Tech News Weekly 385 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 - Mikah Sargent (Host)
Coming up on Tech News Weekly. Abrar Al-Hiti is here. We talk about ChatGPT adding shopping to its search, then where things stand with Apple, the Epic and the App Store. After that, leon Island antitrust reporter at Bloomberg stops by to tell us about the antitrust case with Google and its remediation portion portion, before we talk to the head of research at the Wikimedia Foundation, leila Zia, about Wikipedia's AI strategy. It's a fantastic show, so stay tuned for the episode Podcasts you love From people you trust. This is TWiT. This is TWIT, and welcome to Tech News Weekly, the show where every week, we talk to and about the people making and breaking the tech news. I am one of your hosts this week, micah Sargent, and I am joined across the Internet by the wonderful Abrar Alhiti of CNET. Welcome back, abrar.

01:17 - Abrar Al-Heeti (Co-host)
Thank you so much. It's good to good to have you back.

01:22 - Mikah Sargent (Host)
Yeah, I was going to say it should be welcome back to me. Thank you so much for doing such a fantastic job hosting the show. I heard lots of great feedback. Maybe calls to just have you be there and I just go.

01:34 - Abrar Al-Heeti (Co-host)
I don my best I think you did wonderfully.

01:43 - Mikah Sargent (Host)
Thank you so much again for doing that, of course. So hopefully many of you are return guests and are very aware of how this works, but if you aren't this is the first time here Welcome. There's nothing under your seat, but there is certainly a lot to enjoy. So we kick off the show by talking about our stories of the week the compelling, interesting, exciting things that we've seen in tech news that we want to talk about, and so, without further ado, let's hear about it, abrar.

02:15 - Abrar Al-Heeti (Co-host)
Okay, so I am mixing things up by not talking about self-driving cars for once, or things in my usual wheelhouse. I realized there's more to technology. So I wanted to talk about ChatGPT Search's new shopping experience, and I'll caveat this by saying that I am not an avid user of ChatGPT or ChatGPT Search, but I was still intrigued by this, because shopping is something that you know. When we think of online searches for shopping, people tend to go to Google and see what comes up there and you click on products or you go to you know a website or an influencer or whatever it is that you choose to select a product that is the best fit for you. So now ChatGPT Search wants to be that source for you, so you can search for whatever product you're looking for, compare options. You can buy directly within ChatGPT. It'll show you those direct links so you never have to leave and you get these.

03:17
You know the idea is, it's supposed to know you pretty well, right? So you get these personalized recommendations, you get all these details and pricing, and this is starting out with fashion and beauty and home goods and electronics, and then eventually will expand to more categories, and it'll remember. This is the part where the magic and the craziness and the creepiness of ChatGPT comes in. It'll remember your past shopping experience. It's the same way that Google would right and pick up on context from your previous conversations to give you that more personalized recommendation. So, micah, are you an avid ChatGPT search user and is this something that you feel like you would use as an alternative to whatever you currently use to kind of find the best products for you?

04:03 - Mikah Sargent (Host)
So a couple of things. I do use Chet GPT a lot. I do not like to use it for search. I don't like to use AI search at all, if I can help it. I continue to see compelling arguments that there have been improvements to the AI search, but anecdotally, no, it's not been my experience, and so, just as it was the case when I tried Siri out for the first time and it disappointed me four or five times and I don't use it now, the same thing applies here, where I just didn't like the results that I got.

04:42
What I will say is that one place that I have seen AI do a great job in the sort of search area is the research functionality, and for anyone who hasn't tried that, it's essentially where you set it on a course of completing research for you across the cross the internet, and so I, for example, tasked Google's AI research with helping me figure out the best cuttable sugar cookie that is also gluten-free, and so our research plan I say our, because it kind of tells you what it thinks it should do, and then you say no, no, I want you to do this involved me basically saying I want you to look at all of the different um, gluten-free sugar cookie recipes that are out there and I want you to find commonalities between them, uh, so that we can make a sugar cookie. That is because the problem with gluten-free ones is that they often kind of like, get runny. So let's find the one that's the most stable and steadfast, and then advice on how to make them even more stable. And so it was able to go to, you know, dozens upon dozens of websites and collect information from there, collect information from the comments, that kind of a thing. So that is a really cool aspect. That's sort of like the search functionality, and I think that if it were to come to shopping, I would love that, because I don't know, I would love to hear, like your, your technique for, um, finding products.

06:18
But mine always starts at Wirecutter. Uh, they do a lot of research and I have. I don't think I've ever purchased something that was a Wirecutter recommendation and then been unhappy with it. I know people have, but I have not been one of those people, and so that's my first source of truth and trust. Yeah, and then from there, you know, I just do what I can to try to find what I'm looking for. But if I could, there are times where Wirecutter doesn't have the answer. So if I could turn to and in fact I tried to do that the other day, now that I think about it Um, and yeah, I like the idea of it having kind of awareness of what's out there and then giving me some information based on that.

07:03 - Abrar Al-Heeti (Co-host)
Yeah, absolutely. I think my method as somebody this is a true Testament to how addicted to TikTok I am. But that's usually where I go because I um, I like to hear from people who I can then see the product and then I can hear about it. But I'm also always skeptical of people who are like big time influencers. But you get such a range of content, from videos that have like 10 likes to some that have like thousands, and so you kind of get this broad range of like.

07:29
This person just really enjoyed this product and decided to make a video about it and it's already been on my mind. So maybe that's something that you know I can dig into a little bit more. And and then sometimes it really is just like going straight to Amazon and and that's the other piece of this that I thought about because Amazon had launched something very recently called Interests and this is supposed to also kind of work in the background and give you recommendations about what you like based off of what it's learned about you right? So, oh, this person's really into football or whatever it may be, whatever sport, whatever. For me, it's going to recommend a bunch of Harry Potter stuff. Whatever it's going to be right for me. It's going to recommend a bunch of Harry Potter stuff, whatever. It's going to be Right, so, and it will.

08:11
You know, there's so much, there's so much stuff on Amazon, right, and then, obviously, it's limited to what's actually on there, but it's so much, it's so broad, and then it kind of tailors it based on your price limits and things like that. So I'm wondering, you know, it's it's twofold. For me, it's how is Amazon feeling about this, and how is Google obviously feeling about this? Because I think Google is already facing this reckoning of, okay, things like chat, gpt, search and in other platforms are already starting to chip away at what I think they probably thought was oh like, we rule this, this is our territory, but there are all these alternative means that people are turning not just for search, but now for shopping. So I have a feeling this is something that will resonate with a lot of people, especially given how many people love Chow, gpt and so, yeah. The other piece of this, though, is that OpenAI says it doesn't make any affiliate revenue off of this, and I have a feeling this has to change right Like at some point.

09:11 - Mikah Sargent (Host)
Yeah, I was going to say, if it doesn't change, then they are. What they're probably doing is sort of proof of concept idea where, because they are known for making deals with different organizations behind the scenes, right, so even if they don't do it, uh, kind of in the foreground I wouldn't be surprised if there is sort of a background of, hey, we're ending up going here and we're sending people here quite a bit, let's talk about this, um, because, yeah, I, there's, there's some affiliate that plays into making things a little less trustworthy, right, and that plays a role for me for sure.

10:01 - Abrar Al-Heeti (Co-host)
Definitely. And then the other piece that I'm wondering. You know you mentioned going to Wirecutter. I mentioned kind of going to people online. Piece of that I'm wondering. You know you mentioned going to Wirecutter. I mentioned kind of going to people online. If something like ChatGPT is in charge of recommending things to you, how do you think it can and do you think it even matters? Is it a concern that people might be like oh well, is there any selectiveness or bias in terms of what it's suggesting, especially as things like affiliate programs are launched through this? Do you think it'll ever really replace that? You know, here's a person that's recommending this product, whether it's through an outlet or a social media platform.

10:37 - Mikah Sargent (Host)
That's having a person do it and having a person who can show, because there's a difference right between synthesizing information that's available online versus real lived experiences and that does like, as things stand, until we get the robotic AI area where suddenly we're having a robot grab the handle of a vacuum and, you know, run it across a floor and then it's able to pipe that into chat, gpt and say, oh, this feels uncomfortable as it stands. That's not a thing. That's possible. And if the human being is testing this and feeling the pain points that a human being feels, that's going to play a role in, in what's going on. And, like we know, with with wirecutter, for example I keep going back to it but with many sites, with cnet, with uh, our site, the stuff that we talk about has affiliate links and we make it very clear that that's the case.

11:42
That's just kind of the what, what is a brass tacks of a situation. But when it comes to actually, I guess there's a difference, I feel, between humans who can make certain choices, versus an AI. That what if, based on the AI's training, it figures out that its role is to make as much money for the company as possible? You know what I mean, there are all these little things that you just don't know, that are happening in the background, and that's where it catches me off where I'm going. Maybe I don't want to trust this for these decisions. Not because it's doing anything nefarious per se it's not going to get me to buy a vacuum that's going to explode in my hand, knock on wood but because it's maybe not the best choice.

12:36 - Abrar Al-Heeti (Co-host)
Yeah, no, that's a great point, I think, having a human be the one to kind of hedge that and say, no, I really do like this product and my company cares if we make money, but I, I'm just here to tell you about this really cool thing. Um, but um, but yeah, the other. The one other thing I was wondering about was if, when you go on Amazon, if you're a frequent shop at all, if you've ever tapped into Rufus, which is always bothering me, but it's the AI powered assistant that pops up and is like can I ask, answer questions for you? That's another piece of AI that I'm like do we want this?

13:09 - Mikah Sargent (Host)
I have never used Rufus. In fact, I don't know how I was able to avoid it, but I don't have it popping up for me. However, because you mentioned it, I clicked on it and now it's probably gonna pop up a lot.

13:20 - Abrar Al-Heeti (Co-host)
There you go. It's the new. It's just gonna bug you for no reason.

13:29 - Mikah Sargent (Host)
Let going to bug you for no reason. Let's see how can I keep kids busy on a rainy day. That's very much something I need to know. I've been wondering Best toys for a toddler who loves trains. I've got one of those, and they do love trains. Best dog Okay, best dog treats for training? Let's see what they say about that. Rufus says, see, but yeah, is Rufus being honest? Or is this what Rufus thinks is the best treat for training?

13:52 - Abrar Al-Heeti (Co-host)
Who knows?

13:53 - Mikah Sargent (Host)
Oh, this contains soft lamb and salmon, which is ideal for puppies. Yeah, these are all prime items. Will it ever suggest non-prime items? Is this an antitrust case waiting to happen?

14:10 - Abrar Al-Heeti (Co-host)
We'll have to wait and see.

14:11 - Mikah Sargent (Host)
Yes, there we go the wait and see of it all. Well, I'm going to keep an eye on ChatGPT search and see if that improves. I think I'll try to throw a few more tests at it when it comes to product recommendations and see what it spits out over time. But speaking of time, it is time for us to take a little break. This episode of Tech News Weekly this week brought to you by Delete Me.

14:41
If you ever wonder how much of your personal data is out there on the internet for anyone to see and then you went to look yeah, you probably realized it's more than you think. Your name, your contact information, your social security number, your home address, even information about your family members it's all being compiled by data brokers and sold online. Anyone on the web can buy your private details Anyone. I could. I'm not going to, but I could. This is going to lead to identity theft, phishing attempts, doxing, harassment, but now you can protect your privacy with Delete Me.

15:16
As a person who exists publicly and is, as they say, extremely online, and especially as someone who shares my opinions online, I have to think about safety and security, and I've had to for years. But with Deleteme, it's easier than ever to find personal information and get rid of it online. That's why I recommend and use Deleteme. Deleteme is a subscription service that removes your personal information from hundreds of data brokers. You can sign up and provide Deleteme with exactly the information you want deleted and then delete me, as experts Take it from there. Delete me sends you regular personalized privacy reports showing what info they found, where they found it and what they removed. And delete me isn't just a one-time service. That's important because the data brokers are not just a one-time service. They keep gobbling up your information. Delete me is always working for you, constantly monitoring and removing the personal information you don't want on the internet.

16:13
I have been using this for a while now, and one of the things that stood out to me was looking up my name before I started using DeleteMe and seeing it listing my family members. That is concerning to me, because there are lots of opportunities for people to get in touch with my family members and then say oh, you know, this is me, grandma, this is Micah, and I need you to send this amount of money to this place because I'm you know, whatever it happens to be, there are those kinds of horrible scams and schemes that are at play, and being able to kind of take control of that is important for those reasons, but I also think it's just important in general If you want just a little bit of satisfaction in fighting back at these companies whose sole purpose it is to collect your data and sell it to others. Oh boy, how does it feel good when you get that little report saying and we took this down, and we took this down, and we took this down and I say thank you, thank you, thank you. You are doing the work. It's wonderful. So, to put it simply, delete Me does all the hard work of wiping your and your family's personal information from data broker websites. So take control of your data. Keep your private life private.

17:29
By signing up for Delete Me at a special discount for our listeners Today, you can get 20% off of your Delete Me plan when you go to joindeletemecom slash twit and use the promo code twit at checkout. The only way to get 20% off is to go to joindeletemecom slash twit and enter the code twit at checkout. That's joindelete mecom slash twit with the code twit T-W-I-T. And we thank Delete Me for sponsoring this week's episode of Tech News Weekly. All right, we are back from the break, joined by Abrar Al-Hiti this week on this the first Thursday of the month, and it is time to talk about my story of the week.

18:14
A federal judge has ruled that Apple can no longer charge fees or impose restrictions on purchases made outside iOS apps. This is a direct result of the company's long running legal battle with Epic Games apps. This is a direct result of the company's long running legal battle with Epic Games. The ruling enforces and strengthens a 2021 injunction that Apple, according to the court, failed to follow. This is pretty big, which is why the headline from the Verge reads a judge just blew up Apple's control of the App Store. Basically, what it looks like we have here is a rather upset judge. The judge looked at the original injunction, looked at what's come out of it and what has happened thus far, and said Apple, you were not following the rules that we set, so here are some new rules for you.

19:05
Judge Yvonne Gonzalez-Rogers ruled that Apple may not impose, quote any commission or any fee on purchases that consumers make outside an app, restrict style, formatting or placement of links to outside payment options, block buttons or other calls to action, or interfere with users, leaving an app beyond a neutral message apprising users that they are going to a third party site. So this is something else, because up to this point Apple has had in its rules and it's like rules for developers that you as an app developer cannot just have a straight up link to pop out somewhere else and make payments, and so this whole time there's been this sort of conversation about how different sites can do this, and you may have run into issues before when you were trying to buy a book or you are trying to get some other content where you couldn't get it in the app and you're going well, what do I do? How do I get to it? And you have to go to the website in Safari or something like that. That's what this was up to this point. Now this judge is saying you can say, hey, you're leaving this page and going elsewhere, but that's the extent of it.

20:24
The court found Apple in willful violation of the injunction. This is what Gonzalez-Rogers had to say in her words. Apple thought this court would tolerate such insubordinate. Wait, let me try that again. That Apple thought this court would tolerate such insubordination was a gross miscalculation. Can we talk about how that is? Like the opening of a great rap Like I want this to be turned into a musical or something that mm delight.

20:55 - Abrar Al-Heeti (Co-host)
That's the next Hamilton. What are you talking about?

20:56 - Mikah Sargent (Host)
Yes, it is the next Hamilton. She also noted that Apple acted quote solely to maintain its revenue stream and referred the case to the US attorney for possible criminal contempt proceedings. Judge is not happy. Yeah, judge is not happy. A few other things. Apparently, within the company there was some conflict on which way things should go. Evidence suggests that Phil Schiller supported complying with the court's order. Judge Gonzalez Rogers wrote Tim Cook ignored Schiller and instead allowed Chief Financial Officer Luca Maestri and his finance team to convince him otherwise, and I quote Cook chose poorly. There's also the 27% fee for commission for outside purchases. After being told in 2021 to allow developers to link to outside payment systems, apple added a 27 percent commission. Again, tied to nothing. Apple's goal maintain its anti-competitive revenue stream.

22:15
Judging happy and honestly, it's kind of hard to. You know, it's kind of sort of our job to figure out the scope of this and not make unfair conclusions. Right, but this is very compelling and I think that people who've been talking about this for a long time have been saying this for a long time that Apple's solution was one that kind of flew in the face of the spirit of what they were trying to do. But they I mean the legal system trying to do, realizing that it can't just do that delay, deflect, double down approach that has been very common in big tech up to this point. I wanted to ask you kind of your experience with payment platforms or in-app purchases and that kind of a thing and where you may have found these pain points and this friction, and then just your general take about this Are you sipping the tea and eating some popcorn as you're looking through this? As much as I am.

23:40 - Abrar Al-Heeti (Co-host)
It's been years of this just unfolding. You see, oh, this is the next plot twist. So it is interesting to see this and the court is not ready to just let it slide. I was thinking about so I am not a gamer, but I've been following this specific case just kind of more from a distance.

23:56
But in terms of my own app store purchases, I don't think I'm someone who I'm like oh yeah, this is going to help me a lot. But I think it's more of the spirit of it of, like you know, reducing those friction points and making it easier and more open for various developers to just kind of facilitate that. I mean, I think a 30% fee is ridiculous, I think that's such a hindrance, and so the 27 wasn't much better. So I'm curious, I think I wanted to ask you actually how you think that you would be able to benefit from this. I don't think I tap into the app store enough. I'm just like I'm on social media too much, so I don't think I take full advantage of of um, of all that my phone could potentially do. But, um, but how do you see this kind of um helping your experience?

24:53 - Mikah Sargent (Host)
Well, it's interesting because I would have had a very easy answer for you. Um, until I think, mid last year, maybe a little bit later. Um, because I am a voracious audio book consumer and I I basically, when my hands are free to do things and there's silence around, there's not silence around me, there's audio books, and so I'm like always listening to audiobooks. It's how I fall asleep, uh, etc. Etc. And so, up to this point, uh, if you wanted to buy an audiobook, uh, I use audible, uh, former sponsor of the network, I should point out um, I would have to go online and make the purchase there instead of being able to do it in the app. Amazon and Audible ended up sort of rolling over but we don't know if there was a special deal, we don't know exactly and finally started allowing purchases of book, audio books, within the app itself.

26:07 - Abrar Al-Heeti (Co-host)
I actually just did that yesterday. I realize now and it was like just double click and ready to go.

26:12 - Mikah Sargent (Host)
So it was seamless, and so that would have been in the before times, when Amazon had not added that functionality. This would have been helpful because up to that point, there was no button to be like take me to this book title in the Safari web browser and let me buy it there. You literally had to just open up Safari, audiblecom, do a search for the book and then you could get it. And yeah, so convoluted and annoying and friction. And so now that that change has happened, I don't really there's not a whole lot of stuff, and I do have to say that there's something to be said for if you just go to a random site and put in your information.

27:15
Two, and I think most importantly, when it comes to subscriptions, no one makes it easier to cancel a subscription than Apple does. And they're all right there in one list. I can change them from there, from monthly to yearly. I can set up a notification that I get an email like that Fantastic, yeah, that is fantastic and I love that. And so in that way, I do worry about people. You know, using this new way of things, it's definitely going to end up happening. And then there are more scams available, there's this and that, and I understand that that's the argument that Apple has made and that's, you know, that's the problem and et cetera. But if there's any grounds to it, I think that that is where I feel Apple is correct. But the fact is everyone using this as a human being who could make their own choices and should not be stifled from being able to make those choices.

28:12 - Abrar Al-Heeti (Co-host)
Yeah, absolutely. I think I didn't even realize how much of a step forward my purchase yesterday must have been, given all of this. So I usually, if I use Audible, I usually just wait for my credit to roll in. But yesterday I was like, no, I need this audio book, I'm going to pay the $15. And it took two seconds and it was painful, but it was worth it. Yeah.

28:33 - Mikah Sargent (Host)
Yeah, you're like dang it. I don't have it, but I really want it. Yeah, I know, sometimes I'm like I get to the last, next one. Yeah, I'm going to get the next one in this area.

28:47 - Abrar Al-Heeti (Co-host)
I know, and then Libby takes too long, so you're like, ah fine, I'll pay for it, absolutely. But that was to think that it could be that seamless, you know, and and uh, easier and not take a chunk out of you know what other people could be making platforms. That's great.

29:08 - Mikah Sargent (Host)
Absolutely Well a bra. I want to thank you so much for being here with us this week and again, thank you so much for hosting the show while I was away. Of course, people can head over to CNETcom to check out the work that you're doing. Is there anywhere else they should go to keep up with what you've got going on.

29:20 - Abrar Al-Heeti (Co-host)
You can check out my account on Instagram Abrar El-Hiti no spaces. Also on TikTok, same thing Abraer Elhiti no spaces. And I'm on X every now and then Elhiti, underscore three.

29:31 - Mikah Sargent (Host)
Beautiful. Thank you so much. Thank you See you soon. Bye, all righty folks. We're going to take another quick break before we come back with the antitrust reporter. Extraordinaire from Bloomberg, but we'll talk about that in a moment. Extraordinaire from Bloomberg, but we'll talk about that in a moment.

29:48
I want to tell you about this episode of Tech News Weekly being brought to you by Drata. If you are leading risk and compliance at your company, well, you're likely wearing 10 hats at once, and boy do those hats look heavy. You're managing security risks, you're managing compliance demands, you're managing budget constraints, all while trying not to be seen as the roadblock that slows the business down. But GRC isn't just about checking boxes. It's a revenue driver that builds trust, that accelerates deals, that strengthens security, and that is why modern GRC leaders turn to Drada. It's the trust management platform that automates tedious tasks so you can focus on reducing risk, proving compliance and scaling your program. With Drada, you can automate security questionnaires, evidence collection and compliance tracking. You can stay audit ready with real-time monitoring and you can simplify security reviews with Drada's Trust Center and AI-powered questionnaire assistance. Instead of spending hours proving trust, build it faster with Drada.

30:54
Ready to modernize your GRC program, visit dradacom slash technews to learn more, and we thank Drada for sponsoring this week's episode of Tech News Weekly. All righty, we are back from the break and I am very excited to say that we are joined by antitrust reporter extraordinaire Leah Nyland. Welcome back to the show, leah.

31:17
Thanks for having me, Absolutely so. When you were last on the show, I believe, we talked about the Justice Department's proposal to force Google to sell Chrome. Could you give us a little update on how that proposal has evolved since then and where things stand now in the remedies phase of the trial? And I have to know too do you and all of your antitrust friends love the remedy portion? Because to me that's so interesting.

31:44 - Leah Nylen (Guest)
I mean it is fun because it's a little bit more like forward looking and talking about technology as it exists today and how it might exist in the future, as opposed to, like you know, rehashing stuff that happened back in like 2009. But, yes, so the Justice Department revised its proposed remedies a little bit in March. They are still seeking a divestiture of the Chrome browser, but they changed. The big change was they eliminated a portion that would have prohibited Google from making investments in AI startups. That had been opposed by a number of AI companies, including Anthropic, which already has some investment from Google. They said it would really, you know, harm the AI ecosystem for Google not to be able to invest in companies, in part because a lot of the investment that Google has been giving to companies is in the form of what they call sort of cloud credits, so they've essentially been giving them free access to their cloud infrastructure and that's, you know, a really big part of what AI companies often have to pay for getting all of that computer space. So that was sort of where we started two weeks ago or a week and a half ago, when this remedies trial started on April 21. We are now a week and a half in, and the Justice Department just earlier this week, on Tuesday, finished its sort of presentation. So for the first week, the Justice Department, presented earlier this week on Tuesday, finished its sort of presentation. So for the first week, the Justice Department presented testimony from a number of different market participants, along with a few Google employees, about how its remedies would sort of change the ecosystem.

33:22
There are sort of like three big buckets. One of them is obviously the Chrome Investiture. The second big bucket is the Justice Department wants to force Google to share some of the data that underlies its search results. This it would have to give to what are called qualified competitors, so the government would get to decide who it goes to. It doesn't get to go to just anybody, but if somebody is labeled a qualified competitor, they would get all of the data, the ranking signals and things that underlie Google Search, the idea being that other search engines or other companies could use that information to either create their own search engines or improve their own search results.

34:05
Or, in the case of AI companies, it sort of gives them a leg up on trying to create large language models that might compete with Google's. And then the third big bucket has to do with Google's contracts which were sort of at the center of the original case. So Google had all of these exclusive contracts with smartphone makers and browser makers to make its search engine the default, with smartphone makers and browser makers to make its search engine the default. The Justice Department wants to bar Google from paying companies for that position and additionally they want to bar Google from paying to make its AI app Gemini the default on browsers or phone makers. Google says obviously it's not a big fan of that proposal. It says it's not really fair because AI is still a very burgeoning space and it has not been found to be a monopolist in AI. But those are sort of the three big buckets of what the Justice Department has been presenting evidence for the past week and a half.

35:03 - Mikah Sargent (Host)
Understood. Wow, I really appreciate you because of how well you take what is just these huge topics and make them easy for us to understand, so thank you. Sundar Pichai, of course, has testified at this point, calling the government's data sharing remedy quote a de facto divestiture of search. Can you talk about the tone of his overall testimony and how Google is positioning its defense at this point?

35:32 - Leah Nylen (Guest)
Yeah, it was very interesting. This is the third time that Sundar Pichai has had to testify in as many years in these antitrust cases. He, of course, testified during the original trial in this case back in 2023. He also testified in another lawsuit over the company's alleged monopolization of the Android system, and then now he's testified again. This was by far the shortest one he was only on the stand for about an hour and a half but he took a very serious tone in talking about how he thought this would ultimately damage both Google and the US's stature in innovation and national security. So he said, by forcing us to divest Chrome and give away this data that we have spent years building, it would ultimately undermine Google as a company. You know they have invested you know he said millions and billions of dollars over the years in building up the search infrastructure and all of the things that go into the search engine, and the Justice Department would be requiring them to give it away, not for free, but for for very little money, and that that, you know, is going to hurt the company's ability to invest further in the future because it's not going to be earning as much money. It's going to have to face more competition from these other folks. That in turn, he said, could potentially harm national security, because you know the US, google and the US are technological leaders. As of right now, the US has not put any limits on the nationality of qualified competitors. It hasn't said that you have to be a US company in order to access this data, though that could potentially, in the future, be a condition. And then they also said that this would definitely damage their ability to compete in AI, because the Justice Department is seeking some of these restrictions on how it distributes its AI model. Right now.

37:30
We learned from the trial, actually, that Google has created an agreement with Samsung in which it is going to be paying billions of dollars to Samsung to put Gemini as the default AI assistant on Samsung's new versions of phones, on Samsung's new versions of phones.

37:46
But Google has stressed repeatedly that these are not exclusive contracts, that these phone makers can also try and install another AI system their own or a different one and users would have the option to choose between them. Some of the other AI companies that testified said well, yes, you might be able to get on the device, but Google's contracts have actually been blocking them from getting the default. So we heard from Perplexity AI, which had actually reached an agreement with Motorola that on all of the new Motorola phones Perplexity is going to be installed in there. Both Motorola and Perplexity actually wanted the company to be the default AI assistant, but Google's contract sort of prevented that. So perplexity will be an option, but not sort of the pre-installed option that comes, you know, pre-working on the phone. So there's been a lot of a little bit in the weeds about whether Google's contracts really prevent this or whether this is a misreading of them, and it probably will end up being up to what the judge decides at the end of the day.

38:50 - Mikah Sargent (Host)
Understood One kind of, I think, interesting part of this aspect, because you talked about how you know we heard from other executives OpenAI, Perplexity, duckduckgo, all came in to testify. What was the overall point, I guess, in terms of the court, in understanding how Google's practices affect competitors? And does it seem like those testimonies have had an impact on the remedies portion of the case? Because, speaking just from my own, I think it would be difficult to sort of figure out what remedies like, to even to even fathom what it means to find remediation in the situation. In the situation so, is it often the case that competitors coming in to testify helps to kind of give the court somewhere to go when it comes to the remediation?

39:55 - Leah Nylen (Guest)
Yeah, so this is a little bit of a new proceeding. So what had happened? You know this case is very much based on the Microsoft case. That happened 25 years ago. But what happened in Microsoft is the judge didn't have a remedies hearing and the court really ended up dinging the judge for that and said you should have had a separate entire hearing before you decided what to do. So here the judge is like a little bit breaking new ground with like how to do this and what to do.

40:23
So the Justice Department brought these witnesses in, as you mentioned, openai, perplexity, duckduckgo and also Yahoo. Fun fact, yahoo still exists and what they had them testify about was one how Google's current practices impact their business. So all of the companies talked about how, you know, google's contracts limit their ability to get distribution through their own contracts with manufacturers, which means that it's more difficult, in their view, for consumers to discover their products. And then they talked about how they think the proposals that the Justice Department put forward would help their business. So all of the AI companies and then the two other search engines you know, sort of described how having either access to Google's data or the ability to, in their view, more fairly compete for defaults on search engines and browsers would sort of increase their ability to compete with Google and make it easier for consumers to discover them. And what was very, very interesting is all four of those companies said that they would be interested in buying Chrome. So I remember.

41:30
I heard when it first came out that the government was looking to divest Chrome, all of these analysts were like nobody would buy it. Why would you buy this? It's really dumb. You could make your own browser. But all of these companies got on the stand and said, yes, we would be willing to pay billions of dollars because so many people use Chrome Like it is a built-in distribution mechanism and it has been a very successful built-in distribution mechanism for Google, both in terms of search and in now, in terms of Gemini, its own AI app. Google has started to integrate Gemini directly into the Chrome app so that sometimes you know, right up there in the bar when you're typing, you're going to start getting answers from Gemini, not just from Google search.

42:17 - Mikah Sargent (Host)
Now I wish, I wish I could keep you around longer. I want to round things out by asking you the judge is expected to rule on remedies this summer, and this is always the final question. What should we be watching for in terms of timing, in terms of next steps, and how likely is it that we're going to see everyone's favorite word, an appeal?

42:42 - Leah Nylen (Guest)
Yes, so the judge has. We're having the remedies trial right now. It's supposed to wrap up next Friday and the judge then is waiting a couple more weeks before he's going to have closing statements. So then the government will sort of get to have a last chance to sort of wrap up all of their arguments for the judge, along with sort of like their legal reasoning. That will happen on May 30th, and then the judge has said that he plans to rule by August, because that would be one year from when he issued his opinion last year and he felt it was very important that sort of we have this remedy in place within a year of when he made the original decision.

43:20
And then, of course, Google has already said it plans to appeal, and so that means that we're going to have to go through this entire process where it goes before another court. That can take another nine months to a year, though I will note the government has been pushing very hard that this is a very like the time is of the essence here, right, they really don't want this lingering for a very long time, and in some of the other cases involving Google, such as the Epic case, people have argued that in order to expedite the appeals, ie to make it go a little bit faster than it normally would. In the Epic case they already had their arguments in February and the Ninth Circuit is sort of expected to rule at any time now. So we might you know you could see that here in this case, so that we might have a decision, I mean maybe by the end of the year, but probably early next year.

44:10 - Mikah Sargent (Host)
Understood Well. Leah Nyland, I want to thank you so much for taking the time to join us today. Always a pleasure to have you on the show. If people would like to keep up with what you're writing about, where do they go to do that?

44:21 - Leah Nylen (Guest)
Bloombergcom.

44:27 - Mikah Sargent (Host)
You can follow me and it will email you anytime I write a story or you can also follow me on Blue Sky.

44:30
Wonderful Thanks so much. Thank you, alrighty folks. We have another great interview coming up in just a moment, but first let me tell you about Club Twit. Hey, if you haven't heard, club Twit just added back the annual plan. Annual plan, club Twit, of course, is our wonderful subscription service, our membership, where when you join the club, you gain access to ad-free versions of all of our shows. You gain access to the TwitPlus bonus feed that has extra content you won't find anywhere else behind the scenes, before the show, after the show, special Club Twit events get published there and access to the members-only Discord server, a fun place to go to chat with your fellow Club Twit members and also those of us here at Twit.

45:09
It is a wonderful time in the Club because we have special events that take place there. We've got my Crafting Corner. We have upcoming coverage of all sorts of news, including Leo and I doing our thing covering WWDC in June. We're looking forward to that. Plus so much more Google IO. That's just one other example. We've got a lot, so be sure to be on the lookout if you are a member of the club. If you're not, we've got a two week free trial, so be sure to try it out and join the club, and I look forward to seeing you there Again. That's twittv slash club twit. All right, let's head back to the show. We are back from the break and, oh, I am so excited about our next interview Joining us on the show today. Head of research for the Wikimedia Foundation. It's Leila Zia. Welcome to the show, leila.

46:05 - Leila Zia (Guest)
Thank you so much, Micah, for having me. Great to be here.

46:09 - Mikah Sargent (Host)
Absolutely so. The Wikimedia Foundation has put out a bit of news about the AI strategy for Wikipedia, and I was hoping that you could start by giving us kind of a high-level overview of this strategy and how it fits into the long-term mission of Wikipedia.

46:29 - Leila Zia (Guest)
Happy to do that. Indeed, we have decided to put out a new AI strategy and in this new AI strategy, we are focusing our attention on reinvesting on the humans behind Wikipedia. The community of volunteers behind Wikipedia are the people who are the most unique element of the success for Wikipedia. Wikipedia is a project that, for almost 25 years, have been developed and cared for by these people and has become a central part of our lives on the internet and web in a variety of ways, whether we come to Wikipedia directly or use its content or knowledge through other platforms.

47:11
In the Wikimedia Foundation, we decided to invest more heavily with supporting these editors with AI.

47:19
More specifically, we decided to do a targeted investment in four areas to support the editors. One is to support the work of moderators and patrollers to assure the integrity of knowledge on the projects. The other one is to focus on any tasks that is basically repetitive and does not require human judgment. It's basically a barrier for editors for doing the work that they're trying to achieve. We also decided to focus on investing in AI for reducing the burden on editors, for creating knowledge that already exists on Wikipedia in a given language and perhaps giving them pathways to bring their local perspective and knowledge to the world. And, lastly, using AI to help the next generation of editors to become editors on Wikipedia through mentorship and assisted mentorship and helping them get on board to the projects. These are kind of the four primary areas that we would like to invest in AI in support of the editors on the projects, with, again like a primary focus to be on the human agency and the editors being on the projects and supporting that.

48:32 - Mikah Sargent (Host)
Absolutely. Now I think we'd be remiss if we didn't talk about the public anxiety that exists around AI replacing, in some ways, human creators and editors. So I'd love to hear you talk about how this strategy reaffirms the role of Wikipedia's volunteer community in this age of generative AI, that the humans are first, as you've mentioned. To hear about how that actually plays out would be great.

49:03 - Leila Zia (Guest)
Yeah, so the primary thesis of the strategy is that we are investing, reinvesting and focusing heavily on editors as humans who are behind knowledge. Really, the reason for this is knowledge is socially constructed and humans are needed for creating knowledge. At the same time, we understand that there are a lot of repetitive tasks, tasks that do not require human judgment, that are happening on Wikipedia by our editors. So we want to invest in AI in areas that human judgments, deliberation, discussion and consensus building is not needed and leave more time for people to do the things that people are best at, if they choose to do so. So that's basically the primary thesis for us, which is invest in humans, reduce, give them the option to invest less time in areas that their AI can help them with and then instead give them more time for things that are very uniquely human, which is around deliberation, consensus building and discussion.

50:14 - Mikah Sargent (Host)
Understood. Now the foundation highlights AI-assisted workflows for moderators and patrollers. I love to hear kind of some nitty-gritty what kind of tasks are you hoping to automate and how will that support content integrity on the platform?

50:32 - Leila Zia (Guest)
Yeah, our moderators and patrollers are key to the success of Wikipedia. These are the people who make sure that the content in Wikipedia is meeting the policies that the communities the content policies that the communities have put together. The policies that the communities the content policies that the communities have put together. What we are primarily focusing on in this space is saying that these individuals are handling a lot of repetitive tasks on the projects, and we know that AI can assist them in some of these repetitive tasks that don't require human judgment. Again, we're going back to that theme of where human judgment is not needed or much less needed. Can we introduce AI to support these editors?

51:13
The other thing that I want to bring your audience's attention to is that Wikipedia, while many of us read it in English, wikipedia is a project that is available in many different languages in fact, over 300 languages, and these moderators and patrollers operate in all of these different languages.

51:32
So what our teams eventually decide to do in terms of supporting moderators or patrollers in a given language with AI also depends on the specific needs of that language. You know the needs of a language in which you have, let's say, less than 100 edits per day. In that Wikipedia is going to be different than the needs of a language in which you have a few edits per second in their Wikipedia right and the affordances that you have for offering AI. To give you just one concrete example, maybe suppose you're a patroller and you want to upload, you want to update information related to a reference that is being used currently on a Wikipedia article, if you want to find all Wikipedia articles that are currently using this reference. This is a hard task right now, but the task of retrieval and discovery is something that we can support patrollers with and moderators with. So this is one area that we think AI can actually help moderators more effectively pull information that is already available and do their work more effectively on the projects.

52:42 - Mikah Sargent (Host)
Understood. Now, one interesting aspect with and you talk about this a little bit the AI-assisted translation to help editors share local perspectives. More broadly, I'd love to hear the goals of this feature and, most importantly because there's that AI anxiety again of sort of stripping the human context out of things how do you ensure the cultural context is preserved in translation?

53:08 - Leila Zia (Guest)
Yeah, excellent question, and this is a topic that, as you may imagine, we care deeply about. So a few things I can share here. Is that one. Again going back to the thesis of this strategy, the focus is on humans and bringing AI in places where humans can do their job, support humans to do their job better on the projects, or more effectively on the projects, if they choose to. So when we talk about the translation, it is still in the context of that primary thesis, which is we want to support humans. So I think that's really important.

53:46
Here. We are not talking about automatically translating content and just putting it in front of people. We're talking about supporting editors for translating that content. At the same time, what we see in smaller Wikipedia languages in terms of article size is that editors are under tremendous amount of pressure for creating vital knowledge, vital encyclopedic knowledge on their project before getting to the point of being able to bring their local perspective and knowledge to the projects. And this is a tension that we're trying to resolve, which is there is a list of vital articles every Wikipedia should have. Can we, for at least this list of articles, support editors with translation methods and translation power to more effectively translate and more efficiently translate this content in their languages and give them back time so that they can go and bring the knowledge that doesn't exist anywhere on the internet today, Because they are the people who have that knowledge and knowledge perspective. So that is the idea behind the translation component of where we talk about translation in this context.

54:56 - Mikah Sargent (Host)
Understood you being the foundation of emphasized values like transparency, human agency, multilingual inclusion. Just hearing how those main things have played out in the decisions that you've made with AI would be great, I think. Once again, I know we keep pulling it back to, but this is about you see, the reaction that sometimes people will have when we hear, oh, this site or this service is adding AI. Oh no, but it sounds like there's a very thoughtful rollout that is taking place. So, yeah, can you talk about that transparency, that human agency, that multilingual inclusion as these kind of core principles as you roll out these AI tools, and maybe how you're continuing to check in on those principles being followed with this?

55:50 - Leila Zia (Guest)
Yeah, this is a topic that is dear to our heart and that's why we talk about it in the strategy as well. I'll give you maybe a couple of examples and I'll be led by you if you want me to go through more details. One is on the topic of transparency. Our machine learning team has developed this concept of model cards, so every AI model that we put into production, or even we propose to put in production, needs to have a model card page, which is effectively a page which is publicly accessible and you can go and check to see what is this model, what is the motivation for creating this model, what kind of use cases we had in mind for creating this model, who is the user that is the target for this model, what are the ethical considerations that we have had for developing these models, what are the caveats that we are seeing and all the technical details about which models we have used and what data pipelines we have used, and all that. All of these are captured in model cards. Model cards are a requirement for any model going to production and by making them a requirement effectively, our machine learning team has built in the process for making sure that we have transparency around, what model is going out and what do we know about the shortcomings and the great things that the model can do?

57:09
When it comes to the topic of languages, I think we have developed some good understanding, use and if we are going to use AI in all of these languages or not. Again, for some languages that are very small, we may want to use AI in certain ways, and is it useful to use AI? Or maybe there are other pieces of technology that are going to be more useful, given where the project is at that point in time? The other principle that we have around multilingualism is about trying to not lock ourselves in a model that we cannot expand more languages. So here the philosophy has been maybe if today we can't use it in ten more languages because these languages don't have need, let's make sure we don't lock ourselves, because our principle is generally, we should be able to go to more languages.

58:22 - Mikah Sargent (Host)
That's wonderful, the model cards. I think, in particular, one thing about, you know, wikipedia in particular, is this aspect that kind of anything that you want to dig into and see how this was built there's I can't think of the word for it right now, but essentially it's all checkable, it's all able to be kind of out there in the open, and so the idea of this model card being an aspect of that is really interesting and, I think, important. And, yeah, the transparency of it is, I think, going to be somewhat refreshing in comparison to perhaps some of the other ways we've seen AI used, perhaps some of the other ways we've seen AI used. This last question that I have for you I am very curious to hear what your thoughts are. You know, wikipedia's content is, as far as we can tell, often used to train large language models.

59:39 - Leila Zia (Guest)
How does this strategy of you know AI being included in the work that is taking place kind of position, wikipedia and Wikipedia with regards to AI, with the lens of this strategy?

59:52
One is our focus on the local perspective.

59:56
I think this is something that we have tried to become really clear about.

59:59
This is something that we see perhaps fewer companies or organizations currently investing in, which is bringing in the local perspectives or the local knowledge of the communities from across the globe, and there are communities that are being left behind with the speed at which AI companies are operating and moving.

01:00:23
So one way that we are thinking about this AI strategy is reaffirming the importance of this local knowledge and perspective and thinking about a Wikipedia that continues to be this model for bringing encyclopedic knowledge to the internet for all of us, for whichever application we are looking at, whether it's training an AI model or, you know, getting an alert on our watch about what we should be, where we should be going next or what is the monument that is around us. The other aspect is really the human-centered approach in this. I think this is again what is differentiating the work of Wikipedia from a lot of other content that is available on the internet and web. In an internet and web that is being constantly polluted right now, with machine-generated content, having this place of human-curated and cared-for content is going to be critical for all these AI models that need to be constantly retrained and built, and in that way, I think Wikipedia is going to play an important role of human curated, generated and cared for content for AI.

01:01:38 - Mikah Sargent (Host)
Leila, I want to thank you so much again for taking the time to join us today to explain kind of more about where the Wikimedia Foundation is in terms of AI being used. I think that we've learned a lot today, Really excited to see these model cards, for example, and continue to watch this roll out. If people want to stay up to date of kind of where the project is what they should be looking for, do you have a place that they should go to do that that we could include?

01:02:11 - Leila Zia (Guest)
For the model cars specifically, if you just search in your search engine of choice model cars, machine learning team Wikimedia Foundation, you will be landing there and you will see our model cars and from there you can be routed to other places.

01:02:25 - Mikah Sargent (Host)
Beautiful. Thank you so much for your time today, and we really appreciate it.

01:02:30 - Leila Zia (Guest)
Thank you so much for having me have a good rest of your day, Michael.

01:02:33 - Mikah Sargent (Host)
You as well. Bye, alrighty folks. That brings us to the end of this episode of Tech News Weekly. The show publishes every Thursday at twittv slash TNW. That is where you can go to subscribe to the show in audio and video formats and, as I mentioned during the show, if you would like to get all of our shows ad free, well, check out Club Twit twittv slash Club Twit. The annual plan is back. We've got the monthly plan $7 a month, $84 a year, so be sure to check that out.

01:03:04
If you'd like to follow me online, I'm at Micah Sargent on many a social media network, or you can head to chihuahuacoffee. That's C-H-I-H-U-A-H-U-Acoffee, where I have all of the places that I exist online. Be sure to check out my other shows, including iOS Today, as well as Hands on Mac and Hands on Tech. Those are all places where I am publishing new shows. Thanks so much for being here this week. It was a great time. I hope you enjoyed it as well, and I will, of course, be back next week for another episode of Tech News Weekly, but until then, I say goodbye to all of you.

01:03:45 - Leo Laporte (Announcement)
Stay on top of tech trends without the time sink. Twittv's short form podcasts are built for busy leaders like you, delivering essential insights in minutes. Hands-on Mac and Hands-on Windows provide quick tips for Mac and PC, while Hands-on Tech quickly addresses common tech challenges to keep your operations running smoothly. If your conference room needs an upgrade, home Theater Geeks explores the best screen and sound systems. And if you like watching the shows, join Club Twit to get full video access, ad-free versions and more. Get tech knowledge that matters on your schedule. Download our short format shows now at twittv or your favorite podcast player.