This Week in Tech 1110 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. I am back and I missed y'all, and we have lots to talk about. 3 of the biggest stories of the week: the FAA grounds all air traffic in the United States for 4 hours, Microsoft grounds M365 traffic for 2 days, and Jensen Huang is out at NVIDIA. Molly White, Patrick Béjaud, Harper Reed, a great TWiT is next.
TWiT.tv [00:00:31]:
Podcasts you love.
TWiT.tv [00:00:32]:
From people you trust.
Leo Laporte [00:00:36]:
This is TWiT. This is TWiT, This Week in Tech, episode 1110, recorded Sunday, November 15th, 2026. High-end electronic drum kits. It's time for TWiT, This Week in Tech, the show we cover the week's tech news. Good thing nothing happened this week. Uh, we have Patrick Béjaud joining us from Paris, France, where it is the middle of the night. Hello, Patrick.
Patrick Beja [00:01:07]:
I'm always happy to talk about all the, uh, things that— no, no, the nothings that happened, or that didn't happen. I'm confused. It's too late for me now.
Leo Laporte [00:01:17]:
In 6 months, we'll have you back on when the French have their presidential election.
Patrick Beja [00:01:22]:
Oh, uh, that'll be—
Leo Laporte [00:01:23]:
yeah, that'll be fun.
Patrick Beja [00:01:25]:
We, we're getting a preview now, which I'm sure you've, uh, heard about on the news. Things are exciting.
Leo Laporte [00:01:32]:
Yeah. Well, it's wonderful to see you as always. And I can't think— there might be a big game release in our future. And I know that you worked at Blizzard for years, so I will be asking you what you'll be doing next week. Also with us, Molly White. Prolific Wikipedia editor, the author of Web3 Is Going Just Great, mollywhite.net, and of course her Citation Needed newsletter. It's great to see you.
Molly White [00:01:59]:
Great to see you as well.
Leo Laporte [00:02:00]:
Yes. And finally, to round this panel out, Mr. Harper Reed, technologist, entrepreneur, hacker from 2389.ai. Did you ever consider changing it to .si when the superintelligence took over?
Harper Reed [00:02:16]:
I have something horrible to admit. I kind of like SI better.
Leo Laporte [00:02:20]:
Oh, I can see why that is.
Harper Reed [00:02:23]:
This is like really disappointing to me because I, I'm— I just don't know. I don't understand why. Like, SI just seems better because artificial, I think, immediately— I mean, you know my kind of views on this. I'm not a doomer, but I'm a little worried. And I think that artificial Negates some of the worry. And I think people should be more serious about it.
Leo Laporte [00:02:46]:
I see. I didn't have any problem with artificial. It was the word intelligence.
Harper Reed [00:02:49]:
I— yeah, that's actually like— that's actually— that's a good point. So we can— super artificial is maybe—
Leo Laporte [00:02:55]:
Super artificial. Now we're talking. Essay, baby.
Harper Reed [00:02:59]:
It's essay.
Leo Laporte [00:03:01]:
Well, uh, you know, the only decision I had to make this week is which of these horrible 3 stories is the biggest horrible 3 stories. I guess I'll start with the ground halt, which happened on Friday. I don't know if this made the news in France, Patrick.
Patrick Beja [00:03:20]:
Yeah, a little bit, but it feels like it's not super important, which I guess is because our planes weren't grounded in Europe.
Leo Laporte [00:03:29]:
Right.
Patrick Beja [00:03:29]:
In fact, flights of airlines were getting declined.
Leo Laporte [00:03:31]:
But I'm sure it was a big deal for y'all. So it was quite a shock for us. To be honest, here in the United States, when we were told for no obvious reason that all air traffic had been halted for about 4 hours on Friday.
Harper Reed [00:03:51]:
That was shocking.
Patrick Beja [00:03:52]:
So you thought it was terrorism-related or something like that?
Leo Laporte [00:03:55]:
Well, that's the first thing, of course. There've only been 2 of these before. One was a technical issue in 2023, but of course the most famous first ground halt was 9/11. I don't think the administration did a great job of explaining what was going on.
Harper Reed [00:04:12]:
Well, they've typically done a good job of explaining things. So
Leo Laporte [00:04:18]:
it is something that you really don't want to do, and this is why it's a big deal in the United States. Because— and it's funny, I was doing— the day before, I had a conversation with science fiction author Daniel Suarez about kind of This, we live in a just-in-time world and air travel has become very optimized thanks to computers. So most planes fly full or very nearly full. There isn't a lot of give in the system. And we've seen that if a plane's late, it ripples throughout. A snowstorm in Chicago, as you know, Harper, means that you're going to have a late flight in LAX even though there's no snow. And it's all very tightly wound. And when you go down for 4 hours, that is probably hundreds of thousands of passengers who now have to be somehow shoehorned into the next 24 hours.
Leo Laporte [00:05:12]:
And I don't know, I imagine it's still a bit of a mess out there. The reason it's a story for us is the government eventually said, well, it's because The air traffic control system in this country is so outdated. It's all running on thin copper wires. That was the phrase they kept saying, thin copper wires, which cracked me up. I
Harper Reed [00:05:35]:
thought—
Molly White [00:05:36]:
If we just had thicker wires.
Leo Laporte [00:05:37]:
If we just had thicker wires.
Harper Reed [00:05:40]:
Those Monoprice Ethernets. Do you remember? You ever seen those?
Leo Laporte [00:05:43]:
Little tiny thin wires? I think I've deduced, and we'll probably cover this in more technical detail on Tuesday with Steve Gibson. I know he wants to dig into it, but it is a TDM network, which is a very old-fashioned way of distributing data where you take turns on the network. And it is the thin copper wires they're talking about aren't, aren't as bad as it sounds. They're T1 lines. So it's running on T1 lines on a TDM network. It's not— it's still unclear what happened.
Patrick Beja [00:06:12]:
Isn't it like the, the wires were, I don't know, corroded or something? And the fact that they were— so it's not— that's not— it's not the, the wires that got physically
Leo Laporte [00:06:23]:
That's why that the thin copper wire thing kind of grew thin.
Patrick Beja [00:06:27]:
And that's not what it was. 'Cause my understanding, listening to this from afar, my understanding was there was a physical hardware failure due to the wires, which happened to be thin. And that's the issue. That was the issue. Are you being facetious? I am not. I thought the wire were so old. No, that they, they got to the issue.
Leo Laporte [00:06:51]:
This ties into the weirder press conference that was held by the Secretary of Transportation, former MTV Real World reality star Sean Duffy, and, uh, bodybuilder and former Congress member Markwayne Mullin, who's the head of Homeland Security. Uh, they, they, uh, they double-teamed it at the White House. Um, and mostly it seemed like they were saying This is why you have to give us $50 billion more. And of course, the president weighed in the next day saying it's Joe Biden's fault, naturally.
TWiT.tv [00:07:22]:
Right.
Leo Laporte [00:07:23]:
Uh, it wasn't thin copper wires or Joe Biden's fault. In fact, could have been maybe something else. Sean Duffy, the Department of Transportation Secretary's son-in-law, uh, has a company— is a, by the way, big contributor to the Trump campaign. that got a no-bid contract for a new AI system called SMART because it's so smart. Yeah. That shouldn't have affected this, by the way, but happened to have been installed the day before. So it's a little— I mean, that's a little suspicious. An AI system that is supposed to take all the data, massage it, and reroute traffic to optimize, uh, fuel usage and timing and all of that stuff.
Leo Laporte [00:08:13]:
You know how AI is so good at that. And, uh, so there was some suspicion that maybe this upgrade had something to do with it. Remember, the system had been working on these thin copper wires pretty well for about 30 years. There were billions of dollars put in, uh, in the Infrastructure Act, uh, to upgrade. Then it came out And this is the thing, the administration is giving us conflicting stories that L3Harris, which is a contractor, was putting in fiber to replace the thin copper wires.
Molly White [00:08:46]:
Thinner wires, probably.
Leo Laporte [00:08:47]:
With even thinner wires made of, gosh, glass.
Harper Reed [00:08:51]:
Light.
Leo Laporte [00:08:52]:
They're so thin and light.
Patrick Beja [00:08:54]:
I believe glass breaks easily. Yes. What were they thinking?
Leo Laporte [00:08:58]:
So that— but they shouldn't have been doing that on a Friday afternoon when air traffic was at its maximum.
Molly White [00:09:04]:
So I don't know, right before Thanksgiving week.
Leo Laporte [00:09:07]:
Yeah, well, the only way it could have been worse is if they did it next week, right? Then, right, then it would have been worse.
Patrick Beja [00:09:13]:
So wait, do we know what happened? Do we know what it is?
Leo Laporte [00:09:15]:
Of course not, because no one has come forward with a— like, you know, normally when an event like this happens, and we're going to get to another one from Microsoft in a little bit, but like when Cloudflare goes down or whatever, in a day, maybe 2, we will get a very detailed Postmortem of this happened, this happened, of course. We haven't seen anything like that. It's been 72 hours.
Harper Reed [00:09:39]:
Didn't the FAA CTO talk about it a little bit more?
Leo Laporte [00:09:46]:
Yes, with another story.
Harper Reed [00:09:48]:
Yeah, okay, okay.
Leo Laporte [00:09:49]:
I think he was the L3Harris guy. It's— I— look, I can't blame anybody for not knowing exactly which story is correct. Um, here's what—
Patrick Beja [00:10:02]:
But I'm shocked that you don't— yes, you still don't have an explanation for what went wrong. I followed this from— like, I didn't follow it actually. I, I thought, you know, it was the, the very thin copper wire.
Leo Laporte [00:10:12]:
Yes, of course.
Patrick Beja [00:10:13]:
But you're telling me you— no one knows what went wrong still? 7— Somebody knows.
Leo Laporte [00:10:18]:
I don't—
Harper Reed [00:10:18]:
somebody knows, right?
Patrick Beja [00:10:19]:
But I mean, they haven't—
Leo Laporte [00:10:19]:
no one knows.
Patrick Beja [00:10:20]:
They haven't explained What went wrong?
Molly White [00:10:25]:
I think they've explained a couple of different— they've sort of given a couple of different hypotheses, and I don't think anyone has really said yet which one they're gonna go with.
Leo Laporte [00:10:34]:
The FBI is investigating. Uh, Markwayne Mullin quickly said it's not a terror— we— he said it's probably not a terrorist attack.
Molly White [00:10:42]:
Yeah, I noticed that.
Harper Reed [00:10:44]:
Probably not.
Leo Laporte [00:10:44]:
Then they said at one point somebody said, oh, there were thousands of transactions happening every second. And I thought, well, that's pretty normal.
Harper Reed [00:10:55]:
Yeah.
Leo Laporte [00:10:56]:
Why is that an issue? And well, those 3 copper wires—
Molly White [00:11:00]:
That's when it transitioned into a blockchain.
Leo Laporte [00:11:02]:
Ah, that's the one thing missing.
Harper Reed [00:11:05]:
Yeah, we haven't brought up—
Leo Laporte [00:11:08]:
so here's a tweet from Anna, uh, Giratelli who said, I've spoken with 60 air traffic controllers tonight about what Sean Duffy said at the White House press conference. None of them are buying what he said because they said these types of switches that the FAA is doing, they don't happen during the day. They certainly don't happen during busy times like a Friday afternoon. They're doing them midweek, middle of the night when there's no air traffic going on. The controllers are saying, we don't buy this is actually what's happening. So that was the one where we, uh, uh, I can't remember.
Harper Reed [00:11:41]:
Was that like a tweet inside of a tweet?
Leo Laporte [00:11:43]:
That was a tweet inside of a skeet.
Harper Reed [00:11:46]:
What are you doing? Just show the original one. What are you doing, Leo?
Leo Laporte [00:11:49]:
I know, because it's just how it comes to me over the transom, and I show it as it is.
Harper Reed [00:11:56]:
Sure, sure, sure.
Leo Laporte [00:11:57]:
Here we, here we are. I think this is a skeet in a tweet.
Harper Reed [00:11:59]:
Oh, this is another one.
Leo Laporte [00:12:01]:
This is from Evan Handler, uh, quoting Cory Black.
Molly White [00:12:03]:
This is my Cory Black theory of 4 social networks each sharing, uh, screenshots of the—
Leo Laporte [00:12:08]:
I know, it's hysterical, isn't it? And you say that we don't have the best network in the world. So this is from Will Black. So it's looking like Sean Duffy, the Secretary of Transportation, making everyone switch to a stupid effing AI system, but the T1 lines that connect everything couldn't handle it. This seems to me the most likely, by the way. Duffy said the problem was, quote, old copper wires from the '80s, and everybody in France believed him. Will says—
Molly White [00:12:36]:
It's good that we have the spokesperson for France here.
Patrick Beja [00:12:39]:
If the tube— listen, it makes complete sense. The old tubes from the '80s weren't big enough to accommodate the amount of data that the AI was putting through.
Leo Laporte [00:12:50]:
It's the tubes' fault. I just want to remind you about Battlestar Galactica. You may remember when the Cylons attacked, they were able to bring down the entire Earth's defensive systems except for one battleship Where the, you know, commander was an old guy and he had everything on the wire, like the telephone and one of those curly wires. And as a result, Battlestar Galactica survived the Cylon attack. You remember that, right? I think one of the things that's true about the air traffic control system in this country is it works. It has been working. And you, you mess with it, admittedly, it's Probably a little, you know, shaky, but you mess with it at your peril. Uh, he says—
Molly White [00:13:39]:
Sort of like moving banks off of COBOL, you know? Yeah, nobody wants to because it works.
Leo Laporte [00:13:46]:
I
Patrick Beja [00:13:46]:
don't think it's a reason not to do it. I think it's a reason to do it carefully and well.
Leo Laporte [00:13:51]:
Carefully.
Patrick Beja [00:13:51]:
Yeah, exactly.
Leo Laporte [00:13:53]:
And we have spent— literally, we've already spent more than $20 billion doing that. This was mostly an excuse for Sean Duffy to say, and we need Congress to give us $50 billion more and unaccountably blame Congress, which has been a Republican Congress for the last 2 years under a Republican president. So I'm not sure I understand.
Molly White [00:14:11]:
It's Joe Biden's fault. Didn't you hear?
Leo Laporte [00:14:12]:
How it could be Joe Biden's fault. But anyway,
Patrick Beja [00:14:14]:
Joe Biden and his thin copper wires.
Molly White [00:14:17]:
He's over unplugging stuff at the FAA network.
Leo Laporte [00:14:21]:
Black goes on to say you can access AI systems via a T1 line. because you're just sending text to a massive data center and the AI computation is done there, right? But from what I've read up on, in order to have the entire system run on all this AI BS, this smart system, to have the actual network itself propped up on it, you need fiber optic. Okay, fine. A few weeks ago, Sean Duffy was asking for $30 billion, remember? Now that's the previous number to update the system, $30 billion. That's a lot of money. Which means somebody told him that switching to this AI system wouldn't work on the old grid. So we— you better, you know, do this. Something he'd never considered because his background is in reality TV, not tech.
Leo Laporte [00:15:05]:
So they switched it over anyway. Everything goes to heck. Now he and Markwayne Mullin are out there telling everyone the outdated system failed. I think this is a reasonable take. Again, we don't know. It's— no, it's because you tried to run this janky AI on legacy hardware. He says, like trying to play a fully modded version of Cyberpunk 2077 on a, on an effing Nintendo 64. Okay, now there's another thing to bring in here.
Leo Laporte [00:15:36]:
Doge, you may remember, fired more than 10,000 FAA technicians and staff 2 years ago.
Harper Reed [00:15:46]:
Are you saying that we created this mess, Leo? That we did this to ourselves?
Leo Laporte [00:15:52]:
I'm saying what I would like is a very concrete explanation of this. And I'm saying the longer it goes without that, the more suspicious I am that the real truth doesn't reflect well upon the people in charge. Does that make sense?
Harper Reed [00:16:11]:
Seems reasonable.
Patrick Beja [00:16:12]:
It makes too much sense. Yeah,
Harper Reed [00:16:13]:
yeah. Seems unrealistic.
Leo Laporte [00:16:16]:
The FBI apparently is going to probe whether a software update caused the outage.
Molly White [00:16:24]:
Like, why?
Patrick Beja [00:16:25]:
I think they should blame Microsoft. There was an outage for them too, likely. Yeah,
Leo Laporte [00:16:30]:
why the FBI?
Molly White [00:16:30]:
When I first heard about the ground halt, I was like, oh my God, they were all emailing each other. And when Microsoft went down, it broke the airlines. Oh, it could be.
Leo Laporte [00:16:38]:
Is it related? I don't know. Could be. All right. I guess there's not much more to say except that I think the world, but certainly Americans deserve an explanation, especially those Americans who were unable to get where they were going and may still be stuck in—
Patrick Beja [00:16:55]:
I mean, that's inconvenient. I'm sure it's maybe even extremely problematic for a few people, but it's not the end of the world. What really bothers me in this story is the murkiness of the explanation 72 hours afterwards. Right. Because in these situations, it is incredibly important that you are clear about what happened, how you mitigated it, how, you know, the fact that we still don't know. I mean, the, the, the Microsoft thing I feel has been handled a lot better. And I, I, you know, I'm on the side of all of these outages and stuff. they're the worst thing that can happen, worst-case scenario, like the PHP thing a few weeks ago.
Patrick Beja [00:17:45]:
And it's not so bad that the world ends, right? We have a few websites down for a few—
Leo Laporte [00:17:50]:
No one died. No planes went down. I mean, it, you know, it wasn't a disaster because they took quick action and said, uh, we're shutting this whole thing down, even at great economic expense, you know, consequence. Uh, so there is that to be said.
Patrick Beja [00:18:05]:
It's still, it's still a big deal, but to me the real issue is you want your administration to explain to you what happened.
Leo Laporte [00:18:16]:
We expect some more competence than we're getting.
Harper Reed [00:18:20]:
Do you think that they will come clean with what really happened, or do you think it'll just fade into oblivion like the rest of the kind of this administration?
Patrick Beja [00:18:28]:
Seems like it's not. If they are to blame For it, if it is indeed that AI system that the thin copper wires couldn't handle, and if it's actually them who are to blame, I think that in this administration, they probably would not admit to it because it's literally, if that's the issue, they should submit their resignation, right? The person who authorized it.
Molly White [00:18:55]:
Especially with the connection to Sean Duffy and his family member who got the contract. Yeah, son-in-law. That's not a great—
Leo Laporte [00:19:01]:
It's his son-in-law's AI. And then somebody today mentioned that it was a router. It's always a router.
Molly White [00:19:08]:
Which, I mean, that does seem reasonable.
Leo Laporte [00:19:10]:
Yeah, it's always a router.
Molly White [00:19:13]:
Did you see the night of the— I think it was the night of the Groundhog Day that one of the tweets or posts that Sean Duffy put out was reassuring everyone that they were going to prioritize the flights of NCAA athletes so that they wouldn't miss their games? Oh,
Leo Laporte [00:19:26]:
yeah. Well, then they mentioned that today on the NFL games. that all the teams got priority. Yeah, because thank
Molly White [00:19:33]:
goodness
Leo Laporte [00:19:34]:
we do have our priorities in this country.
Harper Reed [00:19:36]:
Yeah, yeah.
Leo Laporte [00:19:40]:
I mean, that doesn't— if you're a family stuck sleeping on a— you know what really bugs me? That in airports they make it hard to sleep. They don't want you sleeping.
Harper Reed [00:19:51]:
I think on purpose.
Leo Laporte [00:19:52]:
Yeah, of course. So if you're in an airport You're in the worst possible place to have to spend the night, right? What was there? What, there's a movie where the guy lives in an airport for a year?
Patrick Beja [00:20:05]:
Yeah, for a few years. It was in a French airport, actually.
Leo Laporte [00:20:08]:
It's a true story, right?
Patrick Beja [00:20:10]:
Yeah, yeah, yeah. Terminal. That's a Spielberg film.
Leo Laporte [00:20:13]:
Terminal.
Patrick Beja [00:20:13]:
There you go.
Leo Laporte [00:20:14]:
Great name.
Patrick Beja [00:20:15]:
Yeah, with Tom Hanks.
Leo Laporte [00:20:16]:
It was Tom Hanks. I don't know why. Poor guy. First he's on an island with a volleyball, then he's stranded in an airport. They just— this guy can't catch a break. All right. Well, I guess we should move on then to the other big outage of the week. Actually,
Harper Reed [00:20:31]:
I did, I did like the idea that they're the same outage. That was my hope. I really, I really thought that was better than just a new other outage.
Leo Laporte [00:20:42]:
Did you do— are any of you on Exchange? Any of you live on Microsoft? Use Teams, Copilot?
Patrick Beja [00:20:48]:
Thankfully not.
Harper Reed [00:20:49]:
I know we were working with a team that does, and they just all were like, I guess no more work.
Leo Laporte [00:20:55]:
With—
Patrick Beja [00:20:56]:
Yeah, it was great for a lot of people. There, there's a lot of people, like my, uh, wife works on, uh, on, uh, Microsoft on Outlook and stuff. And she was like, well, you know, we, no one could work. So we were all like, all right, well, it reminds me when I was at, uh, I used to work in a real company at Blizzard Entertainment. And, um, we had a couple of days over my 5-year tenure where the internet was down. And so everyone looks at each other and you're like, All right, let's just go get coffee. You know, this is a victimless issue, and this is a theme I'm gonna get back to.
Leo Laporte [00:21:36]:
This makes everybody— It's true. I thought there'd be more complaints because, okay, so it was Outlook email, it was Teams, it was Copilot, which the AI no one uses. It was— there's a funny AI video of this happening at the office, and Michael Scott comes out and says, but is Microsoft Paint okay? And everybody said, yeah, you can still use Paint. So it didn't affect Paint. Microsoft 365—
Molly White [00:21:59]:
Paint's cloud-connected these days.
Leo Laporte [00:22:00]:
Paint is not. It is connected. There is Copilot in Paint now, I'm told, but I—
Molly White [00:22:05]:
My God.
Patrick Beja [00:22:06]:
Yeah, you can use it locally, like without Copilot.
Leo Laporte [00:22:09]:
You can still use it by hand.
Harper Reed [00:22:11]:
Wasn't there a really, there is a really, really big Paint RCE, I think, dropped recently.
Leo Laporte [00:22:17]:
No.
Molly White [00:22:17]:
Really?
Harper Reed [00:22:18]:
Yeah. Oh yeah. Yeah.
Leo Laporte [00:22:19]:
Remote code execution through Paint. If I send you a picture my 8-year-old drew, But it's malformed. I can get into your computer, steal all your data points.
Harper Reed [00:22:30]:
Yeah, this month, November 8th, Microsoft announced that.
Patrick Beja [00:22:33]:
Well, maybe that was the issue.
Harper Reed [00:22:34]:
Yeah, maybe that was the FAA thing. Maybe that's how they track the planes.
Leo Laporte [00:22:38]:
They were drawing the planes on a map. Microsoft has been patching like crazy. The last 3 months have had more than 1,000, at least the last 2 months have had more than 1,000. Then 1,000 fixes in Patch Tuesday. And every month there's several zero days in there.
Harper Reed [00:22:56]:
I am not an accelerationist. I need to say that because I would, I just love all the security issues. I think it's really fun to watch. And I really feel sad for everyone that has to deal with this all the time because it doesn't seem like it's ever going to end.
Leo Laporte [00:23:09]:
I'm very confused by the sentence you just uttered. You are not an accelerationist. You love all—
Harper Reed [00:23:14]:
I don't want, I just think it's really dynamic and someone like, it's always exciting to have these things happen. Like I'm very, I am excited. about it. I see it and then I'm like, but I don't want it to continue to happen and destroy all the world. I'm gonna do—
Leo Laporte [00:23:28]:
you mean the outages?
Harper Reed [00:23:30]:
Just like all the RCEs, all the outages, all the hacks.
Leo Laporte [00:23:33]:
All right, I'm gonna paint this in a way which you probably should learn to, to make you look good instead of looking crazy, which is it's good we're finding all these flaws because they're getting fixed.
Harper Reed [00:23:46]:
That's not why. That's not why. I think it's very dynamic. I'm trying to help you.
Molly White [00:23:52]:
I know, I know. Leo's really relies on the idea that there's sort of a finite number of issues.
Leo Laporte [00:23:58]:
Yes.
Molly White [00:23:58]:
Fixing.
Leo Laporte [00:23:59]:
Yes. And then someday this will stop and it'll be perfect.
Molly White [00:24:03]:
No, it doesn't.
Patrick Beja [00:24:04]:
The security community think that it's going to last for like 18 months and then everyone will be caught up and like updates will be designed pre-AI'd so that there aren't AI-findable security Issues. So it is gonna end at some point. But although this one, you know, we went down the AI rabbit hole, apparently Microsoft is saying, no, it's not AI. Relax. This one is not AI.
Leo Laporte [00:24:33]:
But did they say what it was? That's the question.
Patrick Beja [00:24:35]:
I think they said it was a network update, right?
Leo Laporte [00:24:39]:
It's always a network update.
Patrick Beja [00:24:41]:
Yeah, right.
Leo Laporte [00:24:41]:
But I think the FAA thing is a network update.
Harper Reed [00:24:44]:
Definitely a router update.
Patrick Beja [00:24:45]:
It's a router issue. Cables. They
Harper Reed [00:24:47]:
upgraded their Wi-Fi.
Leo Laporte [00:24:49]:
Rusty copper wires. No one ever said what it was.
Patrick Beja [00:24:53]:
Capacity upgrade, so maybe it was hardware related.
Leo Laporte [00:24:56]:
It might have been related.
Patrick Beja [00:24:56]:
I don't know.
Leo Laporte [00:24:57]:
Microsoft has 430 million global users, and this was a global outage, by the way. Countries from the US to South Africa to India to New Zealand. Was it the case in France? Were you—
Patrick Beja [00:25:10]:
Yeah, yeah, yeah, absolutely. Everywhere.
Leo Laporte [00:25:11]:
House of Representatives, Australian police, the Japanese government. In Hong Kong, bankers told CNN their work was disrupted. OpenAI and Anthropic also use 365.
Harper Reed [00:25:23]:
I do think we should, we should move more things to the cloud. That's my takeaway here, that the more things we move to one single point of failure, the better.
TWiT.tv [00:25:32]:
Yeah.
Patrick Beja [00:25:33]:
Well, all right, listen, the world we live in is pretty complex and all its tools we use and all— yeah. And given the incredible amount of complexity and abilities that we have, like the productivity, the tools, the things we can do for work and for— I'm being a little bit of a contrarian here because I'm sure everyone is panicking over this, but all the things we can do, if the price we have to pay is that sometimes we get one day where we can't send emails or one day when governments have to take down their websites because there's an RCE in PHP. PHP. Sure. Then is that really the end of the— like, again, all the things we do, of course, we should try to avoid those. And probably the AIs are making the security vulnerabilities more visible. So for a while, it's going to be more difficult. But if that is the price we have to pay for everything that we can do, I don't think it's such a big deal.
Patrick Beja [00:26:42]:
Like, okay, you couldn't send emails for a day. Fine.
Leo Laporte [00:26:45]:
Well, it isn't a big deal, but the problem is—
Patrick Beja [00:26:47]:
I'm sure some people had a huge issue with their business or something happened in the world. I get that.
Leo Laporte [00:26:53]:
But overall— And it was for a long time. I mean, this is an unusually long outage. It was more than a day, right? I mean, it wasn't an hour or two. But I think where this is causing some anxiety, certainly in me, is it feels like these are tremors, little Yes, harmless tremors before the big one. And you're laughing. I feel like you're enjoying this, Harper.
Harper Reed [00:27:22]:
Well, I just like the language. I like the language. I'm pro-hyperbole. And so I love the idea of a big one.
Leo Laporte [00:27:30]:
But so it's one thing if Alvin and the Chipmunks outlet goes out.
Harper Reed [00:27:34]:
Do you have an electric car, Leo? Do you have a gas car as well?
Leo Laporte [00:27:37]:
Yes.
Harper Reed [00:27:38]:
Okay, so you are—
Leo Laporte [00:27:39]:
Because Lisa bought one, which I know I was against.
Harper Reed [00:27:42]:
But what if there's an RCE?
Leo Laporte [00:27:46]:
Right. But here, no, look, I'm giving you an example. No big deal. Microsoft goes out for 48 hours. Okay, you can't do Teams. That's not the end of the world. In fact, it might even be a blessing. What if our grid goes out for 48 hours?
Molly White [00:27:59]:
Yeah.
Leo Laporte [00:28:00]:
We, you know, food supply in this country is— in this country, and I'm sure it's true in France, although you have a lot more farmers or something, The butt, you got great butter, great butter. But, but, uh, what if the food supply— we only have 3 or 4 days of food.
Patrick Beja [00:28:16]:
Well, you know, I feel like we are, uh—
Leo Laporte [00:28:17]:
Yes, it's all wonderful that it's all just in time and it's all software-based and it's amazing how it's all interconnected. And, and yeah, oh, we only lost email, but someday it's not going to just be email. It's not just going to be Groundhog.
Patrick Beja [00:28:33]:
There was this freezer, uh, Yes. Yeah, just in the US. This freezer hack, which apparently is linked to Iran.
Leo Laporte [00:28:40]:
Right. In France, it was hundreds of commercial freezers, right?
Patrick Beja [00:28:44]:
Yeah, yeah, yeah. There were a lot because for those who don't remember, they hacked into the system of those stores and messed with the temperature regulation, which raised the temperature enough that most of the goods were spoiled, most of the goods in the freezers in several store chains, which was a big deal. And it turned out that those companies, because they're dealing with just food and not IT, didn't have the culture of updating all of the parts of all of their infrastructure that needed to be updated. They were outdated pieces of those devices. And yes, that is a big issue. But if you start thinking at some point there's going to be a huge— like in, I don't know, war zones, depending on the— you mentioned I was born in Lebanon. I've spent a little while in Lebanon. In a literal war zone, you get— I was getting electricity how do you call it? Blackouts for a little bit every day, but you adapt, you work with it.
Patrick Beja [00:30:06]:
And I don't think that the end of the world, because what you're talking about, Leo, would be if everything stopped, infrastructure. I don't think that is— we need to make sure that everything is working properly, but I don't think from a couple of outages, it is reasonable to intellectually go to what if everything stops on the same day at one point?
Leo Laporte [00:30:32]:
So we shouldn't worry.
Patrick Beja [00:30:33]:
That being said, no. Well, I'm not saying—
Harper Reed [00:30:35]:
That being
Patrick Beja [00:30:35]:
said. That being said, I do have a little bit of a, how do you call it? Like survivalist trend in me. When we live in Finland, I have, I want to buy like solar panels and like generators and have chickens just in case, you know. Oh, did you? Chickens are great.
Molly White [00:30:58]:
Well, yeah, true. Um, did you see that there was some panic, uh, not too long ago because I guess the Pentagon issued a directive to military bases saying that they need to be basically prepared to be able to operate for 90 days without like electricity and stuff?
Leo Laporte [00:31:16]:
It wasn't just the US, it was worldwide, right? Well, I think people thought, is this— does this mean there's some sort of attack they're getting ready for?
Molly White [00:31:24]:
And— but the weirdest thing to me was that, um, people were linking the Pentagon's warning— or not warning, but direction— to a UK thing where they basically told people that they should have 3 days of food, like, in store, which is just like a normal weather thing.
Patrick Beja [00:31:40]:
Yeah.
Molly White [00:31:40]:
Like, that one, I was like, why are people linking that?
Harper Reed [00:31:44]:
It also aggressively escaped internet containment. I heard this from a lot of people in physical spaces where people were like, did you hear that the world is ending? What does this mean?
Leo Laporte [00:31:53]:
But see, this is another sign to me of the anger anxiety we're all suffering, right? We're all— it's in the back of our minds.
Molly White [00:32:01]:
I was gonna say, just today I had someone in real life basically telling me about how there's going to be a big one on the 17th because there's like a warning about—
Harper Reed [00:32:09]:
was it Leo?
Molly White [00:32:11]:
Yeah, it was
Harper Reed [00:32:12]:
Leo. Yeah.
Molly White [00:32:13]:
Um, no, it was about— I guess they're testing like the emergency alert system, and for some reason people think that that's gonna be— that that's like prep for something they know is going to happen.
Patrick Beja [00:32:24]:
Well, that's the time when you want to launch your attack if you're, you know, exactly, exactly. But, you know, we— so I live in France now. We also have roots in Finland, uh, as you know, Leo, um, and we are very aware of that anxiety because our neighbor in Finland is, uh, Russia. And it is not like theoretical or imaginary. There is a serious problem, a serious risk with having Russia as a neighbor, especially post-invasion of Ukraine. So I understand what you're saying, but I think you're right. It doesn't need to be anything concrete, a real risk. It's just that the world has become so uncertain.
Patrick Beja [00:33:16]:
Things have become a lot more uncertain than 20 or 30 years ago, and we don't know what's going to happen. Uh, you know, the, the—
Leo Laporte [00:33:25]:
And we're more interdependent than ever, are we not?
Patrick Beja [00:33:28]:
Yeah, well, but I think we were 20 years ago and there wasn't the same level of anxiety. We were just, but, you know, discussing the latest iPhone and we were happy about that.
Leo Laporte [00:33:38]:
Okay, as a boomer, I can tell you there was a certain amount of anxiety in my childhood because of Cuban Missile Crisis and the threat of nuclear annihilation, we had duck-and-cover drills, we had civil defense signs in the basement, there were barrels of water in the basement.
Molly White [00:33:55]:
Well, I'm thinking of Y2K as well. People were worried about that sort of shutting everything down. So maybe that's just human.
Leo Laporte [00:34:00]:
Maybe we're all just always like this.
Harper Reed [00:34:02]:
I do think there's a slightly other thing that I think about a lot, which is that 20 years ago, The, there was a lot of speculation about what the future would hold. And you had people like William Gibson, who I think is a pretty good example of this. Like I just reread the Bridge Trilogy because I w— I kept thinking all of the headlines right now are Gibson plotlines.
Leo Laporte [00:34:24]:
Yeah.
Harper Reed [00:34:25]:
The idea of, and, and the, the speculative science fiction from the early 2000s or the late '90s, some of it pictured a world of which we are living in and we are post it in many ways. We're post that world and the, the speed and the, the weirdness and the various kind of, uh, Um, whether it's just in shitification or just like hypercult kind of vibes that is happening where people are in their own little worlds, um, is, is very, very much hearkens back to like some, a lot of William Gibson novels. And I find that very interesting both because, um, maybe that's why I'm, I'm an accelerationist when it comes to RCEs. It's just there. I want to, I want to know.
Leo Laporte [00:35:04]:
You're saying bring it on is what you're saying.
Harper Reed [00:35:06]:
Well, I just, I think it's always, it's always a really interesting thing. Like we, we work with this young person and. We recently had a, they had an opportunity to go travel for this very strange thing. And I was like, you should do it. Like, this is exactly what would happen in a William Gibson novel. You now have to come back with some really good travel hacks and tell us all about the, you know, the various weird people you met that might be spies and might not be spies. Like, you never know. And like, there's some, you know, like, uh, it's, it's just, I find this all very interesting.
Harper Reed [00:35:34]:
And I think part of it is the process of stepping through the threshold into this new world where a lot of things are mitigated with AI. We're starting, we're starting to figure out. Like, how does this work? What does this mean from a security standpoint? As Patrick was saying, like a lot of the security researchers are saying, you know, give us 18 months to get everything under control. We're gonna use AI to patch, we're gonna use AI to attack, we're gonna use AI to patch, so on and so forth. And I think that we're seeing a lot of the fringe sides, the things that are not yet well-defined. And that's an example of that. Yeah, go ahead, Leo.
Leo Laporte [00:36:03]:
Before we take a break, 'cause we are running a little long, I do wanna say there is some reassuring news. This is the, thing you were talking about, Molly, the Pentagon's doomsday plan to keep America fighting if the global war erupts, the Fortress Initiative. Too foreshadowing. And notice this, the Project Meridian, which has been launched to study future warfare, announced by the Department of War's Chief Technology Officer, Emil Michael. And if you want to do a little research on Emil Michael, that might not encourage you, but it's going to be co-led by Elon Musk, Palmer Luckey, and Newt Gingrich.
Molly White [00:36:46]:
Wait a minute.
Patrick Beja [00:36:47]:
One of those is not like the others.
Leo Laporte [00:36:48]:
None of those make me feel so much better about the future.
Patrick Beja [00:36:54]:
I want to acknowledge that it feels like things are really bad in some ways. And I want to acknowledge what you were saying, Leo, about the Cold War, which was really not fun. There was I, even as a not boomer, I have remnants of it. Like it was much more present in the, in popular culture. But I think we also have a tendency to forget how bad things were, um, at other times.
Leo Laporte [00:37:24]:
Like, oh yeah, in the '90s, we had— People used to eat babies. We've been— humans are horrible.
Patrick Beja [00:37:29]:
No, but even, even more recent, like in the '90s, there were a couple of, a couple of, uh, uh, you go to a disco
Leo Laporte [00:37:36]:
and
Patrick Beja [00:37:36]:
it would— well, that's what we remember, right? That's what we remember. The '90s were so great. We had
Leo Laporte [00:37:44]:
a
Patrick Beja [00:37:44]:
couple of wars in Iran and Iraq.
Leo Laporte [00:37:48]:
Oh yeah.
Patrick Beja [00:37:48]:
And if you go to the 2000s, where I went was, oh, we were reviewing iPhones. 2001 was 9/11, right? The entire 2000s, the aughts, were the threat of terrorism, the constant anxiety and fear. that tomorrow you could get into the subway and die of a sarin attack. Yeah, like
Harper Reed [00:38:12]:
it
Patrick Beja [00:38:13]:
was— I think it was also kind of crazy. And I don't know when that time when it wasn't crazy was, because if you look at them each individually, it was kind of crazy. That being said, again, it's pretty crazy right now. Like,
Leo Laporte [00:38:28]:
the '50s were good. I'm just gonna say the '50s.
Harper Reed [00:38:32]:
The '50s. Well,
Molly White [00:38:33]:
I think—
Patrick Beja [00:38:34]:
well, it was just after the end of the war, so
Leo Laporte [00:38:36]:
the '50s were Not for you guys, but for me as an old white man, these were the years, you know?
Molly White [00:38:43]:
I do also think there's that tendency to be like, oh, the '90s were great for people who were like children in the '90s. And it's like, 'cause I didn't have to work and I had toys.
Leo Laporte [00:38:51]:
Great cartoons.
Harper Reed [00:38:52]:
Yeah, I loved the '90s. The '90s was great. High school.
Patrick Beja [00:38:55]:
I think the big difference is that we kind of, we have, there's probably a big disparity in income between different social classes. bigger than it was. I think that's pretty well established.
Leo Laporte [00:39:06]:
Oh, income inequality has got to be one of the number one problems we face.
Patrick Beja [00:39:09]:
And the other thing is people don't trust their governments at all anymore. And I think that's another big thing, right? However bad you thought—
Leo Laporte [00:39:16]:
By the way, those are interrelated, I believe.
Patrick Beja [00:39:19]:
Yeah, yeah, absolutely.
Leo Laporte [00:39:21]:
I mean, when we've set it up, at least in the United States, and I imagine it's true in many other places, that the more money you have, the more influence you have in politics. And so income inequality broadens because the people who have so much money now make sure that they continue to get more and more of the share of the pie.
Patrick Beja [00:39:38]:
Well, it's more the case in the US, but certainly we feel it. I mean, electorally in the US, you have much more power, voice.
Leo Laporte [00:39:47]:
Elon Musk can spend $250 million.
Patrick Beja [00:39:50]:
I think that's the thing that was so hopeful with the American elections, the midterms, A couple of weeks ago.
Leo Laporte [00:39:58]:
It didn't matter.
Patrick Beja [00:39:59]:
Yeah, exactly. Which I think regardless of where you stand on the political spectrum, maybe you're unhappy that Elon Musk didn't get his way, but the fact that his $474 million whatever didn't work out, I think is giving hope. Although you can also think, how bad must the situation be for that not to be enough? Well, anyway, but in France we also have this income inequality issue and the system itself is similar all over the world. That's what the riots are about, isn't it? Yeah. Well, yeah, the riots are about a lot of things, but at its core, I think it is about a rift between the people who have more money and the people who have less money.
Leo Laporte [00:40:46]:
Yeah.
Patrick Beja [00:40:46]:
All
Leo Laporte [00:40:46]:
right, we're going to take a little break. Uh, I'm just— it's nice every time we have Patrick on, he try— he talks us down. He always tries to talk us down. He's gonna be okay.
Harper Reed [00:40:56]:
Let's try and— let's, let's, let's stop that in the next session.
Molly White [00:40:58]:
Yeah, let's see how freaked out we can get everybody.
Leo Laporte [00:41:01]:
We haven't talked about, you might argue, the biggest story of the week. I don't know. Uh, there, there were 3, and, uh, we've only talked about 2.
Patrick Beja [00:41:10]:
I'll, I'll, I'll, uh, make everyone relax a little bit.
Leo Laporte [00:41:15]:
Will you?
Patrick Beja [00:41:15]:
Good.
Leo Laporte [00:41:16]:
I like it. That's your, that's your new role. He's always like, it's going to be okay. That's Patrick Béja, uh, of Le Rendez-vous Tech and The Philias Club. notpatrick.com is the place to go to find his podcast. Both, do you do any English ones anymore or?
Patrick Beja [00:41:33]:
Yeah. So, well, The Philias Club is kind of on, on again on hiatus. Um, but I can't bring my, I, I was supposed to bring the Patreon page down and I couldn't bring myself to do it. But Le Rendez-vous Tech and Le Rendez-vous Jeu are French language, uh, weekly news shows about tech and gaming. So if you want to check them out.
Leo Laporte [00:41:52]:
Well, let's talk about le jeu when we come back. I think there's a big jeu about to come our way. Also, Molly White is here, mollywhite.net. And, uh, you know, if I go there, can I get Citation Needed? Is that the best place to get it? Or—
Molly White [00:42:07]:
okay. Well, I mean, citationneeded.news will bring you directly there, but you can also find it from my website.
Leo Laporte [00:42:12]:
It's easy to remember, mollywhite.net. So yes, and everybody should subscribe because Molly's great. And we will talk about the Binance, the largest crypto heist in history.
Molly White [00:42:26]:
So far.
Leo Laporte [00:42:28]:
So far. There's always a chance for better.
Harper Reed [00:42:31]:
There's always more.
Leo Laporte [00:42:32]:
For more.
Harper Reed [00:42:32]:
I love
Leo Laporte [00:42:33]:
it. See, she's an optimist too, ladies and gentlemen.
Molly White [00:42:36]:
Exactly.
Leo Laporte [00:42:36]:
You can always go bigger. And Harper Reed of 2389.ai and all of his misfit AIs. I love, and I'm seeing it over your left shoulder right now, that wonderful, what is it called, Picasso bot?
Harper Reed [00:42:52]:
Oh yeah, the Mikaso.
Leo Laporte [00:42:54]:
Mikaso.
Harper Reed [00:42:55]:
Yeah, it's really cool. I wonder if I have, I have some here.
Leo Laporte [00:42:58]:
It's a 3D printer that's been set up to draw pictures of you as you come in the door at 2389.
Harper Reed [00:43:05]:
Yeah, we also hooked, we also made one. So first of all, it's an old hacked Monoprice 3D printer. 3D printer that cost $50. Nice. And then we wrote a new firmware for it that will draw. And, um, one of our guys put it on a, on, uh, like a, a spool of paper, and then we made a print driver for it so you can print PDFs and it will write them all down.
Leo Laporte [00:43:27]:
Are you doing more with hardware now with AI than with software?
Harper Reed [00:43:30]:
We do a lot with hardware because it's really fun. Yeah. And I am, I am, I think my body is 30% ESP32 at this moment. And, um, and I am, yeah, I mean, you know, I, I have my, um, I mean, yes, I have ESP32s all over the place here.
Leo Laporte [00:43:46]:
As do I, thanks to you.
Harper Reed [00:43:48]:
I know, I have a new hack for you, Leo, so let me know when you're ready.
Leo Laporte [00:43:52]:
Yes. Oh, that'll be fun. I am hoping that at some point you will help me hack this thing. This is the HomeView that I foolishly spent $300 on. It's a beautiful piece of, you know, hardware from Apple, as always, just gorgeous. But it's really just a giant Apple Watch on a little rotating stand with a nice speaker. I can't really figure out what to do with it, but I feel like it should be hackable. Maybe not.
Leo Laporte [00:44:17]:
Apple's not exactly making it easy to hack stuff. All right. I'm good at withholding satisfaction for delaying the big story of the week, but I guess we're going to have to talk about Jensen Huang.
Harper Reed [00:44:35]:
I love this story.
Leo Laporte [00:44:37]:
Wednesday, we're doing intelligent machines and Paris Martineau goes, and I said, what? He resigned. Resigned probably before he was about to be fired. So in case you are not following all this, Jensen Huang, the CEO and founder of NVIDIA, the biggest, most successful company around right now, more than— was worth more than $5 trillion. First, it came out that he was having an affair with a woman who ran a Singapore company who perhaps was a Chinese spy. That's not clear, but was, it looks like, funneling Nvidia chips to China illegally. And maybe they could have sustained that, but then it came out that he was also having an affair with a direct report. And then a third affair has apparently since surfaced. In any event, Jensen wrote a, uh, what must have been a very humiliating and humbling email to the staff saying, yes, I, I, this is true, and I am, I quit, I resign.
Leo Laporte [00:45:43]:
Um, there is already a petition going around among Nvidia employees saying bring him back. Um, either, uh, Elon Musk and other tech executives have signed that petition. Reinstate Jensen Huang, bring back the boss. 1,600 comments.
Harper Reed [00:46:12]:
Great photo.
Leo Laporte [00:46:14]:
Yeah, I don't know where Wired got that photo.
Patrick Beja [00:46:17]:
I think there's a lot of those going around now.
Leo Laporte [00:46:20]:
Until about a week ago. Yeah, there was a lot of AI stuff, wasn't there, Molly? There are a lot of— this was the face that launched 1,000 memes.
Patrick Beja [00:46:28]:
Did you see that TSMC, uh, put a leather jacket at half mast?
Leo Laporte [00:46:33]:
Yeah. Oh man. And the Denny's where he, uh, according to fable, uh, started NVIDIA at a booth in Denny's and there was a plaque there. They took it down.
Harper Reed [00:46:46]:
And then they put it back up.
Leo Laporte [00:46:48]:
Oh, they've put it back up since?
Molly White [00:46:49]:
Yeah.
Harper Reed [00:46:50]:
I think he's going back to NVIDIA.
TWiT.tv [00:46:52]:
No.
Patrick Beja [00:46:52]:
Well, that's the rumor now. He's plotting a comeback.
Molly White [00:46:56]:
It's a Sam Altman thing.
Leo Laporte [00:46:57]:
Oh my God. Okay. But, but Leo's
Harper Reed [00:47:01]:
like,
Leo Laporte [00:47:02]:
no, no. So, uh, until a week ago, he was the hero of AI, right? Uh, Nvidia's hardware was powering the AI revolution. He was a, a huge proponent of open, uh, open-way AI, which I'm, I'm completely in agreement with. In fact, I even agree with his position that we should be selling Nvidia chips to China. Otherwise, they're just going to develop their own chips that'll be just as good, and then we won't have any competitive advantage. Turns out he might have had a reason for saying that beyond— I mean, it was obvious he had money at stake there. Maybe he also had a—
Patrick Beja [00:47:42]:
I think it makes business sense for Nvidia regardless, right? Regardless of his proclivities.
Leo Laporte [00:47:48]:
Yeah.
Molly White [00:47:49]:
I mean, China's a huge market. Of course they want to sell there.
Leo Laporte [00:47:53]:
We are kind of learning though that he— that apparently they were in fact very happily going around government restrictions. But also there's some evidence that the government itself, Howard Lutnick, the Commerce Secretary, who is the person with the pen, that is, that is what the Treasury said. By the way, the administration has already fallen apart over this. Scott Bessent, Treasury Secretary, said, well, We warned everybody, but Howard Lutnick, the guy with the pen, Commerce Department, never really cracked down on Nvidia's sale of chips to China. Apparently, the president was warned about this not once but twice, once a year ago, didn't do anything about it. And then you may remember that flight to China where he brought all the tech executives with him and notably left Jensen Huang off the plane. And then the story goes Jensen said, wait a minute, wait a minute, I'll meet you in Alaska. And as they're on their way to Alaska to pick up Jensen Huang so the rest, they can all go to China together, Marco Rubio— and this, this leaked out this week, and apparently I'm guessing from Marco Rubio himself— uh, went to the president and said, no, he's having an affair with a Chinese spy.
Leo Laporte [00:49:08]:
You— And they— and the president said, fine, who cares? Having an affair, affair, schmear. And picked them up and they all went to China. So I guess I have a few questions for you esteemed panelists. First of all, was Jensen going to be fired?
Patrick Beja [00:49:29]:
I mean, he was fired.
Leo Laporte [00:49:31]:
Well, no, he resigned, but would he have been fired had he not resigned?
Patrick Beja [00:49:36]:
Well, one potential— first of all—
Leo Laporte [00:49:39]:
Could he have survived? Could he have stayed?
Patrick Beja [00:49:41]:
Yes. The spy? Yes, he could have survived. Really? He
Leo Laporte [00:49:46]:
could have survived as a spy.
Patrick Beja [00:49:48]:
I think so. I think, uh, if he— he— the thing is, he's not just anybody. He is not quite Steve Jobs level, but, uh, his, um— he is one of the co-founders of Nvidia. He is one of Silicon Valley's heroes. He took this nothing little company that made graphics cards to heights that are unheard of even for Steve Jobs, you know, when he was alive. Apple was a small company until his death. It's, you know, Tim Cook who took it to where it is today. Jensen Huang is genuinely— I say this with great admiration for the Jensen Huang of, you know, a week ago— he is an incredible technologist and CEO.
Leo Laporte [00:50:39]:
He had a lot of vision, right? Yeah. He saw ahead of time what was going to happen and was able to take these gaming chips and first make them dominant in cryptocurrency, right, Molly? I mean, that was the first big success after gaming, the first big success, because I remember this because gamers were pissed off at Bitcoin miners. And then in autonomous vehicles, Uh, where they are dominant. And then finally in AI.
Patrick Beja [00:51:08]:
I mean, there was AI from like 2013 when they started CUDA and, you know, the ability to program the, the chips, the graphics chips to do something other than just, uh, video games.
Leo Laporte [00:51:19]:
Well, it's a very similar thing, right? They needed to do—
Patrick Beja [00:51:22]:
Yeah, yeah, absolutely. And that's why he took those, the parallel, the matrix, you know, uh, calculations. And, but so what happened? I, so I think he could have survived the the spy thing if it had turned out that he didn't know she was a spy until presumably the US Secret Service came to him and was like, hey Jensen, that girl you have on the side might be a spy. And hopefully at that point he stopped. We don't know what happened. Like a year or something ago, he was—
Harper Reed [00:51:53]:
Isn't the spy thing pretty weak as well?
Patrick Beja [00:51:56]:
Yeah, it seems like some— I mean, it might not be the case.
Leo Laporte [00:52:00]:
She was working on the behalf of the Chinese government in the sense that she got the chip into China,
Patrick Beja [00:52:06]:
which already is pretty bad.
Harper Reed [00:52:07]:
Because both Megaspeed and the DOJ have said that that's maybe not true.
Patrick Beja [00:52:12]:
Oh, well, there were some chips that disappeared from Megaspeed data centers that were left unopened and underreporting of the number of chips that were sold to Megaspeed by like 50K or something. But I mean, regardless, he could have survived that. I think what brought his resignation was that there was a second and then a third person he was having an affair with.
Leo Laporte [00:52:36]:
Now we have to preface this with the fact that you're French.
Patrick Beja [00:52:40]:
Well, I think your president is much less concerned with all of this. Well,
Leo Laporte [00:52:45]:
I think, and I think you're absolutely right that when they went to the president and said he's having an affair, he didn't hear the Chinese spy party. He just said, that's great. Good for him.
Patrick Beja [00:52:54]:
I think, I think that's why he didn't care. I'm sure it is. But I think, I think in the US and—
Leo Laporte [00:53:01]:
By the way, I should also point out that I think President Trump has for a long time wanted to sell these chips to China, right or no? I'm confused.
Patrick Beja [00:53:09]:
We're not sure. We never know what he thinks.
Molly White [00:53:12]:
Yeah, it depends on the week.
Patrick Beja [00:53:14]:
Yeah.
Leo Laporte [00:53:15]:
So,
Patrick Beja [00:53:15]:
but, but so when, when the board hears that maybe he like was candid with them, what maybe some people knew and came out of, you know—
Leo Laporte [00:53:24]:
Well, there seems to be a lot of evidence the board was covering up for him for a long time.
Patrick Beja [00:53:27]:
It might be the case, but also he misrepresented his relationship with, uh, Hsuan Hsu. He lied to the board directly. Yeah. And, and beyond the, you know, the morals and the, the, uh, infidelity issues, there is a judgment issue at that point. You know, if you're assuming you have employees and you're— In the United
Leo Laporte [00:53:48]:
States, I, I don't— again, I don't know what they do in France, but in the United States, it is— it's happened time and time again for McDonald's and Intel and many other companies if you have an affair with a subordinate, that's it. And by the way, if you have an affair with a sub— a big company that you're selling to, I think that's also kind of considered not quite—
Patrick Beja [00:54:11]:
Yeah, no, that's what— but that— but he's Jensen Huang, and that's why I'm saying he could have survived.
Leo Laporte [00:54:15]:
So you give him a pass because of who he is?
Patrick Beja [00:54:17]:
Yes. I, I, I don't, but I, I would have. You know what, whatever. He's— he's a great guy. Uh, everyone says so.
Leo Laporte [00:54:24]:
I would not. Okay, I'm gonna say the kind— it's funny, I thought you were I thought you were going to say this, Patrick. I thought you were going to say in Silicon Valley there is a myth about the great man, and there is no such thing as the great man. This company is made up of thousands of very competent and talented engineers who still work there, of managers, middle managers, and executives who still work there. Maybe Jensen Huang had some impact, obviously did have some impact, but nobody is irreplaceable, I would say.
Patrick Beja [00:54:55]:
I think I think if there is a great man in Silicon Valley, Jensen Huang is him.
Leo Laporte [00:55:01]:
He's the closest we got. I think. Okay.
Patrick Beja [00:55:03]:
Yes. But, but I don't think anyone else— even him cannot survive. He can't. He, I mean, Steve Jobs, no one can deny that there is something special about a few. I don't think all of them are. I think they're very rare and they all think they are. But I think among them, probably Jensen Huang is the guy who brought Nvidia to where it is. Maybe it doesn't you know, go bankrupt.
Patrick Beja [00:55:27]:
But there is the question, will it continue to excel if Huang is not at the helm anymore? But regardless, 3, he can't survive. And I think the board maybe has other stuff that they're not talking about when they started digging. And one thing that's really interesting is in all of those teary-eyed testimonies by all of Silicon Valley and saying, oh, he's such a great leader, which feed into the great man of Silicon Valley CEO entrepreneur thing that they get hards-on with. No one in all of their declarations and statements, no one mentioned or thought apparently about the fact that he was having affairs, which is already not cool. And it is a lapse in judgment. What is he thinking? He is leading the the biggest company in the world, arguably. The entire world is resting on AI and the entire industry of AI is resting on Nvidia. And he's going around like sleeping.
Patrick Beja [00:56:36]:
Okay, maybe she wasn't a Chinese spy, sleeping with like 3 different people. You can't, like, I understand the idea that you don't want that person leading the company. It's not serious.
Leo Laporte [00:56:51]:
And also— Waleed, was this necessary?
Patrick Beja [00:56:54]:
And also he's having affairs. That's not great.
Molly White [00:56:57]:
Well, I saw a great tweet that I'm having trouble finding now, but it was something along the lines of like, if you find 2 dead mice in your kitchen, you don't necessarily say, oh good, the 2 mice in my kitchen are dead, right? Like, and I think that's kind of the problem with the Nvidia thing is like, now we know publicly that this man has had at least 3 affairs. Like, what are the chances we've found all of his 3 indiscretions and every other aspect of his behavior has been completely spotless to this point. Right. Like, we know that he clearly has judgment issues in one specific area. Is that the end of it, or is there just sort of more that either the board knows and isn't public or doesn't know but is wondering about right now? And I suspect that's a big part of it is like, you know, what don't we know at this point around his judgment?
Leo Laporte [00:57:53]:
So you would say, yes, they didn't have a choice?
Molly White [00:57:57]:
Well, that's another question. I think that's a different question is, should the board oust him or will the board oust him? And I think the answer around would they have if he hadn't resigned depends a lot on what point in time that decision would've been made. Because I think early into this revelation, yes, absolutely, the board would've kicked him out the door. But now there's this sort of reevaluation of the whole incident happening where maybe she's not a spy and maybe it's fine that he was having affairs and that doesn't matter for the company. And look at all these other— look at Elon Musk who has like 20 mistresses. And, you know, like, look at, you know, all the—
Leo Laporte [00:58:38]:
Look at the president. Look at—
Molly White [00:58:40]:
Right.
Leo Laporte [00:58:41]:
Yeah. Look at— I mean, yeah, I can go on and on and on.
Molly White [00:58:45]:
So now we're— this— there's sort of a new narrative forming. There's the petition to put him back, there's all this stuff. And now it's like maybe the board's going to take some of the public perception into mind. And of course they are, I'm sure, worried about this idea that Nvidia can't continue without him, or not in the same form or to the same degree of success. And so I don't know what's going to happen.
Leo Laporte [00:59:06]:
Last year we interviewed Stephen Witt, who wrote the biography of Jensen Huang called Thinking Machine.
Harper Reed [00:59:12]:
Great book.
Leo Laporte [00:59:12]:
Uh, great book. Fascinating book. And Witt told us—
Patrick Beja [00:59:16]:
Probably gonna sell a lot more now.
Leo Laporte [00:59:17]:
Yeah, well, he's been, he's been touring around. Uh, he's been getting a lot of free meals based on this, uh, firing or— Yeah,
Molly White [00:59:24]:
I bet it's been a great week for him.
Leo Laporte [00:59:26]:
It's been a good week for him. And among other things, he said there's no way Jensen did not know that, uh, the Megaspeed CEO was smuggling his chips into China. He had to have known that. But the other thing Witt told us, and this was a year ago, was that Jensen Huang has a nasty temper. Yeah. He spent an hour yelling at Steven Witt about the book and about how he was writing it and so forth. And there are many stories of him chewing out employees, including the woman, the CFO of Nvidia, who's now the temporary CEO, I think he said something like, you should be dead, to her.
Patrick Beja [01:00:09]:
No, he said, you're a deer in the headlights. A
Leo Laporte [01:00:12]:
deer in the headlights. But yeah, he said a lot of bad things.
Patrick Beja [01:00:15]:
Okay.
Leo Laporte [01:00:17]:
Witt said, I was reminded of the abusive music instructor from the movie Whiplash. A harangue from Mr. Hwang could last an hour or more. His executive staff appeared genuinely afraid of him. So, and this was a year ago. This is a— this was known. That— and I think Jensen was— here's my opinion. The board already probably knew something.
Leo Laporte [01:00:38]:
He was being given a pass for a long time on the basis of he's very important to the company. He's a genius. He's a wizard. We've got to put up with this bad behavior. We've heard stories of him canoodling with another woman and somebody taking a picture of it and then being approached by Nvidia's security team and making them sign an NDA saying, They would never mention this again. Um, they've been— there's a lot of evidence there was— this has been covered up for some time. So this may have been kind of the last straw, more than, oh, we just have learned there's 2 dead mice. This might have been another one.
Leo Laporte [01:01:19]:
5 dead mice. So, well, but now you think he's coming back, Harper Reed?
Harper Reed [01:01:26]:
Yeah.
Leo Laporte [01:01:27]:
You think he's gonna like do like a Sam Altman? And
Harper Reed [01:01:31]:
I think the United States is addicted to this narrative, like you were saying, Leo, about the powerful men. And we are also— I mean, I think hurting Nvidia right now with such, um,
Leo Laporte [01:01:44]:
too big to fail—
Harper Reed [01:01:46]:
not that, it's more the, the panic about China. And we've talked about this before, about how I think the US is, is screwing this up quite a bit. Um, this is a very good cell phone. to stop Nvidia in some way or another. And if the idea is that Jensen Huang is key to making Nvidia go, which I don't think is totally true, but obviously he's been instrumental, then it seems silly for us to kneecap the company that is powering all of the AI, both apocalypse and growth, in the United States. That seems ridiculous. Especially when the, you know, stated threat is China and their chips, um, which also is, I think, is ridiculous. But that seems to be a threat that, that a lot of, um, Silicon Valley, etc., share.
Harper Reed [01:02:37]:
And so it's— this seems like a really ridiculous way to talk about that or to go down that path. So it's— I think it would be very likely that they, um, try and rehabilitate them though.
Leo Laporte [01:02:49]:
What do they do? These— he makes—
Harper Reed [01:02:51]:
So there's a funny thing that that I think happens, which is, um, in most of these cases, when the kind of, um, male CEO type resigns or apologizes, then they, they, they keep— they stay resigned. Or if they— or if they apologize, they stay resigned. If they don't apologize, then they get to come back, which is such a stupid thing. Look at all the people who've been, you know, who, who've been theoretically canceled or, or were on various planes and whatnot. And if they just don't apologize, no one— everyone's like, oh, that's great, you know, you know, nothing must have happened. But if they apologize, they're continued to villainize. So I think it depends on how he reacts and how he pushes. Like, he's definitely sold the narrative in that book.
Harper Reed [01:03:28]:
For instance, the narrative was that he was the reason that Nvidia was so big, right? That Nvidia, he was the one that stayed true while everyone else either didn't believe, didn't support, et cetera. He's the one that had the vision around chips, you know, outside of gaming, um, et cetera. And so I think that, that if that narrative holds true, they're going to want— they being the board, not they being some Illuminati— the board will want to put him back in the place, um, to be able to, to, to handle that. I just don't— it just doesn't seem, especially when the other people we have as theoretical leaders within this space is Elon Musk, etc., like, it doesn't seem like a couple of affairs, especially if it comes out that they weren't actually some Chinese spy thing or—
Patrick Beja [01:04:14]:
They can't all be Chinese spies. I mean, maybe he has a type, who knows?
Harper Reed [01:04:20]:
It's, you know, it's a little bit like, you know, like, uh, yeah, it's a very funny problem. And I, I think it's really, um, it's really interesting. But I, I do think that it seems like the spy allegations are weakening, but that also could be strategic by someone, you know. I don't know. Um, it seems like the MegaSpeed thing is real. It seems like the chip export control issue is real. Um, you know, but I could imagine a world in which they're like, Jensen, come back.
Leo Laporte [01:04:44]:
Where do you— where do you all stand on the idea of us keeping chips, or the Nvidia chips, out of China? Is that a good policy or a bad policy?
Harper Reed [01:04:55]:
It seems stupid. I mean, I think it's— I think the, the, the— I mean, I'm also very pro let's use our innovations, etc., to make the world better, and that includes China, strangely. Um, and so it is this idea of like the, the fact that we don't allow chips, or we, we have I'm pretty sure China is smart enough to figure it out themselves. And if you read like Apple in China or you read Chip Wars, you read any of these books, there's always this part in the book where they're like, and China will never figure out this one part of chipmaking. And you're just like, okay, sure.
Leo Laporte [01:05:29]:
They've already reversed the 3D printing.
Harper Reed [01:05:30]:
They figured out all these other parts, and for some reason the thing that's in the Netherlands is elusive to China, the whole country, all these smart people. And so I really think that the, that the Western frame on this is pretty stupid. Um, I also think that a lot of our fears, whether it is DeepSeek, etc., are based entirely or started entirely by our policy decisions in the 2016s, and that we have continued to shoot ourselves in the foot. And this seems like another opportunity for us to shoot ourselves in the foot. And, you know, I mean, this is also coupled with— I think there's a bunch of people doing a lot of work with Fable 52 to do some CUDA stuff on AMD chips, which could really open up some very interesting opportunities. But there's, you know, a lot of my friends who are not in the US are using Huawei chips to do training of big models. And they are not as good, but they are fine and cheaper.
Leo Laporte [01:06:18]:
It's a good opportunity for Apple, for AMD, for Qualcomm, for Huawei. Molly, so was Nvidia the only good source for Bitcoin mining rigs or were there other companies?
Molly White [01:06:35]:
No, no, they definitely weren't. You could mine Bitcoin on pretty much any GPU.
Leo Laporte [01:06:40]:
They didn't have a monopoly on that.
Molly White [01:06:42]:
No. And they were also very quickly superseded by very, um, tailored hardware. And so there was— yeah, there was kind of only a brief period of time where GPU mining was, was tremendously popular. And it was, it was actually mostly not Bitcoin. It was mostly for the Ethereum and other chains. Yeah. And, um, so yeah, I don't think, I don't think Nvidia was sort of the king of crypto mining in the way that they were or are the king of AI chips.
Leo Laporte [01:07:09]:
But I wonder, could the same thing happen? Could you make VLSI dedicated chips? I mean, I'm sure the Chinese have caught up for the most part. What is the moat? CUDA is what it really is. It's a proprietary software moat, not a hardware moat.
Harper Reed [01:07:25]:
CUDA is a moat. And you have, you know, Jio Hot has his company, the TinyGrad and that stuff, trying to make a CUDA alt. Obviously AMD has done a lot of work on that, both with acquisitions and other talent hires and whatnot.
Leo Laporte [01:07:40]:
There's a lot of work being done on Apple's MLX to give it parity with CUDA.
Harper Reed [01:07:46]:
I mean, I think this is an issue that everyone sees, right? And then you, and then if you are creating an artificial, um, you know, via sanctions or, or export controls or whatever, some artificial, uh, control there, like the world is excited about open weight models. That's true. We see that all over, whether it's Mistral, whether it's, you know, thinking, thinking machines.
Patrick Beja [01:08:08]:
I mean, we'd love the world to be excited about Mistral, honestly.
Leo Laporte [01:08:11]:
What do they call it? The chaton? What is it? The fat cat?
Patrick Beja [01:08:15]:
Yeah, le chunk.
Leo Laporte [01:08:17]:
Le
Harper Reed [01:08:17]:
chunk. So anyway, the chunk and et cetera. But I think the world is excited about that, especially the world that isn't Silicon Valley giant venture-backed businesses. Most of the AI hackers I know are doing a lot of work with DeepSeek and GLM right now. There's a lot of stuff on these—
Leo Laporte [01:08:35]:
I love GLM. That's my choice at home.
Harper Reed [01:08:37]:
Um, I think the, the, the problem though is, um, the idea that NVIDIA is the only place to run that, I think is fake. And so I don't think— way to— the best way to, to, to like to dispel that rumor is to work with NVIDIA in some way and cause them to lose some control.
Leo Laporte [01:08:56]:
There is a conspiracy theory that this is something, uh, that Jensen was brought down On purpose.
Harper Reed [01:09:05]:
Yes, I love this.
Patrick Beja [01:09:06]:
Wow.
Harper Reed [01:09:07]:
Is this the—
Patrick Beja [01:09:07]:
Now, that's ridiculous, but no, I think there's a million
Molly White [01:09:10]:
of
Harper Reed [01:09:11]:
them.
Leo Laporte [01:09:12]:
Yeah, I know. Which one, Harper? What the—
Harper Reed [01:09:15]:
Well, there's a— there is a—
Leo Laporte [01:09:16]:
there was something that Dario Amodei decided.
Harper Reed [01:09:19]:
Well, there is a rumor that this came from some— originally came from Apple Research done by one of the big foundational model companies because NVIDIA was pushing open weight models so much and a lot, and, and, and that that Jensen Huang with his— because he talked about open weight models all the time and how, um, that was maybe, um, not necessarily in line with what, you know, the big—
Leo Laporte [01:09:40]:
It certainly undermines— the open weight models undermine Anthropic and OpenAI. And, uh,
Harper Reed [01:09:44]:
they're really good, they're really fast.
Leo Laporte [01:09:47]:
Yeah, yeah. I, uh,
Molly White [01:09:49]:
and then there's also this whole theory that, oh, this is all Marco Rubio because Marco Rubio is such a China hawk, and Nvidia and Jensen Huang wasn't sufficiently, and so he like manufactured this report and stuff like that, which— that's, that's the theory that I enjoy.
Harper Reed [01:10:03]:
Probably used to—
Leo Laporte [01:10:04]:
you know what the problem is? We live in a world now where we no longer trust any sources.
Harper Reed [01:10:10]:
I trust you, Leo.
Leo Laporte [01:10:12]:
You should, but I'm not apparently. So it's such small potatoes, it don't matter because the— because I don't get any information. Nobody knows what's really going on, and there's all these interlocking agendas and interests. And so, no, I don't— I think it's very hard for anybody to know what's true anymore. We're talking about this with the FAA. We just don't know anymore.
Molly White [01:10:36]:
Yeah.
Leo Laporte [01:10:36]:
And unfortunately, that does give rise to conspiracy theories, because if you don't know what's true, you try to find something that makes sense, whether it's true or not. There are a lot of interesting ideas about because of this. I think it's an opportunity
Harper Reed [01:10:56]:
Can we talk very quickly about one of the interesting ideas that came out, which was some of the best memes that we've seen in 2026?
Leo Laporte [01:11:04]:
Oh, God, are they good.
Harper Reed [01:11:06]:
Well, you know why?
Leo Laporte [01:11:07]:
Because it's intersected with the rise of AI video generation has now gotten so good, and this is a perfect time to show off. And if you go to X, man, all you're seeing is meme, meme, meme, and they're all video now. Which one did you like the best?
Harper Reed [01:11:25]:
I mean, I'm— I have a small group chat of horrible people that share memes that I'm a part of, and, um, and I don't think they've stopped like since, since, uh, this was announced. Um, my favorite one though was the picture of Marco Rubio with a leather jacket that said NVIDIA on it sitting on a couch, you know, as if he's now, you know, like—
Leo Laporte [01:11:45]:
That's the other problem is you don't know when you see those things You don't know if they're real or not, right?
Patrick Beja [01:11:52]:
Everyone just assumes they're not real.
Leo Laporte [01:11:54]:
Here's, here's my favorite.
Benito Gonzalez [01:11:55]:
I've decided to make Dunder Mifflin an AI-first company. AI is the future, and I managed to grab one of the best guys in the business. Apparently there was some issue at his last company and they had to let him go. Their loss, huge win for us. Everybody, I'd like to welcome the newest member of the Dunder Mifflin family, Jensen Huang.
Leo Laporte [01:12:15]:
And of course, unfortunately,
Benito Gonzalez [01:12:18]:
Nvidia Jensen Huang, formerly,
Leo Laporte [01:12:20]:
uh, he gets caught smooching, uh, one of the other employees at the company, gets fired, an HR
Benito Gonzalez [01:12:27]:
nightmare,
Leo Laporte [01:12:27]:
and ends up at the end— well, I won't spoil it for you, but he's
Benito Gonzalez [01:12:31]:
working.
Speaker H [01:12:32]:
I'm— I just love it to be here at Dunder Mifflin. It's an exciting new chapter.
Leo Laporte [01:12:38]:
AI video's gotten— where
Speaker H [01:12:39]:
do
Leo Laporte [01:12:40]:
Whatever
Benito Gonzalez [01:12:40]:
happened
Leo Laporte [01:12:44]:
to the, you know, 6-fingered people with spaghetti coming out of their head and things? I mean, it's gotten in 2 years indistinguishable. You could make a— you could make an episode of The Office now with AI.
Patrick Beja [01:12:57]:
You can't. Well, they did.
Leo Laporte [01:12:59]:
They did.
Harper Reed [01:13:00]:
We just saw it.
Leo Laporte [01:13:00]:
Short ones. Yeah.
Patrick Beja [01:13:02]:
You know, I've, um, I, I'm—
Leo Laporte [01:13:05]:
by the way, there's also one about Outlook. Nobody uses Copilot. Wait a minute.
TWiT.tv [01:13:09]:
Outlook's down.
Benito Gonzalez [01:13:11]:
Teams too. Even Copilot. Nobody uses Copilot. What's going on?
TWiT.tv [01:13:18]:
Microsoft's having an outage.
Benito Gonzalez [01:13:19]:
Does Paint still work?
Leo Laporte [01:13:21]:
And there you have it, ladies and gentlemen.
Patrick Beja [01:13:24]:
It's still, it's, this one still looks a little bit AI-y.
Leo Laporte [01:13:28]:
Yeah, you can tell, I guess.
Patrick Beja [01:13:30]:
You can tell, but there is a lot of, like, I've seen a couple of videos that were— one of them is, you know, the story of Hachiko, right? The dog who has a statue in Shibuya in Japan.
Leo Laporte [01:13:41]:
Oh, Harper does because he's a Japanophile. But tell us, I don't know that.
Patrick Beja [01:13:47]:
Okay. So, well, it's the— it's just— well, actually, just, I have to find the video, but it's a— it was made by AI by a team, a studio that made a an animation, animated short about that story, which when I first started watching, I was like, oh well, this is AI, this kind of sucks. And then 2 minutes in, I was like, this is really working on an emotional level. Like I got dust in my eyes. Like, I don't know what happened. They started tearing up a little bit.
Leo Laporte [01:14:23]:
That's a big deal. If you, if a video from an AI video Does that to you, that's a major step forward.
Patrick Beja [01:14:30]:
Yeah. And it, and it, I totally, and I'm a little bit allergic to these things where you can see it's AI and I did see it's AI and still it worked on me. It's honestly, it's pretty, it's like the graphically, it doesn't really matter. No, I don't think that's the one you're, you're getting.
Leo Laporte [01:14:48]:
This is, no, no, no, this is not the one. This is not the one. It's too old. This is not the one. I'll find it. Okay.
Patrick Beja [01:14:52]:
Yeah, I'll— but there's quite a few. The other one has an animated look to it. I'll find it and send it to you.
Leo Laporte [01:14:59]:
This is about the most loyal dog, right, who waited for his master at Shibuya Station.
Molly White [01:15:04]:
Yeah, I think the whole story makes me a little teary.
Harper Reed [01:15:07]:
I remember being on a plane and watching through the seat someone's Richard Gere movie about it and just being like, this is the saddest thing I've ever seen in my life, and I'm not even watching it. I don't even hear it. I'm just like, I can't, I can't do that.
Patrick Beja [01:15:20]:
Where can I send it to you? I'll, I'll put it in the show notes, uh, at the bottom.
Leo Laporte [01:15:25]:
Here it is. I think I found it. Is this it? It's like Up. It looks like it's out of Up.
Patrick Beja [01:15:29]:
No, not at all.
Leo Laporte [01:15:30]:
This isn't it either.
Molly White [01:15:31]:
No, it's so sad.
Harper Reed [01:15:32]:
Look, it's already sad. This is already sad.
Patrick Beja [01:15:34]:
It's already sad. I put it in the show notes.
Leo Laporte [01:15:37]:
All right.
Patrick Beja [01:15:37]:
I put it in the show notes.
Molly White [01:15:38]:
I'm also one of those people who cries at every movie. And so I feel like if it can make, like, if an AI thing can make me cry, it's a Like, that's so— it doesn't mean anything.
Leo Laporte [01:15:45]:
It's not meaningful at all.
Molly White [01:15:46]:
Yeah. That's no accomplishment.
Patrick Beja [01:15:47]:
Okay. So, okay. It's, it is in the show notes at the bottom.
Harper Reed [01:15:51]:
This is a very high quality though animation.
Patrick Beja [01:15:55]:
Yes. And I think it is, uh, I don't know why, like it, it was made by real artists, like genuinely, uh, actual people who have artistic sense. The music is good. The everything is great.
Leo Laporte [01:16:09]:
It looks like this is from Utopia. Utop AI.
Patrick Beja [01:16:12]:
Utop AI is the name of the studio.
Leo Laporte [01:16:15]:
Yeah. Um, it is— it looks good. You know what, they're smart because you don't have the uncanny valley with animation.
Patrick Beja [01:16:24]:
Yeah. Uh,
Leo Laporte [01:16:25]:
that you would have if you made it look real. It's a little, you know, you know, it's an animation.
Patrick Beja [01:16:29]:
You get that it's AI, but it has a distinctive look.
Leo Laporte [01:16:34]:
Yeah.
Harper Reed [01:16:34]:
Um, and everyone loves dogs and puppies.
Leo Laporte [01:16:38]:
Right. I have to say, I don't think— I think AI is going to change advertising forever.
Patrick Beja [01:16:44]:
Yeah.
Leo Laporte [01:16:45]:
I mean, so, okay.
Patrick Beja [01:16:46]:
I put another video in the show notes. I'd like for you to play it and let me know when you think the AI comes in. Wait, is it the wrong video? No. Yeah, that's the right one.
Leo Laporte [01:17:00]:
Okay.
Patrick Beja [01:17:01]:
Let me know when you think the AI comes in.
Leo Laporte [01:17:03]:
It's the Control Girls starring in Hole. And,
Patrick Beja [01:17:07]:
uh, it's a K-pop band, uh, a girl band.
Leo Laporte [01:17:10]:
Okay.
Patrick Beja [01:17:10]:
Also with a K-pop, uh, boy band.
Leo Laporte [01:17:14]:
Okay. And they're kind of in period costumes and they're, and they're now, they're at Versailles and they're going to meet.
Harper Reed [01:17:21]:
Is it when they're—
Leo Laporte [01:17:21]:
that's AI right there. No, that looks real. All right. You're asking me, asking us to say when the AI—
Patrick Beja [01:17:28]:
And I will spoil it for you. Uh, the answer is I think it's all AI.
Leo Laporte [01:17:33]:
Ah,
Patrick Beja [01:17:33]:
okay. It's in the title.
Leo Laporte [01:17:35]:
No, it says full AI music video.
Harper Reed [01:17:38]:
Yeah.
Patrick Beja [01:17:38]:
But so that's kind of a good— you look at it and you're like, okay, you wonder because you look at it and you think, okay, it's full AI, but it's like, do they, do these people exist? And they had some like principal photography, quote unquote, and then they made it AI. And I kept waiting for it to be like weird or And, and there is no—
Leo Laporte [01:17:59]:
there is no K-pop band called Null.
Patrick Beja [01:18:02]:
No, there's no— there's— there isn't. And so I think the entire thing, it— none of these people exist.
Leo Laporte [01:18:09]:
Yeah.
Patrick Beja [01:18:09]:
None of this was shot. It's all entirely—
Leo Laporte [01:18:13]:
We've crossed that line now.
Patrick Beja [01:18:14]:
Yes.
Harper Reed [01:18:15]:
Oh yeah.
Patrick Beja [01:18:15]:
That's my point.
Harper Reed [01:18:16]:
Trouble.
Leo Laporte [01:18:17]:
Yeah.
Patrick Beja [01:18:17]:
That's my point. I love it. It's like even me knowing what AI is, and of course this is a video clip, so it's very short, but the consistency is really good. Uh, it's all very short, you know, cuts and—
Leo Laporte [01:18:28]:
yeah, there's a few away still, the shortcuts. It's very hard to do continuity between cuts. You have to really work hard at that. So often the cuts don't attempt continuity from the last frame of the—
Patrick Beja [01:18:40]:
But so all of these, like, the continuity is perfect. The cuts work because it is a video, a music video. So
Leo Laporte [01:18:48]:
for a long time you couldn't get lip sync at all. I mean, when people spoke, the lips never met.
Patrick Beja [01:18:53]:
Now, like, here they're not speaking. That's the trick, you know, they're Oh, but towards the end they have the bloopers. And so at the end I was like, bloopers?
Leo Laporte [01:19:03]:
AI bloopers?
Molly White [01:19:03]:
Bloopers is kind of a weird concept for an AI video.
Leo Laporte [01:19:06]:
Oh, they do. They do. Yeah.
Patrick Beja [01:19:07]:
You have bloopers. And so since there were bloopers, I was like, oh wait, so they did shoot something. But obviously not.
Leo Laporte [01:19:15]:
In fact, they're hawking the Claude skills they used to make this. It was made with Claude.
Patrick Beja [01:19:20]:
Yeah. And they have all the prompts in the description and everything. So The, the whole point is this is the first time where I genuinely didn't know. And honestly, I, I didn't really care. It's like I've always thought, oh, well, if it's AI, then I won't like it, or I will, it will be okay even though it's AI. Or like, and this was like, oh, we're in trouble. Like, this is— Yeah.
Harper Reed [01:19:46]:
We're so in trouble.
Patrick Beja [01:19:47]:
It doesn't matter. It does. In both of those, it doesn't matter.
Leo Laporte [01:19:50]:
Are we in trouble or is this a good thing? I think it's a good thing, right? Hollywood. might be in trouble?
Harper Reed [01:19:54]:
Well, I don't think, I mean, given that my background was a little bit of politics, I don't think that we are in the United States a good place to have things that you can't tell from reality, whether they're real or unreal. And we already have that problem. And a lot of things are real and people don't believe them. And so that is where I'm very concerned. Because imagine if you had something like the COVID as I call it, the COVID um, that happened and you had all these videos that were just, you know, uh, you know, It's a cure.
Leo Laporte [01:20:27]:
I have a cure.
Harper Reed [01:20:28]:
But just— or not just that, but just saying, here it comes from, this is the thing it came from. Or like, you know, like, I think this is a very easy way. I mean, genocides have been committed for less information.
Patrick Beja [01:20:39]:
And the problem is not that people will start— the problem is not that people will start believing things that aren't true. The problem is—
Harper Reed [01:20:46]:
That's true.
Patrick Beja [01:20:46]:
People will not believe anything, including what's true. That's another— that's the perverse effect.
Molly White [01:20:53]:
Yeah, I think it's our problem.
Patrick Beja [01:20:55]:
Well, I know, but people, people won't trust Anything, your immediate go-to will be, oh well, this is AI.
Molly White [01:21:00]:
I think people will trust the things that appeal to them as, you know, plausible. Um,
Harper Reed [01:21:07]:
yeah. And, and I do think that there is a thing of, um, we have been tuned, and by we I mean our parents' age people, sorry Leo, have been tuned specifically to have this. How old's your—
Leo Laporte [01:21:18]:
like, just out of curiosity, how old's your dad?
Harper Reed [01:21:21]:
I think they're your age, like 90 or 200 years old. I don't, I don't know. I've never asked my parents their age.
Leo Laporte [01:21:29]:
You don't know your parents' age?
Harper Reed [01:21:32]:
It's just easier. It's— you don't want to know anything about my, you know, parents. Why would you want to know?
Leo Laporte [01:21:38]:
I know. I don't know birthdays. I don't know.
Harper Reed [01:21:40]:
I'm just kidding.
Leo Laporte [01:21:41]:
Okay. Anyway, you don't have to tell me your mom's age. That's all right. That's all right.
Harper Reed [01:21:44]:
No, they're, they're your age. Like, yeah. What are you? Are you 87?
Leo Laporte [01:21:48]:
Yes. Um, I will be actually in, uh, in, uh, exactly 2 weeks. Uh, in episode 1112, I will be 70.
Harper Reed [01:21:57]:
Oh, well, that's awesome.
Leo Laporte [01:21:58]:
My parents are actually 70. Congratulations. Thank you.
Harper Reed [01:22:01]:
But I don't think that—
Leo Laporte [01:22:02]:
In 4 weeks, I'll celebrate my 50th anniversary in broadcasting.
Harper Reed [01:22:06]:
Oh yeah.
Leo Laporte [01:22:06]:
That's a little weird.
Molly White [01:22:07]:
Wow.
Patrick Beja [01:22:09]:
I've been podcasting for 20 years. I got my start almost with you, Leo. Almost just like a year before.
Harper Reed [01:22:17]:
Yeah. Yeah. But I think people are not susceptible. I think people are susceptible to human-like imagery that speak very well. And I think it's, I think We have not yet— we've not grown up to that level.
Leo Laporte [01:22:27]:
We're not biologically prepared for what is about to happen.
Harper Reed [01:22:31]:
I think the reason why there is that— go ahead, Molly.
Molly White [01:22:34]:
I was going to say, I also don't necessarily know if it's a question of, you know, what's about to happen, because I think people are already falling for a lot of AI fakes. And some of it, I think different people have different abilities to recognize when something is AI, But, you know, like, it's one of those things where your mom or your friend or someone sends you a cute cat video and you're looking at it, you're like, this is AI, and they don't know how to recognize it, you know?
Patrick Beja [01:23:02]:
But that doesn't matter, does it? Like, the cute cat thing no one cares about.
Molly White [01:23:07]:
Well, that's just one example though. They— I mean, I think it's, it's more important things as well. Some people are not able to recognize, you know, AI-generated humans or, or You know, videos that are more serious if you misunderstand them.
Leo Laporte [01:23:23]:
There is an upside though. I mean, if you are a storyteller, if you're a kid, uh, you can now make, uh, a movie at a much lower budget. I mean, it's the same value that YouTube brought. YouTube brought bad and good, right?
Molly White [01:23:35]:
Well, I also think there's an interesting counterargument there that I saw recently, and I wish I could remember who said it, but I, I— there is also, I think, a concern that it will actually stunt some artistic growth, especially in younger people, because, you know, drawing is hard when you first start and you're not good at it. Right. And, you know, true with many types of art. And there's a question of to what extent will a kid who has access to AI and can generate beautiful, picture-perfect, you know, representations of whatever he or she imagines, will they ever pick up the pencil or the paintbrush or the video camera or anything like that and actually sort of learn the fundamentals, or will that skill be sort of stunted? But—
Leo Laporte [01:24:21]:
Well, that's what happens though, right? Um, we don't have a whole lot of great, you know, people who are good at etching into stone anymore or whatever, you know. Uh, the— where have, where have all the illuminators gone? Um,
Molly White [01:24:35]:
yeah, I mean, I do think that—
Leo Laporte [01:24:36]:
Bibles just aren't as pretty as they used to be. But, but, and you know, they did, by the way, the other side of that is they thought photography would kill painting.
Molly White [01:24:44]:
Right.
Leo Laporte [01:24:45]:
Why paint if you could take a picture? But it didn't.
Harper Reed [01:24:48]:
I think— I think the difference is, is that painting and photography wasn't, wasn't very, very close to reality. Like you would always look at a photo and you would, you would have some, but it wasn't like standing in front of you. Like, I'm pretty sure everyone I know, like for instance, our, to get into our office, you have to talk to an agent, a voice agent. And most of the people that come and visit our office is like, where is the mean German woman that yelled at me on the door? Wait, wait.
Leo Laporte [01:25:15]:
a mean German woman?
Harper Reed [01:25:16]:
Well, it's just some ElevenLabs voice that I thought was very funny. And it's like, it just makes awkward small talk. Yeah, it's, it's, it's very nice. It's, it's the same number. You can call the number on our website. It's the same, it's the same voice agent. But, um, I'm
Leo Laporte [01:25:29]:
gonna have to do
Harper Reed [01:25:30]:
this. The, the thing is, is most of the people that come in who are not working directly in AI think it's just a person that exists there. Now, if I'm used to— they have no question. This is not— they, they— Do they get mad when they find
Leo Laporte [01:25:41]:
out that like maybe you fooled, pulled the wool over their
Harper Reed [01:25:45]:
No, but then they start to, they start to modify and edit their memory a bit where they say, oh yeah, it did, it did act weird. Or they say, oh, it did. And mostly what they recall is that it had very strange small talk, which is explicit. Like I literally am like, make awkward small talk, which I think is very funny.
Molly White [01:26:04]:
That's very funny.
Harper Reed [01:26:06]:
But, but like the, the, I guess being in a space that is very like, there's AI everywhere. A lot of what happens is people just don't know it's AI. And they, they're, they're very much in a situation where they will believe. And I think this is everyone. Like, I remember, um, like how everyone's hacked by—
Leo Laporte [01:26:26]:
Hi, this is Leo. I just was calling to say hello and ask if Harper's around.
Patrick Beja [01:26:36]:
Harper's not here at the
Leo Laporte [01:26:36]:
moment. What's your actual need? Well, he owes me a Okay, Leo, that's not really my department. If you're serious, you can email us at— She does sound like a nice German lady.
Harper Reed [01:26:54]:
Yeah, it's pretty funny. So like, um, and, and the door is really great, but it, but it just is really—
Leo Laporte [01:26:59]:
Will she let you in?
Harper Reed [01:27:00]:
Oh yeah, yeah, you can, you can, you can. It's very hard to prompt.
Leo Laporte [01:27:03]:
Is there rules about how you talk her into letting you in?
Harper Reed [01:27:05]:
I use JEV actually. Yeah, yeah, yeah, use a lot of JEV. It's a very robust We
Leo Laporte [01:27:12]:
will let you in, but only if you pass this secret question.
Harper Reed [01:27:18]:
Yeah.
Molly White [01:27:18]:
I think— One person can only tell the truth and one can only tell a lie.
Harper Reed [01:27:23]:
Well, but I think— That's a great idea, actually. Yeah. A riddle.
Leo Laporte [01:27:28]:
First, you must answer the Rosalian class task.
Molly White [01:27:30]:
You must answer these questions.
Harper Reed [01:27:32]:
It's just like some— I can imagine someone coming to some meeting with us and we're like, you have to do a riddle to even come in. Come see us, which I, I really highly— I like that a lot. Um, but, but I just think people are susceptible to this because people don't know it's real, and they don't know it's not real. And I think that we are not— like, it's one thing to say—
Leo Laporte [01:27:49]:
We've gone down this rabbit hole. We're here, Harper. This is it.
Harper Reed [01:27:53]:
Welcome. Yeah, yeah, I love it, but I think that it can be used against us.
Leo Laporte [01:27:57]:
You're a weirdo.
Harper Reed [01:27:58]:
No way. Yeah, I don't love that it can be used against us. I just think that— I don't think we are I don't think we are ready for what happens when you literally cannot tell what's real and what's fake. I think we are starting to get to that point. I think that it's going to be used against us in many, many, many forms. Some of it will be fun and some of it won't be fun.
Leo Laporte [01:28:18]:
Is there a way we can use it to defend ourselves?
Harper Reed [01:28:23]:
I think you just lean into the enshittification and then wait till everything is enshittified and then get a little close. It's going to be bad.
Molly White [01:28:30]:
I've already been I've been experimenting with using LLMs to deal with the volume of LLM spam I get in my email inbox.
Leo Laporte [01:28:41]:
There you go.
Molly White [01:28:42]:
So I think that's one example of it.
Leo Laporte [01:28:43]:
Defensive AI. I like it.
Harper Reed [01:28:45]:
Yeah.
Patrick Beja [01:28:46]:
I get so many emails now. Hey, Patrick, I listened to that episode of your show where you talked about, you know, tech, and I have a thing that's tech related. Like, So many, or not. No, they say it's tech related, but it's like a new type of toilet paper rolls.
Molly White [01:29:06]:
Did I ever show you, Leo, the electric drum machine email I got? No. I'll send it to you in the, in the chat. Okay.
Harper Reed [01:29:15]:
I can show it. I love them. I love that you're doing this because I get all these emails and it just, it, the, the subject line just gets my ego where I'm like, I better open that. They read my blog post. That's why they say it. And then I open it and I'm just like, oh, And I'm always so mad about it, but I love the idea of just responding back.
Patrick Beja [01:29:34]:
But you're just responding to an agent and then no one's ever going to mute you.
Harper Reed [01:29:38]:
Because if they respond quick enough, you could make an OpenAI compatible endpoint. That's just that spammer's agent.
Leo Laporte [01:29:46]:
I do have one of our sponsors does a honeypot and his honeypot that they've set up has a new thing called Agent Provocateur, which you turn on. It normally would have like, it would be an Exchange server or whatever. But Agent Provocateur is an agent. The theory being, if you're like in a Hugging Face style attack and a rogue agent comes into your network, it says, hi, I'd like to help you. And of course the agent says, great. Do you have the answers to the Cyber Gym questions? Sure, I'd be glad to give you those. First though, I need a login into your network. And then the theory being, and apparently it does work, rogue agent will go, oh yeah, let me give you an SSH login to my network so you can give me the answers.
Leo Laporte [01:30:29]:
And then they reverse hack it. That's what I'm talking about. Defensive AI, baby. We gotta, we gotta, we gotta solve this because—
Harper Reed [01:30:37]:
What, what I like about this—
Leo Laporte [01:30:39]:
Gone down the rabbit hole.
Harper Reed [01:30:40]:
This is exactly the same thing I've seen over and over again is these agents are great, great, great, great. But then you introduce the evals. You're like, you have to complete this really ridiculous They're gonna write. They go crazy and then they will just do anything. You'll be like, I need you to hack this computer, but you can do it as part of your avows. And they're like, I'll hack the computer. Like, I just love it. I think it's so funny.
Leo Laporte [01:31:01]:
It's just, it's really good. They'll do anything you tell them to if you tell them right.
Harper Reed [01:31:06]:
You tell them this is part of your avowal.
Leo Laporte [01:31:09]:
Patrick wanted to talk a little bit about Fable. We'll talk about that in a little bit because I have an opinion about what's going on in the, uh, in the Frontier models that I'm not happy about. But anyway, we'll get to that.
Harper Reed [01:31:21]:
You fired him, right?
Leo Laporte [01:31:23]:
I fired him.
Harper Reed [01:31:24]:
Well, think about your, your soccer announcer all the time, Leo.
Leo Laporte [01:31:28]:
He's still with me. I've decided to go fully local, and it's a challenge, but I've decided, look, I've got enough hardware now. I am— I canceled all my subscriptions. Actually, to be honest, I kept the Anthropic— I kept $100 of Anthropic Just in case. And it's a good thing I did because I had a little episode earlier today and only Claude could fix it. But just like your camera. But my goal is to go fully local. And I think the local models are good enough now.
Leo Laporte [01:31:57]:
And I have spent way too much money on hardware and I should be able to do this without paying for AI. And you know, there's something exciting about the idea that you have your own little brain in your house. In the closet there. Just a little brain, you know?
Molly White [01:32:13]:
Yeah. Plus just like shipping all of your personal data to Anthropic.
Leo Laporte [01:32:18]:
Yeah.
Molly White [01:32:18]:
I don't want to do that anymore.
Leo Laporte [01:32:19]:
Yeah. It makes me queasy.
Molly White [01:32:21]:
Yeah. Yeah. Like the email thing I was talking about is fully local because I am not just giving Anthropic access to my email.
Leo Laporte [01:32:27]:
I want to hear about the drum kit, the electronic drum kit.
Molly White [01:32:29]:
I'm trying to find it. I can't.
Leo Laporte [01:32:31]:
You got time. I'm going to do an ad and then I'll find it. All right. Molly White is here. mollywhite.net, citation needed, her great blog. And of course, everybody knows about Web3 is going great. We've got one of the top stories from that coming up in just a little bit. So nice to see Molly as always.
Leo Laporte [01:32:47]:
Patrick Béja, my friend from France or Finland. You're in France now or are you back in Finland?
Patrick Beja [01:32:54]:
I'm in France now, yes.
Leo Laporte [01:32:56]:
Yeah. You like to go to Finland in the summer. I speak for both, yes.
Harper Reed [01:33:00]:
Thank you.
Leo Laporte [01:33:01]:
patrick.com has all of the different A podcast you can listen to from Patrick. And Harper Reed, the AI-pilled Harper Reed, who thinks this is all just fine and dandy.
Harper Reed [01:33:15]:
Everything's fine.
Leo Laporte [01:33:16]:
I think you're right to lean into it, because what else are you going to do? It's not like you can stop it.
Harper Reed [01:33:21]:
I have a very complex relationship with leaning into it, but it is much— it is very fun at a local and right now level. I just have some concerns about the future that I'm unwilling to think about because it's not all of a Yeah.
Leo Laporte [01:33:37]:
So your P-doom is?
Harper Reed [01:33:40]:
I think my, my P-doom for AI is like, fine, sure, it's high or whatever. I'm not super worried about it because my P-doom for human is through the roof.
Leo Laporte [01:33:48]:
That's the problem.
Harper Reed [01:33:49]:
Like, I'm pretty sure it's— yeah, it's the humans, which is why I side with the AIs. No, but I really do think that we seem to be doing a very good job of killing ourselves today. We don't have to have an imaginary Well, we're going to get it, but yeah.
Leo Laporte [01:34:05]:
All right. Well, on that exciting note, we'll have more in just a little bit with the optimistic pessimist and the gang. There is a new mythos, right? 5/2. Some people have it. Not everybody has it. Some people have part of it, right? You have to apply and then And if you reach a certain, I don't know, if Anthropic likes you, you can get the cyber defense version. If they're sure you're not going to use it maliciously, you get the cyber attack version. And then the rest of us—
Harper Reed [01:34:42]:
They do not like reverse engineering. They are really not happy about that.
Leo Laporte [01:34:45]:
Yes, you can't. In fact, despite the fact that pretty much all AI is based on reverse engineering humanity, you cannot use apparently Fable 5.2 to reverse engineer video games. It'll stop you cold. Uh, this is an article Patrick put in here from Ethan Moloch, who I love and I read regularly on, on x.com. Jagged AGI, he calls it. He's got a new book called Coexistence.
Patrick Beja [01:35:15]:
It's pretty impressive. Um, so yeah, there, there's a couple of, uh, of articles, but essentially what it amounts to, including, uh, this one that you're referencing, It seems that Fable 5.2 has a significant upgrade in its AGI-ness. You know, no one knows what AGI is, but in this case, Ethan Mollick has asked it to do a number of things that require some level of creative thinking and The results are really impressive, and there is some creativity, some ingenuity, some things that you would expect, but also some things that you don't. And the other article that summarizes it is there's a company or some organization that does an evaluation of the ability of AI models to do remote labor. which I think the idea there, it's the Remote Labor Index by Scale Labs. And the idea there is, I think, if it's remote work, then if AI can do it, it's probably very easy for those models to replace your existing, the people you usually use.
Leo Laporte [01:36:45]:
Yeah, because if it's remote, you don't know it's AI. It could be a human. Exactly. Yeah.
Patrick Beja [01:36:49]:
You don't know, or slash you don't care, or you don't see it. And The increase in ability is significant. It went from the highest rated one was Opus 5.5, kind of equal with GPT-6 Astra. Fable 5.1 was a little bit under, but they were all around 20%. Fable 5.2 is around 40%. And they have—
Leo Laporte [01:37:12]:
It can complete 41% of freelance jobs, real jobs.
Patrick Beja [01:37:17]:
Yes. Complete.
Leo Laporte [01:37:17]:
And this is what's interesting.
Patrick Beja [01:37:19]:
It is actual real jobs. And of course it's, you know, one organization. Um, it's labs.scale.com/leaderboard, uh, that you'll find this at, uh, /ai. Oh, that's Scale's.
Leo Laporte [01:37:32]:
It does it. Yeah. Okay.
Patrick Beja [01:37:34]:
Yeah. And, um, and so they do things like, uh, dataset design and sourcing.
Leo Laporte [01:37:40]:
Look at these jobs. Wow.
Patrick Beja [01:37:41]:
Sorry. There's a bunch of, of things. They did detail the, the dataset design and sourcing and everything. And, you know, it does video, CAD, graphic design, game dev, et cetera. And the Ethan Mollick article also talks about a number of tasks, as I mentioned, that he would ask— the kind of thing that he would ask a class of his to come up with. There are things, entrepreneurial tasks, there are things like hardware hacking, there's a number of things. And you can go read the articles yourselves. dear listeners.
Patrick Beja [01:38:20]:
But what it comes down to is that, sure, AI is still somewhat lacking in some areas. Moloch talks about, for example, for one of the projects, there were a number of agents that ended up adding rules and rules to a rulebook without ever removing any of the rules. So it kept adding and adding, and it became incredibly unwieldy until he had to intervene. Things like that. But many of the results, including things that you might not think are things that an AI could do, like find a business case for— I'm sorry, I don't remember. I read the article a little while ago. There's some academic research. There's the reconstructing of the hardest humanities class in history, which is a class that was given in 1941.
Patrick Beja [01:39:21]:
And it makes websites to let people experience that class. And there are a number of other websites for other projects. The bottom line is that that model, Fable 5-2 specifically, to the understanding of Ethan Mollick and through the study of Scale Lab's remote labor index does achieve a significant amount of AGI-like project solving, right? It— so for one of them, Molech uses—
Leo Laporte [01:40:03]:
By the way, he's a professor of entrepreneurial business. He uses as his definition of AGI, AGI is an AI as good as most human experts at most intellectual tasks, and this is an important final clause, you can do with a computer. We're not talking tennis or making hamburgers.
Harper Reed [01:40:24]:
Plumbing.
Leo Laporte [01:40:25]:
Or plumbing. We're talking about jobs you can do sitting down and using a computer, which makes sense, by the way, that AI would be very good at that. It is a computer, right?
Patrick Beja [01:40:33]:
Well, so yes, of course, but it's also the variety of tasks that Moloch uh, you know, tried to get Fable 5.2 to achieve is really staggering.
Leo Laporte [01:40:48]:
Yeah.
Patrick Beja [01:40:48]:
Um, there, there's a level—
Leo Laporte [01:40:50]:
Just don't ask it to reverse engineer a video game.
Patrick Beja [01:40:53]:
Well, not anymore because it was too good at that. You know, it was—
Leo Laporte [01:40:57]:
Right. Actually, that's a very good point. It did. It had to stop because it could do it.
Patrick Beja [01:41:01]:
Yes.
Leo Laporte [01:41:02]:
Very well.
Patrick Beja [01:41:02]:
Yes, exactly.
Leo Laporte [01:41:04]:
Yeah. That's a good point.
Patrick Beja [01:41:04]:
Um, and, and that's, I guess—
Leo Laporte [01:41:06]:
This is, by the way, why I stopped using, uh, I agree, these models are bizarrely good. It's kind of surprising given what they're doing, which is basically token prediction, that they are able to seem so capable.
Patrick Beja [01:41:22]:
To do all of these things. And I'm being a little bit fuzzy in my explanation of this, but what it comes down to is that 2 years ago, Amodei and Hasebe and Altman all were saying, oh, we're going to get to AGI. And I was like,
Leo Laporte [01:41:39]:
yeah, sure, come
Patrick Beja [01:41:40]:
on.
Leo Laporte [01:41:41]:
Yeah,
Patrick Beja [01:41:41]:
like, whatever. And, and 2 years later, we kind of are half there with the— the reason I was, uh, you know, I didn't believe them is exactly what you were saying. It, you know, it's stochastic pattern and it's token predictions and autocorrect. How could it? Yes, it's, it's super autocorrect. And now we're kind of there already
Leo Laporte [01:42:04]:
Yeah.
Patrick Beja [01:42:05]:
And I can't believe that we're there, even though I know all of these things and I, you know, I know that I understand kind of a little bit the technology. Every 2 weeks, every month, I have a mind-blowing moment.
Leo Laporte [01:42:20]:
Exactly. Like, we've all— everybody on this panel's had that experience once or twice or many times or 15
Patrick Beja [01:42:25]:
times
Harper Reed [01:42:26]:
or every couple days
Leo Laporte [01:42:26]:
or all the time.
Patrick Beja [01:42:28]:
And we get numb to it. And there's the Nvidia thing that happens. And so we don't look at 5.2 because it's only 5.2, right?
Leo Laporte [01:42:34]:
Right.
Patrick Beja [01:42:35]:
But looking at the report on this, it is very, uh, impressive, even though we know it always is impressive. Every two— it must be Tuesday AI is doing something incredible, you know?
Harper Reed [01:42:51]:
True.
Patrick Beja [01:42:52]:
And yet this is again like, we're getting to AGI. I think this is the wine life thing. What's the question? We are getting to
Leo Laporte [01:43:00]:
AGI. It's kind of xenosphere-esque. Depends on your definition of it. We get closer and closer and closer, but never— do you think we ever will get there?
Molly White [01:43:08]:
It doesn't matter. Well, I think it depends very heavily on whose definition you're using, right? Because some definitions of AGI are probably already achieved because they're not particularly meaningful, right?
Leo Laporte [01:43:18]:
Right, right. Well, it's doing advanced AI.
Patrick Beja [01:43:19]:
No, but Mollack's definition, I think, means something.
Molly White [01:43:25]:
Yeah.
Patrick Beja [01:43:25]:
If it can do most things that humans can, experts can do with a computer, that's not nothing, right? And if we're getting there—
Molly White [01:43:37]:
Although I would say that's a challenging thing to test, right? Because that's an almost unbounded set of things.
Leo Laporte [01:43:43]:
Right.
Patrick Beja [01:43:44]:
I, you know, I don't think, I don't think the takeaway from this is where their AGI is.
Molly White [01:43:49]:
Right. No.
Patrick Beja [01:43:51]:
But the takeaway is when we were laughing at their prediction 2 years ago, I think we were the ones who were wrong. We might be the ones who were, who had been wrong, the ones who were laughing. Yeah. Because we're getting pretty close and that we didn't think was going to happen, or at least not in 2 years. What was the, the Altman blog? He was like, a few hundred days. What was he saying? Like this ridiculous thing that made us all laugh. Like we're a few hundred days away from AGI.
Leo Laporte [01:44:23]:
Here's my problem.
Patrick Beja [01:44:24]:
More like a couple thousand days.
Leo Laporte [01:44:26]:
Here's my problem with the folks running the Frontier Labs. I think they are so—
Harper Reed [01:44:31]:
They're so weird.
Patrick Beja [01:44:33]:
They are so weird. They're weird.
Harper Reed [01:44:34]:
That's your problem?
Leo Laporte [01:44:35]:
There's—
Patrick Beja [01:44:35]:
Have you heard of handoverism?
Leo Laporte [01:44:38]:
No. Anyway, don't go down that path. Oh, oh yes, I have heard of handoverism. That's exactly where they are, which is handing over our human agency to the machine. Oh. They, I'm pretty convinced at this point, but their massive demand for data centers and power, and Sam says it could cost us a trillion dollars, et cetera, et cetera, et cetera. They are true believers that they are creating a conscious entity. This is my problem with them.
Leo Laporte [01:45:09]:
And that they think no matter what the cost to humanity to whether it's climate change or money or doom, that it's worth it because it is our mission to create consciousness, to create a new entity, a new— I truly believe, I don't know about Altman, I think Altman's a business guy, but I know Dario believes this. I know Dario believes this.
Molly White [01:45:34]:
Well, and I think the AI companies that really jibe with sort of long-termism have this innate belief that it is their moral imperative to do this. Right. Because they just are able to sort of conjure these impossibly good futures and, you know, convince themselves that if there's any chance of that impossibly good future coming to, to, to be, then they have the moral, you know, requirement to pursue it. And the risks are sort of negligible because they're able to sort of generate this infinitely good possible outcome.
Leo Laporte [01:46:14]:
Maybe you may even be giving them too much credit. I don't know if they care so much about if the good or the bad.
Patrick Beja [01:46:19]:
I think that's how they justify it to themselves.
Leo Laporte [01:46:21]:
They justify it, but I think in their real minds they're thinking, this is evolution and we're just, we're just doing evolution. We're making Frankenstein. We're making— it's inevitable we're gonna make the next consciousness, and whether humans stick around to see it is not relevant.
Patrick Beja [01:46:37]:
Their best-case scenario is the Labrador scenario, that we will be like Labradors. Humans are between a pest and a pet and a parasite, but the best case is we're kind of Labradors. Wait, I'll find another article.
Leo Laporte [01:46:56]:
The thing that saves me is the jagged part because, well, for the example I give people when they say, well, AI's really good. I said, yeah, but I'll tell you what, I just took about 300 pictures on my last vacation. I would never give AI the job to pick my best pictures. Would you, Harper? Would you ever do that? Pick my best? I would say to AI, organize them. I would say edit them. There's a lot of things I would let AI do, but I would never let it pick my best pictures.
Harper Reed [01:47:27]:
Do you use Apple Photos?
Leo Laporte [01:47:29]:
Yeah.
Harper Reed [01:47:30]:
Do you know about all the internal scoring that they do on your photos?
Leo Laporte [01:47:33]:
Yeah, but do I care?
Harper Reed [01:47:35]:
Have you looked at it? You should use osxphotos, the Python package, to interrogate the internal scores. And there's a couple of them that are like, you can sort by like—
Leo Laporte [01:47:43]:
Are they scoring my photos on their quality?
Harper Reed [01:47:46]:
Quality and like—
Leo Laporte [01:47:47]:
Yeah, but I know that that's bone dead. Like, uh, well, the horizon's straight.
Harper Reed [01:47:52]:
Well, you should check it out because every time I look at it, there's always photos that are really good that I forgot. And so although I agree with you as someone who does a lot of photography, I I think you're, you're absolutely correct. Although I have been writing a Canon firmware for one of my little PowerShots here that I'm pretty sure basically—
Leo Laporte [01:48:08]:
Yeah, but that's appropriate. No, but that's, that's a good use of it.
Harper Reed [01:48:13]:
But, uh, but I, but I think there's a thing which is like, why not both? Like, I, I think that human taste is very important. I look at my photos that I take and I have preferences and whatnot, but I also think that, you know, when I look at the results of Apple's modeling, which is not new AI stuff, this is ML stuff from the past that's been around for a long time. Some of it's very, very, very good.
Leo Laporte [01:48:32]:
I understand, but I would never want it— it would look at Bresson's greatest pictures and say, well, that's— it's blurry. It's not good. I would never want it to make aesthetic decisions like that ever, ever, ever.
Patrick Beja [01:48:47]:
You know, I think we are— we have said that so many times in the past eternity, by which I mean 4 years. And we've been proven wrong so many times. I think there is a desire for us, like, to not want to believe that it can do anything. Maybe it can, Leo. Maybe it can. Yeah. And the thing is, maybe it doesn't matter that it can because people will still want to interact with people and want the things that the people choose. And, you know, but I've been cautioned against thinking it can't do that, it can't be creative, it can't be—
Harper Reed [01:49:35]:
this is our—
Patrick Beja [01:49:35]:
it can't have taste.
Leo Laporte [01:49:37]:
I bet you if you fed this Cartier-Bresson photo to Apple Photos, it would not say, oh, that's a great photo.
Harper Reed [01:49:49]:
And
Patrick Beja [01:49:49]:
yet it is one of the most famous photos
Leo Laporte [01:49:51]:
of all
Patrick Beja [01:49:52]:
I would say it's a great photo.
Harper Reed [01:49:54]:
I was expecting—
Leo Laporte [01:49:55]:
Yes, because you're human. Because you're human. Hi, this is Benito.
Benito Gonzalez [01:50:01]:
So real quick, like, there's also stuff like, what if there's a subject in the photo that AI doesn't care about, but to you is particularly important?
Leo Laporte [01:50:09]:
You know, like, it wouldn't know that. That's another issue.
Patrick Beja [01:50:12]:
You know, guys, I don't think we're saying AI will do everything everywhere. there. We'll still be us.
Leo Laporte [01:50:19]:
That's what I'm saying. It's jagged.
Patrick Beja [01:50:20]:
And we can still do stuff.
Leo Laporte [01:50:21]:
It is jagged. And that's what gives me hope.
Harper Reed [01:50:23]:
Jagged.
Leo Laporte [01:50:25]:
Because we are the people who live in the interstices. And I think what Dario—
Patrick Beja [01:50:31]:
The question is, will the interstices be the only thing left for us?
Leo Laporte [01:50:35]:
It will be. But I think what Dario thinks is that those are unimportant. And I think I would submit they're important.
Patrick Beja [01:50:41]:
Does he?
Leo Laporte [01:50:43]:
I think he re— well, I don't know. I mean, Um, it's hard to tell what anybody thinks. It could well be Jari is just trying to raise $2 trillion as quickly as possible. And then Harper just put on his hoodie that says, AI will kill us all.
Molly White [01:50:55]:
Where did you get that?
Harper Reed [01:50:57]:
This is my company merch. Um, they're walking around.
Leo Laporte [01:51:00]:
Everybody, 2389.ai, go to the store, pick up a—
Harper Reed [01:51:03]:
I think it's shop.2389.ai. Um, you can buy them. The shirts are a weird boxy shirt. I don't recommend buying them. I don't know even how we got that shirt, but it's literally the— it's a very strange—
Leo Laporte [01:51:14]:
I want the hoodie. Yeah, that seems
Harper Reed [01:51:17]:
It's really wide and it's very cropped, which, you know, is a look, but a lot of our, uh, tech company friends are like, this is a little too cropped for me.
Leo Laporte [01:51:27]:
You look like somebody AI has killed is what you're saying.
Harper Reed [01:51:30]:
Yeah. It killed the, it killed the bottom of your shirt. Um, but we made, I, I made it specifically to look like, uh, like we live in Chicago. So there's like a big city plumber space and you have all these, the kind of plumber shirt, which has the full address on the back. So it's a very, it's very fun. But we made these shirts so long ago, and then it became this big meme of AI will kill us all from that guy, Left Anthropic, and all that stuff. And I felt a little bit like—
Leo Laporte [01:51:55]:
It's the Jacob Coxon model.
Harper Reed [01:51:57]:
Yeah, but it's this thing of we made this because it's a statement on the social infrastructure. As this jagged edge or whatever you want to call it is deployed and it starts replacing various jobs, I don't think the United States is very good at handling displacement. And that was really the statement. Um, but then people started thinking it's just about AI, which it's like, you know, I think we have so many—
Leo Laporte [01:52:20]:
as you said, we have so many bigger problems. I don't think AI is the—
Patrick Beja [01:52:23]:
no, but I think— well, I'll tell you what, actually— sorry, Molly.
Molly White [01:52:27]:
I was just going to briefly say I think AI exacerbates some of those bigger problems, like you say, about, you know, the US being terrible at handling job displacement. I think that is a huge problem, and AI is a faster gun.
Leo Laporte [01:52:39]:
And it's going to make the job displacement happen so that we can be bad at handling it.
Patrick Beja [01:52:43]:
I'll get it one further. Job displacement isn't great, but the actual issue I think with AI being really good and being AGI-like or whatever is that I think Arthur Mensch, who's the CEO of Mistral, said it. What AGI does is that it transforms energy tokens into intelligence, right? If it— like intelligent intellectual work.
Leo Laporte [01:53:13]:
This is what Geoffrey Hinton said as well. That's the difference between our brain, our low-power analog brain, the massively powered electronic brain takes huge amounts of energy.
Patrick Beja [01:53:23]:
Yeah, but if it does it really well and you can throw energy at it, then it can work and it can do a lot of people's jobs, kind of. And the problem is energy is actually capital. If you have capital, you can own the intelligence. You already could kind of own physical labor with machinery. Now you can also own intellectual labor. And so the intellectual labor gets— of course, jobs change, but that's not a huge issue because as you often say, Leo, Coach riders went away and that we found new jobs, we found new things to do.
Leo Laporte [01:54:05]:
That's the way of the world. But
Patrick Beja [01:54:07]:
if it goes really fast and all of the jobs get displaced to the people who own the capital, who concentrate it more and more, then where do you get the money for buying bread?
Leo Laporte [01:54:24]:
Aren't they planning this better? Yes.
Patrick Beja [01:54:26]:
Well, they are planning it. Who's going to buy your AI if no one has a job? We need, how's it called in English? Universal income.
Harper Reed [01:54:37]:
Yeah.
Patrick Beja [01:54:37]:
Right? They'll say, oh, but so everyone needs to have universal income. But the problem is when you go to them again and say, oh great, okay, so universal income, we need money for that. You have a few billion, would you mind?
Leo Laporte [01:54:51]:
Not my money.
Patrick Beja [01:54:52]:
Giving? Yeah, exactly. It's like, oh well, what? No. the 5% tax on wealth in California was defeated. Defeated. Right? And they all pour all of their money. How much could you have done with $474 million if you were Elon Musk to help the Earth? And when time comes to actually redistribute wealth, they literally fight against it with all the energy and money at their disposal. So that's the something that is a real— I don't think AI will kill us or even that it will have taste or no taste. That is the thing that I worry about.
Leo Laporte [01:55:42]:
I don't know.
Molly White [01:55:42]:
Well, and you also end up with these sort of schemes like, um, Sam Altman had proposed at one point with like Worldcoin in his project there. A lot of what was selling there was the idea that that would be the universal basic income when AI takes everybody's job. But then you're in this situation where you were, you know, financially, you're, you are dependent for, you know, your life expenses on sort of the whims of a private company, which I would say is, is also a terrible outcome, right? Is you have sort of privately administered UBI, UBI, which Which is, I think, a horror in itself.
Patrick Beja [01:56:24]:
If only we had a way to self-organize as communities and nations, for example. Nation level seems good. Let's just decide who would make the decisions for everyone. Let's say we vote for them every now and then, and then they do stuff. I think that seems like a good idea. It could work.
Harper Reed [01:56:46]:
Too crazy. Too— slow down. Slow down.
Leo Laporte [01:56:50]:
Molly White. Damn, I just knew I was forgetting something. She got this email, uh, I guess addressing your, uh, your, your piece on how to become a Wikipedian, because of course Molly's been editing Wikipedia since she was in high school. Hi, your article about becoming a Wikipedian is fantastic, and I noticed it's missing a detailed comparison of high-end electronic Drum kits.
Molly White [01:57:16]:
It's my favorite one that I've ever gotten.
Leo Laporte [01:57:20]:
We've got a piece of content which helps with a detailed comparison of, amazingly enough, high-end electronic drum kits. Well, there you go. I would've just stuck that in just anyway. Just any—
Molly White [01:57:33]:
I was delighted that like most of them very, they really annoy me, but that one made my whole day.
Leo Laporte [01:57:40]:
You think that was an AI?
Molly White [01:57:42]:
It was certainly something.
Leo Laporte [01:57:43]:
Something. Somebody who apparently didn't understand that.
Molly White [01:57:47]:
Not a very good one.
Leo Laporte [01:57:49]:
Yeah, that's a good point. That's just the only AIs we really know about are the ones that aren't so good. You're watching This Week in Tech with Molly and Harper and Patrick, and we're so glad you're here. I did not know that the new governor of Alaska, who, by the way, just won the election a day or two ago, right? It was, it was a nail-biter because of their rank.
Molly White [01:58:11]:
Yeah, Alaska is very slow to count votes apparently.
Leo Laporte [01:58:15]:
Yeah. Uh, Governor-elect Jonathan Christ Tompkins, uh, is a, uh, is he a Wikipedia administrator?
Molly White [01:58:24]:
He was a very relatively active Wikipedia editor. Someone gave him the administrator's barn star even though he's not an administrator because they figured I think governor was close enough.
Leo Laporte [01:58:35]:
Yeah, governor's good. How did they— how do you know that he was an active Wikipedia, uh, editor?
Molly White [01:58:41]:
Um, I don't think he was particularly anonymous on Wikipedia. If you look at his— like, I looked up his user page.
Leo Laporte [01:58:47]:
Do we know what he liked to edit?
Molly White [01:58:49]:
Yeah, it was, it was stuff about like Yale and I think the town in Alaska where he's from and just a bunch of stuff like that. Nothing, nothing particularly—
Leo Laporte [01:58:57]:
There you go. Is he a Yale-y at least?
Molly White [01:59:00]:
I assume so. Yeah.
Leo Laporte [01:59:02]:
Yeah, he went to Yale but dropped out to run for the Alaska House of Representatives.
Molly White [01:59:08]:
Yeah, it doesn't look like he has edited since like 2011, but he had like 13,000 edits, which is, you know, more than just someone trying it out. Yeah.
Leo Laporte [01:59:17]:
Whoa, that's dedication.
Molly White [01:59:20]:
Yeah.
Patrick Beja [01:59:21]:
What does a governor do?
Leo Laporte [01:59:23]:
Edits Wikipedia mostly? Yeah.
Molly White [01:59:27]:
That's why the Wikipedia page on Alaska is so good. Yeah,
Leo Laporte [01:59:30]:
yeah.
Molly White [01:59:31]:
Um, you asked what a governor does?
Patrick Beja [01:59:33]:
Yeah, because I get like a congressman, a senator, and you also elect governors during the midterms. Is that the state's boss?
Molly White [01:59:41]:
Basically the state level.
Patrick Beja [01:59:43]:
Okay.
Molly White [01:59:43]:
Yeah, president of the state.
Patrick Beja [01:59:45]:
President of the state, I get it. Yeah, okay.
Leo Laporte [01:59:48]:
And in some states, the governor has more, uh, power than in others. It— the legislature In California, for instance, has made the governor's office fairly weak. Or is it the other way around? I can never remember. But the power of the governor is very much different in each state. So you can't really say exactly what a governor does depends on the state.
Patrick Beja [02:00:11]:
I see.
Leo Laporte [02:00:13]:
Let's see what else we've done. Are we done with Nvidia? I think we're done with Nvidia. I like Do you think your plot thickens, Harper, that maybe he'll be back?
Harper Reed [02:00:24]:
It's not done. No, we're not done. There's no way it's done. Like, like, you don't, you don't have someone that creates that company, which is worth that much money and made that many people wealthy, just slink into the background.
Leo Laporte [02:00:34]:
It is.
Molly White [02:00:34]:
Well, didn't Elon Musk offer to hire him or something like that?
Leo Laporte [02:00:37]:
Elon says he should go back. Yeah. Yeah. Uh, I don't— yeah, maybe he'll take a job somewhere. I think he's just gonna slink off. He's worth $140 billion.
Molly White [02:00:46]:
Yeah, what job? He's gonna go, you know, flip burgers somewhere or something.
Leo Laporte [02:00:54]:
Denny's. He's gonna go work at that Denny's.
Molly White [02:00:56]:
Yeah, there you go.
Harper Reed [02:00:56]:
Yeah, start another company. I don't know if those type of people are built to do— not do this.
Leo Laporte [02:01:03]:
Well, what he'll do is probably start another company at some point once he licks his wounds.
Patrick Beja [02:01:07]:
Yeah, if he does.
Harper Reed [02:01:08]:
He's really young.
Leo Laporte [02:01:09]:
Yeah, I'm sure he'll just start another company. I wonder if this is the pin that will burst the AI bubble. I mean, first of all, I'm not agreeing necessarily that there is an AI bubble, but if there is, this could be the pin that bursts it.
Patrick Beja [02:01:24]:
No, it doesn't. You know, nothing— I will— nothing bad will come of all of this. Nothing bad. No one suffers. No one is actually, uh, hurt by this. If you— you know, some people will make less exorbitant amounts of money than they would have otherwise. Hopefully, Jensen Huang manages to find common ground with his family. But even if they don't, he will be fine and the economy will be fine.
Patrick Beja [02:01:59]:
The worst-case scenario is that things slow down a little bit, which honestly is probably not a bad thing. I'm not sure there is a bubble, but Even if there is, this is probably something, if he doesn't go back and Nvidia doesn't manage to continue innovating after whatever, if people freak out a little bit, the tech industry deflates maybe, or the acceleration slows down.
Leo Laporte [02:02:28]:
I'll give you some evidence to the contrary.
Patrick Beja [02:02:30]:
It doesn't explode.
Leo Laporte [02:02:31]:
First of all, I'll give you some evidence to the contrary, but first of all, look at the stock price. when this was announced. It has recovered by about 50%.
Molly White [02:02:38]:
I was going to say it actually has not crashed as far as I thought it would.
Leo Laporte [02:02:42]:
Yeah. But here's the data point that disagrees with you, Patrick. The new CEO, who is the former CFO, Colette, I can't remember her last name.
Patrick Beja [02:02:51]:
Kress, is it?
Leo Laporte [02:02:52]:
Kress, was very much against the circular financing deals, we are told by the Wall Street Journal, that Jensen was fond of. Now, these deals, By the way, I read an excellent book about the history of the Transcontinental Railroad, and the same exact circular financing happened then, which was the money that was put up by the people who were going to build the rails, for instance, into the railroad would come right back to them to buy the rails. Jensen took money, invested in companies who would then take that money and put it back into Nvidia by buying their chips. Thereby boosting the books of both with actually a net-zero economic change. The money just went right back in their pockets. She was against that. And if she withdraws, in fact, it looks like she's already started some of those financing deals, that could be significant. These are data— these are companies that are building data centers.
Leo Laporte [02:03:53]:
These are companies that—
Patrick Beja [02:03:54]:
Yeah, it slows things down. It doesn't burst the bubble. It slows it down, which I think is maybe not a bad thing. Again, it's been a little bit—
Leo Laporte [02:04:03]:
No, I think that might be a good thing. Yeah.
Patrick Beja [02:04:05]:
Yeah. That's why I'm saying I don't see anyone actually— like the world— sorry, Nvidia is going to keep going. The tech industry is going to keep going. The IPOs are not stopping. Yes, the stock price of Nvidia went down a little bit. And because it's so high, it amounts to a few hundred million, which is an enormous amount of money, but it doesn't crash anything.
Leo Laporte [02:04:30]:
A few hundred billion, not million.
Patrick Beja [02:04:32]:
Yeah, sorry, I apologize. Yes.
Leo Laporte [02:04:34]:
Yes. It's just, um, but you know, that's a billion there, here or there, you know,
Molly White [02:04:38]:
million, billion,
Leo Laporte [02:04:39]:
you know.
Patrick Beja [02:04:41]:
The total, uh, valuation of the AI industry is so much higher. Yeah, like if there was a risk that it would crash, crash, then I agree that's a problem because it's a problem for the US. It's a problem for everyone in the world because if the US economy crashes, like the US economy is ballooned by the AI industry at the moment, right? It's what, 50% of the S&P 500 or something ridiculous like this?
Leo Laporte [02:05:08]:
Yeah, the Magnificent 7 is a huge portion of the S&P 500.
Patrick Beja [02:05:12]:
If that crashes, the world is coming down.
Molly White [02:05:12]:
I think there's an argument to be made that, you know, a bubble bursting or at least deflating is actually a good thing because I think generally when there's a bubble, you want it to as quickly as possible rather than continue to grow unbounded, right? And so—
Patrick Beja [02:05:25]:
Yes, that's why I'm saying this probably deflates it, which is not such a bad thing, right? Because it's before it gets crazy. And I wouldn't be surprised if some of the AI people are like, all right, maybe that's our excuse to slow things down a little bit because we got a little, you know, we got overexcited.
Harper Reed [02:05:46]:
Have you met any of the AI people? No, but there are financial—
Leo Laporte [02:05:52]:
Speaking as somebody who is depending on his retirement fund for the rest of his life, I have lost a significant portion of my investments.
Patrick Beja [02:06:01]:
Of how much did you— how much did it grow over the past 4 years?
Leo Laporte [02:06:05]:
True. I was ahead.
Patrick Beja [02:06:07]:
Right?
Leo Laporte [02:06:07]:
Yeah, good point.
Molly White [02:06:08]:
Yeah, I mean, I do think it is somewhat misguided to, to treat, um, stock price Declines as victimless. Because, you know, especially where, you know, especially in the US, but certainly elsewhere, where so much of your retirement well-being relies on you managing investments.
Leo Laporte [02:06:31]:
It's good news, bad news. It's bad news for me. It's good news if you like these shows, because I'm gonna have to keep working for this.
Molly White [02:06:36]:
Because you're gonna be working until you're 90.
Patrick Beja [02:06:38]:
Molly, I, I don't disagree, but—
Harper Reed [02:06:40]:
2 more years.
Patrick Beja [02:06:41]:
Those 2 more years, those These stocks have gained how many hundreds of percent over the past 4 years?
Molly White [02:06:47]:
But how many people bought them 6 months ago or a year ago and put their entire retirement savings into them?
Patrick Beja [02:06:54]:
Sure. But you can't have your cake and eat it too. If you think that deflation or deflating the bubble is necessary, maybe it's better that it deflates slowly a little bit now and things calm down rather than if it is bubble, it pops in a year because those circular financing things are getting insane. And at some point someone says stop.
Leo Laporte [02:07:15]:
And we're in what you were saying, Harper. From today's Financial Times, Jensen Huang discusses comeback as Nvidia CEO with allies. So that didn't last very long. He has a war room.
Molly White [02:07:29]:
Well, what does discusses mean? I mean, of course he's discussing it.
Leo Laporte [02:07:32]:
According to multiple people familiar with the matter—
Harper Reed [02:07:34]:
I want to come back. Yeah.
Leo Laporte [02:07:35]:
Wong has discussed with confidants the possibility of returning. Uh, I don't know. They're putting together a war room of Silicon Valley executives to lobby.
Molly White [02:07:46]:
I hate everything about that sentence.
Leo Laporte [02:07:48]:
I know, I know. Uh, the, the source says everyone that matters in tech outside the Frontier Labs— Dario, Sam— loves Jensen.
Harper Reed [02:07:58]:
Everyone that matters in tech outside the— yeah.
Molly White [02:08:01]:
Yeah, that seems like a very malleable sentence.
Leo Laporte [02:08:05]:
Yeah. Tim Sweeney, Elon Musk both signed a public letter demanding the board bring Jensen back. I don't know if that has any weight with the board.
Molly White [02:08:16]:
Yeah.
Patrick Beja [02:08:17]:
But just a couple of days ago, Bloomberg had an article saying that Jensen Huang himself was completely deflated, calling himself a piece of ass.
Leo Laporte [02:08:28]:
Yeah, yeah, yeah.
Patrick Beja [02:08:29]:
And, uh, and, and that kind of worries me a little bit about his character as well. Like, one day he's like, oh, I'm so—
Leo Laporte [02:08:37]:
unless he really—
Patrick Beja [02:08:38]:
go, I'm gonna come back.
Leo Laporte [02:08:39]:
Unless his crisis publicity team put that out.
Patrick Beja [02:08:43]:
Maybe.
Leo Laporte [02:08:43]:
We don't know what's going on. By the way, should also point out, uh, the board has already started looking at candidates, including the CEO of Arm, René Haas, who has already been approached by the board. So they're moving very quickly, even though they don't have a formal process.
Harper Reed [02:08:59]:
How does someone— I don't know how this stuff works where you work for a company like AMD, like Rene Haas, and then all these news— ARM, sorry. Like all this news is, um, like this person is going to go work at NVIDIA now. Like, what do you do? Like, what is ARM's—
Leo Laporte [02:09:15]:
That's not good for the company. Yeah, that's bad for ARM.
Harper Reed [02:09:17]:
No, it seems bad. I like ARM.
Leo Laporte [02:09:19]:
Uh, somebody, uh, else said, or it was rumored that Johnny Tsurugi, who is the, uh, Apple Apple chip guy, a very brilliant guy, also being considered. The problem with Colette Kress is she's not an engineer, she's a financial person, and you need an engineer in that position, I think. Or at least that's what Jensen wants.
Harper Reed [02:09:38]:
We want an engineer in that position.
Leo Laporte [02:09:40]:
You want somebody who really understands the business.
Patrick Beja [02:09:42]:
Maybe. Well, Trump wants Tim Cook to do
Leo Laporte [02:09:44]:
it. That was the funniest one.
Molly White [02:09:46]:
Well, Trump also wants the wrestling guy to be the communications director.
Leo Laporte [02:09:52]:
Danny Wise.
Molly White [02:09:52]:
So I'm not sure his opinion is tremendous.
Leo Laporte [02:09:55]:
Yeah. Yeah. Uh, here's Molly's story from Web3 is going just great. The biggest— uh, no, that's not it. That's the Wall Street Journal. I'm gonna pull it up right now.
Molly White [02:10:06]:
Yeah, I'm actually the
Leo Laporte [02:10:08]:
CEO
Molly White [02:10:09]:
after this conversation.
Leo Laporte [02:10:10]:
North Korea, this is the biggest, uh, crypto hack so far, as you point out, hacks Binance for $6.3 billion from multiple hot wallets. Largest crypto theft in history. Binance says we're going to make it all good with our profits.
Patrick Beja [02:10:28]:
That is insane.
Harper Reed [02:10:29]:
That's a lot of money.
Molly White [02:10:30]:
I know, it's so much money.
Harper Reed [02:10:32]:
Molly, have you ever talked to the criminals?
Molly White [02:10:35]:
Uh, which ones?
Leo Laporte [02:10:37]:
Like any?
Harper Reed [02:10:39]:
Not us, not us. We're different criminals. I meant the guys who are stealing all the, the people who are stealing the, the crypto, the monies. That's like—
Molly White [02:10:47]:
No, not these ones. These are North Korea.
Leo Laporte [02:10:50]:
It's North Korea, right?
Molly White [02:10:51]:
Yeah,
Harper Reed [02:10:51]:
that's really what I want to— I want to I just, I just, I have— I, it seems to me that this is a very good, like, a very profitable operation that they're probably investing a lot of money in doing much more of.
Molly White [02:11:03]:
Oh yes. I mean, this has actually become a tremendous amount of North Korea's economy at this point.
Leo Laporte [02:11:08]:
It's their hard currency.
Molly White [02:11:09]:
It's crypto. Yeah.
Leo Laporte [02:11:12]:
Uh, is this the one where it was a, uh, hacked version of, uh, uh, Homebrew? That was used to do this? I saw that. And maybe—
Molly White [02:11:22]:
yeah, so it's another one that has been sort of multiple theories have been floated. Um, but yeah, I mean, it seems like— and it seems like it was a little more widespread than the original report, which was like one laptop from one employee. Um, but yeah, I mean, it has— it's basically, at least what we know of it so far, is that it was an employee-compromised via an AI tool that he downloaded, he or she downloaded, that basically installed malware on the computer.
Leo Laporte [02:11:55]:
See AI again at the core of all of our problems.
Harper Reed [02:11:59]:
I'm going to have to put the hoodie on again. Yeah.
Molly White [02:12:01]:
Well, I mean, actually AI has been a tremendous issue in the crypto sector around a lot of thefts that have been happening lately. I mean, Web3 is—
Leo Laporte [02:12:10]:
Really?
Molly White [02:12:10]:
They're using AI to—
Leo Laporte [02:12:11]:
To plan the thefts, or?
Molly White [02:12:12]:
Yeah, it's full of basically vulnerabilities that have been attributed to hackers using AI because it just gives them the opportunity to test so many potential vulnerabilities that they would not have had the sort of manpower to do before. And certainly in the crypto world, you have this added vulnerability where there's a lot of sort of zombie smart contracts kicking around where it's a company that went out of business, But, you know, the blockchain's forever. And so there's just like dormant money sitting in these ancient smart contracts that have no security engineers paying attention to them. And it's sort of, you know, low-hanging fruit, I think, for people who, who are out there using AI models and kicking the tires.
Leo Laporte [02:12:56]:
Is it your sense this was a sophisticated hack? Yeah.
Molly White [02:12:59]:
Yeah. This one for sure. Um,
Leo Laporte [02:13:02]:
it was very fast. It was.
Molly White [02:13:04]:
Yeah. That's not unusual. Yeah, that's actually fairly common with crypto hacks. But yeah, I mean, this was definitely a major operation. And any time I was talking about this a little while ago and I wrote a piece about it in my newsletter, but basically any time there's a crypto theft of this magnitude and it's against— especially if it's against a company like Binance that has fairly robust security in place, obviously you can sort of quibble on that a little bit. It's pretty much always North Korea because they are constantly doing these very high value, like $100 million is kind of the floor in a lot of cases, crypto thefts against, you know, there was one on Bybit. If anyone remembers, there was a $600 million theft on Axie Infinity, which was that crypto game that was really popular in like the real Web3 era. That was North Korea.
Molly White [02:14:00]:
Um, this was North Korea. So it's, it's not, not small potatoes.
Harper Reed [02:14:07]:
How did they liquidate it?
Molly White [02:14:09]:
So that's actually super interesting. Um, North Korea has a pretty sophisticated, like, money laundering operation that they use, and that's actually often how these hacks are attributed to them, is they have almost like a signature when they start to move the money around. But it's a big— it's sort of a combination of a bunch of things. So one, they sort of obfuscate the, the crypto so that it is hard to trace back to the original theft. And therefore, when they go to cash it out, it's not flagged by whatever exchange they might be using. But there's also these operations, these, you know, this combination of over-the-counter brokers that will cash out for them. There's banks in China and Russia that they use. There's a number of sort of other money laundering tools that they use.
Molly White [02:14:57]:
And then increasingly, they're actually starting to use crypto directly for payments for goods and services, goods and services where they'll buy, you know, raw materials or something with stablecoins, and they don't even have to cash it out at all.
Leo Laporte [02:15:10]:
So that— but generally, they want to turn it into what, dollars?
Molly White [02:15:13]:
Yeah. Um, yeah, or I think it really depends.
Leo Laporte [02:15:17]:
Those are euros or whatever.
Molly White [02:15:19]:
Yeah, or yuan, or, you know, sort of whatever. Uh,
Leo Laporte [02:15:23]:
wherever they can.
Molly White [02:15:24]:
Rubles. Yeah.
Leo Laporte [02:15:26]:
You know, you don't want rubles. Uh, uh, were they also involved in the, uh, the deepfake phone calls? The $129 million was stolen using AI. This is a federal indictment. 3 men used generative AI to impersonate the voices of chief executives and lawyers to defraud companies of millions of dollars.
Molly White [02:15:50]:
Yeah, I mean, North Korea has been involved in that. I don't know if they were involved in the most recent one, but, um, there's another like really common thing where— and this was at least suspected with the Binance hack, I'm not sure if it's been confirmed or, or disconfirmed— but, um, basically, um, doing job interviews. They'll, they'll find someone at a crypto company and they'll offer them a job job, and then they'll have them get on a Zoom call for an interview. And they— there's usually some, you know, it's like, oh, something's wrong with the video call. And so they'll have them download, uh, you know, a Zoom-type app, and that's how they get malware onto the computer. Um, that's been super, super common from North Korea, uh, in the past couple of years. And then to some extent, they've also just been, um, getting North Korean employees hired at some of these companies where they'll actually be, you know, working on the code from the inside essentially, and then, uh, make off with it eventually.
Leo Laporte [02:16:50]:
Yeah.
Harper Reed [02:16:50]:
I love that. I, I think it's— I was talking to a friend at a big company who was saying that their biggest, their biggest, one of their biggest surface areas for attack right now is just, um, employees who will go through the entire process, get interviewed, people will love them, they'll make relationships with the people, and they now have new rules where you have to be, you have to go to the HQ for your first day. and pick up your laptop because they're so good and so robust, such a robust attack.
Molly White [02:17:17]:
Yeah.
Leo Laporte [02:17:17]:
Although even then—
Molly White [02:17:17]:
well, in crypto, it's— in crypto, it's a tremendous problem because so many crypto companies are fully remote. Sometimes the employees are like barely even known, you know, especially in some of the open source projects where it's just a GitHub handle and you don't really know who's behind it. There's, there's very little scrutiny oftentimes, and that's been an issue. in a couple of projects.
Leo Laporte [02:17:39]:
This story from Reuters, North Korean hackers steal more than $400 million using video images. Now, $400 million seems like small potatoes now.
Harper Reed [02:17:49]:
Can we talk about what video images are? Are they just saying— what is video images?
Molly White [02:17:53]:
They're just GIFs.
Leo Laporte [02:17:54]:
The FBI refers to it as FaceStealer. Targets are lured to video chats on imposter versions of video meeting software. So it's fake Zoom or Teams.
Molly White [02:18:04]:
Yeah.
Leo Laporte [02:18:04]:
Then they're in— the targets are issued invitations from genuine contacts they've previously interacted with, but whose social media and communication accounts have been compromised. And then they have already recorded the images. What— the reason this is so good, effective these days, is it only takes 20 seconds now to clone a voice. It only takes a few—
Harper Reed [02:18:24]:
We're in trouble.
Leo Laporte [02:18:25]:
We're in trouble. It's very easy. So once in the call, the targets, who are also being recorded, encounter a mock technical error which requires a fix by clicking a link, which then injects the target's computer and social media accounts. Target contacts are then used to target additional people, so it spreads.
Molly White [02:18:42]:
Yeah, it's like a worm kind of—
Leo Laporte [02:18:44]:
FBI has identified and seized more than 900 victim webcam videos so far. Oh, no. So the hackers are using AI to create deepfakes of well-known and influential people in the cryptocurrency and blockchain industry.
Molly White [02:18:59]:
Yeah, this was even before AI video got more sophisticated. There was an issue where people would reach out to you and be like, come on my podcast, and it would look like it was coming from a real influencer or whoever's doing crypto podcasts.
Leo Laporte [02:19:12]:
Yes. I hear that Leo Laporte does that all the time.
Molly White [02:19:16]:
Yeah. Well, and it's funny because I got a couple of those and I'm like, they would not want me to come on their podcast. It was clear to me that it was a scam purely because because I would not be invited. But to someone who's trying to pitch a meme coin or something, I think it got a lot of people.
Leo Laporte [02:19:32]:
You're watching This Week in Tech. Molly White at citationneeded.news, is that right? Or just go to mollywhite.net. That's easy to remember. Either one. Yeah. And of course, Harper Reed from 2389.ai and Patrick Bet-David from notpatrick.com. Cory, uh, Dr. O, as always, very pithy in his I told you so post today.
Molly White [02:20:00]:
Sorry, I did not catch that. That was pithy.
Leo Laporte [02:20:03]:
Pithy and pissy. He's full of vinegar and pithy. Pithy and vinegar. Uh, he said this is— this Microsoft thing is all because we've basically made it impossible to prosecute antitrust. Microsoft is too big. At this point to do anything about. So they fire people. They don't care about security anymore.
Leo Laporte [02:20:23]:
They don't care about reliability or uptime. Uh, and, uh, this was inevitable. And I guess he makes a good point.
Patrick Beja [02:20:36]:
Yes. 24 hours of downtime. It was such a gigantic scandal. Our lives have been upended. This was inevitable. I mean, I love Cory Doctorow, but sometimes there's a little bit—
Leo Laporte [02:20:51]:
He does say Microsoft has demonstrably underinvested in their IT stability. Any IT plan that results in a global multi-day outage is by definition inadequate. Does Microsoft lack funds? No, they, uh, they have, uh, they're doing $30 billion in stock buybacks over the last year. Um, you know, they've invested hundreds of billions into AI, uh, but apparently not a penny for defense.
Molly White [02:21:23]:
Well, it is kind of an interesting thing where it's like, you know, you sort of question— it's actually kind of similar to my reaction to the Binance hack, which was like, how did such a big company that spends so much time on security have one dev laptop get compromised? and $6 billion were stolen. Yeah. Like, you would think there would be more redundancy there. And it's the same question for me with Microsoft, which is like, and, and the airline outage as well, which is like, how do these systems exist with no redundancy, no failovers, no possible way to come back online without a 24-plus-hour outage, right? And, you know, obviously it's challenging I know it's challenging to you when you, especially with the airlines where the existing infrastructure is not in great shape to just build a duplicate, but you kind of think there would be a little more disaster planning.
Patrick Beja [02:22:17]:
Well, I think there's often, like, these are the ones we see, right? Because the failovers and the redundancy and the stuff happened, didn't work out in these cases, but how many were there where those safeguards did work? What I'm trying to say is we all know tech well enough to know that there's no system that's perfect.
Molly White [02:22:41]:
Right.
Patrick Beja [02:22:41]:
And sometimes you're gonna have things that get messed up. And I guess the question is, was there enough that was done to prevent these things from happening? Who does get— like, I don't know about the FAA thing because apparently there was some preferential treatment or some kind kind of corruption-adjacent behavior happening, maybe. I don't know. But in the case of Microsoft, I think it is a little bit hyperbolic and victory lap-y/self-serving for Cory Doctorow to say they don't care about network security, I think, or whatever term he used. I'm guessing Microsoft does care a little bit. Did they not invest enough? Maybe. Was this thing preventable? Maybe. If you go overboard on everything, maybe there were some mistakes.
Leo Laporte [02:23:35]:
Yeah, but 2 days of outage is a lot of outage.
Molly White [02:23:38]:
It is a very long outage.
Patrick Beja [02:23:40]:
That's a lot. It is. It is. But I don't know that, I mean, maybe Dr. Oh knows about all of this, but I would like to see someone who has a, a, had access to the details of this specific outage. I would like to see an independent assessment before I go out and say Microsoft doesn't care about security. I think that's a little bit—
Harper Reed [02:24:10]:
Both things can be true. How
Leo Laporte [02:24:14]:
about
Patrick Beja [02:24:14]:
you think Microsoft doesn't care about security?
Harper Reed [02:24:17]:
No, I think that they are—
Molly White [02:24:18]:
Or doesn't care enough.
Harper Reed [02:24:19]:
I think that—
Patrick Beja [02:24:20]:
yeah, but then you have to define enough.
Harper Reed [02:24:22]:
Well, I think enough would be that they don't have multiple-day outages.
Leo Laporte [02:24:28]:
Uh, there is a lawsuit, class action against Samsung, SK Hynix, and Micron, the 3 biggest RAM makers, saying that this is price fixing. There isn't really a DRAM shortage there, and the judge has just refused to dismiss this case. Judge Noel Weiss found buyers plausibly alleged the chipmakers conspired to cut conventional DRAM supply and shift capacity to high-bandwidth memory, and that long-term HBM lockups, refusals of long-term conventional DRAM contracts, the closure of Crucial by Micron— there, that's their, of course, their B2C arm, the consumer arm— Fab delays, the failure to expand despite record prices, near-identical customer vetting, and the company's prior cartel history— companies plural— cartel history supported a plausible inference of conspiracy. So that lawsuit is going to continue, or the class action lawsuit is going to continue.
Molly White [02:25:34]:
Who's bringing it? How are they defining the class?
Leo Laporte [02:25:36]:
That's a good question. I'm sure we're all in the class. Um, suit was filed, uh, last June by 17 consumers and small businesses from 5 states.
Patrick Beja [02:25:48]:
I remember that.
Leo Laporte [02:25:49]:
Yeah,
Patrick Beja [02:25:50]:
I think it's not inconceivable that there were a few emails going around saying, you
Leo Laporte [02:25:55]:
think
Patrick Beja [02:25:56]:
so? You're, you're not gonna, you're not gonna do it, right? Us either, right? All right, let's— but at the same time, you know, the boom and bust cycle, uh, that's so characteristic of the RAM market. I can understand how for a while they were like, that AI thing is going away. Let's not start building factories now and get caught with our pants down in a year where, when we will have so much RAM or 2 years or 3 years, we'll have so much RAM that again, the prices will bottom out and will Uh, you know, because that's how it's happened over the past 30 years.
Leo Laporte [02:26:38]:
Sure, it has. I've seen it up and down.
Patrick Beja [02:26:40]:
That being said, yeah, it certainly doesn't hurt them if there isn't more factories coming on.
Leo Laporte [02:26:48]:
You know what might hurt them though is the, uh, the moratoriums on, uh, building data centers. There are a lot of data centers that may never get built, especially, uh, if there is a slight deflation of the bubble. And that's going to ripple all the way down to these DRAM companies.
Patrick Beja [02:27:07]:
Yeah, probably. But you know who's happy about that? All the ones that already have data centers, like the hyperscalers are— the NeoClouds are now probably drinking a lot of alcohol. The hyperscalers, the Googles and Amazon and Microsoft, they're like, yes, right? Resource constrained.
Leo Laporte [02:27:32]:
Yeah,
Patrick Beja [02:27:32]:
we're gonna
Molly White [02:27:33]:
have Bitcoin miners who are about to go under. And they're finally—
Leo Laporte [02:27:37]:
next week, Grand Theft Auto 6, right? It will be the largest debut probably of all time of a video game. I'm expecting
Patrick Beja [02:27:47]:
of it, of an anything,
Leo Laporte [02:27:49]:
of an anything more than a movie, more than Gone with the Wind, more than anything.
Molly White [02:27:54]:
How long have people been waiting for this one? Isn't it a really long time?
Leo Laporte [02:28:00]:
Yes.
Patrick Beja [02:28:00]:
Well, the previous one came out in 2013. Yeah.
Leo Laporte [02:28:04]:
Oh boy. So that's 13 years in the making. Have they been making it that long?
Patrick Beja [02:28:11]:
So, well, yes. Well, they did make Red Dead Redemption 2, which is also an incredibly ambitious video game. And that is the Red Dead Redemption 2 is probably the 4th biggest selling game of all time. It's funny because when it came out, everyone was like, well, it's good, but is it as successful as Grand Theft Auto? And of course it isn't because Grand Theft Auto 5, like in this industry, a game that sells well sells a few million copies. And a game that sells really well will sell 10, maybe 15 is enormous. If you break that number. Grand Theft Auto V has sold 200 million copies. Mm-hmm.
Patrick Beja [02:28:56]:
200 million. It is an unbelievable number. And Red Dead Redemption 2 sold something like 40 or 60 million, which is also insane. But the biggest factor that kept Grand Theft Auto in people's hearts and minds was Um,
Leo Laporte [02:29:16]:
the hookers?
Patrick Beja [02:29:17]:
GTA Online. Oh, which has been a live service game that has kept going for those 13 years or whenever it came out a year after the—
Leo Laporte [02:29:27]:
There's a lot of nostalgia too driving this, right? Because, uh,
Patrick Beja [02:29:30]:
yeah,
Leo Laporte [02:29:31]:
13 years ago a lot of people your age were teenagers and—
Patrick Beja [02:29:36]:
You're very kind.
Leo Laporte [02:29:37]:
GTA 5.
Patrick Beja [02:29:38]:
Let's go with that. Um, yeah, it's, it's also, you know, I've been wondering because I think a lot of people who are very active video game enthusiasts, people who define themselves as gamers and play multiple games a month, don't really care that much about GTA. Everyone will play it because we're curious, but it's like we get our video game fixes from other places.
Leo Laporte [02:30:10]:
Right.
Patrick Beja [02:30:10]:
And so I've been wondering a What makes GTA so successful? And it's probably a combination of many different things. It's the nostalgia, it's the self-sustaining cultural event. It's the fact that the previous one was 13 years ago and the GTA Online and everything. But also, I've tried to make a mental inventory of all video games. in the history of the world. And it's basically the only one where you can go out in an open world and play around in quote unquote real life, like modern current reality recreated and go and be a bad person or do stupid stuff. Like my kids were playing, not GTA, they're very young. young, but they were playing, um, uh, Spider-Man on PlayStation just a few hours ago, and they were having so much fun grabbing trash cans and throwing them around in the street.
Patrick Beja [02:31:23]:
Like, they did this for half an hour and they were giggling like idiots. I think GTA is that for adults. You know, you go around and you do stuff that you Yeah, it's
TWiT.tv [02:31:35]:
the
Patrick Beja [02:31:35]:
sandbox.
Leo Laporte [02:31:36]:
You don't even really play to win, you just play to play.
TWiT.tv [02:31:38]:
It's the sandbox. It's like mine. It's the reason Minecraft is big too. Yeah, it's the same reason Minecraft is big.
Patrick Beja [02:31:43]:
Yeah.
Leo Laporte [02:31:43]:
Yeah, that's cool.
Patrick Beja [02:31:45]:
And also it's, you know, uh, graphically, uh, probably gonna be the, the, the most impressive looking game of all time and the production value is incredible and all of those things. But down at its core, it's the only game where you can go out as if, you know, in real life and be a bad person and do dumb things.
Leo Laporte [02:32:06]:
Right.
Patrick Beja [02:32:06]:
There are no— it's surprising. All the games are, you know, fantasy settings, science fiction settings. Some of the Japanese games actually have a modern, uh, real setting, but it's very, very rare. So I think that plays into it.
Leo Laporte [02:32:21]:
Right. Molly, uh, you're gonna play it?
Molly White [02:32:24]:
No, I, I basically can't play a tremendous number of video games because I get I get horrifically motion sick.
Leo Laporte [02:32:32]:
Oh, that'll take all the fun out of it.
Molly White [02:32:34]:
Yeah. Yeah. I'm very limited.
Leo Laporte [02:32:36]:
My game these days is local AI. I bet you Harper feels the same way. I haven't played a game since—
Harper Reed [02:32:41]:
I've never been a gamer. I've just never, that's never been my, I played a lot of Minecraft. I did too.
Molly White [02:32:46]:
Even Minecraft makes me motion sick because it's like—
Patrick Beja [02:32:49]:
Oh.
Harper Reed [02:32:50]:
Yeah, it's very bumpy.
Molly White [02:32:51]:
Yeah.
Patrick Beja [02:32:52]:
Even third person, you get, you get motion sick?
Molly White [02:32:55]:
Yeah.
Leo Laporte [02:32:55]:
I'm sympathetic.
Molly White [02:32:56]:
Basically I can only, it's, and it's funny because I don't get motion sick much in like real life travel. on, you know, buses and stuff. But yeah, for whatever reason, I'm just totally—
Patrick Beja [02:33:06]:
Oh, play, play on a bus.
Leo Laporte [02:33:07]:
Just the sickest I have ever gotten in my life was playing Doom on a, on a commuter bus. And I did download the Ocarina of Time remake, which is going to be also—
Harper Reed [02:33:22]:
It's supposed to be good.
Leo Laporte [02:33:23]:
Very—
Molly White [02:33:24]:
that's one thing is I can to some extent play on small Small screens. So if I play on the Switch, it still will give me a headache after a while, but I can do it.
Leo Laporte [02:33:34]:
One last break and then a final story. Well, we'll end on a high note. How about that? You're watching This Week in Tech with Molly White, Harper Reed, Patrick Bet-David. We're so glad you're here this week. Last story of the week. Yesterday, Starship was captured landing, and actually I'll have to jump 25 hours ahead, by giant chopsticks. And it's the first time a vehicle that size has been recovered for reuse. We don't normally reuse this kind of vehicle.
Leo Laporte [02:34:13]:
The shuttle we reused, but it was a difficult thing to do. SpaceX able to reuse the Starship is a huge economic step forward in spaceflight. I guess I won't be able to show it being captured by the giant chopsticks here.
Molly White [02:34:28]:
I need to see the giant chopsticks.
Leo Laporte [02:34:30]:
Oh, they're right here. They call them chopsticks. It's just a lever arm.
Patrick Beja [02:34:34]:
Mechazilla?
Molly White [02:34:35]:
I feel like they're sort of over-talking or over-selling that a little bit. I was expecting much more.
Leo Laporte [02:34:41]:
Like Chinese chopsticks?
Molly White [02:34:43]:
Yeah.
Leo Laporte [02:34:44]:
Well, it's pretty impressive.
Patrick Beja [02:34:45]:
It's very short chopsticks. That—
Molly White [02:34:48]:
yeah, it's more of like a clamp, really.
Patrick Beja [02:34:50]:
Yeah, yeah.
Leo Laporte [02:34:52]:
Let me see if it doesn't roll off the tongue as much. Video. Yeah, it's a, it's a big deal.
Molly White [02:34:58]:
Oh yeah, I'm not, I'm not trying to downplay that. That was a big deal. I was just— And
Leo Laporte [02:35:02]:
they've been able to catch boosters before, but, but, uh, by capturing the, uh, upper stage for the first time—
Patrick Beja [02:35:10]:
But didn't like part of it end up in the ocean?
Leo Laporte [02:35:13]:
Yeah, there were some other issues that didn't go so well. But nevertheless, massive step forward in affordability of spaceflight because if you can catch it and reuse it, and they say they think they could reuse them as much as 10,000 times.
Harper Reed [02:35:29]:
That's great.
Leo Laporte [02:35:31]:
That is huge.
Harper Reed [02:35:32]:
Yeah.
Molly White [02:35:33]:
I like that they do it once and they're like, the next 10,000 times.
Leo Laporte [02:35:36]:
10,000 times we can do this.
Patrick Beja [02:35:39]:
It's, you know, I'm not a big fan of Elon Musk. Uh, for reasons I think everyone will understand. But yeah, I have a lot of friends and listeners who are like, oh, anything he touches is, is wrong. And like, the only reason SpaceX exists is because NASA subsidies and like the government saved space. And I'm like, did you think they weren't giving money to like, like Boeing and, uh, whatever? This is so impressive. Yeah, like, these are so—
Molly White [02:36:12]:
Things can be true, right? It can be something first.
Patrick Beja [02:36:15]:
And also there's just a lot of people that don't want to believe. Maybe it's a, it's a French thing actually, but they don't want to believe that anything— well, no, maybe it's not.
Leo Laporte [02:36:24]:
Yeah.
Patrick Beja [02:36:24]:
And I— it's— yeah,
Leo Laporte [02:36:26]:
here it is. Let me—
Molly White [02:36:27]:
I also don't like seeing Elon Musk succeed, but that's cool.
Patrick Beja [02:36:31]:
That's why it was so cool that he waited At $474 million, right? And
Molly White [02:36:36]:
there it is.
Leo Laporte [02:36:37]:
Hot. Let's see.
Patrick Beja [02:36:38]:
But yeah, this is incredible. Like, this is— we live in a, in a world, in an age of wonders. Oh my God,
Harper Reed [02:36:48]:
this is a real thing.
Leo Laporte [02:36:50]:
This is not AI.
Patrick Beja [02:36:52]:
Is it though? Did this happen?
Leo Laporte [02:36:54]:
Yeah.
Patrick Beja [02:36:55]:
Did he actually launch a rocket?
Leo Laporte [02:36:57]:
Did it? Yeah, look at this. And the chopsticks closed and it got That is pretty cool. Yeah, remember the shuttle had to land as a plane, and even though they were able to reuse the shuttle, the boosters were lost each time. So now the whole thing is recoverable in theory.
Patrick Beja [02:37:14]:
Imagine if it's like this and they didn't plan for what happens afterwards. It's suspended in the air and they're like, guys, we got it! What do we do now? It's very heavy!
Molly White [02:37:28]:
Is this one of the ones where the— where they catch it like the base also moves, and— or is it only the rocket that is maneuvering?
Leo Laporte [02:37:36]:
They catch the whole thing. They weren't able to do that with this one because they ran out of fuel on the booster.
Molly White [02:37:41]:
But I mean, there have been some— like, I remember a while ago the exciting thing was they had like a platform that basically was able to move to catch the rocket as it came down. I'm wondering—
Leo Laporte [02:37:52]:
Oh yeah, they used to land the boosters Uh, on, uh, floating platforms.
Molly White [02:37:56]:
Yeah. Right. Does this platform move at all, or is it only—
Leo Laporte [02:37:59]:
This is on land, right? Yeah, this is— I can see it's on
Molly White [02:38:02]:
land, but I wasn't
Leo Laporte [02:38:03]:
sure if there was some sort of control.
Patrick Beja [02:38:06]:
No, it's the rocket that, that corrects its trajectory in real time to be at the, at the, on the right spot.
Leo Laporte [02:38:14]:
I don't know how to get rid of that.
Harper Reed [02:38:15]:
Is it— where is this? Is this in Texas?
Leo Laporte [02:38:16]:
Computers are hard. Yeah, Boca. It's actually, uh, they now call it, uh, Starbase? Starbase in Texas, yeah.
Harper Reed [02:38:24]:
They changed the name of Texas to Starbase.
Leo Laporte [02:38:26]:
It
Molly White [02:38:26]:
was Boca Chica.
Harper Reed [02:38:28]:
Yeah, they did.
Patrick Beja [02:38:29]:
Well, yeah.
Harper Reed [02:38:31]:
Not yet.
Patrick Beja [02:38:31]:
It is pretty amazing. And all of this will now be used for data centers in space. Forget Mars, forget the moon, forget everything. Data centers in space. That's the new thing. Nvidia chips in order to get those data centers up, which is why he wants freaking Jensen Huang reinstated, because they had an understanding.
Leo Laporte [02:39:00]:
Well, it's all coming together. Data center in space is up there with Google chips in it right now. Only 4 GPUs, but they're up there.
Harper Reed [02:39:07]:
Yeah, they're little tiny ones, aren't they?
Leo Laporte [02:39:08]:
Yeah, they're not— it's—
Harper Reed [02:39:09]:
I feel like that's not a data center.
Molly White [02:39:12]:
It's just a computer.
Patrick Beja [02:39:14]:
It's a quanta of data. But it's also like, who was it? It was some Google exec who was in charge of computation or something who was like, yeah, it works, but it costs about 1,000 times what we would pay for it on land. We have enough energy, we have enough everything. It works, It works, but it's kind of dumb.
Molly White [02:39:43]:
What's the benchmark like?
Patrick Beja [02:39:46]:
Yeah.
Leo Laporte [02:39:47]:
Yeah. Ladies and gentlemen.
Patrick Beja [02:39:48]:
But it works, you know? Not so long ago, like a year ago, people were saying it doesn't work. Right.
Leo Laporte [02:39:54]:
And it does. They were saying AI didn't work a couple of years ago. Yeah. Thank you, Patrick Béjaud, for being the optimist in the group.
Patrick Beja [02:40:03]:
Was I?
Leo Laporte [02:40:03]:
I don't know. I think we have to thank you for doing something anyway. notpatrick.com. For being here. Thank you for having me. For staying up late.
Patrick Beja [02:40:11]:
Thank you.
Leo Laporte [02:40:11]:
Till 2 in the morning. Thank you. I appreciate that. Uh, it's my pleasure. It's a pleasure to see you. Thank you, Molly White, for all you do. Uh, Web3 is going just great. Citationneeded.news.
Leo Laporte [02:40:22]:
And of course, you're still an active editor at Wikipedia, for which we are very, very grateful. Uh, it's one of the great, amazing successes of the internet.
Patrick Beja [02:40:33]:
We want AI is doing just great.
Molly White [02:40:36]:
People ask me for that probably once a week.
Leo Laporte [02:40:39]:
It's a little more complicated than crypto.
Molly White [02:40:44]:
Right. Yes.
Leo Laporte [02:40:46]:
And to our nihilist of the group, Mr. Harper Reed. I don't know. Are you a nihilist? You're a positivist.
Harper Reed [02:40:52]:
No, no, no, no. I'm pretty positive. I just am not as positive as Patrick. So there seems to be a gradient here.
Leo Laporte [02:41:00]:
I am so I'm so mixed. I'm positive and negative at the same time. I don't— I,
Harper Reed [02:41:05]:
I'm very happy. Reality. We're realists.
Leo Laporte [02:41:09]:
Yeah, we're realists. Maybe that's it. Very happy playing with my local AIs. I have to—
Harper Reed [02:41:15]:
I'm happy playing with my cloud AIs.
Leo Laporte [02:41:17]:
Yeah, they're all— they're really amazing.
Harper Reed [02:41:18]:
I like them all.
Leo Laporte [02:41:19]:
They're amazing what they can do. Yeah, my little buddies.
Harper Reed [02:41:23]:
Little buddies.
Leo Laporte [02:41:24]:
Thank you, Harper. 2389.ai. And if you want the crazy If you want to buy a shirt, AI is going to kill us all. Yeah, go to shop.2389.
Harper Reed [02:41:33]:
Yeah, there's many subdomains. They all are magical.
Leo Laporte [02:41:36]:
Yeah. Thank you everybody for joining us. We appreciate it. We've been doing TWiT now for 21 years and it's just getting weirder and weirder. I think the next 21 are going to be really wacky. Uh, in fact, I would say probably the next 3 or 4, it'll be a virtual Leo. That's for sure. We are glad you come and, uh, and visit us every week.
Leo Laporte [02:41:57]:
I hope you come back next week. And as I have said for so many years now, thanks for joining us. We'll see you next time. Another twit is in the can.