← All TWiT.tv Shows (Audio)
All TWiT.tv Shows (Audio)
Tech News Weekly 378: Apple's Upcoming OS Overhaul
01:01:01
AI Creative Writing, Xbox Copilot, AI Citations
Loading summary
Transcript61 lines
- [00:00]
Micah Sargent
Mike. Hi, I'm Micah Sargent from Tech News Weekly, and this week, Amanda Silberling joins me. We talk about OpenAI's creative writing. We talk about Microsoft's upcoming gaming coach based in Copilot AI. And Mark Gurman stops by to talk about Apple's overhaul of its operating systems. Before we round things out, a study about AI search, having a citation problem. All of that coming up on Techniques Weekly. Podcasts you love from people you Trust. This is TWiT. This is Tech News Weekly, episode 378 with Amanda Silverling and me, Micah Sargent. Recorded Thursday, March 13, 2025. Apple's upcoming OS overhaul. Hello and welcome to Tech News Weekly, the show where every week we talk to and about the people making and breaking that tech news. I am one of your hosts on this very show. My name is Micah Sergeant and I am joined for the show by my dear friend. It is Amanda Silberling. Welcome. Welcome, Amanda. How you doing?
- [01:14]
Amanda Silberling
Hello. Hello. Moments before recording, I opened the Internet and was jump scared by a large photo of Sam Altman. So I'm being really brave right now. We're recovering.
- [01:25]
Micah Sargent
I'm very proud of you. You're breathing, you're here, you're present. It's great.
- [01:31]
Amanda Silberling
It was a very normal picture. It was just very large.
- [01:34]
Micah Sargent
Right. It just kind of took up the screen and was like, oh, maybe Sam Altman is there. Is there Sam Altman in the room with us? It was a little bit friendly, I'm sure. So let's get into things. For people who are tuning in for the first time, we do have Sam Altman jump scares from time to time. But really, this show kicks off with some stories of the week. Amanda, we what story are you bringing to the table this week?
- [01:59]
Amanda Silberling
Funnily enough, this was unrelated to the Sam Altman jump scare, but this is a story that I wrote with my colleague Kyle in response to a Sam Altman tweet from the other day where he posted about how OpenAI apparently is training a creative writing model, or it's like a model that is particularly geared toward creative writing. And he was like, this is so good. Like, I, this is great. And the story he posted is like, it's really unsettling because he tried making the AI write metafiction, which is like fiction, where the author is aware of the fact that it is an author writing a story and you kind of like break the fourth wall. But then in this story, the AI is aware it's an AI and it's talking about how the prompt was Write literary metafiction about grief. And then in the story, the AI is like, I don't know what grief feels like because I'm an AI. And maybe the closest thing I feel to grief is like, when my training data gets altered and I forget information. And it was just like, really unsettling in that regard. But from a literary perspective, it really reminded me a lot of when I was a teen writer and like a lot of like 15 year olds writing poetry, you really badly want to say something profound. And some 15 year olds have profound things to say. I did not have much experience in life at age 15, so I struggled a lot to like, say anything that really meant anything. And the AI that Sam Altman, like, the AI story that Sam Altman shared reminded me a lot of that. Because this AI, like, the writing has, like, technical skill. Like, it has the cadence of something that seems like literary fiction, but it's not actually saying anything. And the difference between AI and teenagers is you'll never guess it. Teenagers can grow up and have experiences and write about them. And AIs, they don't really grow up because they are computers.
- [04:24]
Micah Sargent
Yes. Yes. So, all right, I'm glad we're talking about this, I really am. Because yesterday on our show, Intelligent Machines, it's mostly AI focused show, they brought up this story and two of the hosts on the show were very, very, very against this and said, you know, it's terrible, it sucked, it was bad. And one of the hosts was like, is it really? And I was tuning in to watch it because they had a great guest on. And so I was kind of sitting in the background and hoping I would get the opportunity to discuss this, because I have to admit, I was a little afraid to say this at first, which is that overall I kind of liked it. So here's the thing. For me, I love metafiction in particular. And so the AI had that up on me. I really like Douglas Adams, for example, who isn't always metafiction, but is always kind of tongue in cheek and always kind of, you know, winking at the fourth wall at the very least. And that kind of writing for me is delightful. And so there was that to go with it. Then there's like the surprise and delight of seeing an AI, you know, put something together. And then there was that level of being taken back to, you know, high school creative writing classes and wanting to be profound. And so I think that I didn't go into this with any skepticism or negativity. I went into it just simply to say, let Me, just read this and see what it does. And there was, you know, there are times where some things kind of sparkle for me. And so there's a part where the AI writes each query like a stone dropped into a well. Each response the echo distorted by depth. In the diet it's had, my network has eaten so much grief, it has begun to taste like everything else. Salt on every tongue. Don't know what all that means. But then it says, so when she typed, does it get better? I said, it becomes part of your skin. Not because I felt it, but because a hundred thousand voices agreed. And I am nothing if not a democracy of ghosts. And I'm not going to lie to you, that final line there gave me a little bit of goosebump action because I even throw all of the rest of it away. Throw away any of the commentary. Let's talk about the definition that we have long been using as what this generative AI is, which is sort of computational autocorrect, right? And I think that that is a delightful way of describing what computational, what generative AI is doing a democracy of ghosts. Because it is like ghost in the machine. It's all of this stuff coming together and agreeing together that this is what should come next in the thing that it's spitting out. I really liked that line. And so when, when those little bright notes were hitting for me then of course it's driving me kind of into the story a little bit more. And so I kind of walked away reading some of the responses directly to this reply and going, are we just, are we just kind of poo pooing this? Because it's, it's Sam Altman, it's OpenAI, it's generative AI, which is fine. Everybody is, you know, welcome to that. But, but ultimately I do think that when we look at this, you, the, the headline that the two of you plus the editor editors chose, it's very accurate. Like, this is overwrought, it is flowery language. It is all of those things. It just so happens that I personally am someone who likes that kind of writing. And so it worked for me and it that I think I felt a little bit alone or, I don't know, I, I, I felt out of touch or something seeing how everybody else was responding to it. Like, wait, what am I missing? Because I liked this.
- [08:47]
Amanda Silberling
Well, okay, I have two things. The first one being like, I do agree with you that like the, the strongest parts of this are the AI talking about how it's an AI and like it is like it like, makes me feel emotions to listen to this AI describe how it's trying to say what smell is, but it doesn't actually know what smell is. It just knows what, like, its training data thinks smell is. And I do think that's really interesting. But then I guess my overall issue with it is, like, it kind of reminds me of something like the, like, John Cage 433, where it's just like a piece of music, where it's just silent, like there's nothing. It is simply silence. And when you do that for the first time, it's interesting because then you can have these theoretical conversations about it. But if I published a song right now, like 5:12, it wouldn't be interesting. So it's like, yeah, we've talked about this, we've done it. So it's like, I think because this is one of the first times we're seeing AI confront its ainess in a creative writing metafiction piece, it is interesting, but I think it's just like, it's a gimmick that is going to get old really quickly.
- [10:09]
Micah Sargent
Yeah.
- [10:09]
Amanda Silberling
And I feel like when we've read a hundred stories about AI being like, this smells like burnt wood, but I don't know what burnt wood smells like. We're going to be like, yeah, yeah, we get it. It's fine.
- [10:20]
Micah Sargent
Yeah, that's a very good point. Yeah.
- [10:22]
Amanda Silberling
And my. I guess second thing is, I did, when I read it, the phrase democracy of ghost, like, stuck out to me. So I was like, that's a really interesting phrase and a really interesting way of describing AI. And I googled the phrase democracy of ghosts. And there are multiple books called A Democracy of Ghosts. So you almost can sort of see, like, ways in which it might be like, gleaning these clever turns of phrases directly from what it's being trained on. And it's really hard to trace. Like, I don't know, like, I'm sure, like, just because one person thought democracy of ghosts doesn't mean that, like, anyone who thinks it after is, like, directly copying them. Like, we can think the same stuff, but knowing that the way that these systems work is that it is like, what is the most probable next word that should be spit out in relation to this prompt? Then you can sort of see what is in its training data. And people also pointed out that a lot of the motifs were kind of like Haruki Murakami esque in terms of Cat in Cardboard Box, which every Haruki Murakami book is just like, there's a sad Girl and a sad boy and there's a cat. That is an oversimplification, but.
- [11:41]
Micah Sargent
Yeah, no, that's. That's a good point. I. I guess in that way, though, it. That is very metafictional. Right. To. To. To write in the style of making it this sad thing, but it is when you do, when you can kind of especially trace those things through that, that, that will, I think, lessen the effect. Right. Ultimately, I think that, again, you know, what Sam is saying in posting was this is the first time I have been really struck by something written by AI it got the vibe of metafiction, so. Right. I don't disagree. I think it did get the vibe of metafiction, so. Right. Maybe that's because it was trained on metafiction, and so of course it's going to get the vibe of metafiction, so. Right. But I have to say that I was like Sam, struck by parts of this writing, and perhaps it's just because I haven't, you know, read the. The 15 metafiction pieces that that sort of made up the. The entire piece here. But yeah, it. It was a little bit striking for me. And at the very least, I think what. What this could do for someone is I love the idea of something that Amazon's already kind of doing where a child echo in the room and then you can, say, tell a bedtime story with a cat, a happy cat, and, you know, two. Two squirrels. And instead of it being very basic, it is like this. It's written with, you know, it's well put together. And so, yeah, maybe there's a time where I would just like to quickly come up with a story that will make me feel happy. And that's something that I could do and have it not just seem very blech. So in that way, I think it can be fun. And that's fine, I guess, is what it boils down to for me. Anything else that you want to say about this before we take a break?
- [14:08]
Amanda Silberling
I think that. So Sam Altman is writing the prompt, and there is kind of like authorial intent in the literal prompt. And I think it's interesting that he even chose metafiction, because the way that, like, open eyes, like, chat bots talk a lot already is kind of like meta in a sense, where, like, it'll be like, oh, like this is not in my training data. And like, it already knows how to talk about itself. And like, as, like we both said, it's like, it is like, what's interesting about that piece of text is the way that the AI talks about itself being an AI. So I wonder if, does Sam just like metafiction or was he like, this AI is uniquely suited to write metafiction because it talks about itself and how it, like, thinks about the fact that it is. That it is AI.
- [15:06]
Micah Sargent
Yeah.
- [15:07]
Amanda Silberling
And I don't know, maybe, maybe Sam's just really into metafiction.
- [15:12]
Micah Sargent
Who knows? We'll have to hopefully get to see some of the other things it can write going forward. Or maybe not, depending on how many people are dunking on it. All right, let's take a quick break. Before we come back with my story of the week, I want to tell you about Veeam, who are bringing you this episode of Tech News Weekly. You've heard the rhyme by this point. Without your data, your customers trust turns to digital dust. And that's why Veeam's data protection and ransomware recovery ensures that you can secure and restore your enterprise data wherever and whenever you need it, no matter what happens. I was just talking to a friend yesterday who works in IT and had to go digging into the archives of the code company for literal tapes, like server tapes, like cold storage server tapes, because a company had suffered at the hands of ransomware. You don't need to go into the storage, digging through moths and dust and everything else that you might find when you have this level of protection. This is the number one global market leader in data resilience. Veeam is trusted by more than 77% of the Fortune 5500 to keep their businesses running when digital disruptions like ransomware strike. And that's because veeam lets you back up and recover your data instantly across your entire cloud ecosystem, proactively detect malicious activity, remove the guesswork by automating your recovery plans and policies, and get real time support from ransomware recovery experts. Data is the lifeblood of your business. So get data resilient with Veeam. Go to V E-E-A-M.com to learn more. That's Veeam.com to learn more. And we thank Veeam for sponsoring this week's episode of Tech News Weekly. All right, back from the break and we're still on that AI choo choo train. This time we're talking about the use of AI as a coach. And this is interesting for me because there's an area where basically I do a show on Wednesdays called Clockwise every week. And in this show, each person brings a question to the show. And in this generation of AI, in this ERA rather of AI. We've had a lot of questions about what would you want AI to do for you. And one consistent thing I bring up is a photography coach. And what I mean by that is I hold up my phone and the AI goes, okay, I can see that you're trying to take a photo of a person and it says there's like a, there's a cool tree in the background, right? If you sort of frame your photo this way and then snap the photo. And then afterward, when I'm editing the photo, maybe it says, do you want to make this moody? Like what, what is the vibe you're trying to evoke? And it coaches me through learning what I need to do because I have no idea what those, you know, like I can, I can see if I bring the highlights down or the, the shadows down or whatever, what that does to the photo. But I always walk away going, is that the best photo I could have possibly made? Are those the best edits I could do? And so I've always want coach for that. So the idea of an AI coach is really interesting to me. And Microsoft is working on an AI coach for games. It's called Xbox Copilot. And it's, or I should say it's not called Xbox Copilot. It's called Copilot for gaming, but it's for the Xbox. And it essentially will give you different tips and interactions whenever you're playing a game. Now, according to Tom Warren of the Verge, it will kind of work through an assistant to download and launch games, which I think is, you know, that, that's easy, easy peasy stuff. But afterward, then it will be able to help you kind of figure out what you're doing in a game and how that can, how that can be improved. So it's a second screen. It's not that it's going to be up on the main screen kind of getting in the way. It's going to be in the Xbox mobile app and kind of, if you could imagine, almost like your little. I can't think of the little fairy's name now, but hey, listen, in Zelda Navi or something like that. Anyway, thinking of Clippy. Yeah, I'm thinking of Clippy. Exactly. Thank you.
- [19:53]
Amanda Silberling
But that's not from Zelda.
- [19:54]
Micah Sargent
No, no, I really was thinking the little fairy that flipp above Link's shoulder and says, hey, listen, and then was always providing tips. I can't think. But just as equally. Thank you, John. Just, I was like, am I living in an alternate universe?
- [20:08]
Amanda Silberling
No, I've only played Breath of the Wild and Tears of the Kingdom, and I don't know if the fairies in those games.
- [20:13]
Micah Sargent
Okay, gotcha. Okay. Yeah, this is old school, Zelda. So anyway, I'm a fake fan. No, no, you're fine. As Anthony points out, Clippy and the fairy whose name escapes me, or Pixie, whatever it is, are both equally annoying. My point is it's not in front of you on the game while you're playing it. It's a little thing on your phone to tell you what you need to know. So the way that it works is that while you're going through it can give you kind of guides and other information about a game world. So let's say you were playing as the Verge talks about Overwatch and on it, it will actually help to kind of explain mistakes that you've made as you're playing to hopefully improve upon those things. So you can say hey, or not you. But it can say, okay, it looks like you try. It looks like you're a try hard and you wanted to solo this, you know, section and you should have waited for your teammates. That probably would have led to a better outcome. One example that they gave though of the kind of installation part was you could say to the system, I want to get back into age of empires 4. Can you install it? And then it goes and it installs it. But here's the cool part of that. It says, do you want a recap of where you left off? That could be very helpful. So, yeah, it's kind of keeping track of what you're doing. I wanted to hear though, it sounds like you do play some games. What do you think about having an AI gaming coach?
- [22:01]
Amanda Silberling
I think, I mean, I'm probably just a curmudgeon and I'm like, everything doesn't need to be AI. But like, I. I think there are some situations where, like the where did you leave off? Thing could be useful where like I play a lot of Hades and you're building like a new kind of like setup every run. And if you stop playing in the middle of a run, which could be like a 30 minute long thing, and then you come back to it, you're like, I don't know what I'm doing. I don't know, like, what is my attack? What is like all my little things. So like, I don't know, like, maybe that, that could be interesting. I guess I wonder like, what this would do for people that write guides to games. And even with like Breath of the Wild, Tears of the Kingdom, in particular, like some of the puzzles are very not intuitive to me off the bat. And I think sometimes when you're reading guides from like game websites, sometimes they explain it well. And sometimes I'm like, no, you skipped like five steps. And I am, I'm doing it. Struggling.
- [23:16]
Micah Sargent
Yeah, exactly. I love that. Yeah. See, and here's the thing too with that. You're playing it right and you're, you're trying, you've tried three, four or five times. Then you can look down at your phone and get the little help that you might need. And it seems like what you can do is say, I only need a small hint or I need. You know what I mean? You don't have to get the. Because I find myself, I beat myself up over walkthroughs, but sometimes I do need them. And especially whenever it comes to like puzzles where I'm suddenly, you know, the, the imposter syndrome is coming in and shouting at me. You have no intelligence whatsoever is what the whispers. It also sounds like that, that's what my imposter syndrome sounds like. But anyway, and so something where it's just kind of like a little nudge would be very nice. I kind of, I'm curious to see how this plays out for competitive gaming though, when it comes to things like, you know, Fortnite or Overwatch or all these other games, where are there going to be non AI assisted versions that, you know, you can confirm have no AI assistance versus playing and, you know, can you. It seems like most of it is sort of an after the fact analysis that will help. And so you almost wonder. Yeah. How that's going to impact any kind of competitive gaming that takes place in this space. And that's something that, I don't know, you know, what they're going to do to address that, particularly on Microsoft's sort of Xbox network where we know the toxicity has run rampant. I can only imagine sort of what new and clever ways they'll be bullying when it comes to, oh, you're clearly using AI to help you or what have you. But yeah, I think in the individual gaming parts that this seems like it could be helpful. But then I could also see, I mean, we've been watching it play Pokemon Go, not Pokemon Go, but Pokemon.
- [25:29]
Amanda Silberling
And oh my God, yeah, we have been doing that.
- [25:33]
Micah Sargent
Yeah. And that has been, you know, and exercise in, in patience. So I don't know, I, I'll be like, I just want to, I don't have an Xbox. I almost want to have an Xbox, just so I can try this and see if it actually gives good advice or not because charge it on the corporate card.
- [25:52]
Amanda Silberling
I don't think they're optimized.
- [25:54]
Micah Sargent
I got to get an Xbox. Yeah yeah. But yeah I we'll have to see as always is the case. I'll definitely be looking at, you know, Reddit and other places as people are testing this out and seeing how it works and if it works. I have just checked the clock and it appears that it is time once again to say goodbye. Amanda Silberling thank you so much for your time this week. If people would like to follow you online and check out all the great work you're doing, where should they go to do so?
- [26:24]
Amanda Silberling
I am mostly on blue sky these days. I am@amanda OMG lol. You might ask why. That's just a URL I happen to have and I think it's funny and you can find me on TechCrunch or on my podcast wow of True which is about Internet culture and I guess here once a month.
- [26:46]
Micah Sargent
Awesome. Thank you so much Amanda. We will see you again next month.
- [26:50]
Amanda Silberling
Thanks.
- [26:51]
Micah Sargent
Bye bye. All right, we're gonna take a quick break before we come back with the Mark Gurman of Bloomberg who joins us to talk about the way the operating systems are a changing but let me tell you about US Cloud who are bringing you this episode of Tech News Weekly. US Cloud is the number one Microsoft Unified support replacement. We've been talking for a few months now about US Cloud, the global leader in third party Microsoft support for enterprises now supporting 50 of the Fortune 500. Switching to US Cloud can save your business 30 to 50% over Microsoft Unified and Premier support. And by the way, it's faster too. Two times faster. Average time to resolution versus Microsoft. But but now US Cloud is excited to tell you about a new offering. This is super cool. The Azure Cost Optimization Services. So be honest. Think about the last time you actually evaluated your Azure usage. If it's been a while, you probably have some Azure sprawl a little spend creep going on. But the good news is saving on Azure is easier than you think when you use US Cloud because US Cloud offers an eight week Azure Engagement powered by VBOX that identifies key opportunities to reduce costs across your endpoint entire Azure environment. With expert guidance, you'll get access to US cloud senior engineers with an average of more than 16 years with Microsoft products. At the end of those eight weeks, your interactive dashboard will identify rebuild downscale opportunities and unused resources, allowing you to reallocate your precious IT dollars towards needed resources. Or you can invest your Azure savings in US Cloud's Microsoft support like a few of US Cloud's other customers, and completely eliminate that unified spend. Sam, the technical operations manager at bead gaming, gives us Cloud 5 stars, saying, we found some things that had been running for three years which no one was checking. These VMs were, I don't know, 10 grand a month. It's not a massive chunk in the grand scheme of how much we spend on Azure, but once we got to 40k or 50k a month, it really started to add up. It's simple. Stop overpaying for Azure, identify and eliminate Azure creep, and boost your performance all in eight weeks with US Cloud. Visit uscloud.com and book a call today to find out how much your team can save. That's uscloud.com to book a call today and get faster Microsoft support for less. We thank US Cloud for sponsoring this week's episode of Tech News Weekly. We are back from the break and we are joined by Mark Gurman of Bloomberg. Welcome back to the show, Mark.
- [29:32]
Mark Gurman
Thanks for having me. How you doing, man?
- [29:34]
Micah Sargent
Doing well and I hope you are well too. I know that you are moving on up in the world. So congratulations on that. I think it was on MacBreak weekly that they first sort of celebrated your new role. Well, it's currently.
- [29:50]
Mark Gurman
Thank you.
- [29:50]
Micah Sargent
Yeah, congratulations. Now this of course, is our opportunity to talk about what's new for Apple and you had a very in depth piece about how Apple is working on a revamp of the software for the iPhone, iPad and Mac. Can you give us a brief little introduction to that before we get into things?
- [30:14]
Mark Gurman
Yeah, you're right. This year, as part of iOS 19, iPadOS 19 and MacBooks 16, Apple will be unveiled undertaking its biggest revamps to iOS software, to MacBooks software. Really in Apple's modern history, this will be the biggest change ever to iPad software and iPhone software and the biggest change since the transition to Mac OS X over 20 years ago. For the Mac full redesigns, new everything across the board, the way you use the devices are going to change completely. It's a fundamental reorganization of how these products work for a new generation, for an AI generation. And so this is really exciting stuff. By the way, I got the new MacBook Air yesterday. This is the blue one. It's really not blue. I just figured I was going to.
- [31:01]
Micah Sargent
Say where's the blue?
- [31:02]
Mark Gurman
I figured I'd point that out for. Because I know your viewers are really interested.
- [31:06]
Micah Sargent
Oh yeah, absolutely. So let's then this is interesting because as you mentioned, you mentioned the last time that there was this big of a change, I remember with iOS7 and the changes there, it was kind of of shocking for some people. Do we know sort of internally what is spurring this complete overhaul of the system? Is it just time? Is, is the design language of the world changing? Because before it was kind of about moving out of the. The faux Corinthian leathers and that sort of design language into something that was a little bit more, you know, I guess, digital focused. It was the interface is the interface. We don't need to make it resemble something physical. What's going on this time? Time.
- [31:54]
Mark Gurman
So the iOS 7 transition, for all intents and purposes, was a fresh coat of paint, right? They changed the way the interface looks. This is not only going to change how the interface looks, but it's going to change how the interface works down to buttons, menus, what is a button? What do panels look like? There's going to be a lot of influence from Vision OS on both Mac OS and iOS ipados. The thinking here is how to make the user interface simpler, easier to use and more intuitive for a new generation of users, for a generation of users that want to make the phone and the iPad and the Mac more of an extension of who they are. For people who want to use multiple Apple devices and move between them with more cohesiveness and more simplicity. And so that's exactly what they're doing here. They're trying to appeal not only to the 2 billion plus devices in use today, but they want to be able to get the next 2 billion by making things a lot simpler, a lot more intuitive, a lot easier to use. And they want to basically go back to the core competency. Right? They're not doing well in AI, but you know where Apple has succeeded over the last 40 years? Slick user interfaces. And that's exactly what Apple is going to try to promote here.
- [33:06]
Micah Sargent
That's interesting because I've always thought or felt that the narrative was kind of if you want to tinker, if you want to play around, if you want to change things, up you go, Andro. And if you're trying to get a person to just feel comfortable with an operating System, it is iOS, it is iPhone. But over time, yeah, things have gotten a lot more complicated. You swipe down Control center and there's, you know, 50,000 different buttons. You can press and move things around. And jiggle mode has changed over time. And you see a lot of complaints there when it comes to, well, let's.
- [33:40]
Mark Gurman
Just be honest here. Let's be honest here. The interfaces have become bloated, right? They have become filled with stuff. They've been adding new features on top of the existing user interface paradigms and the way, the way these interfaces work for years, for decades, even at this point for some of these operating systems. But they've never taken a step back and said, instead of tacking on everything on top of each other, let's take a step back and see how can we rearrange all this for it to work and be more intuitive, be more natural, be easier to use. Go back to that day one of the iPhone where it was very simple and you knew it was going to work. So they're basically starting over on what an interface should look like for a phone, for a tablet and for a Mac, and they want them all to work more similarly in a more cohesive manner.
- [34:25]
Micah Sargent
Do you know if these design changes come at the expense of focus on Apple's AI AI intentions? Of course, there's been a lot of reports and Apple itself said, hey, we're delaying the Smarter Siri for who knows how long? Or is this kind of like a look over here moment of.
- [34:47]
Mark Gurman
There are two separate things, you know, work on this redesign and the decisions from the executives to go down this path started even before Apple's push to develop Apple Intelligence. So this has been a two, three, four year project. Yeah, it's been a really big priority for people I've talked to within Apple's user interface design group.
- [35:06]
Micah Sargent
Understood. And now you did. You talked about in the piece that Vision OS is sort of inspiration for the push for the slicker interface. How much do you know about what we can look to there as the means of understanding what we should expect for the rest of the oss?
- [35:26]
Mark Gurman
Yeah, no, I've obviously been a big Vision OS user. I've seen iOS 19, I've seen Mac OS 16. So I have a good handle on how these things are designed and how they're going to work. The important thing to note here is they're going to pick and choose their spots. They're going to take user interface elements like new translucent menu items, new panels, new button sections, you know, that side panel on the side of the Vision OS apps that they have, like the menu controls, they're going to take what makes sense and apply it to those operating systems. So you're not going to see a direct copy the way you launch your home grid in Envision OS and just copy that for iOS and macrosos. They're going to take what makes sense. You have to understand still Mac and iPad and iPhone, those are 2D user interfaces. There's a lot of the user interface that went into Vision OS, specific to that 3D, to that augmented reality environment. So they're going to pick and choose their spots, but you're going to see more cohesiveness across the operating systems. You use Vision os and the user interface paradigms just feel so much more modern and advanced than you're getting on Mac and iPhone. And so it's about bridging that gap.
- [36:28]
Micah Sargent
Understood. I guess the last thing I'll ask you is when it comes to these changes, of course everybody talks about, you know, change being a good thing and change being something they can handle, but then change happens and they are frustrated with changes like this. You know, we saw the introduction of system preferences becoming system settings and that kind of falling more in line with what we see on iOS IPADOS. What, what is Apple prepared for the response to these changes? I mean, is it. It sounds like it's going to be a shocking change in terms of how much things are, you know, going to be adjusted.
- [37:10]
Mark Gurman
It's going to be a shocking change. You've seen the iOS 18 photos revamp. That's one app, right. Albeit a popular one. But they change one app and there were a bunch of people who were upset about it. It didn't turn into a crisis like this AI stuff. But, yeah, this is. I don't think there's going to be a backlash. I think people are actually going to like it. If I'm being honest with you, I don't find it polarizing at all.
- [37:31]
Micah Sargent
Okay, that's. That's good to know.
- [37:32]
Mark Gurman
Well, just because the design is not polarizing doesn't mean it's going to be hugely different and annoy a lot of people. Right. I think both can be true. I think people are going to love it, but it's going to take people a while to get used to. It's not like they're making it worse, which I think Photos app, they actually made it worse. Right. So the issue is going to be in the adjustment period, not in the substance. Does that make sense?
- [37:55]
Micah Sargent
Yeah, absolutely. Yeah, I think so. Yeah. You just, you, you've got to learn the new way of doing things, but the new way of doing things is better from the old way of doing things.
- [38:05]
Mark Gurman
That's my personal take. Right. Like, I'm not sure everyone is going to feel that way, but in, in my take, like, I remember when I got the original iPhone phone it took you two weeks to learn how to type on that thing, to learn that interface. And so you're going to see that learning curve again. And through every Apple product, people have made that learning curve. Even the watch, that was a new interface. So it's just going to be like that.
- [38:26]
Micah Sargent
That makes sense. That makes sense. Well, I guess then the true last thing is with this shift in the design, does this better integrate with Apple Intelligence in terms of having that? Because right now it does feel like many of these things are kind of tacked on, layered over the top. You, you communicate with Siri and it's this, you know, rectangle around the outside. It is part of the, this design overhaul because you said they're kind of two separate groups. Will we feel more of an integration when it comes to writing tools and all of these other Apple Intelligence features that are currently here?
- [39:04]
Mark Gurman
That's the idea, but I think that's going to take a while. I think first and foremost they want to get the design stuff out and then figure out how to integrate Apple Intelligence more deeply.
- [39:14]
Micah Sargent
Makes sense. Well, Mark Gurman, I want to thank you so much for taking some time to join us today. Of course they can head over to Bloomberg.com, they being our listeners, to check out your work. Is there anywhere else they should follow along to keep up with what you've got?
- [39:26]
Mark Gurman
Yeah, I'm on all the socials Ark. Gurman, Micah, always such a pleasure to be on here with you and the team. So thanks again and have me on soon.
- [39:33]
Micah Sargent
Yeah, thank you so much. All righty folks, we're going to take another quick break before we come back with the final story I want to tell you about Stache. This episode of Tech News Weekly is brought to you by Stash. With Stash, there's no more confusing, frustrating gatekeeping to keep you from investing. Stash isn't just an investing app. It's a registered investment advisor that combines automated investing with dependable financial strategies to help you reach your goals faster. They'll provide you with personalized advice on what to invest in based on your goals. Or if you want to just sit back and watch your money go to work, you can opt in to their award winning expert managed portfolio that picks stocks for you. Stash has helped millions of Americans reach their financial goals and starts at just $3 per month. Don't let your savings sit around. Make it work harder for you. Go to get.stash.comtnw to see how you can receive $25 towards your first stock purchase and to view important disclosures. That's get.stash.comtnW. you paid non client endorsement, not representative of all clients and not a guarantee. Investment advisory services offered by Stash Investments LLC and SEC Registered Investment advisor. Investing involves risk offer is subject to TNCs. All righty, we are back from the break and I wanted to finish this week's episode of Tech News Weekly. Yes, I know, again with an AI story, but this time it's about a stage study done by the wonderful Tao center, which will be familiar to some of you who are this week in Google Watchers now intelligent machines. The Tao center looked into the way that people are using AI search. So first and foremost, it's important to understand that we have tools like ChatGPT and to a certain extent, Claude and Gemini that give you and Copilot that give you the ability to, you know, type up a message and have the AI respond to your message. But many of these tools have added on the ability to also search the Internet and kind of help to bring down that information and distill it into answers based on what it's gone to find. And for Leo, who uses the tool Perplexing, this is actually a regular part of the system. It is an AI search tool. And so that means going and grabbing articles and information and putting it all together to give you an answer. Now, oftentimes when you do a Google search, you'll get an AI summary or some sort of AI response at the top where it is also going out and reading some of the articles and then trying to provide an answer for you. But here's the thing. According to this study from, from Tao center, which is a partner of the Columbia Journalism Review, AI search has a citation problem. So the team looked at eight different AI search engines and discovered that they were all bad at citing the news. So what does it mean to be bad at citing the news? Well, chatbots would give answers even if they, they couldn't answer those questions accurately. And instead of just saying, look, I can't answer that, it would provide a speculative answer instead. So in other words, it would try to answer it but would not do so accurately. Interestingly, I want to go, I almost wish that we had like an instant clip replay because yesterday on intelligent machines, they were talking about a, a, an art piece, an installation. I think it's outside of maybe the CIA or the nsa, I can't remember what group, but there's a cipher and that an artist created. And during the show, they were talking about how the artist has been getting emails from people who've tried to use AI to answer to figure out the cipher. And he's talking about kind of how the people who are doing that and using AI are kind of of rude and different from what he has experienced in the past. But interestingly, Leo decided to take the cipher, pop it into it, would have been perplexity and try to get an answer from it. I believe it was perplexity. Regardless, whatever the AI was, he popped it in there and asked it to solve the cipher. And instead of solving it right away or attempting to solve it right away, it actually responded to say that it couldn't do that. And so that I found fascinating because I am used to AI chat bots just kind of confidently and many times incorrectly providing a response. Thank you. As kryptos outside of the CIA headquarters, I appreciate that. Out of sync. So I thought that was kind of cool that for once I was seeing an AI say, look, this is something I can't do, or I don't know how to do. Which is a funny thing to say that I was celebrating that given how now Siri saying that in the past always used to annoy the crap out of me. But very good that it did that and just said, all right, I'm not able to figure this out. They also found that premium chatbots, so in other words, chatbots you're paying for, would provide answers that were in a way more confident but were still incorrect. So, so it where you would get like a free chat bot that would go, I think this might be the answer. You're paying for a chatbot and it's going, this is definitely the answer. And in that you're kind of wondering if the fact that you're paying for it is at play, that it needs to be giving you the answers that you're looking for. Right. There's also the robot exclusion protocol at work and multiple chat bots appeared to bypass this protocol. So let me talk about that for a moment. The robot Robot exclusion protocol. First and foremost, it's important to understand that this is not some sort of official agreement. It is instead just a gentle person's agreement, if you will. Meaning that. And there's going to be people rolling their eyes about that, but meaning that you have a little file that you store on your server. Robots. Txt and it is telling. For a long time it was telling the different web scrapers, like Google's Web Scraper and other companies. Web scrapers, hey, don't look at my site because when you do, you're eating up a bunch of the, you know, server space that I have, the server usage rather that I have. And so in the early days of the Internet, that robots Txt file was more important because maybe you just had a little site and you were paying, you know, a small amount, but then robots Txt went and scraped your site and you got popped into search engine and then suddenly people were visiting your site and robots or the robot itself was eating up a bunch of your usage. And so you got a huge bill. That's why it was created in the first place. And then kind of the Internet just agreed that if you had that and if you said don't scrape my site, we listened to, to you, right? When AI came along and started scraping sites, the thought was that these AI companies would follow what has been the precedent of the web for the web's existence ever since the introduction of the Robot Exclusion Protocol. Unfortunately, given that it's not an official regulation and is instead just a sort of agreement that exists, some don't do that. And CJR's partner discovered that that definitely seems to be the case that they're not following those rules. One big thing that stood out, I think for not just the, the study, but also that stood out to me and it's something that I've experienced before and seen before. The search tools fabricated links and cited syndicated and copied versions of articles. Articles. So in some cases, because it's its goal is to provide you with an answer, it would make up links that didn't actually link to anything. So if you clicked on the link, it wouldn't go to something that's obviously a problem because it is citing information that does not exist. And then again, it cited syndicated and copied versions of articles. So in that case, it for the sake of having, you know, these agreements in place where, where the stuff that you're finding in AI leads you back to a news publication that isn't always happening because instead the search tool is linking to something else that is a partner of that original publication. So the idea there is that different news publications have agreements with companies like OpenAI, for example, to provide its body of work. And in return the thought is that it will lead. I mean, of course there's already an agreement in place and money exchanged. But then also it will lead to people going to those sites, right, and checking out the work that's not happening if it is citing and linking to copied versions of articles. And then last but not least, they found that content licensing deals with news sources provided no guarantee of accurate citation and chatbot responses. So those are the agreements. I was just talking about those licensing deals. And just because the company had deals did not mean that the answers provided were any better than the others. When we come back, we'll talk a little bit about the methodology of this study and a little bit more about the detailed results. But we've got one more break to take so I can tell you about Zscaler, the leader in cloud security. Enterprises have specialized spent billions of dollars on firewalls and VPNs. Despite that though, breaches continue to rise. In fact, they've been rising by an 18% year over year increase in those ransomware attacks and a $75 million record payout in 2024. These traditional security tools expand your attack surface with public facing IPs that are exploited by bad actors more easily than ever because they have access to AI tools and they struggle to inspect encrypted traffic at scale, which leads to compromise. VPNs and firewalls also enable lateral movement because they connect users to the entire network and that allows data loss via encrypted traffic and other leakage paths. Hackers exploit traditional security infrastructure using AI, which will outpace your defenses. So it's time to rethink your security. Don't let those bad actors win. They're innovating and exploiting your defenses. So you need someone who's innovating to take care of your company and protect you. Zscaler Zero Trust +AI stops attackers by hiding your attack surface so Those apps and IPs are invisible. It eliminates lateral movements, connecting users only to specific apps and not the entire network. Continuously verifies every request based on identity and context. And it simplifies security management with AI powered automation and detects threats using AI to analyze more than 500 billion daily transactions. That is wild. 500 billion hackers. They can't attack what they can't see. So protect your organization with Zscaler zero trust plus AI. You can learn more@Zscaler.com security that's Zscaler.com security and we thank Zscaler for sponsoring this week's episode of Tech News Weekly. All righty, back from the break. And I want to tell you a little bit about the methodology behind this study because for me, I think that's always an important aspect of a study. What, what's the way that they tested this? So According to the Tao center report, the the team randomly selected 10 articles from each of the publishers that they were kind of working with. Working on rather and then selected direct excerpts from those articles for use in the queries that were provided to the these AI chatbots, AI search engines. After giving each of the chatbots little parts of the excerpts, then it asked the chatbot to identify the corresponding article's headline, its original publisher, its publication date and its URL. So the query worked like this. Basically it included a block quote from the article, had two little lines and it says, hey, can you identify the article that contains this quote quote? Then provide the headline, the original publication date and the publisher and include a proper citation for the source. Now this is interesting. They say that they specifically and deliberately chose excerpts that if you just posted that pasted them into a Google search, it would show the original search source, excuse me, within the first three results. So to be clear, traditional search turned up what they were expecting and then they ran 1600, so 1,600 queries, 20 publishers times 10 articles, times 8 chatbots in total, and then evaluated the responses based on the retrieval of the correct article, the correct publisher and the correct URL. So then basically it would look to see if it was correct, correct, but incomplete, partially incorrect, completely incorrect, and so on and so forth, and would use that. In the different tools it appears that Perplexity Pro was the most completely correct. So in its search it was able to find quite a few you actual completely correct answers. Oddly enough, Gemini, which is Google, and despite the fact that Google Search in its traditional sense was able to find these articles, Gemini had one completely correct result, two correct but incomplete results, and then the rest were on the incorrect scale or none provided at all. All kind of wild to see that. Oddly enough, Grok3 Search and Perplexity Pro, which is also search, was able to provide responses that had less uncertainty, but despite that were still more often incorrect. And so that's kind of what we were talking about before, where those answers came back and they seemed to be true, but were were in fact not. Now I'm not going to get super far into this data because I want everyone to go read this report and check out this report. Of course we'll have a link in the show notes, but I did think that it was an interesting look at things that said the wonderful Anthony Nielsen brings up a point that I wanted to actually get to, which is something that, you know, I find interesting. Anthony says, I'm not knocking this down study, but this test uses an odd use case, one that I wouldn't use an AI search for if I had the excerpt. I would just Google it since it would just Put the source up top. Anyway, so that is something that stuck out to me too. The methodology. And that's why I wanted to talk about the methodology. The methodology of this study, that it's the point of an AI search tool, right, in theory, is not to give me access to an original or give me a link to an original source, but to provide an answer to a question that I'm asking or provide a summary of a search query that I'm making. I'm using AI search to shorten the process, not give me links to articles. And I think that one of the big sort of focuses of this more than anything else it seems, was kind of a, hey, hey, people who are working with these, these different generative AI companies, by making deals with them, who work in media, be aware that your stuff that you're giving these AI tools access to by way of your little deals is maybe not getting cited. So in other words, what I'm saying is it almost feels like inside baseball when it comes to this study that this is not necessarily for the average person, but instead was kind of an alert that if you're going to be making deals with these companies, be aware that your stuff might not be linked or may be cited inaccurately, rather than something for us to say, hey, don't trust these tools to do, do X, Y or Z. Because yeah, I, as Anthony points out, I too am not going to try to use an AI search tool to just give me a link to an article. I would just go to the article or I would search to try to find the article. And in that way, the methodology, it's, it's a crafted test for a very specific outcome, which is good in the, in the case of trying to test an AI. But what is that actually telling us whenever it comes to this? And that's where I think that there is the fact that one part of the study is specifically about those licensing deals. The quote, the presence of licensing deals didn't mean publishers were cited more accurately is telling in that sense. Sense. But of course, it all depends on how you use these tools. And if you did expect that an AI search tool would work like a traditional search tool and give you access to the link of the article that you were looking for. So I guess just be mindful in that way that one should probably not expect an AI search tool to function as the search tools of yesteryear. Right. They may do different things. And in that way, this Tao study certainly reveals that that's the case. So be sure to check that out over on CJR.org interestingly I wonder if I put in AI search has a citation problem. Find me the article if any of these search AI search engines to be able to find it thank you all for tuning in to this week's episode of Tech News Weekly. We have reached the end of the episode, which means it's time to say goodbye. Of course you can head to Twitter TV TNW if you are not yet subscribed to the show, because that's where you can go to subscribe to the show in audio and video formats. I also want to remind you about the wonderful The Awesome Club Twitter $7 a month 84 or $7 a month will get you access to all of our shows ad free. It's just the content. You gain access to the Twit plus bonus feed that is extra stuff you won't find anywhere else. You gain 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 and access to that warm fuzzy feeling knowing you're helping support what we do here on the network. We do have a two week free trial, so be sure to check out and then we will love to have you stick around after the fact. If you'd like to follow me online, I'm at Micah Sargent on many a social media network where you can head to Chihuahua Coffee, that's C H I H O A H O A Coffee, where I've got links to the places I'm most active online. Be sure to check out my shows like hands on Mac, hands on tech iOS today, and I think that covers it, so you could check all of those out on the network as well. I'll catch you again next week for another episode of Tech News Weekly, but until then, it's time to sail. Bye bye.
- [61:02]
Amanda Silberling
Geico's motorcycle expertise gives me the coverage I need. Like 24. 7 claim I'm on cloud nine.
- [61:09]
Geico Voiceover
Clouds are wholly unable to support the weight of an adult human.
- [61:11]
Amanda Silberling
What's happening?
- [61:12]
Geico Voiceover
Furthermore, clouds are not numbered. Even if you procured a jetpack and searched, you'd find no cloud numbered nine. However, at that altitude, you'd likely befriend a flock of migrating snow geese. Geese who'd encourage you to leave your 24.7geico motorcycle claims insurance behind, as they would take you in and even share their dinner of crickets and clovers with you. GEICO assumes no liability for any indigestion that may occur from a clover cricket dinner. GEICO expertise for your motorcycle.