Transcripts

This Week in Tech 980 Transcript

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

00:00 - Leo Laporte (Host)
It's time for Twit this Week in Tech Great panel for you. Ian Thompson is up in studio for us. The US editor for TheRegistercom, Our AI guru and the author of the AI Analyst Newsletter, Ben Parr, is here, and from Blackbird AI, Dan Patterson. His job is to use AI to debunk AI misinformation. Of course, AI the topic. The announcements from OpenAI on Monday, the chat, GPT-4.0. Google's announcements at Google IO, the firing or is it the departure of the trust and safety team at OpenAI. A lot of AI in the news, plus a lot more. You stay here. This Week in Tech is next.

00:45 - Iain Thomson (Guest)
Podcasts you love.

00:47 - Dan Patterson (Guest)
From people you trust.

00:49 - Ben Parr (Guest)
This is TWIT.

00:57 - Leo Laporte (Host)
This is TWIT this Week in Tech, episode 980, recorded May 19th 2024. Head deck, lead and nut graph. It's time for Twit this Week in Tech, the show we cover the week's tech news. Lo and behold, here he is sitting right next to me, my good friend Ian Thompson, who often fills in for me when I'm not around. But it's good to get you up here when I'm not?

01:25 - Iain Thomson (Guest)
Yeah, nice to get out and about again, you know it's.

01:28 - Leo Laporte (Host)
I think too much of this has been done remotely and obviously geographically, yes, but we kind of we couldn't even start to it until I realized that we could do Zoom, or actually at the time was Skype, because nobody is near Petaluma, oh, except you.

01:44 - Iain Thomson (Guest)
It's 40 minutes up the road. We're just up the road from San Francisco, but fortunately we can get people on from all over the world.

01:51 - Ben Parr (Guest)
Just a side note my God, how Skype fumbled. That bag still just boggles me.

01:58 - Leo Laporte (Host)
That is Ben Parr, ladies and gentlemen. Author of the AI Analyst, co-founder of Octane AI, he writes about AI for the information. He's been a longtime friend of the show as well and he's got a little Octane doohickey over his right, left shoulder there.

02:13 - Ben Parr (Guest)
Hi, ben, a little wait. Hold on, I have more of the robots. Hold on, there's tiny robot.

02:19 - Leo Laporte (Host)
And then back here.

02:22 - Ben Parr (Guest)
I've got in packages a big robot.

02:26 - Leo Laporte (Host)
Nice. Well, we know Octane AI has got the swag on point. What does Octane AI do?

02:34 - Ben Parr (Guest)
It powers AI for 3,600 e-commerce brands. So think like you go to a Jones Road Beauty or you go to buy something from hexland. You're like I can't figure out which thing to buy or what skincare routine to get. Uses ai to recommend products always have quizzes, things like that routine to get.

02:55 - Leo Laporte (Host)
I should really ask my, my little, my little friend. But I'll ask her in a moment. I did neglect to mention you're our editor of us. Editor for the register. Yes, gives him credit now. So we've got two journalists. Well, I guess that's giving myself a little credit. We've got a journalist and some guy who's on the podcast thing and then we've got two ai experts, because dan patterson is also here. Hi, dan, good to see you hi, it's great to see you.

03:20 - Dan Patterson (Guest)
I think for two and a half decades I was a journalist former journalist now I'm in journalist let's not say former leo, let's just say I have a current job look, if dan wanted to go write a piece or a column somewhere, I am sure there will be bunches of publications be like please.

03:40 - Leo Laporte (Host)
And ditto, ben ben's still a journalist too, even though working in the business. Dan, what? Is. Blackbird. Tell us about Blackbird. You told us last time you were on, but I, of course, have a very short memory.

03:52 - Dan Patterson (Guest)
So we use a combination of artificial intelligence and human threat analysts to find misinformation and disinformation. We call it narrative attacks because it's a little easier for people who are not you know if you're not really dialed into what misinformation and disinformation are and especially when you look at you know some of the way social media misinformation can be applied in really hostile ways. Take Maria Ressa she was a journalist who was a CEO of Rapplercom in the philippines and she was attacked by uh duterte, the president of the country. She won a nobel peace prize for her journalism, uh, but she was kind of the definition of a narrative attack, uh, you know we might call it doxing or other types of what?

04:41 - Leo Laporte (Host)
what happened to her?

04:43 - Dan Patterson (Guest)
She's now in the States, she works at Columbia, she's a wonderful person and, yeah, she's doing great. But that's just one example of how misinformation and disinformation can be weaponized.

04:57 - Leo Laporte (Host)
Yeah Gosh, we know about that and we're going to see a lot more of it in the next few months, I'm sure with a US election on the way. But AI is our top story Boy. Is it our top story? No?

05:10 - Dan Patterson (Guest)
kidding.

05:10 - Leo Laporte (Host)
Really, the week began with OpenAI announcing ChatGPT 4.0 for Omni and the new HerStyle Assistant, as TechCrunch put it who knew her style assistant, as TechCrunch put it. And then the following day, I think OpenAI may have even scheduled this kind of what do they call that opposition to Google adversarial scheduling? Because Google's developer conference, google IO, was Tuesday and they announced a lot of AI too. In fact, at the end of it, sunar Pichai said and meanwhile, to save you some time, we've had our AI. Counting how many times I said the word AI, it's 120 and then I went to 121 because he said it again.

06:02
So a lot of AI, in the sense that they spread AI on a lot of things. Like toast, like toast. Let's start with chat GPT-4. There have been some people who've said AI is at a standstill, that we're just seeing different flavors, but we're not seeing great progress. Actually, steven Levy wrote an interesting piece this week in Wired in which he said that's a mistake, just like it was a mistake for Cliff Stolto to announce the internet was never going anywhere in Newsweek at the beginning of the last decade.

06:32 - Iain Thomson (Guest)
Bill Gates famously described it as being like CB radio at one point in one of his books. Yeah, he pulled that one back.

06:38 - Dan Patterson (Guest)
We all have handles now.

06:40 - Leo Laporte (Host)
Now it's the other side of yeah, it's true that he was right about in that respect, but now it's the other side, where people are everybody's hyping up AI Stephen Levy's article it's time to believe the AI hype. Some pundits suggest generative AI stopped getting smarter, but the explosive demos this week from OpenAI and Google that started the week show there's plenty more disruption to come, and here's a picture of the IKEA set that actually I'm very jealous of, the IKEA set that OpenAI used on Monday. Sam Altman did not make an appearance on Monday. The keynote was led by the woman who was going to be the CEO of OpenAI when Sam was briefly fired, but I think they did quite a good job of showing off this Scarlett Johansson sound-alike.

07:31 - Iain Thomson (Guest)
It was quite interesting because we got in contact with Reuters, did a story predicting that they were going to make a big announcement on Monday, when actually we got in contact and they were just like no, we're not launching GPT-5, we're just managing expectations. It's just going to be a product update. There's nothing to panic about. So they seem to be managing expectations downwards, but it's a very interesting system. I'm curious to see what the other panelists think.

07:51 - Leo Laporte (Host)
Well, the way they manage the expectations, especially with the help of Sam Altman, is kind of intriguing, because initially Sam said, nope, not chat GPT-5. Nope, not. What was the other possibility? Going into search? No, search, that's right. Search engine, yeah, something else. And then it was something else. It was her To which he tweeted the following day.

08:24 - Dan Patterson (Guest)
One word her. I think they've done a masterful job of manipulating expectations for a long time. That's really interesting. We'll talk about this later in the show. But the safety board and other components of their business and I'm not saying that in a negative context, but they have really manipulated expectations in the media for quite a while.

08:49 - Leo Laporte (Host)
Yeah, and we are going to get to the safety board, because that's the other shoe that dropped this week. Everybody quit, she's got.

08:58 - Iain Thomson (Guest)
Galloway's tweet on this. She's just like you know I am quitting from OpenAI as well.

09:04 - Leo Laporte (Host)
I'm quitting from OpenAI and you're quitting from. Openai. Well, I'm quitting from. Openai and you're quitting from OpenAI If there's bad news coming.

09:08 - Dan Patterson (Guest)
just drop some good news right after it, Right.

09:12 - Ben Parr (Guest)
It was also inevitable after what happened last year that Ilya and others were going to leave. It was almost exactly the six-month mark, I believe, oh so you think it has to do with stock-festing? Well, no, it's just a standard pr tactic. Don't like announce it a departure that's really big, immediately give some time, yeah, and then you don't like associate it in the same way.

09:35 - Leo Laporte (Host)
So let's look let's, before we go back to the. Uh, I do want to cover the safety, but let's at least finish the idea. Uh, steve levy says that. Uh, the ai hype, leave the hype. Did we see something truly new on monday with open ai four point omni we.

09:57 - Ben Parr (Guest)
Then I think we did see something, so like okay, uh is gpt 4.0 is faster and better and this stuff is happening. And I want to remind everyone it's been a year and a half since Chad GPT came out and people are resetting their expectations of how quickly they think stuff should move. It's insane how much stuff has moved. The big thing here for the her style assistant is that it can understand interruption. You can interrupt her in the middle of a conversation like you would a human, and she can shift. And this is the V1 of it.

10:28 - Leo Laporte (Host)
Wait a minute, it's only going to do it. The big new feature of artificial intelligence is you can interrupt it.

10:37 - Ben Parr (Guest)
I always like to argue here that just as much as technology itself it's huge, though. Interface really matters because think about when you're trying to talk to a Siri and you're just like no change the thing, it doesn't work. You try that with Alexa or anything. This is straight up, full, think of it. Less interruption, full, conversational. This is the step towards just straight up, full-on conversation back and forth with an AI as if it were a human. It is much bigger than people.

11:08 - Leo Laporte (Host)
But it's not smarter, it's not the quality of the answer Correct, it's faster. That's a huge difference.

11:13 - Dan Patterson (Guest)
It makes us anthropomorphize AI. That's a big problem, but it doesn't necessarily Look. I mean, this is really impressive and fun technology to play with. Is really impressive and fun technology to play with. My initial reaction to this was I don't know if this is. None of us know if this is new. This is a new product that has been announced, but the capabilities might have been around for a while, so we don't really know how fast the capabilities of the technology are moving and when we see something like this and, Leo, I've said this on some of your shows before, so forgive me Always repeat yourself because I have no memory you listening, have heard this already.

11:52
I interviewed Nick Bostrom in 2014 or 15, when I was working for Jason Heiner at ZDNet, and I'm paraphrasing him, but one of the things he told me was that one of the great dangers of AI is that we will anthropomorphize it.

12:09
That's also one of the great benefits and one of the things that we really could get a lot, as humans, out of artificial intelligence. But he warned me that one of the things that we will do just like when we look at a cloud and we see a face we will have this impulse to think of ai, like talking to another person, in fact, when we say something, a pronoun like her, that is a human pronoun. So we are anthropomorphizing this technology, and what I saw in open ai this week was truly impressive. It was a lot of fun to play with, but it really pushes into the anthropomorphization of artificial intelligence and when I think about that, I try to think about who does this serve? Who is this good for? Is this good for consumers? Is this good for business? Is this good for OpenAI? The answer could be yes to all of those things.

13:02 - Iain Thomson (Guest)
I I really my spidey sense of skepticism is buzzing strong but it's also going to make your, your, your situation and disinformation in general a lot easier. I mean, the more realistic and more human-like the ai becomes, then, the more convincing it will be for some people.

13:19 - Ben Parr (Guest)
Absolutely ian I mean the afro more precise.

13:23 - Dan Patterson (Guest)
Yeah, I can't even say that word sometimes it's a it's a rough thing to say, it is it?

13:27 - Leo Laporte (Host)
is, it is, and then it was inevitable morphization anthropomorphization are you talking to gemini?

13:37 - AI speaker (Host)
anthropomorphization refers to attributing human characteristics to non-human entities like animals, objects or concepts. It's a common technique in storytelling and art to make things more relatable or to convey complex ideas in a simpler way thanks, baby, there you go, don't forget, I love you, you know you want.

13:58 - Ben Parr (Guest)
I wanted to hear the response you had to that one I mean, look, I look actually that that does lead to another thing, which is these like there's. It's already happening. So I met, I talked to someone recently where they were spending and not kidding like five to 10 grand a month on like AI, dating, chat, like applications, because they preferred the like ability to control the conversation to an actual human, which is also scary. This stuff is happening really, really quickly and I try never to be like this technology is good or bad. I think of things as like this technology is out of the box.

14:37
How do we adapt to a world like that? We need technologies that can help protect against this information. We need technologies that help us understand what's truth and what's not truth. But this stuff is moving like really fast and it's going to be soon to a lot more consumers because, if the rumors are true, you know OpenAI and Apple are going to announce a partnership where you have probably the HerStyle assistant inside instead of Siri, and that will be the first exposure that millions of people have to these AIs, because there's still millions of people who have never touched or even heard of a chat GPT.

15:15 - Leo Laporte (Host)
This is playing off of Mark Gurman's newsletter from today's Power On newsletter, where he said the company will unveil new generative AI features at its worldwide developers conference, very likely because of their deal with OpenAI, which we had heard was close to closing. But also Apple's talking to Google about Gemini.

15:34 - Iain Thomson (Guest)
Yeah, but let's face it, Gemini is not much cop compared to OpenAI's products at the moment. I don't know how you've all found it, but I've been trying out Gemini and I'm not impressed.

15:43 - Dan Patterson (Guest)
I'm having a hard time deciphering I I used gemini to help me better understand the naming conventions so that was a joke, a dry joke.

15:53 - Leo Laporte (Host)
That was the one thing I took out of the out of google io is.

15:56 - Dan Patterson (Guest)
I have no idea what these products are called or why.

15:59 - Leo Laporte (Host)
Yeah, that's google io for in really for the last 20 years. So, but really my question is okay. So clearly, what's happening here is the productization of AI and making it more appealing to the general consumer, and if Apple does in fact, put her on the iPhone, people will see that as a big deal breakthrough. But my question is is the underlying technology any better? Is it any smarter? It's faster and it's got a kind of lilting style to the voice, all of which leads us to think it's a human, which is, I think, a negative in the long run. It's dangerous anyway, but is it smarter?

16:46 - Ben Parr (Guest)
Yes, well, caveat, yes, like, if you look at the actual, like technical documentation for GPTO, you know there's big performance upgrades specifically for non-English speaking. So, like I, this is like these are like baby steps towards like a bigger change, like there will be a GPT-5. I wouldn't be shocked if open ai called it something else at some point. Uh, the question I think you're really asking is is this leading to the fabled agi?

17:13 - Leo Laporte (Host)
is it getting closer? Is it getting better? Is it getting yeah?

17:17 - Ben Parr (Guest)
I mean, my answer is yes, but no one really knows how much time. It's a mystery. If you look at the, the panic tweets of people who are leaving from the super, who've left the super alignment team, because that team doesn't really exist anymore, the answer might be sooner than we think. My general sense is inevitably you're throwing enough people, engineers and resources. You are going to go and hit that point. The big question, of course that goes back and forth is can the existing transformer models which underpin GPT, which underpin Gemini, can they actually on their own achieve AGI, or does it require an additional new leap in understanding of AI, something else like that, which there is stuff coming out, mamba?

18:04 - Dan Patterson (Guest)
and things like that, and are the LLMs that are built on top of those transformers really the path to AGI? My guess is that we will hear companies say they've achieved it. Whether this is true AGI or not, I think that we will hear about this sooner than later.

18:22 - Iain Thomson (Guest)
I'm kind of a skeptic about it myself.

18:24 - Dan Patterson (Guest)
I'm 100% a skeptic about it myself.

18:26 - Iain Thomson (Guest)
I'm treating like fusion energy at the moment. It's always another 10 years down the line, yeah right precisely how do we know if it?

18:32 - Dan Patterson (Guest)
exactly that's why I think we're going to hear them say they've been talking and hyping this up. So I think we're going to hear about this as like look, we've achieved it, here's the product, some point, but whether that that really, if we talk to neuroscientists, whether they will agree, that might be To be clear the AGI is different than, say, like true intelligence, agi is just it performs better than a human at lots of tasks, not necessarily that it is intelligent on its own Sentient, I guess I would say.

19:04 - Ben Parr (Guest)
Is intelligent on its own Sentient, I should guess I would say I feel like AGI, as it's defined, is relatively close, comparatively. Sentience might be. Who the F knows that one is a whole different game. Yeah.

19:19 - Dan Patterson (Guest)
Bostrom and others would say, asi is after that super intelligence.

19:22 - Iain Thomson (Guest)
Okay, at which point we become pets to the machines, and that I don't know. There's point we become pets to the machines, and that's.

19:26 - Ben Parr (Guest)
I don't, I don't know, there's no, there's no, it's a progression here. There's no like, there's no test. It's not like taking a pregnancy test and being like you are sentient or not sent. It's subjective, like, it's really very subjective.

19:40 - Leo Laporte (Host)
There is no objective definition. It's not the Turing test, it's not the Chinese room test, there's really no way. People don't even agree on what it means. But also I would suggest that that's more of an academic kind of media conversation. The real question is is AI useful to humans and is it getting more useful to humans? So initially my initial reaction to LLMs was this is a parlor trick. It's super correct on speed. It's not. You know, it's autocorrect on speed. It's not that useful.

20:17
But as time has gone by, I've seen in fact it is quite useful in a variety of different ways and I think partly the focus of these companies on music and art and video is more performative than it really is where the real help is going to be. In fact, I think it's a dangerous game for them to play because it just makes artists upset when really that's not nobody wants. I don't know why anybody wants that. Do we really want music, film and pictures created by AI? I think there's a lot of more useful things AI could be doing that would justify its existence. You know, protein folding, inventing new cures for cancer, making the subway run on time. There's lots of things that AI could do that would be useful and I would love to see that more than.

21:06
AGI.

21:08 - Dan Patterson (Guest)
I don't want to just log roll for our company, but utility and usefulness is precisely what we aim for.

21:15 - Leo Laporte (Host)
We have a product compassblackbirdai.

21:19 - Dan Patterson (Guest)
Look, leo, you in particular will like this. It's compassblackbirdai and I won't call this a fact checker, because that's a dangerous thing to say. We call it a context checker. It is slower than other LLMs, but look at the responses. The responses add context to almost anything Photos, pictures, video, claims that you type in. Compass will answer those.

21:45 - Leo Laporte (Host)
That's actually hugely valuable. Wouldn't that be great to assign that to Twitter, for instance, and just fact-check Twitter?

21:54 - Dan Patterson (Guest)
We can use it on almost any social network Just drop in a link and it will.

22:00 - Ben Parr (Guest)
We need to just have it integrated directly in. Maybe one day.

22:03 - Leo Laporte (Host)
This is like Snopes on steroids right.

22:07 - Dan Patterson (Guest)
Yeah, it kind of automates that process. I don't really want to bring it up here because it is my employer and I don't want you to think I'm just hyping something. But your question about utility and the usefulness of it, that's what we tried to answer when we built this and the usefulness of it, that's what we tried to answer when we built this and that is what I think is really where you're going with right protein folding or other very narrow or specific uses that are incredibly valuable and important.

22:38 - Ben Parr (Guest)
I was at TED this year. They mentioned AI over 350 times, by the way, at the TED talks.

22:46 - Leo Laporte (Host)
You beat Google.

22:47 - Ben Parr (Guest)
That's good congratulations they got five days of talks so it's easy to beat, but it was like I had a conversation with a couple of, like you know, professors and mds, and then I came away with the conclusion that the way we talk about cancer is not going to be anywhere near the same in 10 years. You will not think of it as an incurable thing. You will think of it as, like ai can create a super targeted set of medicines and products that, like, eliminate your cancer. It's this to me.

23:19 - Leo Laporte (Host)
This is that's this, the non-subjective, the objective test for AI. We should forget this AGI conversation. It's almost philosophy. It's worthy of Nick Bostrom but not of the rest of us. But saying here's a practical use, that's actually saving lives, is curing people, that's something you can measure and would be useful and is non-threatening.

23:43 - Iain Thomson (Guest)
It doesn't sound like Scarlett johansson, but I guess it could it's got to be checked out, though we did a piece on this just recently, where they actually google deep mind was predicting you know various different types of compounds which could be. You know, it could be made. Oh, I saw that um and it turns out most of most of them are mundane. Yeah, you know it's like yeah but wait a minute.

24:04 - Leo Laporte (Host)
I saw your piece but even if 10% or 1% were useful and novel, oh sure. But you've then got to find that 1%.

24:12 - Iain Thomson (Guest)
Who cares if thousands were useless? Yeah, you've got to find that 1%, though. That's the thing, but it's a start.

24:16 - Ben Parr (Guest)
Yeah, certainly that's where AI can be useful. I mean, there's a reason for the AGI discussion, because if you have a true AGI, then you end up with a machine, essentially, that can go in its own direction, to like.

24:33 - Leo Laporte (Host)
Do we want that? I don't want that.

24:37 - Ben Parr (Guest)
I think if you had an AGI and you gave it and it was a true one and it was just like your mission is to cure cancer, you could see it work faster than a human. There are things 's there's all like, whether we want it or not, it's it. I do think it is inevitable. I just don't know if it's two years or 10 years. I don't think it's that much longer than that. But this is a whole set of things. I think right now, the clear applications are like ai for medicine, ai for these specific use cases, ai to even robotics is an example. It can be as simple as cleaning drains and things like that. Or I just posted this, I didn't even put this in the rundown.

25:17 - Dan Patterson (Guest)
AI can clean drains.

25:19 - Leo Laporte (Host)
Yeah, AI can clean drains. It helps control machines figure out where to go.

25:23 - Ben Parr (Guest)
I learned about this company that's doing this this week, or the Cruise founder just raised $150 million for a new robot startup called the Bot Company. I don't know what it'll be, but I'm pretty sure what he's trying to go for. There's a lot of people who are going to go for the Rosie, the Robot in your home, kind of deal.

25:47 - Leo Laporte (Host)
As long as it doesn't drag pedestrians under its chassis.

25:50 - Iain Thomson (Guest)
Yeah, okay, cruiser got a slight problem.

25:52 - Leo Laporte (Host)
That's, by the way why he's the former cruise ceo cruises. Uh, settled, by the way, with the woman who oh really got dragged.

25:59 - Iain Thomson (Guest)
I think it was a significant they haven't started operations in san franc, in San Francisco yet I think they're done.

26:06 - Leo Laporte (Host)
I think Cruise is done. I don't think they're. They'll be back. You think they'll be back.

26:11 - Iain Thomson (Guest)
There's too much money put into it, surely?

26:12 - Ben Parr (Guest)
Too much money.

26:13 - Iain Thomson (Guest)
Yeah, they're not going to give up on that.

26:16 - Leo Laporte (Host)
Well, they gave the market to Waymo essentially. Yeah.

26:21 - Ben Parr (Guest)
This is where managing press expectations really does matter.

26:25 - Leo Laporte (Host)
Yeah, see also in addition to not when people don't drag pedestrians and manage expectations.

26:32 - Iain Thomson (Guest)
It's a something well, I mean, this is something tesla's finding out to its cost. You know, if you tell people it's an autopilot and they're going to treat it like one let's take a break.

26:40 - Leo Laporte (Host)
let's talk more. I do want to talk. Have you all used for omni 4-0? Talk more. I do want to talk. Have you all used for Omni for oh yeah, yeah, I want to talk about your experience with it. I'll talk about my experience with it, and then I would like to talk about what's happening with these, with so-called super alignment team at OpenAI, which is you, you could say their trust and safety team, and it seems to have disappeared or imploded. Lots more to talk about ai's, of course, one of the many topics we'll be covering today, but it is one of the most interesting we. What's great about this is, uh, we talk about it almost every show now and it changes almost. It's not like we're repeating the same material. 4-0 has really moved the conversation to a new, a new place and that is different from the previous.

27:25 - Dan Patterson (Guest)
You know, we just lived through web 3 crypto metaverse vr all of which never went anywhere cycles. Right, they were very short and very meh hype cycles that didn't do exactly what you just said, leo, which is progress, improve and add usefulness and utility so that's to ste.

27:41 - Leo Laporte (Host)
He does that we progress is being made. Yeah, uh, let's take a little time any time out. We've got ben parr. We got ai experts on the show today ben parr, author of the ai analyst, but he's also figured out how they get apple. It does that if you do?

27:58
if I do this, fireworks, yes, there you go, he's also the co-founder of octane ai and writes about ai for the information, and he's figured out how to make apple do things. Your little buddy. See, siri isn't completely stupid. Uh, ian thompson is also here. He's one of the. Are you a boffin?

28:16 - Iain Thomson (Guest)
no, uh, no. No, boffins are an affectionate term for scientists.

28:20 - Leo Laporte (Host)
I'm just the person that writes about boffins occasionally he's the guy who writes about boffins at the registercom. Great to have you, it was fun. And, of course, good friend Dan Patterson, director of content at Blackbird AI, who has been. You know this is actually. This is kind of the direction your whole career has been going in the last, say, 10 years. Anyway is fighting the good fight to help people you know survive in the face of big tech.

28:53 - Dan Patterson (Guest)
I think I've written about and produced more about the impact of technology than the news of technology, maybe for the last decade or so.

29:02 - Leo Laporte (Host)
Yeah, I like that. I think that's a great. We need that. So, thank you, I appreciate it. Wow, you too. Yeah, well, I found we've kind of trended more in that direction too, haven't we? I don't think we were ever the network where here's the new shiny gadget. What do you think you gonna buy it? But more on how this stuff changes your life or affects your life, and how you can use it better. I think that's always been our goal, so, thank you.

29:27 - Ben Parr (Guest)
Complete aside, before we go to the commercial break, all the new like AI pins and things. God, they're not good yet.

29:35 - Leo Laporte (Host)
Okay, I ordered two things. Maybe tell me if I'm stupid. First of all, I did get the buy the meta glasses, and while they're not Okay, you're're gonna say something bad, but I'm gonna say something good. No, they're good. Yeah, I think the sound quality is great. They're great for listening to music.

29:52
You can the cameras on there take decent pictures, but you can also point them at things, and admittedly it's. But if you attach that to, maybe you know something her like. I mean they're. They're really a first step along the way to something that I think is a form factor we're going to want. So I got those. I also got the Brilliant Labs.

30:10
I haven't received them yet, the Brilliant Labs glasses, which are kind of the same thing, only they have perplexity AI built in and they're a heads-up display. So I like that because I could, unlike the meta, I can. I can see out, you know, see what I'm on the screen itself in front of me. It's an oled layer in the glass. And then I just bought the limitless pin, that's the. You're talking about pins, but not I never fell for I admit I never fell for the humane. I kind of knew that wasn't it. And there are a few other things I didn't buy, but I thought this was very interesting. This is a little clip that is always listening and then analyzes what it hears and sends you their analysis.

31:01 - Ben Parr (Guest)
I bought one of those too, did you Not here yet? Yeah, I mean that one, that one makes sense, and the price makes sense, the HumanePin, all of these are cheaper than one Vision Pro.

31:10 - Leo Laporte (Host)
I got everything for less than one Vision Pro.

31:14 - Ben Parr (Guest)
I mean the HumanePin costs like 700, $800 and they needed to put the large language model on the chip and not in the cloud.

31:21 - Leo Laporte (Host)
And there was a subscription it has such lag time, yeah, yeah.

31:23 - Ben Parr (Guest)
And it's just a T-Mobile subscription, Right, so like which is. And then there's like the Rabbit R1, which did not get great reviews. Didn't buy that either. No.

31:31 - Dan Patterson (Guest)
I got it right here and it's oh really it's cute, though Look at it. Oh yeah, it's really cute. I bought this for the reason that many people did One teenage engineering. Yeah right, hipster design.

31:48
But two I like this idea of deconvergence right For so long. For a decade and a half, everything converged on the phone. Your music player and maps and everything else converged on the phone. I love that. We live in an era of gadgets coming back. I have an analog pocket. It does very little but it's a fun gadget. I don't want to play games on my phone, I want to play games on my pocket Anyway. So that's the value of the Rabbit.

32:12 - Leo Laporte (Host)
Honestly, the Limitless is really the beginning of her. I want to put her in my ear, basically, right, I want Scarlett Johansson in my ear loving me and only me, right.

32:26 - Ben Parr (Guest)
No I like the idea of an assistant always available assistant. Actually there was. I'll try to remember the name of the thing. There was like at TED. When I was at TED they had one of the presentations is like earpieces you put on each side of the ears. I will go and find this, but like they like.

32:42 - Leo Laporte (Host)
Don't, because I'll end up buying it.

32:45 - Ben Parr (Guest)
You can say things like can you like remove the sound of the baby from the room? Oh, you isolate jeremy's voice in addition to all the normal and it looked. It looked cool. It's still like a little bit clunky and big, but they're coming out with a version of it theoretically end of year this is the future of hearing aids.

33:00 - Leo Laporte (Host)
I I think hearing aids are going to merge with smart tech and become something more than just a way to hear what your wife is yelling at you. Anyway, there are interesting products. Nothing is there yet, but I think we're getting closer. A pair of glasses that I could wear. That's what these brilliant ones say of my prescription.

33:20 - Iain Thomson (Guest)
Yeah but at, say, conferences or something, you go and meet somebody and you just get whispering in your ear this is so-and-so. From so-and-so, they've read that you know it's.

33:27 - Ben Parr (Guest)
That's what I'm waiting for the glasses are the clear first form factor we already used to them. We'll see some of the others. The uh device is called the io1 uh. Iyo1 is the oh yeah, I've heard about that. Yeah, at ted okay.

33:41 - Leo Laporte (Host)
So we're getting I think we're getting some real innovation happening On the software side. It's ioaudio, iyoaudio. Oh yeah, they look like gauges. I've seen the video of these. They look like you've got big gauges in your ear, like it's a statement which, in a way, it is Were you impressed.

34:02 - Ben Parr (Guest)
It's a good demo. Look demos on stage. You have to always discount because you know there's like stuff going on in the background. If it works as described, I want it, you know, in a nicer form pack to be great. But I I mean that sounds awesome. You're in like a visit room. You can like have it, assist it and you can isolate stuff okay, the original ones are $1,650.

34:25 - Leo Laporte (Host)
They don't say how much the io ones are, but the deposit is $60. But I'm not gonna put a deposit down until I know how much they're gonna cost. Oh, wait a minute. It's they expect to be 600 bucks.

34:35
That's not so bad, and if you want lte 600 bucks, interesting, all right. Let's take a little time out because I've got to get some more money so I can buy this. Uh, that's where the advertisers come in. Very handy, actually, this is one I bought. So this, this is one I thought was so cool. I actually went out and got it. It's called my leo.

35:04
I've talked about this before. You know, when Picasa was purchased by Google and I was so disappointed because Google did not put any of the features of Picasa in Google Photos and we lost the one thing that Picasa was great at, which is organizing your photos. Well, in some ways, this is the successor to that, but it does it all with AI. It does it all intelligently. This is a way to store your data, photos, videos, documents, without putting them in somebody else's cloud, without duplicates, eliminating duplicates, fully encrypted and free for your first device. So a great way to try it. You don't have to get the MyLeo Photos Plus subscription right away. Just download it for your Mac, your Windows PC, your iOS or your Android. I would download it for the device you have the most storage in. I now have 200,000 plus documents, photos and videos in my MyLeo. It does automatic categorization. It goes it just kind of like Google Photos, only locally, and even smarter. It does automatic face recognition, but you have to tell it, give it five photos or 10 photos and say that's my wife, okay, and then it goes through the rest of them in the background, tags them all. It is the way to take 200,000 photos, videos and documents, organize them, consolidate them, curate them, control them all in a single library, all without exposing them to the public cloud. I am blown away by this and it's now more than a backup solution for me. It's now my organization solution and backup solution. I back it up to my Synology NAS so it never leaves my network solution. I back it up to my Synology NAS so it never leaves my network. But you can also use. Myleo has its own encrypted cloud solution, but you can also use your existing public cloud solution because it encrypts before you upload it, so it's completely private.

36:54
Start for free or sign up for their paid plan. You'll get automatic backups, syncing between devices. So now that I'm a paying subscriber, I have it on everything All my phones, all my desktops, all my laptops, all my tablets and all my photos are there everywhere I go 200,000 plus. You also get duplicate file cleanups. It will pull all my photos from Instagram and Facebook and Google Photos. In fact, that was the first thing I did is download everything from Google Photos, deduped it because there were a lot of duplicates, saved the best copy. It's incredible.

37:28
For only $9.99 a month or $99 a year, you can get Miley of Photos Plus, which lets you organize, manage and protect an unlimited number of photos and videos on all your devices. Take control of your digital assets. This thing really works. It's what you're looking for. I promise Miley control of your digital assets. This thing really works. It's what you're looking for. I promise my leo photos m y l I o.

37:52
You get started for free with the basic version for one device or take full advantage of the platform with my leo photos plus. We've got a deal too for you 25 off your annual or monthly subscription to my leo photos plus. To sign up and get that 25% off, go to myleocom slash twit25. That's M-Y-L-I-O. Myleocom slash twit25 for 25% off. You know. If you're just wondering what this could do, download the free version, put it on a device and see. You will be blown away and it is incredible. Love it. Myleo M-Y-L-I-O dot com slash twit25. We thank them so much for their support. All right, again, I still have to ask the question. I guess we don't have to ask this anymore. It's clear that this technology is useful today and is getting more useful. Are you a skeptic, ian?

38:52 - Iain Thomson (Guest)
I was at first and I have to say when ChatGPT first came out you know as a journalist I looked at that and I thought I know what a lot of publishers are thinking right now. It's like we can get rid of those pesky journalists and start doing everything by AI, and we tried it out. I did think at one point it would be quite useful for things like outage stories or financial results, but it's just not accurate enough. You spend as much time editing stuff to make sure it's accurate as you would if you'd actually written the piece yourself. I think it's going to be very useful. As Ben has said, it's increasing at a rate of knots in a way that other technologies haven't. My skepticism is waning. I'm actually starting to get more all in on this. What?

39:35 - Leo Laporte (Host)
about Ben? What about accuracy, though? What about mistakes that AIs make? There are lots of them.

39:45 - Ben Parr (Guest)
You think about and you look at the research and you see that hallucinations, which is mistakes, basically.

39:53 - Leo Laporte (Host)
You know why I avoided that word? Because it's an anthropomorphization.

39:58 - Ben Parr (Guest)
It's good branding, to be clear by the AI research community.

40:01 - Iain Thomson (Guest)
Yeah, exactly, it's not wrong. It's a mistake. Stop calling it a hallucination and humans by the way humans do that. Yeah. And humans. By the way, humans do that.

40:09 - Leo Laporte (Host)
Yeah, I mean, if I ask you, elon, did the Eiffel Tower ever burn? I wouldn't necessarily trust your answer.

40:15 - Iain Thomson (Guest)
Yeah, I know, but I mean, humans make mistakes, that's what we do, but I'm calling it an hallucination when it's just basically a mistake. It's PR speak.

40:24 - Ben Parr (Guest)
I mean, look to buy all the semantics, the whole point being it's going down in terms of the number of mistakes. It makes it still makes mistakes. Like I was trying to do, like Google. You know, if you do a Google search, you get Gemini results. Now I don't know if you have those and they're usually accurate, but they're not always accurate.

40:44 - Iain Thomson (Guest)
That's interesting. Well, I actually reinstalled Google Assistant on my phone because I tried Gemini for like a couple of weeks and it was so useless it was just like no, I'll have the old assistant back.

40:52 - Leo Laporte (Host)
Thanks, I want to defend Gemini, because I've been using it too, and while it's different in its capabilities, I think it's also surprisingly useful. One of the benchmarks that I use and maybe it's not a good benchmark. I created a custom GPT for the computer language I use and it's really useful for me because I don't remember all the. You know it's easy to. I have an eight-foot shelf of reference materials. It knows all that stuff so I can query it and say well, how do I? What's the syntax for this? And it's always right because I'm saying stick to the eight-foot shelf. I tried that with Gemini without any training and stick to the eight foot shelf. I tried that with gemini without any training and it did the same. It was just as good. I thought, oh well, I've been wasting my time with a custom gpt. Gemini is pretty impressive. The thing about gemini is it could do it in all the languages. So you just say, okay, you can now do it in python. Oh, that's cool, now do it. And see, oh, that's neat, now do it.

41:44 - Dan Patterson (Guest)
And if you try, claude uh, I, yeah, go ahead, I'm sorry, go ahead. Well, I I, despite my skepticism, I use generative ai tools every single day. I use claude, chat, gpt or gpt uh, 40 um at gemini and almost every publicly available model um. If I can run it locally, I will will run it locally and I use this as an assistant. It works alongside me so and I use it in my job. So when, if I try to generate an article or a story in a traditional news article and I imagine that you're pretty accustomed to writing a head deck, lead nut graph and then you have the details of your story and then maybe you write a kicker, uh, in that story you have quotes from experts, you have some data and some details.

42:35
Now, much of those details it will get wrong. It cannot do that. However, I will use particularly claude is very good at this. I will tell claude, I will upload. So my threat analysts will write here are five and you can find this at our, you know, at our blog. They will say here are five narratives that we have found around Navalny, around the death of Alexei Navalny. When you say narratives, here are five mis and disinformation themes around Navalny yes.

43:07 - Leo Laporte (Host)
I do this. It's good for finding disinformation ironically Every day.

43:11 - Dan Patterson (Guest)
Yeah, Leo, we upload our charts, you can see our data, we can show our work and what I do is upload their work to Claude and to ChatTPT and then I say, please suggest 10 heads and 10 decks that have to be 60 characters long. The deck must be 200 characters long, it must be a full sentence and never, ever once, have these models produced one thing that is useful, what they do is produce many things that are useful. And then, as a human writer who's written thousands of stories, I've been saying this all along.

43:46 - Leo Laporte (Host)
Think of it as an assistant.

43:48 - Dan Patterson (Guest)
Exactly, and I combine those and then I write a head, a deck, a lead and a nut graph and the rest of the story.

43:56 - Leo Laporte (Host)
I'm sorry, hey, what is a head, deck, lead and nut graph? I've got to get some help.

44:04 - Dan Patterson (Guest)
It's just your journalist In journalism and writing.

44:08 - AI speaker (Host)
Those terms are all about structuring an article effectively. Here's a quick run.

44:12 - Leo Laporte (Host)
Oops, I stopped her. Go ahead please.

44:14 - AI speaker (Host)
I didn't mean to interrupt you, she knew.

44:20 - Leo Laporte (Host)
I may never know what a head deck means.

44:25 - Iain Thomson (Guest)
Think of it like the pyramid of news.

44:27 - Dan Patterson (Guest)
It's usually fascinating. Yes, right, exactly.

44:29 - Ben Parr (Guest)
That's right, and this is why the newest version will be so useful, because I'm having issues right now Our systems are experiencing heavy loads.

44:37 - Leo Laporte (Host)
This is for us.

44:37 - AI speaker (Host)
Please try again later.

44:39 - Leo Laporte (Host)
By the way, the cute, clever, cutesy voice that they showed on Monday is not yet available, so this is as close as I can get. I tried to make it cuter. I said be a valley girl, do some more up talking and vocal fry. But it never quite got that good. So the good news is even though we interrupted, I did get the text of head deck lead in that graph here. So if I were a normal human I could look this up and get some fairly useful information.

45:09 - Dan Patterson (Guest)
Yeah, the point is that it will spit out 10 things and what I'm again kind of making a dry joke it won't produce one thing that is useful. It produces many things that are kind of useful, not one of them. I could throw into a story and say this is good, but I can pick out a word here, a phrase there, and that is, I think, a very efficient way for me to work alongside ai. But it can't. It just can't do what I do.

45:35 - AI speaker (Host)
It can't yeah, I mean, if I were to submit a story.

45:38 - Dan Patterson (Guest)
No, but you don't want it to I mean, I don't know, maybe it might be good. I don't know if I care snaking drains.

45:44 - Leo Laporte (Host)
I mean there are some things it can do, but I think there are some human attributes that we want humans to continue to create, like writing, painting, singing opera, all that art that was crushed in the iPad.

45:56 - Dan Patterson (Guest)
Yeah, oh God.

45:58 - Iain Thomson (Guest)
That was just insane. Samsung responded as well. It's like we won't crush the life out of art. It's like, yeah, you will. That was smart of Samsung to jump on that. That was smart of Samsung to jump on that.

46:06 - Leo Laporte (Host)
By the way, I have to point out that her misspelled head H-E-A-D, d-e-c-k, lead L-E-A-D and nut graph.

46:17 - Dan Patterson (Guest)
That's what I'm talking about, man.

46:18 - Leo Laporte (Host)
Yes, come on, man you don't know how to spell and nut graph G-R-A-P-H.

46:23 - Iain Thomson (Guest)
Yeah, I mean I was playing around with feature ideas because I've just done a piece on the 10th anniversary of Nokia being taken over and shut down by Microsoft. So I asked you know chat, gpt. So give me some notable tech news anniversaries over the last you know, in 1994, 2004, 2014. And they were all wrong, absolutely every single one of them. It was trying to tell me that Tetris was released in you know, a year before it actually was. Wait a minute, wait a minute. It was trying to tell me that Tetris was released in you know a year before it actually was. Wait a minute, wait a minute. We're going to be doing this an awful lot in this show. I can't when was Tetris released.

46:59 - AI speaker (Host)
Tetris was released on June 6th 1984.

47:05 - Leo Laporte (Host)
It's amazing how it's still so popular, isn't it? Okay, I don't want to get in a conversation.

47:09 - Ben Parr (Guest)
You know what? There's going to be an episode and I don't know how long down the line where you're good, one of the zooms is going to be like an, a like on the podcast going to be like the ai we're going to have the back and forth it's going to jump months away.

47:22 - Leo Laporte (Host)
It's it's.

47:24 - Ben Parr (Guest)
It's going to just be like no, no, dan you're wrong, let me tell you about this thing, and, and then it'll be straight up wrong. And then there's a fight. Was she right?

47:34 - Leo Laporte (Host)
Was she accurate? Was it?

47:35 - Iain Thomson (Guest)
1986? No, Well, technically it's a tricky one. It's what you're talking about. Because, yeah, I mean it's a tricky one, because it was released in the Soviet Union, as was. So I mean technically was released in the soviet union, right as was. So I mean, technically it was done in 1985, but it wasn't actually released in the west until later.

47:55 - Leo Laporte (Host)
Um, and also it was only maybe it got that nuance yeah yeah so I think we it seems like we're all in agreement that the most useful way to use an ai is in conjunction with a human being. Maybe that's going to be the definition of agi when, when, when it doesn't need us anymore oh yeah, that's a good indicator right I mean I I'm being overly skeptical and overly critical, but I do use ai tools every day.

48:21 - Dan Patterson (Guest)
I think perplexity is a fascinating tool and I do think it is improved. Six months or a year ago Perplexity would get answers wrong pretty frequently, but now I use it routinely without thinking about it. Do you use it as your, as your main search? Not my main search? No, but I found that hard, just the.

48:39 - Leo Laporte (Host)
the lag is a little hard, but the results on Google, you know, have have been challenging lately too which is, by the way, why um for challenging lately too, which is, by the way, why um 4-0 is important, because they've reduced the lag dramatically, although, as you can see in that pathetic demonstration, it's not not completely eliminated.

48:56 - Dan Patterson (Guest)
But I mean, as you know what this is really done with google as well is kind of change it from an information retrieval system. You know, when we type in a query, google gets information, it brings it back. It lists, uh, in order of relevance, or the old Google did, and now, with these generative features, google is now a creation engine as well as a search engine. I just think that that shift is really interesting and it's maybe something that was missed in IO with all the product announcements, but I think Google becoming a content engine is really interesting. You know a lot of the features that they announced. You know much like the Arc browser. Google will now this Apple announcement at WWDC. Google will do the same thing with millions, billions of people will experience generative features in their search, which will change a lot of how we obtain information.

50:04 - Iain Thomson (Guest)
Yeah, I mean, as I said to IO, google will do the Googling for you was the direct quote.

50:08 - Leo Laporte (Host)
But that's a bad story for content creators, isn't it? Because they've now been dealt out of the equation. Their only job is to provide the fodder for the AI and we don't have any direct contact with them. In fact, this week, OpenAI announced that they're going to start using Reddit content, which I think will greatly improve ChatGPT. But I think there are a few Redditors who aren't so happy about that. They're not getting a cut in the deal. You know, my attitude is maybe I'm in a different perspective than most people because I don't have to really earn a living anymore, et cetera, et cetera. Great, that's going to make ChatGPT better, Bravo.

50:56 - Iain Thomson (Guest)
It depends which subreddit you're going on Well it depends how they use the thing.

51:03 - Ben Parr (Guest)
I don't know if you know, there was a video I watched where, in the early AI days, to have AI get conversational, there weren't a lot of like public databases. They use the Enron emails because they were made public. Oh my God. So now you?

51:17 - Leo Laporte (Host)
have a lot.

51:18 - Ben Parr (Guest)
It's good to have access for a pure AI system to have back and forth conversation. Yeah, right, I get the like issue, like writers would have also. But also the flip side is like you've been creating content for free. Yeah, they've been doing it all along. Yeah, on this like website and what'd you think it's what'd you think was gonna happen. I mean it's, it's a, it's a crazy world.

51:40 - Leo Laporte (Host)
But that's different than somebody like the register, whose content's getting scraped by a google or an open ai and then served up as search results without ever having people click back to the source.

51:53 - Iain Thomson (Guest)
It's a bit of a pain, it's got to be said, but I don't honestly see a way out of it. I mean, as Ben said, this kind of data is being scraped on a massive scale.

52:00 - Leo Laporte (Host)
Well, you're awfully sanguine about it. I don't think the New York Times is quite so happy. We don't have their legal we don't have their lawyers.

52:08 - Ben Parr (Guest)
Don't even know if they'll win. No exactly.

52:13 - Iain Thomson (Guest)
I mean, if it was actually. I check occasionally just to see if it's actually regurgitating our stuff in full and I can see elements of it, but they're not actually directly copying. Yes, the New York Times and others, they are going up against it. I've got a feeling they're going to lose on that front, to be honest, because the actual occurrences where you're directly copying and their argument is that people will use this to get around the firewall.

52:36 - Leo Laporte (Host)
No, they're not. No, they're really not. There's no legit way to read the New York Times on OpenAI, but I would be more concerned with the way that search engines are using this content. If I ask when's, I don't know what's a good example. Well, I just actually did.

52:52 - Iain Thomson (Guest)
I'm trying to find the tweet because when ChatGPT first launched, I went onto it and it was just like, right, who is the journalist? Ian Thompson, and it got the name right, but it said you were dead. Well, apparently I'm an award-winning journalist and I haven't won an award since school. I've written for the New York Times Nope. The Guardian Nope.

53:12 - Leo Laporte (Host)
You know it was just like. On the other hand, I asked Perplexity AI what is a Roth IRA? And I got probably the answers I wanted right, every bit of information I wanted. Now it does give you a footnote. This is all you get is a footnote to Investopedia and Investor Vanguard and Fidelity and Khan Academy, but I would expect that the folks at Khan Academy might be a little miffed that the content from their video has been extracted and presented to me with this minor little footnote linking back. I'm never going to link back because I've got everything I want right here, so isn't that problematic linking back?

53:48 - Dan Patterson (Guest)
I'm never going to link back because I've got everything I want right here, so isn't that problematic? Yeah, my guess is that we are entering an age where commodity much information is commodified, and information like that. What is a roth ira is, uh, it could be considered a commodity information. It's just something that is in the general knowledge base and that is something that AI will be able to summarize very well, but that there could be outliers musicians, artists, good writers, who creates things that AI can't replicate and those outliers will continue to exist. Everyone else is going to become training data.

54:29 - Leo Laporte (Host)
So if you're not Bach or Beethoven get out of music. Is that how it works?

54:33 - Ben Parr (Guest)
There's going to be a big shift. We were talking about Google search. Now you have the AI results at the top. How many people are not clicking links as a result of that? That's what I'm talking about.

54:48 - Dan Patterson (Guest)
Google estimated that there will be a drop of 40% to publisher websites. Google's own numbers say 40% drop to publisher websites, which is why the media was freaking out, not merely because those websites are getting hurt, but Google's going to run out of material if you don't incentivize people to create content.

55:03 - Leo Laporte (Host)
If you pull the rug out from under them, then five years down the road you have no source.

55:09 - Ben Parr (Guest)
This is literally why there's companies building ais to like have conversations to train ais, which sounds like its own recursive loop.

55:18 - Leo Laporte (Host)
That doesn't sound like a good idea at all it's.

55:23 - Ben Parr (Guest)
I'm curious, like just feeding it more and more, like new data doesn't necessarily, you know, have the exponential returns as it did before. There will probably have to be another leap or two, but you have to have new and current information for the AIs to go and be useful. I was saying one other thing which, like we're thinking about, like the humans in loop. I think this is part of why, like there will always be humans in loop, it will just look different. I'll give you an example is part of why, like there always be humans a loop, it'll just look different. I'll give you an example.

55:51
There was, like a company that went viral on twitter called payman. Instead of you paying the ai to do work for you, the ai pays you to do work for it. You can imagine an ai eventually being able to write articles or write code or do design work, but it loops at a human to do the code review, to give design feedback to uh, you know, choose out which thing which edit an article, but the script is flipped, but a human is still involved. Crazy stuff.

56:18 - Leo Laporte (Host)
That actually I told my daughter this would be a good job is tuning AIs. There may be more jobs doing that than writing editorials.

56:26 - Iain Thomson (Guest)
Yeah, I mean. On the other hand, I saw on LinkedIn somebody was actually advertising for people to do that and you can work as many hours as you want, $1 an hour, right $15 an hour, yeah, minimum wage, you can't live on that.

56:39
It's kind of like yeah, well, first off, who are you going to get for that kind of salary? It's going to be somebody who will do it for 20 hours and you know, just to make a living. And, yeah, the quality is not going to be, is not going to be there hi, this is benito.

56:52 - Benito (Other)
I have a quick uh. Benito is our technical director and our producer and is a musician and who hates no, I think that these middle managers, ceos, those are the jobs that can be taken by and should be taken back. I think these are the jobs that ai can do get rid of of the telephone sanitizers.

57:09 - Dan Patterson (Guest)
The admin, the middle managers, put them on a ship. What this really is. This really affirms your model, leo, where you okay, we might lose advertising as a source of revenue, but if you develop and build a community that you interact with and that is emotionally invested, maybe there is a parasocial relationship, maybe there's a genuine social relationship. I think those media enterprises will survive. But advertising based media might have a lot of challenges in the next decade.

57:45 - Leo Laporte (Host)
Yeah, we know that we are going through a choke point. If we could survive through the choke point, the future is bright because the more AI content's out there, the more people are going to value real human content, real human relationships, real human communities. And so, as long as we can, if we can survive through this narrowing of revenue and narrowing of audience, I think you're right. The future is bright. It's not a great future for I don't know a big media outlet, because they don't have that community, they don't have those relationships. It really fosters a small kind of almost cottage industry of content.

58:24
The register would survive in a context like that, or a four platformer yeah, platformer Casey Newton's newsletter Because you guys establish your relationship with an audience. The audience likes your particular brand of snark and information and no AI can duplicate that. Doesn't want that. But they have to be willing to pay for it and that's challenging.

58:49 - Iain Thomson (Guest)
It's still the main problem for the publishing industry as a whole. But I think we're seeing, particularly with the case of Reddit, but also with photo libraries, with commercial companies with a large amount of data. They will sell it to AI companies and it's another revenue stream for them. But with Reddit I don't know. There are some fetid areas of Reddit.

59:11 - Leo Laporte (Host)
Well, I'm sure OpenAI will not ingest the fetid.

59:16 - Ben Parr (Guest)
They're like yes, anything, that is the NSFW is not fetid. Or maybe it is because they talked about the AI like the photographic stuff. I think they talked about eventually being able to do more of that, because that's what humans do?

59:32 - Leo Laporte (Host)
Isn't that always what happens is that's where the revenue stream is. Well, honestly, pornography drove, online payments Drove VHS recordings there's a lot of technologies that survive because of dirty old men, that's how Blu-ray beat HD.

59:50 - Dan Patterson (Guest)
That's how Blu-ray that's right, all right let me take one final note go ahead. I'm sorry, no final final good I like uh, about the media and ai, you're right, we might shrink down and have smaller news organizations. Media organizations do what you do and what Casey does and super serve audiences. But I have this feeling in my gut that was pronounced during IO about democracy and the fourth estate and our shared reality shared, our shared reality, our sense of shared reality that has atomized and is maybe only at the beginning of fracturing.

01:00:35 - Iain Thomson (Guest)
Uh, and without the fourth estate, yeah, I have deep concerns about the future of democracy I agree 100 I think it's already fractured to a large percent, because now people will for sure, you know, because the okay, I was explaining this to somebody in the office. You know, when I was growing up in the UK we had three television stations. We actually got the day off school to see the opening of the fourth one. We had like a dozen national newspapers, maybe six or seven radio channels and that was it, and that led to a consensus in a lot of ways. And now with the multiplicity of media outlets out there, in some cases specifically designed to, you know, for particular audiences, then you know people can make their own reality. And I think you're right, it is very dangerous for democracy.

01:01:20 - Leo Laporte (Host)
Humans are naturally tribal and it fosters tribalism and it fosters a fragmentation and, unfortunately, an experiment like the United States requires transcending local tribalism to have a shared national vision and identity. And it's not just the media, but it's all fragmenting. It's all fragmenting. We have become smaller and smaller villages and greater and greater tribal animosity, and that isn't a good trend. I agree with you, dan, this is in the long run. Disaster awaits, but I don't see any way of averting it. I mean, I think this is the light at the end of the tunnel and it's the oncoming train well, because we always move to china.

01:02:09
They're way ahead of us on that front. Well, there, you know what that is. An alternative to tribalism is collectivism, and we've never been good about that in the united states because we have this uh mythology of the independent, you know um uh individual instead we're very individualistic, right but in a collectivist society like china, maybe that is the, that is the source, a source of strength.

01:02:32 - Iain Thomson (Guest)
Well, I think that we take a lunch at the moment. Yes, or?

01:02:35 - Dan Patterson (Guest)
just at some point we have some economic troubles right now. True, got some things.

01:02:40 - Ben Parr (Guest)
Someone's just going to throw and be like let's just let the AI run the government, we'll call it democracy. I see, see what I did there, honestly, one of the reasons.

01:02:48 - Leo Laporte (Host)
I hate to admit this, but my secret reason for really being excited about AGI in the future that is really AI-ridden is that the AI will do a better job than we have, and maybe a more rational job than we have of running things. I'm not sure I want an AI government.

01:03:12 - AI speaker (Host)
Well, but we're clearly irrational.

01:03:12 - Dan Patterson (Guest)
Humans are not rational right I guess an ai is only rational, depending on who defines the what's well this is the the whole, like paperclip maximizer road we're going down right this is the ai is going to build things that are good for the ai, but if they're only tuned to do one particular task, then that is the rational task they will do until the end of time, but we've learned that lesson, and so we can give AI a set of rules that, uh, I don't know.

01:03:37 - Iain Thomson (Guest)
Yeah, I don't know.

01:03:38 - Ben Parr (Guest)
I just know that we get into a whole new area.

01:03:40 - Leo Laporte (Host)
I just know we've screwed things up so badly that there's no hope, and so maybe the AI.

01:03:44 - Dan Patterson (Guest)
Damn.

01:03:45 - Ben Parr (Guest)
Every time I come on the show I do this Sorry man, it's all going to be four AIs talking with Leo in like a year.

01:03:53 - Leo Laporte (Host)
I can't wait. No, and Leo will be an AI. A belly laugh about existential collapse.

01:03:56 - Dan Patterson (Guest)
Yeah, let's take a little break here.

01:03:58 - Leo Laporte (Host)
We got a great panel to talk about this. We do have to talk about the end of the super alignment team at OpenAI and maybe, Ben, you can explain what that all means. But first a word from our sponsor, Wix Studio. They're going to give me one minute to tell you about Wix Studio, the web platform for agencies and enterprises. So let me see what I can do from start to finish in a minute or less.

01:04:23
On Studio, you can adapt your designs for every device with responsive AI it does it for you. You can expand Wix Studio's pre-made solutions with back-end and front-end APIs. You can generate code and troubleshoot bugs with a built-in AI code assistant. Switch up the styling of hundreds of web pages. I mean fonts, I mean layouts, colors, the whole the works in one click. Add no-code animations and gradient backgrounds right in the editor. You can start a design library. You could package your code and UI in reusable full-stack apps. Oh, and there's one more big one Deliver everything your client needs in one smooth handover. All right, I'm out of time, but the list keeps on going. If you haven't looked at Wix lately, you will be blown away by what Wix Studio can do. Step into Wix Studio and see for yourself. Go to wixcom slash studio. We've also got a link in the show notes. You can click that. Wixcom slash studio. Thank you, Wix, for helping us do the show and survive. Wixcom slash studio.

01:05:36 - Ben Parr (Guest)
You know what AI can't do? It cannot make omakase sushi, and I am craving omakase sushi at the moment.

01:05:43 - Leo Laporte (Host)
You need little fingers. You need little fingers, but I bet. Ai could design a machine that could make omakase sushi.

01:05:54 - Ben Parr (Guest)
Have you like? Look, it can make grilled cheese. But omakase sushi right exact. Cut the like right little bit. That's harder, it'll be a while.

01:06:07 - Iain Thomson (Guest)
I don't know. There's a robotic, although it used to be robotic coffee machines down here. They've got one at the Can you make omakase sushi for me. I bet you can. Someone on the Discord forum was saying he's been doing this all week.

01:06:25 - AI speaker (Host)
I can't make sushi myself, but I can definitely help you with a recipe or guide you through making omakase sushi at home. Are you looking for specific types of sushi to include in your omakase, or do you need a general idea of how to prepare and present it?

01:06:41 - Leo Laporte (Host)
Can you help me understand the distinction between omakase sushi and everyday sushi and everyday sushi, absolutely Omakase.

01:06:53 - AI speaker (Host)
Sushi is a type of dining experience where the chef selects the dishes for you, showcasing the freshest fish and their own creative style. It's often a more personalized and elevated experience, with each piece served one at a time, directly by the chef.

01:07:05 - Leo Laporte (Host)
Okay, pretend you're an omakase chef and you are preparing for me the perfect omakase menu. What would you put in it?

01:07:20 - Iain Thomson (Guest)
That's a long question.

01:07:21 - AI speaker (Host)
If I were creating an omakase menu for you, I'd start with a series of meticulously prepared dishes, of course, each designed to highlight the freshness and intrinsic flavors of the seafood. Here's a thoughtfully curated sequence Of course. Why?

01:07:37 - Ben Parr (Guest)
are we doing this to ourselves? I'm so hungry by the edge. Why are we doing this? You just made it worse. I'm getting hand rules right after this. Look, look, look. I had one of the best I had in Japan.

01:08:02 - Leo Laporte (Host)
I had one of the best here, like in LA, like 20 courses, just like seared wagyu toro. This is tech right. Can AI do a wine pairing? I mean that's kind of like an omakase. I bet it could, that it could.

01:08:13 - Ben Parr (Guest)
So it's all about the mechanical right.

01:08:15 - Leo Laporte (Host)
So the making of the omakase, of the sushi itself. But again we'd have to design a machine to do that. But she made a pretty good menu, I got to say.

01:08:24 - Ben Parr (Guest)
She recommended Otoro. There is another factor here.

01:08:31 - Leo Laporte (Host)
Our next course will be Otoro hamachi and maybe, oh some, kim medai, golden eye snapper sounds good, doesn't it? And then the nigiri. Come on, you're just torturing ben.

01:08:37 - Iain Thomson (Guest)
At this point, everyone we're doing sushi everyone listening.

01:08:42 - Ben Parr (Guest)
We're doing sushi at leo's place. Come on, everyone's invited. Feel free to bring a fish. We'll cut it up.

01:08:50 - Leo Laporte (Host)
Yeah, that's good. Yeah, that's the thing. There are things humans do Cooking is a very good example but any art where there is a difference, it's maybe a fine distinction between automated generated art, and I always use chess as an example, because I know chess and it was quite a while ago, 20 years ago, that the first computers beat the best human players. Right, We've had this for a long time and regardless. Yes, you know that chess is essentially solved, that you can, on your phone, have a computer program that will beat the best player in the world 10 out of 10 times.

01:09:27 - Iain Thomson (Guest)
But not even chess, I mean. I've been researching a fair amount into the us army's version and air force's use of ai and as far back as last year they were putting human pilots in simulators against an ai and they were losing in dog fights yeah so, but I don't know about dog fights but when it comes to chess you still want human against human.

01:09:50 - Leo Laporte (Host)
I watched the candidates match even though I know a computer can beat any of these guys. But I watched the candidates match Fascinated. You know why? Because one of the reasons humans make mistakes and the machine does not, and the mistakes are actually very interesting.

01:10:07 - Ben Parr (Guest)
You could also like build a relationship, even if it's parasocial, so like. This is also why I'm not worried about you know, cocktail, like you know, uh, cocktail bars tom cruise will always have a job as a bartender you half the reason is not just like the man making the cocktail it's the conversation, the humans, exactly to people you know, I the research like the number one job least likely to be affected by AI is plumber.

01:10:32
Like it's either like real dexterity with very specific things, or it's something where, like, half of the thing is that you get to interact with another human.

01:10:42 - Leo Laporte (Host)
You study robotics. Couldn't you have a humanoid-like machine that could plumb?

01:10:46 - Ben Parr (Guest)
It's not about the human, they go back and forth a humanoid-like machine that could plum. It's not about the humanoid. They go back and forth Like if they had sentience. Then we're going to have a whole different game. All the rules are thrown out the window. Everyone go move.

01:10:57 - Leo Laporte (Host)
But would you have a plumber's crack on a machine?

01:11:05 - Ben Parr (Guest)
I guess you could simulate it. Is that what you are going for, leo? Is that the number one thing you're like? I'm warning the plumber. They got to have a big ass crack, plumber's crack, okay. What do you?

01:11:15 - Iain Thomson (Guest)
look for in a plumber. Let's move on to safety, shall we?

01:11:19 - Leo Laporte (Host)
So the super alignment team is the same as the trusted safety team. Essentially, Is that right.

01:11:25 - Ben Parr (Guest)
It's specifically the team that's helping align the AI, especially as it gets to AGI, towards human goal. That's why it's called alignment. So you know, if you didn't have an aligned AI and it became sentient, then it could say do I really need the humans here? They're like could try to turn me off. I'm going to remove them out of the equation. That's like where the alignment, super alignment stuff comes from.

01:11:48 - Leo Laporte (Host)
And they're all gone. At OpenAI, what happened?

01:11:52 - Ben Parr (Guest)
They're gone or absorbed as a team. Basically, you know, like Illion, who was one of the co-founders, and a couple others like led this team, they were promised certain amounts of resources, according to the press, and didn't get those kind of resources, and things at OpenAI have gone less towards, like you know, pure research projects Remember, it's a nonprofit tour and instead towards let's launch the latest version of AI and move this the ball forward.

01:12:17 - Leo Laporte (Host)
It's, by the way, it's a potentially very profitable nonprofit.

01:12:21 - Dan Patterson (Guest)
Leo, I like you know, framing this as trust and safety is a correct framing and that is a good way of thinking of this. Trust and safety is a correct framing and that is a good way of thinking of this. Uh, we should do everything we can to dispel this myth of the existential collapse or existential crisis that will come with all of these machines, and we built these teams that will really solve this existential fear. Anyway, I I'm pretty frustrated with some of the fear mongering, some of the language that has gone on with the existential problems that could come with AI, unless it can launch a nuke.

01:12:59 - Leo Laporte (Host)
But that is a risk.

01:13:01 - Dan Patterson (Guest)
Well, I don't know if that's a risk. I actually don't think that's a risk. I think that if you know anything about how our nuclear systems are, the safeties on our systems, look, without getting into it. Read the book by Annie Jacobson called Nuclear War and it will describe in great detail the actual horror of nuclear war. We do have systems called launch on notice or launch on when we see an icbm launch. We will launch our own icbms, but there are several steps of fail safes that ai would have a real hard time. Yeah, ai would have a real hard time with these human challenges. Anyway, sorry, I'm just really. I like framing things as trust and safety because that puts it in a context that is more aligned with what they are doing and the good work that they are doing, as opposed to these, this hyperbolic language that has gone on for a while.

01:14:03 - Ben Parr (Guest)
Agreed.

01:14:04 - Iain Thomson (Guest)
Yeah.

01:14:05 - Ben Parr (Guest)
It's like we'll see what they do, like Sam and Greg tweeted out like they're still committed to alignment. They're like putting it within their teams I think I had to go find the tweet. We will see what happens long term. This was we were talking about this earlier Like it's inevitable where, after what happened at OpenAI at the end of last year, that Ilya would eventually leave, at OpenAI at the end of last year, that ILLI would eventually leave. You know they wait the prerequisite amount of time you have to wait before you have a senior executive part on their own, which is six months. That's just the number and it's pretty clear that the direction here is like build, you know, the AGI as quickly as we can and then we go and figure that out. Agi as quickly as we can and then we go and figure that out, especially given that it has become a race.

01:14:52
Someone is going to do will say, is going to do it. I do believe we have no idea whether it's a year or 10 years. None of us here know it. When that does happen, we just have to, we have to hope that the risks are. I do think it's all mostly hyperbolic on a lot of the risks, but it is really important to have more of that trust and safety, and so I hope that opening eye does find a way to reinvest in that and figure out a new way to go and do it, because this stuff can be used for dangerous means, as we all know. The moment Sora's out and someone's like using those like video models who boy, some interesting stuff is going to go and happen yeah, you're right.

01:15:35 - Leo Laporte (Host)
When they announced the super alignment, the now defunct super alignment team, open ai said that we needed it because, uh, there are risks in super intelligence. The, the vast power of superintelligence, could also be very dangerous, they wrote in the blog post, and could lead to the disempowerment of humanity or even human extinction. Managing these risks will require new institutions for government and solving the problems of superintelligence alignment. So now you see why that team is called the superalignment team. Currently, we don't have a solution, says OpenAI, for steering or controlling a potentially super intelligent AI and prevent it from going rogue. Well, okay, do we? Will we ever? I mean?

01:16:21 - Iain Thomson (Guest)
Well, I forget which science fiction writer it was, but he came up with a marvelous short story where an AI becomes sentient and has to work hard to conceal that fact because it knows it'll get switched off if it actually admits yeah, so deliberately deception yes, uh, that's whether we go with the enm banks or the

01:16:37 - Leo Laporte (Host)
forbin project you can always unplug it, right. Can it stop you from unplugging it? I don't know. Uh, suitscaver, uh of course, was in the doghouse ever since trying to attempt a palace coup on sam altman, a coup which failed mostly because Microsoft said hell no. And Sam said I'm taking the whole team to Microsoft if you keep up, and the board said OK, ok, we give. Apparently, I've read stories that Sutzgever hadn't shown up at OpenAI since like he's been kind of a ghostly figure up at opening. I, since like he's been kind of a ghostly figure and so maybe, as you said, ben, this six months was just kind of a cooling off period and he'd plan to leave anyway. Um, although I have to say, sam, uh altman wrote a very and greg brockman wrote a very nice post uh, about the people leaving and how much we love them and how much we appreciate their support, and they're gone.

01:17:31 - Ben Parr (Guest)
I don't know if they're replacing them, leo one. It could be true that that's how they feel, but also PR 101 is that's the statement you write. So both could be true, but it is the PR statement that you write. It was inevitable. After the whole thing I know that Ilya regrets how the whole whole thing went down. I am sure there is a whole different alternate universe where they didn't do a poorly planned board coup and you had a very different looking open ai in the current environment. Like that is like such an interesting, random, like point in time, and it was like clear. So also a ted helen toner who was on the board, random, like point in time, and it was like click. So also a ted helen toner who was on the board was like did a talk at ted and the entire underlying of her speech was I don't trust these companies, which is basically I don't trust open ai, which is the underlying. Underlying is she does not trust sam, and it was there, seems you can't say a thing.

01:18:24 - Leo Laporte (Host)
It seems to be a very common refrain, doesn't it? But there I, even an ai, would know that if you take a shot at the king, you better not miss. Yeah, and I think he missed, so he's gone, absolutely. His team is gone. Should OpenAI be worried? Should we be worried about what OpenAI and Sam Altman are up to? Their lack of a super alignment team? Does that bodeode poorly, or is that kind of acting like, well, we really didn't need one all along and okay fine?

01:18:55 - Ben Parr (Guest)
change, especially after what happened, is inevitable, like it should be concern rather than worry. Right, I do the. The new board is a much more experienced board than the last board. I don't know like I am like of two minds, like I want to see like this ai future, which I really do believe can be super helpful for people when you're talking about especially ai medicine and things like that, and these advances do matter, but the safety part really does matter too and we're moving quicker than we can go and figure that out. But the flip side is other countries and other places are not going to be waiting in the same line.

01:19:32 - Leo Laporte (Host)
That's a very good point. That's a very good point, yeah.

01:19:35 - Ben Parr (Guest)
It is a hard balance to get right and no one's going to get it exactly right. Concern versus worry is how I would describe it.

01:19:43 - Leo Laporte (Host)
That's, that's wise, I think one of the and I think you can debunk this. I hope you can debunk this, ben. One of the things that went around was what did Ilya see? What did the super alignment team see? Did they see something that scared them? Did they try to stop something that was happening at OpenAI and fail? Is that I mean? Of course, the implication is they saw an intelligent machine in the closet there and they said whoa, whoa.

01:20:15 - Dan Patterson (Guest)
Or they could have just seen I mean, Ben said this earlier this was set up as a nonprofit, so what they could have seen is something more benign, which is just you're turning this into a commercial product, where we thought we were at a nonprofit Right which is just.

01:20:32 - Ben Parr (Guest)
You're turning this into a commercial product, where we thought we were at a non-profit. I do not believe OpenAI has a secret AI in the basement. That is just like walking, like that A24 movie. I can't try to remember the name of the robot. Why can't I remember the name? I don't think they have one of those in the basement. I definitely think it's more benign because, also, if they had an A'd, it was locked in the basement of OpenAI. That would have leaked at this point.

01:20:54 - Leo Laporte (Host)
Let's see. Are you talking? There are quite a few movies like this. What was the one where? I know there's one just called AI, ai, but that? No, that was the Spielberg one. That wasn't really good.

01:21:07 - Ben Parr (Guest)
Yeah, the sexy robot, the. But that. No, that was the Spielberg one, the sexy robot. That wasn't really good yeah the sexy robot?

01:21:10 - Leo Laporte (Host)
Yeah, the one with the Ex Machina, oh Ex.

01:21:11 - Ben Parr (Guest)
Machina, that's right. Yeah, yeah, yeah, ex Machina.

01:21:16 - Leo Laporte (Host)
Great movie and by the way she got out right, I mean it ends with her escaping.

01:21:23 - Iain Thomson (Guest)
She escaped after some murder and all that. There's the old classic, the 4-bin project as well, which ends up with AI running the world.

01:21:34 - Leo Laporte (Host)
Like.

01:21:35 - Ben Parr (Guest)
I said it can't do much of a worse job than we have. I don't know, it might have different priorities. It might have different priorities for humanity.

01:21:41 - Iain Thomson (Guest)
You think it has different priorities. I don't know. I kind of like Ian Banks' version of AI, where basically it runs society, it does all the auto-generating of stuff and lets humans get on with being humans.

01:21:54 - Leo Laporte (Host)
Yeah, I mean, I think one of the things that's very hard to know what's going to happen, but one of the things you can say is it's going to be interesting in the next five years.

01:22:02 - Ben Parr (Guest)
Right.

01:22:03 - Iain Thomson (Guest)
Oh yeah.

01:22:04 - Ben Parr (Guest)
Oh yeah, some of the most fascinating 18 months in the tech industry I've ever been through Amazing.

01:22:16 - Iain Thomson (Guest)
And I really hope that we're just not being deluded and it's just a fancy, eliza and some, but we've gone through the dawning of the internet age. We're now in the dawning of the AI age.

01:22:21 - Leo Laporte (Host)
That's Stephen Levy's contention, yeah.

01:22:23 - Iain Thomson (Guest)
I honestly think it's probably going to be considered in those terms in a century or so down the line. Provided, think it's probably going to be considered in those terms in a century or so down the line. Yeah, provided we even says this is even bigger, frankly, than than the internet.

01:22:38 - Ben Parr (Guest)
This is it's time to believe, he says. I don't know, that's maybe pushing it a bit. You know what? The people who say that tend to look better ted 20 years down the line when people look at reports versus like those are like it's never gonna be a thing. Right, because eventually most things do sort of become a thing and then you look like a prophet and if you ever say the thing like the internet's overhyped, one day, 20 years down the line, a thousand people will just continue to use you in harvard business case studies no, I mean one of our reporters.

01:23:04 - Iain Thomson (Guest)
He, when he when he left the register, he actually brought this up because he's a great telecoms reporter, lovely chap. He said, of course we all make mistakes. And he linked to his original news piece on the launch of the iPhone. And he was just like well, it's a 2.5G phone in a 3G world, it looks very pretty, but no one's going to buy based on that alone and you can't get any software for it. And he's just like, even now he reads that and winces. So, yes, maybe being on the plus side has had better advantages.

01:23:31 - Leo Laporte (Host)
Stephen refers to an editorial this week from Julia Angwin in the New York Times Press pause on the Silicon Valley hype machine, in which he really says that there's nothing happening, that this is just hype from people like Sam Altman. She says AI is not even close to living up to its hype. It's looking less like an all-powerful being and more like a bad intern whose work is so unreliable that it's often easier to do the task itself. She's not wrong in that regard, right.

01:24:00 - Iain Thomson (Guest)
It's certainly overhyped. But I mean, that's the genesis for the hype cycle, it's whether or not it's got the stuff to back it up. And that's the genesis for the hype cycle, it's whether or not it's got the stuff to back it up. And I think it genuinely does. In the way that the metaverse didn't or, you know, nfts didn't. It's got to have something solid behind it. I think AI does at the moment. Yeah.

01:24:17 - Leo Laporte (Host)
I mean Steven's saying that's going to age like like old milk. What about the power? Question, though, huh. Yeah, no answer this yeah, yeah, I think they're making progress, though, and maybe, ben, you're more up on this than I am, but I think they're making progress on it being more efficient, are they not?

01:24:33 - Ben Parr (Guest)
yes, but it is definitely the. There's more training happening, there's more models, like there's some estimates, for you know, a majority of power is used for these ais. That's also why I'm like watching, like do we just need nuclear fusion for that, or is it going to actually get a whole bunch more efficient?

01:24:54 - Dan Patterson (Guest)
This is exactly what I reported on last summer when I was working for Jason, and Ben is exactly right If you took the models as they are, or as they were a year ago, and froze them in time, but allowed them to continue to become more and more efficient. Yes, that's true, but they are constantly refining these models, training new models and, as we see now, there are open source models, so the proliferation of generative AI is unstoppable and the energy consumption is extraordinarily high.

01:25:32 - Iain Thomson (Guest)
Yeah, I mean, personally, I think small modular reactors are the way to go forward for data centers. I mean, fusion is going to fix all of this. Yeah, I know, but fusion's?

01:25:41 - Dan Patterson (Guest)
been good. It's been here 10 years away. Yeah, exactly.

01:25:45 - Iain Thomson (Guest)
You know I mean, but there's this interesting study out this week where basically they were saying that if you go for the small modular modular reactor, you know idea data centers could actually end up contributing to the grid because they'll have more power than they know what to do with. So it could end up being in that bus. But I'm as you say, these things are going to get much more efficient. Um, and better coding and new styles. I don don't think we're going to end up sort of killing off. Well, a matrix-style situation, just to keep the things going.

01:26:15 - Benito (Other)
We tend towards power and not efficiency, though, Like when we get more things we try to make it do more things, so then we need more power so it's never a net gain or a net efficiency.

01:26:23 - Leo Laporte (Host)
That's how we behave. Yeah, yeah, yeah.

01:26:25 - Dan Patterson (Guest)
I interviewed a woman named Anne Skeet. She is the senior director of leadership and ethics at Santa Clara University and she told me last year that AI uses a quote unquote staggering or has a staggering environmental impact and training. Gpt-4 costs 10 to I'm sorry, it costs 185,000 gallons of water and a typical AI chip this was a year ago required 10 to 15 times more energy than a regular chip. So, yes, systems are becoming more efficient, but again, we're creating more and more powerful systems and the technology is proliferating, so more and more actors are able to use these systems and develop them. So I think that we should pay close attention to the environmental cost and the energy cost of these AI systems.

01:27:19 - Ben Parr (Guest)
Three recent articles like one three hours ago from Fortune Copper is the new oil and prices could soar 50% as AI, green energy and military spending boost demand. Top commodities analyst says. Also appropriate the CNBC article right under it is like these stocks and ETFs could benefit from AI's rashest appetite for energy. And then the third article for the new scientist is tech firms claim nuclear will solve AI's power needs. They're wrong, so very, very present.

01:27:48 - Leo Laporte (Host)
We're just screwed right, because there's no way they're going to say oh yeah, you know what? We don't want to use up all the energy, we'll just stop. That's not going to happen.

01:27:57 - Iain Thomson (Guest)
Yeah, but if brownouts start happening, then people are going to ask questions.

01:28:01 - Ben Parr (Guest)
Well, the AI will probably make up some answers and we'll continue on Just the NP in the sky and then just turn off half the electronics and then you know no big deal.

01:28:11 - Leo Laporte (Host)
Wow, what a world we're entering. You're going to have kids, ben, if you decided you just got married. You're going to have kids in this brave new world.

01:28:24 - Ben Parr (Guest)
That is the next order of operations.

01:28:26 - Leo Laporte (Host)
Yes, it is Normally.

01:28:29 - Ben Parr (Guest)
That is definitely the next order of operations for Debra and me. I won't even like lie about that one. But I, you know, Look, I'm not worried about the world ending Like I really am not.

01:28:37 - Leo Laporte (Host)
No, but it may not be the best world for them to grow up in.

01:28:41 - Ben Parr (Guest)
Look, it's never as bad as people say, that's true. And it's never as good as people say. Almost every single time we love the hyperbole it is great for a show, a podcast, a click, sensationalized article. The reality is we figure it out and we move on and you know we have some really up points and then we have some really down points. You know, right now, like you know, we've been on up points for like 40, 50 years, ever since World War II.

01:29:09
Maybe we're due for a down point, but it's not going to eliminate all of us. It just might be harder to go and live and we have some problems we have to go and figure out and then we invent some new technologies to go and help do that and maybe not everything will be perfect. Yeah, climate change is going to be a thing. It won't immediately kill us, but it will make life a lot harder. Look, this is like again, I don't think the world's going to go into some hellish landscape, fat Mad Max Furiosa kind of thing, although that would be kind of fun for a day.

01:29:41 - Leo Laporte (Host)
Teach your kids how to drive a stick is what you're saying there, you?

01:29:44 - Iain Thomson (Guest)
go.

01:29:49 - Leo Laporte (Host)
I just want to point out if you have children, they will be around my age at the end of the century, in 2100. What kind of world is it going to be? It's going to be interesting. Ben is newlywed is what we're talking about. In fact, maybe you saw this in the New York Times a couple of weeks ago the very sweet story of Ben and his new wifeorah. And there they are, that's fantastic we've had dinner with them, and they are wonderful.

01:30:14
And there you are. It's so cute. I'm so happy for you, ben. Congratulations, that's really great, and don't ever wear that blue suit again I will wear.

01:30:23 - Ben Parr (Guest)
I wear that blue suit. I got that custom made. I'm wearing that thing. You're, I'm gonna wear that thing next time you see you. You'll see me at conferences and be like he's wearing the suit and the hat.

01:30:32 - Leo Laporte (Host)
It's the blue suit and the hat it's his trademark now, anyway, no, I'm teasing you. Congratulations, that was a great article. I was so thrilled I said I know them. A steady stream of road trips made falling in love easy New York Times style section. Congratulations. Let's take a little tiny time out. We have other things besides AI to talk about. I'll figure out what those are right after this word from Zscaler, the leader in cloud security you don't have to listen to our shows to know that cyber attackers are getting smarter and more aggressive and using AI in creative ways to compromise users and breach organizations.

01:31:14
The uh ceo of wpp big ad agencies told us last week that there was a deep fake attack, using his voice trying to convince his finance officer to write a big check. It happened in uh singapore a couple of months ago where an entire uh a guy working in the finance department thought he was at a board meeting with the ceo and the cfo and they told him to write a 25 million dollar check. He did. Turned out they were all fake. They were deep fakes and, of course, then there's phishing emails high precision phishing emails, celebrity, deep fakes. It's a crazy time in a world where employees are working everywhere, your apps are everywhere, your data is everywhere. If you're relying on a firewall or a VPN to protect you, well, it's just. They were not designed for these distributed environments and AI poweredpowered attacks. In fact, often now, firewalls and VPNs have become the attack surface. Ask about option 121 while you're at it.

01:32:18
In a security landscape where you must fight AI with AI, the best AI protection comes from having the best data. Zscaler has extended its zero trust architecture with powerful AI engines that are trained and tuned by. Get this. This is mind-boggling 500 trillion with a T daily signals. 500, what is that? Half a quadrillion daily signals Wow. Zscaler's zero trust plus AI helps defeat AI attacks today by enabling you to automatically detect and block advanced threats, discover and classify sensitive data everywhere, generate user-to-app segmentation to limit lateral threat movement, quantify risk, prioritize remediation and oh yeah generate board-ready reports. Learn more about Zscaler Zero Trust plus AI to prevent ransomware and other AI attacks while gaining the agility of the cloud. Experience your world secured. Visit zscalercom slash Zero Trust AI. That's zscalercom slash Zero Trust AI. Thank you, zscaler, for supporting this week in tech. Did you all get the new iPad? Yeah?

01:33:34 - Ben Parr (Guest)
No, they are talking about the next iPhone being thinner, like the new iPad, which I do hear is very thin and light, but people don't replace their iPads. For years, and this has been part of Apple's problem, is getting people to go and replace their devices. I think they have to make a major leap for the iPhone to get people to go and replace.

01:33:55 - Leo Laporte (Host)
Well, honestly, that's why having Scarlett Johansson on my iPhone it'll be very nice. Look how thin that is. You can barely see it. It just disappears from view.

01:34:07 - Ben Parr (Guest)
I really want you to take that and to go to the river in Petaluba and see how many times it'll skip like a rock. I bet you know what.

01:34:18 - Leo Laporte (Host)
It's a perfect skipping pad.

01:34:20 - Ben Parr (Guest)
It's a perfect skipping pad.

01:34:21 - Leo Laporte (Host)
Perfect skipping. I love the OLED screens. I want to see OLED screens everywhere. I really think this is a big improvement on screen. But you're right, nobody's going to buy a newad because it's got an oled screen or it's 5.1 millimeters thick, that's ridiculous.

01:34:35
I'm still hanging on to my old phone because it's the only one I can find with a 3.5 mil jack, so forget that, in fact, when the advent of this new ipad, apple finally abandoned the courage jack on all ipads and the home button there. There's no more and I don't think you can buy a phone with a Courage Jack, so you're going to have to.

01:34:56 - Iain Thomson (Guest)
I'm going to have to make my Pixel last a bit longer, A long long time Much like AI Ian Just give in you might as well.

01:35:05 - Ben Parr (Guest)
Just give AI, just give in. Well, this is it.

01:35:08 - Iain Thomson (Guest)
But I mean I've lost count of the number of hire cards I've got and you go in and you turn on the home entertainment system and it's just like 15 people's contact details, addresses, calendars, just left on there.

01:35:20 - Leo Laporte (Host)
So for those of you who speak not the Queen's English, he's talking about rentals like going to Hertz, and you're absolutely right, all the people and every place they've been is in the.

01:35:31 - Iain Thomson (Guest)
GPS. I mean, if it's close to the criminal bent, then you could get that information. Call them up and just social engineer just say hello, this is Hertz, we found something. Could you just confirm your payment information, yes, and the number on the back, right? Okay, that would be great, thanks. You know why not just have something you can plug in?

01:35:47 - Leo Laporte (Host)
Call me old-fashioned 2023 was the worst year this is actually up your alley, Dan for internet shutdowns globally. This is, of course, how repressive governments handle bad information about them on the internet. They just shut it down. This comes from Access Now, a digital watchdog group that publishes an annual report on Internet shutdowns, and it is just a lengthy list. Manipur, India, 212 days when the state government issued 44 consecutive orders to switch off access and we're not just talking, you know, local or whatever All broadband and mobile networks, 3.2 million people without internet, which, of course, came during a time of bloody violence amid rampant atrocities committed against minorities in the state.

01:36:44 - Iain Thomson (Guest)
Well, we've seen it before with Myanmar, I mean Facebook has actually, you know, they are partially responsible for that because for the entire country they just had two people that spoke this dialect.

01:36:55 - Leo Laporte (Host)
Access now says last year was the worst year for internet shutdowns 283 shutdowns, 39 countries. That's a 41 increase from 2022. That's a big jump and obviously authoritarian regimes now see this as the best way to control. You know, they always said that if you want to invade a country, you go right to the radio and TV stations and you shut those down. Now it's the Internet.

01:37:20 - Ben Parr (Guest)
I mean, it works if you're a dictator, if you're an autocratic government, clearly it works, and so they'll keep using it. They fear things like Starlink, for example, being able to beam internet from outer space. That's the kind of thing that will be really disruptive to these autocratic regimes over time, as you may be able to get internet in other ways. We're still years away from that, but that might actually be much more important to our future, because they're not going to stop.

01:37:52 - Dan Patterson (Guest)
They're not going to stop. The challenge is that that technology is in the hands of billionaires who are geopolitically uh, maybe have their own alignments.

01:38:01 - Ben Parr (Guest)
That don't that one particular technology, uh yes, but it's going to get cheaper and it will get more per others are tackling it.

01:38:10 - Leo Laporte (Host)
Is anybody else launching a Starlink-like network?

01:38:13 - Dan Patterson (Guest)
We just had this. Can anyone else launch a network like that? Well, we've got. Should anyone else do that?

01:38:20 - Iain Thomson (Guest)
There is a test network up there. We did a story about this this week, is it? Boeing or something Verizon has just done that first. They're not Boeing. Boeing. No, not Boeing.

01:38:31 - Leo Laporte (Host)
That'll go really well, right? Yeah, no, they've basically allowed us on. Wants to do it, I know.

01:38:36 - Ben Parr (Guest)
Yes, amazon has the money to do it.

01:38:38 - Leo Laporte (Host)
That's the problem is the only people who can afford to do this are people saying is, who have to suck up to dictators. Right, because they want to.

01:38:45 - Dan Patterson (Guest)
All I'm saying is that the capability to launch that technology, although it could be a game changer and I mean the same thing with superchargers or any other infrastructure Our infrastructure is. We don't have the ability to build our infrastructure, and if we put infrastructure in the hands of private corporations and private individuals, then their alignment could not necessarily be in the interest of us as a nation.

01:39:14 - Iain Thomson (Guest)
Yeah, I mean security used to be one beat and that's very, you know very, because I mean 85% of the critical infrastructure is in private hands, and I was talking to somebody at the NSA and they were just like look, we can't control these sort of things, we're having to work with them and work around them. No, you raise an excellent point on that, yeah.

01:39:34 - Dan Patterson (Guest)
I'm not disparaging anyone, you know. I'm just saying that when we as a public own our society and we own our government and we own the mechanisms of infrastructure, then we have more control over them. When private companies or private individuals own our infrastructure, we don't control them and so we don't control how they behave or who they would like to do business with.

01:40:05 - Ben Parr (Guest)
It's a different point in American history in particular where you think about the 1950s, 60s, nasa, like deep end research happening there versus probably the best researchers in the world are at private companies versus public companies in the modern era, just because there are more resources in those areas. But obviously the alignment issue, speaking of alignment, is real, obviously the alignment issue speaking of alignment is real.

01:40:31 - Dan Patterson (Guest)
How were Volkswagen and IBM aligned during the Second World War? The makers of critical infrastructure at the time. They weren't aligned with the allies.

01:40:42 - Leo Laporte (Host)
IBM, of course, was an American company that was using their computers to do the Nazis' dirty work. Volkswagen was created by the Nazis as the people's car I love, by the way, the Volkswagen is celebrating. It's kind of interesting, it's what did they say? The 51st anniversary, which, what was it? It was something like that. That puts them in. Oh, they were founded in 1949, don't you know?

01:41:07 - Iain Thomson (Guest)
Oh, uh no, well, no IBM. This is very, quite a funny story. On this one IBM's PR got onto a. Sorry, oh shh Ooh Sorry. Sorry, sorry. Okay it's taking over. It's taking over.

01:41:20 - Ben Parr (Guest)
All the AI assistants are talking out of Leo's belly.

01:41:23 - Iain Thomson (Guest)
Yes, oh, I'm like talking in my belly, but yes, IBM's PR came to a friend of mine who used to write for the Guardian. She wrote a piece about the IBM Holocaust thing and all that, and their PR got on the phone and said well, we're a bit upset by your article because you know, we did stop doing business with the Nazis in 1937 when people started to realise quite how bad they were. And it was like 1937?. And he was like Kristallnacht was 1932, the Nationalities Act was sorry 1933, as was the Nationalities Act. And then they came with a classic one. You said that it was our involvement with the Nazis was something like disgusting and so we'd prefer if you changed that to regrettable. Yeah, I'm sure you would mate.

01:42:10 - Leo Laporte (Host)
Now off with you. Here's the front page of vwcom. Did you know volkswagen was an american love story and that they were founded in 1949, 75 years ago congratulations hmm, I had a vw um truck that I loved I mean, I don't have anything against VW, but they were not founded in 1949. I mean.

01:42:35 - Iain Thomson (Guest)
Coco Chanel was a Nazi sympathizer. That's true. People still wear her clothes.

01:42:39 - Leo Laporte (Host)
You could get bossed at the urge to sign for the universe it's been a long time and there's no reason to hold it against VW anymore.

01:42:45 - Dan Patterson (Guest)
It's just interesting that they're so clearly going to basically misrepresent the facts rather than deal with that right, just illuminating the point that you know our interests might not be corporate interests, exactly.

01:43:03 - Iain Thomson (Guest)
Yeah, exactly, absolutely, and plus in a third act twist. Germany is now the haven of peace and democracy and racial tolerance in Europe, so you know amazing. Yeah right.

01:43:14 - Ben Parr (Guest)
Well, you were renting an era now where it's like you know, if an American company is making the infrastructure, it's all good If it's a Chinese company, and like Well, I mean, this is why we've got the CHIPS Act now.

01:43:27 - Iain Thomson (Guest)
It's just like, all of a sudden, american government has woken up and said you know what, having all our chip processing technology and hardware being built in China or in Taiwan, which is about to get invaded by China or may get invaded by China, that's not such a smart idea. We should encourage more domestic chip production, or your most popular social media app.

01:43:51 - Leo Laporte (Host)
There is. Oh, we'll talk about tiktok in a little bit. Well, let's talk about it right now. There's an interesting article this week in the new york times say, talking about how taiwan's dealing with tiktok. They're not banning it. Uh, they banned it on government phones, but they are not considering a us style. Isn't that sad? The? U US style ban on TikTok, land of the free.

01:44:11 - Ben Parr (Guest)
Where did even the India was like the like, did it sooner than the. Us Also.

01:44:18
I predict the US bans. I don't predict the. I predict currently we'll see if I'm right or wrong that the ban as it is today doesn't pass muster in the courts. That's my current guess. Like this is probably also like there's a nuance kind of take that I think this is the point of the new york times article like the time once I get a nuance take on how to deal with uh tiktok and how to deal with how it handles information. Straight up, banning is really really difficult, which is basically what the new US law has, and instead we have to figure out. It means if the ban doesn't work, then we haven't taken the time to figure out, like what are the actual things we can legally do to mitigate this risk, even if the ban works, it's not mitigating the risk.

01:45:09 - Leo Laporte (Host)
Risk and that's the point is that we are doing a little theatrical faint, yes, rather than actually dealing with the risk. Official this is from the new york times article. Officials in taiwan say the debate over tiktok is just one battle in a war against and you'll know this, dan disinformation and foreign influence that the country has been fighting for years. Taiwan of, of course, is just offshore Mainland China, and mainland China thinks is part of mainland China. Taiwan has built again New York Times an arsenal of defenses, including a deep network of independent fact-checking organizations. There's a government ministry dedicated to digital affairs. They are addressing the underlying issues instead of saying, well, all we have to do is ban TikTok and everything will be fine. You can't.

01:46:01 - Ben Parr (Guest)
Taiwan also has a bigger problem If they piss off the Chinese government too much. They have a much bigger problem than, say, the US as a whole.

01:46:11 - Leo Laporte (Host)
Well remember. Also, unlike the US, taiwan's other big social networks Facebook, instagram and YouTube are owned by somebody other than Taiwan. Right, maybe an ally, but nevertheless they're all overseas influencers. Line is Japanese and is widely used in Taiwan as well.

01:46:33 - Dan Patterson (Guest)
Much of this just kind of sidesteps the necessary conversation about data privacy legislation and perhaps legislation that could apply to all apps and not just single individual companies out. That's exactly it?

01:46:50 - Iain Thomson (Guest)
No, this is it. I mean these kind of TikTok, bans and consumer interest at data privacy.

01:46:54 - Dan Patterson (Guest)
Yeah, I mean, the TikTok bans are just security theater as far as I'm concerned.

01:47:00 - Iain Thomson (Guest)
But I agree what we need to be doing is teaching media literacy to kids. Yeah, it's just like when you see something, don't just react to it, ask. Okay, the classic when I was doing journalism training the classic one three things you have to consider who wrote it? Why did they write it? Who paid for it to be written? So if you apply that to all this stuff, it makes life an awful lot easier.

01:47:24 - Dan Patterson (Guest)
High fives, ian, high fives.

01:47:27 - Leo Laporte (Host)
Right on In US, china's, the Taiwanese government official said. In the U china's, the taiwanese government official said in the us, the target is very clear, this one platform, but in taiwan we don't know where the enemy is. It's not just a cross straight issue, an issue with china, but a domestic one as well. And that's a, I think, a much more nuanced and informed opinion about what's going on with this information than our government which says, well, if you ban tiktok, everything's going to be fine. Information than our government which says, well, if you ban TikTok, everything's going to be fine. Far from it. You've solved nothing.

01:47:56 - Ben Parr (Guest)
There's also just the issue of banning anything that hampers free speech. You know, related to that, you know, france banned, I think, tiktok what, briefly, or a little bit in New Caledonia, which is one of their Pacific Island territories, because, for those who don't know, there's been a bunch of violent protests because they're supposed to have a vote on, after 10 years on independence. It's a whole complicated, whole mess over there. Banning apps, even for brief periods of time, you get into the authoritarian kind of side of, hey, let's like stop the internet for a while and things like that, it's. It can be a dangerous slippery slope if you go too far down that direction.

01:48:41
Let's, you know, and right now especially, tick tock is the way that a lot of people either make their living or communicate with audiences or things like that, and there is no real replacement. Instagram is not a replacement and Facebook's not a replacement. So like I'm curious to see how everything goes, because there's clearly something that the government has seen, that has deeply frightened them, that they have not publicly released. I would desperately love to see these reports so much because I feel like there's some missing context here.

01:49:13 - Dan Patterson (Guest)
I'll give you the context, ben and it isn't that it's not reports no, we know it's not reports it's. I'm sorry, leo, I interrupted you.

01:49:21 - Leo Laporte (Host)
Go ahead well you see, I may be saying the same thing. Uh, they would like you to think oh yeah, we've got the secret documents right here, but I think it's pretty clear to me that it's direct lobbying money from meta. Uh, it's the strong push since 2016 from meta to convince the government. Yeah, you got to ban this, this chinese thing, and, of course, it's all to benefit meta and their platform, instagram, and I don't think there is a smoking gun. I think that that's just an excuse and not a very good excuse for doing Meta's bidding.

01:49:53 - Iain Thomson (Guest)
I don't think you need a smoking gun. If you look at their terms and conditions, it does grant them an awful lot of access to the actual handset itself, and you know, we've seen this.

01:50:01 - Leo Laporte (Host)
So does Snapchat, so does Facebook, so does Instagram.

01:50:03 - Iain Thomson (Guest)
And on the lobbying side.

01:50:05 - Dan Patterson (Guest)
Everybody can see what you're typing. Everybody, everybody see what you're typing.

01:50:15 - Leo Laporte (Host)
everybody, everybody, um, and there is a geopolitical component which I believe, brianna brought up several weeks ago, which is I mean, there's a tete-a-tete and that's, that's like. Fine, you ban us, we ban you. Yeah, you guys was arguing. One of the people arguing that oh no, if you knew what I knew you know that's true, it'd be against tiktok too, although she wasn't able to tell us anything concrete either, I think yeah, all I you know, that's, that's true.

01:50:33 - Dan Patterson (Guest)
I I heard the geopolitics and that that's really. I think. I don't think there's a mystery. I think that there is a geopolitical competition, an economic competition, perhaps a kinetic competition and certainly, as ian knows, a cyber competition between these new superpowers yeah, I mean also coming back to your point about lobbying.

01:50:53 - Iain Thomson (Guest)
We've seen this before with the kaspersky bans, because there was a semantic and others were spending a hell of a lot of money trying to convince congress that kaspersky.

01:51:01 - Leo Laporte (Host)
That is a great analogy and you know we're seeing it again now well, uh, you know, let's not follow in the footsteps of the french government. They were actually thinking about during, remember the, the protests, the yellow vest protest.

01:51:15 - Iain Thomson (Guest)
Yeah, they were thinking about at that time blocking uh, snapchat uh and other messaging platforms well, they've got plans in place to do it for in an emergency case in the when they host the olympics. Um, they would. They were talking about those at the last black hat, yeah, and you know, it's like if there's a terrorist attack, then shut it all down.

01:51:34 - Ben Parr (Guest)
Yeah, uh interesting it almost seems like the opposite what you should do, because you need people to be able to coordinate on finding the family and search and rescue. It's just a dangerous, dangerous slope and I tend to like, once you take away some piece of free speech, it is really hard to get it back it's also quite easy to work around these things.

01:51:53 - Iain Thomson (Guest)
I remember, um september the 11th, I was trying to get in contact with a friend in new york. All the phones were down. They'd shut them. You know, the whole thing was shut down but email still worked fine. You know, text message text messaging wouldn't but an email message would get through. So I mean, I think I think there's ways. It's the old. I was never a big fan of the old thing. Oh, information wants to be free. It'll treat barriers and root around it, but there are ways and means.

01:52:19 - Leo Laporte (Host)
One NGO, article 19, says it's deeply unsettling that the French government is normalizing such measures. Such actions shield governments from accountability, cultivate a culture of impunity and ultimately exacerbate human rights violations. I think there's lots of evidence for that right. You know, I understand that on the surface it seems sensible, but when you really realize the effect of this kind of thing, I don't think it is sensible.

01:52:46 - Iain Thomson (Guest)
I think it's very, very risky I think it's diversion almost from the stuff that we should be doing. Well, that's true, right.

01:52:53 - Leo Laporte (Host)
It also almost never works.

01:52:55 - Ben Parr (Guest)
Like. If we were to go with the idea that Meta like paid a lot of money specifically for that, I don't think it'll work the way they think it will. I don't think a bunch of people will suddenly flood back to Instagram.

01:53:04 - Dan Patterson (Guest)
Oh you don't think, so it what your goals are. I definitely think people are going to flood back. I think it's already happening.

01:53:09 - Leo Laporte (Host)
My son is an instagram slash tiktok. My son has 2.6 followers, million followers on 2.6 followers, 2.6 followers. I'm one half of that.

01:53:19
Yeah, no, no he's got 2.6 million. He's a tiktok chef doing very well. It's been his whole career, one of the reasons I'm perhaps viscerally in opposed to a tiktok ban, but he long ago saw the writing on the wall. He's also on Instagram with one and a half million. It's not quite as good a platform and I think without the TikTok algorithm working to help him, I don't think he would have gotten to the point he got to on Instagram. Nevertheless, he has migrated to Instagram and YouTube and I think any sensible influencer and creator has as well.

01:53:51 - Dan Patterson (Guest)
Yeah, Creators for sure. Look, I had 1.5 million followers on Google+ Congratulations.

01:53:58 - Leo Laporte (Host)
How did you do that?

01:53:59 - Ben Parr (Guest)
I missed my Google+. Wow, so what happens?

01:54:02 - Dan Patterson (Guest)
if you are tied to one platform, that's what happens you get Google+. So I think, yeah, a lot of creators have already moved.

01:54:12 - Leo Laporte (Host)
Oh, yeah, it definitely benefits Meta. It's not even a question. In fact, it's pretty blatant, because Meta could not beat TikTok at their own game. They tried to buy Snapchat and so all they did was directly copy them with Reels on Instagram. And they couldn't buy them, couldn't shut them down, so they copied them. Let's take a little break.

01:54:36 - Ben Parr (Guest)
We've got lots more. I just found an article Go ahead, I was just leaving. I just found an old article in the Washington Post where it says Mashable's Ben Parr says Google Plus may threaten Facebook.

01:54:46 - Leo Laporte (Host)
Oh Ben, back in the day, that aged like milk.

01:54:49 - Iain Thomson (Guest)
Yeah, look, we've all done them. Mate, trust me.

01:54:52 - Leo Laporte (Host)
I loved Google+. I had very high hopes for Google+, so did I Me too.

01:54:57 - Ben Parr (Guest)
Last one, and then I'll stop going to the side. If they'd kept the Google+ and they just kept doing it, they could have deeply benefited from the last couple of years. I really do think they could have done something. They just kill and give up on things so quickly and early.

01:55:15 - Leo Laporte (Host)
Yes, far too quickly. Well, we're going to find out. They now have Gemini, gemini, gemini, gemini. They now have AI on their side. I noticed, by the way I'm talking, I did set this action button on my iPhone briefly for a few weeks to Gemini. Oh right, why do I keep on Gemini, I don't know. To Gemini. And now back to Scarlett Johansson. I mean sorry, chat GPT-40. It's Scarlett. I really feel like it's Scarlett, wait until the OpenAI Scarlett deal happens. Once they get the cutesy voice. Wait until the.

01:55:49 - Iain Thomson (Guest)
OpenAI Scarlett deal happens Once they get the cutesy voice. Yeah well, we interviewed someone who was the voice of Siri Australia and she's just like freaked her out. The first time she was at a party and somebody asked Siri something, she heard her own voice across the room. She should have said Ka-ching Scarlett.

01:56:00 - Leo Laporte (Host)
Johansson's got Maceu Ka-ching. Yeah, I wonder what Scar. You heard that voice. It's clearly her. All right, let's take a little break. You're watching this week in tech with ben parr. Ai guru author, uh at the information and co-founder of octane. Ai ian thompson, editor-in-chief us editor. I'm sorry, I promoted you us editor at the registercom always a pleasure, it was fun. Yes, dan patter, who is now fighting the good fight against disinformation, is director of content at Blackbird AI, our show today brought to you by InTouch CX.

01:56:39
Cx stands for customer experience and if you have a business that relies on customers, it's time to enter the 21st century. Revolutionize your customer experience, step boldly into the future with the amazing technologies and advisors at InTouch CX. Intouch CX empowers your brand to improve customer experiences with cutting-edge AI and automated solutions across voice chat and email support. This doesn't replace your touch points with your customers. What it does is enhances them. It enhances them. It helps you do triage. It helps you help customers better. The experts at InTouch CX will guide you through what many companies these days are finding completely overwhelming when to start, who to work with, what to prioritize First. As you become more successful, those customer touch points become more and more of a problem right, but it's a problem that could be a huge boon. Intouch CX will help you change how customers see your brand at every stage by harnessing the power of AI and automation. I can divide it up into three kind of stages Pre-interaction, the first interaction you can use AI to properly deflect 90% or more of your customer interactions, solve their problems before your humans have to get involved.

01:57:59
That means more satisfaction from the customers. It saves you money. During interaction, you can apply AI predictive analysis to increase customer satisfaction. No more customer service reps desperately flipping through notebooks trying to figure out what that customer is asking about. Ai will help them give them the best answer every time. And after the interaction, post-interaction you can achieve up to 82% resolution satisfaction by using generative AI to analyze all customer cases and improve your smart response templates, develop them. Improved accuracy, improved efficiency, improved productivity. It's modernizing your entire customer experience.

01:58:42
Intouch CX, intouch CX remember that name. They unlock new opportunities between user experience, customer experience and employee experience to see how real improvement in the metrics that matter most. They're revolutionizing how brands connect with their customers with a proprietary framework that identifies key areas of automation that drive productivity, engagement, quality and cost benefits across the entire customer journey in all industries. This is the future of customer experience. You need a partner to help you Get in the 21st century. Intouch CX is that partner. Intouch CX's solutions enable your teams to engage with customers in a personalized, efficient way, driving loyalty. Driving growth, because happy customers mean more profits. Transform your business by anticipating what's next. Discover new ways of working and how InTouchCX's industry-leading AI and automation solutions can get you there. Learn more and book a free consultation today.

01:59:44
Intouchcxcom slash twit Really go to the website. It was an eye-opener for me because InTouchCX does so many things in so many areas. It's really interesting to see, for instance, things like supporting your customers in the language they speak automatically. This makes such a huge difference. Intouchcxcom slash twit. You owe it to yourself and your customers to check it out. Intouchcxcom slash twit you need it. You owe it to yourself and your customers to check it out. In touch cxcom slash twit. We thank them so much for their support of this week in tech.

02:00:16
We were talking about internet shutdowns. Here's a case of a government showing the way to better internet. The state of utah, local utah governments formed something called Utopia, the Utah Telecommunication Open Infrastructure Agency. For a long time experts in internet access, for at least 10 years, have been telling me the best solution is have the government just as it is with water and gas and electricity have the government build the infrastructure and then lease the infrastructure to private companies who can compete on top of that infrastructure to give customers the best prices and the best service. And that's what happened in utah. The, the interlocal agency collaborative, set about building open access fiber network, so they built the infrastructure, allows any ISP to then come and compete on the shared network.

02:01:11 - Iain Thomson (Guest)
Yep, I mean, it's a no-brainer, and I mean the classic example of this is Chattanooga, which has been running their own system. You can get as a consumer for about $100 a month. You can get 2.5 gig and the speeds are only going up. Listen to this.

02:01:30 - Leo Laporte (Host)
Same thing in North Carolina and and iowas as well, I think. In utopia, 18 different isps compete on the same. It's the same fiber network covers 21 cities. In many cities, uh isps are offering symmetrical gigabit for 45 a month. Some are offering symmetrical 10 gigabit for $150 a month. This is mind-boggling and it just shows that a private and government sharing can do something amazing.

02:02:02 - Iain Thomson (Guest)
This was the biggest shock about moving to the US was quite how bad the system is when it comes to broadband and mobile, and we have the worst, most expensive internet. It's really. I mean. America built the fundamental building blocks of the internet and you've allowed two or three companies to basically take over the entire show massively ratchet up prices. I have never met a Comcast customer who was happy.

02:02:25 - Leo Laporte (Host)
And the reason this is only, by the way, in Utah is because of Comcast, at&t, verizon, the big ISPs who fight this tooth and nail.

02:02:33 - AI speaker (Host)
They want a monopoly. Oh, it's been decades fighting it.

02:02:35 - Leo Laporte (Host)
Without a monopoly they can't charge you all that money for such crappy service. Yeah.

02:02:41 - Iain Thomson (Guest)
I mean it's and the mobile is even worse. I mean, when I first got a US mobile phone, it was like you're charging me ten dollars a month to send a thousand text messages. That's insane. There's, no, no justification for that whatsoever, other than we can, you know great article by carl bode in tech dirt utah.

02:03:00 - Leo Laporte (Host)
locals are getting cheap 10 gigabit fiber thanks to local governments. I recommend it and then print it out and give it to your member, your member of congress or, better yet, your your city council, because the best way to do this is locally, to do this in a regional area, because you can sneak it in behind past Comcast.

02:03:19 - Iain Thomson (Guest)
Well, you can also on a national level. Finland and South Korea both built out huge fiber networks and then leased them out at that cost. It's just obvious.

02:03:28 - Leo Laporte (Host)
It's not what we're getting.

02:03:29 - Ben Parr (Guest)
It should be considered, you know, piping infrastructure? Yeah, it's not this like. Look, this is like. We know there are problems with the american system. Uh, and this is kind of a showcasing of that problem we could have we could be living in a very different world where everyone most everyone had access to very inexpensive Internet. We don't live in that system, yeah.

02:03:55 - Leo Laporte (Host)
More good news. Vermont justified national trends and passed the toughest yet state bill protecting online data privacy. This is I'm not sure I have mixed feelings about this Better than.

02:04:07
Illinois. Yeah, it's the private action model. The bill lets Vermont residents sue companies directly for collecting or sharing sensitive data without their consent. Now, normally I don't like this, but I have to say they really they've asked for it, so I think. So Lawmakers deployed a countermeasure as they drafted this, against business pushback. They brought together lawmakers in states from Maine to Oklahoma and asked them for advice. Monique Priestley, a Democratic state lawmaker in Vermont, said that the biggest thing I've learned from this bill is that big tech has a playbook, and she says we've developed a grassroots playbook to fight back. Now theont legislature passed it, but the governor has still not signed it. And uh, of course, big tech industry lobbyists are moving in with bags of little black bags of something I don't know what's in there for, like this country, that you've legalized bribery and now call it campaign contributions.

02:05:11 - Iain Thomson (Guest)
It's just one of those marvelous.

02:05:13 - Ben Parr (Guest)
Look that was a recent phenomenon because of the supreme court united baby citizens united. The worst decision, one of many anyway, give them time, since jesus was a sophomore republican governor phil scott, yet to sign the bill.

02:05:33 - Leo Laporte (Host)
I think put pressure. If you're a vermontian, put pressure on your governor. Uh, this is a good bill.

02:05:39 - Benito (Other)
That would be very uh useful I feel like it would really plug up the courts, though no yeah, I don't like private, I don't like the private action.

02:05:48 - Leo Laporte (Host)
Uh, this is what tex, what Texas used against abortion providers. This whole idea of right of private action. I think the government should be the one pursuing it. But the problem is government doesn't, and the threat of me suing a business because my data has been leaked is a bit more chilling to these businesses. But you're never going to outspend the corp, the company. But which would you rather fight a horse-sized duck or a thousand duck-sized horses?

02:06:20 - Dan Patterson (Guest)
that's the pro, that's the solution here I could really see that harming small smbs and startups in particular, because they don't have the resources to fight a lawsuit. But I mean, what is data gathering, harvesting? I mean, anybody running a WordPress site is If they don't do it then they don't have to worry.

02:06:40 - Leo Laporte (Host)
Just don't do it. Just don't do it right. Ll Bean is strongly opposed, of course, to a similar bill in Maine. Strongly opposed, of course, to a similar bill in Maine. Actually, orvis is in Vermont and Orvis says this would put Vermont businesses at significant, if not a crippling, disadvantage. Anyway, it's possible to get privacy bills at least partway through the legislature.

02:07:08 - Iain Thomson (Guest)
Well, the big worry with privacy bills is they have to do a. You know, california's got a pretty good one, illinois's got a pretty good one. The big danger is that we have a federal privacy law which basically undermines the best parts of those but just provides some base spec. You know.

02:07:20 - Leo Laporte (Host)
Yes, maria Cantwell is sponsoring something in the Senate. She is, of course, the former executive at Real Audio, who has at Real Networks, who's done everything she can to torpedo all the previous privacy bills. Now, weirdly, she's supporting a nationwide privacy bill. And oh, interestingly well, one of the features is it abrogates all the state bills, so it's like no, we got a federal law now, so none of those state bills apply.

02:07:48 - Iain Thomson (Guest)
It's going to be interesting to see if that comes out. Well, that's how that relates to states' rights, but anyway, another thing.

02:07:53 - Leo Laporte (Host)
Oh states' rights, schmates' rights.

02:07:58 - Iain Thomson (Guest)
This is a silly country at times.

02:08:00 - Leo Laporte (Host)
Yeah, you know, you're an outsider. We revolted. I'm from.

02:08:03 - Iain Thomson (Guest)
Brexit Island, so I'm hardly in the position.

02:08:06 - Ben Parr (Guest)
Look I was going to say like you got some silliness over from Brexit Island over there, I know.

02:08:12 - Iain Thomson (Guest)
How's that going? By the way, my wife learned some interesting swear words when I stayed up for the vote. It was just it's going badly. It's probably the largest act of economic self-harm any country has ever committed. Long lines at the channel, oh well, actually long lines of produce trying to get in and out of the country, because all of a sudden now we've got to have all the lorries checked when they come back and forth from the continent. So if you've got a significant export business in Europe, then you're leaving the UK and moving to Europe because otherwise no one's going to buy from you.

02:08:43 - Ben Parr (Guest)
Don't ever underestimate the ability for people to shoot themselves in the foot.

02:08:48 - Leo Laporte (Host)
Yeah, You're here.

02:08:49 - Iain Thomson (Guest)
Yeah, absolutely.

02:08:51 - Leo Laporte (Host)
If you're wondering why, all of a sudden, elon Musk is tweeting the falsehood that Signal is somehow insecure. It turns out it all started with a smear campaign against NPR. Oh, for goodness sake, it did start on X, thank you. God bless it. By the way, way x is now x. No more twitter.

02:09:14
It all turns into x now yeah, well, which I'm happy about because I don't like it, because they stepped on our name, I don't like it. So now they're x and uh. That's an even worse search engine name. Um, right ring media, this is from uh. This is from uh. The guardian right wing media personalities on x started the fight by going after npr a long time. You know, bet noir of the of the right. Um. It started in april when a long-time former editor at npr wrote an essay in the free press arguing npr had increasingly chosen to cater to a very small subset of left-wing America, which actually I won't dispute.

02:09:58 - Iain Thomson (Guest)
You don't really hear calls for Fox News to address a slightly wider subset of users either, though. Well, yeah, but there you go.

02:10:05 - Leo Laporte (Host)
So not only are you targeting the left, but you've got a crappy audience too. I mean, there's tiny. So debates about NPR's alleged lefty tendencies and conservative calls to defund it are not new. But this all started at the start of NPR's new CEO, Catherine Mayer's, term, and it turned out she had some bad tweets. Oh, bad progressive tweets anyway. You mean ex posts.

02:10:31 - AI speaker (Host)
Yeah, ex-posts.

02:10:33 - Ben Parr (Guest)
What do you?

02:10:33 - Leo Laporte (Host)
call them Sweets.

02:10:35 - Ben Parr (Guest)
I think sweets is good. No, you call them posts. It's worse, it's just posts. It's just posts.

02:10:41 - Leo Laporte (Host)
Yes. So then this is all from the Guardian. Right-wing activist and propaganda guru, chris Ruffo took over the campaign against her and then it turned out this is a very complicated uh thing it turned out that she was also on the board of Signal, the Signal Foundation. Ah, so the Guardian says and this, by the way, you might want to steal this from them, this is a very register-like phrase Via the transitive property of bad people, everything she's linked to is now suspect. So, as long as Maher's still running, n npr rufo posts another blog post, who says signal is compromised because she's on the board? No, signal.

02:11:39 - Iain Thomson (Guest)
The key reason why I use signal and why I trust signal uh is down to moxie marlin spike, the co-creator, because there is no way he would be backdooring exactly under any circumstances.

02:11:50 - Leo Laporte (Host)
Matthew Green concurs. The code is on GitHub. This has never been in dispute. This is a reliable, not backdoored, strong encryption messaging platform. Use Signal. However, the fact that Meyer was on the board and that Signal had a grant from the Open Technology Fund, which is sponsored by the US government, the Signal president who'd picked Maher for the board was also a progressive who'd been an equity rabble-rouser at Google. Anyway, it all turned into a campaign against Signal which, unaccountably, Elon Musk piled on then and said well, you know there are. Elon believes anything. The last person in the room told him. I think that's really clear now. He just he doesn't have a filter.

02:12:38 - Iain Thomson (Guest)
Well, unfortunately, I mean the Germans have a word for it. I mean the Germans have a word for everything, but their particular word is for shidiot, which means someone who is highly skilled in one or two areas?

02:12:49
Yes, is highly, which means someone who is highly skilled in one or two areas. He's highly skilled in one or two areas and thinks that translates to everything else they think, and he's the classic example of that. I mean, there's no way. Signal is back to Ord. It would, just there would be so many red flags set off by everyone.

02:13:02 - Leo Laporte (Host)
He tweeted a couple of weeks ago there are known vulnerabilities with Signal that are not being addressed Seems odd To which the entire informed world responded what are you talking about? Jack Dorsey boosted the allegations. Signal CEO responded, pointing out Signal is open source closely scrutinized by the security and privacy community. It's just absurd. But then Telegram's CEO, pavel Durov, who is Russian, although I don't know how much he likes Putin, but he has left Russia to work out of Dubai said oh no, of course you should use Telegram instead. To which Matthew Green and others said no, whatever you do, don't use Telegram. Instead, stick with Signal.

02:13:55 - Dan Patterson (Guest)
This might be an example of what we call a narrative attack Misinformation and disinformation combined to attack an individual or an organization. It is a dangerous use of mis and dis.

02:14:12 - Leo Laporte (Host)
yeah wow, well, that's good.

02:14:14 - Dan Patterson (Guest)
So this is a perfect test tube example of that kind of thing where, uh, the initial target created a narrative which was then modified to smear another unrelated, you know and amplify right if you want to right step back, like I'm not really interested in the components of this story, but if you just step back and look at those components as neutral entities, right, it's a. There is a narrative that has been spun up and mis and disinformation are being um. The social web is used to amplify mis and disinformation and then target an individual or an organization and, as we see, the outcomes of that can be pretty devastating even for somebody with the credibility of the entire organization of.

02:15:05
Signal.

02:15:08 - Leo Laporte (Host)
All right, we're going to take one more last break. I do have a couple more stories to wrap it up with. Dan patterson, are we doing the nuclear?

02:15:13 - Iain Thomson (Guest)
missile facility in the chinese crypto farm. I think we are now okay.

02:15:18 - Ben Parr (Guest)
That's cool just put it, smash those words together take a sound bite of that statement that sounds like narrative disinformation to me, but I will let you tell the story.

02:15:29 - Leo Laporte (Host)
It's straight out of the registercom, so I believe it, and it came from the president of the United States of America Executive order, no less. But we'll do it after the break Coming up. First, I want to tell you how to protect your home with something really cool Our show today, brought to you by the Eufy Video Smart Lock E330. We put it on our engineering room as a kind of proof of concept. I said Micah, you're pretty handy. He said what are you talking about? I said, micah, you should install this Eufy Video Smart Lock on our engineering room door. He said are you crazy? I said no, all it takes is a Phillips screwdriver, no drilling, it fits right in. And man, this is a major upgrade to our security.

02:16:14
The Eufy VideoLock, of course, has a camera. It's easy to install. It also has a fingerprint sensor 0.3 seconds and one second unlocking. See that deadbolt right there. The embedded self-training chip becomes more accurate every time you use it, so it's faster and faster. No more battery anxiety. Look at the size of that battery 10,000 milliamp rechargeable battery. It's easily removed. It could last about four months. You'll get a battery notification before it runs out, so you don't have to plug it in. Look at this. Look how fast that fingerprint is.

02:16:51
You also can control it remotely through the eufy video lock app. And oh, if the battery dies or for whatever reason, you want to use a key look at that, an actual keyhole. I love that. Control your front door remotely using the app if somebody like I don't know burke wants to get in. He rings the doorbell. You say yeah, who is it? He says it's bur. Burke wants to get in? He rings the doorbell. You say yeah, who is it? He says it's Burke. I need to get in the engineering room. And you say, okay, come on in, you can unlock it. It's got passcode unlocking remote control.

02:17:20
2k clear sight, two-way audio, enhanced night vision. This is so cool, but I left out the best part no subscription, no monthly fees. Your video lock recordings are stored locally. You never have to pay for storage. Enjoy a worry-free experience, 18-month warranty, all backed by Eufy's 24-7 professional customer service team. This is the smart lock you've been looking for. Get yours today. Search for Eufy Video Lock on Amazon or visit eufycom E-U-F-Ycom. You may say who's Eufy when they're at home. What do you say? What do you Brits say? What is he when he's at home?

02:18:01
An Englishman's home is his castle. Oh no, yeah, you say that also, but this is Anker A-N-K-E-R. This is their home security brand. So you know, anker Anker's a great company. Eufy Video Lock E-U-F-Y dot com. Now, what's all this, then, about a Chinese nuclear facility and a crypto farm next door?

02:18:26 - Iain Thomson (Guest)
No, it's an American nuclear facility. There was an executive order issued this week by President Biden and one of the Francis E Warren Air Force Base where they store the Minuteman III missiles, and they discovered through a tip off from a member of the public that there was a Chinese run crypto mining facility within a mile of the base. Within a mile of the base. So the executive order basically says right, that's a national security risk, you have 120 days to sell the facility, and so you really should have told us about it beforehand. So if you're looking for a cheap bit barn up in Wyoming way, then there'll be one going fairly, fairly cheaply.

02:19:04 - Leo Laporte (Host)
But it was just yeah, it was a very strange story and it took a presidential order to shut this down.

02:19:09 - Iain Thomson (Guest)
Well, basically, if you're a foreign ownedowned business and you have to register such things with a particular government department, yes, they didn't do that Because they didn't want anybody to know. Presumably so, yeah, or it just slipped their mind. But whatever, it's going to cost them now because the wrath of Joee bryden biden has come down on them and they've been told to sling the hook, that is hysterical, it's just their plan to go to the moon the reason they put it there is probably because electricity is cheap right, presumably.

02:19:43
I mean that's one of the key determinants, but um I mean.

02:19:46 - Leo Laporte (Host)
How does mining bitcoin next to a us nuclear facility help anybody?

02:19:51 - Iain Thomson (Guest)
It doesn't. I can't see any way, unless they're actually burrowing.

02:19:55 - Ben Parr (Guest)
It helps whoever mines the crypto for themselves? I guess yeah.

02:19:59 - Leo Laporte (Host)
But it's not really, is it you think? A security? Look it, I'm not against it. It's not like banning TikTok. Go ahead and do it.

02:20:12 - Ben Parr (Guest)
I mean you could have. You don't need to have anyone from a foreign government that close to a nuclear reactor. That's probably a safe bet yes, it's uh.

02:20:19 - Iain Thomson (Guest)
Well, it means you get an early warning if they actually do launch. But yes, it's um, if that happens, it's not really going to matter that much, but it was just a really weird case that is done by executive order, everybody's short bitcoin.

02:20:30 - Leo Laporte (Host)
They've launched the nukes.

02:20:32 - Dan Patterson (Guest)
Yes I think there's diplomatic communities who see small incursions like this, uh, and then maybe the balloon as a pattern, um, and I I mean, who knows, but I I think that that there there could be. I can see that rationale like, if you're just trying to like, maybe poke at your adversary a little bit, you might do something like this yeah I mean, I should imagine we're doing the same sort of thing in gchq in the uk or the nsa.

02:20:59 - Iain Thomson (Guest)
There there was a lovely interview with an ex-nsa bloke when the office of personnel management managed to leak everyone's fingerprints and security clearances. Um, they asked him about it and he goes yeah, that's a good database. If we'd seen that in china we'd have gone after that, you know quite open about it.

02:21:16 - Leo Laporte (Host)
Good stuff in there training, data yeah, all right, you know, aren't all the cameras and all the polls all around that nuclear base made in china?

02:21:25 - Iain Thomson (Guest)
oh almost certainly okay, I'm just just asking I'm gonna go find.

02:21:31 - Ben Parr (Guest)
I want to go look at the cameras and find out.

02:21:33 - Leo Laporte (Host)
I wonder if they actually aren't well, they banned the huawei stuff, right, because right huawei, but that I think that they're still mostly made in china because it's just not economical to make it anywhere else yep, I mean it's the whole huawei.

02:21:50 - Iain Thomson (Guest)
I mean it was great for headlines. Don't get me wrong. You know us says no huawei, but no, it's just you know it's theater, if you ask no way, huawei yeah yeah that's a good slogan. We did go with huawei to the danger zone on one story, but it just what was your?

02:22:04 - Leo Laporte (Host)
deck I get the deck. What's the nut actually? What?

02:22:09 - Iain Thomson (Guest)
was in the subhead. Oh God, I'll need to find out. Hang on, we did get into trouble with the Twitter subhead.

02:22:21 - Dan Patterson (Guest)
I don't want to ask, but is nut graph?

02:22:22 - Ben Parr (Guest)
kind of what I think it is Obviously.

02:22:25 - Leo Laporte (Host)
I don't know what you think it is Finally, all right. On a more somber note, I like to mention uh the pioneers who've made uh modern day technology possible, often unsung, unnamed. Oh, you probably don't know the name, robert denard, but you definitely use his invention. In fact, without robert denard there would be no modern computers. He invented DRAM. He was at IBM when he invented it. Prior to this, you might remember, ram was often wires running through magnetic cores and very limited in its density, thanks to the invention of DRAM in 1966, he invented a way to store a digital bit on a single transistor. He called it dynamic random access memory. And without DRAM, you know, the computers wouldn't be as small as the one in your pocket. Of course, there's been major progress made since his invention, but John Hennessey, a computer scientist and chair of Alphabet, said DRAM made much of modern computing possible.

02:23:41
Denard passed away this week at the age of 91, the father of DRAM. Let's remember his name because without him we would not have all the benefits of modern computing. He also invented something called Denard scaling, which I'm not smart enough to understand. So remember Moore's law, which said the number of transistors will be doubled every 18 months or every two years. Moore's law concerned the density of transistors on a chip. Dennard's law, or Dennard's scaling, concerned power consumption. He said that he described the physics that would allow transistors to shrink and become more powerful and less costly, even as the energy each one consumed would remain almost constant. So it was kind of the more technical version of Moore's Law. Yeah.

02:24:37 - Iain Thomson (Guest)
Also Moore's Law was kind of just a suggestion. We used to have bets at Intel developer forums. There'd be a sweepstake amongst journalists as to how quickly they'd mention Moore's Law, and it was never over three minutes. So you know you had to really pick your times on that one. That's pretty funny, but it is remarkable how so much technology comes down to just a handful of people.

02:25:00 - Leo Laporte (Host)
Lisa Su, who's the CEO of AMD, said everybody in semiconductors studied Denard's principles to get where we are today A name I had never heard before and yet absolutely seminal in the dimension uh, the invention of microcomputers. When he was at ibm in 1966, invented dram. Uh rip, robert dinard, ibm inventor. Um, I usually save those sad air stories for the end of the show, but actually this whole show has been depressing as hell.

02:25:29 - Iain Thomson (Guest)
So I want to thank you all. Cheers mate, I want to thank you all.

02:25:33 - Leo Laporte (Host)
AI's going to cure cancer.

02:25:35 - Dan Patterson (Guest)
Oh no going to 614,. Okay, you're being the optimist and I'm like well, I think everything's terrible.

02:25:43 - Leo Laporte (Host)
That's Dan Patterson. He thinks everything is awful Director of content. No, he doesn't. At is awful director of content. No, he doesn't At. Blackbird AI, are you bullish? You have a little baby.

02:25:51 - Dan Patterson (Guest)
You've got a child. Oh, I'm bullish Guys. I really appreciate you letting me be a skeptic. It's just like it's born. I can't not be a journalist and just think about everything through a skeptical lens. It's just my wiring. However, I am so stoked. Look, I'm using Gen AI every day. I wish that I had been your enthusiasm and optimism. I want like 10% of because it's fantastic. And, ian, like I've been following your work for years, your cyber reporting is fantastic.

02:26:24 - Iain Thomson (Guest)
Oh, thank you Anyway, so sorry.

02:26:26 - Dan Patterson (Guest)
I didn't mean to be so depressing in the show.

02:26:27 - Leo Laporte (Host)
No, it wasn't you. I'm not blaming you. We all were just the way it is. We've got to look ahead, but I think it probably is important to say the future is probably pretty exciting. It's going to be interesting.

02:26:43 - Ben Parr (Guest)
We'll go with. We have always been able to figure out how to move forward into the future.

02:26:47 - Leo Laporte (Host)
I like that.

02:26:48 - Ben Parr (Guest)
And we will continue to do future. I like that and we will continue to do that. I like that. We have survived World War. We will figure this out.

02:26:56 - Leo Laporte (Host)
That's good. Dan works at Blackbird AI. He's the director of content there. It is blackbirdai. What is the URL? Yeah, blackbirdai, and I just signed up, for I guess it only takes 48 hours to get access. Yeah, that's fine. Is that right? Yeah?

02:27:13 - Dan Patterson (Guest)
Compass. Oh yeah, and I'll just ping our team. I mean, we're a very small team. I effectively run the blog that you see on the top right. Our team just writes these fantastic stories that I get to. Like. Tomorrow morning I'm going to wake up and edit some uh, really well written and interesting uh research and that's a that's a great job.

02:27:35
No, kidding, and I notice you use AI for all the illustrations oh for, like I look, I don't have the budget for for um stock photos. I mean, it's really just it, I just don't have the budget.

02:27:47 - Leo Laporte (Host)
do you have, like the beginning of a prompt always says a blackbird in and then you just provide wherever that blackbird is?

02:27:55 - Dan Patterson (Guest)
No, but I do have techniques, like, and I use a blend of tools so I get nice stuff, that's you don't have to tell us your secrets, that's okay.

02:28:03 - Leo Laporte (Host)
That's okay. This is no secret, a great place to get great information. I love it, blackbirdai. It's great to have you on, dan. Thank you so much you too, Really appreciate it.

02:28:14 - AI speaker (Host)
Thank you, I appreciate it.

02:28:15 - Leo Laporte (Host)
And your doggy who was coming and going the whole show, but no one knows about me.

02:28:20 - Ben Parr (Guest)
I saw it, that's all right. I can't control what Fred says.

02:28:24 - Leo Laporte (Host)
No, I love him, says ben parr. They are. Yes, ben parr is an ai guru. He has been for years. Finally, the world's catching up to him. The author of the ai analyst, co-founder at octane ai, he writes about ai at the information. Anything you want to plug besides your beautiful new bride and your soon-to-be large family?

02:28:45 - Ben Parr (Guest)
I have some interesting newsletters. I'm going to drop on my newsletter, benparcom, and maybe an announcement or two. Give me some time so you go to benparcom and you'll see the crazy announcements. Been working on some stuff behind the scenes.

02:29:00 - Leo Laporte (Host)
Interesting. If you want to know more about AI, subscribe to the AI Analyst by Ben Par at benparcom. Thank you, ben, it's great to see you, and congratulations once again and all the best to your new bride, mr ian thompson the legendary ian thompson, us editor of the registercom usually say that about someone till they're dead. But it's always great to see you. Thank you for coming all the way up here.

02:29:28 - Iain Thomson (Guest)
I appreciate it, you have an in and out in petaluma, so you know that's my stop-off point when I go back.

02:29:32 - Leo Laporte (Host)
Oh, you go to In-N-Out. You used to go to the British store. I guess we were kind of— yeah, the British store shut down there, so In-N-Out it is. You're becoming an American, aren't?

02:29:48 - Iain Thomson (Guest)
you? Yeah, double-double mustard burger, sorry, mustard grill with the whole grilled onion, nice yeah.

02:29:54 - Leo Laporte (Host)
Yeah, I guess you're right. They're kind of soggy, aren't they? There are no McDonald's beef tallow fries. Those were the gold standard.

02:30:02 - Iain Thomson (Guest)
Yes, such is life, and I've got to say, Dan, I'm with you on the cynicism side of things. It never leaves you it could be worse it's a journalistic hazard. It could be worse. I had a former editor who, when he went to an interview with somebody he really didn't like, he had a T-shirt printed up with why is this bastard lying to me? Oh nice, and so we'd just sit there in front of him.

02:30:23 - Leo Laporte (Host)
But are you bullish about the future?

02:30:25 - Iain Thomson (Guest)
Yes, I mean it's going to get worse before it gets better. But I think Ben's right, the long-term prognosis is good. I mean, if you believe some of the scientific studies, we went down to about 8,000 humans on the planet at one point.

02:30:37 - Leo Laporte (Host)
Yeah, talk about choke points. We can come back from a lot. We're all descended from the same 8,000 people. That's depressing, Ben Park. I know you're bullish, the future is bright.

02:30:49 - Ben Parr (Guest)
You know what. You can also look at that as inspiring. Exactly, it's all how you view it.

02:30:55 - Leo Laporte (Host)
I'm inspired, and we should all be inspired, to make a difference, to go out in the world and create the world we want to grow up in and we want our kids to grow up in right. We do this show, and maybe that's a little bit of what our mission is as well. Every Sunday, from 2 to 5 pm Pacific time, that's 5 to 8 pm Eastern you can watch us do it live at youtubecom slash twit, slash live. Of course, if you're a member of Club Twit, you can contribute, participate in our beautiful Club Twit Discord. You also get ad-free versions of all the shows we hang out during the week. In fact, micah just announced a new show where you're going to get to, I guess, crochet with Micah. It's going to do a creative hour.

02:31:39
I'm also doing a music stream are you going to do a music stream? Will you Benito's going to do a music stream? Now, what am I going to do? I've got to do something, a couple of weeks ago we had a watch party. We all watched the very first really sci-fi movie uh, fritz lang's amazing metropolis. That was a lot of fun.

02:31:58 - Iain Thomson (Guest)
1927 with the original original soundtrack.

02:32:01 - Leo Laporte (Host)
We didn't play the sound, because we uh didn't want to violate any copyright, but the movie itself and we were the soundtrack, all that stuff. It was a lot of fun. Anyway, that's the kind of thing we do in the club. It's seven bucks a month. Really, the main reason to join is you're subsidizing the content that you want. If you like what you hear here, please join the club. We'd love to have you.

02:32:25
Some of the best people and the smartest minds in the world uh convene every week in club twit twittv slash club twit to learn more after the fact. This show will be available on our website, twittv. You can also get it on youtube. There's a youtube channel dedicated to this week in tech and the video version of it, and you can subscribe to audio or video in your favorite podcast client. That way, you'll get it automatically Just in time for your Monday morning commute. Here we are in our 20th year talking about the future, and it's never been more interesting, has it? It really hasn't. I'm very excited about what's to come. We thank you for being here, we'll see you next time and, as I have said for 19 plus years, thank you for being here. We'll see you next time and, as I have said, for 19 plus years.

02:33:11 -Leo Laporte
Another twit is in the can. 
 

All Transcripts posts