This Week in Tech Episode 999 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. We're going to kick things off with a visit with my son Salt Hank. His new book is coming out and speaking of new books, we also have Parmi Olson on. She is the author of a brand new book about AI fascinating called Supremacy. Denise Howell, my internet attorney, joins us. And from Windows Central, daniel Rubino. Lots of AI talk, but lots of other stories too. Twit is next. Podcasts you love. From people you trust.

00:30
This is TWIT. This is TWIT this Week in Tech, episode 999. Recorded Sunday, september 29th 2024. Bananas and browsers.

00:57
It's time for Twit this Week in Tech, the show where we get together with some of the best names in the tech journalism business talk about the week's tech news. I'm going to break format a little bit briefly here because, uh, I'm a proud papa and my son's new cookbook is coming out on tuesday and I thought I just I thought I'd get salt hank on just to talk a little bit about salt hank, about the cookbook. Give him a little bit of a plug and then we'll get to our panel. We've got a great panel for you, ladies and gentlemen. Uh, I'd like to introduce you to somebody I know pretty well, a long time member of the family. Since he was born, I've known this guy, my son henry laporte, better known as salt hank what's going on? The author of a brand new cookbook which comes out t Tuesday from Simon Schuster. Yeah, salt Hank, a five-napkin situation, got it right here. Oh, look at this. Look at there, he has stuff in his mouth. Kind of a disgusting cover.

01:56
It's more than a cookbook, though Flip through it a little bit, because it's really a picture book.

02:00 - Salt Hank (Guest)
I mean you could put this on your coffee table. It's bon. I mean this is on your coffee table, is bonkers. Oh my God, what is that Poutine? This is like, basically, animal fries. It's recreated animal fries from In-N-Out but we had to call them feral fries for copyright reasons. But they are like just very gourmet versions of animal fries. That's one of the honestly my favorite.

02:23 - Leo Laporte (Host)
Somebody's asking Keith in our discord. Don't parents just teach their kids how to cook so they can get them to make lunch for them? I want to tell you something I have never had anything. Hank has made. That's not completely true. I don't think I've had your French fries, which are a miracle. You cook them like three times.

02:40 - Salt Hank (Guest)
Well, it depends what kind of French fries you're making. If you're doing shoestrings, cook them once. If you're doing like the Michelin star fries, where you boil them, take them out, fry them, take them out and fry them again, that's the three-time cook thing.

02:51 - Leo Laporte (Host)
Henry's on his houseboat where he does his Salt Hanks studio. It's actually called the Salt Lovers Club. We should get that straight the Salt Lovers Club.

03:01 - Salt Hank (Guest)
Oh, that's just the salt company, so that's a little bit separate. We sell seasonings and pickles and stuff like that and that's the name of it, yeah, salt Lovers Club, and we have a big neon sign above, kind of where I cook.

03:13 - Leo Laporte (Host)
Yeah, like when I see you cooking on the Insta, you're there in front of your sign that says the Salt Lovers Club right there, like that. Well, that's just to promote the heck out of the salt company, but right now I'm like not, but isn't the salt company really just like a sideline for salt hank?

03:34 - Salt Hank (Guest)
what do you mean? Like a side hustle, side business?

03:36 - Leo Laporte (Host)
side hustle.

03:37 - Salt Hank (Guest)
Well, right now, mr beast's burgers and yeah, it's kind of like that. I mean, if you want to be like, if you want to start a product, it's kind of like that. I mean, if you want to start a product, that's kind of a way into the longevity of the influencer world a little bit, and then you can make cool stuff like pickles. I literally can't wait and, by the way, I'm really pumped to send you some of these pickles.

03:55 - Leo Laporte (Host)
Disclaimer I am an investor in the pickle business. That's true, yeah.

04:00 - Salt Hank (Guest)
You have to legally make people aware of that.

04:04 - Leo Laporte (Host)
I think I have to. I think FTC regulations say that if I'm going to tell you, promote something that I have a financial stake.

04:11 - Salt Hank (Guest)
You're an investor in all this. You gave me my first camera, so technically you you basically like own a chunk of this entire business. Well, where?

04:20 - Leo Laporte (Host)
okay, where do I get paid off? Owning a chunk means mainly I just yeah, I gave you cameras.

04:28 - Benito (Announcement)
I you were into d drones, you have drones I wanted to support you in this stuff, and little did I know but you, when you were a kid, all you ever did is watch those extreme food videos on youtube yeah, really good food porn.

04:41 - Salt Hank (Guest)
maddie matheson, my boy, my daddy, my dog, matty who loves you now.

04:45 - Leo Laporte (Host)
By the way, Isn't that cool that you've become friends with Matty Matheson. That part's pretty wild.

04:50 - Salt Hank (Guest)
I don't know if he knows how crazy it is for me when he calls me. I'm like dude, you're my hero.

04:56 - Leo Laporte (Host)
You were on his show just a week or so ago. He woke you up.

05:05 - Salt Hank (Guest)
Yeah, it was literally. Literally, I was laying in bed right there completely asleep. He's like hey, what's up? You're on the show, like what's going on? Describe what you're doing right now he's like.

05:10 - Leo Laporte (Host)
He's like cursing you out for these steak sandwiches, and yet he's making it and it's incredible.

05:16 - Salt Hank (Guest)
Well, he thought it was gonna. I don't know what he was doing. He was like hank got famous for this one sandwich. He's it's not true.

05:21 - Leo Laporte (Host)
That's not true. You know what for many?

05:23 - Salt Hank (Guest)
things whatever he wants. He's the king, he's kind of the godfather of like food content and food media as far as the internet goes. So he's, he's got carte blanche access to say whatever he wants about anybody and yeah you know he's a sweetheart, but honestly, uh, there, there he is. By the way, there you are with maddie madison eating something I had to breakfast, like get him to come do a video with us.

05:48 - Leo Laporte (Host)
Oh, look at that oh, dipped in gypped in queso. Oh man, huge maddie's doing the cooking.

05:55 - Salt Hank (Guest)
I recognize those tats I literally didn't cook much at all during this video. I was just filming them like kind of go crazy. I made the brown patties and that little vodka things that we chopped up and put in the burrito, but other than that, maddie was just going nuts in the kitchen.

06:09 - Leo Laporte (Host)
It was so if you like food porn, cause this is what Henry grew up on his food porn. Follow salt Hank on Instagram salt underscore Hank. Thank you, dad. Yes, and the cookbook get the cookbook, because the cookbook is out tuesday and you can get 40 off, a target right now you get 40 off on walmart. Yeah, uh, what's the? Sweepstakes. I can't do sweepstakes. That's illegal so I won't. You could get buy salt from the salt lovers what are you looking at right now?

06:41 - Salt Hank (Guest)
I'm looking at your salt, hankcomiio oh yeah, this is like the link tree thing.

06:46 - Leo Laporte (Host)
Yeah, it's got all there's got everything you can buy the signed copies.

06:50 - Salt Hank (Guest)
Look at that oh yeah, and a crap load of them, but uh yeah, 40 off at walmart target and amazon right now how many did you sign? Like 2500 or something. Oh, or something. Oh, my God, your hand must be killing you. No, actually, it taught me how to. I didn't have a signature before and I did enough. Oh, you do Once. I got to 500,. I was like oh okay, here it is.

07:14 - Leo Laporte (Host)
October 2nd, Saul Hank will be at the Barnes Noble in Mira Mesa, San Diego, California, cooking. Or do you do a reading from your book?

07:25 - Salt Hank (Guest)
No cooking or do you just, uh, do you do a reading from your book? No, I wanted to do pop-ups. I thought that would have been like so cool to like, do like, if you buy a book, you get a free sandwich, like you know, and just make a bunch of sandwiches in all these cities. But but we're just going to be in like barnes and nobles kind of doing little chats with people from those. Okay, it's funny to do a reading from a cookbook.

07:40 - Leo Laporte (Host)
no, then you take the onions and you slice them. It's not that he will also be in Los Angeles on the 3rd at Diesel Brentwood. Owen Hahn, his sandwich buddy, will join him. San Francisco Book Passage up our way on the 5th. That'll be fun. We love Book Passage. That's one of the best bookstores in the world. In Chicago at Anderson's North Central College on October 7th and in Brooklyn, brooklyn, october 8th, powerhouse Arena.

08:06 - Salt Hank (Guest)
Brooklyn's a big one. If anyone knows, wishbone Kitchen or Olivia T they're giant. They're going to be co-hosting with me for that one. So if you're fancy, you should come out to that.

08:15 - Leo Laporte (Host)
It's going to be sweet and the book comes out Tuesday. You can pre-order now, 40% off at Walmart. If you go to Hank's Instagram page, you will see that, um, I didn't really get to interview you. We're going to do the uh, the regular show now, hank I know yeah, do your thing. I don't want to come back wednesday on twig, because I know paris and jeff want to hang and I will do. I will interview then and find out how you got into this.

08:40 - Salt Hank (Guest)
Well, wednesday is the first book tour, do it from the hotel. It's okay, possible. I'll let you know.

08:46 - Leo Laporte (Host)
Pacific I got to check in with my own son. Won't even come on.

08:49 - Salt Hank (Guest)
I would love to. Of course, you're the one kicking me off right now, dad he's such a star this is the first day of the book tour, but I absolutely will, that's okay.

09:01 - Leo Laporte (Host)
Henry, I'm very proud of you. You've done.

09:03 - Salt Hank (Guest)
You've done good son okay, thank you guys, sorry for interrupting. I love you.

09:08 - Leo Laporte (Host)
Anybody that's listening, take care all right, don't forget salt Hank, a five napkin situation. He did not pay me for this ad, by the way. I just, you know, I just thought, hey, it's my son, I could do that, you could do that. When Isla writes her first book. Parmy Olson is here from her daughter's bedroom.

09:31 - Parmy Olson (Guest)
I love it. Thank Isla for letting you use her bedroom tonight. She's going to be so pleased to have had a shout out on this week in tech.

09:34 - Leo Laporte (Host)
No, she's not. No, she's going to go. God, I'm so embarrassed.

09:37 - Parmy Olson (Guest)
She will in a few years when she starts listening to this. That's right.

09:39 - Leo Laporte (Host)
That's right. Parmi is the author of a fan, a technology columnist at Bloomberg. She's written for the journal and many other places, but he's just got a brand new book. By the way, I love the cover of Napoleon. And then it says supremacy, ai, chat, gpt and the race that will change the world. And it is a fascinating look at Sam Altman at OpenAI about. I mean, there's so much politics. The history is incredible and your timing is pretty good because Sam's back in the news. We'll talk about that in just a second. Parmi, thanks for staying up late with us. I appreciate it. Also here from editor-in-chief from Windows Central Magazine, our Windows guru, daniel Rubino. Always good to have you on, daniel, thanks for joining us today. Thanks for having me. Yeah, there's a little news, a little Windows news, a little Microsoft news we can talk about just a little bit. Also here, denise Howell, who is my online attorney, host of two podcasts at hearsayculturecom, one of which I was on not so long ago. It was a lot of fun. Hi, denise.

10:43 - Denise Howell (Guest)
Hi, it was a lot of fun.

10:44 - Leo Laporte (Host)
Good to see you see you 999th episode of twit can you did you see that I did terrifying. Uh, we are going to do a 1000. You know I don't like I feel it's too self-congratulatory to celebrate all that, but people have insisted so we're going to get some of the old timers on uh the show next week to uh people who are on the original twit, not 999 episodes ago. 1000 is next week, but this, this is a pretty good one.

11:14 - Denise Howell (Guest)
This is a 999 is a pretty good, I feel like this is the bottles of beer on the wall, one, yeah, exactly thank god we're not doing it backwards.

11:23 - Leo Laporte (Host)
I'm not sure.

11:25
Had I known when I started that we would be doing a thousand of these, I might not have signed up for that. I don't know. It's been pretty amazing. You know, what's been amazing is watching, because that's 20 years, and 20 years in technology years is like a hundred years. I mean it's watching things happen like crazy. And the biggest of course, change has come right at the end, which is the advent of chatbots, of AI, of deep fakes, of AI imagery. Parmi, I'm really glad we could get you on because Sam has just written an AI manifesto. That I think is quite interesting. The book Supremacy starts with Sam Altman, his youth. He was a very smart, precocious kid and always kind of bossy. It sounds like Always a leader. You tell one anecdote in here of him in his high school getting all the water polo guys up and having them stripped down to their-.

12:29 - Parmy Olson (Guest)
Speedos.

12:29 - Leo Laporte (Host)
Speedos which he got in trouble for, but then he kind of talked his way out of, I guess.

12:34 - Parmy Olson (Guest)
He was very good at talking his way out of trouble.

12:36 - Leo Laporte (Host)
Still is, is he not? Yeah, you also talk about Sir Demis Hassibis, who is deep mined now, but that's one of the things I got from the book is everybody's moved around a lot, right? Ai is a very incestuous business. It almost feels like there really are only. In fact, somebody's told me that there are only five kind of seminal papers that were the foundation for everything we're seeing today in LLM One of them, probably being the transformer paper. Yes.

13:09 - Parmy Olson (Guest)
Attention is all you need, yeah.

13:11 - Leo Laporte (Host)
That's right, yep. So the point of that was that there aren't any secrets. There's no secret sauce, despite what some of these guys would pretend. Yeah, sauce. Despite what some of these guys would pretend, yeah. Do you think Sam Altman has the secret sauce that his next big thing is? You know, they talk about reasoning in his this, this, um.

13:35
He didn't call it a manifesto, but I think it is the intelligence age that he wrote. He uh says in a couple of decades, next couple of, we'll be able to do things that would have seemed like magic to our grandparents. A little shout out to Arthur C Clark, but I think he makes an interesting point because we benefit from society. Society makes us smarter and more capable. The infrastructure makes us smarter and more capable than any individual. He says in an an important sense, society itself is a form of advanced intelligence. He also talks about how ai is going to take us to that next level. Ai will give people tools to solve hard problems and help us add new struts to the scaffolding of human progress that we couldn't have figured out on our own. Parmy, from what you've written and from what you've studied of all this, is this threats to the scaffolding of human progress that we couldn't have figured out on our own. Parmi, from what you've written and from what you've studied of all this is this as some say, just nonsense or does it make sense?

14:36 - Parmy Olson (Guest)
I think yes and no. I think his pitch around using AI to benefit humanity and sort of lift us out of like, improve the wealth of everyone on Earth you know, sam often uses the word abundance and from the beginning he has said that if he can build artificial general intelligence, that he will help elevate the wealth of billions of people. So this is incredibly grand, almost galactic in scope ambitions for what this technology could bring. And it's interesting that he posted this last week, not too long ago, just before a number of his executives left the company, just as he's also trying to raise as much as $6.5 billion for the company. And I think what this all kind of says to me is that selling AI, you have to sell the future, and that's kind of what he's doing. He's selling this future of AI benefiting humanity.

15:40
And one of the reasons I wrote this book is when Chachi T came out. I was awestruck by it. I thought it was amazing, but I think a lot of people didn't know. The story behind the scenes was that Sam and this other guy, the founder of DeepMind, demis Hassabis ChatGPT, had kind of come as a result of their rivalry to build artificial general intelligence, but they had both started out with these incredibly altruistic goals of you know Demis wanted to cure cancer and solve climate change and Sam wanted to make everybody in the world more wealthy. But what ended up happening along the way which is what you've probably seen over you know 999 episodes, leo of doing this is mission drift right In Silicon Valley, the tech companies will often say they're trying to make the world a better place.

16:33
Hbo's Silicon Valley, very famously, the TV series made a wonderful scene making fun of that of tech startups wanting to all pitching their startups as making the world a better place, and it's an incredibly successful selling point and you can see Sam doing this yet again in this blog post, just reinforcing this notion that and he uses the word prosperity this time that AGI will bring prosperity.

17:01
But what I wanted to point out in the book is that what we've really seen is a kind of greater control taken of this technology by the world's largest tech companies, and the main stories I really wanted to point out is that both Sam and Demis tried to protect their technology from the control of these companies. That's why Sam started as a nonprofit, and there's a lot in the book about how DeepMind tried to break away from Google and become a nonprofit because the founders were concerned about where AI would be steered when it was controlled by these large companies with a growth objective and profit objective. So, yeah, I mean to answer your question. Good and bad stuff, of course, but there's always going to be a price to pay. Every tech revolution has come with a price, whether it was the internet, mobile, social media especially and I think we're definitely going to see some kind of human price come with generative AI as well.

17:59 - Leo Laporte (Host)
It's interesting because this is he's right on the cusp of turning open AI into two companies, a nonprofit and a for-profit right, and that was one of the reasons he got fired briefly, because it wasn't clear what was going on. And, in fact, meena Muradi, who took over as interim CEO when Altman was on the outs for about three seconds, just left the company, along with many of the other top leadership. What's going on there? Is that a lack of confidence in OpenAI's mission? Is it that they don't want to work for a for-profit company? Do you have any insight into that?

18:40 - Parmy Olson (Guest)
I think it's probably a mixture of a few things. So what I heard from speaking to people who were on the OpenAI board was that Mira was not comfortable with Sam's leadership. Neither was Ilya Sutskiver, who also left. He was the chief scientist he's disappeared almost yeah. He's made his own startup, which is worth. You know, raised a ton of money, called Super Safe Intelligence. It's in the name, it's in the name, it's in the name, and the joke is that maybe Mira should make another startup called Super Extra Safe. Intelligence, super Super Safe, super Califragilistic Safe.

19:16
Because the insinuation, of course, is that OpenAI is not actually making safe AI, or they don't trust Sam to do it actually making safe ai, or they don't trust sam to do it. Um, and as we know that, when he was fired, the big the quote that came out, came out from the board, was that he was not consistently candid, ie he was not telling the truth. He was, he was lying, he was or he was being manipulative and there was this.

19:37 - Leo Laporte (Host)
We never got the receipts on that, like well about what they never?

19:41 - Parmy Olson (Guest)
I think they were very uh cagey and they still are about all this yeah, and they haven't, yet they haven't been able to really give, like, uh, clear examples that I think, but I think yeah I.

19:53
I think that's partly due to to ndas and things that they signed, that obviously that stopped them from from doing that. And then what's interesting is that when mira fall after that, mira actually sent a note to the open ai staff saying kind of dismissing all of that and denying that there was any kind of big rift with Sam. But you know how true is that it's actually very hard to trust some of the public statements from these people. But the fact that there are so many have left means in a way, that Sam can sort of run open AI now as his show.

20:26 - Denise Howell (Guest)
So he can mold it how he wants.

20:28 - Leo Laporte (Host)
There is a difference between him and Demise, which is that he is not really a scientist or a technologist. He is a startup guy. He's a money raising machine.

20:38 - Parmy Olson (Guest)
Demise is a technologist right, he was formerly a neuroscientist. He wrote a PhD on memory, a very well-received PhD. He was a former game designer as well. He was a former chess champion. He is a brain. He is just like if you think of him as the brain, and he tried to build artificial intelligence and model it on the human brain. That's why he studied it. He's just been obsessed with AI his whole life and he's a very different kind of character, but also very charming and captivating and has this kind of cult-like devotion among his staff, in much the same way that Sam Altman did for OpenAI and Demis was the first entrepreneur to start a company to try and build AGI. He did that five years before Sam did and then, when Sam started OpenAI with DeepMind and then when Sam did, it turned into this really kind of bitter rivalry between the two the fight seems to be and you include Anthropic in this.

21:38 - Leo Laporte (Host)
it seems to be over safety and it's interesting, the title of your book, supremacy, is kind of intriguing is a. Is a super intelligence safe? I guess is the thing people debate, and clearly ilya suitskiver, uh, demis, um, the anthropic folks, all are worried that a super intelligence will be threatening our existence, an existential threat to humankind. Sam does not seem to think that, or does he?

22:10 - Parmy Olson (Guest)
Well, he talks about it and he's talked about it to the Senate and he's kind of made suggestive. He's paid lip service to it, yes, yes.

22:18 - Leo Laporte (Host)
It also feels like he's the guy who said $7 trillion I'm sorry with a T and we're going to have intelligence and let's go full speed ahead. Openai seems to kind of act like, well, we're just going to ingest everything and worry about it later and let's just. Let's go full speed ahead and not worry so much about safety.

22:38 - Parmy Olson (Guest)
Yeah.

22:38 - Leo Laporte (Host)
Is that not?

22:39 - Parmy Olson (Guest)
right yeah, no-transcript the world coin Iris, and that is Samuel, and it's also Elon Musk and it's also other. You know, jack Dorsey ran two companies at the same time. It's like these tech billionaires You're. You know, jack Dorsey ran two companies at the same time. It's like these tech billionaires. You're not really a true tech leader, unless you're like juggling three or four different companies, it seems. But in Sam's case it seems like he may have spread himself a little bit too thin because things are really looking out of control at OpenAI.

23:38 - Leo Laporte (Host)
Daniel, I want to include you in this because, of course, microsoft is a big partner with OpenAI, and Microsoft is probably the way most people encounter it with Microsoft's co-pilot, and they've actually changed the rules on recall. They're about to push it out, though, aren't they?

23:55 - Daniel Rubino (Guest)
Yeah, next month, October, insiders will be getting it. You know, I think this is a great example of how Microsoft sometimes just makes really boneheaded decisions. I'm referring to the initial rollout of recall, where they didn't take security as seriously as they probably should. So now they've gone back to the drawing board and it's the way. You know. I don't want to give them like a lot of credit here. It's like the way it's going to come out is the way it always should have come out.

24:21 - Leo Laporte (Host)
Right difference, that's an opt-in thing and you can turn it off. You can uninstall it now, right, yep?

24:28
yep the whole reason security experts are worried is because, uh, for those who don't remember, who don't recall, the idea was it was going to do screenshots periodically, like every 30 seconds, of what's going on and then use ai to ingest it, understand it which was interesting, in fact, a variety of ais, depending on what the screenshot was of and then use AI to ingest it, understand it which was interesting, in fact, a variety of AIs, depending on what the screenshot was of and then for you to be able to query it. Microsoft said but don't worry, it'll never leave your computer, only you will have access to it. But security experts said that's just too valuable a treasure trove, everything you've done on the computer. Bad guys, guys will go after it and it really is going to be a problem yeah, all the information was stored on encrypted uh and that's right.

25:13
Initially it wasn't. It wasn't file, it wasn't a bitlocker. It was.

25:16 - Daniel Rubino (Guest)
It was open no, anyone with top level access could technically go through another user's files. So so there's all sorts of questions coming up. Yeah, there's all sorts of questions coming up, even about like with parents and children who, like children, may be going especially with the trans stuff, right, you know, and whether or not they're exploring different ideas and could parents get access to that and it can endanger people. So now it's going to be all encrypted and a VBS enclave obviously still never leaves the device. You need to use Windows Hello, visual facial recognition to set it up in the first place, or fingerprint From there on, though you can use a pin if you want. So it's going to have all the right security stuff in place.

26:00
I still think it's an excellent tool. I think people who I get the concern over the privacy and stuff, like, for instance, right now, it will also automatically filter out credit card information, id, license numbers, passport information. It'll block that stuff automatically. So even if someone did get access to it, that stuff would be blocked. I get the safety concerns of this and of course, it should be an open discussion and they should be as transparent as possible, but at the same time, it's like when people kind of freaked out about this stuff. I'm like well, what do you think AI is going to do? Like, the whole point of making these tools is to make our lives easier, to augment the human brain, basically and this is just one specific use case of it.

26:46
But in order for AI to be successful, it is all about, at some level, surrendering information about yourself. You can't have a personal assistant that knows nothing about you. It's like hiring a personal assistant in real life and not telling them anything about what your likes are like. How are they supposed to do their job? Now, of course, a lot of people may not want that experience, and that's totally fine and they're understandable too. In which case, yeah, you don't need to opt in. If you did opt in and you don't like it, you can delete all the info. You can uninstall the entire program. So I think all these options are good, but I think this is what we want with AI at least some of us. Otherwise, I don't know what we're talking about, right.

27:31 - Leo Laporte (Host)
Are we just talking about image generators and making silly videos on, you know, social networks? Well, even image generators have to ingest huge amounts of images, and artists don't like that. Ai is nothing unless it ingests all the content we can give it, whether it's our stuff or somebody else's. Denise, do you use a chatbot or any of the AI tools?

27:52 - Denise Howell (Guest)
I tinker Like.

27:53 - Leo Laporte (Host)
I don't use it as a productivity tool in my daily workflow really we were talking before the show about and Jeff Jarvis has demonstrated this on Twig about Notebook LM. Talking before the show about and and Jeff is Jarvis has demonstrated this on twig about notebook LM, which is Google's rag based AI bot, which which rag is retrieval augmented generation, where it takes documents, you provide it and then you can query it about those documents. It's kind of what recall would also be, and Jeff Jeff took a bunch of his writings, put it in there and it generated a podcast with a male and female host who paused and you know they sounded like totally like podcasters explaining, talking about what Jeff's writings said. I said, Jeff, is it accurate? He said, yeah, it's pretty accurate. It's actually would be one way you could understand what I'm writing about, if you prefer that.

28:47 - Denise Howell (Guest)
It sounds like a cool parlor trick, but is that super useful If you're looking for one particular piece of information? You don't want to have to listen to two people doing a podcast.

28:57 - Leo Laporte (Host)
Well, wait a minute, don't say that, don't ever say that.

29:02 - Daniel Rubino (Guest)
I think it's an interesting example though, because, like right now, what we're seeing a lot of is like I know Microsoft had this a few years ago which was it would summarize your emails. But it would not even summarize, it would just kind of read you your emails, which itself can be kind of boring. But if you can add a narration to it, tell it as a story kind of thing when you throw your headphones on and you're commuting in the morning. That's kind of the point of this.

29:22 - Leo Laporte (Host)
You're trying to yeah, and Apple has promised, with Apple Intelligence, to do that if it ever comes out next year. Yeah, I guess that's something people would like. You used two words that I've always thought kind of described AI, which is parlor trick. Denise, it often feels like, especially with the chatbots it's just a trick, it's just showing something that doesn't seem that useful. Having Scarlett Johansson's voice, by the way, I just got the new Meta voices and one of them definitely sounds like Scarlettohansson. So I don't think meta has made a deal with uh, scar joe, but they have made deals with a lot of um, famous uh voices like uh, I think they just announced judy dench.

30:12
And yeah, judy dench is on there yeah, and cena and stuff that's a big get that's an expensive get, but they're, they're you know I. So this ties into the event this week, the MetaConnect event that Mark Zuckerberg came on talked about what they're doing. He calls Lama their large language model, open source AI. I don't think that's actually an accurate way to describe it, but it is more open in the sense that you can use it right. You can download it. In fact, you can download the models and run them locally. Do you talk about I in the book? I don't think you talk much about meta and their AI efforts um, I don't.

30:53 - Parmy Olson (Guest)
I think that that requires. That would probably require like a couple of chapters because open the whole open source effort in AI is so interesting.

31:02 - Leo Laporte (Host)
What is open source when it comes to AI, I mean? What does that mean?

31:07 - Parmy Olson (Guest)
Well, it means that you don't have to. You don't have. It's in the same way that open source software works for, although, as you said, like is it actually open source? I think there's one of the main open source associations actually said that Lama is not strictly open source, and so the phrase that people are using is open, not in the traditional sense of open source. Anyway, yeah, apparently the most accurate way to describe it is open weights, but it's just you know.

31:33 - Leo Laporte (Host)
I suppose for a lot of-. So they tell you what the weights are that they're using.

31:36 - Parmy Olson (Guest)
And it's a cheaper way for businesses to use generative AI rather than have to pay a ton of money for API access to OpenAI's GPT, latest GPT model or Gemini from Google or even Anthropix. Claude, you don't have to, you know it's just a much cheaper option and that you've got similar efforts coming from France, from Mistral there's a lot. If you go to Hugging Face, that's like a real hub for all the different open source models that are currently available.

32:10 - Leo Laporte (Host)
Yeah, and I've played with. There are a number of tools you can download, like Ollama and anythingai, that you can download models from Hugging Face or other sources and try them. And you know, no model is perfect. All the models are good at some things and not good at other things. It's funny because Mira Marotti was one of the people who was on stage sitting there with ChatGPT 4.0. She must have liked OpenAI at that time. I don't know what that was only a couple of months ago. This stuff is moving so darn fast. That was only a couple of months ago. This stuff is moving so darn fast.

32:45 - Denise Howell (Guest)
I'm looking forward to reading your book, parmi, and can you give us a preview, because we haven't touched on? We've touched on sort of the hand-wringing of the former board members at OpenAI saying we're not being super safe enough and what dangers we know are there, what guardrails legitimate companies are going to put on their products. But then there's the dark web and there's open source or whatever.

33:15 - Leo Laporte (Host)
Yeah, and there's Russia, I know.

33:17 - Denise Howell (Guest)
There's going to be so much out there that is so evil and so good at say here's an analogy. Evil and so good at say here's an analogy.

33:31 - Parmy Olson (Guest)
Um, we're going to talk probably later on this show about uh, do not pay and its founder um oh yeah yeah, that's right up your alley.

33:36 - Denise Howell (Guest)
He just got smacked right and and some of these, some of these early folks in the legal space like Josh. There's another guy named Andrew Arruda whose latest product is going to go through judicial opinions and try and profile judges, give lawyers intelligence about judges' tendencies, and people are constantly trying to read the tea leaves as to their judges and AI could help you do that constantly trying to read the tea leaves as to their judges and AI could help you do that. If you put in everything you know about this person and every legal decision they've ever written or made, you could probably gain some insights. If you fed an AI with information that was strictly like grooming techniques for underaged minors to be put into terrible illegal situations, you could probably really fine tune things to be very effective, you know.

34:33 - Leo Laporte (Host)
I just came up with. See, I've always wondered what's AI safety mean. By the way breaking news this just happened as we begin the show. Gavin Newsom has vetoed the ai safety bill and that was gonna happen yeah, I mean, even nancy pelosi was saying, oh, you can't say yes to this, um, but this was an ai safe, this was an attempt to make ai safer. Why was it always going to happen parmi?

34:59 - Parmy Olson (Guest)
oh sorry, I just when I was um, when I took a trip to silicon valley and talked to a lot of people in the industry there. On the policy side, and they said they hate it.

35:08
Not only that, but also they believe that it's politically damaging for Newsom if he signed it, for his potential chances if he ever wants to run for president in the future. So because it's had pushback from Democrats as well from other Democrats, so just there was this kind of acceptance that it wasn't going to happen. He did sign those other bills, right, but just not this one that everybody hated.

35:32 - Leo Laporte (Host)
He's been busy. We have a whole host of bills, including banning cell phones at California schools. We'll get to all of that, he said when he vetoed the AI safety legislation. It doesn't take into account whether an AI system is deployed in high risk environments, involves critical decision making or the use of sensitive data. Remember this was a bill that was tasking companies creating AIs with proactively making sure that they were safe, he said. Instead, the bill applies stringent standards to even the most basic functions. So, as long as a large system deploys it, I do not believe this is the best approach to protecting the public from real threats posed by the technology. But it was.

36:12 - Parmy Olson (Guest)
It was politics, ultimately, that that tanked this uh bill um yeah musk supported it he was probably the only person in that support Silicon.

36:23 - Leo Laporte (Host)
Valley liked it.

36:33 - Denise Howell (Guest)
My point in concern, I think, is that we can impose all the regulations that we want you know, this bill or some other but only the good guys are going to follow the regulations and the bad guys are going to do what they want.

36:46 - Leo Laporte (Host)
So, but does that mean we shouldn't have any regulation at all in order to keep up with China and the dark web and and whoever else is is doing this? At the beginning part of me, we were talking about the fact that that there isn't, there aren't really any secrets in creating LLMs and transformers, because all of these papers have been published openly and and everybody kind of understands how it works. Um, so there's nothing to stop bad guys or a nation state from going full speed ahead with unsafe whatever that is unsafe AI.

37:13 - Parmy Olson (Guest)
Yeah, if they want to pay for the access or if they want to find a way to use a bit of corporate espionage and steal the secrets. I mean these companies have become more secretive, DeepMind has stopped releasing research in the way that it did just a year ago.

37:29 - Leo Laporte (Host)
Kevin Kennedy 1, Even OpenAI which was supposed to be open, kim Kido 1.

37:32 - Parmy Olson (Guest)
Oh, the opposite, kevin.

37:33 - Leo Laporte (Host)
Kennedy 1. Quite the opposite. As soon as they started making some headway, they said oh yeah, we're not going to tell you how we're doing this, but China has been at this for decades and they've been using AI for things like social punishment and you know. So it's. I mean, I presume that China has all of the capabilities we have, and then some the Chinese government.

37:56 - Parmy Olson (Guest)
I think they have very strong capabilities, but in different areas, so for example, with facial recognition and vision recognition. China is incredibly strong because there's demand for it.

38:06
For the surveillance you know the state surveillance network Also. You know large language models. Not so much, but there's also a lot of. There's been a lot of government subsidies for companies out there, and building AI in China has actually been a lot cheaper than it has been in the US and in the west, partly for that reason, and the cost of actually accessing this technology has been low such a big topic because we're, so we're.

38:30 - Leo Laporte (Host)
There's also an attempt to keep them from having the chips. For instance, the nvidia chips video in fact had to make dumbed down chips for the chinese market because the chips act prevents them from sharing their best technology with Chinese government. It's very, very complicated. We got to take a break, but I appreciate. I'm so glad to have you on, parmi, to talk about this. I think everybody should read the book supremacy Because it'll give you a good grounding and what's going on and we have lots more to talk about, not just AI and just a little bit. Parmi Olson writes for Bloomberg. Her last book was we Are Anonymous. Was that about the anonymous movement?

39:11 - Parmy Olson (Guest)
It was. It was about the hacktivist collective and the splinter group of hackers that wrought havoc across the internet.

39:17 - Leo Laporte (Host)
Do you think they're going to have? I would imagine the same activists would have something to say about AI. I think it's only a matter of time before people start to sabotage AI. I know some artists have done that with their work. They've put stuff in the art that supposedly confuses AI.

39:36 - Parmy Olson (Guest)
Yeah, and some people are taking advantage of it. I could totally imagine people from Anonymous finding ways to exploit Because that's what they do.

39:43
They're very opportunistic people, yeah Well, they certainly were 10 years ago and they're probably finding ways to use generative ai, you know to, to social engineer people. Um yeah, and artists are. Artists are exploiting it too. I know there's a damien hurst who's like an extremely well-known and wealthy artist in the uk, has been generating paintings using ai and selling them and he's made millions of dollars as a result. So you know, there's a ton of artists out there who may have had their work ripped off by AI because it's been kind of just sucked into the training data, but then there are others, especially the well-established ones, who have found ways to actually capitalize on it.

40:23 - Leo Laporte (Host)
Do you think if you were a smart artist, you would adopt it and say, oh, this is the future.

40:30 - Parmy Olson (Guest)
It depends how you define smart, because would you sell your soul to make a smart decision and use AI I don't know that gets a little bit philosophical about whether, as an artist, whether you would use AI to create your work, and where you draw the line between the control you have over what you're creating versus where you let the computer step in and create something with you or for you.

40:53 - Leo Laporte (Host)
Right. That's what's so interesting about this whole thing. There's no agreement. No one knows what's going on. It's very confusing to us us mere mortals, trying to understand any of this, but I'm so glad we got Parmi Olson to help us out. Of course, Daniel Rubino, editor-in-chief of Windows Central Lots to talk about in just a bit. In fact, I want to talk more about Copilot and Denise Howell, my personal attorney. No, she's not, but she's my internet attorney. That's better Internet attorney. She's not responsible for all the stupid things I do, let's put it that way. She's just responsible for trying to talk me out of them and failing. Every once in a while I get an email from Denise saying you know, you really shouldn't be doing that.

41:40 - Denise Howell (Guest)
I would never. We've never had an actual attorney-client relationship.

41:44 - Leo Laporte (Host)
No, you have no privilege. When I started a Mastodon instance, she told me all the horrible things that could happen to me Every one of them. And did I pay attention? No, I went full speed ahead. Of course, because that's just me. Anyway, great to have all three of you.

41:57
Our show today brought to you by Lookout Couldn't be more timely. Really, these days, every company is in the business of managing data. Data is the core of it all. Right, that means every company is the risk, increased risk of data exposure and loss. You see about seeing the headlines every day cyber threats, breaches there's a, I get a. I get mail now from companies say, oh yeah, we lost all your information, there's been another breach, constantly leaks. And cyber criminals are not, you know, taking it easy. They're getting more sophisticated every day.

42:30
Modern breaches now happen in minutes before you could, even you know, respond. You're done at a time when the majority of sensitive corporate data has moved to the cloud, traditional boundaries for your data no longer exist and the strategies for securing that data have fundamentally changed. That's why you need Lookout, from the first phishing text to the final data grab. Lookout stops modern breaches as swiftly as they unfold, whether on a device in the cloud across networks or working remotely at the local coffee shop. You've got an employee there, so is your data right? Lookout gives you clear visibility into all your data, at rest and in motion. You'll monitor, assess and protect without sacrificing productivity and employee satisfaction for security With a single, unified cloud platform. By the way, you will love it. Lookout simplifies and strengthens. Reimagines security for the world. That's coming today. Visit lookoutcom right now and learn how to safeguard your data, secure hybrid work, reduce IT complexity.

43:35
Lookoutcom. We thank them so much for supporting this week in tech. We really appreciate it. You support us when you go there and if ask you, you say, oh yeah, I, I saw it on twit. Thank you, look out. Uh, back to the show we go. We're talking about ai. We're talking about all sorts of things. What's the uh current state of the co-pilot key? I think daniel microsoft has now said you can. You can change what it does yeah, I mean technically.

44:06 - Daniel Rubino (Guest)
I mean technically, you could always remap any key in Windows.

44:08 - Leo Laporte (Host)
I guess you can with the third-party software, right, but now it's in the settings, right.

44:13 - Daniel Rubino (Guest)
Yeah, now it's going to be in settings, which is fine. I mean, I still think it's a good move in terms of just trying to get people to use it more. But even though I use Copilot quite frequently, even I don't use the key as much oh, copilot quite frequently. Even I don't use the key as much, oh, yeah. Yeah, I actually subscribe to pro because I find it useful. Actually, in bird, when I am writing, I do like how you can also use it to change the tone sometimes of what you're trying to write and get across. So there are some valuable like aspects to it and I'm looking forward to getting you know.

44:41
Creating slides I think it's going to be someone who hasn't done PowerPoint in a very long time very much looking forward to AI being able to build slides based on input. You know that direction I give it, so I still find it quite useful. I don't know if it's actually worth $20 a month that's pretty expensive but the rumor is that some big news is going to be coming out with Copilot soon. You know you've been following Copilot. They've been pretty quiet on it the last few months. Actually, there haven't been a lot of announcements around it, um, and so we're starting to hear rumblings that there's going to be a lot more coming out pretty soon on that front.

45:16 - Leo Laporte (Host)
So is it as far as gpt 4-0 that copilot uses, or do we know what model it's using, or yeah, it uses 4.0.

45:23 - Daniel Rubino (Guest)
I think the pro version uses a little bit more of an advanced one. Um, okay, with the pro, you do get the the latest and whatever the fastest one is out there, plus you get the priority on the server. I will say I mean I was just at Samsung's event in New York City on um Thursday where they announced the new uh apps. Yeah, the Galaxy s tab, sorry, tab s10. Uh, the new keyboard has a ai button. No, oh man, yeah for galaxy ai, which you know. We all know samsung's pushing galaxy ai pretty hard and they just did the same thing. They created their own keyboard. So apparently they like the idea. Yeah, I mean, I could see everybody kind of doing this right, because this idea of an AI agent on your system, I think, is going to be one of the most critical things in our technology, and Siri is going to do that too for Apple. So I see this as a trend. I'll continue.

46:18 - Leo Laporte (Host)
Stevie Batiste at I guess it was at Ignite last year really kind of catalyzed, I thought, in a very nice way the the, the different roles ai would play over time in your computer, from like a sidebar, which is kind of where copilot is, to something much more integrated into your daily operation and Microsoft's moving full speed ahead with that. Is that the case?

46:48 - Daniel Rubino (Guest)
Yeah, definitely. I know some stuff that's coming.

46:53 - Leo Laporte (Host)
You can tell us it's okay, nobody's listening.

46:55 - Daniel Rubino (Guest)
Just you and me. You know this. I think this stuff goes. It's pretty obvious where it goes. I mean where Apple's taking with Siri. You know this idea of you know they tried it a couple of years ago right With Cortana, and Cortana was famous. It was cool, yeah, yeah, it was a cool idea. It was often faster and better than what Siri was at the time. But at the end of the day, these are still systems.

47:20 - Leo Laporte (Host)
Comparing it to Siri is not exactly the highest bar.

47:25 - Daniel Rubino (Guest)
But like none of this was truly ai right, there was no like an actual agent. There was pre-recorded things and it wasn't very smart. Now we're getting to this like fuzzy logic thing where you could just kind of tell it roughly what you want and it'll figure it out even in the most rough context.

47:40 - Leo Laporte (Host)
So I feel like richard campbell is listening. He said stevie's presentation was at build 2023.

47:45 - Daniel Rubino (Guest)
That's right, I remember it seems a great guy. I've met him quite a few times. He's so smart, very smart, very down to earth too, just uh.

47:53 - Leo Laporte (Host)
Yeah, he's doing some really interesting things over at microsoft and he's been able, in a way I guess he's uh, he's like a fellow, he's in applied sciences, yeah, you know above like any other role, but he has, in a way that I think others have not been able to kind of explain what the role of AI is in how people work and what Microsoft's vision is for incorporating that in. And that, to me, is a much better response to AI safety than a bunch of hand-waving saying gonna you know, we're gonna make sure it's safe. Um, talk about the role it's going to play, how it's going to integrate in, what data it's going to use and what user controls there will be.

48:34 - Daniel Rubino (Guest)
That's much more concrete to me and yeah, they're always been trying to get ahead of this stuff, for the simple reason is that they don't want government to come in. It's not that they're against regulation, they just don't want dumb regulation on this and they don't want to get to. They don't want government to come in. It's not that they're against regulation, they just don't want dumb regulation on this and they don't want to get to the. They don't want a bunch of companies doing a bunch of stuff, having massive blowback government step in and doing you know kind of like what this bill was going to be in California, right? Maybe an overreach or kind of limited technology. So Microsoft tries to get ahead of this stuff by being kind of open about what they're doing, transparency, putting everything they want on websites so you can go read what they're doing, in a way to curb the government from doing this. I think ultimately, they do support legislation, but they just want it smartly done, and that's a tough thing with this technology sense. Even all of us have a tough time struggling trying to understand it.

49:24
You mentioned about AI safety. What does that even mean, right? Right, uh, it's. It's a very weird question because right now we're talking about large language models mostly, which are. They're cool, they can do neat things. They're foundational for a lot of other aspects that may come for from for ai. Then you have, like, image generators and video generators, but like there's no, this idea of, like you know, agi is like what everybody is fearful of. I personally don't think we're anywhere near agi. I think we're still because I don't think we've even defined the problems and we know what to solve there general intelligence.

49:58 - Leo Laporte (Host)
That's what supremacy is right parmi is. Is a general uh like smarter than humans?

50:05 - Parmy Olson (Guest)
yeah, the question is about the quest to build agi. Yeah, yeah, which is this very abstract? But you know, and and it's interesting when tech companies talk about, or what the likes of sam or or demis or others talk about, how many years it will take to get there, um it just, it's almost like a carrot being dangled in front of us, like oh it's coming.

50:25 - Leo Laporte (Host)
Salmon Manifesto said in some number of thousands of days Thousands of days.

50:31 - Parmy Olson (Guest)
It kind of reminded me of when someone puts 99 cents on a product instead of $1, just to make it sound like it's sooner than it actually is.

50:39 - Leo Laporte (Host)
If you want to do the math, a thousand days is about three years, so he could be talking a decade out. But it sounds closer if you say thousands of days and of course Ray Kurzweil has been saying it's just a couple of years out for decades.

50:54 - Parmy Olson (Guest)
For decades now, and it's always getting closer and closer Every minute. Singularity is nearer.

51:00 - Daniel Rubino (Guest)
I do think it's interesting the idea of where we are with large language models, because you can feed it so much information and we have so much computing power these days. There really is a lot they can do. Like you know, the idea of like it figuring out potential cures for diseases is super fascinating, right, because it could just do so much data and modeling that humans couldn't do. But I think, you know, this question of AGI does become a little bit more interesting when you start to bring in. Can AI as we know it today, with these large language models and advanced systems, kind of figure its own self out on how to create AGI, right? Is it going to lead us down that path? For us, I think, is an interesting thing, because it's going to take all the world's knowledge and being able to model that right knowledge and being able to model that Right.

51:48 - Leo Laporte (Host)
So I saw just the other day I saw somebody who works in a medical lab saying AI is not going to cure disease, but what AI might do and help us cure disease is tell us what alleys to investigate, what experiments to do, and over and over again. That's kind of what I see people say is that AI is useful hand in hand with a human, like there's things we bring to the table that AI cannot, but an AI can be useful for a writer to inspire, for a researcher to kind of synopsize a bunch of stuff, but ultimately the human brain does something qualitatively differently to the material. So it's a good hand in hand thing. When I think of AI safety, I worry more about AI and weapons and companies like Anduril and Palantir companies. We're seeing it right now in the Ukraine-Russia war. It's a test tube for AI and and autonomous weaponry and it worries the hell out of me. That's a lot more scary than a super intelligence than you know.

52:55 - Parmy Olson (Guest)
The forbin project, um yeah well, that's something that deep sorry go ahead go ahead sorry yeah, go ahead for me oh, I was just going to quickly say that deep mind that was one of their big red lines was not using their AI to be used for military purposes, because they knew that Google had this Project Maven contract with the military, with the Pentagon, a number of years ago. But it was one of the reasons why they wanted to break away from Google and become an independent kind of pseudo nonprofit organization was to protect. When they eventually made AGI, they didn't want Google to have control of it. But they failed and Google ended up taking control of DeepMind more closely.

53:41 - Denise Howell (Guest)
Sorry, Denise, that's all right. What I was going to ask is and I see this in Sam Altman's latest manifesto- you can call it a manifesto. I like that. Yeah, and elsewhere just in the discourse, leo, you were just saying how you know, right now we're sort of working hand in hand and the humans have something to bring to the table, but of course nobody talks about the next step anymore, which is artificial super intelligence.

54:10 - Leo Laporte (Host)
And you know Do you think we're going to get there, denise?

54:12 - Denise Howell (Guest)
I mean, I'm not the person to ask that question of you, know Well I don't know anybody who knows better.

54:17 - Leo Laporte (Host)
This is the problem. You can ask 12 different AI scientists and you're going to get 12 different answers, because nobody really knows if we can make that final leap into intelligence. There are people who believe and I don't know what I think, but there are people who believe, like, uh, steve gibson security now that humans are just mechanisms, that we're just deterministic mechanisms, we're just, uh, like a computer. That's really, really good, and that there's nothing to distinguish what AI is doing from what a human is doing. And eventually AI will get fast enough, big enough, whatever, to duplicate what we do. So there's nothing special about what we do.

54:56
And then there are people and Ray Kurzweil, by the way, is one of those people who believes that. Then there are people who think, no, there's something different that a human does that a von Neumann device cannot duplicate. There is some magical quality that humans have that machines can't have. So far. Those people are right. Right, I mean, the art, the writing, the stuff AI generates is mediocre in the extreme. It is not think, but I think that's just a matter of time?

55:29 - Daniel Rubino (Guest)
I think so, so you fall in that first camp you think eventually it's going to catch up I, yeah, and I think, um, well, so part of the problem here with agi is the simple fact that we don't even know what consciousness is. We haven't defined it, uh, we don't know how to study it. Uh, we don't know what it is necessarily. We know, okay, it it comes from the human brain, but we don't know you know its exact description and how it and why it's here, like how it actually happens, right? So there's all sorts of weird theories there about that. So when it comes to like this super intelligence, the thing is it's like deterring effect, right, where if I give you, uh, a computer and it simulates a human being, you may think it acts like a human being and therefore it is. But, uh, you know, is it really.

56:13
That's why the turing test is a crap test, because you can do that yeah, because tricking people is super easy so if you have, yeah, so if you have like ai and it's like, oh, it seems like it has consciousness, Like we don't know even how to test for consciousness or like how to define it, so it comes a really strange question. But I'm in that camp of you know, I've been reading Sapolsky's book determined, and you know which he argues.

56:40
I love him, by the way his stuff and I'm I'm just in that camp like sorry, we are just kind of machines and cogs. We do have a lot more going on in our brain, but at the same time it's ultimately just we're to some of our experiences.

56:52 - Parmy Olson (Guest)
Yeah and I think that's how sam altman and demis hasabis also view humans is that we can. Our brains can effectively be replicated by machines. Machines are extensions of our brains and they are being modeled on our brains. That's how we're kind of building them, based on the brain as an inspiration. But like Leo, when you were saying earlier that we'll always need to have a human, you know working with a machine and have some element of the human in, I also question that, Like in 10, 15, 20 years.

57:20 - Leo Laporte (Host)
Will we need that input?

57:22 - Parmy Olson (Guest)
Yeah, we need that input. Yeah, and I don't know that that necessarily means that computers will be conscious, necessarily. But we're going to outsource so much more cognitive labor to these things over the next 10, 20 years just because we can, because it's easier, because we're going to save money from doing it. Companies are going to make a ton of money from doing that, so why not? So they will, and then our definition of consciousness maybe will change. That's kind of semantics in a way, but I think that's where these things will become powerful and that's just going to that's. I see that as inevitable over the next decade or so.

57:57 - Daniel Rubino (Guest)
Just two things too. This is why I always find weird about Elon Musk, where he's, at the same time, talking about AI, all the time started his own company of course, talks about the fear of it, and then he's also going off about human population decline all the time. We all got to make tons of babies, but all the arguments for this kind of AI is exactly what you're talking about, which is we're going to use it to offload a lot of jobs, actually, and it's going to be the economies with declining populations who will benefit the most from this, since they'll be able to offload a lot of that work to computers, to AI, to robots, the very things he's building. But some reason he thinks that I don't know, he hasn't made that connection yet, where everyone else kind of has. The other thing I would say is there's a great movie from 1970. It's a very. Not many people have ever seen it.

58:46 - Leo Laporte (Host)
It's a British one called Colossus the four, the forbin project.

58:47 - Daniel Rubino (Guest)
I referred to it earlier. I love that movie. Oh okay, yeah, that movie is fantastic writing this down yeah, oh yeah, because, like, what's funny about that, though? When we talk about ai, it's like the movie starts off. They're like turned on the computer. They're like there, it is connected to every single infrastructure in the united states, including our weapons. And you're like there, it is connected to every single infrastructure in the United States, including our weapons. And you're like what are?

59:07 - Leo Laporte (Host)
you doing. That seems a bad idea, right, yeah. So, like you know, maybe Just on the surface of it.

59:12 - Daniel Rubino (Guest)
If you do create a supercomputer, an AGI, maybe the first thing you shouldn't do is connect it to your entire infrastructure. Yeah.

59:22 - Leo Laporte (Host)
But see that's the thing that's scary. That's where I'm more worried about ai safety. Uh, that's what's happening in ukraine right now, right um kind of right there. So they just attached a robot dog to a drone so they can fly it into russia and drop it in russia.

59:40 - Parmy Olson (Guest)
I can't imagine anything more terrifying but I think those kinds of I, if you want to continue to go down this route, like I think that's one element of AI safety. But then there's also a more subtle, nuanced consequence that's squishier and harder to foresee, which is kind of like the consequences we got from social media, for instance. Like who would have guessed, with social media, which came to us to connect the world, that was Markuckerberg's big mo, um, that it would also have these side effects where, um, there would be misinformation, a lot of teams would find their their mental health hampered, especially young girls on instagram. Polarization, political polarization, filter bubbles, all those things. But it's also useful.

01:00:25
It's not like we're going to shut social media, all those things, but it's also useful. It's not like we're going to shut social media down. It's a utility, it's practically our social, it's an infrastructure for us. So it's very difficult to regulate. But it does also have these kind of real world human consequences and that's also an element of AI safety that I'm concerned about. That we'll have similar kinds of erosions of various things that we value as humans, whether it's our relationships or our critical thinking skills, or our creativity, the value of human creativity. I think those kinds of values will start to get eroded as we rely more and more on generative AI models and all the amazing things they can do over the next decade Again, I don't see this happening in the next year or two, but really over the next decade or so.

01:01:14 - Denise Howell (Guest)
I'm 100% on board with that and I'm going to make a groundbreaking prediction right here that Gen X is going to be the one that sells its soul, because so many things become so convenient.

01:01:29
Yeah, we don't have one anyway, but so many things you know that bug us are going to get taken off our plate If, for example, copilot can stop my dad from calling me every time he can't find the right draft of the document that he's looking for on his computer. And so you've got it coming down from that side, because our parents are young enough to use computers, but not old enough to really get them, or not young enough to really get them Worst combination Right, exactly.

01:02:01
And then on the other side, there are kids who are constantly in need of help and support. And if they would just do the homework literally the homework, or the research or whatever themselves, they could figure it out. But no, they always come to mom and dad.

01:02:15 - Leo Laporte (Host)
So here's the question for you all. Let's say we knew 10, 20 years ago that social media was going to cause these problems. What would we have done differently? What's the what's the model for which we should do now with ai that we could have done 20 years ago with the advent of the internet and social networks?

01:02:35 - Parmy Olson (Guest)
well, we know these algorithms.

01:02:39 - Leo Laporte (Host)
But would we have done that?

01:02:40 - Daniel Rubino (Guest)
well, no algorithms, let's not use algorithms no, open source, open source them, or yeah, sorry, go ahead.

01:02:46 - Parmy Olson (Guest)
So by that do you mean like more transparency for like third party auditing, by whether it's like an auditing firm or a regulatory agency. That's really like if, in an ideal world, that really had the expertise to look at algorithms and say, okay, look, you guys have started optimizing for engagement. What effect is that going to have, now that your news feed is going to start putting all the outrage posts, outrage-inducing posts, at the top of the feed? But yeah, so I think I agree, transparency would be.

01:03:27 - Leo Laporte (Host)
We could have. By the way, we could have known that, because we've seen that happen with, for instance, broadcast television, which went from from uh, something that was going to transform society to a vast wasteland simply by pandering to what people wanted uh, their own kind of slow algorithms. So we got fast algorithms with social media and got the same thing happened much more quickly, but I don't know if we have the will to to intervene.

01:03:43 - Daniel Rubino (Guest)
If we would have had the will then, or if we have the will to to intervene, if we would have had the will then, or if we have the will now to intervene in these things or whether we would intervene even intelligently true, I mean, but, like you know, we did have the idea of like uh, the idea of moderation, right, and that if you own a property, if you own a website, it could have required that huh yeah, you, you know you're entitled to moderate that network, and according to what. But now there's blowback against that, where people are conflating it with. You know free speech somebody in our discord.

01:04:16 - Leo Laporte (Host)
Santa barbarian, says we now know that section 230's reach was too broad.

01:04:21 - Daniel Rubino (Guest)
I would disagree with that, but yeah, I mean that's one of those people like if you, if you own a website, or if Twitter is a website, if you own that and you set up the rules and people agree to your terms of service, you're letting them into your property. You can do whatever you want in terms of regulation and moderation. So I don't understand how people are conflating this idea that, like well, when a social network gets to a certain size arbitrarily, we won't even define that it now becomes a public commodity, where it's now under different rules, rules and we don't need to moderate it and do all these different things. I don't know.

01:04:58 - Leo Laporte (Host)
It's who are you going to trust Anti-capitalist method? Who are?

01:05:01 - Daniel Rubino (Guest)
you going to?

01:05:02 - Leo Laporte (Host)
trust to decide what's OK and what's not OK, and that doesn't, at least in the united states. That changes every four years. So I I don't in. In the uk changes even faster, doesn't it bar me? So I don't. I don't really know who you're going to trust.

01:05:22 - Daniel Rubino (Guest)
Who's who you're going to ask where you have private networks, right, but like so, if you know the idea is it give you a facebook. It moderates a certain way. It's supposed to deliver an experience that people enjoy and go to for that experience, right. If they don't enjoy that experience, a competing website will come out and offer something. You're saying the free market. The free market, yeah, because you have extremes.

01:05:44
Or you have like 4chan, which is like we're not going to really moderate. And then, of course, they did do some moderation, right, you know. And but then you have other companies who then host them and are like well, we're going to express our own moderation and we're not going to host you, right, I mean, this is technically how this is all supposed to work, but for some reason, everybody got sensitive feelings all of a sudden and it's like, well, that private company can't actually tell its users what it can and can't do. It's like since when, I don't know, just because it's virtual, if you walk into someone's store and they start behaving erratically and you're like could you please leave? Who is someone going to come in and be like no, you can't tell them to leave it's a public space.

01:06:22 - Leo Laporte (Host)
It's like this is an age-old conversation between the invisible hand of the free market uh versus every other thing. We've thought of planned economies and we've never come up with a good solution, but the free market and capitalism seems the best of the worst solutions. I don't know if we've ever solved this.

01:06:46 - Denise Howell (Guest)
Right. And then you can throw in government's role, which you've never been able to. Your point about Section 230, leo, to do illegal things on Facebook, right, because Section 230 doesn't shield them from that.

01:07:01 - Leo Laporte (Host)
That's appropriate, yes, right.

01:07:02 - Denise Howell (Guest)
But then it becomes a question of enforcement and the government has to step in and say, hey, look at all those terrorists. Let's, let's have you do something about them.

01:07:12 - Leo Laporte (Host)
And we know I mean Facebook was very instrumental in the genocide in Myanmar. I mean we've seen these platforms be used, despite every law against it, in horrible ways. So I don't, I don't, I don't know what the answer is.

01:07:27 - Daniel Rubino (Guest)
I just and I think that's what a lot of governments are fearful of right now this social engineering they're losing control. You're losing control. No one knows what to do. It reminds me of early COVID. People always are in a hindsight now like, oh, we screwed up, they overreach yeah.

01:07:42 - Salt Hank (Guest)
Yeah.

01:07:46 - Daniel Rubino (Guest)
But it's easy to know that happened.

01:07:47 - Leo Laporte (Host)
No one knew what was going on, like we used to wipe down all our products, because we don't know, maybe we don't know, maybe that's how you get it right, yeah but that's the same thing with you know how we're over.

01:07:53 - Daniel Rubino (Guest)
We may be overreacting on social networks and regulation and government steps in, tells facebook you, you know you got to take that down. Yeah, it may be wrong, but it's because governments are are worried about losing control and it's a it's a real concern. I don't know what the solution is and we're going to get some things wrong, but there is going to have to be this kind of give and pull, you know, like give and take when it comes to this stuff.

01:08:15 - Leo Laporte (Host)
We're going to take a little break. Come back with more. This is a fascinating conversation. We've had it before. I don't know if we've ever come to a solution on it, but it is a very interesting world we are living in, isn't it?

01:08:29
Parmi Olson is here. She's the author of a new book called Supremacy, ai, chat, gpt and the Race that Will Change the World. Wow, just came out. It's available right now. Uh, from uh, st Martin's Press. Is that right? Let me look. Yes, st Martin's Press. Is that right? Let me look. Yes, st Martin's Press. Thank you for being here, parmi. We appreciate it. And thank you, isla, for letting us use her bedroom. Daniel Rubino is also here. Editor-in-chief of Windows Central. Great to have you. I like having smart people on, because this is such a challenging conversation and such a challenging world we're growing up in, uh, we're living in, our kids are growing up in that uh, it feels like we should be able to come to a solution and yet it's so difficult, so challenging. And denise howell, uh, who is a parent like me, who is looking at our kids and saying what do we do, what? What a world, what a world. How's your, how's your son doing?

01:09:25 - Denise Howell (Guest)
uh, he's great he was actually home from college this weekend.

01:09:28 - Leo Laporte (Host)
Is he liking it?

01:09:29 - Denise Howell (Guest)
Yeah, well, you know, everybody likes some aspect of college and is challenged by others, and he is having very much that experience, but he goes to school in San Diego, so it can't be all bad.

01:09:41 - Leo Laporte (Host)
No.

01:09:41 - Denise Howell (Guest)
No.

01:09:42 - Leo Laporte (Host)
You can surf and baby, yes, yes. Our show today we're so happy to have them as a sponsor is brought to you by the great folks at one password. I know, you know the name, uh, I. One password has found an interesting solution to a problem that is very common in business. Let me ask you a question it's kind of rhetorical question Do your end users always work on company-owned devices and only use IT-approved apps? No, of course not. So how do you keep your company's data safe when it's sitting on all those unmanaged apps, those BYOD devices?

01:10:20
1password has an answer to that question. They call it extended access management. This is more than password management. This is really, really a good idea. 1password extended access management helps you secure every sign in for every app on every device. It solves problems traditional IAM and MDM just can't touch.

01:10:42
The best way to think of this is, as your company's security is like the quad of a college campus somewhere maybe that your son might be Denise with a beautiful green lawn, brick buildings, ivy covered and, of course, nice brick paths leading from building to building. Those would be the company-owned devices, the IT-approved apps, the managed employee identities. Everything's perfect there, right? And then there are the paths. People actually use the shortcuts worn through the grass, the actual straightest line from building A to building B. The real world, those are the unmanaged devices, the shadow IT apps, the non-employee identities like contractors in your network. Most security tools just assume oh, you got happy brick paths, everything's going to be fine, we're going to only work there. But really, where do the security problems take place? They take place on the shortcuts, on the pads that users are wearing themselves into the green sward One password. Extended access management is the first security solution that takes all those unmanaged devices, all those apps, and puts them under your control. All those identities. It ensures that every user credential is strong and protected, every device is known and healthy and every app is visible. It's security for the way we work today.

01:12:02
It is now generally available to companies with okta and microsoftra and they're in beta for Google Workspace customers. So if you're using Okta for authentication, it takes the next step. Make sure those devices are also authenticated. It's a really a great idea. Check it out 1passwordcom slash twit, again for Okta and Entra, and in beta now for Google Workspace. It's the number one P-A-S-S-W-O-R-D dot com slash twit. What a great idea. This is 1passwordcom slash twit. We thank them for their support of this Week in Tech and your support too. When you go to that address, make sure you do 1passwordcom slash twit.

01:12:44
We were talking about image, image generation, artists using that. Uh, of course, there's been a lot of artists, filmmakers, upset with uh stable diffusion and its ability to create not just stills now, but video. Uh, the company behind stable diffusion, stability ai, has just uh named james cameron to their board of directors. That's so. That is wild. The great filmmaker who has always been, I have to say, on the cutting edge of technology, director of avatar, the avatar movies and the terminator movies in titanic. Uh, this is obviously good pr for a stability ai right uh, and it means, I think, that hollywood is going to kind of embrace the idea, and I can imagine this is the case for of of ai generated short clips and films, coverage, uh, transitions, things you forgot to shoot. The ai we've seen it can do a pretty amazing job.

01:13:48 - Daniel Rubino (Guest)
I went down a rabbit hole on YouTube where people are taking clips from movies and then using AI to rewrite part of the scene. Wow, so you'll watch the scene. It's like really playing. At some point, ai steps in and completely changes the outcome of it.

01:14:05 - Leo Laporte (Host)
How do I search for that? That sounds wild.

01:14:08 - Daniel Rubino (Guest)
Yeah, I think it's like AI.

01:14:13 - Parmy Olson (Guest)
Yeah, I'm changing auto complete for movie scenes.

01:14:17 - Daniel Rubino (Guest)
Yeah, yeah, kind of. It got me thinking just about how people remix music today. You know, this is on a small scale. These are great when people redo, uh, like movies and stuff like that, but like in the 1950s panavision, those are amazing. Um, but like I can envision, you know, this stuff like people doing fan edits of movies.

01:14:38 - Leo Laporte (Host)
Right, this is this is the terminator redone in 70 millimeter panavision. Look at that. This is done by an AI, right? It looks so cool wow, it's only gonna get better too yeah, I

01:14:54
mean well, and this raises an interesting topic because I've already seen uh pictures of Joe Biden in a wheelchair at a McDonald's. Oh yeah, you know, I've seen some really interesting deep fakes. I don't think they're good enough to fool anybody yet, but they are memes, they are meme generators and, as we know, memes can carry a lot of psychological import. They can really change your mind about things.

01:15:23 - Parmy Olson (Guest)
I think with this news, this is really also a reflection of Stability AI's ability to get a really big name on its board.

01:15:31 - Leo Laporte (Host)
That is a big name yeah.

01:15:33 - Parmy Olson (Guest)
And so, after I saw this news headline, so I interviewed the founder of Stability AI back in August 2022. His name is Emad Mostak. He's not CEO anymore. He had to step down a few months ago because investors were not happy with the way he was running things.

01:15:53
But I remember in that I looked through the transcript of our conversation and because I remember him mentioning James Cameron when we met and he and I looked and he did we're we want to try and make an AI generated movie and we're going to spend X million dollars on this and we're talking to James Cameron to be the director. So this was he was already planning this back in August 2022. And for them to have got him on the board, I think it's a huge. He's so influential in filmmaking right, he is like an icon for filmmakers and, just as he said in his statement, like he has used, whether it's in Avatar or in Terminator, he has absolutely embraced technology for CGI as part of his filmmaking, so it's probably going to just I don't know if it's going to alleviate any of the fears about AI and creative industries and and filmmaking, but I think it's going to help create this maybe the sense of acceptance of using it.

01:16:56 - Leo Laporte (Host)
Meanwhile, of course, sag-aftra is calling a strike against League of Legends because they're using AI voices instead of real players voices to get around the video game strike and also non-union performers. So there is definitely a push-pull going on in Hollywood over AI. It's very interesting. Speaking of AI, did anybody buy the Rabbit R1? No, rabbit is now saying only Miss. They sold a hundred thousand. I think I don't know how many were returned. Only five thousand people use it daily. This, this benito, and I like this because it came from teenage or it was designed by teenage engineering, so it really looked cool. But then we found out it. Well, it was just an ai app running on android, running on a cool looking uh unit. Um, did you buy one, leo? No, I can't. I was, you know what I was. I pushed the buy button and then it didn't take my credit card and I thought, oh, and I ran away from that because I buy all this yeah, I buy all this stuff.

01:18:06
Uh, they have pushed out 16 software updates. They sold a hundred thousand. This is this is like that. What was that projection pen that you pin? Humane pen humane pen, and so these hardware, and I'm waiting.

01:18:20 - Denise Howell (Guest)
I did order it works as a phone, right, no Well?

01:18:25 - Leo Laporte (Host)
it doesn't do anything, it just does the one thing.

01:18:28 - Denise Howell (Guest)
Well, in the article, in the rundown, there is the text the R1 had become somewhat useless to many, leading some to turn it into a more capable Android phone.

01:18:39 - Leo Laporte (Host)
Yeah, you could, and I think that it came with cellular connectivity because that's how it went out and it did things. So the theory was oh, it could call an Uber for you. You don't need a phone because you just have the AI and the R1 get you an Uber. I mean honestly, I think in some time 10 years maybe these will be commonplace. I ordered the pin that's supposed to record everything that I happens in my life and then tell me what happened I ordered this friend, the friendcom no, what was it called?

01:19:13
it was um, you know, it's never come. It's supposed to come back in august. Limitlessai, they made a pin that you clip on and then it would, it's magic, it would record everything. And if it didn't know the voice it would, you could ask hey, do you mind if I record this for later for notes? And if the person said yes, it would then add that voice to your list of approved voices. Anyway, I ordered it, don't have it someday I'm s I'm so intrigued by those devices.

01:19:57 - Parmy Olson (Guest)
The other one is is friendcom, which is like it's a pendant and it does similar things. It's got a mic, always on microphone, and records your conversation so you can ask it through your phone. What about that chat I had with tom? Did he? Did he come off as capacive, aggressive or oh interesting how am I? Yeah, and you can. It can make judgments about your conversations um it says friend, that you just had not imaginary just in case you weren't sure about that, just it might be 99 and it will come someday.

01:20:31 - Leo Laporte (Host)
I'm also waiting for my uh, my glasses with the ar uh glasses that are from brilliant labs. There's a lot of stuff out there and I'm a sucker now. What I do have uh and mark zuckerberg talked about it uh this week on uh at the connect event is the meta glasses, which are getting smarter and smarter. Yeah, people really like those. This, this, you know what? Because they attempt so little. Yeah, there is an ai in it. You could say what am I looking at? And it kind of gives you I. I asked you what am I looking at? And they said, well, there's a lot of screens and wires. You, it didn't really. It did. Let me see if I can. What am I looking at? I can't remember what I'm, how I'm supposed to do this? Hey, meta, what am I looking at?

01:21:24
it's thinking well, it's thinking you're looking at oh wait a minute possibly used for live streaming and video production. It actually understood what it was seeing, which is yeah, there you go pretty wild. Uh, you can take videos with it and pictures. And um, and mark says we're going to have a new version of these, which I think they are selling quite well because they look like.

01:21:49 - Denise Howell (Guest)
They just look like glasses, right well, and and that that's a problem. And that's a problem that friend not imaginary has too. At least 13 states have two-party consent laws.

01:22:02 - Leo Laporte (Host)
You can't just go around recording people well, that's, that's what that limitless was supposed to get around was you have to ask permission? It won't record it, and if it doesn't recognize the voice, but if you ask permission and Denise, and you said yes, okay, then it would add it and would start recording you as well. This, this is a little light that comes on when I, when I record. Yeah, so you know, I mean it's not, it not. It's not red, it's interesting, they didn't make it red.

01:22:26 - Parmy Olson (Guest)
They make it white yeah.

01:22:27 - Salt Hank (Guest)
So it's not the most obvious.

01:22:28 - Parmy Olson (Guest)
It's not like it doesn't strike that same triggering feeling of like seeing a red light on a camera. Suddenly. You're on, I'm holding you. Right now it's a white light, but it's interesting that I'm really impressed by the success of these glasses, the Ray-bans that that meta's had out, and I think a part of it is just that they don't look like tech glasses, they look like normal glasses and they don't make you look really nerdy and awful like the google glass did and and other like snaps augmented reality glasses also look really silly and I think a lot of these companies just underestimate how important people care, how much people care about what they look like when they work.

01:23:04 - Leo Laporte (Host)
Did you see it disappear? Care how much people care about what they look like when they were? Did you see it disappear? This so, uh, apple has, I think. Um, I've said this all along, but oh the overshot. I think apple's kind of missed the boat. With vision pro, it's 3 500 bucks and I don't think anybody wants to strap a heavy computer on their face. I think that's really a lot of it.

01:23:20
Meta is still going to sell their meta quest, but one of the things they talked about is orion, which is not available yet. But but in fact, mark, the guy came out with a briefcase that was handcuffed to his wrist and he said this is the only one of these, but they did show it to some people. They are working on it, they, they look like these, only they're a little thicker. Um, I think you know this is this is something. There they are.

01:23:48
There's the orion glasses, and these are augmented reality. These have, uh, an interesting technology to kind of shoot video into your pupils as opposed to reflect them off the glass. I think that they're on the right track, but who knows how much it'll cost, whether you know what they'll be like or whatever. This is their first time they've talked about this. I'm very interested that Meta has has basically. By the way, at MetaConnect, the developers conference, mark didn't even say the word Facebook once he mentioned Instagram, but I think really that zuckerberg has decided we're going to take all the money we make off facebook and ride it down into the sunset because the only people who use it now are older people.

01:24:37 - Daniel Rubino (Guest)
Uh, and then um, go into this vr ar metaverse thing I mean kind of has to right, because all this, the problem with all these social networks is, at some point you read, mark market saturation, right, and you just can't grow anymore. So as a company, what do you do then? You can't just be like, well, we're kind of fine this quarter, too right, but if you go into other growth areas, uh, or you could sell it to elon musk if you're, if you're, cagey.

01:25:04 - Leo Laporte (Host)
Yes, yeah, he'll buy it.

01:25:06 - Parmy Olson (Guest)
The hard part is that they don't have a platform associated with this technology, right? Because at least Facebook and Instagram they have. Horizons but no one. I mean. I have a Quest 3 and I've been in Horizon events in the Horizon rooms and there's no one there. Yeah, there's no one there. It's like tumbleweeds. Yeah, there's no one there it's like tumbleweeds.

01:25:25 - Leo Laporte (Host)
That was even the story a couple of months ago that even Meta's own employees don't ever use Horizon.

01:25:31 - Parmy Olson (Guest)
Yeah, why should they? There's nothing to do and it's really. There's a hardcore group of like I think small. There's always like a community of people who will keep going because they really get something out of it, but it's nowhere near getting mainstream success. If anything, I think user numbers have really declined significantly.

01:25:50 - Leo Laporte (Host)
Do you think the case with Vision Pro as well With Apple's?

01:25:54 - Parmy Olson (Guest)
product. Well, I'm totally with you, leo, and what you were saying. It's almost like they're trying to do too much in one go and people just aren't ready for it. But they don't do that. I don't think Apple like a social platform for people to meet and mingle in virtual reality, like, like, like meta does.

01:26:11 - Leo Laporte (Host)
Honestly, is not vr, the future is ar and the idea that you could have these glasses on, look completely normal well, nerdy but normal and be getting information through them. They showed a simultaneous translation, which is nerdy because you say something and then the person has to wait and and then the translation gives them the translation, then he responds. So there's this kind of lag in the conversation, but it is. You know, you're having a real-time conversation with somebody in a language you don't speak.

01:26:45 - Daniel Rubino (Guest)
That's pretty cool, I think you know what the best part is, what it's going to be used for ads at some point. Oh.

01:26:51 - Leo Laporte (Host)
God, it's going to be rocking around.

01:26:53 - Daniel Rubino (Guest)
It'll be just ads popping up in your real world.

01:26:57 - Denise Howell (Guest)
It'll be minority report.

01:26:59 - Parmy Olson (Guest)
Yeah, of course it will, because some you know, of course it will it was like you were saying, daniel, in the very beginning when we were talking about Microsoft Recall like how do people expect to use AI? Like it's just going to be ingesting information and people are just it's going to become normalized and people are going to get used to it. And I think it's a similar situation with these glasses. You know, the more the tech disappears and it doesn't look like tech anymore and people don't realize or care as much that they're being filmed or being recorded, then you know there's just going to be a whole boatload of new data and information for meta to ingest and use for ad targeting.

01:27:41 - Daniel Rubino (Guest)
Yeah, look at the watch.

01:27:43 - Denise Howell (Guest)
The watch, the Apple watch, was so novel when it first came out Not very many people had them. Now so many people have them we don't even think twice about it and to that point Apple at one point partnered with Hermes to do an incredibly expensive watch band. I could see them Prada makes eyewear. I could see them partnering up with somebody very high end and maybe trying to take the nerd edge off the product.

01:28:10 - Leo Laporte (Host)
Yeah, we don't know that Apple's not doing the same thing. Meta is and working on AR glasses at the same time as they did.

01:28:16 - Daniel Rubino (Guest)
I mean, there is that rumor that they're doing a lighter version, right to bring down the price and like we won't have as many features, which I think seems obvious. I still think it's. You know, we talked about this on the last month. On here, they're starting from the wrong end. They went from let's put everything we can imagine and this crazy technology and then whittle it down as we figure out what people don't want, versus starting off very simple, which is kind of what Facebook and Meta is doing here with these glasses, and then adding features as people want more things.

01:28:46 - Parmy Olson (Guest)
I've always said these AR glasses. Yeah, to your point. That's how they did it with the iPhone, right? They didn't have a ton of apps on there, it was just a few apps, start small and then grow from there. So it's weird that they did it in that opposite direction.

01:28:59 - Leo Laporte (Host)
Your colleague at Bloomberg, who is probably the most connected Apple rumor guy, mark Gurman, wrote in his Sunday newsletter today, meta's new headsets show Apple has lost its way with the Vision Pro. Yeah, he believes that Apple, while they're working on other stuff, they just haven't really found a market for the Vision Pro. He says it wants to turn the product into a line of devices with different features and price points, but it's not starting from a rock solid foundation, to say the least, the headset is more of a technology showcase than a genuine consumer product. There's little reason for someone to buy a Vision Pro instead of a computer.

01:29:44 - Parmy Olson (Guest)
Yeah, I mean, I suppose if you were going to be generous, you could say yes, this is a technology showcase, so now they're going to whittle things down and start with something a little bit more basic. But I don't know, it sounds like a strange strategy for Apple to do.

01:29:59 - Daniel Rubino (Guest)
They're going to cut out a lot to get down to where the Ray-Ban meadows are. At this point, I mean you're just going to like throw everything out. I've always had the jr glasses too, sorry. I was just gonna say that, like, what would make them really fascinating is what those cameras, of course, is facial recognition. When you're talking to someone, it immediately looks them up and brings up their I love that, their name yeah, yeah, it's like.

01:30:18 - Leo Laporte (Host)
His name is daniel rubino. You used to work with him.

01:30:21 - Daniel Rubino (Guest)
Just their linkedin, yeah, right you know, in 20 years I'm going to need that.

01:30:24 - Denise Howell (Guest)
I'm just telling you what school they went to and like conversations, how many kids they have. Yeah, yeah, that's what I meant when I said gen x is going to sell their soul.

01:30:38 - Parmy Olson (Guest)
They need that yeah, that's another example of that. They're going to be the early adopters. I don't care if I look like.

01:30:44 - Leo Laporte (Host)
Look at this picture of mark zuckerberg from uh, the german's column. I don't care if I even adopters, I don't care if I look like. Look at this picture of Mark Zuckerberg from the German's column.

01:30:47 - Parmy Olson (Guest)
I don't care if I even look like that.

01:30:49 - Leo Laporte (Host)
I don't care.

01:30:52 - Daniel Rubino (Guest)
Speaking to Denise's point there, it's like when you know clear came out for the airport, I was like, yeah, the TSA breed that I stacked them. I'm like I don't care, Take all my fingerprints. Bodily fluids, yeah, Go to security quicker.

01:31:08 - Leo Laporte (Host)
I don't care. Would you do Sam Altman's orb? Would you look into get like he's doing Iris scans? You can't change your Iris. He's doing Iris scans and giving people a handful of world coin all over the world. In fact he's been kicked out of a number of countries because it's a little skeezy. Um, but would you give him your iris? You gave the federal government.

01:31:27 - Denise Howell (Guest)
I gave it clear you gave it to clear and that's a private company. I gave it to the tsa for um. What do we call it? Pre-check it's worth it right but I, to my point there being I'm, I guess, more sanguine about giving it to the us government, that, or a state government, than I am to a private company. Witness 23andMe yeah.

01:31:52 - Leo Laporte (Host)
I gave them my spit and now I don't know what's going to happen with it.

01:31:57 - Parmy Olson (Guest)
But also that that's clear is like that's one to one. Facial recognition right, that's that's. And when you use facial recognition to unlock your phone, that's one-to-one. But using it in glasses for a more kind of one-to-many approach, where you're looking at a crowd of people or a random face comes up to you and then somehow that gets identified, that's a little bit more creepy and I think in a problematic area when it comes to privacy especially, that's where we're in surveillance territory said it had that capability years ago and declined to do it because of stalking and other issues you know that you would.

01:32:34 - Leo Laporte (Host)
Well, there's that business down, yeah yeah, and in fact I I used to have a hello doorbell, uh, and it was supposed to recognize people when they came up to the door and it didn't, and I think that that's why they just didn't want to risk that.

01:32:48 - Daniel Rubino (Guest)
There is a company, though, that that was doing this or is doing this right, but they're scraping the Web for all these images.

01:32:53 - Parmy Olson (Guest)
Clearview, Clearview AI.

01:32:55 - Leo Laporte (Host)
Yeah that's it. You know, and very kind of suspect on it, on the other hand, by the way, there you go to Clearview AI. Look at the headline how Clearview AI Helped Shape the War in Ukraine Great, great. On the other hand, you know, if Taylor Swift uses AI, as she does at her concerts, to keep stalkers out, I support that. I think that's fine. So, as with all this stuff, it's really hard to find a bright moral line. Uh, yes or no?

01:33:28 - Denise Howell (Guest)
on this. Well, anything that helps you keep the lawyers out, a la madison square garden, do you remember?

01:33:34 - Parmy Olson (Guest)
that.

01:33:36 - Leo Laporte (Host)
That's hysterical that was a great, that was a crazy story msg is owned by the dolans uh and uh Dolan's and he is kind of famously litigious and he used the AI at both. It was not just Madison Square Gardens but also, I think, at Radio City Music Hall and, very famously, last Christmas an attorney and her family tried to see the Rockettes Christmas show. They let the family in but they said you cannot go in because you work for a law firm that we are in litigation with and I don't know if she did she sue? Does she have any grounds for suit? I don't know. She said it was very humiliating and I guess it would be.

01:34:21 - Denise Howell (Guest)
Yeah, I think there was some litigation over that. I don't know exactly what happened. I will.

01:34:26 - Leo Laporte (Host)
Google in the background? Yeah, the Dolans are. He also owns the Sphere in Las Vegas, you know, I just feel like we shouldn't have billionaires. We should just legislate them out of existence. Forget all the other stuff. Let's just legislate billionaires out of existence. Forget all the other stuff. Let's just legislate billionaires out of existence. I don't know. Let's take a little break, because I want to be a billionaire and I am going to do everything I can, with ad after ad after ad, to make as much money as possible. No, that's not me, I'm sorry to say that's why I'm in my attic.

01:35:02
I never quite worked that part out Our today, though I'll tell you early on in the show. You remember we had salt hank on and I didn't ask him this, but I know for a fact that his whole business is powered by shopify. This episode of this week in tech is brought to you by shopify. So many businesses, big and little, in the, in brick and mortar or online, use Shopify. If you think about businesses whose sales are rocketing Allbirds, for instance, I'm wearing my Allbirds right now or Untuckit I have many Untuckit shirts you think about an innovative product, a progressive brand and button-downed marketing. Brand and button-downed marketing.

01:35:47
But an often overlooked secret is actually the businesses behind the business making selling and for shoppers, buying simple. For millions of businesses, that business is Shopify. That's it. I love Shopify. They've made it possible for my son to literally create his own business. All of his sales are done by Shopify. Nobody does sales better than Shopify, home of the number one checkout on the planet. And the not-so-secret secret with shop, pay conversions are boosted up to 50 50%. Far fewer carts going abandoned, way more sales being made, yeah.

01:36:25
So if you're into growing your business, your commerce platform better be ready to sell whatever your customers are scrolling or strolling because Shopify is everywhere your customers are, and that's the beauty of Shopify On the web, in your store, in their feed and everywhere in between. Businesses that sell more sell on Shopify. Upgrade your business and get the same checkout that Salt Hank uses and all birds and untuck it. Sign up for your $1 a month trial period at shopifycom slash twit. That's all lowercase shopify S-H-O-P-I-F-Y dot com slash twit. Upgrade your selling today. Shopifycom slash twit. Upgrade You're selling today. Shopifycom slash twit. There you go. I played that sound effect one time too many last week and they had to edit it out. They don't want too many of those, so I'll use this when the commercial's over, and then we will. We will all know.

01:37:25 - Denise Howell (Guest)
Can we screen that out with AI?

01:37:27 - Leo Laporte (Host)
Yeah, the commercial or just the sound effect.

01:37:29 - Denise Howell (Guest)
Just the sound effect.

01:37:30 - Leo Laporte (Host)
Yeah, I think Benito used a razor blade and tape, but I think he got it done. Yeah, something.

01:37:37 - Denise Howell (Guest)
So update on the lawyer and Madison Square Garden she did sue. There was an injunction in place. The trial court said no, they really shouldn't be able to do that, at least while the trial is pending. So there was an injunction put in place. Then there was saying no, you can't screen out your opposing lawyers anymore. Then there was an appeal and the appeals court set it aside and said yes, indeed, of course you can.

01:38:05 - Leo Laporte (Host)
You're a private enterprise, you can do anything you want.

01:38:07 - Denise Howell (Guest)
Right, the lawyer is good at making his case for the quote unquote mother separating a mother from her daughter, and. Girl. Scouts From her children and a.

01:38:19 - Salt Hank (Guest)
Girl.

01:38:19 - Denise Howell (Guest)
Scout.

01:38:20 - Leo Laporte (Host)
Girl Scouts.

01:38:21 - Denise Howell (Guest)
She was watching over.

01:38:22 - Leo Laporte (Host)
She had a troop with her, that's right, yes, exactly. I do remember the New York City tried to prevent him from doing it at Madison Square Garden by threatening his liquor license. Yeah Saying okay, you can block people, but we're not going to let you sell liquor at Madison Square Garden. Right.

01:38:38
And Dolan said fine, no liquor. You tell all those Bruins fans there's no liquor for you Thanks to the New York City liquor, borda Liquor. Yeah, the FTC actually has weighed in on AI. So we've all experienced this. You're looking for in certain product categories it seems to be more common you're looking for a mattress and you go to a website which looks like a real website, has all these mattress reviews, but there's something a little off about it. The FTC has said Lena Kahn went after AI, generated best lists and they have a little section of rules just for them. Now, barring this exact thing. It's a violation for anyone quote to use an unfounded oh, wait a minute, okay, this I'm sorry. So you know about those ftc bans, product review suppression. You better help me out, lawyer. Uh, the ftc says it's a violation for anyone to use quote an unfounded or groundless legal threat, a physical threat, intimidation or a public false accusation. Oh, this is when you get upset at a review like uh yes remember the book the Everything Store about.

01:40:06
Amazon. Oh, I love that book.

01:40:07 - Parmy Olson (Guest)
It's a great book, yeah.

01:40:07 - Leo Laporte (Host)
Brad Stone. Brad's fantastic and at one point in the reviews on Amazon, jeff Bezos' ex she wasn't his ex at the time got in there and said this is a terrible book.

01:40:22 - Denise Howell (Guest)
Anyway, you cannot threaten, so it's more like for uh, I, I got this wrong. I'm no, you're right. There's a new rule on the use of consumer reviews and testimonials, and they are indeed cracking down in the way okay suggesting so I'm confusing two different you are confusing two different things. The interesting thing about the? Well, I mean, it's very interesting that they're cracking down on fake reviews and yay, thank goodness and good, the salon article in the rundown is just written with such snark and irony.

01:40:55 - Daniel Rubino (Guest)
It's so well written.

01:40:57 - Denise Howell (Guest)
Yes I encourage people to go read it. Who's it by here, hold?

01:41:01 - Leo Laporte (Host)
on. I couldn't so this. Yeah, this is Ray Hodge writing in Salon AI Smackdown how a new FTC ruling just protected the free press. Now it's going to be hard to get to that point, but she writes I couldn't be more delighted to be the bearer of such bad news from publishing AI-generated articles and fake product reviews, which posed journalism what I just described. Then. Ftc chair Lena the Lion Khan just landed on your wallet with a WWE-style people's elbow drop from the top rope. So you might want to lawyer up, pal, because this one's going to hit you where it hurts.

01:41:45
She's got style, so what she didn't comment on, unfortunately, your style confused me completely, because it seems like it's two different things now. So explain.

01:41:57 - Denise Howell (Guest)
Well, it's very much all about stamping out these fake review sites, any sorts of you know, pumping up anything on Amazon with fake reviews, all of that. But it also is sort of down far in the text of the new rule. You can no longer buy followers for your social media account. Yeah.

01:42:18 - Daniel Rubino (Guest)
Wow, that's interesting you can't buy reviews anymore.

01:42:23 - Leo Laporte (Host)
So if you're a business, you know you're a business and you're putting fake reviews on yelp, the fdc is going to go after you, yeah, uh. Nor can you threaten bad reviews, which happens all the time. Somebody writes a bad review of a restaurant and the restaurateur goes after them. Uh, it's a violation for a business to provide compensation or other incentives in exchange for a condition, expressly or by implication on you lawyers. Can we get an AI to rewrite this?

01:42:49
the writing or creation of consumer reviews expressing a particular sentiment, whether positive or negative, regarding the product, service or business anyway, I'm not a congress person.

01:43:00 - Denise Howell (Guest)
That's true, that's or?

01:43:03 - Leo Laporte (Host)
or law yeah or a regulation.

01:43:22 - Parmy Olson (Guest)
This is the big battle to try and keep authenticity on the web, which is such a tough one because already, even without generative AI, it's so hard to trust what you see on social media, whether it's a bot or is it a real person. Is the review I read on Amazon really written by a real person or was that a bot? Can I trust all these reviews on Airbnb or whatever you know? It's like we're kind of an SEO the gaming of Google search. How can, how much can I trust that these are like objectively relevant web links for me whenever I do Google search? So I don't know. I think it's going to be really tough because another statistic I've heard is that in the next few years, the web is just going to be absolutely awash with AI generated content, like something like 80 to 90% of the content will be AI generated. So this is a huge battle. It's very heartening, but it's a huge battle.

01:44:04 - Leo Laporte (Host)
It's funny, the prologue of your book Parmi. After picking up this book and reading these first few words, you might be wondering if a human wrote them.

01:44:13 - Parmy Olson (Guest)
I had to put that in there. I had to put that in there in the very beginning.

01:44:17 - Leo Laporte (Host)
I saw when I was in New York a couple of weeks ago previews of a new play which I would recommend to everybody. It's Robert Downey Jr's Broadway debut. It's called McLean and the premise of it is he is a Nobel laureate fiction writer and it starts to come out that he's been using AI to write his fiction. So obviously it's the near future Relevant, really good play. I really enjoyed it and uh it's at the um lincoln center right now.

01:44:52
I imagine it will uh do well on broadway because it's got robert downey jr in it. He's quite good in it, but it but it is. It's exactly. I think it's just a matter of time before there's a some scandal that a well-known novelist has been using ai to to polish up. It was kind of scandalous when we found out that some of our favorite you know uh fiction authors used ghost writers or, you know, hadn't, like hadn't written a book in in years, but had somebody writing it for them like I'm not going to name any names danielle steele or anybody like that but I think that that's not uncommon.

01:45:28
But imagine if an a they started using ais to do it.

01:45:31 - Denise Howell (Guest)
I think that's just a matter of time I'm going to offer a controversial take and say there's a market for it. And you know I mean if some people buy it and read it.

01:45:42 - Leo Laporte (Host)
I guess it's just as good, right.

01:45:44 - Denise Howell (Guest)
Right and there should be disclosure.

01:45:47 - Leo Laporte (Host)
You shouldn't win the Nobel Prize for it.

01:45:48 - Denise Howell (Guest)
Yeah, well, yeah, I don't think you would.

01:45:52 - Leo Laporte (Host)
It's actually. They leave it a little ambiguous. It's really an interesting play because it really talks about all these things that we've been talking about with AI. That's what I love it when Art kind of talks about this.

01:46:04 - Benito (Announcement)
But back to the FTC. I've got a quick question about that real quick, leo. Yeah, if a book is written by AI, could I not just go into ChatGPT and have it write another book, the same book?

01:46:14 - Denise Howell (Guest)
Right, you could.

01:46:16 - Benito (Announcement)
So no one would buy that book then, because I could just go into the AI, it wouldn't be identical.

01:46:27 - Denise Howell (Guest)
It's all about the prompt the crafting and the human creativity is now all just in the prompt. And of course, the author you, because you've wholesale taken their book to train. Sue me but not the ai but the ai wrote the original sue you and see you know, some lawyers are thorough. They're going to sue you both. There are a lot of those lawsuits pending, by the way.

01:46:50 - Parmy Olson (Guest)
I have to share this. I spoke to someone in book publishing recently who's relatively high up and I asked about their whether they would consider publishing books AI-generated books, but trained where the model has been trained on the work of a particular novelist who has died but who's like, really, really successful and they want to keep that revenue stream going. Would they consider publishing books that are AI-generated? And they said yes, they said they would.

01:47:19 - Leo Laporte (Host)
And I think it's really really interesting.

01:47:21 - Parmy Olson (Guest)
Denise, you said that there is a market for this, because I think the question is would people read it? I actually think they would.

01:47:27 - Leo Laporte (Host)
It would have to say in big letters on the front though, an AI homage or something right.

01:47:32 - Parmy Olson (Guest)
Right, yeah, they would have to disclose it.

01:47:34 - Leo Laporte (Host)
That's the thing that was interesting about McNeil.

01:47:36 - Parmy Olson (Guest)
Well, it would be obvious if the novelist is dead.

01:47:38 - Leo Laporte (Host)
In the play. It's kind of pretty savvy. He actually throws in King Lear, actually throws in king lee. He has he's doing rag, he's throwing in king lear and you know as a mandias and a bunch of different poems and things and it says now write me my nobel speech. It's really, it's really great.

01:47:54 - Daniel Rubino (Guest)
I have to say, um, so much potential for this stuff. That can be used for both good and bad, of course, but it's like I mean, for instance, take ai, feed it all the beatles albums, live performances, everything and then get new Beatles About the band. Yeah, you know about their philosophy why hasn't somebody done that? And then have them write new Beatles music right.

01:48:16 - Leo Laporte (Host)
Benito, you're a musician. Okay, if that happened, would it be new Beatles music that?

01:48:19 - Benito (Announcement)
hasn't happened and me personally, I wouldn't care about any of that music because it wasn't the Beatles. John Lennon didn't write that, paul McCartney didn't write that, some computer wrote that. I wouldn't give a crap.

01:48:29 - Leo Laporte (Host)
But, benito, you have a lot of people. Benito's our producer, I care about craft.

01:48:32 - Benito (Announcement)
I don't care as much about output as I care about craft. He likes the process.

01:48:35 - Daniel Rubino (Guest)
But there's also some heart.

01:48:41 - Leo Laporte (Host)
There's a piece of John Lennon. Would you know that it's not there?

01:48:45 - Daniel Rubino (Guest)
I don't know, it would almost be like a theoretical possibility that this would be music you could kind of write. I don't granted there would be people, oh that's not.

01:48:54
Yeah, yeah, you know it's not real, it's not 100% authentic, but if it's good it doesn't matter. Like if it's a good song, people would listen to it. I mean, that's what pop music is. It doesn't matter. Pop music is all manufactured today, but it doesn't matter. If the song is good and catchy and gets stuck, then you wouldn't need to pass it off as a Beatles song.

01:49:12 - Benito (Announcement)
If it's a good song, then you wouldn't need to pass it off as a Beatles song, right? Did you ever see the movie Danny? Oh no.

01:49:18 - Leo Laporte (Host)
Yes, yeah, I love this idea. The idea is, this guy is a huge Beatles fan. He gets clonked in the head and when he wakes up, the Beatles never existed, except he knows all the Beatles songs. So he starts singing them as if he wrote them and becomes a massive star and he gets into a songwriting match with Ed Sheeran and wins it hands down.

01:49:46 - Parmy Olson (Guest)
Just by remembering one because he writes yesterday.

01:49:49 - Leo Laporte (Host)
Uh, imagine, oh, let me write a song. I got a song, let me just write it. Um, the best part of his meeting meets John Lennon, because Lennon never became the Beatles, right, so he had a happy life. He's an old man now. I, if you haven't seen it, it's an interesting and I guess, with, uh, with AI, it might even be something the future brings. So, to go back to the FTC uh thing, the reason this is an issue and and you, you know, you work for Windows Central, daniel I mean, yeah, uh, this comes up all the time in the future. Yeah, these publishing companies are suffering, they're dying. Anand Tech just went out of business, imore just went out of business this week Breaks my heart. So when companies like Red Ventures, these private equity companies, come along and they start posting AI written reviews on their sites because it's link bait, it's cheap to make and you can make hundreds and hundreds of link bait-y review articles, this really devalues our business. Devalues our business.

01:51:04 - Daniel Rubino (Guest)
Yeah, 100%, and it comes up all the time at Future, our parent company, whose position right now and for the foreseeable future is we don't use any AI-generated content on the site.

01:51:17
They should put that in a big banner when actually a lot of us are arguing for that exact thing, although part of it is sort of like the presumption is it's already you're not doing that, so it's kind of weird to be. It's sort of like showing up to um a casino wearing a shirt says I'm not cheating. It's like wait why? Why you're wearing that shirt. You know were you cheating.

01:51:38 - Parmy Olson (Guest)
But that's crazy that our contact sites like yours might eventually put up a banner like a disclosure just saying human made. That's amazing to me.

01:51:49 - Daniel Rubino (Guest)
Yeah. So this comes up. Even if I want to do something AI generated, I'd have to run it up the chain and get all this permission. I can do it as, like you know, explicitly saying this thing I'm about to show you is AI generated. As an example, I could do that, but I can't like do anything. Articles. Now, I think the irony here is that Google is going to be one of the companies that could potentially save this right. It's because AI generated content can also be detected by AI. That's how a lot of this stuff works. Openai has said they have their own tool that can do this, that has like a 99.5 accuracy or something. Um, I think it's going to be really imperative that, uh, social networks, search engines, integrate this stuff into their back end. Um, it doesn't mean it necessarily has to be removed, but it should all be flagged and alerted to people that you know you're looking at a general ai generate, assuming that you can tell yeah, but you can't always tell no.

01:52:48 - Denise Howell (Guest)
And also also, assuming we're talking about a final product, things like co-pilot are right. Can I start that letter for you? Can I start that article for you and then you take it from there, which is fine, right, so there's what is it?

01:53:02 - Daniel Rubino (Guest)
that's my question yeah, I think it is. I think there's. There's ways to. If you're doing a review, there's definitely ways. It goes to this level of how can you prove you did the review, and one way you can prove you did a review is one um having photos, videos, your reporter's notebook unique yeah, having real, yeah, having real data run real tests, show how you'll know the test.

01:53:26
Yeah, you know, there's like an idea, because that's kind of your issue right now. You know up a couple years ago what websites would do is they would do like best tvs, best 4k tvs, and the site wouldn't have reviewed any of the tvs. What they would do then those they would go to best buy, they go to amazon, they look at what other people are saying in their reviews. That's horrible, and they would basically do a human summarization of that and put it into their buying. It's as bad as AI, right? Yeah, and that's what Google has now caught on to is saying that, like you know, now websites are getting penalized for doing that kind of content.

01:54:01
So now it's all about proving how much that you actually did do this review, like how original is your content. The more original it is, the better you will do with Google versus AI generated content tends to be very superficial and won't provide images and video. If I showed you a video with me on screen with a laptop and I'm talking about it, there's a pretty good chance that's real. Now you can get into the whole thing that well, maybe that was the I generated too, but you know this idea that I I'm a real human being. That's going to be what's important here, um, but I think google will be. It's in their interest to kind of solve this problem, uh, because I think long term it's going to blow back. So some of these companies may be winning for now, and that's another reason why future doesn't even touch this stuff, because they're paranoid that at some point google will detect this and then will absolutely punish all those websites, which is what happened a couple years ago, and yeah, they changed their algorithm and legitimate review sites were hurt as well.

01:55:04 - Leo Laporte (Host)
Right, yeah?

01:55:05 - Denise Howell (Guest)
yeah, who cares about the ftc? We?

01:55:07 - Parmy Olson (Guest)
we don't show up in search google's much, much more more significant than anything the ftc could do but I do question whether google would be able to detect this stuff, because maybe they could detect let's say they were perhaps they could detect stuff that's generated by gemini their own model but could they detect stuff generated's generated by Gemini their own model but could they detect stuff generated by an open source model or open AI's model? Because there have been instances I know it was a few months ago there was.

01:55:34 - Leo Laporte (Host)
You know, you could Google a celebrity Like there was one instance with this Hawaiian singer named Israel Kamakawiwole.

01:55:43 - Parmy Olson (Guest)
I think he's somewhere over the rainbow guy. Yeah, kamako Wewole, I think he's somewhere over the rainbow guy. Yeah, yes, that guy. If you Googled him, the first picture that came up it was a photograph. But it's actually not a photograph. It was an AI-generated photograph of him. That was the top Google search result for this guy, and the same happened for I think it was some sort of famous. Dutch painter. Beethoven, oh, it was.

01:56:01
Beethoven. Yeah, so I think. So Google's like I don't know, that it can get a grip on this, especially if the models get better, more realistic, the language is more human-like. I don't know. I think it's a tough one for them.

01:56:16 - Daniel Rubino (Guest)
Yeah, it's going to be a multi-part solution. Like part of it, like right now, google supposedly weighs on this thing called EEAT expertise I forgot what the other one is. It's authority and trustworthiness and it's this idea that, like, I've been doing this now for like 17 years, so, like I have a very long reputation on Google that follows me around and all this where I come across as more authentic versus if someone just wrote an article on even my site and did a review for the first time. They don't really have any history with Google and, by default, google would probably downrank them compared to if I wrote something, and so there's ways of balancing these things off to detect authenticity.

01:56:57
A site like Tom's Guide has been around for a very long time, so they have a reputation that's been there for a while, versus a new site that just popped out. That doesn't mean that, like, tom's Guide couldn't be putting out fake content, of course, but I think there'll be multiple ways to detect this stuff. But yeah, it's always going to be a cat and mouse game, but I actually do believe like AI can be used to fight AI. Like I think that's going to be a thing, but it will be back and forth right. There'll be moments where, no, we did fool the system, but I think companies would be really playing a dangerous game if they start to bet that they won't get caught, because we've seen what happens when Google does decide to change those algorithms and it does decimate websites, and a lot of legitimate websites got decimated too. So it's it's a really interesting problem this is the image of israel kamako.

01:57:48 - Leo Laporte (Host)
I can never get his name on the left.

01:57:50
Ai generated right you can kind of tell yeah, you can kind of tell, but still, the fact that that was the top google image search result uh, is kind of scary. He's, of course, no longer with us. So the ftc this is a cbs uh quote, uh from the ftc. Oh no, I just closed that window. Let me put pull that salon article uh, back up here. Um, they're quoting, uh, the ftc's ad practices director, serena Viswanathan. We're all using them now to make decisions on whether to buy a product, these review sites or these reviews, where to stay on vacation.

01:58:35
But fortunately, with the rise in online reviews, we've seen that bad actors can manipulate or fake reviews to deceive customers for their own benefit. The FTC has seen a massive increase in online reviews in the last few years. So the new rule prevents secretly advertising for yourself, which is why I had this claim that I have a stake in Salt Hanks Pickle Company, right, denise? Denise told me that a long time ago Can't secretly advertise for yourself while pretending to be an independent outlet or company. It bars, quote the creation or operation of websites, organizations or entities that purportedly provide independent reviews or opinions of products. This is a big problem in the VPN space. Search for a VPN review A lot of the sites you go to are owned by the companies that make the product the site reviews. Yeah, I've seen that happens all the time. Uh, it is now illegal, and I guess you know.

01:59:30 - Daniel Rubino (Guest)
I mean, the ftc has limited enforcement capabilities and a lot of fish to fry, but it's good that it's illegal yeah, it's definitely the right direction, and throwing fines out as well as the potential of, like, search engines downgrading you is, yeah, you know, a potential double whammy for sites, you know, attempting to do this stuff. But we could go further. I mean, maybe one of those things at some point where authors and reviewers need to like somehow authenticate themselves, you know, to be on the web and to get a kind of an approval kind of thing, where you just know you're verified, right. That was the whole verification thing with Twitter years ago that we all enjoyed, which was you knew that was the real person, because at some point you had to show your ID to someone to verify it. That's still a valid way to get around this stuff. But then people get you know squishy with the idea of, like, well, how are you going to see my license?

02:00:25 - Denise Howell (Guest)
You know, so I don't know. And, as you were pointing out, daniel, google essentially verifies you without giving you a blue check. We used to call that Google juice, right that you would? Just you know you had that influence with Google based on what you had published in the past.

02:00:40 - Leo Laporte (Host)
It's always been easy to impersonate people online. I mean, it's only going to get easier, right? I've had people take my images name and make fake uh, that's the real me, because I got a blue check right. But make fake uh.

02:00:53 - Daniel Rubino (Guest)
It's easy enough to take these images and make a fake account, and it's happened to me but it's one of those things that as soon as you start to pull out that, that string a little bit, it's usually pretty easy like if you just start looking, one would hope you would figure it out usually pretty quickly. But someone has to be looking, right it also makes it difficult.

02:01:10 - Leo Laporte (Host)
When I went back because I'd left facebook for years and I thought I ought to have a facebook account just to see what's going on. So I made an account and invited some friends and they said is this you? Because you're not on facebook, it also makes it harder for you to convince people this is really you. And then there's the case of the AI lawyer. The FTC sued I. I'm a little sad about this, because Do Not Pay is really cool. Do Not Pay was created to get you out of parking tickets, right, denise?

02:01:42 - Denise Howell (Guest)
Yeah, and for anybody interested, you can go back and see an interview on this week in law with the founder, episode 352 was he an attorney? Um, I want to say yes, I don't remember off the top of my head, but he noticed that you could.

02:01:59 - Leo Laporte (Host)
If you fought a parking ticket, yeah you, you could pretty much always get out of it kind of automatically.

02:02:05 - Denise Howell (Guest)
Yeah, his name's Josh Browder and yeah, he'd made this really cool site. But you know, not just the FTC but every state bar is really really trigger happy and sensitive on this issue of the unauthorized practice of law, and you have to tread that line carefully and so he started do not pay, not really to practice law, but to get you out of parking tickets.

02:02:34 - Leo Laporte (Host)
That probably was okay, but then he got ambitious and he expanded to cover over 200 other areas of law breach of contract, restraining orders, insurance claims, divorce settlements. But he never acknowledged that these services were provided without lawyer oversight.

02:02:57 - Parmy Olson (Guest)
Right, it's a pure robo-lawyer, yeah.

02:02:59 - Leo Laporte (Host)
Robo-something. Yeah, robo something. Anyway. The ftc uh said that uh do not pay. Begin charging subscribers 36 every two months while making false claims in ads to drive up subscriptions. Deceptive ads included a prominently featured quote, purportedly from the la times that hyped up the service by saying quote what this robot lawyer can do is astonishingly similar, if not more, to what human lawyers do. The quote was actually from a high schooler's opinion piece in the Los Angeles Times High School Insider website, which is a user-generated content platform for young people. I mean, I guess that's the LA Times.

02:03:40 - Denise Howell (Guest)
No, no, it's not. Okay, not at all, all right.

02:03:44 - Leo Laporte (Host)
Also beyond the deceptive ads. I'm reading from the Ars Technica piece by Ashley Belanger Also the service was essentially just a chatbot trained to recognize relationships between words. Wait a minute, that sounds a little bit like ChatGPT.

02:04:06 - Parmy Olson (Guest)
They used ChatGPT.

02:04:07 - Leo Laporte (Host)
They were using open ai's model for their stuff yeah, none of the services technologies fcc said has been trained on a comprehensive and current corpus of federal and state law regulations and judicial decisions. Well, is that the only mistake they made? Would it be okay to use an ai lawyer if it were properly trained?

02:04:23
someone's gonna do it the ftc said they have the do not pay employees have not tested the quality and accuracy of legal documents and advice generated, they have not employed in term attorneys. They've not retained attorneys, let alone attorneys with relevant legal expertise, to test the quality, accuracy of the features. Anyway, he settled $193,000. So it's kind of a tiny little settlement.

02:04:50 - Denise Howell (Guest)
It's a lot of feedback. You're Josh Browder.

02:04:52 - Parmy Olson (Guest)
Oh, but actually if you look at the top comment of that article that you're reading, leo, and that sort of makes a really good point. There they say, okay, so this company has 200,000 subscribers. They're paying 36 every two months. That means they're paying 216 a year, which means they're the company is making an estimated 43 million dollars in revenue a year. So that fine is basically half. It's tiny, it's half a percent of their revenue. Wow, it's. It's nothing for them.

02:05:23 - Leo Laporte (Host)
It's a cost probably what the FTC was more going after was like stopped it, knock it off right, and I'm sure they got that yes, as a deterrent to others, which I think is healthy.

02:05:34 - Parmy Olson (Guest)
It's definitely an issue. I mean this, this guy, josh Browder for years and I've written about him before, even back when I was at Forbes he's one of these guys that's just really good at selling the future, very, very good at it, um, and has that tendency to fudge the truth a little bit, maybe with the marketing. I remember when I interviewed him once he was saying that oh, our next thing is we're gonna invent a robo lawyer that can talk to you through your airpods when you're in court and it will just whisper you instructions so you know what to say defend yourself, wasn't he?

02:06:07 - Leo Laporte (Host)
I believe that they stopped that. Yeah, that one in the bud but it was coming.

02:06:12 - Parmy Olson (Guest)
It was the future he he wanted to.

02:06:15 - Denise Howell (Guest)
He was uh prevented. He was gonna go to jail for bringing a um can't do that a an ai lawyer into a courtroom and his. Airpods.

02:06:25 - Leo Laporte (Host)
There's another I have though what I do is I just call Denise, and I have her in my AirPods.

02:06:31 - Denise Howell (Guest)
To your point. There was another company, and this was the guest on Twill with Josh Browder. It was founded by Andrew Arruda and it was called Ross Intelligence and they were basically sued out of existence by I forget who sued them. It was the database that runs Westlaw, and that is a huge repository of legal opinions, statutes, basically what the FTC was calling out. Hey, if you trained something on a huge composite of actual legal data, then you might have something legitimate to talk about, and that's what Ross was trying to do. But copyright considerations basically sued them.

02:07:18 - Leo Laporte (Host)
Yeah, Westlaw won that one.

02:07:20 - Parmy Olson (Guest)
Yeah, but this is kind of a necessary area that I think the FTC is getting into, which has been an issue for a number of years, which is companies kind of riding on the coattails of AI hype to get higher valuations and to embellish the capabilitiesder's product can do and has done and I've seen that he has Do Not Pay has helped a lot of people get out of parking tickets. But from what I've read, like broadly the FTC is trying to crack down on businesses that are just really egregiously overhyping the use of AI and deceiving customers and investors too, and it's been a problem for years. This is not a recent thing.

02:08:06 - Leo Laporte (Host)
They have something called Operation AI Comply, which that's the one. Yes, yeah, lena Kahn says well, monitor companies, attempts to quote, lure consumers into bogus schemes or use AI tools to turbocharge deception. God bless Lena Kahn. I'm glad that she's paying attention to this because, you know, typically you'd say 10 years later the government say, oh, we should do something about this. But she's, she's pretty smart. She said using AI tools to trick, mislead or defraud people is illegal. The FTC's enforcement actions make it clear there is no AI exemption from the laws on the books. By cracking down on unfair or deceptive practices in these markets, ftc is ensuring that honest businesses and innovators can get a fair shot. Bravo.

02:08:52 - Daniel Rubino (Guest)
And it also points to the risk of I mean, I'm glad she's there and doing this kind of thing, but imagine if someone from the AI industry became the Fc chair, which we know how lobbying works and how people get that's why your vote counts, folks. Yeah, you get someone in there it's like, well, you know, maybe this ai stuff isn't so bad.

02:09:13 - Denise Howell (Guest)
A lot of companies or, god forbid, private equity yeah, right, yeah, yeah, because I'm tempted to think that's what happened with do not pay. I have no basis for saying that. I don't know how they're funded or who they've partnered with, but somebody got greedy.

02:09:30 - Leo Laporte (Host)
That's what happened. We don't know who, but somebody got greedy. Let's take a little break. We will have more, actually a few more minutes with our great panel. Parmi Olson is here, the author of a brand new book, supremacy. I hope you'll come back when you don't have a book to flog.

02:09:45 - Parmy Olson (Guest)
I love having you on fantastic no, absolutely okay, I'd love to come back okay, uh, we just used the book as a hook.

02:09:53 - Leo Laporte (Host)
Uh, go get it. It's great if you want to know how we got to where we are today and you want to know about the. More importantly, I think it's important to understand the personalities. People like sam altman and demis hasibis, uh, the people behind it gives you a little bit more insight into what's going on and what their motivations are. Uh, brand new just came out from st martin's press also. Uh, wonderful journalist. Editor-in-chief of windows central. It's always great to have daniel rubino on and Denise Howell, attorney at law. Tell me about your new podcast that here say culture.

02:10:28 - Denise Howell (Guest)
Well, I co-host a show with Dave Levine, who's a professor at Elon university, and that is called R and D with D and D.

02:10:38
I like the name, so we do a two and two, two on one interview kind of format on all things sort of tech, culture, society. We just try and dig deep and help people have a better understanding of things. And then my own show is called Uneven Distribution, which is what you were on, leo, and for everyone listening, I definitely encourage you to check out my episode with Leo, because we had great fun and I learned a lot and I know you would too. And my current episode is with a woman named Marina Gerner. She's a journalist who's done a ton of research on women's health care and startups.

02:11:21 - Leo Laporte (Host)
Thank you so much. Sorry, I started playing it.

02:11:33 - Denise Howell (Guest)
Awesome health care and startups. Thank you so much. Sorry I started playing it. Awesome, yeah, um. And uh, she has a book out now called the vagina business. Uh, that just came out this month. Yes, it's so good. It's all about um startups, you know, struggling against the traditional vc infrastructure and biases, trying to serve women's healthcare ideas.

02:11:50 - Parmy Olson (Guest)
I want to order it just for the name alone, but then you described it and now I definitely want to order this book.

02:11:56 - Denise Howell (Guest)
Oh, and you have to order it for the graphic on the front too, which shows a sort of diamond shape with a USB cable running down the middle of it.

02:12:06 - Leo Laporte (Host)
You know, lisa my wife is always complaining about how medical care is geared around men, and they're often. For instance, she needs a knee replacement, but all the knees are for men, and then they just. Well, we'll give you a small man's knee. No, no, that's just not right.

02:12:25 - Denise Howell (Guest)
Uh, Marina did just hundreds of hours of research for this book. And you're going to learn so much about a lot of really great startups that are out there that you probably never would have heard of.

02:12:36 - Leo Laporte (Host)
What is the economic? What is the what is the economic uh force that's doing that Cause? I mean, women have always been 51% of the population and yet ignored for so long in medicine and many other businesses.

02:12:49 - Denise Howell (Guest)
Well, I I forget the exact quote, but one of the people she interviewed at vc for the book she was trying to say you know why? Why don't you fund more of these companies? And he said I don't want to be thinking about vaginas with my morning coffee oh god.

02:13:03 - Salt Hank (Guest)
And he said I don't want to be thinking about vaginas with my morning coffee. Oh God, oh my God, I mean you know, it was just that bad that bad.

02:13:09 - Leo Laporte (Host)
Yeah, yeah. Apparently 2.1% of venture capital dollars go to companies founded by women, 2.1% Oof and the simple answer is the patriarchy right. Yeah.

02:13:20 - Daniel Rubino (Guest)
The patriarchy VC is always the worst.

02:13:25 - Salt Hank (Guest)
It always comes back to that. That's the real.

02:13:26 - Leo Laporte (Host)
You know, if you want to know what people really are all about, find out where they put their money, cause that tells you everything. Yep, well, it's great to have you, denise. Here's say culturecom. That sounds like an episode well worth listening to. I look forward to that, Our show today, brought to you by Veeam.

02:13:46
Now, this is a business where everybody ought to know about it. I don't understand. You hear it all the time ransomware attacks and companies paying millions of dollars to bad guys to get their data back. Your data is everything in your business. Without your data, your customers trust turns to digital dust. That's why veams data protection and ransomware recovery ensures that you can secure and restore your enterprise data wherever and whenever you need it. Everyone needs this. No, stop paying ransomware folks. That's crazy.

02:14:22
Veem is trusted by over 77% of the Fortune 500 to keep their businesses running when digital disruptions like ransomware strike. Now they're having a big event, october 1st. It's open to all of you, it's virtual the 2024 Veeam On Data Resilience Summit, and I think you're going to want to attend to this. A world without data is one that grinds to a halt, but data resilient organizations don't let the threats of cyber attacks or disruptions define them. They put the plan, the expertise and the solutions in place ahead of time, so their businesses will keep running no matter what. That's why you want to find out about this.

02:15:00
Join the virtual Veeamon Data Resilience Summit. You'll learn about the industry's leading playbook to keep your data safe, protected and available, keynote-worthy news about Veeam's latest product updates, how to take full control over where and how your data is stored, and best practices from data resilience experts who are leading the way. You've got to do this. Go to Veeam, v-e-a-mcom V-E-A-Mcom. Register. Join it, it's free, it's online. There's no reason to miss the Veeam on Data Resilience Summit coming up October 1st, veeamcom. And if you don't do this, you gotta be nuts. I mean it's just, it's obvious, veemcom.

02:15:47 - Denise Howell (Guest)
Thank you, veem, for your support hey, can I follow up on something we were just talking about, because literally I didn't I'm not making this up. I went to go buy henry laporte's cookbook thank you reading the ad copy. Thank you, it's called.

02:16:04 - Leo Laporte (Host)
Salt Hank, a five napkin situation.

02:16:07 - Denise Howell (Guest)
Yes.

02:16:08 - Leo Laporte (Host)
And it's 40% off at Target and Walmart right now, so go buy it.

02:16:12 - Denise Howell (Guest)
It is also 40% off at Amazon, or that's what it's telling me you bet, you bet, yep. And when I launched Amazon on Firefox, a little beta window popped up from Firefox that's called Review Checker. Try our trusted guide to product reviews.

02:16:33 - Leo Laporte (Host)
Oh.

02:16:34 - Denise Howell (Guest)
See how reliable product reviews are on Amazon before you buy. Review Checker. An experimental feature from Firefox is built right into the browser.

02:16:42 - Leo Laporte (Host)
Good for them.

02:16:44 - Denise Howell (Guest)
It works on Walmart and Best Buy too. Using the power of FakeSpot by Mozilla, we help you avoid biased and inauthentic reviews. Our AI model is always improving to protect you as you shop, so they're fighting fire with fire. They're using AI to detect fake. I didn't know.

02:17:03 - Daniel Rubino (Guest)
Yeah, that's interesting. I didn't know, yeah, that's interesting. I didn't know firefox. I guess the mozilla foundation owns fake spots. Fake spots been around for a while and you could just install the extension. Oh, you can install the extension on any browser. Okay, so, but it's kind of cool they're building it right into, uh, firefox too, because you have to go out and find it now right and it just shows you how bad the uh fake review situation is.

02:17:24 - Leo Laporte (Host)
Oh, yeah, yeah, you need a plug-in in your browser to tell you no, no, that that one's fake it does two things now.

02:17:30 - Daniel Rubino (Guest)
Now it looks at uh, both the reviews on the site from the people to you know determine how many of those reviews are real and fake. And then it also now uh rates the buyer to where you're buying from, because a lot of times on amazon you're not getting it from amazon, you're kind of getting it through service through amazon, through another company, and so it'll also review that to tell them if they're like a scam would have issues and stuff like that.

02:17:54 - Leo Laporte (Host)
So it's uh. Firefox 119. Version 119 introduces review checker.

02:18:01 - Daniel Rubino (Guest)
That's very good and some of these browser companies, like firefox and brave.

02:18:04 - Leo Laporte (Host)
Brave has done a lot of really interesting things with privacy as well, building into our browser, into theirs and ad blockers and stuff you know why I support firefox, because brave, along with arc, which I'm using right now, edge, they're all chrome based, they're all chromium browsers and I just feel like there's a real risk to a monoculture in anything, whether it's bananas or browsers. And, uh, you, really it's good to have another. There's really only two engines now there's chromium and apple safari engine. So it's good, firefox is dwindling, but let's support them because we need that. We need that third engine. I think that third way of doing things and they're doing great stuff. I'm a big fan. Firefox review checker. Oh, and let's not forget salt Hank, a five napkin situation. Esquire magazine, in a real review, said it was the best cookbook of 2024 so far. Wow, so it's beautiful, it's a. It's a. It's actually a beautiful, uh, picture book as much as a cookbook. You're just going to look at all the pictures and go I'm hungry now and it's my son and I'm so proud of him.

02:19:14 - Denise Howell (Guest)
That's such a great title.

02:19:16 - Leo Laporte (Host)
Isn't it? It's his life. We went to visit my dad, who's in the nursing home. He's 92, now 91, and, uh, my dad wrote a book when I was a kid. I wrote a book when henry was a kid, and now henry has written a book. So I said say from the third generation of of authors isn't that cool.

02:19:38
That's really cool yeah, yeah, very cool. I have his, my book and my dad's books on my shelf over here and I'm going to add Hank's. Oh, perfect, wouldn't that be cool? Wouldn't you like that, parmi, if you have created kind of a generational thing?

02:19:54 - Parmy Olson (Guest)
No pressure on my daughter. She's like get to writing, Don't use AI.

02:20:00 - Leo Laporte (Host)
By the way, good news as long as we're talking about California law, they have just put a law into practice. Actually, it's going to be a few years probably before it's real. That requires one-click subscription cancellation. It should be as easy to cancel as it is to subscribe Now. The law passed the Assembly. I don't know if the governor has signed it yet. Yeah, it takes effect january of next year and it would cover purchases and contracted contracts entered into from the following july. So I think the ftc is all I know. Lena khan's been working on this as well, but you know what california huge economy. As california goes in in cases like this, so goes the nation let's. That's good. That's a good law.

02:20:49 - Daniel Rubino (Guest)
It should be as easy to cancel because they hide it right? Yeah, oh yeah, they make you click to sign up but then to cancel. They like you have to call us. You know, absolutely, it's just like they know a lot of people, like even myself, I'll do it tomorrow, or it's like eight o'clock at night, you know, like they're not going to be open, right?

02:21:06 - Leo Laporte (Host)
yeah, um yeah, they do that all on purpose oh, you sign up online and they say, well, if you want to cancel, you're going to have to call us and yeah, we're going to talk you out of it. And blah, blah, blah, blah, blah. Speak speaking of uh, might as well do the the government legal segment. The DOJ is suing Visa, the big credit card company, saying they tried to smother competitors like Square. They filed an antitrust lawsuit saying that they have an illegal monopoly over debit network markets and has attempted to unlawfully crush competitors, including paypal and square. Multi-year investigation since 2021, the doj has been investigating that right, I mean, think about it.

02:21:55 - Denise Howell (Guest)
Have you ever had a non-visa debit card?

02:21:59 - Leo Laporte (Host)
yeah, I've had a mastercard debit card you have yeah through your bank yeah okay usaa, but they changed to visa a couple of years ago. They were mastercard wonder why.

02:22:12 - Parmy Olson (Guest)
How do they do that?

02:22:14 - Leo Laporte (Host)
what would they get off from incentives?

02:22:17 - Daniel Rubino (Guest)
they might. With scale they can. It would scale. You can offer better packages and deals, and for years I thought mastercard and visa were like the same parallel.

02:22:27 - Leo Laporte (Host)
There was discover, which nobody used mastercard and visa and then the american express, which nobody used because it was the merchants don't like it because they take a bigger cut. Visa makes more, according to the verge, makes more than seven billion a year in payment processing fees alone.

02:22:41 - Daniel Rubino (Guest)
I'm not talking about interest on your credit card, just the fees 60% of debit transactions in the US run on Visa's network and it's why a lot of some places will. I've seen it recently where some places will offer a lower price for cash. Yeah, that's right, because when you use a Visa or debit card, whatever they get charged to use that network. But if you pay it in cash it can be cheaper, so they end up charging you more for the product to make up for it. So that's what the big thing here is, that they're saying it's not just that it affects all these retailers, but everybody ends up paying more money for all these services and goods because of this, because the patent, the cost gets passed on to the consumer.

02:23:29 - Leo Laporte (Host)
According to the complaint, visa entered into paid agreements with potential competitors as part of an effort to fend off competition from newer entrants. The doj said this has allowed visa to build quote an enormous moat around its business. When Square launched Square Cash in 2013, the cash app right Visa was so worried that it would threaten their payment volume. They entered into a series of contracts to discourage Square from competing aggressively against Visa. Or, as a Visa executive said and I guess this is in the emails or something in discovery, quote we've got square on a short leash wow, right, and they considered it doesn't an existential threat oh yeah, I use apple, pay for everything, right doesn't sound monopolistic at all.

02:24:20 - Denise Howell (Guest)
No, and now we've got paypal out there with the will ferrell ads trying to encourage venmo and paper oh yeah, what will ferrell's dancing around saying use paypal.

02:24:29 - Leo Laporte (Host)
I don't even know how you would use paypal to pay for something apparently the way you would use do you put it on your watch or no, you'd have to I'm sure venmo is a paper, venmo is paypal. That's what I'm puzzled by the whole thing. But I never used paypal you can send direct cash to people on paypal too.

02:24:53 - Daniel Rubino (Guest)
It's basically the same, yeah I don't think anybody wants to use paypal I'm sorry the us on websites like where it's a pay option, you know?

02:25:01 - Leo Laporte (Host)
uh, because you have to fill out your own credit card, but then it's just a credit card. I mean, yeah, because you, I don't want to log into paypal, and then I have to fill out your own credit card, but then it's just a credit card. I mean, yeah, cause you, I don't want to log into PayPal, and then I have to do the authentic K. It just takes forever.

02:25:09 - Benito (Announcement)
I actually use PayPal for all of my online purchases. When they when they add, when they offer it, because it's one click yeah.

02:25:16 - Daniel Rubino (Guest)
And you can and you can set it up to pay from multiple sources, right?

02:25:29 - Parmy Olson (Guest)
yeah, so it is, it is easy. I mean, ebay is primary primary uh choice for purchases. I just it'd be nice if this leads to if this leads to lower fees. Right is not the for consumers, that's that's the hopeful end game of the lawsuit oh, apple Pay is a Master card.

02:25:42 - Leo Laporte (Host)
Never mind, I do have a, I do have a master card, it's my apple pay. But that's. That's not a debit card it comes right out of my bank. Okay, oh no, it doesn't. No, it doesn't you're right it's not, it's a it's a charge card and then a month later it comes out of my bank, right?

02:26:00
I don't know. I think everything's free. I just tap my watch and I just take it. It doesn't feel like I'm paying for anything. It's amazing. I love it. Hey, here's some really good news from NIST N-I-S-T-A what is that? The National Institute of Standards and Technology. Somebody, way back when, when NIST was writing password recommendations, somebody I won't name his name, we do know his name wrote that oh, passwords should be changed every six months. Just made it up because there's no real reason for that. Nevertheless, how many of you does your work demand that your password be changed every few months? Yeah, you got to come up with a new password, so annoying. The guy who wrote that recommendation later recanted and said I'm sorry. I should never have written that it's wrong.

02:26:59
Well, last week, nist released the second public draft of its digital identity guidelines, which, by the way, runs to 35,000 words, but the section devoted to passwords finally says the new rules bar the requirement that end users periodically change their passwords, that end users periodically change their passwords. This requirement came into being decades ago, when password security was poorly understood. Since then, most services require stronger passwords. They're random, people use password managers, et cetera, et cetera. The real problem with changing your password Parmi, you can vouch for this. Is it Bloomberg that make you change your password more than every three months.

02:27:48 - Parmy Olson (Guest)
I don't know if it's every three months. It might be every three months or every six months.

02:27:51 - Leo Laporte (Host)
I just had to change it last week.

02:27:53 - Parmy Olson (Guest)
It was so annoying.

02:27:54 - Leo Laporte (Host)
So what do you do? You take the old password and you add something to it, or you don't come up with a better password. In many cases, it's just a bad idea.

02:28:05 - Parmy Olson (Guest)
But isn't also the possible impact that people just choose weaker passwords as a result, so they can remember it. Exactly.

02:28:11 - Leo Laporte (Host)
So they can remember them.

02:28:12 - Parmy Olson (Guest)
And then that defeats the whole purpose of keeping the network secure there's another thing that they wrote into.

02:28:19 - Leo Laporte (Host)
Another requirement that often does more harm than good is the required use of certain characters. You've seen it. You've got to have one number, one special character, one upper and lower case. That actually weakens your password. It should. If you dan gooden's writing about this in ours technique and dan's a very good security guy, he knows about this stuff that weakens your password. The new NIST guidelines say verifiers shall not impose composition rules for passwords and they shall not require users to change passwords periodically. They can force a change if there's evidence of compromise. Of course that's sensible.

02:29:01 - Parmy Olson (Guest)
Why did they say shall and not should? That's just really oddise howell knows, or something well and again, these are just guidelines, right?

02:29:09 - Denise Howell (Guest)
nobody has to nobody.

02:29:11 - Leo Laporte (Host)
Yeah, shall implies like, shall, like you will, you will, shall not is trying to make it stronger, even though it's just guidelines, right? Yeah, in previous versions of the guidelines actually it's good you asked this question, parmi, you're a good journalist in previous versions of the guidelines, some of the rules used words should not, which means the practice is not recommended as a best practice. Shall not, by contrast, means the practice must be barred for an organization to be in compliance. So they do, I guess, have some sort of compliance standard. They don't have the rule of law.

02:29:46 - Parmy Olson (Guest)
So this is great, so now I can send an email to my IT.

02:29:50 - Leo Laporte (Host)
Yes, send them this article Link this.

02:29:52 - Parmy Olson (Guest)
Without this article and these guidelines, they would just like. It would be like throwing a message into the abyss. They wouldn't listen to me, but now they will because they've said shall in the guidelines.

02:30:02 - Denise Howell (Guest)
So they Now they will, because they've said shall in the guidelines.

02:30:04 - Leo Laporte (Host)
But Parmi, should you, I'm guessing you already bugged them, she shall it drives me crazy iHeart the last time I worked for, a big company was iHeartMedia, and they required this every few months, and it drives me crazy, because I only logged into my account once every six months or so. Every time I had to make a new password.

02:30:24 - Daniel Rubino (Guest)
It's like I just barely got this one bigger question is now can parents use should and shall for their kids in terms of rules?

02:30:35 - Parmy Olson (Guest)
now shall just changing the culture but this is important distinction yeah, I remember from school there there shall is stronger than should or would or should it just sounds like something people would say on downton abbey like shall we retire for a cigar, shall we? Yeah?

02:30:56 - Leo Laporte (Host)
shall we wrap things up with the final commercial, shall we? And then maybe, uh, we'll find something funny to talk about.

02:31:02 - Daniel Rubino (Guest)
A little amuse-bouche to conclude this episode Shall we, I think, I say indeed at the end Indeed, indeed we shall.

02:31:10 - Denise Howell (Guest)
And you have to say it, like Maggie Smith.

02:31:13 - Leo Laporte (Host)
Oh absolutely. Passed away this week at the age of 89. One of the great actresses.

02:31:20 - Parmy Olson (Guest)
Oh, you're finding out now.

02:31:22 - Leo Laporte (Host)
She was so good in Downton Abbey as the salty. What was she, great aunt?

02:31:28 - Parmy Olson (Guest)
The Dowager, the Dowager, lady Grantham, no, she's the.

02:31:33 - Denise Howell (Guest)
Duchess? No, she was the Archduchess.

02:31:36 - Parmy Olson (Guest)
I don't remember Dowager Countess.

02:31:39 - Leo Laporte (Host)
Something. What does Dowager mean? It means you never got married, right?

02:31:43 - Denise Howell (Guest)
No, it means there is You're a parent of the king.

02:31:47 - Leo Laporte (Host)
A parent of the king.

02:31:48 - Denise Howell (Guest)
So Lady Grantham is the Duke's wife? Yes, the dowager used to be Lady Grantham, ah so she's the former Lady Grantham.

02:31:58 - Leo Laporte (Host)
Right Just like the dowager, queen Elizabeth was Elizabeth's mother. Yes, we have learned so much about the English language on today's episode. Final thought here this episode of this Week in Tech brought to you by Flashpoint. Now, for security leaders, 2024 has been a year like no other. Cyber threats and physical security concerns have continued to increase. Now you've got geopolitical instability. You know the world is a crazy place and every new incident adds a new layer of risk and uncertainty. Let's talk numbers. Last year, this is this is terrifying. There was a staggering 84 rise in ransomware attacks, a 34% jump in data breaches and, of course, the result trillions of dollars in financial losses and threats to safety worldwide.

02:32:54
Now, if you're a government, if you're a world leader, you've got intelligence, people who bring you this information, who keep you up to date, who let you know ahead of time what's going on. What do you do in business? You need Flashpoint. Flashpoint empowers organizations to make mission critical decisions that will keep their people and assets safe. How? By combining cutting edge technology with the expertise of world class analyst teams Some of the best people in the business work for Flashpoint and with Ignite, which is Flashpoint's award-winning threat intelligence platform, you get access to critical data, finished intelligence, alerts and analytics all in one place. Some of the Flashpoint customers avoid half a billion dollars in fraud loss annually a 482% ROI in six months. Probably why Flashpoint earned Frost Sullivan's 2024 Global Product Leadership Award for unrivaled threat data and intelligence.

02:33:52
You need this information as an SVP of cyber operations at a I won't name it, but a large US financial institution said quote Flashpoint saves us over $80 million in fraud losses every year. Their proactive approach and sharp insights are crucial in keeping our financial institutions secure. They're not just a solution, they're a strategic partner helping us stay ahead of cyber threats. No wonder Flashpoint is trusted by both mission-critical businesses and yes, governments worldwide. You need this information, you need this intelligence To access the industry's best threat data and intelligence. Visit flashpointio today, otherwise you're just working in the dark. Flashpointio I talked to these guys.

02:34:42
I was kind of blown away. I did not even realize shows my ignorance, uh, that this kind of private intelligence exists. Uh, and if you're a business, you need it, you desperately need it. All right, I've got a few extra stories from our great panelists. Denise howell. Stories from our great panelists denise howell some mad genius put chat gpt into a ti-84 graphing calculator. I know. You know about this because of tyler right. You had to buy him this calculator, didn't you? When he took the sats, or whatever long ago.

02:35:18 - Denise Howell (Guest)
Yes, and they say you know, he's a business major in college, so he's oh, so he still uses it uses graphing calculators? Yeah, we.

02:35:31 - Leo Laporte (Host)
We bought these for our kids, many of them because they kept losing them. Yeah, imagine, though, adding chat, gpt, because they allow these in testing right.

02:35:37 - Denise Howell (Guest)
So the reason I actually threw this in is in our regulatory section one. One more law that California just enacted is the one that says California schools must restrict phones, like the LA Unified School District is already doing, and all I can think about is okay. So how's this cat and mouse going to go? And that's why I asked about Rabbit and whether it was a phone or friend, not imaginary, or you know whatever. What are kids going to use? That's not a phone that they're going to get around this law. It's going to be something. I just don't know what it is.

02:36:18 - Leo Laporte (Host)
This is the video describing how he put ChatGPT into this graphing calculator.

02:36:25 - Parmy Olson (Guest)
That's amazing, yeah, so he was able to access chat GPT through the calculator, basically.

02:36:32 - Leo Laporte (Host)
Apparently he's playing Tetris on it as well.

02:36:35 - Denise Howell (Guest)
Right, and when you read his sort of goals in doing this job, one was get the TI-84 banned from the sat yeah, yeah, there's actually.

02:36:48 - Leo Laporte (Host)
This is a whole story of how ti basically made so much money by getting these. Yes, the lock in on these, right, yeah, yeah, I don't blame him. God bless him. He's doing, he's doing the Lord's work. Putting chat GPT on a TI-84 graphing calculator. That is wild. Let's see what else you put in here. The final rule on use of oh, we did that one on the use of consumer reviews 23andMe. So this week the entire board quit at 23andMe entire. This week the entire board quit at 23 and me the.

02:37:29
This is the the uh ann wajiski's company that I was really excited about. I think they were even an advertiser for a while. You spit into a tube and then they would do some genetic analysis. It turned out that the genetic analysis they were doing was kind of a little. You know they. They were doing statistical matching between this genome and the questionnaires they sent people and so forth. It's gotten better over time. There are other, maybe better, uh genome. In fact I just I did one recently, much more expensive, where they'd sequence the entire genome. But it does raise the issue who has control of the spit vials? 23andme has been talking about going public. They just settled a data breach lawsuit for 30 million dollars and there really is some question about their financial future. And if 23andme does go out of business or go public, who gets the spit?

02:38:34 - Daniel Rubino (Guest)
I think it's also just interesting that in hindsight, setting up a company where you charge people one time for one thing and there's no reason for them ever to come back and pay for it again, it's not a really good, long-term good, good model that was the problem from the start.

02:38:52 - Benito (Announcement)
They'd never figured out any kind of subscription model a reason for people to keep spending money yeah um, also, I felt like 23andme was very arbitrary in what they're telling you because, like, how far back do you want to go? Yeah, right, because at the end of the day, we're all from the same place.

02:39:10 - Leo Laporte (Host)
Right, right yeah.

02:39:13 - Denise Howell (Guest)
It's just such a cautionary tale. Whenever you're giving any kind of sensitive data to a private company, just know that you know this is a temporary relationship between you and that company and private equity exists. They're bound to gobble your company up at some point.

02:39:33 - Parmy Olson (Guest)
I wasn't too familiar with. I'm not too familiar with 23andMe, but just like reading the reports on it, it really strikes me this is just another example of this. The whole culture of the vision that the founder knows best in Silicon Valley, you know, and that's why founders of tech companies like Mark Zuckerberg have, like you know, more than 50% voting control of Meta and the Google founders had a lot of control of the company is this prevailing belief that the founder has a long term vision and should be allowed to see that vision through, and usually investors will get behind them, but in this case not. And it was interesting to read in the report that Anne Wojcicki still has some kind of. She says we have to stay. The course's going to be difficult, but we can make this work.

02:40:25 - Leo Laporte (Host)
Um, but it's really unclear what her strategy is I hope she has, and you know her sister recently passed. She was the former susan was just his former ceo of youtube, the uh her.

02:40:37
Their mom, esther wojcicki, rented the family garage to a couple of Stanford graduate students named Sergey Brin and Larry Page, and in fact, that's how Susan Wajiski became an early employee at Google. So they're smart people. I would hope that Ann Wajiski would have some integrity and if it looked like the company was going to go under, not sell the spit, do not sell the spit, throw it out. But you can see, maybe, that the value of the company is the fact that they hold all these Right.

02:41:26 - Denise Howell (Guest)
Lewis Columbus has an interesting article at VentureBeat where he's basically just reporting on what's going on here, but he ends it with five strategies where to start to take this forward, and whether these are aimed at 23andMe or at whoever acquires this company or its assets, it's spit. But all five are very security and privacy focused, so he seems to be saying that's where they drop the ball is figure out first how to keep this data secure and then, I guess, get to the problem of how you're going to charge them more than one time.

02:42:05 - Leo Laporte (Host)
Yeah, I had a guy named you can show that George Churchon, who is considered by many the father of genomics. He started a company called Nebula Genomics. Same thing, I think. I sent him spit, but they, unlike 23andMe, they actually give you and you can download your entire genome. It's a huge, it's terabytes. Uh, they decode 100 of your dna, which 23 and me does not do. Uh, and they say we guarantee your privacy. But I'm such a sucker. I just you know. Fine, I sent him my spit. Um, somebody pointed out to me, you know, it's not just you, leo, it's your kids, because they share your genome. So you might be revealing something about your children that would be me.

02:42:48 - Denise Howell (Guest)
I pointed that out to you, was it you? Thank you, see again.

02:42:52 - Leo Laporte (Host)
I should listen to my online attorney. Um, yeah, so it's.

02:42:57 - Denise Howell (Guest)
It's your children, it's your cousins you know nebula.

02:43:01 - Leo Laporte (Host)
So 23 and me does what's called genotyping same thing with ancestry DNA and it's just a small fraction of your DNA. These guys decode 100% of the DNA, which is why it's so expensive. They do a huge job on this and Church made a really good argument. By the way, he's also the guy who wants to bring back woolly mammoths to save the climate. So if you're interested, search church, george, church, woolly mammoths. It's fascinating, the Woolly Mammoth Project. But his point he said everybody should do this before they get married, for instance, to make sure that you and your prospective spouse aren't going to propagate some genetic flaw to your kids. So you at least have the knowledge that there's that risk. We are kind of wandering blithely into, you know, into parenthood without a lot of information that we can today get. I didn't learn anything particular from uh doing the nebula thing, but yeah, again I'd I.

02:44:02 - Denise Howell (Guest)
I just do it medically. If you're concerned about propagating a, you know, have your doctor run that test through a lab I don't do they will.

02:44:13 - Leo Laporte (Host)
Doctors offer that? They do when you're pregnant, that's for sure, oh nice oh, yeah, yeah for things like sickle cell and yeah, yeah, yeah, yeah that makes leo just get pregnant uh, let's see what else is going on. Anything else you want to talk about before we wrap this up? I know it's what is it? It's it's late, it's midnight in the uk, almost 1 am. I'm so sorry, but thank you and is your daughter going to come back to her bedroom or is she going to sleep on the couch tonight?

02:44:44 - Parmy Olson (Guest)
once she's out, she's out oh, that's good she will sleep right. She's fine on the couch please, parmy.

02:44:50 - Leo Laporte (Host)
Thank you. Thank you, your wonderful daughter, for letting you have her room. Is it's the only room you could do a podcast from? Is that it?

02:44:57 - Parmy Olson (Guest)
yeah, yeah, yeah, pretty much, you sound great.

02:44:59 - Leo Laporte (Host)
Two other kids and yeah, it's great, it's wonderful. Uh. The book supremacy is out now from st martin's press. Highly recommended I was it's. It's really uh engrossing the story uh, and for somebody like me who has to cover ai, it really helps to kind of understand the personalities and their goals and all of that uh, very nicely done. Thank you for being on the show. Supremacies and bookstores. Now you'll see parmi's writing at Bloomberg Opinion as well and other places. Former Wall Street Journal reporter. Thank you for being here, parmi, I appreciate it.

02:45:32 - Parmy Olson (Guest)
Oh, thanks for having me.

02:45:33 - Leo Laporte (Host)
My pleasure you can go to bed now. I'm sorry to keep you up so late. Daniel Rubino, always a pleasure to have you on Editor-in-chief at windows central. Keeping up with microsoft these days is a full-time job. Thank you for doing it. Uh, anything you'd like to plug coming up?

02:45:52 - Daniel Rubino (Guest)
oh, I'm probably doing a podcast this week, uh, because we expect some big news from microsoft, so you said that several times.

02:45:58 - Leo Laporte (Host)
Can you give? Us a hint of any kind.

02:46:01 - Daniel Rubino (Guest)
New hardware, new software, 24H2, anything you want to talk about 24H2, yeah, Some of the topics we touched upon here we've kind of alluded to so, yeah, it's going to be really interesting, it's yeah, it's cool stuff.

02:46:17 - Leo Laporte (Host)
Also, Prime Day coverage right, we got Prime Day 2 coming up. That's got to be the thing you dread the most.

02:46:24 - Denise Howell (Guest)
We always had two.

02:46:25 - Benito (Announcement)
I always thought there no no, no, this is no this is days.

02:46:29 - Daniel Rubino (Guest)
It's the second yeah, yeah.

02:46:31 - Leo Laporte (Host)
You can't go to amazon with a big, big banner saying yeah.

02:46:34 - Daniel Rubino (Guest)
A week from tuesday yep, and now it's funny too, because walmart and best buy all do counter programming. I mean it's good for consumers, I guess, but it's also this weird thing of like feels like you entrapping people to go buy things. I agree.

02:46:50 - Leo Laporte (Host)
Great article.

02:46:50
There's a lot of cool tech out there to buy, so Great article on the front page of Windows Central about Sam Altman podcasting. Bro, I love it. Thank you so much for being here. It's always a pleasure, daniel. Thank you so much for being here. It's always a pleasure, daniel, thank you, and Denise Howell thank you so much. We appreciate your being here. Denise is at hearsayculturecom. That's where her two podcasts are, including the one that I was a part of and anything else you would like to plug. The most recent interview sounds very interesting.

02:47:23 - Denise Howell (Guest)
Yeah, definitely Give that a listen. Um. Denisehowellinfo is my main website. Nice, I'm d howell, most places online thanks for joining us, really appreciate it.

02:47:35 - Leo Laporte (Host)
Lisa left her uh, her water-based curling iron on the boat and we're very sad.

02:47:40 - Denise Howell (Guest)
No, no, we have to buy another one.

02:47:43 - Leo Laporte (Host)
Denise, among all of her other things, gives us hair care tips, so it's. She takes no responsibility for my hair, however. That's all my fault. I know I need a haircut.

02:47:55 - Denise Howell (Guest)
My barber moved I'm one of those people that it. Had it were I not using something like a steam powered straightening iron, my hair would look like rosanna rosanna dana yeah, same with lisa.

02:48:06 - Leo Laporte (Host)
She's got very, very big curly hair. Yeah, she loves that thing. Yeah, it's a good one. Is it from dyson? Who made it?

02:48:14 - Denise Howell (Guest)
no, it's uh l'oreal l'oreal, that's right yep, it's called the steam pod okay, so yes oh cool I've got the dyson air wrap.

02:48:23 - Parmy Olson (Guest)
It's very good yeah yeah, yeah, I have.

02:48:25 - Denise Howell (Guest)
I have air wrap envy.

02:48:27 - Parmy Olson (Guest)
Haven't tried it yet, oh man you got to try it, I will see we got everything on this show hey, it's their technology.

02:48:38 - Leo Laporte (Host)
No, absolutely it's the latest stuff. Uh, we do twit every Sunday, 2 to 5 pm. Sunday evening. That's Pacific time, 5 to 8 Eastern time and, for those of you in the UK, 2100 UTC. I'm so sorry, late, late, late.

02:48:53
You can watch us do it live on seven different platforms YouTube, twitch, kik, xcom, facebook, linkedin and for our club members right there in the Discord. If you're not a club member, first of all, a big thank you to our club members for making all of the shows we do possible, and if you're not a club member, I'd love to invite you to join. We keep the price low $7 a month, it's nothing. You get ad-free versions of all the shows. We don't need to show you ads because you're already a club member. You get free versions of all the shows. We don't need to show you ads because you're already a club member.

02:49:26
You get access to the club twit discord special programming we only do in the discord. For instance, if you missed my photo uh review with chris marquart, we talked about photography news on friday that is available to club members in the twit plus feed. Lots more stuff like that, and. But the best reason to do it is because it keeps us going. We don't. We don't do the fake reviews, we don't do the link bait, none of that stuff. We try to give you great content with high integrity, the best information, and I think it's worth seven bucks a month, if you agree.

02:49:59
Twittv, slash club, twit, and thank you. I really appreciate it. After the fact, on-demand versions of the show are available at our website, twittv. There is a youtube channel dedicated to the video. In fact, you'll find the link there at twittv. But the best way to get all of our shows is subscribe and your favorite podcast client, uh, you can subscribe to audio or video. Uh, and that way you'll have it. If you subscribe now, uh, every monday morning for your, your beginning of your day, weekly commute if you actually commute I just go upstairs my commute is very short.

02:50:33
Thanks for joining us everyone. Thanks especially, uh, to our great panel, denise howell, uh and parmi olson great to have you, daniel Rubino. Thanks to all of you. Next episode, episode 1000. But as I have said, now for 999 episodes for 20 years. Thanks for joining us. We'll see you next time. Another twit is in the can Bye-bye.