 minutes later, I'm getting the data on my phone, right?
Leo Laporte [01:04:23]:
That's really interesting.
Larry Magid [01:04:24]:
So when the doc walks in, I can ask him or her or them questions. So that's one thing. And the same thing, they did a couple of scans. But the amazing thing is I took the CT scan, I took the image, ran it through ChatGPT without giving it the radiology report. And it gave me its own radiology report, which was almost identical to what the radiologist said. Then I said to ChatGPT, I don't understand exactly— I understand what happened, but tell me visually what happened. Where is the blockage? Why is it happening? And as you can see from that image—
Leo Laporte [01:04:55]:
This was generated by ChatGPT?
Devindra Hardawar [01:04:56]:
Yeah.
Larry Magid [01:04:57]:
This was— so you see the yellow line? I can't quite see your image. The yellow line is where a previous surgery happened. And the red line, I believe, is where the current blockage is. And basically, I now know— I knew I had scar tissue from a previous surgery. I now know exactly where that scar tissue is. I know exactly where the blockage was. That bile duct that's surrounded with a marker is full of stuff that shouldn't be there, fecal matter that should have passed through. So I know exactly why I had this blockage.
Larry Magid [01:05:28]:
Now, I don't know if it's going to help me prevent it, but at least I have a clear image of what happened. And what's funny about this is I happen to be attending I was giving an AI seminar at Stanford Medical a few weeks ago, and I brought this up to the head radiologist at Stanford. And he at first said, oh, you know, radiologists are not threatened. And I said, I agree, we need radiologists. But from a patient education standpoint, no doctor, no radiologist has ever explained to me what's going on inside of my body as well as this illustration has. And so now whenever I get any information, any medical information, I get one of the AI models and I vary between ChatGPT, Claude, Gemini, and a few others. I get them to explain it to me.
Leo Laporte [01:06:13]:
Did you show this to the radiologist at the hospital?
Larry Magid [01:06:15]:
I did. He was kind of blown away. I also showed it to my gastroenterologist who said, you know, I don't know how to read a CT scan.
Leo Laporte [01:06:23]:
He goes to the radiologist.
Larry Magid [01:06:24]:
He goes to the radiologist. But this gave me more than the radiologist. So yeah, I showed it to the head radiologist. He was pretty impressed.
Leo Laporte [01:06:30]:
You know, I think one of the concerns besides losing their jobs, and you know, it's very famously, um, I can't remember who it was, uh, maybe Dario Amodei, I think it was, the CEO and founder of, uh, Anthropic, said, oh, radiologists, I won't have a job in 10 years. He said that 10 years ago. They do still have a job.
Larry Magid [01:06:47]:
No, he's still moving.
Leo Laporte [01:06:49]:
Yeah, but, uh, I think not only are they threatened about jobs, they also are worried that you're gonna get bad information.
Larry Magid [01:06:55]:
Yeah.
Leo Laporte [01:06:56]:
And then they're going to have to be arguing against a machine.
Larry Magid [01:07:00]:
But, you know, they make mistakes too. They're human. They make mistakes. And these things can see things that radiologists don't see. Now, you still need, I think, some wise person to put it in perspective.
Allyn Malventano [01:07:11]:
Right.
Larry Magid [01:07:12]:
Because the problem— and I don't care, you know, I do the same thing with my Apple Watch.
Leo Laporte [01:07:15]:
Jeffrey Hinton. Thank you. Joe corrected me. It was Jeffrey Hinton.
Larry Magid [01:07:18]:
I will run my Apple Watch reports or I'm now wearing this Google Fitbit Air. I'll run— although the Fitbit Air actually has its own AI, which is fantastic, the Google Health. And it will explain technical details in ways that not only can I understand, but it puts it in perspective. So for example, Google Health will take all of my vitals in the context of my activity for the day and my health information because it's linked to my medical records. And it will put it all into context in ways, frankly, that internists typically do.
Allyn Malventano [01:07:48]:
Yeah.
Leo Laporte [01:07:49]:
So, and increasingly don't have the time to do. Oh yeah, that's part of it is that, uh, you know, my doctor went from 1,000 patients to 5,000 patients.
Allyn Malventano [01:07:59]:
Oh my God.
Devindra Hardawar [01:08:00]:
Yeah.
Leo Laporte [01:08:00]:
You know, and so he just doesn't have the time to explain stuff to me. When I went on Ozempic a couple of years ago, he just said, here's this prescription, go, go get it. And I had to do a lot of reading and studying.
Devindra Hardawar [01:08:12]:
Yep.
Leo Laporte [01:08:12]:
And I, and I had symptoms that I didn't know anything about. And yeah, I could email him, but it's a long process. It was very helpful to get that information. And I've had my genome done by Nebula Genomics, which does the full genome. I've had biome tests. I wear all the rings and watches and my scale and my blood. I get all the stuff, right?
Allyn Malventano [01:08:34]:
Yeah.
Leo Laporte [01:08:34]:
So I feed that into the AI and I find that I have more information than before. The AI told me, for instance, based on your biome test, this is what you should be eating now. Now we'll see. I'm going to do another— it said, and then 12 weeks from now, do it again, and we'll see if you've improved.
Larry Magid [01:08:52]:
I want to put one caveat. I'm not going to do a medical procedure or take—
Leo Laporte [01:08:56]:
No, I wouldn't have it operate on me.
Larry Magid [01:08:59]:
I might, I might change my diet a little bit, but I'm not going to take—
Leo Laporte [01:09:01]:
Yes, I don't mind eating more, you know, blueberries. Okay, fine.
Larry Magid [01:09:05]:
That's not going to hurt you, right? So you are starting to—
Leo Laporte [01:09:08]:
You have to use your head. Yeah, but I'll pay for it. But my doctor will never tell me that stuff. You know, he told me drink more green tea and eat carrots. Well, I had—
Allyn Malventano [01:09:16]:
I had—
Leo Laporte [01:09:17]:
I love him. He's a great doctor and he wants to do more, but he doesn't have time to do more.
Larry Magid [01:09:21]:
I had a skin infection in Prague at 3 in the morning. There was no way I was—
Leo Laporte [01:09:25]:
You're just falling apart, dude.
Larry Magid [01:09:26]:
I know. No way I was gonna get real medical advice. So for that, I actually did use AI. And luckily I had the medication with me that it recommended. It recommended a particular medication. I happened to have it with me and fine, it worked. But by and large, I don't take pills based on what AI says.
Allyn Malventano [01:09:43]:
The, uh, so the reason that like a radiologist is still gonna have a job is the same reason that like I'm— I've run into this a lot myself. So here's an example from my work. Yeah, it would be great if we could just make like the all-knowing, all-seeing, all-doing, uh, you know, SSD benchmark AI as an example, right? Just pick an example. Problem is you need the expert to recognize when the thing is trying to BS you with the wrong answer.
Devindra Hardawar [01:10:08]:
Yeah.
Allyn Malventano [01:10:08]:
answer.
Larry Magid [01:10:08]:
Right.
Allyn Malventano [01:10:09]:
Right. So if anything, you have to be a little bit— it forces you to be a little bit smarter when reviewing its results, because it will try to trick the person that doesn't know any better into going down a bad path, which is one of the other things to, you know, exercise caution with. Right. And that's the reason you're not going to see a bunch of radiologists going out of work, because like, yeah, they might use AI as a tool to assist their work. Maybe they can be better at presenting more detailed chart. Maybe eventually their tools at their hospital will just do what Larry's AI did for him to draw on the scan for them, but still be reviewed. That output will then be reviewed by the radiologist to make sure it drew the circles in the right place and then pass it on to the patient.
Larry Magid [01:10:53]:
At the risk of being self-serving to the 4 of us, in the hands of an expert, AI is an incredibly powerful tool.
Allyn Malventano [01:11:01]:
Yep.
Larry Magid [01:11:01]:
Now, I wouldn't want to be the guy who gets hired by the PR firm to write draft press releases, and that guy doesn't exist anymore. I wouldn't want to be the junior lawyer who's trying to break into a law firm to do the grunt work, because that, for them, they're being replaced by AI. But the experts, the senior people, are able to greatly improve their productivity and the quality of their work.
Leo Laporte [01:11:21]:
That's the job now, is to become an expert. I mean, that's really why— How do you become an expert?
Allyn Malventano [01:11:26]:
How do you get there? Well, how do you get there 10 years from now? How do those people get there? Right.
Leo Laporte [01:11:32]:
Start now.
Devindra Hardawar [01:11:33]:
We'll show up.
Leo Laporte [01:11:34]:
And the thing is, it's not even an expensive job.
Larry Magid [01:11:35]:
Who's gonna hire you now? That's the problem.
Allyn Malventano [01:11:37]:
That's, that's the point. Well, I understand. There's a different kind of experience. Like, so, okay, so—
Leo Laporte [01:11:42]:
It doesn't take that long. If you, if you know, I start it all, my eyes opened with Opus 4.5, November 24th, 2025, a day I will never forget. I started using it and I went, holy moly.
Larry Magid [01:11:55]:
You're talking about being an AI expert. I think I'm talking about being a subject matter expert.
Leo Laporte [01:11:59]:
Oh, well, okay. Well, I'm a subject matter expert in podcasting. I guess.
Larry Magid [01:12:03]:
Yeah. But you're not gonna be as smart as Alan is about SSDs in any situation.
Leo Laporte [01:12:06]:
No, never.
Larry Magid [01:12:07]:
You know?
Allyn Malventano [01:12:08]:
Oh, I'm sorry.
Leo Laporte [01:12:09]:
I misunderstood you.
Allyn Malventano [01:12:10]:
How does— yeah. Yeah. So how does the next person get to that level in that particular field with, without having gone through the same—
Leo Laporte [01:12:17]:
Couldn't AI help you learn all that?
Allyn Malventano [01:12:20]:
I mean, it could, but that's not the same as experience.
Devindra Hardawar [01:12:22]:
And it's not the same as like critical thinking. Did you guys hear about the David Eggers went to OpenAI?
Allyn Malventano [01:12:28]:
Yeah.
Devindra Hardawar [01:12:28]:
To do the speech and he spent the entire time saying, you guys are ruining the world. But the one thing he said is that ChatGPT is silencing an entire generation.
Allyn Malventano [01:12:36]:
Right.
Devindra Hardawar [01:12:37]:
And I think about that.
Larry Magid [01:12:38]:
Silence.
Devindra Hardawar [01:12:38]:
Because schools are starting to push AI tools and in colleges it's already a problem and we don't know what impact it's gonna have on education for at least the current generation of kids.
Larry Magid [01:12:51]:
How does it silence people?
Leo Laporte [01:12:53]:
Make me—
Devindra Hardawar [01:12:54]:
Well, because it means, I think there's a lot of offloading going on versus critical thinking. Right.
Leo Laporte [01:13:00]:
Well, there are a lot of kids which they need to knock this off. who are saying, I don't need to study or learn anymore, right, because I can just use AI. Now knock it off, kids, if you're doing that, because that is a bad idea. That's what the tool is.
Devindra Hardawar [01:13:12]:
You're giving— you're giving this thing that gives all the answers, right?
Leo Laporte [01:13:16]:
Well, people said the same thing when you had Google Search, right? Oh, now everybody doesn't have to— nobody has to go to the library anymore. How are the kids ever going to learn? Well, I mean, you need to adapt to the new way, except Those things that, like, the calculator is a good example, right?
Devindra Hardawar [01:13:31]:
The calculator, yeah, it gives you the answer. And to be really great, to, you know, in a math-heavy field, you have to actually know what you're doing, right, to get those answers. Except now we've created this thing that also offloads critical thinking and creative thinking in ways that could be potentially dangerous. Yeah, I saw people talking about this on Blue Sky, like, even the act of, like, bullshitting a paper, right? Say you had a paper due the next day, you had to write 5 to 10, you know, pages on something, and you barely knew the subject The, the act of, you know, BSing your way through that paper at least teaches you something. You gain something from that subject versus kind of just like slopping up everything that the AI will tell you.
Larry Magid [01:14:11]:
Yeah.
Leo Laporte [01:14:12]:
Anthony Nielsen, our creative director and a very proficient AI user, says AI is both a shortcut and a tool to learn, but humans are wired to take the shortcut.
Devindra Hardawar [01:14:21]:
I mean, look at all that happened with the OpenAI things. Like, It wanted the short— it didn't want to figure out the answers, right? It wanted to go figure out where to just steal the answer.
Allyn Malventano [01:14:30]:
So— Yeah, it took the shortcut too.
Leo Laporte [01:14:32]:
Yeah, AI did the same thing.
Devindra Hardawar [01:14:34]:
So humans are going to do this. And the thing we can't just— there is so little preparation for teachers, for professors. Like, there are professors saying, my entire class is using AI to write papers and to do things. I can't stop this. I have no way to stop this. I can't really detect this.
Leo Laporte [01:14:49]:
What do I do?
Devindra Hardawar [01:14:50]:
All these companies kind of threw Kids cheat too, right?
Leo Laporte [01:14:53]:
I mean, you just have to explain to kids, you're cheating yourself. You know, yes, you're going to get the grade, but you're cheating yourself.
Larry Magid [01:15:01]:
You're—
Leo Laporte [01:15:01]:
this is your chance, your one chance to get an education.
Larry Magid [01:15:04]:
I know it's not exactly the same thing, but is it cheating to use a power saw instead of a handsaw? There are people who actually—
Leo Laporte [01:15:09]:
Or ride a bicycle instead of walking, or drive a car instead of riding a bicycle.
Larry Magid [01:15:13]:
Well, if you're riding a bicycle in a foot race, you are cheating. It depends on what it is you're trying to accomplish. There are people who think it's, you know, artisans who believe you should be using hand tools, but that's not how we build houses.
Leo Laporte [01:15:26]:
Here's the thing.
Devindra Hardawar [01:15:26]:
For specific things, yeah.
Leo Laporte [01:15:28]:
There are artisans who will continue to use hand tools because that's their art.
Devindra Hardawar [01:15:32]:
Yeah.
Leo Laporte [01:15:34]:
When, you know, there were a lot of people in North Carolina who worked in the furniture business, building furniture, beautiful, wonderful, better furniture than you get today, gorgeous stuff. But when automation came in, they lost their jobs.
Allyn Malventano [01:15:46]:
Yeah.
Leo Laporte [01:15:46]:
And we now have cheap furniture that falls apart in a few years. Uh, there are still people who hand-make furniture, but it's very, very expensive. If you care about having furniture that's going to last, you'll go to artisans to get that built for you. That's the way of the world. Technology comes along and changes things.
Devindra Hardawar [01:16:07]:
It's different. It's, it's not the same thing.
Leo Laporte [01:16:08]:
And you can't— by the way, you can't stop it. You're not going to be able to stop it. You might not like it.
Devindra Hardawar [01:16:12]:
I don't like this line of thinking. Like, here, we stopped the nuclear proliferation to a certain degree once we realized it could explode the world, right? We collectively as a society, it was like, hey, this is not a great thing that everybody should have these world-ending missiles. And at least something was done to try to rein that in. Alan, I'm sure you could probably talk more about that. But I do think we have to think about the proliferation of AI as something on that level, right? It's not just a tool.
Allyn Malventano [01:16:42]:
Yeah.
Devindra Hardawar [01:16:42]:
It's not just the bicycle, it's not just the child's toy.
Leo Laporte [01:16:44]:
It should be treated like the atomic bomb, you think?
Devindra Hardawar [01:16:45]:
It is a thing that can potentially obliterate the way we think, the way we create.
Leo Laporte [01:16:49]:
That's the debate going on right now. Washington would like to shut down open models. China was apparently thinking about that, although President Xi, I thought, gave a very good speech, believe it or not, at the AI summit last week, in which he basically said, we believe in OpenAI. We believe in open models. And actually, if you're China, that makes a lot of sense.
Devindra Hardawar [01:17:16]:
But also, like, that's— China wants to do this because it means awesome levels of productivity, right? Every worker can all of a sudden do a crap ton more. And that is— that's what the corporate level of AI adoption wants. I don't— I do think we are starting to lose something fundamentally human about how we read and how we explore and how we create things. And that's why I'm worried about AI when it comes to schools. schools, we have no idea what we're doing. We just dropped this tool on teachers, on administrations. They are not prepared for it pedagogically. Like, we don't know what, what, how to really incorporate AI into education.
Devindra Hardawar [01:17:52]:
And it's just going to land there, and an entire generation of kids is going to go through school with this heavy, like, AI tool usage.
Allyn Malventano [01:17:58]:
Yeah.
Devindra Hardawar [01:17:59]:
It's, it's sort of like the thing that happened when, um, they were teaching, uh, what the, the skill to teach kids how to read, right? The way We taught kids how to read in America. Now there's a lot of people going to look back like, oh, maybe we should have— maybe that wasn't the best way.
Leo Laporte [01:18:13]:
Well, they're pulling back now on putting laptops in classrooms, right?
Larry Magid [01:18:18]:
Yeah.
Devindra Hardawar [01:18:18]:
I mean, I can see why. My daughter is going through that right now, and especially Chromebooks. Like, whatever we thought Chromebooks could do, like those, it depends on how it's used, and a lot of school districts can barely even get that right. So now you give of you have this AI, you have AI as something that could completely destroy the way kids learn, or at least like change it in ways that we're not prepared for, that's an even bigger deal. Like, I do feel like the, the problem is this stuff was just dumped on everybody, right? And we are just like holding on to the armrest trying to figure out how to like deal with it.
Larry Magid [01:18:50]:
But going back to my hand tool analogy, I mean, where is it written that we have to do things the way we've always done You know, where is it written that we have to know how to write? I know that sounds absurd, but, but I mean, especially a guy who made his living as a writer.
Devindra Hardawar [01:19:05]:
Yeah.
Larry Magid [01:19:06]:
But, but the point is that just because I made my living as a writer doesn't mean that, that Johnny and Susie necessarily have to be— have to learn those skills any more than you— I need to have to remember how to do long division anymore. I've long forgotten how to do long division. I don't care.
Devindra Hardawar [01:19:21]:
Because I've had dark ages before. Right? Like, what brings about the Dark Ages? If you go see The Odyssey, at the, you know, like the way Nolan has twisted, has kind of like told his story of The Odyssey, it is very much about a civilization falling apart. And even towards the end, like, no spoilers, but there's a reference to like, oh yeah, people will forget how to write. Our stories will be told to the people—
Larry Magid [01:19:45]:
Socrates worried about that. Socrates wrote about—
Leo Laporte [01:19:47]:
Yes.
Larry Magid [01:19:48]:
He wrote about how the quill The quill is going to destroy memory. Actually, not they forget about— he worried that the quill would destroy memory. I actually have a presentation that's on Connect Safely, which is the history of techno panics, moral techno panics. And, you know, everywhere along the line, the telegraph, the radio, obviously the railroad, every single technology that's come along, there have been critics.
Allyn Malventano [01:20:12]:
For sure.
Larry Magid [01:20:12]:
The word Luddite had to do with people who were worried about Luddites.
Devindra Hardawar [01:20:17]:
Well, they were worried about their jobs. They were worried about, like, their very existence.
Larry Magid [01:20:21]:
You know, like, they didn't take technology. You know, there were people who wrote articles about railway madmen. They were— that women shouldn't ride railroad car trains because men will go crazy from the speed and attack them.
Allyn Malventano [01:20:30]:
For sure.
Devindra Hardawar [01:20:31]:
You know, there were bad societal responses to all this stuff. My thing is, this is nuclear level. This is in terms of, like, you're not just— it's not just another tool. It's the brain is being offloaded. Our thinking, our ability to create, like what a lot of these companies want. They spent years trying to get us interested in generative AI images and generative AI video. Sora just completely died because there's no real market for it. Crap, nobody wants it.
Devindra Hardawar [01:21:01]:
We need actual human creativity, right? Like the thing, the reason the Odyssey and other stories have been around for so long is like we, We need stories, we need human thought. I think like there is something, there is something that worries me.
Leo Laporte [01:21:14]:
I love it that Devendra—
Larry Magid [01:21:15]:
I'm much more of an optimist than you are.
Leo Laporte [01:21:17]:
Devendra is the young guy here. You and I, Larry, are the old guys.
Devindra Hardawar [01:21:20]:
I'm 42 years old.
Leo Laporte [01:21:22]:
I'm not young.
Larry Magid [01:21:23]:
Trust me.
Devindra Hardawar [01:21:24]:
I'm mostly gray.
Leo Laporte [01:21:25]:
And the old guys are the ones saying, this is exciting, the future is great. And the young guy's going, oh man, you kids today. You and your long hair and rock and roll. Cover story of The Atlantic magazine this month, the end of reading is here. Rose Horowitz said optimists once believed universal literacy was inevitable. Now it seems the age of reading might be a short anomaly in human history because we're all moving to video. Look at the world's changing. Technology is changing it faster than ever.
Leo Laporte [01:21:55]:
I mean, that's the thing that scares people the most about AI is it's changing.
Allyn Malventano [01:21:59]:
I mean, I mean, Leo, that's part of the reason I got out of doing reviews.
Leo Laporte [01:22:02]:
Yes.
Allyn Malventano [01:22:04]:
The written part of that was a dying art, and that was like a decade ago.
Leo Laporte [01:22:08]:
We used to think that the best way— and it did used to be the best way for us to promote our network— was to do lots of reviews that would get searched for on YouTube, and then people would see them and they would promote the network. Nobody does that anymore. It's dead. But I mean, dead in my lifetime. It's dead in the last 15 minutes. That's what's so amazing about the speed with which this is happening. I started writing reviews for Byte magazine. That's how I got started in technology in the early '80s.
Allyn Malventano [01:22:35]:
Yep.
Leo Laporte [01:22:36]:
Nobody writes or reads.
Devindra Hardawar [01:22:37]:
I mean, I should be going all in on short-form video at this point, right? I should just stop writing.
Leo Laporte [01:22:41]:
Absolutely. You should be a TikTok guy.
Devindra Hardawar [01:22:44]:
I'm trying to do things.
Larry Magid [01:22:45]:
Yeah. The problem is, Leo, when you were writing for Byte magazine, even if we didn't know who Leo Laporte was, we knew who Byte magazine was, and we knew they had some standards that they didn't just hire anybody off the street.
Leo Laporte [01:22:56]:
That's probably what happened over here.
Larry Magid [01:22:57]:
Like, for example, my, my good friend Walt Mossberg, I don't agree with everything he said back when he was a journalist, but I, I, I agreed with the fact that he, when he said it, he was a trustworthy guy who put some research into it. And I could trust, and the same thing with DeVore, any, anybody who I trust. And that was the beauty of the, the legacy media companies. They only hired people who were competent for the most part.
Leo Laporte [01:23:16]:
Aren't we learning though that we, yeah, we trusted the New York Times, we trusted these And aren't we learning, in fact, that they probably weren't all that trusting of people?
Larry Magid [01:23:26]:
They hired me. They couldn't be that trusting of me.
Leo Laporte [01:23:28]:
And then me. Let's take another break. No more AI. We got other stories. Google made a ton of money. Actually, it's an AI story. Never mind.
Allyn Malventano [01:23:38]:
We'll talk more.
Larry Magid [01:23:38]:
It's all about the AI.
Devindra Hardawar [01:23:39]:
It's all an AI story.
Allyn Malventano [01:23:41]:
You said no more AI. I was like, well, there goes the rest of the show.
Leo Laporte [01:23:44]:
There goes the rest of the show.
Allyn Malventano [01:23:45]:
We're in it right now.
Larry Magid [01:23:45]:
There goes our lives.
Leo Laporte [01:23:46]:
I got an email from somebody. who's a longtime listener says, I can't listen to the show anymore. I don't even know what you guys are talking about. This is, you know, when we— when the iPhone came out, I got a lot of email from people saying, will you guys stop talking about the iPhone? But we knew this was going to change everything. It did. 2007 was a watershed moment.
Larry Magid [01:24:04]:
All your listeners have to do is take the MP3 of this broadcast, put it into Gemini, and it will explain to them in simple language what we've been trying to say.
Leo Laporte [01:24:13]:
Actually, I did that. I got in a big fight night with Paris, who much like you, Devendra, is skeptical on our AI show about all of this, had very much your point of view. And I got, I got really got my dander up. And I put the transcript of it into my agent and I said, was I wrong? It said, yeah, you were so wrong. And it gave me, and it said, this is the apology you should give to Paris. Let me explain why you overclaimed. Here's what's true. Here's what's not true.
Allyn Malventano [01:24:38]:
Wow.
Leo Laporte [01:24:38]:
It was actually really, and because it came from a machine, Not my wife, who also said the same thing.
Allyn Malventano [01:24:45]:
It's just cold truth, right? It's just like—
Leo Laporte [01:24:47]:
It was like, oh yeah, I guess you're right. Didn't get my hackles up so much. I just saw a Pew study that said some huge number of people now turn to AI for advice and counsel.
Devindra Hardawar [01:25:01]:
Oh, absolutely.
Leo Laporte [01:25:02]:
You know, this has become— maybe again, it's kind of sort of Because same thing with doctors, we just don't— there aren't enough therapists to go around. There aren't enough—
Devindra Hardawar [01:25:14]:
You can't afford therapists.
Larry Magid [01:25:15]:
There isn't enough mental health to go around.
Allyn Malventano [01:25:17]:
Yeah.
Leo Laporte [01:25:17]:
And so, uh, we are sort of turning to AI for those things. I don't know if it's bad.
Devindra Hardawar [01:25:23]:
Um, I, you know, I'm sure there is something about our society that this is where we are. Yeah, that's all. Yeah.
Leo Laporte [01:25:30]:
I'm more bullish. I'm very optimistic, but partly it's because of my own experience with this.
Larry Magid [01:25:38]:
But I want to make one comment. So I was around in the '60s when fellow Berkeley students and Harvard students were experimenting with LSD. And for the most part, we had good experiences.
Devindra Hardawar [01:25:49]:
That was me.
Leo Laporte [01:25:50]:
Experimenting. And we thought the world was— we thought like everything's changed.
Larry Magid [01:25:54]:
But as it turned out—
Leo Laporte [01:25:54]:
Blew my mind.
Larry Magid [01:25:55]:
When it got in the hands of millions and millions of people, it was a little different than Berkeley and Harvard students. And I'm not trying to be an elitist here.
Devindra Hardawar [01:26:02]:
Yeah.
Larry Magid [01:26:02]:
But Leo, you are a real— you are smarter than the average bear. And I'm a little nervous saying that.
Leo Laporte [01:26:08]:
I'm definitely smarter than a bear.
Allyn Malventano [01:26:10]:
I agree. Yeah, no, barely.
Larry Magid [01:26:11]:
But no, I'm a little nervous saying this.
Leo Laporte [01:26:13]:
I will grant you that. Not smarter than Devendra or Alan or you, but a bear, yeah, I'll go for it.
Larry Magid [01:26:18]:
But when you put this in the general population, and as much as I hate to agree with Devendra, when you put this out to millions or billions of people who don't have the education, the critical thinking skills, Whatever it is that you have.
Leo Laporte [01:26:30]:
I think that's kind of patronizing.
Larry Magid [01:26:32]:
It is patronizing.
Devindra Hardawar [01:26:33]:
I agree.
Allyn Malventano [01:26:33]:
It's patronizing, but it's necessary.
Larry Magid [01:26:35]:
It's patronizing, but it's not really true.
Leo Laporte [01:26:37]:
It's too good for you guys. No, I think it's quite the opposite. I think this is going to empower people.
Larry Magid [01:26:45]:
That's my hope.
Leo Laporte [01:26:45]:
In a way that they've never been empowered before. And it's, uh, to me, very exciting. But again, you know, we'll see. A lot of people think there's going to be an AI bubble that's going to burst. I don't think it is. I think this is— I think this is—
Devindra Hardawar [01:27:00]:
Oh yeah, all these companies are making a full profit, Leo. They're fully sustainable businesses.
Leo Laporte [01:27:03]:
Well, we don't know because none of them are public yet.
Larry Magid [01:27:08]:
Google is.
Leo Laporte [01:27:09]:
Yeah, well, Google's making a lot of money and the market doesn't like it. All right, let's take a break. We'll talk about that. I'll just give you a hint, it's a lot of money. Uh, you're watching This Week in Tech. Alan Malventano, smarter than the average bear. Larry Magid, smarter than the average bear. Devendra Hardwar, smarter than the average bear.
Leo Laporte [01:27:29]:
Me, I'm the average bear. Alphabet's profits quarterly report out this week. Profit quadrupled. Can you say 400% in one quarter?
Devindra Hardawar [01:27:44]:
Yeah.
Leo Laporte [01:27:45]:
This is profit. $112 billion in 3 months. And, you know, Google says it's because, uh, our investments in AI are paying off. It's funny because I don't think anybody thinks Gemini— I know you use Gemini Gems, but I don't think anybody thinks Gemini is the best AI out there. But it is widely used by real people. Maybe that's why.
Allyn Malventano [01:28:09]:
It's built into all their— they're all being used more and more.
Larry Magid [01:28:11]:
Yeah.
Leo Laporte [01:28:12]:
And it's in search.
Larry Magid [01:28:13]:
Yeah.
Leo Laporte [01:28:13]:
Right.
Larry Magid [01:28:14]:
Search, it's in Google Docs, it's everywhere.
Allyn Malventano [01:28:16]:
Yeah.
Leo Laporte [01:28:17]:
The stock market, however, was a little skeptical because of Google's CapEx, its expenditure to build that AI. Revenue was up, but so was the amount of money they're spending. Let me see if I can find the CapEx expenditures here. I'm looking at the New York Times report, Google estimated spending would be— now they had said, I think, $180 billion last quarter. They now think it's going to be $195 billion to $205 billion this year. That's double what they spent last year. Yeah, the earlier projections was $190 billion. So it's— they're going, yeah, we're gonna spend even more.
Leo Laporte [01:29:02]:
Sundar Pichai said our AI investments are redefining what's possible across every part of our business. I know I'm saying Google Well, it's Alphabet, but it's Google, right? Alphabet, okay. But does anything else in Alphabet's portfolio matter?
Devindra Hardawar [01:29:16]:
No.
Allyn Malventano [01:29:18]:
Google.
Leo Laporte [01:29:18]:
Google Search, Google Ads, Google AI. Google said in June it was raising $80 billion to fund its data center projects. It did that by selling stock. That was before they raised their spending forecast. Stock markets, as a result, Well, punished Google's stock price. I actually didn't look to see if they came back in the rest of the week, but they went down sharply when they announced this. Let me just look real quickly at their last 5 days.
Allyn Malventano [01:29:55]:
They're—
Leo Laporte [01:29:55]:
you know what? They're coming back up a little bit. They're coming back up.
Devindra Hardawar [01:30:00]:
Were they wondering, should Google have paid out dividends or something?
Leo Laporte [01:30:03]:
Yeah. I mean, that's one of the things Google does, right? To, uh, you know, they buy back stock. Um, they, uh, they give you dividends. Google is currently worth $3.9 trillion. We are in the era of, you know, $4 trillion companies at this point. Apple's worth more than $4 trillion.
Devindra Hardawar [01:30:22]:
Billion-dollar man.
Leo Laporte [01:30:24]:
It's amazing. But is it a bubble? It's going to all pop and these companies are going to go worthless down to nothing.
Devindra Hardawar [01:30:33]:
Not Google, but—
Leo Laporte [01:30:34]:
well, Google's—
Devindra Hardawar [01:30:35]:
the others could be bad.
Leo Laporte [01:30:36]:
Google's current stock price, $319. That's off from $404, their 52-week high. So yeah, the market's not excited about the amount of money they're spending.
Devindra Hardawar [01:30:49]:
Um, isn't it great as we face a climate apocalypse that we're just burning through resources and money to build AI data centers? It's awesome.
Leo Laporte [01:30:59]:
Zero emissions thing we were talking about, not so much.
Devindra Hardawar [01:31:02]:
All their environmental goals, everybody's gone.
Larry Magid [01:31:05]:
Shot.
Allyn Malventano [01:31:06]:
I mean, Microsoft bought that reactor plant. Yeah, right.
Leo Laporte [01:31:10]:
Yeah, Microsoft's, uh, saying we're gonna slip, we're not gonna make our goals. Google's saying we're gonna slip, we're not gonna make our goals. Um, you know, but climate change, climate change, people, that's all— that was so yesterday.
Devindra Hardawar [01:31:23]:
That's a problem for my children.
Leo Laporte [01:31:25]:
Yeah, that's your kids. That's not— my kids are already out of that. Anyway, it was a— it was a— now that was the very good news. The bad news is the EU is now going to fine Google a billion dollars.
Larry Magid [01:31:38]:
Yeah, pay.
Leo Laporte [01:31:39]:
Good shit. This, this, you know, they've been going back and forth. They've decided, yeah, we're gonna fine you for prioritizing our own services and search. This all started with Google Shopping. This is a European shopping company said Google isn't putting us high enough in the search results. Um, Google said it would consider appealing the penalty. Uh, President of Global Affairs Kent Walker said this isn't fair competition, it's product degradation driven by a small group of self-serving complainants with European businesses and consumers taking the hit. You'll be sorry, you Uh, the funny thing is the EU had just fined, uh, AliExpress, Alibaba, for fraudulent, uh, advertising.
Leo Laporte [01:32:28]:
The biggest penalty yet, €550 million. It was the biggest penalty for about 2 days. AliExpress failing to stop the sale of fake and unsafe goods. Well, isn't that why we go to AliExpress? Isn't that exactly what we want? So that €550 fine was dwarfed by Google's.
Larry Magid [01:32:52]:
Europeans take counterfeiting very seriously.
Leo Laporte [01:32:53]:
They do. They do.
Larry Magid [01:32:55]:
Don't try to sell a fake bag in Florence. You'll get a €10,000 fine.
Leo Laporte [01:32:59]:
Meanwhile, the US administration is saying, you fine our guys, we're gonna tariff the hell out of you. If anybody's going to punish big tech, it's going to be us, not you guys.
Devindra Hardawar [01:33:15]:
But barely, except not—
Leo Laporte [01:33:16]:
but a little, but only if they don't kiss the golden ring.
Larry Magid [01:33:20]:
Yeah, yeah.
Leo Laporte [01:33:23]:
Uh, speaking of kissing the golden ring, oh boy, Apple is being sued. Apple's being sued by customers because they— now this is interesting.
Devindra Hardawar [01:33:36]:
I don't—
Leo Laporte [01:33:37]:
I wonder if this case will go very far. I don't know. No, customers say that you allowed a fake Bitcoin wallet— actually, there were several— in the App Store, and we lost a total of $1.8 million due to this fake wallet, and it's your fault, Apple, because you're supposed to protect us.
Larry Magid [01:33:54]:
It wasn't Trump Coin, was it?
Leo Laporte [01:33:56]:
No, it's a Bitcoin wallet called Sparrow Wallet. It's, by the way, not just the App Store. Oh, wait a minute. No, wait a minute. There is a real— I'm sorry, I don't want to confuse people. There is a real wallet app called Sparrow Wallet. It's available on Windows, macOS, and Linux, not iOS. Scammers said, oh, you know what's missing on iOS is something that looks just like Sparrow Wallet that can get people to give us their Bitcoin.
Leo Laporte [01:34:23]:
So they put it on the App Store. Apple did not stop it.
Devindra Hardawar [01:34:27]:
Brilliant, actually.
Allyn Malventano [01:34:28]:
I thought, I thought that was one of the things they were supposed to check for, honestly, in their app submissions.
Larry Magid [01:34:33]:
Is it not?
Leo Laporte [01:34:33]:
I think that's the contention of the lawsuit.
Larry Magid [01:34:36]:
Yeah, yeah, they screwed up.
Leo Laporte [01:34:39]:
The complaint alleges Apple violated California's Consumer Legal Remedies Act and other laws by failing to adequately review and monitor apps distributed by the App Store despite marketing it as a, quote, safe and trusted place to discover and download apps. That's what Apple says, right?
Allyn Malventano [01:34:54]:
Yeah, yeah, that's how they justify their crazy percent markup on app purchases and everything else, right? Like, that's—
Leo Laporte [01:35:01]:
The plaintiffs said, well, we trusted the The fake Sparrow wallet app that we downloaded. Uh, users found their accounts have been wiped out, their cryptocurrency assets transferred to the scammers' private wallets. The complaint says the fraud succeeded precisely because Apple exploited the consumer trust it had deliberately cultivated through decades of marketing.
Devindra Hardawar [01:35:24]:
Get 'em.
Leo Laporte [01:35:27]:
Well, you think, or will the judge say, well, Well, I don't know. I don't know. The judge might say this is Section 230, that Apple's not responsible, but Apple's saying we're gonna prosecute you.
Devindra Hardawar [01:35:39]:
The entire thing about the App Store is that Apple does such a great job of keeping it safe and secure, and they vet the apps. So yeah, I don't know if that'll stand.
Allyn Malventano [01:35:47]:
Yeah. You can't consider Apple App Store like an open marketplace, like where anybody can submit anything.
Leo Laporte [01:35:53]:
They're not just a publisher of other— they're not like Twitter where they're just saying, oh yeah, right.
Devindra Hardawar [01:35:57]:
It's a very curated marketplace.
Larry Magid [01:35:59]:
Yeah, yeah, yeah.
Leo Laporte [01:36:01]:
Well, good news, because the number of apps on the App Store are doubling this year because of vibe-coded apps. They're all AI apps now.
Allyn Malventano [01:36:08]:
Oh, great.
Leo Laporte [01:36:10]:
Yeah, what could possibly go wrong? Um, Apple's, uh, actually not releasing the facts, but Sensor Tower, an analytics company that does this— are they pretty accurate, right? Uh, I mean That's the only numbers we ever have.
Devindra Hardawar [01:36:27]:
Yeah.
Leo Laporte [01:36:27]:
Uh, last year the number of new apps in the App Store grew 30% to about 600,000. In the first half of this year, new apps doubled in the first half to 560,000. So almost, almost the same number in the first half of this year as all of last year.
Allyn Malventano [01:36:45]:
I would ask how Apple's even keeping up, but of course it's—
Leo Laporte [01:36:48]:
Well, that's why I think that's why stuff sneaks through, honestly, is that they just don't have the They just can't.
Allyn Malventano [01:36:54]:
They're right.
Leo Laporte [01:36:56]:
So they probably shouldn't—
Devindra Hardawar [01:36:57]:
Knowing them, they wouldn't be leaning on too much AI for the actual vetting. Like, it's— from what I've heard from developers, there's a lot of human interaction with the review board.
Leo Laporte [01:37:05]:
I think you have— there has to be. Yeah, but they miss stuff. For instance, uh, one of the scams— there are a number of apps that if you install them outside of Brazil look like kind of dumb game apps, but if your IP address is from Brazil, they turn into gambling apps, which Apple bans. But, you know, the Apple testers aren't saying, well, let's— what happens if we use a Brazilian IP address? They don't— you know, there's only so much you could test, I guess. If now, if they had Alan Malventano there, he could look at the—
Allyn Malventano [01:37:33]:
Just reverse engineer it.
Leo Laporte [01:37:34]:
Yeah, and you could say, oh, I know what this app does.
Allyn Malventano [01:37:37]:
I'm kind of surprised that— so I've never submitted an app for— like, I'm surprised that you don't have to submit your source. Like, I would have thought that you had to do that.
Leo Laporte [01:37:45]:
That's interesting. Yeah, I don't think you do. Yeah. You don't have to, Darren, right? Darren has written— Darren released— Darren Oakey in our club, 10+ handwritten apps back in the day. Now he's released lots of vibe-coded apps. He says his vibe-coded apps are so much better. How many? Okay, now I'm gonna have to go. He's got a link to all of the— is all of his iOS apps.
Leo Laporte [01:38:10]:
Darren, I trust you. Darren's amazing. Apps by Darren. So these are all vibe-coded, Darren? OL Reader, AI Narrated EPUB. Look at all this. OL DJ, OL Bridge, OL Four, OL Sudoku. Once you— these are better than your hand-coded apps. Darren's a professional coder.
Leo Laporte [01:38:32]:
Uh, how hard is it to get these on the App Store, Darren? Do you— you don't submit source code?
Allyn Malventano [01:38:36]:
There you go, we can find out. Yeah, I'm assuming you don't Probably. No, like, I'm sure you don't, but I can imagine.
Leo Laporte [01:38:45]:
It's easier than it used to be, which maybe is the wrong direction.
Allyn Malventano [01:38:48]:
Hmm. Well, I mean, there's— it's there. I mean, there's a— there's an inherent conflict, right? Because like, this is how Apple makes money on the App Store. One of the ways to make money.
Leo Laporte [01:39:00]:
That's a very good point.
Allyn Malventano [01:39:02]:
Right? So what—
Leo Laporte [01:39:03]:
the more apps, better for Apple.
Allyn Malventano [01:39:04]:
More apps equals more money, right?
Leo Laporte [01:39:05]:
Like, why would you not Yeah, well, they only make money on apps that cost money.
Allyn Malventano [01:39:11]:
True.
Leo Laporte [01:39:12]:
I'm sure they want more apps.
Allyn Malventano [01:39:14]:
Well, if it's, if it's ad-supported, they make a cut of that too.
Leo Laporte [01:39:18]:
Oh, that's right, they get a cut of the ads.
Larry Magid [01:39:20]:
Don't they get a cut of any purchases you make on the app as well?
Leo Laporte [01:39:22]:
And any purchases, in-app purchases, in the big bucks. Actually, most apps are free.
Larry Magid [01:39:30]:
Well, no, it's true. I mean, one of the things that, you know, historically I used to, for example, feel good about downloading software from download.com, which was run by CNET.
Leo Laporte [01:39:39]:
Oh, I remember that.
Larry Magid [01:39:41]:
Oh yeah. And you know, most of the time this stuff was good, but not 100% of the time.
Allyn Malventano [01:39:45]:
No, not anymore.
Devindra Hardawar [01:39:46]:
Yeah. I, I have had conversations with executives at CBS and they're like, oh yeah, we, we stuck so much stuff into download.com.
Leo Laporte [01:39:55]:
They admit it.
Devindra Hardawar [01:39:56]:
Oh man, that's amazing.
Leo Laporte [01:39:57]:
Well, download.com, pretty soon you started to have to download a downloader, which had other apps that it would install. I mean, it became—
Larry Magid [01:40:04]:
yeah, but it's still safer than the unregulated web. At least you felt there was someone looking over the shoulder. But yeah, you know, again, it was a false sense of security.
Devindra Hardawar [01:40:14]:
It was, it was, because those execs like totally knew that they had this platform where they could just like sneak a little.
Larry Magid [01:40:20]:
Are you saying when CBS bought CNET?
Leo Laporte [01:40:21]:
Yeah.
Larry Magid [01:40:21]:
So I was—
Benito Gonzalez [01:40:22]:
so this is Benito.
Benito Gonzalez [01:40:23]:
I was, I was working there.
Benito Gonzalez [01:40:25]:
I was working there when that buyout happened, and I saw that happen in real I saw the enshittification of download.com happen in real time, and it was really, really sad to watch because I was a heavy user of download.com.
Leo Laporte [01:40:37]:
Same. Sure, we all were back in the day. Yeah, yeah. But I did start warning, uh, my listeners on the radio show pretty early on, don't stop, don't—
Devindra Hardawar [01:40:45]:
no.
Leo Laporte [01:40:45]:
And if any download site asks you to download an app to make your download go faster, just say no.
Larry Magid [01:40:51]:
Even if I had known that, I would have warned them on my CBS radio show.
Leo Laporte [01:40:56]:
You should have listened to my show, Larry, and you would've known it.
Devindra Hardawar [01:41:00]:
I blew it.
Leo Laporte [01:41:02]:
I blew it. I have to say, I miss doing the radio show because that was the— unlike the shows we do here, which are aimed at enthusiasts and assume that you understand what we're talking about when we talk about 1-bit quants, the radio show was really designed for regular people. It was a bridge to regular people to help them understand it.
Devindra Hardawar [01:41:20]:
This is the perspective of a lot of my reporting too. So this is why I'm very much Like, I look at what regular people are going to be doing with AI and I'm, I am a little worried.
Larry Magid [01:41:29]:
So, oh, speaking of regular people, I have a question. Is, is iOS 27 beta safe for regular people yet? Absolutely.
Devindra Hardawar [01:41:36]:
Right now the, the first public, uh, beta is out.
Allyn Malventano [01:41:39]:
Yeah, the public's out.
Leo Laporte [01:41:40]:
Always wait for the public, but this one's very stable. I've got it on everything. On Golden Gate. Now you, when you press the button for Siri, it's this pretty little, let me see, I guess I have to unlock it.
Devindra Hardawar [01:41:50]:
The thing at the island.
Leo Laporte [01:41:51]:
Yeah. It's the pretty little island. And Siri's a little bit smarter than she used to be. You can actually ask her questions about your text messages and your emails and stuff, and she knows all about it. Is anybody—
Allyn Malventano [01:42:00]:
This answer's gonna be interesting.
Larry Magid [01:42:01]:
You know who's really getting funny? I'm gonna— the product that Amazon has that runs on the Echo, and I won't use the word because if I use the word, everybody's Echo devices will light up, has gotten so silly with the Plus version of it. Every time I ask—
Leo Laporte [01:42:16]:
Oh, you gotta choose the right personality. You have the sassy personality.
Larry Magid [01:42:20]:
I don't know what I've got wrong with it.
Leo Laporte [01:42:21]:
No, no, no, look, there are, there are, there are different personalities. There's one personality that will like talk back to you, but we'll say, hey, hey, A-word, set a timer for broccoli, and it'll go, timer for broccoli set. That's gonna be a delicious meal.
Larry Magid [01:42:37]:
Yeah, I know, it drives me nuts. I'll ask it to write a routine and it will explain to me what routines are, and I say, well, I wouldn't be asking you to write a routine if you didn't know what a routine was. Just write it. Yeah.
Devindra Hardawar [01:42:48]:
Well, or like when you ask one question, it's like, by the way, would you like to add this to your—
Larry Magid [01:42:52]:
That I hate.
Leo Laporte [01:42:52]:
That's an invocation. I'm like, stop.
Devindra Hardawar [01:42:54]:
I did not ask you for this. Quiet down.
Leo Laporte [01:42:57]:
I think unfortunately that most people's experience of AI is that kind of thing. And I think that's why there is kind of this general feeling of why are we building all these data centers so that it can sass me about my broccoli? This is a waste of money.
Devindra Hardawar [01:43:10]:
Yeah.
Larry Magid [01:43:11]:
And ask you questions on, like, I'll ask it a question once and it'll Well, are you going to make that tonight?
Allyn Malventano [01:43:17]:
Do you care?
Larry Magid [01:43:18]:
So, Larry, if anybody at Amazon really cares—
Leo Laporte [01:43:20]:
if this were the radio show, uh, we got Larry on the line from, uh, the East Bay. Uh, Larry, there is a setting in your Alexa app that you could change the personality of Alexa Plus and just change it to like the— I think there's concise or something like that where it's just like business.
Devindra Hardawar [01:43:35]:
I know, but the point is still the Amazon like alerts.
Leo Laporte [01:43:39]:
I turned on sassy and I regret it immediately.
Devindra Hardawar [01:43:43]:
I mean, I want a sassy robot, but I also—
Leo Laporte [01:43:45]:
well, who doesn't?
Devindra Hardawar [01:43:46]:
At what cost?
Leo Laporte [01:43:47]:
At what cost?
Larry Magid [01:43:49]:
I don't mind sassy, but what I do mind is when AI wants to be my friend. Didn't we say we're not going to talk about AI? Here we are.
Leo Laporte [01:43:55]:
We can't, we can't stop. Can't stop.
Larry Magid [01:43:59]:
Maybe we're all AI. Maybe, maybe listeners need to be informed.
Allyn Malventano [01:44:03]:
We still haven't gotten to Leo's. Leo wants to do that story about the DRAM and the SSD prices.
Devindra Hardawar [01:44:07]:
Okay.
Leo Laporte [01:44:08]:
Let's talk about it. Now, let me tell you. People are upset because they blame AI for the fact— well, Western Digital and I think SanDisk said, we sold out our entire capacity for 2026.
Devindra Hardawar [01:44:22]:
Yeah.
Leo Laporte [01:44:23]:
AI manufacturers, uh, like Micron stopped selling to consumers and then said, and we're sold out too. And Micron says, and it's going to be till 2030 before you can get any. Alan, is that true? What's going on? Why aren't they building up capacity?
Allyn Malventano [01:44:42]:
I mean, the only thing I consider fortunate is that, uh, my company got out of client, but that was because of the post-COVID dip.
Leo Laporte [01:44:52]:
Solidime sells solid-state storage, right?
Devindra Hardawar [01:44:56]:
Yeah.
Allyn Malventano [01:44:56]:
So, so I previously worked on the client side of Solidime, and that folded, you know, 3 or 4 years back now. It was a while back.
Leo Laporte [01:45:04]:
Well, that's one thing that Micron, I think, said.
Allyn Malventano [01:45:09]:
Well, they got out of it. See, they got out of it right in the middle of the AI boom. So it was pretty—
Devindra Hardawar [01:45:14]:
Right.
Leo Laporte [01:45:14]:
They said, we're not going to sell. Crucial's gone. We're not going to sell.
Allyn Malventano [01:45:17]:
Yeah. Crucial's gone. And it was during the upswing and it was pretty clear that that was the reason. I think they even stated in their press release.
Leo Laporte [01:45:24]:
He even blamed Apple. He said, if Apple hadn't pushed us so hard on DRAM prices in 2024, We would've been able to build out capacity. But the problem with the RAM market is it's boom and bust. And when we're in the bust, we don't have the money to build up capacity. So when it's the boom, we can't keep up. So he blamed Apple, literally said they drove prices so low. Now that's BS, obviously.
Devindra Hardawar [01:45:50]:
Right.
Allyn Malventano [01:45:51]:
But there is a balance, right? So when you're in the middle of a bust, obviously you can't, you don't have a bunch of funds to reinvest into all of the infrastructure because You know, you can't, you can't, if you build it, they will come for too long in a row. Case in point, Intel Optane memory, of which I have a bunch on the shelf back there.
Leo Laporte [01:46:08]:
Right, which they killed. You worked at Intel?
Allyn Malventano [01:46:11]:
I worked at Intel during like in the heyday of that, and I was, before that, I was reviewing those products, right? And I was very enthusiastic about it at the time. But of course, if there's not enough demand for the thing and you keep building the thing, right, that's, that's a bad way to roll a business. So that's why you would have a company like Micron that, you know, maybe not because of Apple, but the argument does stand that, you know, when times are tough, you're not necessarily gonna have all the funds to invest in a bunch of capacity. And that combined with a sudden boom of a thing that has a crazy demand on both DRAM and storage, I mean, I think part of the— I think in another universe, like The fact that it bled all the way down to hard disks is what gets me, right? That there was— because a lot of the industry—
Leo Laporte [01:46:59]:
You mean not just SSDs, not just solid state, but spinning drives too?
Allyn Malventano [01:47:03]:
Correct.
Larry Magid [01:47:03]:
Yeah.
Allyn Malventano [01:47:03]:
Well, when you— well, you just said Western Digital and SanDisk, like they were talking about hard drives they're sold out of, right? And it was because the industry was moving towards solid state little by little and off of spinning rest, as I love to refer to it. But, you know, that whole thing turned back on its head when all of a sudden the supply was dried up for SSDs seemingly overnight, right? Then people are like, well, where am I gonna put all of our stuff? Well, if not on solid state because the price is insane, then we have to fall back to hard disk. So it puts a squeeze on the hard disk supply as well, which was already kind of winding down.
Leo Laporte [01:47:39]:
Yeah.
Allyn Malventano [01:47:39]:
Somewhat because—
Leo Laporte [01:47:40]:
I haven't bought a spinning— the only thing I buy spinning drives for is my NAS because of capacity. I haven't bought a spinning drive in ages.
Devindra Hardawar [01:47:46]:
Correct.
Allyn Malventano [01:47:46]:
And anybody that's trying to do that now, is experiencing great pain.
Leo Laporte [01:47:51]:
Yeah, they've tripled in price, right? In some cases, 6 times more. Framework, I think, is saying 6 times more for SSDs.
Devindra Hardawar [01:47:59]:
Is this worth it, Leo, for you to play with your agents?
Leo Laporte [01:48:02]:
Like, yes, and I don't care if you don't have water for your gosh darn golf courses. I don't care. Let's drain the Colorado nuclear plant. I want to write a new first-person shooter.
Allyn Malventano [01:48:19]:
I don't get the water thing. Kind of boggles my mind. Like, the water consumption thing doesn't make any sense to me. If you look at like any of the, like any of the real numbers on the water thing, it's not just gobbling up all the water.
Leo Laporte [01:48:31]:
That's not how it works. And most, almost all the data centers now are recirculating the water. They're not just spraying it on the chips and letting it evaporate.
Allyn Malventano [01:48:39]:
No, it's not efficient to do it that way.
Leo Laporte [01:48:41]:
Yeah, they don't want to do that.
Devindra Hardawar [01:48:43]:
stories like the Google one, which they were clearly trying to hide the fact it was like Oregon or somewhere.
Leo Laporte [01:48:47]:
Right.
Devindra Hardawar [01:48:48]:
Like Google was building a data center and it was using a lot more water than the actual local municipality was. And it is true, like fresh water, which is what they still need, we don't have enough like for people. So the idea that we're spending, we're shoving all this and we're still dealing with the droughts in the West and the Southwest, like it does It seemed bad. It seemed like a bad choice.
Leo Laporte [01:49:11]:
I'm gonna say something here.
Allyn Malventano [01:49:13]:
You shouldn't get it from the same source. That's the key, is like, what source are you tapping into?
Leo Laporte [01:49:17]:
But I have to point out, and if you read Cadillac Desert, it's a great book, highly recommend it. We for decades built, overbuilt in areas where there isn't any water.
Allyn Malventano [01:49:28]:
Right.
Leo Laporte [01:49:29]:
And we dug canals, we made dams, we provided water. Vegas is in a desert. Most of Arizona is a desert. We intentionally built there and then built massive projects to provide water for these places. But as he points out in Cadillac Desert, it never was tenable and it was never going to last with or without AI data centers.
Devindra Hardawar [01:49:54]:
Humans making poor infrastructure decisions based on boom economics?
Larry Magid [01:49:58]:
Yeah.
Devindra Hardawar [01:49:58]:
I wonder if that applies right now.
Leo Laporte [01:50:00]:
The economics of it were that the government wanted to build these dams. I mean, it was a crazy rationale. It made no sense. And so, yeah, okay, there are places, there are deserts, and we are living in the deserts. This was going to happen whether or not we built data centers. I don't know why they keep building data centers in Arizona.
Devindra Hardawar [01:50:20]:
Yeah, people living in Colorado or along the Colorado River, like a lot of California, or at least Southern California, is also affected Absolutely.
Leo Laporte [01:50:29]:
Absolutely.
Devindra Hardawar [01:50:31]:
It's rough.
Leo Laporte [01:50:31]:
It is rough.
Larry Magid [01:50:32]:
We're not even talking about electricity. That's a whole other reason.
Devindra Hardawar [01:50:35]:
And then there's the electricity, and then there's the, uh, the output. Uh, let's just build a freaking data center with methane gas, right?
Leo Laporte [01:50:42]:
That's a mistake. I will grant you that Elon's nuts to do that, and it's highly polluting, and it's a terrible idea.
Devindra Hardawar [01:50:48]:
He doesn't care because he doesn't care about the community.
Leo Laporte [01:50:50]:
He doesn't care.
Devindra Hardawar [01:50:51]:
It's actually affecting—
Leo Laporte [01:50:51]:
Because he thinks we're all moving to Mars, so he doesn't care.
Devindra Hardawar [01:50:54]:
Well, no, he doesn't care because it doesn't affect people like him. So that's, that's the main thing for Elon.
Leo Laporte [01:50:59]:
I have to also point out we have a terrible grid, terribly planned because of our belief in free market, and, uh, we are suffering for other reasons than data centers. This is not a good— uh, our electrical system is awful. And there are countries like China that invested heavily in sustainable energy and solar power We're doing just fine.
Larry Magid [01:51:21]:
Yeah.
Devindra Hardawar [01:51:22]:
And by the way, it's not just 100 years.
Allyn Malventano [01:51:24]:
Yeah.
Larry Magid [01:51:24]:
It's not just AI. I hate to say it because we're streaming this program, but streaming—
Leo Laporte [01:51:29]:
Oh, do you like the internet? Guess where you're getting it from?
Larry Magid [01:51:33]:
I know.
Leo Laporte [01:51:34]:
A data center.
Larry Magid [01:51:35]:
But Netflix uses an enormous amount of electricity.
Leo Laporte [01:51:38]:
Of course. Everything we do today in the modern world uses data centers.
Devindra Hardawar [01:51:42]:
But at least those are, those are things that are actively being used by consumers and like delivers a certain sense of good to them, right? The thing is, we are building a crap ton of AI data centers for potential, potential, you know, AI usage by consumers. We're seeing some of it, but talk to Microsoft about how Copilot's doing.
Leo Laporte [01:52:00]:
You think Stranger Things and the dink are worth it, but creating, creating useful AI tools are not?
Devindra Hardawar [01:52:10]:
It's not at the same scale. Those data centers are certainly not the same scale, but also that was delivering a thing people actually wanted. And the thing I keep hearing from AI skeptics and like general people's like, who want— who was asking for any of this? And that's the thing, like I've noticed with, um, you know, Windows users as well. That was the main pushback of Copilot. And even Microsoft has scaled that back because—
Leo Laporte [01:52:32]:
But that was pushed down people's throats. I mean, that really— I don't—
Allyn Malventano [01:52:34]:
yeah.
Devindra Hardawar [01:52:34]:
Why is there a Copilot in Microsoft Paint? I don't— what if I don't wanna engage engage with this thing. And that's kind of the big problem there.
Leo Laporte [01:52:40]:
Well, I do not deny that the AI companies have really screwed themselves. I mean, look what Anthropic did with Mythos. The reason Mythos and Fable were banned is because Anthropic scared the hell out of everybody. And the government, Trump administration said, oh boy, we better stop this. These AI companies are fairly inept, partly because I think they have a worldview that doesn't match normal human.
Devindra Hardawar [01:53:06]:
Yeah, again, which is why I'm worried that these are the companies building these models.
Leo Laporte [01:53:11]:
No, that's why I use Chinese models, because those are good.
Devindra Hardawar [01:53:15]:
Those are, those are good, even though there are reports that some Chinese models are stealing from other models. And you know what, I don't care, because those existing models, they're all stolen.
Leo Laporte [01:53:23]:
Yeah, give me a break. Oh yeah, Anthropic says we will, we stole everybody's data, but you can't steal that back.
Devindra Hardawar [01:53:30]:
Have you guys had like a true AI believer on though, Leo? Like People who are like, we are building the AGI, we are building the second coming of God.
Leo Laporte [01:53:37]:
Oh no, I don't believe in AGI. No, no, no, no, no. I don't believe in that.
Devindra Hardawar [01:53:41]:
That's what all these companies— like you talk about Dario Amado, like that guy believes in insane things. Like this is a full-on religion for a lot of these folks.
Leo Laporte [01:53:50]:
I'll tell you what I think the right solution long-term is for open weight models, small language models that are purpose-built. If we built a model just for Larry's you know, CAT scan so that, you know, that these smaller models you could run locally, they'd be— they wouldn't require giant data centers. We also— I think the research they're doing on things like self-improving models, better post-training, there's all sorts of things. NVIDIA is building chips that are, you know, 3 or 4 times more efficient than the current crop of chips.
Larry Magid [01:54:22]:
Yeah.
Leo Laporte [01:54:22]:
There are absolutely— there'll be progress in a lot of areas. That isn't Anthropic and OpenAI.
Larry Magid [01:54:28]:
And also it can be more accurate. That, that data, that, that model designed to analyze CT scans doesn't— shouldn't have access to the entire internet.
Devindra Hardawar [01:54:37]:
No.
Larry Magid [01:54:37]:
So it can hallucinate. It should only have access to—
Leo Laporte [01:54:39]:
It doesn't need to know who won Wimbledon in 1925.
Larry Magid [01:54:41]:
Well, my point is it not, but it also doesn't need to know some crackpot who interprets CT scans as being something that they aren't.
Leo Laporte [01:54:48]:
That's an issue also.
Devindra Hardawar [01:54:49]:
You want to see that.
Larry Magid [01:54:50]:
If it only has vetted information, It's going to give you better information. Maybe not as much, but probably better quality.
Leo Laporte [01:54:57]:
Okay, that's the last AI story. Wait a minute, there's one more. That's what you think. Anthropic, that did admit, as every AI company has done, to ingesting a bunch of books, pirated. Uh, actually Anthropic did 2 things. Some of the books they bought, scanned, and then threw out. So that, you know, and those, the judge said, that's fine, that's fair use.
Larry Magid [01:55:22]:
Wait, the fact they bought one copy?
Leo Laporte [01:55:24]:
Yeah, they bought it. But the problem, the trouble they got was the pirated database. They've— the judge has now approved a settlement for $1.5 billion, um, to go to authors, like $3,000 a book. Did you apply for your, uh, benefits? I, uh, being an AI, uh, enthusiast, I did not.
Larry Magid [01:55:46]:
I did, but the problem is my books, I don't think anybody wants to pirate my Windows XP book.
Leo Laporte [01:55:52]:
Well, 2 of my books, I've written 13, 2 of them were in the database. Now, of the $3,000, had I applied for it, some goes to the publisher, some goes to my co-authors, you know, but it wasn't about the money. It was like, I'm glad that these crappy books I wrote 20 years ago have a new life in AI. I don't mind. Scan away. Anyway, there are plenty of authors who are going to get the money. 350 opted out hoping that they could sue further, but the rest are settling with Anthropic.
Larry Magid [01:56:25]:
I haven't heard a word. I don't even know if I'm getting a dime. I don't know if my book's worth anything.
Leo Laporte [01:56:29]:
Oh, you will. You will. How many books were in that database of yours?
Larry Magid [01:56:33]:
Oh, I can't remember. I've written 12. I can't remember how many are in there.
Leo Laporte [01:56:36]:
Yeah, 2 of my 13 were in there. Weirdly, though, like, not the bestsellers. It was just kind of random.
Larry Magid [01:56:43]:
And mine are so out of date. I haven't written anything in a while.
Leo Laporte [01:56:46]:
Yeah, well, that's more likely the older it is. There were half— more than half a million works in this pirated database. In a statement to Reuters, lead plaintiffs said the settlement Brings us one step closer to real accountability for Anthropic.
Allyn Malventano [01:57:07]:
No. Okay.
Leo Laporte [01:57:10]:
And puts all AI companies on notice. They can't shortcut the law or override creators' rights. Actually, the judge's decision was really, I think, a decision in favor of fair use.
Allyn Malventano [01:57:21]:
Yeah.
Leo Laporte [01:57:22]:
Uh, because, um, a lot of the infringement was deemed fair use. It was just this pirate database.
Larry Magid [01:57:28]:
I mean, if you think about it, how do we, how does any of us know anything? We all read.
Leo Laporte [01:57:32]:
Read the books, ingest it.
Larry Magid [01:57:34]:
Or newspapers or something. I don't have a bunch of reporters that work for me that are reporting to me what's going on in the world. I rely on— so if I tell you what's going on in the Iran War, it's because I probably read it in the New York Times or somewhere. It's not because I went to Iran to observe it. And certainly not from the government.
Allyn Malventano [01:57:53]:
All right.
Leo Laporte [01:57:55]:
That was the last AI story. Now we're gonna take a break and come We'll be right back to Devendra Hardiwar, Alan Malventano, who did not fully explain why we are running out of AI, I mean, memory and storage.
Allyn Malventano [01:58:06]:
It's just supply and demand. There's just too much demand for the supply.
Leo Laporte [01:58:09]:
Too much demand, too little supply.
Allyn Malventano [01:58:10]:
There's your simple answer.
Devindra Hardawar [01:58:11]:
Very simple, yeah.
Leo Laporte [01:58:12]:
That is simple. It's not a surprise. It's called economics.
Allyn Malventano [01:58:16]:
Yep.
Leo Laporte [01:58:18]:
But it is frustrating that, you know, I wish I had bought that 512 gigabyte Macintosh last year when I could have.
Allyn Malventano [01:58:26]:
Well, I mean, I'm putting off upgrades right now. Like, I can't— like, I literally just can't, can't afford the DRAM.
Leo Laporte [01:58:33]:
Can't do it.
Allyn Malventano [01:58:34]:
Yep.
Leo Laporte [01:58:35]:
And, uh, Larry Magid, connectsafely.org. How's the bowel? It's good, right?
Larry Magid [01:58:40]:
Bowel's doing great. Yeah.
Leo Laporte [01:58:41]:
But now your Achilles is bad. Now, yeah, I looked at it earlier.
Larry Magid [01:58:44]:
Emergency room. I tripped over a crack in a New York sidewalk. No, I thought I sprained my ankle, but it turned out I must have ruptured my Achilles.
Leo Laporte [01:58:51]:
Wait a minute, you injured yourself on the way to the emergency room for something else.
Larry Magid [01:58:54]:
Something else. I know it's a long, crazy story, but bottom line is I got to go into surgery on—
Leo Laporte [01:59:00]:
It's kind of appropriate. You should go see The Odyssey before you go in for Achilles surgery.
Devindra Hardawar [01:59:05]:
No Achilles in The Odyssey, but—
Leo Laporte [01:59:07]:
Oh, he's not in The Odyssey?
Devindra Hardawar [01:59:08]:
No.
Leo Laporte [01:59:09]:
Oh, that was the real heel.
Larry Magid [01:59:10]:
I'll tell you, the real heel. Damn it. Yeah, I mean, it's—
Leo Laporte [01:59:13]:
The Trojan Horse is from The Aeneid, so it's all, it's all mushed together.
Devindra Hardawar [01:59:17]:
It's an interconnected story. That was the MCU of its time.
Leo Laporte [01:59:20]:
Did you see it yet, uh, Peter Venner?
Devindra Hardawar [01:59:22]:
I've seen it 3 times.
Leo Laporte [01:59:23]:
3?
Devindra Hardawar [01:59:24]:
It's a very good movie.
Leo Laporte [01:59:25]:
Okay, okay.
Devindra Hardawar [01:59:27]:
We should talk about it because it's very anti-AI, and I think I kind of love it.
Leo Laporte [01:59:31]:
Christopher Nolan.
Devindra Hardawar [01:59:31]:
We're not supposed to talk about AI anymore.
Leo Laporte [01:59:33]:
No, but we could talk about it, like how bad it is. We can talk about that. That'll make—
Larry Magid [01:59:37]:
Peter Venner, that's all he talks about.
Leo Laporte [01:59:40]:
Uh, thanks for watching, Twit, regardless. So, uh, what was it, $265 million The Odyssey, uh, grossed in its first week.
Larry Magid [01:59:50]:
Huge.
Leo Laporte [01:59:51]:
Is it Christopher Nolan's biggest movie ever? I think it is. You're muted, I think, Devendra.
Devindra Hardawar [01:59:58]:
Yeah, can you hear me?
Allyn Malventano [01:59:59]:
Yeah. Okay.
Devindra Hardawar [02:00:02]:
His biggest movie ever.
Allyn Malventano [02:00:03]:
We can go on and on about The Odyssey now.
Leo Laporte [02:00:05]:
We are.
Devindra Hardawar [02:00:05]:
I mean, I can, I can.
Leo Laporte [02:00:06]:
It is 70mm IMAX. Did you see it, the film version of it?
Devindra Hardawar [02:00:10]:
I saw it 70mm normal, so like a normal-sized theater. I saw it 70mm IMAX at the Regal Mall of Georgia, one of the few theaters that can project that.
Leo Laporte [02:00:19]:
Yeah, there's only a few dozen, right?
Devindra Hardawar [02:00:21]:
Yeah, and I saw it in digital IMAX as well, which is still a very big screen but not like the full—
Leo Laporte [02:00:26]:
Which format did you prefer?
Devindra Hardawar [02:00:28]:
Obviously the 70mm IMAX, but it is worth a journey of your own to get to make it out to go see that. I wrote about that in Engadget. I've seen people drive several hours, you know, book a hotel stay just to get to a theater that plays The Odyssey. It is so wild Isn't it wild that there are so few true IMAX theaters around, and even, like, fewer that can do 70mm film projection?
Leo Laporte [02:00:52]:
Well, because it's so expensive, both for the theater and for the filmmaker. Only somebody like Christopher Nolan can really do it.
Devindra Hardawar [02:01:00]:
And this is the first film fully shot in 70mm IMAX.
Leo Laporte [02:01:03]:
So, Oppenheimer— I saw that Oppenheimer, which was on 70mm, wasn't fully 70mm.
Devindra Hardawar [02:01:07]:
It wasn't fully. Like, he's done more and more. It's started with The Dark Knight for him, but like he slowly added more and more.
Leo Laporte [02:01:13]:
The Dark Knight was weird because it would switch back and forth, and I didn't know why.
Devindra Hardawar [02:01:17]:
I mean, all the movies would kind of do that a bit. Um, he's the one that popularized IMAX as like a big thing because he's an analog film guy, and that was the best high-quality analog film, you know, technology available. But what's weird is that it's like, it's a dying technology, you know? Like they're not building more projectors, they're not building more cameras. So IMAX eventually will have to go fully digital on something, but nothing can really touch like the— it's like a 12 to 16K effective resolution.
Leo Laporte [02:01:44]:
You can tell.
Devindra Hardawar [02:01:46]:
It's very clear, like seeing it at the full projection, it is otherworldly because it's like you are being transported to another world. Everything looks obscenely clear. And there are a lot of like large vista shots.
Leo Laporte [02:02:01]:
That's what I didn't like about Oppenheimer. Oppenheimer, because I'm looking up Cillian's nose. I saw Oppenheimer on the 70mm film, and then I went home and watched it on my nice TV with a nice setup, and I far preferred it because I could follow the story on a normal-sized screen. IMAX is all about that, the picture, right?
Devindra Hardawar [02:02:22]:
It is. This is a much better use of IMAX than Oppenheimer was.
Leo Laporte [02:02:25]:
This is spectacle, isn't it?
Devindra Hardawar [02:02:26]:
Yeah, it was. That movie is mostly conversation. of people inside small rooms, you know, and this is— Yeah, that's not good. You get the full spectacle of it. And also you get like scenes with thousands of extras, scenes with thousands of people doing things that take real human effort. So obviously there's CG somewhere, but they didn't like go and like green screen large portions of this movie. It's not like filmed on the volume, like so much of the Marvel and Star Wars stuff these days. It feels real and authentic.
Devindra Hardawar [02:02:54]:
And it's a story that's relevant to our times, but also So man, it does make me think humans did this. This is a story we've been retelling for 3,000 years. Homer wrote down a version of it.
Leo Laporte [02:03:06]:
But did it really? You think it really happened?
Devindra Hardawar [02:03:09]:
No.
Leo Laporte [02:03:10]:
Oh, okay. So humans made the movie, not humans.
Devindra Hardawar [02:03:13]:
This is their Marvel Universe.
Leo Laporte [02:03:14]:
They're like gods and stuff in this. There's a Cyclops in it.
Devindra Hardawar [02:03:18]:
I mean, maybe the oral storytelling was like based on a city that could have been something like Troy. Like, there are probably bits of reality.
Leo Laporte [02:03:26]:
I've been to what they say, which Schliemann discovered and said was Troy. And if you— they say, well, we don't really know.
Devindra Hardawar [02:03:33]:
We don't know.
Leo Laporte [02:03:33]:
But they did put a giant wooden horse out front, so you think it's Troy.
Devindra Hardawar [02:03:39]:
But it is amazing to see, like, one of the oldest recurring stories we've had as a species in a crowded theater with the, like, latest technology that we have.
Leo Laporte [02:03:47]:
And it works.
Devindra Hardawar [02:03:47]:
This is a story that persists. And for a reason.
Leo Laporte [02:03:50]:
It worked.
Devindra Hardawar [02:03:51]:
It works. It absolutely works.
Leo Laporte [02:03:52]:
It's a good movie.
Devindra Hardawar [02:03:53]:
I saw it 3 times because first of all, the first time I saw it was for a press screening. I had to go back to see it in full IMAX. And I didn't like— I was not bored one bit. This is a 3-hour movie that like moves like gangbusters. So I went back to watch it again because my wife didn't get to see it. So I did that on vacation, even took some vacation time to see it again.
Leo Laporte [02:04:12]:
That's ironic because of course, as Burke says, it's a The whole story is the one Homer made up when he wanted to leave his wife. So, uh, honey, I was— I— there were sirens and cyclopses that just— I couldn't get back fast. So it took me a long time to get home. He was out drinking with the boys.
Devindra Hardawar [02:04:35]:
Come on! Definitely Leo. Everybody like, Leo, you can get to the Metreon, which happened.
Leo Laporte [02:04:39]:
I'm gonna go to the Met. That's where I saw Oppenheimer. You could, but this is the problem. Yeah, go look, it's sold. I mean, it's not sold out, but all that's left is front row or the— I mean, you— the problem with the IMAX is not only that there are only something like 20 70mm film IMAXs, but there is only a small number of seats that you want, right? You want the middle of the middle, kind of.
Devindra Hardawar [02:05:01]:
You want the middle or the back, but it depends on the theater. Like, I was front— I was second row at the Regal Mall in Georgia. And it was like, it was actually, I'm a front row kind of weirdo in general. Like I love to be up close whenever I can for any theater, but even for that one, like you could still see the entire picture and the way Nolan frames it is that it's really the center of the frame that he focuses on. 'Cause that, that's what they crop into for the smaller film formats.
Allyn Malventano [02:05:24]:
Right.
Larry Magid [02:05:24]:
Right.
Devindra Hardawar [02:05:24]:
So if you're looking there and then peripherally, oh, there's a huge sky or there's more sea below, you don't have to look at everything all at once. So even, even if you're close, you'll get a good good experience there, I think.
Leo Laporte [02:05:35]:
All right. I, I'm just waiting until I, I don't have to sit in the front row because my neck, it's like—
Devindra Hardawar [02:05:42]:
I would also recommend, um, keeping it like, look daily for tickets, like for a day you're free, because people cancel tickets and something, sometimes things will pop up.
Leo Laporte [02:05:49]:
I, I was thinking of going like, uh, at 8 in the morning, but even then there's no seats.
Devindra Hardawar [02:05:55]:
There are midnight screenings, there are 3 AM screenings, and they're all being sold out.
Leo Laporte [02:05:59]:
Yeah, you can't, because everybody's got— there's— I mean, they're really only probably in each theater maybe 100 good seats.
Larry Magid [02:06:07]:
Is this going to be the best box office of all times?
Devindra Hardawar [02:06:10]:
It sounds like it.
Leo Laporte [02:06:12]:
$265 million in the first week.
Devindra Hardawar [02:06:13]:
Yeah, yeah.
Larry Magid [02:06:14]:
And there are—
Devindra Hardawar [02:06:14]:
the IMAX tickets are so much more expensive, and I think like it seems like people are really digging this. I don't— I have no idea why Oppenheimer became such a phenomenon. I thought that movie was fine, but not like the billion-dollar You know, blockbuster.
Leo Laporte [02:06:27]:
It was fine for home, let's put it that way.
Devindra Hardawar [02:06:30]:
Yeah, yeah.
Larry Magid [02:06:30]:
I wonder how long before it streams. Be quite a while, I'm sure.
Leo Laporte [02:06:34]:
Well, will you even want to watch it at home?
Larry Magid [02:06:36]:
Yeah, that's something you'd want to watch in a stream.
Devindra Hardawar [02:06:38]:
Yeah, it's a great movie, so like you would have a good time watching at home, but the whole point is to be, I think, overwhelmed by it. And also like, hey, when last have you been to a fully jam-packed theater filled with people like enjoying a story that feels so generous generationally important, and the angle he takes here is also very much like relevant to today. You can totally read like today's headlines into how he frames The Odyssey.
Larry Magid [02:07:03]:
The movie theater in Mountain View that I go to is often empty, so I'm sure—
Leo Laporte [02:07:07]:
Well, this is what I thought— movies were dead. Our local little local theater in Petaluma, nobody was ever there. There were rats. It was terrible. But they put in an IMAX, not the big IMAX. You know, IMAX could be— there's a lot of different things IMAX could be, but it was a— it's a giant screen. Yeah, they put in a weird one called ScreenX where the walls become part of the screen.
Devindra Hardawar [02:07:31]:
I hate ScreenX, but it's a thing.
Leo Laporte [02:07:32]:
I know, it's weird. And then they put in, you know, the stadium seating and the trays and the things, and they serve liquor and, you know, they come around with food. But I— now they're sold out.
Larry Magid [02:07:46]:
Good.
Leo Laporte [02:07:47]:
And it's young people. Young people have always been the main audience for movies, I think, but it's young people and they want to go to the movies again, which I think is very interesting.
Devindra Hardawar [02:07:55]:
And it's the young people who are also like not really getting into AI as much as I think a lot of these companies want to.
Leo Laporte [02:08:01]:
So there's—
Larry Magid [02:08:01]:
I got my girlfriend to hold my hand when I was a teenager.
Leo Laporte [02:08:06]:
Larry, did you do the stretch? And then you just put your arm around her very, like, suavely and coolly.
Larry Magid [02:08:14]:
I'm not gonna tell you or AI what I did.
Allyn Malventano [02:08:17]:
I'm sorry.
Larry Magid [02:08:19]:
I do.
Leo Laporte [02:08:20]:
Yeah, you know what? Yeah, yeah.
Devindra Hardawar [02:08:22]:
You have to tell the AI. You have to give tips. Yeah.
Leo Laporte [02:08:26]:
Uh, it was definitely a few firsts.
Larry Magid [02:08:29]:
No, but the movie theater was— that is the place we dated. That, that, that and backseat of the car. I mean, yeah.
Leo Laporte [02:08:35]:
Yeah, so they've, uh, the gross, uh, for the first 2 weekends and week now is $286.3 million. Uh, they made it, they had another great weekend so far. Uh, they're making, uh, big bucks.
Devindra Hardawar [02:08:50]:
I think people are going back too because like once, once you see it, if you vibe with that movie, the experience is worth—
Larry Magid [02:08:55]:
Well, that's just—
Leo Laporte [02:08:57]:
I could see how it'd have a very strong opening, but I'm, I'm, I was thinking The Odyssey, are— it's 3 hours. Are people really— is this what people really want to see. I thought this might not be mass appeal, but I guess it is.
Larry Magid [02:09:09]:
Is it worth rereading the book before you see the movie?
Leo Laporte [02:09:12]:
In case you forgot what happened?
Larry Magid [02:09:14]:
It's been a long time since I read that book.
Devindra Hardawar [02:09:16]:
I spent a large portion of time in high school, like translating the Aeneid and the Odyssey.
Leo Laporte [02:09:22]:
From the original Greek?
Devindra Hardawar [02:09:24]:
I did the Aeneid from Latin, and we did portions of the Odyssey in Latin, not from Greek. So, it was a Latin class. So, it was not for But it was still like, we spent a lot of time studying the text and everything. And that is— there's a reason that story has stuck around, because I think humans can read a lot into it. And certainly in our society today, it is certainly saying a lot about like the state of the world, which is surprising that no one found a really relevant way to frame that story.
Leo Laporte [02:09:48]:
It was a great way for me to fall asleep in college. And all I remember is something about a wine-dark sea, like 100 times.
Devindra Hardawar [02:09:56]:
Yeah. There is that newer translation by Emily Wilson. I have not read that one.
Leo Laporte [02:10:00]:
Apparently, that's a very good translation. Actually, I have that. I bought that because I've read that it's a very— but it's controversial.
Devindra Hardawar [02:10:06]:
That's what inspired him to do this, I believe.
Leo Laporte [02:10:08]:
Oh, is it? Yeah, it's very controversial because it is not a word-for-word. It's more poetic. But that's maybe what we need.
Devindra Hardawar [02:10:17]:
That's how the story was told.
Allyn Malventano [02:10:18]:
It was oral.
Devindra Hardawar [02:10:19]:
We shouldn't— yeah.
Leo Laporte [02:10:20]:
It was an oral tradition.
Devindra Hardawar [02:10:21]:
Yeah.
Leo Laporte [02:10:23]:
Well, there's our movie review for the show. Aren't you glad we didn't cover any AI?
Devindra Hardawar [02:10:29]:
Well, again, anti-AI, because man, the human effort, the human power in that. Somebody released like an AI Odyssey movie ahead of the actual Nolan movie.
Leo Laporte [02:10:38]:
Elon did, and it's terrible.
Devindra Hardawar [02:10:42]:
Well, he announced—
Leo Laporte [02:10:42]:
Oh no, he tweeted it. He retweeted it.
Devindra Hardawar [02:10:44]:
He tweeted that something will happen, and we know what Elon means when something will happen, probably never will. It will happen in our lives. Some other guy, some other like indie director had produced like—
Leo Laporte [02:10:54]:
Yeah, yeah, yeah, I forgot. We watched a little bit of it.
Allyn Malventano [02:10:56]:
It's—
Leo Laporte [02:10:56]:
yeah, maybe I'll show a little bit of it. It's so awful.
Devindra Hardawar [02:11:01]:
It astounds me, like coming at this from like— I'm also like a part-time movie critic, so like looking at this from the way the tech companies are presenting entertainment, it's like, have you guys ever seen a movie? Do you know what people actually want? The— when something looks cold and dead, it really turns us off.
Leo Laporte [02:11:17]:
It does. It's the, it's the, it's the dead eyes that kind of give it away.
Devindra Hardawar [02:11:22]:
I think this is, uh, this is the Elon Musk one.
Leo Laporte [02:11:25]:
This is the one he retweeted.
Devindra Hardawar [02:11:27]:
He retweeted.
Leo Laporte [02:11:28]:
From Heavy.
Devindra Hardawar [02:11:29]:
Even though people are like, um, you know, the, uh, the ship—
Leo Laporte [02:11:31]:
It was built on grok. That's why Elon retweeted it.
Allyn Malventano [02:11:33]:
It's built on grok.
Devindra Hardawar [02:11:34]:
But, uh, ships— ship wheels are a fairly recent seafaring invention. So if you're gonna be accurate to the Odyssey—
Leo Laporte [02:11:41]:
He's eating a potato chip. Wait a minute, or is it a Pringle? Wait a minute, let's see what he's eating here. Hold on a second. He's gonna— she's gonna feed— I think that's a Pringle. I don't think they had Pringles. Oh, heavy, heavy. Now he's drinking from the cup while the girls plant tomatoes, apparently. I don't know what's going on.
Leo Laporte [02:11:59]:
At least he has 5 fingers. That's good. We've gotten past that. Will Smith can eat spaghetti. The AIs think that acting involves a lot of pauses, a lot of dramatic sighs.
Devindra Hardawar [02:12:14]:
Yep.
Leo Laporte [02:12:14]:
A lot of long staring into the distance. There's a hole in the rock.
Devindra Hardawar [02:12:22]:
Go see the real Odyssey.
Leo Laporte [02:12:23]:
Go see Christopher Nolan's Odyssey in a theater.
Devindra Hardawar [02:12:25]:
Actually, Nolan intended—
Leo Laporte [02:12:27]:
the fact that Elon thought, oh, this is so good, and posted on Twitter really underscores the gulf between the AI bros and humans.
Devindra Hardawar [02:12:37]:
For sure.
Leo Laporte [02:12:39]:
ISP— I love this headline from Ars Technica. ISP's long nightmare of having to list all the fees they charge is finally over. Thank God the FCC— remember that brilliant idea the Biden administration had, the, the nutrition labels for your ISP so you'd know exactly What'd it cost? Well, the FCC today voted to eliminate those rules so that ISPs don't have to go through all that trouble.
Devindra Hardawar [02:13:06]:
So those poor ISPs can just like, you know, just give you a number, a simple number.
Leo Laporte [02:13:11]:
That's all you need.
Devindra Hardawar [02:13:12]:
Yeah, just pay it.
Leo Laporte [02:13:14]:
The burden of listing fees. Broadband lobby group U.S. Telecom complained to the FCC, providers must create and update hundreds of different labels to account for geographic variability ensure their systems properly queue the label specific to the proper location when the customer inputs their address. You don't think they do that when they send you the bill?
Allyn Malventano [02:13:35]:
Right.
Leo Laporte [02:13:36]:
Oh, it's so much extra work. No, you're already gonna do that when you bill me, for crying out loud. Well, the FCC, we know who they work for. They've always worked for— uh, the government lawyers now say the Trump administration, uh, kept can use TikTok again. So that's—
Larry Magid [02:13:57]:
Of course, it's run by his buddy.
Leo Laporte [02:13:58]:
On their government phones.
Devindra Hardawar [02:14:00]:
Yeah. Yeah.
Larry Magid [02:14:01]:
Why not?
Devindra Hardawar [02:14:01]:
Because, uh, yeah, it was, uh, TikTok was incredibly dangerous.
Leo Laporte [02:14:04]:
So dangerous.
Devindra Hardawar [02:14:05]:
Which nobody has revealed to us just yet. Alan, I don't know if you have any thoughts about that one, but I, I'm still struck by like, we did this whole thing based on nothing.
Allyn Malventano [02:14:16]:
Well, I mean, I can, I can offer. So like, I don't understand why government phones are doing any service of even similar to TikTok.
Leo Laporte [02:14:24]:
You shouldn't have TikTok on your government-issued phone.
Allyn Malventano [02:14:28]:
That's my, that's my point. Like, it's the type of— so like, it's nothing about TikTok in particular. That's just—
Leo Laporte [02:14:32]:
You shouldn't have Instagram on it. You shouldn't have Facebook on it.
Allyn Malventano [02:14:35]:
Right, exactly. That's not what the government phones are for. Like, just—
Leo Laporte [02:14:39]:
Don't you have your own gosh darn phone?
Allyn Malventano [02:14:41]:
Yeah, exactly. That's— I mean, that's what you're supposed to use your own phone for, that kind of stuff.
Leo Laporte [02:14:45]:
Incidentally, it's still against the law. It's not like the law went away. The Trump administration said you can ignore the law.
Larry Magid [02:14:52]:
Yeah, well, that's not uncommon.
Devindra Hardawar [02:14:53]:
Yeah, that's typical.
Larry Magid [02:14:55]:
Yeah, yeah.
Allyn Malventano [02:14:57]:
I mean, there's— I mean, the simple concept, defense in depth. Like, you don't— you don't have government phones with— like, that sort of stuff shouldn't be on a government phone that you would assume is connected to a government closed network. So that means that the poor folks up in Norfolk, Virginia Uh, sitting at the Navy Cyber Defense Operations Command, for example, like where I used to work back when I was calling into the show. You were?
Leo Laporte [02:15:25]:
You didn't tell me that.
Allyn Malventano [02:15:26]:
Oh yeah, that's where I was working in the early days when I did the Drobo review and took the Drobo apart. I was on the base in Little Creek, Virginia.
Leo Laporte [02:15:34]:
Did we know that?
Allyn Malventano [02:15:35]:
Yeah, I think I said, I might've said something. Um, yeah, I was, it was in my barracks room on the base, but like, that's awesome.
Leo Laporte [02:15:43]:
That's so cool.
Allyn Malventano [02:15:44]:
But, uh, yeah, like those poor people are going to have to try to defend the entire like Navy intranet or DOD intranet from people trying to use other services that like, I mean, it's hard enough to defend networks as it is with all sorts of adversarial kind of stuff.
Leo Laporte [02:16:00]:
I gotta have my TikToks.
Allyn Malventano [02:16:03]:
Yeah. I feel for those guys trying to, trying to keep the network.
Devindra Hardawar [02:16:07]:
Maybe they should keep the network safe by just doing, you know, war planning on Signal on their personal—
Leo Laporte [02:16:11]:
oh, there you go.
Devindra Hardawar [02:16:12]:
So much easier. Just put the Navy.
Leo Laporte [02:16:15]:
The problem was they weren't using Signal. They were using a third-party Signal-compatible app.
Devindra Hardawar [02:16:21]:
Yep.
Leo Laporte [02:16:22]:
Oh yeah, which made it worse. It wasn't secure.
Allyn Malventano [02:16:25]:
Yeah, yeah, known to be insecure. And there were plenty of black hat talks on it after the fact too, right?
Larry Magid [02:16:34]:
My question is, who do you trust least, the Chinese or the elephants, when it comes to your information? I mean, I—
Leo Laporte [02:16:40]:
Who are the elephants? What do you mean? Yeah, the Ellison. I thought you said elephants.
Devindra Hardawar [02:16:46]:
Elephants.
Larry Magid [02:16:48]:
Elephants have memory. No, Larry Ellison owns TikTok.
Leo Laporte [02:16:51]:
All of these guys, you know, it's like their one interest is making more money, like they don't have enough. That's their one interest. And they're gonna use your information, they're gonna sell it. Of course they are.
Devindra Hardawar [02:17:02]:
But they also have like political, uh, you know, angles which are different than the Chinese government did. So that was my main worry, like, but even that was like Like, there was supposedly some big, you know, big reason TikTok had to be sold in the US, and we still don't know what that is.
Leo Laporte [02:17:16]:
Well, interestingly, a judge has halted the Paramount acquisition of Warner Brothers. The Trump administration approved it, but the state said, whoa, hold on there.
Larry Magid [02:17:26]:
You mean Bari Weiss isn't going to get to run CNN?
Leo Laporte [02:17:29]:
That's good news. You know, it's fun watching the CNN anchors get more and more outspoken as they thought, we're going down.
Larry Magid [02:17:37]:
We might as well.
Leo Laporte [02:17:37]:
We got nothing to lose.
Devindra Hardawar [02:17:38]:
Just tell people what's happening.
Leo Laporte [02:17:40]:
We might as well just go for it.
Devindra Hardawar [02:17:42]:
The Trump administration didn't just approve it, they like hand-holded this whole deal to make it happen when Netflix was the, the previous buyer.
Larry Magid [02:17:50]:
Well, did you hear what Trump said about CBS, about a regime change at CBS, an enemy of America, and now it's, and now it's run by good people? I mean, yeah.
Leo Laporte [02:18:01]:
Yeah, I noticed when ABC and NBC did not carry the presidential primetime address about how the election was stolen in 2020, uh, that he said they should have their licenses revoked, right? You didn't mention CBS.
Larry Magid [02:18:16]:
No, CBS is a friend of theirs now. Friend.
Leo Laporte [02:18:19]:
Yeah, friend of the family.
Larry Magid [02:18:20]:
He loves CBS.
Leo Laporte [02:18:22]:
All right, one more commercial, uh, and then we're— let's talk about, uh, prices, money, and, uh, other stuff. Actually, speaking of money, the AI industry has now spent $65 million on the campaigns in November. Wonder what they're lobbying for.
Larry Magid [02:18:44]:
Oh, I know.
Allyn Malventano [02:18:45]:
I thought you were going to say during this podcast.
Leo Laporte [02:18:47]:
Well, yeah, during this podcast they've spent $65 million. They want favorable regulation. They want favorable language. Meanwhile, the Trump administration is saying we need an AI kill switch. I bet you'd like that, Devinder, wouldn't you? There should be a big button in the White House that says, stop the AI.
Devindra Hardawar [02:19:03]:
I wouldn't want it in the White House.
Larry Magid [02:19:04]:
I wouldn't want it in today's White House. Can we put it somewhere else?
Leo Laporte [02:19:07]:
Where do you put it? Give it to Alan. Alan, would you like that button in your basement?
Allyn Malventano [02:19:11]:
I'll take the button. I got a button here somewhere.
Leo Laporte [02:19:13]:
Just don't put it next to the that was easy button and confuse the two.
Allyn Malventano [02:19:18]:
That's the button I was going to find.
Leo Laporte [02:19:20]:
I have one over here if you want. Uh, we'll have more in just a little bit. You're watching This Week in Tech. Devendra Hardwark, always good to see you. I'm glad to have you on to, to talk some sense into me.
Devindra Hardawar [02:19:35]:
I gotta be the crank now, Leo, so I'm gonna crank it up.
Leo Laporte [02:19:38]:
I just think it's so funny that the old guy is saying, I love the future, and the young guy's going, no, no, don't do it, don't do it, old man. You kids today.
Larry Magid [02:19:47]:
I went to a demonstration at Berkeley of undergraduates saying, we want to take control of our technology. We don't want these big companies.
Leo Laporte [02:19:54]:
No, the kids are really—
Larry Magid [02:19:56]:
yeah, yeah, yeah. It's the—
Devindra Hardawar [02:19:58]:
Slow tech. Slow tech is how we're going back to it.
Leo Laporte [02:20:00]:
Yeah, slow tech. Yeah, local tech.
Devindra Hardawar [02:20:03]:
Local tech.
Leo Laporte [02:20:04]:
Yep, farm grown. Uh, Alan Malfentano, he's down on the farm.
Devindra Hardawar [02:20:10]:
Look at that. Look at that.
Leo Laporte [02:20:11]:
Look at his shelf. Unbelievable. Unbelievable. You got everything, man. What's your favorite tchotchke there? What's the baseball? Is that a signed baseball?
Allyn Malventano [02:20:21]:
That's Smarter Every Day.
Leo Laporte [02:20:23]:
Oh, cool.
Allyn Malventano [02:20:24]:
This is my favorite tchotchke there.
Leo Laporte [02:20:27]:
Oh, that's my favorite too. I got one of those.
Allyn Malventano [02:20:29]:
There you go.
Leo Laporte [02:20:31]:
Lovely guy named Joe, uh, in Hawaii makes those. He's so cool. Uh, yeah, I have a bunch of those over here. They're off camera, you can't see them. They're good. I'm gonna give you this though. I want to get Joe to make this.
Allyn Malventano [02:20:45]:
Uh-oh.
Leo Laporte [02:20:46]:
Little badge.
Allyn Malventano [02:20:49]:
Works on my what?
Leo Laporte [02:20:50]:
He made this for me too, just like you.
Allyn Malventano [02:20:52]:
Yeah, yeah. Joe Noel.
Leo Laporte [02:20:53]:
He lives in Hawaii, but I want Joel to turn this into a wooden coaster.
Allyn Malventano [02:20:59]:
Works on my machine.
Leo Laporte [02:21:01]:
Every IT guy should have just, you know, have that just on their forehead.
Allyn Malventano [02:21:06]:
Is that, is that Is that particular one 3D printed? It looks—
Leo Laporte [02:21:09]:
Uh, it's, it's— what's weird is it's Velcro, so it's like, I don't know what it's—
Allyn Malventano [02:21:13]:
Oh, it's a patch.
Leo Laporte [02:21:14]:
I should put it on my coat.
Allyn Malventano [02:21:17]:
Yeah, I guess.
Leo Laporte [02:21:18]:
It's on my machine. Jeff Atwood gave me that, the, uh, Coding Horror guy, founder of Stack Exchange. I love that. I'm gonna get one of that. All right, we'll have more in just— oh, and I, I forgot to say, and Larry Magid is also here, the other old guy. When are you going for your surgery?
Larry Magid [02:21:34]:
Friday.
Leo Laporte [02:21:35]:
Oh, I hope— it's a simple outpatient procedure, right?
Larry Magid [02:21:39]:
It's an outpatient procedure. The doctor is not nervous at all. Oh, good.
Leo Laporte [02:21:43]:
It's arthroscopic. They just put a little hole in there.
Larry Magid [02:21:45]:
Oh no, not when you're transplanting, not when you're grafting a tendon.
Leo Laporte [02:21:49]:
Where are they getting the tendon from?
Larry Magid [02:21:50]:
Either from my big toe or the side of my foot. He's not going to decide that till he opens it up.
Leo Laporte [02:21:55]:
When they have to open one side to put it in the other side.
Larry Magid [02:21:58]:
Yeah, I should have, I should have treated it I didn't, you know, hey kids, if you have a sore foot and you go to the doctor, ask him to check your tendon. Don't assume it's a sprain and don't walk 100 miles with a—
Leo Laporte [02:22:10]:
So you actually made it worse?
Larry Magid [02:22:12]:
Oh, much worse. I walked all over New York, Washington, Toronto, Montreal, you know, mild pain.
Leo Laporte [02:22:18]:
Getting old's a bitch.
Larry Magid [02:22:19]:
Minor little pain. You know, the thing about nobody took it seriously 'cause I wasn't in horrible pain.
Leo Laporte [02:22:26]:
Don't get old, kids. Actually, do get old.
Devindra Hardawar [02:22:27]:
No, don't get old.
Larry Magid [02:22:28]:
It's the alternative.
Leo Laporte [02:22:29]:
It's not that bad. Better to get old, get old, get old.
Larry Magid [02:22:33]:
But stay healthy as much as you can.
Leo Laporte [02:22:34]:
Stay—
Larry Magid [02:22:34]:
that's right. I was thinking, you know, I'm walking, thinking that's good for— walking's good for you.
Leo Laporte [02:22:38]:
Walking's good for you.
Larry Magid [02:22:39]:
I said, okay.
Leo Laporte [02:22:40]:
Old people are supposed to walk.
Devindra Hardawar [02:22:42]:
Yeah, you're supposed to walk.
Leo Laporte [02:22:43]:
My mom passed away a couple of weeks ago.
Larry Magid [02:22:45]:
Oh, I'm so sorry.
Leo Laporte [02:22:46]:
Oh, it's fine. She was 93. She had not a pain in her life. She felt great. She just went to sleep one morning. She got up, got showered, yard, took a little nap, never got up. And that's perfect. That's what I want.
Leo Laporte [02:23:01]:
93, not a pain, not a care in the world. She was a wonderful woman, very happy. But now this is what happens, Larry. People start dying around you.
Larry Magid [02:23:11]:
Well, you know, we talked a lot about John Dvorak, and I'm not that much younger than John at all. And it got—
Devindra Hardawar [02:23:18]:
I thought about it.
Larry Magid [02:23:18]:
I mean, I'm sad for him, but I'm also aware that—
Leo Laporte [02:23:22]:
We're all Well, you know, every day I asked my AI, I said, what should I do about this? It said, just live every day. And you know, the AI is never wrong. Our show today—
Larry Magid [02:23:34]:
No response to that one.
Leo Laporte [02:23:35]:
Our show today brought to you by— I don't know how the AI knew. It just said, you know what, Leo? Don't try to leave a legacy. Don't try to write some big story in the sky. Just be honest and truthful and live every day as if it's your last. I thought, That's probably pretty good advice. Uh, let's see, there's so many, uh, stories still. We only have a few minutes. Galaxy Fold Z— Galaxy Z Fold Ultra.
Leo Laporte [02:24:04]:
Are you excited about this, Devendra? Have you taken a look at the new— they had their Galaxy—
Devindra Hardawar [02:24:08]:
I've been on a beach for the past week, so—
Leo Laporte [02:24:10]:
Oh, you didn't have to do that?
Devindra Hardawar [02:24:12]:
Yeah, yeah, I didn't have to deal with this.
Leo Laporte [02:24:14]:
I wonder if it's a bellwether for Apple's Fold foldable, which we think will be coming out in a couple of months.
Devindra Hardawar [02:24:19]:
Well, Samsung has one now that is more like a little, um, square, I guess.
Leo Laporte [02:24:23]:
Or yes, they do. Don't they?
Devindra Hardawar [02:24:25]:
A little differently, which is kind of what Pixel—
Leo Laporte [02:24:27]:
Apple's is supposed to be, right?
Devindra Hardawar [02:24:28]:
That's what the rumor is. And that's what like the case leaks have shown. So I am, I'm all for that if Apple moves that way. At this, like the first couple generations of foldables were so flimsy and so expensive. And now I feel like they at least figured out how to build these things better and more.
Leo Laporte [02:24:45]:
We know Apple is getting its screens from Samsung, so I'm thinking maybe this is kind of a way to see kind of maybe what Apple's gonna look like.
Allyn Malventano [02:24:53]:
I don't know.
Leo Laporte [02:24:54]:
It's very thin, and the crease is better. Apple's been very worried about the crease. One of the reasons Apple didn't do a folding phone is the crease.
Devindra Hardawar [02:25:04]:
Apple would not live with, you know, producing something that had that So, right, I think that's a major reason.
Leo Laporte [02:25:11]:
Well, the price, $2,099 to start, and I bet you Apple's phone's going to be in that price range.
Allyn Malventano [02:25:20]:
Yeah, half of that's DRAM.
Devindra Hardawar [02:25:22]:
Given, yeah, given how much they've already, uh, jumped up prices, and the rumor is we're not going to see a base-level iPhone 18 this year either, that may wait till next year, right? Um, maybe around $2,000 for sure. I could see them going even higher because it's Apple, right? It's like a $2,500 smartphone. Who's gonna be the first to do it with a base price?
Leo Laporte [02:25:45]:
And I'm gonna have to buy it because—
Devindra Hardawar [02:25:48]:
You have to.
Leo Laporte [02:25:49]:
I have to.
Devindra Hardawar [02:25:50]:
I don't wanna.
Leo Laporte [02:25:53]:
I do have the Z Fold 7 and it's really thin. It's really nice. It's beautiful.
Allyn Malventano [02:25:58]:
But I Android.
Leo Laporte [02:26:01]:
Android. Yeah, Android.
Allyn Malventano [02:26:04]:
I like Android.
Devindra Hardawar [02:26:06]:
I—
Leo Laporte [02:26:07]:
well, I like Graphene. Have you tried— I put Graphene on my Pixel and I was really impressed. Now here's something to be aware of. Graphene has this, uh, PIN number that you can put on it that will erase the phone. So if you are, uh, being stopped by law enforcement And, uh, you don't want them to take your phone, you can— and they say, well, just unlock this for us. No, unlock it for us or you're going to airport jail.
Allyn Malventano [02:26:38]:
Okay.
Leo Laporte [02:26:40]:
And you use— and you say, here's my PIN, 1234, but it's that special PIN. It will erase it, except maybe you don't want to do that. Uh, there's a guy—
Larry Magid [02:26:53]:
It's probably backed up.
Devindra Hardawar [02:26:54]:
Helpful.
Larry Magid [02:26:54]:
Huh? It might be backed up.
Leo Laporte [02:26:56]:
Well, yeah, no, no, the whole idea is that you have, you know, at home or somewhere you have a backup.
Devindra Hardawar [02:27:02]:
Yeah.
Leo Laporte [02:27:03]:
There's a guy named Sam Tunic who was a big protester about Cop City, which is the— was this in Atlanta? I think it was. This is your neck of the woods, Devendra. The training center in Atlanta known as Cop City. He was trying to get in the United States. He had Graphene on his phone, which is a really nice privacy-forward operating system. In January, he was stopped at the border coming back at Atlanta's Hartsfield-Jackson Airport after vacationing in the Dominican Republic. He was already on a watch list because of his— merely because he was protesting Cop City, but he was on a terrorism watch list. So Border Patrol said, give me your phone, unlock it.
Leo Laporte [02:27:52]:
He gave them that special PIN. An agent took him to a room with a table, several chairs, and a camera and microphone. Several agents began a, quote, secondary inspection. They pretended they were looking for child sexual abuse material, but it was because he was a protester. He was a terrorist.
Devindra Hardawar [02:28:10]:
He would have been better off not even giving like any PIN. Right, because they can't—
Leo Laporte [02:28:14]:
He would have, because now he's in trouble for destroying evidence.
Devindra Hardawar [02:28:20]:
Well, he didn't do it.
Leo Laporte [02:28:23]:
Well, that's what we're gonna find out.
Allyn Malventano [02:28:26]:
Yeah, when they typed it in.
Leo Laporte [02:28:28]:
Yeah, yeah, well, that's what we're gonna find out, because, uh, he is now being prosecuted by the Department of Justice because— for destroying, uh, evidence. This will be the first time the law has been aimed at an operating system on Pixel phones. It's very interesting. Graphene says, well, we— you don't have to use that. We have other ways of, you know, Graphene is secure. It's going to be hard for them to get the information out of your phone. But if you unlock it, I think, you know, there is a—
Larry Magid [02:29:01]:
there are—
Leo Laporte [02:29:01]:
there's lots of stuff on Graphene you can have. Deniable, you know, storage and so forth.
Devindra Hardawar [02:29:06]:
Isn't the general advice though is that cops can basically push you to use biometrics to unlock your phone, but they can't do the PIN pad? At least that's what I've heard.
Leo Laporte [02:29:15]:
Because it's in your brain. Well, some courts have said that, but not all courts, and I don't think there is a universal doctrine on this.
Devindra Hardawar [02:29:22]:
Gotcha.
Leo Laporte [02:29:23]:
So he— they somehow, they convinced him to give them a passcode. The screen went blank, flashed several times, the phone appeared to restart. And now they're saying you destroyed the evidence and you're going to jail, not because of what we found in your phone, because what you did. So just a word of warning if you're planning on doing that. I'm going to bring, um, we're going to Black Hat, Alan. Are you gonna be at Black Hat this year?
Allyn Malventano [02:29:48]:
I don't usually. I mean, I should have went a bunch in the past, but I ended up not, back when I was in the Navy. But, uh, I mean, now I don't have a reason to go anymore, right?
Leo Laporte [02:29:55]:
Should I be nervous? Lisa's all worried. Because we're going to do, uh, we're going to do security now there and Windows Week.
Devindra Hardawar [02:30:00]:
Don't bring your agents earlier.
Allyn Malventano [02:30:01]:
Be careful, be careful what you bring.
Leo Laporte [02:30:03]:
Well, yeah, so she— so what should I, uh, an iPhone's all right, right?
Allyn Malventano [02:30:08]:
If you say so.
Leo Laporte [02:30:12]:
So, uh, some people have said turn off your Bluetooth, don't use Bluetooth, don't use Wi-Fi. Yeah, is that okay?
Allyn Malventano [02:30:21]:
I would say, I would say cell phone network only, but that's not safe either at Black No, because they put stingrays everywhere that look like real cell towers, but they're not, right?
Leo Laporte [02:30:31]:
Yep, they're man-in-the-middle attacks. So don't bring my phone at all?
Allyn Malventano [02:30:35]:
Well, I mean, you could bring stuff, just don't bring anything with information on it that you don't want someone else to see.
Leo Laporte [02:30:40]:
Bring my graphing device.
Devindra Hardawar [02:30:40]:
You have multiple devices, right, Leo? Like, bring a non-primary device without—
Leo Laporte [02:30:44]:
Yes, correct. All right, that's what I'll do.
Allyn Malventano [02:30:46]:
Yeah, I don't want a clean install. Yes, and not signed into anything.
Leo Laporte [02:30:52]:
Certainly, but I got to talk to my agents. I can't leave these guys alone. Well, who knows what they're going to do?
Devindra Hardawar [02:30:59]:
Open them up to a vector at Black Hat.
Leo Laporte [02:31:01]:
Awesome. Yeah, there you go.
Allyn Malventano [02:31:02]:
That's what you want, someone from Black Hat.
Leo Laporte [02:31:04]:
I did, I said, um, I did a security audit yesterday and they said, you don't have a password on this. You know, if somebody got in your network, they could wreak havoc. I said, oh yeah, nobody's gonna get in my network. They said, you don't have a password on it. They made me put, they made me put a password on it.
Devindra Hardawar [02:31:20]:
On what, the phone?
Leo Laporte [02:31:21]:
On my agent.
Allyn Malventano [02:31:22]:
No, on his, on his network.
Devindra Hardawar [02:31:23]:
Oh my God, my network agents are unpassworded?
Allyn Malventano [02:31:27]:
Yeah.
Larry Magid [02:31:30]:
Yeah. Okay.
Devindra Hardawar [02:31:33]:
Yeah.
Leo Laporte [02:31:34]:
All right, last thing I'll say, uh, how— okay, show of hands, how many of you love the cookie banner?
Allyn Malventano [02:31:41]:
The what?
Leo Laporte [02:31:43]:
You know, the thing when you go to a website, it says we use cookies, okay, opt out, opt No, I hate that. I love that.
Allyn Malventano [02:31:48]:
That's the most appealing thing ever.
Devindra Hardawar [02:31:50]:
But it's also, I kind of, I like what it represents, even though it's a bad implementation of it.
Allyn Malventano [02:31:54]:
Yeah. It's a bad implementation is the problem.
Leo Laporte [02:31:56]:
Cookies aren't, come on, third-party tracking cookies. Okay. But every website uses cookies, which means every website has to put up a banner.
Larry Magid [02:32:04]:
Yeah.
Allyn Malventano [02:32:04]:
Right.
Leo Laporte [02:32:05]:
Our banner on our, if you go to twit.tv, our banner says, yeah, we use cookies. What about it?
Devindra Hardawar [02:32:14]:
It was a simpler time when we worried about cookies and not about what the AI agents can do to our data.
Leo Laporte [02:32:19]:
You can fingerprint a browser. You don't need cookies. Cookies are old school, man. Anyway, the EU is finally thinking of doing something, putting a setting in the browser that you set and you'll never see another cookie banner. The tracking industry is pushing back. And so far they've been successful. So I'm pushing everybody to this site. Kill the cookie banner.
Leo Laporte [02:32:45]:
Because it's not just the EU, right? It's killthecookiebanner.eu, but it's not just the EU because it's an EU GDPR rule. No, actually predates GDPR because it's a rule in the EU. Every website in the world has it, including ours, because I don't want to say, well, if you're in the EU, you can't visit our website, so we put up a cookie banner.
Allyn Malventano [02:33:03]:
Right.
Leo Laporte [02:33:04]:
Killthecookiebanner.eu. Take action, sign the petition. Uh, cookie banners bad. There's a better way, and I would love it if I didn't have to click a cookie banner. You know what, I actually— you can, you can use uBlock Origin and just— it's one of the annoyances you can turn off. No more cookie banners. All right, that was my, uh, that was my plug. That is it.
Leo Laporte [02:33:28]:
You guys have survived the marathon known as Twit.
Larry Magid [02:33:33]:
John C.
Leo Laporte [02:33:35]:
He would have said get on with it a long time ago.
Devindra Hardawar [02:33:37]:
RIP.
Leo Laporte [02:33:38]:
Salute. RIP, everybody. Pour one out for John.
Allyn Malventano [02:33:42]:
Uh, we would have lost him during the third break.
Leo Laporte [02:33:44]:
Yeah, he'd be off stealing mugs.
Allyn Malventano [02:33:47]:
Yeah, stealing mugs.
Larry Magid [02:33:49]:
Yeah, your $5,000 bottle of brandy or whatever. Got any wine here?
Allyn Malventano [02:33:52]:
That's true.
Leo Laporte [02:33:53]:
Yeah, here's the wine. Yeah. Uh, we'll miss you, John. I, I sure missed having him on the show. He was always Right. Devendra Hardawar, I am so glad you're still around. Keep up the good work. You are the best for a long, long time to come.
Leo Laporte [02:34:08]:
Senior editor at Engadget. Uh, anything you want to plug?
Larry Magid [02:34:13]:
Sure.
Devindra Hardawar [02:34:13]:
I mean, I, if you like me talking about movies, uh, check out The Film Cast at thefilmcast.com. We just reviewed The Odyssey and we're gonna be, we're gonna be reviewing the, uh, the Avatar movie that just came out. Um, Avatar—
Leo Laporte [02:34:25]:
There's another Avatar movie?
Devindra Hardawar [02:34:26]:
When is he gonna stop? The Last Airbender. So The Last Airbender show is now producing new stuff after The Legend of Korra. So Avatar: Aang: The Last Airbender is the new movie, just hit Paramount Plus. They have a new series coming in fall as well. And I think this is one of the best series for kids, best animated series ever made. So check out the movie, check out Avatar: The Last Airbender for sure.
Leo Laporte [02:34:52]:
We saw The Backrooms in Petaluma where the director is from, and he was there. Zach was there.
Devindra Hardawar [02:34:57]:
That's cool.
Leo Laporte [02:34:57]:
What a great story. I mean, it's not a great movie. It's okay, but it's a great story. The guy, you know, started it on YouTube. It's done very well. What is he, 22? And he's a Hollywood film director. They gave him a $10 million budget.
Devindra Hardawar [02:35:09]:
Tim Parsons is his name. But yeah, he was tapped by A24 when he was like 16, 17.
Leo Laporte [02:35:15]:
Yeah, because they said, you know, the kids today.
Devindra Hardawar [02:35:18]:
The kids.
Leo Laporte [02:35:19]:
The kids. A lot of good— boy, you know, I gotta— I haven't seen Obsession yet. I just saw—
Devindra Hardawar [02:35:25]:
Obsession is incredible.
Allyn Malventano [02:35:26]:
Yeah.
Devindra Hardawar [02:35:27]:
Okay, good. Very good.
Larry Magid [02:35:28]:
Yeah.
Leo Laporte [02:35:28]:
I just saw Disclosure Day. I'm, you know, it's a Spielberg, right?
Devindra Hardawar [02:35:32]:
That's a, that was more fun in the theater. I think if you watch Disclosure Day at home, you see how it's like kind of a mess, but you know, it was fun. It was fun.
Leo Laporte [02:35:39]:
I enjoyed it. You know, it was fun.
Larry Magid [02:35:42]:
It just kind of showed though how, how detached Spielberg is from how the world works today. I know. Yeah.
Allyn Malventano [02:35:47]:
Yeah.
Devindra Hardawar [02:35:48]:
Yeah.
Leo Laporte [02:35:48]:
It's a Spielberg. It's a, there's bright lights and there's a lone guy and nobody believes him. Great chase scene in that movie. There is an amazing chase scene in that movie with invisible fire trucks. I love it. I'm not gonna spoil it. Anything great coming out in the next few months that we should be looking for besides The Odyssey?
Devindra Hardawar [02:36:13]:
Summer is ending. I mean, Ridley Scott has a new movie that's coming out that I've heard very little about that's like kind like a post-apocalyptic thing. So that'll be interesting. But I was super excited for Avatar: Aang, and I just saw it this morning with my daughter, and it's incredible. So check it out if you like The Last Airbender. Okay.
Leo Laporte [02:36:30]:
Thank you, Devendra. thefilmcast.com. Uh, there it is. 603 episodes.
Devindra Hardawar [02:36:38]:
Since 2008.
Leo Laporte [02:36:39]:
And there's going to be more because the movies never stop. And I love Jeff Kanata. Give him my best, by the way.
Devindra Hardawar [02:36:44]:
Oh, I will.
Leo Laporte [02:36:44]:
Yeah. Uh, thank you, Alan Malventano. So great to see you. Yeah, I'm sorry I missed you guys when you were in town. Shoot, I saw your email too late, but at least we got you on the show. Next time you're in the Bay Area, come up and say hello.
Allyn Malventano [02:37:00]:
When was I— when was that? When was I in town?
Leo Laporte [02:37:02]:
I feel like it was like recently. Maybe I'm wrong. Weren't you in Sonoma County or somewhere? Was that you? Maybe it wasn't you.
Allyn Malventano [02:37:09]:
No, I don't think it was me.
Devindra Hardawar [02:37:10]:
Oh.
Leo Laporte [02:37:11]:
Oh, oh, was somebody posing as you?
Larry Magid [02:37:13]:
Maybe.
Allyn Malventano [02:37:14]:
Oh no, his agent.
Leo Laporte [02:37:15]:
Maybe it was— I was being catfished.
Allyn Malventano [02:37:17]:
It was one of your AIs. Yeah, there's catfish.
Leo Laporte [02:37:18]:
I was being catfished.
Allyn Malventano [02:37:20]:
Yeah.
Leo Laporte [02:37:21]:
You and your wife weren't in the Sonoma area?
Allyn Malventano [02:37:24]:
Uh, we— I don't think recently.
Leo Laporte [02:37:28]:
Maybe it was a really old email I found. I know it was probably a really old email.
Larry Magid [02:37:32]:
Yeah, I know that, Leo. I've learned that with you. I don't send you emails.
Allyn Malventano [02:37:36]:
No, I've been knocking on your door.
Leo Laporte [02:37:38]:
Text A year or two ago?
Allyn Malventano [02:37:40]:
No. Text you? No, now I gotta look at my texts.
Devindra Hardawar [02:37:44]:
No, I gotta get through the agents.
Allyn Malventano [02:37:45]:
I got Leo's number. Listen, I used to work at the NSA. I got Leo's number.
Leo Laporte [02:37:48]:
You probably can find it. Great to see you, Alan. Thank you for being here. I appreciate it. Thanks for everything you do. Good luck with the hard drive thing. I'm sure they're gonna really take off someday.
Allyn Malventano [02:37:59]:
I mean, we're making them as fast as we can make them. That's all I can tell you.
Leo Laporte [02:38:03]:
It must be a nice— must be nice. And of course, Larry Magid, good luck with the surgery. It is great to see you. Thanks for all you do. connectsafely.org, everybody should go there. There's stuff for young people, there's stuff for old people.
Allyn Malventano [02:38:18]:
Yeah.
Leo Laporte [02:38:19]:
You actually have a guide to Uber senior accounts.
Larry Magid [02:38:22]:
Seniors, yeah, guide to Uber, yeah.
Allyn Malventano [02:38:24]:
Yeah.
Larry Magid [02:38:24]:
And we have an article on there that I wrote about how we need to regulate social media, but not ban it from kids. We don't want kids off it. We want social media to be fixed. So kids can go on it safely.
Allyn Malventano [02:38:36]:
Yes.
Devindra Hardawar [02:38:37]:
Nice.
Leo Laporte [02:38:37]:
I agree. It's too bad it's run by people like Mark Zuckerberg who really don't have children's interests at heart. But at the same time, you know, why are countries banning YouTube? There's so much useful stuff there. There's so much it could be.
Devindra Hardawar [02:38:51]:
Anyway.
Allyn Malventano [02:38:52]:
Leo, uh, it'll be August 5th of 2025.
Leo Laporte [02:38:56]:
Oh, well, I'm sorry I missed you last year.
Larry Magid [02:39:00]:
That was yesterday.
Leo Laporte [02:39:01]:
Really? It was last year?
Devindra Hardawar [02:39:03]:
You should have asked the age.
Allyn Malventano [02:39:04]:
You were having a contractor battle and you were busy.
Leo Laporte [02:39:06]:
I was gonna take you out to dinner. Well, you know, guess what? There's still— it's still happening. There's scaffolding out here. They didn't put scaffolding there yet. We've had— oh Lord, it's been going on for 2 years.
Allyn Malventano [02:39:19]:
Oh my goodness.
Leo Laporte [02:39:20]:
So when you come out in 2028, maybe we can talk.
Allyn Malventano [02:39:23]:
Yeah, maybe you'll be done.
Leo Laporte [02:39:24]:
It's a nonstop— Lisa says, I am so sick of this house. She says, just once, just once, I'd like to be able to get out of bed and there's nobody looking in the window just once. Thank you, Larry, Alan. Thank you, Devendra. Great to see all 3 of you. Thanks to all of you for joining us. We do TWiT on Sundays, 2 to 5 Pacific, 5 to 8 Eastern, 2100 UTC. You can watch us live if you wish in the Club TWiT Discord.
Leo Laporte [02:39:57]:
If you're not a member of Club TWiT, look into it because there's a lot of great benefits— ad-free versions of all the shows, chapter markers so you can skip the AI crap if you want, just jump right by it. You also get special shows like our AI user group if you want more AI. We now do twice a month. We also have the Stacy's Book Club, Micah's Media Club. We have the photo club. We have travel. We have all, I mean, so much great stuff in the club. $10 a month.
Leo Laporte [02:40:23]:
It supports what we do and it's a big part of our operating expenses. We really need it and it makes a big difference. So if you support independent podcasting, if you believe in it, please help us out. twit.tv/twist. But you don't have to be a member to watch live. We also stream to everybody, and there are people watching on YouTube, Twitch, X.com, Facebook, LinkedIn, and Kick right now. 727 of them. Hello to all of you.
Leo Laporte [02:40:48]:
It's good to have you here. After the fact, though, you can watch whenever you feel like it. You can go to our website, twit.tv, download a show there, or get links to the RSS feed, or the link to the YouTube channel. There's a dedicated YouTube channel. We do audio and video of every show.
Larry Magid [02:41:04]:
Yeah.
Leo Laporte [02:41:04]:
Uh, if you subscribe in your favorite podcast client, audio or video, you'll get it automatically. You don't have to think about it. You can have it for next time you're on a submarine. There's something about these shows, people just— so Alan loads them up on his— it probably was an iPod back in the day, right?
Allyn Malventano [02:41:20]:
No, it was a Dell Axim.
Leo Laporte [02:41:23]:
Oh, I remember the Axim.
Devindra Hardawar [02:41:25]:
Yeah, yeah.
Leo Laporte [02:41:27]:
Wow, very early media player. It's funny, Steve Martin, uh, told me that he used to listen to the radio show. He was up in Vancouver, Washington doing a, uh, it's a, a big— it was called The Big Year. It was a show about birding. And he said, we're in the middle of nowhere, I got nothing to do. So he did the same thing.
Devindra Hardawar [02:41:46]:
I was—
Leo Laporte [02:41:47]:
this show's great for lonely people stuck in the wilderness or underwater.
Allyn Malventano [02:41:50]:
I was a superhero on the submarine. I had Dell Axim, uh, loaded with an offline copy of Wikipedia.
Larry Magid [02:41:57]:
Ooh.
Allyn Malventano [02:41:58]:
On a 5 Gigabyte IBM MicroDrive. Sure. I think.
Leo Laporte [02:42:05]:
Sure.
Allyn Malventano [02:42:05]:
Yeah.
Leo Laporte [02:42:06]:
That's huge.
Allyn Malventano [02:42:06]:
Back in the day. So I basically had the internet on the submarine. I was like the only guy.
Devindra Hardawar [02:42:11]:
Wow.
Allyn Malventano [02:42:12]:
Yeah. Very impressive. It was good times. Yeah.
Devindra Hardawar [02:42:15]:
I was watching.
Larry Magid [02:42:17]:
I was watching. Take a bite.
Allyn Malventano [02:42:18]:
Oh, he's leaving.
Leo Laporte [02:42:19]:
Oh, bye Larry. Go, go take care of your ankle.
Devindra Hardawar [02:42:21]:
Bye Larry.
Larry Magid [02:42:21]:
I'm good. I gotta rest my ankle. Take care guys.
Leo Laporte [02:42:23]:
Go rest your ankle, sir.
Allyn Malventano [02:42:24]:
Well, we gotta, we gotta close the show anyway, so like, gotta get Yeah, we're done. Have a great—
Leo Laporte [02:42:30]:
have a wonderful, wonderful life, everybody. We'll see you soon. Okay, maybe next Sunday.
Devindra Hardawar [02:42:36]:
Like we're all going away forever.
Leo Laporte [02:42:38]:
I don't know. Everybody's dying around me. I don't know if I'll ever see you again. I don't know.
Devindra Hardawar [02:42:42]:
Take care, guys.
Larry Magid [02:42:43]:
Good show.
Leo Laporte [02:42:44]:
I might die. Who knows?
Larry Magid [02:42:45]:
Anyway, great show.
Leo Laporte [02:42:46]:
Take care. Another twit is in the can. Bye-bye.
