 This 10th year of Daily Tech News show is made possible by you the listener. Thank you. That includes Pepper Geesey, Eric Holm, and Carmine Bailey. Coming up on DTNS Orbital Pursuit Jackals are coming for satellites. Why some folks think AMD is price gouging. And why is every other tech story about chat GPT? Turns out there's a good reason. This is the Daily Tech News for Thursday, February 2nd, Groundhog Day 2023. This is the Daily Tech News show for Thursday, Groundhog Day, February 2nd, 2023. In Los Angeles, I'm Tom Merritt. And from Studio Wedwood, I'm Sarah Lane. Sarah Lane. Sarah Lane. And once again, I'm Justin Robert Young. Yeah, and I'm the show's producer, Roger Chang. Let's start. Let's start with the quick hits. Okay, enough of that joke. Meta reported its Q4 revenue was down 24% and net income was down 55% on the year. But that was better than expected, especially considering cost associated with layoffs this year, breaking office space leases, and data center redesigns. The good news was that users were back up. Daily active users across Meta's family of brands was up 5%. Facebook specifically added 16 million daily active users, and WhatsApp reached 2 billion active users in October. Another reason you might see Meta's stock rising is that Mark Zuckerberg called 2023 the Year of Efficiency, a.k.a. they're going to keep costs under control. However, one of Meta's biggest cost is its Reality Labs, which is developing AR and VR products for a future metaverse of some kind. Reality Labs revenue rose in Q4, but losses did too, and those are expected to continue to rise this year. In a series of tweets from its Twitter developer account, Twitter announced it will discontinue the free access to the Twitter API for both Legacy V1.1 and the new version 2, starting next week. You got one week till February 9th. They will launch a paid version in its place. No word on how much that will cost, but you have one week to prepare to pay them something. Twitter recently changed the terms of its API, causing third-party Twitter clients like Tweetbot and Twitterrific to shut down their mobile apps. Developers have been using this free version of the API, not for those clients. That was a policy change. The free version of the API is used to track things like account changes or give you alerts. Scientists use it to study human behavior on the platform. They want to look at the spread of misinformation. You use the free API for that. Well, not anymore. Yesterday, we talked about Netflix's upcoming plans to crack down on passwords sharing in the U.S. and elsewhere in the world. The discussion was based on text found on Netflix's help center, first documented by The Streamable. Turns out Netflix wasn't prepared for The Streamable to shine quite a light on that section of its U.S. help pages and has since removed them. The Streamable wasn't imagining this, though. They didn't make up a bunch of news. The Verge found that Archive.org had a version of that page from yesterday to confirm it had indeed been up there and had been shared. So what happened? A spokesperson told The Streamable, quote, for a brief time yesterday, a help center article containing information that is only applicable to Chile, Costa Rica and Peru went live in other countries. So, oops. How different is it really going to be what they finally do? Because they said they're going to anyway. Sony sold 7.1 million PS5s in Q4. It's best ever sales quarter for PS5s. Sony has now shipped 32 million PS5s since launch. Sales revenue for Sony's Game and Network Services segment was up to 53% year over year. Operating profit rose 25%. Sony did well selling game titles, sold 86.5 million video games, 11 million of which were God of War Ragnarok, and PlayStation Plus subscribers rose from 45.5 million to 46.6 million. Not a huge jump, but a jump. Sony also announced that starting today, beta testers will be able to join Discord voice calls on a PS5 console if you are in the US, Canada, Japan or the UK. You link your Discord and Sony accounts first, then use Discord's mobile app and transfer the call to the PS5. Sony also announced it added variable refresh rate support for 1440p, and there's a few other dashboard improvements that came down as well. Password Manager software maker Dashlane has published the source code for its iOS, Android, Apple Watch and Mac apps on GitHub for public auditing. It plans to publish the code of its web extension as well. It was issued with a Creative Commons attribution non-commercial 4.0 license, meaning you can copy, you can share, you can build it, but it's not exactly open source. Dashlane says it also didn't publish some key elements which prevent you from using the code to build your own Dashlane. Oh, that's why I prefer Sarah Lane. Alright, let's talk about space. Space is getting crowded, y'all, and when fresh new resources get crowded, there can be spats, there can be fights, there can even sadly be wars. Ars Technica has an article about True Anomaly, Inc., a startup formed last year by former U.S. Air Force Major Evan Jolly Rogers. No relation to Roger K's Twitter account. For their own Jolly Roger. Yeah, Major Evan Rogers. True Anomaly will launch its first orbital mission in October. SpaceX is going to carry two orbital pursuit spacecraft that they call Jackals. They will have no weapons, okay, these are not going to go around shooting down other satellites. That is technically against many agreements about space. They will have sensors, though. They're capable of something called RPO, Rendezvous Proximity Operations. Basically, they can get up close to another satellite and detect whether it's got certain surveillance equipment or if it has weapons as well as possibly intercepts some communications. The Jackal's first mission will be to spy on each other, you know, just to make sure they're loyal. No, it's actually to see if the systems work because they know how they work. So they'll look at each other and if that looks right, then they know it's working. Eventually, the company imagines hundreds of autonomous Jackals patrolling orbit. Well, that doesn't sound frightening at all. I'm very, very glad that we're getting some RPO news as the Eagles and Chiefs are about to play in the Super Bowl. It is important that we point out here that these RPOs are not new. There are dozens of these crabs already in orbit over US, Russia and China. They're just not pursuit vehicles. Right. You could use them for repairing satellites. That's been done before. Most of them just swing around in orbit and look at each other. They could be used for debris cleanup. So RPOs themselves are not bad. Calling them Jackals, making them pursuit spacecraft is a little more aggressive than RPOs have been in the past. But yeah, this from the guy who once ran a company called 3720 to 1. Yeah. A reference to Han Solo navigating the asteroid field in the Empire Strikes Back. So obviously the guy has a sense of humor. He's Jolly after all. He's Jolly Rogers. He wants Roger Chang's Twitter account, I bet. I don't know. I guess the worst case scenario, if that's even how you would describe this is someone saying, well, okay, let's say that the US gets into spat with another country that has a cornucopia of satellites up there too. And we shoot them all down. And then what happens? I mean, the war thing is, okay, that's not happening today. Most of this is being done to repair equipment and just do stuff that's in orbit. But yeah, what does this mean at some point when these satellites are able to go to bat for themselves? The privatization of space has made it possible for everyone to be in space, which means that you're going to have the same things happening there that humans do on the ground. Surveillance, you know, try jockeying for position. That's what the jackals are. They're a defense thing that you may or may not love, but is sort of inevitable when humans are in the equation. AMD announced some shipping dates for some of its gaming focused processors for those who celebrate. February 28th will be the day that you can get the Ryzen 9 7900X3D and the 7950X3D and also the 7800X3D will follow on April 6th. But there's a headline submitted by KV underscore 87, shot to the top of our subreddit. Thank you, by the way, KV. It's from PCWorld and reads, AMD is under shipping chips to balance CPU GPU supply, less supply to balance out demand and keep prices high. However, the article was originally called, quote, AMD is under shipping chips to keep CPU GPU prices elevated. A little bit different wording there. Well, during AMD's earning call Tuesday, CEO Lisa Sue said, quote, we've been under shipping the sell-through or consumption for the last two quarters. We under shipped in Q3, we under shipped in Q4. We will under ship to a lesser extent in the next Q1. This is one of the situations where when you're not familiar with the lingo, you start to think like, oh, they're doing something on purpose. They're under shipping. They're like, haha, we won't sell them things. And some folks, including PC Gamer's headline writer, apparently jumped to the conclusion that AMD was trying to do something like price gouging here. AMD's VP of communications told PC Gamer, we are shipping below consumption because there is too much inventory in the channel and that partners want to carry lower levels of inventory based on the demand they are seeing. In other words, there's already too many of these things available. That's why we're not shipping as many as we normally would because there's already too many of them. In fact, Nvidia's CFO Colette Cress said a similar thing on the Nvidia earnings call last November. She said, we have been under shipping gaming at this time so that we can correct that inventory that is out in the channel. Or as Cobra 8089 in our subreddit put it, AMD is following basic supply and demand. Is Cobra right? Is this just Econ 101 or is this collusion? Well, you also have to remember is how the players work here. So the chip makers make the chips. The chip makers then either sell them directly to chains or sell them through another distributor that go to chains. And in both of those situations, either you are selling a distributor that has to warehouse it and that costs X amount of money or you are selling it directly to these chains, these final places that you can either walk up and buy it or buy it online from. And they have to hold them. So what I really, really do want to underline for people is that what happened in our economy, what happened in our supply chain from 2000 on was something extraordinarily catastrophic. And we do have to keep in mind that while our memories move on fast from that, a lot of these mechanisms are extraordinarily slow moving, especially in the chip sector. Roger, I was curious if you had a thought on this because I know you follow the chip sector really close. I would add that a lot of these GPUs are kind of, if not exactly at the MSRP that they were announced at, maybe five or for 8% over. I mean, historically in the past three years, that's a good thing. I mean, back in two years ago, you were looking at GPUs that are selling upwards of 80, in some cases 100%, depending on demand for a scarce product. I mean, one of the things people need to understand is they're not going to give you these things 50% off, even if they are two year old GPUs because they're still very potent technology and they work incredibly well. So, I mean, I think some people are just so used to the cycle of a new technology comes out. Obviously, that means the older stuff gets discounted by like 10% to 15% off the bat and I don't think that's the case anymore. One is GPUs are now priced at a much higher level than they were five years ago just because of the amount of functions they put into it. But also what Justin was saying is like the supply chain is still sort of storming itself out and these companies need to make money. They're not going to sell things at a loss. And you could rightly say like, well, what about MSRP? Maybe MSRP is too high right now. And to that, I would just point to the general inflation that's happening all around the world. Prices are going up. The prices are set because they think they can get people to pay that much. If they are unable to sell that many, then the price goes down. And remember, these final outlets that you are buying these products from have bought them at a certain price. So if they sell it lower, then they are taking a loss on that and that their business is damaged. This is not the farm to table solution that I think a lot of people simplify it to when you just look at, well, why is it this costing less? Yeah. And under shipping is a weird term. I get that. If you're not using it all the time, it probably sounds more nefarious than it is. But it basically means we would normally ship at this level, but there's so many already in the channel or on the shelves. If you want to think about it that way, we can't sit. If Best Buy shelves are full of NVIDIA or AMD chips, they're not going to sell a bunch more to Best Buy. Best Buy doesn't want them. They want to sell the ones they've got on the shelf first. And that cascades through the system. I know I'm oversimplifying, logistic expert, big Jim. But that kind of gives people the general sense. Hey, we got that story thanks to the subreddit. You could be somebody who helps us find those kinds of stories by just submitting them in the subreddit or even easier, just voting on the ones that have already been submitted. Head on over there, dailytechnewshow.reddit.com. Well, you may be tired of hearing about Chat GPT. Sorry in advance. It seems to be in every other tech news story out there, although the other ones are mostly layoffs in the technology sector. Also a very popular story these days. But Chat GPT may be the most important story in tech this year. And hey, it's only February. We're going to debate that. But first, here are a few things you need to know to understand these stories. Yeah. So just to get the lingo down, in case you've been hearing it and you're like, you know, I hear all these acronyms. OpenAI's GPT stands for Generative Pre-Trained Transformer. And the important part of that is the transformer. Google invented the transformer part of that. Google's version is called Lambda or Language Model for Dialogue Applications. The transformer is the thing that helps it predict what text it should write. So you got GPT is OpenAI's, Google's is called Lambda. It's the engine. These kind of large language models, sometimes called LLMs, look at it prompt and what if anything is written so far, then estimate the probability that one of the words in its vocabulary is the best choice for the next word. It's not thinking. It's not looking at a list of facts. It's looking at what you wrote, what it's written so far, and then going, huh, the logical next word would be this. That is all it's doing. And then it repeats the process. Chat GPT uses OpenAI's GPT. GPT-3 was launched in June 2020. But Chat GPT is using GPT-3.5, which was launched last March and was used to fine tune Chat GPT. It is the current version being used in public products, but there is another one on the way called GPT-4 that is improved and Microsoft apparently has access to that already under its deal with OpenAI. That's part of the deal that Microsoft got. Now, I think if you know that, you know enough to understand the lingo to set the table for our discussion. These are just some of the stories about large language models, mostly GPT-3, that have been published over the past 24 hours. Yeah, so CNBC says that Google is testing a Chat GPT-like tool called Apprentice Bard, not Bored, Bard, B-A-R-D, and might replace the I'm feeling lucky button that you possibly forgot is still on the Google homepage. Maybe you use it, many don'ts. So this could be a place where you might see a Chat GPT-like tool end up with a prompt for asking AI questions about your search. Semaphore says Microsoft is working to incorporate GPT-4 into Bing Search. Semaphore also says OpenAI is planning a mobile app for Chat GPT and a way to get Dolly to generate video. Chat GPT is also now available in Microsoft Word thanks to a third-party add-on from Creative Data Studios called Ghostwriter. Versions for Excel, PowerPoint, OneNote, and Outlook are also in the works. Microsoft launched the Teams Premium that uses GPT-3 for Intelligent Recap, which automatically generates notes, tasks, and highlights of meetings. Swiss international airlines at Lufthansa have both been using forecasting models from Google to optimize their flights in their network and reduce flight delays because of wind concerns. And for those who don't even cover news, we previously reported that Chat GPT-plus is now available for $20 per month. Yeah, so finally, Reuters reported Wednesday that Chat GPT reached an estimated 100 million active users last month, two months after launch. That sounds like a lot. It is. UBS Research calls it the fastest growing consumer application in history by comparison. TikTok took nine months to get to 100 million active users. Instagram took two and a half years to get to 100 million active users. Facebook took five years to get to 100 million active users. Chat GPT took two months. If you're one of the people who've been asking, and I forget these questions all the time, when's the next big tech innovation going to hit? When are we going to get that next iPhone moment? Why haven't we had anything new in technology in so long? You might have your answer here. So Justin, let's start it with you. Are large language models like GPT-3 a turning point in tech history, like the ones I just mentioned? Are they along the lines of the browser and the iPhone? Yes, I genuinely believe that they are. All the examples that you mentioned are the bridge between what the audience can understand benefits them and where the technology is. Very famously, with everything that you said, the browser, the iPhone. This was not wholly new technology. And often we tend to conflate the technological breakthrough with the consumer breakthrough. There's no doubt that based on how people have understood Chat GPT, and if you even look at the conversations that are happening now with Chat GPT, the price of the product, whether or not it's biased in how you train the model. These are secondary conversations that you can only really have once you've found utility in this technology. That has not happened, in my opinion, since the big, big, big moments where we found this harmony between tech and consumer. The iPhone, YouTube, the browsers, things that materially move the culture forward by way of technology. I believe that we are at the beginnings of that with AI and Chat GPT right now is the Coca-Cola of that brand. Yeah, I mean, not to poo-poo on these numbers because obviously Chat GPT has garnered quite a bit of interest. Some of the 100 million active users that were reported last month, two months after launch, it's like, wow, yeah, that's crazy. Crazy growth, crazy interest. Not all of those people are going to use it. Some of them, like me, for example, are just kind of going in there and seeing what's going on and saying, okay, here's how it works. Here's how I want to understand it working better. Am I using it on a day-to-day basis? Not currently. I might later. Don't have a need for it exactly right now. I think those in the educational sector have really a more robust way to be able to use it kind of out of the gate, even though clearly it's not totally ready for prime time, again, depending on who's using it and why. So these numbers may flatten out a little bit and not be so insane over the next couple of months. But yeah, we're talking about something like this being incorporated into all search engines. I mean, it only takes one for it to work pretty well and we'll see it in all of them soon enough. And everybody's kind of working on that, trying to figure out, okay, very, very powerful tool. How does this get incorporated into things that people already use, rather than this new thing where it's like, well, I don't need chat GPT to write me my next book report because I don't. But I sure could use it in various facets of my professional life, certainly, and even my personal. Yeah, I see so many elements of this that are familiar with things that became big. People saying, I don't understand what you'd use this for. That doesn't mean anything to me. People saying, I think it's overhyped. Sure, it's flashy, but people won't stick around and keep using it. That doesn't mean much to me. What means something to me is, are there a lot of different uses and what kinds of fears has it evoked? Because if it evokes fears, that usually means it's really good at something to get that kind of fear evoked. And you didn't see Facebook evoke a bunch of fears about privacy when it launched, but you did have people wondering about whether it was good for students and are people going to spend too much time? Then there started to be privacy stuff pretty early on. I think this has all the hallmarks of something that is used in a multiple variety of ways, and it's misunderstood. USC associate professor of computer science, Jonathan May, wrote on the conversation, chat GP doesn't try to write sentences that are true. It tries to write sentences that are plausible. And I think wrapping your head around that gives you an advantage in understanding chat GPT. It's danger isn't that it's going to replace book reports that stop people from thinking because it's not good at that, at least not yet. Maybe a future version will be, and then we'll deal with that. What it's good at is constructing sentences and giving you first drafts and doing things that we haven't even figured out how to do with it yet. Tom, let me ask you this. You've been in this tech podcasting news game for a while. What is your hall of fame of why are we talking about this thing every single show? Yeah. The thing that I over and over and over have gotten so tired of hearing about X. Oddly, the first one I ever experienced was when I was interning at National Public Radio. One of the reporters walked in and was like Bosnia again because it was just in the news every day in the mid 90s. But Microsoft in the earliest days of tech TV and CNET, oh, you guys over cover Microsoft. Why don't you ever talk about Apple very soon after that? Oh, Apple, that's all you ever talk about. Why don't you talk about Microsoft or something else? Facebook, Lately, Twitter, I don't know. I'd have to think about it some more to come up with the true hall of fame. But those, those would all be entrance right there. And I think all of those were big, big, big news. They all fall into this, this genre. Well, folks, we're not quite done talking about generative machine systems. If you've ever wondered why those machine generated images can't do hands. Well, they've always got like six fingers or their mouth or three. They look like a claw. Yeah. Sarah, explain, please. I hope BuzzFeed's Pranav Dixit wondered the same as you, Tom, and sought to find the answer. He talked to artist and university of Florida associate professor of AI, Amelia Winger-Bariskin. Turns out it's all in the training. Models like Mid Journey, for example, are trained on images that just don't show hands very well. And those are images that are photographs of people. If a hand is in a picture at all, it's usually holding something or pointing, making a fist. Maybe it's in a pocket. Maybe you don't see it at all. Few fingers are often showing. So it learned it from watching us. Yeah. We're the problem. You've never shown me a picture that definitively tells me how many fingers are on a hand. How am I supposed to know? I'm just going to do my best and make fingers look real weird. Everybody put your hands where AI can see. Yes. Hands up for the AI. It's an interesting example of training set bias. We usually talk about training set bias in relation to ethnicity and skin color and things like that, and we should, but this is another example of that. We have biased it to not know how many fingers are on a hand because we didn't put enough hands in the training set. The thing, though, is that I get that this does happen. However, having used some photos, like a nice photo of a friend where I thought, oh, maybe I could add like a little, I don't know, art quality to it and get it printed out or something and just do something kind of fun and creative. Some of those images do have hands that I think are pretty easily shown, but the AI version of it is totally messed up. The rest of it is mostly fine until you look at the hands and you're like, oh gosh. I'm not sure I want to give you this. Just like we described with the chat GPT, it doesn't know how to draw hands. It just goes based on all the hands I've ever seen or things that look like hands. Here's what I expect it could look like, and then, yeah, and then it does its best. So really what we're saying is, lend AI a hand. Yeah, show them what's up, humans. All right, let's check out the bell bag. Let's do it. Josh wrote in and said, this came up a few times this week in previous episodes, just wondering your take. Josh says, Netflix has refused to be part of the Apple TV app, and I totally get that. They want people in their app watching their shows, followed by another one of their shows. If they're in the TV app, people may watch another services show next, etc. Josh says, when they were far the biggest streamer, it made sense. But lately, most of the others are in the app, and I find myself going there mostly and consequently rarely opening Netflix itself. Seems like at this stage, it's backfiring. I assume I'm not alone. I guess my question is, do you think Netflix will give in and allow themselves to be fully integrated, or continue to hold out? Good question. This is a great question. I've had these exact same questions myself. Here's the answers that I have come up with in looking into this sort of thing. First of all, Netflix does not seem to be suffering from it, and my personal anecdote back to you is, I thought the same thing, like, oh, I keep forgetting to go to the Netflix stuff, because it's not up there in the top row of Apple TV. But then I did, then I remembered, then I found something that was like, oh, this is only on Netflix, and it wasn't hard to go launch Netflix. It doesn't seem to be the problem with Netflix's subscriber numbers. So I don't think you can show evidence that not being integrated in either Apple TV or Android or anybody else's lineup is causing a problem. The second thing is, it's not that they're worried about people watching other shows than theirs if they're up there, because then it wouldn't make sense not to be there at all, because like you say, then people might forget that Netflix has shows they want to watch. It's about the relationship with the audience. Apple wants to control all the information that goes into selecting those shows. Netflix wants to control all the information that goes into selecting those shows. And so Netflix is like, we don't need to compromise on that. We get better data by keeping it all in house so that every time somebody's clicking on something and watching it, we see that click. That is what I have come to find out is why they don't want to be there. They can't come to an agreement about the data sharing. Yeah, can I add one more thing there? It's also leverage. Netflix, like many other gigantic companies that deal with Apple, have constant ongoing negotiations about many, many things up to placement and cuts and stuff like that. This is something that Apple would like from Netflix. So considering Apple's intransient position on some of their issues with the App Store, Netflix, I don't believe is in any kind of mood to be doing favors. Now, if they could strike a grand bargain, then maybe they would be more aimable to do it. But right now I don't believe something that would be all encompassing. It benefits Netflix's interest to do anything for Apple that would make their technology more of a must buy for consumers. Yeah, think of it this way. The 30% that Apple takes from Netflix subscribers if they were to offer subscriberships through the App Store directly, if you could reduce that, that might cover the value lost by letting Apple manage that relationship with Netflix. Yeah. Yeah. Well, thanks, Josh, for writing in and to anybody else who has thoughts on anything we talk about the show or might talk about on a future show, feedback at dailytechnewshow.com is where to send that email. Also, thanks to you, Justin Robert Young. Let folks know where they can keep up with your latest. I do a podcast called Politics, Politics, Politics, and we are gearing up for the big primary season, which is currently in swing. In fact, I did an episode before Newsbroke that Nikki Haley was going to announce that she is running for president on February 15th in Charleston, South Carolina, just by the grace of the podcast gods. I did a whole explainer not only about Nikki Haley, her entire career, and the hokey pokey that she has played with the Trump right. You can download all of it at Politics, Politics, Politics, and many lives of Nikki Haley wherever you get your podcast. Also, a special thanks to Michael Perkins, who was one of our top lifetime supporters here on DTNS. Michael, we see you. Thank you for all the years of support. Thanks to all of you who support us. Be like Michael if you're not already. If you're already a patron, just hang in there. We'll get to you. There's a lot of you. But if you're not, Michael, we'll get in there and we'll actually, you know what? I know we shouldn't do this, but if you join right now, we'll get to the front of the line. We'll mention you tomorrow. Patreon.com slash DTNS. Speaking of patrons, there's so many things you get, especially good day internet. Stick around for our extended show. We roll into it right after DTNS wraps up. But just a reminder, Daily Tech News Show is live. Monday through Friday, 4 p.m. Eastern, 2100 UTC. Find out more at DailyTechNewsShow.com slash live. Tell a friend. Maybe they'll like the show as well. We're back tomorrow talking municipal broadband with Shannon Morse and Len Pearl to joining us as well. Talk to you then.