 Daily Tech News show made possible by you, its listeners, thanks to all of you, including Dale McKahy, Maz Aglen, Kelly Cook, and Andrew Pearl. On this episode of DTNS, Meta goes all in on AI. Again, what does Google's new big reorg mean for hardware? And Dr. Nicky explains nanomotors and how they may help against bacterial infections. I'm Dr. Nicky Ackermans. And I'm the show's producer, Roger Chang. How are the goats today, Nicky? They are in heaven. Oh, okay. Well, just to say, they just ate a lot of ice cream and are happy. They did. They actually did. All right, well, we do have some tech news to talk about. So let's do that starting with the Quick Hits. Twitch is launching a TikTok style for you discovery feed that has been testing for about a year to all Twitch users later this month. The new feed will launch as a tab within the mobile app. And next month, at least some users might see this feed as the homepage app itself. Twitch says all live streams and clips are eligible to appear in its discovery feed as long as they meet the content guidelines. The new feed will include ads, but you can scroll past them like before without having to pause your overall experience. On Wednesday, the U.S. House of Representatives approved a bill requiring government authorities to obtain a warrant before purchasing commercially available data from third parties to carry out law enforcement activities. The Fourth Amendment is not for sale act past 219 to 199 enforces law enforcement and other government entities to get a warrant before buying information from third party data brokers who purchase information gleaned from apps. Huawei unveiled four models of its high-end Pura70 smartphone series believed to be powered by an advanced chip, a slight upgrade from last year's Mate 60. The Pura70 has premium features and price points that directly compete with Apple's iPhone 15 lineup. According to Huawei's official website, the Pura70 costs $54.991, the Pura70 Pro $64.991, the Pura70 Pro Plus $7,9991, and the Pura70 Ultra $9,9991, which is a range from about $760 to $1,380 U.S. dollars. Nothing introduced two new earbuds, the nothing ear for $149 and the nothing ear A for $99. These are the fourth and fifth headphone products from the company. The ear keeps the overall design of the ear one and ear two, but has a new personalization option, better battery life and a new base-enhancing setting. The ear A also comes in bright yellow as for what rival earbuds offer no spatial audio in either model, but the company plans to add chat GPT integration to both its smartphones and earbuds. On Thursday, the Defense Advanced Research Projects Agency, or DARPA, announced an AI-controlled jet successfully faced a human pilot in simulated air-to-air dogfight as part of the Air Combat Evolution Program. The test was carried out by the U.S. Air Force at Edwards Air Force Base. It pitted an F-16 against the experimental X-62A fitted with an AI system. Both plans were crude, but DARPA says the pilots on board the X-62A never needed to intervene. Both aircraft demonstrated high aspect nose-to-nose engagements and got as close as 2,000 feet at 1,200 miles per hour. No word, though, on who actually won. All right, Rob, let's talk about what META is up to today. So, yeah, there's lots of AI news from META today. Let's start with Lama 3, the company's latest open source large language model. META says Lama 3 is pre-trained on over 15 trillion tokens that were all collected from publicly available sources. Our training data set is seven times larger than it used for Lama 2 and includes four times more code. Around 5% of Lama 3's training data is high quality non-English data that covers over 30 languages, though META admits English is going to be higher performing overall to smaller Lama 3 models. An 8 billion parameter model and a 70 billion parameter model will both be accessible on all the major cloud providers starting today, both in the META AI Assistant and outside to developers. A larger multimodal version is set to release in the coming months. You know, when I was reading about Lama 3, because I think sometimes you can get a little into the weeds on, okay, there's Lama 1, Lama 2. Now, we've got Lama 3. META's obviously going to say it's better than Lama 2 in every single way. But one of the examples that the company gave that I thought was interesting was it can decrease false refusals over Lama 2. So Mark Zuckerberg announcing Lama 3 gave an example of if you asked it to make a killer Margarita in the past, it might say, you know, I can't answer a question like that, even though it's a harmless answer because it's a harmless question, but Lama 2 tended to get confused. I think this is kind of a lot of the real world stuff that people find AI is still lacking on. Yeah, this just seems like it's just getting better. Like it's learning, okay, wait a minute. This didn't really mean it was going to kill someone. This is just a drink that is a really, really good drink. So it seems like it's getting colloquialisms and things like that that it didn't necessarily get in Lama 2. That is a good thing. And we want to see these things improve until the conversation you have with them is just very, very conversational. And you may not even notice that you're talking to a bot. The Verge also notes that META is a little cagey when talking about exactly how Lama 3 is trained, but said the total training data set is seven times larger than Lama 2, as Rob mentioned, says no META user data is being used here, but it's using a mix of public internet data and synthetic AI generated data. So Nikki, does it bother you that AI is helping to train other AI? I think it's interesting, and I'm sure they're thinking about this, but I wonder about the redundancy error that that's going to create because we know that AI has its own gaps and it can't catch them, then it might create some loops, but I'm sure they're working on that. We've seen that in a lot of the, I don't know what you would call this, but the like image graveyards that are happening on Facebook where you just have AIs commenting on images and boosting them and then recreating more images and they just get weirder and weirder. I know it's not the same as a language model, but it'd be interesting. Hopefully they don't fall into that pitfall. I'm old enough to remember photo copy machines. You remember back in the day, you just kept making copies of copies and it progressively got worse? I kind of wonder is something like that going to happen here to where if you have AI that's not 100%, it is making AI clearly that AI that it makes can't be 100%. So at what point does it get really, really bad? Or does it get really, really funny? Well, yeah, because you, you know, some of the AI generated data that's training more AI started with public data from humans. But yes, there's something, you know, you can only make so many photo copies before. It does start to get weird and that's where I think we're still in that situation, particularly in photos and videos. Well, you know, in written AI as well, where it's like, where did that come from? That's just weird. That's, you know, where did it learn from unless it was just, you know, getting bad data in the first place? It's not unlike we don't get those now when, you know, it's trained off of human data. So we know it's not great. So like I said, I think there's going to be some, wow, I can't believe it did that as well as some good comedy that'll come out of these things when it's just going to be so clear that something was created by an AI that was created by an AI that was created by an AI that was created by an AI. Well, in more sort of real world meta news, Meta is also bringing its AI assistant, Meta AI, which I introduced last year into pretty much everything. Instagram, WhatsApp, Facebook, specifically in the search boxes of each app, plus inside the main Facebook feed. And now it has its own home on the internet, Meta.ai standalone website so you can play around with what it's offering. So Meta AI assistant integrates real time search results from both Bing and Google has an image generation tool that can create animations and high res images in real time as you type, kind of fun. Previously only available in the US now being rolled out in English to a variety of countries, Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia, Zimbabwe with other languages and countries on the way. Okay. Who wants AI in WhatsApp? I've seen some use cases where people use this as like a chat companion. I could see how that could be kind of fun until it gets boring. Sure. I don't, I use it. I'll use some AIs, not necessarily this one for assistance sometimes, but I still feel that I find the pitfalls too quickly to keep using it. As long as it works, I'm fine. And then it's like, I'm sorry, I can't help you with that. I'm like, all right, I'm done. I don't know how good this one specifically is. Have you guys used this one before? I mean, no, I don't do a lot of chatting within Instagram. I mean, I have a fair amount of folks that I chat within Facebook Messenger with, but not in a way that I'd say, okay, we're planning the friend camping trip. Let's use AI to figure out when we should all leave the house to get there at the same time type thing. Although that is an example of what it's supposed to be used for. Kind of real world stuff. I'm not a big user of meta tools either with the exception of WhatsApp and the exception of threads, which this is not applying to at this point. So when I think about WhatsApp, I just, I have noticed the icon that pops up for the AI, but it's just like, would I want to click that for? I just wanted to tell my wife to, you know, that I'm going to be home a little bit late because I stopped to get milk or something like that. I'm just, I'm not really getting that deep into where I feel like I need an AI bot in it. I'm not anti the AI bot being there. I just think I might need someone to show me a use case to where I say, oh, okay, that's what I use that for. 100%. I think meta.ai just as a standalone place to just play around might be a good place to start. They've got ideas, writing, fun, support, learning and imagined sections. And, you know, this is exactly what you're talking about, Rob. This is meta saying, you might just go, what do I need this for? Well, let's, you know, let's walk through making your next email sound as professional as possible or your school assignment being, you know, tip top or, you know, you create some beautiful watercolor painting type thing. I think there are so many of us who are still like, these are great. Wow, these models getting smarter every day, but how do I use it again? Yeah, it's just eventually, eventually we'll get there. But like I said, somebody please show me a use case. Make this be really cool for me because I use what's up a lot. It is my primary, it's my primary messaging app. So I'd like to see it be useful. I just don't know how it is yet. Well, all right, Rob. Let's talk about the shakeup within Google. Are we, are we, are we in dire straits here? What's up? Well, there is an article out today on The Verge about some fairly substantial reorganizations. Google CEO Sundar Pichai announced including the creation of a brand new team. Platforms and devices will oversee all of Google's Pixel products, all of Android, Chrome, Chrome OS photos and more and is being headed up by Rick Osterloh who was previously SVP of devices and services overseeing all of Google's hardware efforts. The goal of this team is to bring AI to your phone, your TV and literally everything else that runs Android. So what is really interesting to me about this is that this is a bit of a, about face for Google. Now they're not just doing it, they've been going through this for about the last, I would say 18 months, maybe as much as two years, but Google historically has been we're going to build Android and we're going to build hardware and they're going to be separate things, but that's not the case anymore. They are, they're bringing them back together. So this is a very Apple like thing that Google is doing where they're saying we've got our hardware, we've got our software. We're not going to show the rest of the world, the Android world with the best of both of those things are. Now one might say, well, hold on, if it was all under Google in the first place, why would the Android and hardware teams even be separated at all? Big reason is because Android has a bunch of OEMs. If you think about Apple, only Apple makes iPhones and iPads, but there's a lot of companies, Samsung, Motorola, Sony. I mean, there's just all kinds of companies that make Android devices. So what Google wanted to do is they wanted to make sure that they didn't feel that the OEMs didn't feel like that Google was giving themselves a lot of advantage by integrating their hardware very tightly with the software that everyone else got. That's changed though, that is going away. And, you know, and it probably is a good thing for Google, you know, in my opinion. The reason being is that I think by any measure, most would say that Google is a bit behind the likes of Microsoft and ChatGBT when it comes to AI. But the fact that they have a phone and even though we are in a iOS dominated world here in the United States, Android still pretty much dominates in the world. I want to say it's nearly 70% market share. So if Google can really focus on, you know, putting their AI on their own hardware using their own tensor AI chip and just show the world some really, really cool things, they may be able to not only catch up but actually surpass some of their rivals when it comes to AI because they are all about AI at this point, you know, Google, you know, for a lack of a better term is an AI company at this point. I mean, I guess if the Pixel was, the Pixel line was outselling Samsung's flagship phones, it might be a different conversation. But at this point, yeah, I think Google can get away with this. Google says, you know, the AI team needs to be talking to the hardware team. And, you know, now that we look at it, maybe it's all just one big team, you know, with one person at the helm, making sure that everybody is as cohesive as possible as they innovate. I also wonder if the OEMs of the world, you know, back in the day, everyone was kind of working on their own Android alternative. You know, maybe we could do, you know, some, you know, our own fork of Android type thing. And you see a lot of that still, but, you know, I mean, where are they going to go at this point? I don't think they're really going to go anywhere. I think that, you know, that if you're an OEM that's using Android, you're, you know, that is the horse you're tied to at this point. So there's that, but I do believe that Google is not going to go so far to where they just don't care about the OEM. They know that they still have to please Samsung in a big way. They know that they still have to please Motorola in a big way. So although this is a bit of an about face, it is not new. They've been doing this for about the last, as I said, 18 months, maybe as long as two years, you know, with the Pixel line of devices. So I think they're just going to lean more into that, particularly when it comes to their other hardware that aren't, you know, when it comes to TVs, they're going to put all the AI stuff in it. When it comes to Chrome OS stuff, they're going to put all the AI stuff into it. So they're really going to lean into this. And as I said, I think it is probably a good move for them because they are behind when it comes to AI overall. Hiroshi Lockheimer, who was the head of Android and Chrome and Chrome OS for quite some time, staying with the company, taking on other projects inside of Google and Alphabet. At least that's what Lockheimer and Austerlo told the Verge. Wonder what he'll be doing. Sounds like he's out of this department completely. It's interesting when a company says that, oh, they're going to go do this other thing. And it's like, yeah, are they going to do another thing at the company that they used to get a check from? Or is it going to go do something somewhere else? I think that remains to be seen. Sometimes it works out. Sometimes they'll move around in the company and still be happy. But when you were running something and you're no longer running it, that sounds like it's time for you to move on. Well, folks, want to recap of the week's tech headlines with insights into how technology affects and disaffects communities of color? Then check out The Tech John, where host Rob Dunwood, that's me, Stephanie Humphrey and Terence Gaines dive into the top tech stories of the week delivered from points of view you don't always hear in mainstream media. New episodes land Tuesday afternoons. Find out wherever you get your podcasts or visit thetechjon.com. That's thetechjwn.com. Last week, researchers from Vanderbilt University uncovered new information about how bacteria swim in the study, which was published in Nature Microbiology. The researchers share new insights on how a bacterial motor powers chemotaxis, a key finding in the process of bacterial infection. Nikki, I do not know what this all means, so you're here to break it down for us. What do they find? Yeah, let's go. This is a classic detainee topic, I think. So first of all, I'll break that down. Chemotaxis is basically how a cell or an organism moves in a certain direction. Specifically, it's either moving towards a increasing or decreasing concentration of a chemical. So it's either trying to find food or running away from something, basically. For bacteria, the chemotaxis is used to swim towards energy-rich molecules for feeding, and they use it also to look for host that they want to infect or to avoid predators. So because of this link to infection, chemotaxis is a really interesting target for new therapeutics, but it's still not very well understood. So how did finding all of this out? How is this going to help? Well, I'll explain this a bit further. So in order to swim towards their target, these bacteria have a flagellum, so it's like a tail kind of appendage, similar to what sperm cells have, if you can visualize that. And that's where this kind of tech aspect comes in. So they have a small motor, a biological motor that spins this flagellum and provides propulsion, kind of like a very, very small propeller, but teeny tiny. So we're at the order, it's about 45 nanometers. So it's a nano motor. And we knew that this mechanism existed, but the researchers on this team just hadn't figured out the specific architecture of this tiny little motor, how the components worked out and how they worked together. And so because of that, they couldn't really find a use for it for targeting chemotaxis with drugs. But in this new study, we got new information about how the motor provides torque to spin the flagellum and how it's able to switch directions so the bacteria can swim forward and backwards. So the researchers didn't quite understand how this all worked before. How were they able to figure it out? I mean, is it just that it was too small in the past or they had to study it for a certain amount of time or both? A little bit of both. Because it's extremely small, one of the main problems is being able to visualize it. And to do that, these researchers used Salmonella and Terica as a model. So you guys know about Salmonella. This is a bacteria that causes over 60,000 deaths a year through infection. Side note, I was looking up on the CDC, what was the main cause of Salmonella. And some people the most at risk are adventurous eaters. So stop being adventurous, apparently. But anyway, the researchers isolated the flagella from these bacteria and they used something called a cryoelectron microscope. So we're zooming in very, very, very closely to get really detailed imaging structures of this motor, which to this point hadn't been imaged at that detail. And seeing that kind of detail allowed them to understand how this motor actually functions. So you said that these work almost like a propeller and I'm thinking of an actual boat propeller. But what do these nature made motors actually look like? Great question. And it's so interesting because it does look a lot like what you would see in a car or an axle or something. And there are visuals in the paper, but I'll try and describe it for the listeners. So the cryoelectron imaging revealed that there's a complex of rings. So there's an inner ring that's called the cytoplasmic ring or a C ring. And it's basically a circle built out of a bunch of copies of the same protein. And this ring is the one that flips the flagellum rotation clockwise or counterclockwise by like actually physically kind of changing gears. And this then generates torque and spins the flagellum through electrostatic an interaction with a stator, which is like a stationary component. And that's kind of similar to the way that a man made motor works and not an engineer, but I'm pretty sure that's how that goes. And then there's another ring that connects the C ring to the flagellum rod. So like basically the tail of the bacteria and then two other rings that help with rotation, support and buffering so that it's not like wobbling all over the place. And this paper basically offered kind of a much better resolution on this main C ring and how it works, how it spins and also how different proteins can make the motor go faster, slower or stop moving altogether. I mean, if you're in the research field, this is obviously a big deal and a lot of future applications are being thought of right now and grants, hopefully being doled out to researchers who can make people be, even if they're adventurous eaters, less sick. So what are some future applications? Yeah, you got it. So basically a lot of bacteria share the same form of chemotaxis, Acherichicoli or E. coli, for example, has like a 95% similarity of the same motor that Salmonella has. So if we're able to stop this chemotaxis within the organism like inside a human, you would be able to stop recurrent infections from happening without impacting your microbiome. So you wouldn't have to take antibiotics. So it's kind of a mechanical way to stop infections, which is really, really interesting, especially considering the antibiotic crisis that we're having. And the authors didn't mention this too much because I think they don't want to say it out loud, but this could serve as some kind of inspiration for something like nano medicine, nano bots. But I think we're a bit far away from that. But yeah, that's kind of the future impacts of this finding, which I thought was really interesting. I mean, even to just replace antibiotics, which as you mentioned, I mean, have their own host of problems, overprescribed and can make you sick in other ways. To be able to defeat a bacterial infection in a different way, I mean, that seems huge. Oh yeah, I mean, we're just at the beginning of just understanding this structure, which is awesome. But yeah, there's a lot more to be done. And somebody has already built this wheel in Minecraft. I saw a video of that earlier. Of course they have. I'm just imagining someone, if they can figure out the shrinking science that someone will make a craft that's powered by this like interspaced in 1987 movie where they struck the ship down and they were traveling around inside the body. That's just kind of cool to me. Was that Martin Short? I was just thinking about Ms. Frizzle. But you know those like, those rides where you're like kind of go inside and it spins and you're just like stuck to the wall. That's what that reminds me of. All right. Well, Nikki, as usual, thank you for bringing the knowledge. It's fascinating what is going on in the research. Scientific departments of the world of which you are one. So thank you again. This was awesome. Oh, no worries. I always love bringing the nanobots to DTNS. And we love having them. We also love your feedback. So in that vein, let's check out the mail bag. Anon Juna wrote in saying that they don't think YouTube is wrong for going after ad blocking apps. We talked about that on Tuesday's show. But Anon also says it would be good to figure out why the number of people opting for this versus premium. I do use an ad blocker and I'm a creator on YouTube. Anon Juna says it's been dormant since I got a cancer diagnosis. I'll be back soon. We very much hope that's true. Get well soon. He says for me, it's because the number and length of ads have made even short videos unwatchable. Three minutes of ads for an eight minute video and that's just on home internet. That's a lot of mobile data if you're not on an unlimited plan. Premium is just out of my price bracket for now. But I'll probably sub once the medical bills are done. Yeah. Good point. YouTube premium. I mean, a lot of people are paying for YouTube premium, but a good point that if you opt for an ad blocking app instead, well, either not enough people know about premium as an option or don't feel that it's worth it. Yeah. Google is going to make sure that they are going, you know, they're saying that we want to make sure that the creators are made whole, but they also want to make sure that they're made whole too since they get half of half of that ad revenue. So yeah, I see them cracking down on this and it's going to be more of the same. You're going to see more and more apps stop working. But Carl also chimed in about a new app air chat on Tuesday show in the ad. He basically says, is it wrong to think of this service as tiktok radio? Feels like there is a void of short form audio content. Video spans the gamut from tiktok to YouTube shorts through YouTube TV shows and movies. Tech spans from tweets through blogs, articles to books. Audio really doesn't have that shortest form and goes from the medium of, you know, of the length of podcast audio books. I feel like there might be a niche for short form audio content and maybe this service could grow to feel that. Am I wrong? I don't know, Carl. That's a really good point. Carl needs to go listen to the Daily Tech headlines. Yeah, right. Hey, I have around five minutes. I have around five minutes every day. But as far as, yeah, like comparing it to tiktok radio, what do you think, Rob? I think Carl might be onto something. So I remember, you know, apps like this. I think now the most famous one is probably going to be X in their spaces. But, you know, you had Clubhouse. Clubhouse during the pandemic, it blew up. The problem is that I think they waited too long to get it over to Android. So it just kind of fizzled out because people went over to Twitter now X and, you know, got into spaces over there. But I'm kind of for this. I like being able to have these audio type of conversations with people. It reminds me when I was a kid, I remember like my older brother and older sister, they would get in like these chat rooms and there will literally be dozens if not hundreds of people on these party line chats and they're just having their conversations going back and forth. And I see this doing something similar to that as some of the other applications that came before it. So I'm a fan if they can get this right. Well, thanks to everybody who sends us feedback. We are fans of you feedback at daily tech new show dot com. Keep the feedback coming and Dr. Nikki Ackerman's keep coming back to the show as well. Thank you so much for being with us today. Let folks know where they can keep up with your work. Always it is at Nicole Ackerman's dot com. That's where all the news and all the stuff is at. And patrons stick around for the extended show Good Day Internet Tic Tac is top dog when it comes to short phone video. But can it realistically take on Instagram in the photo space? We'll talk about that in the company's latest effort to try. Just a reminder, you can catch our show live Monday through Friday at 4 p.m. Eastern twenty hundred UTC and you can find out more at daily tech new show dot com slash live. We're back doing it all again tomorrow talking about the funding of social media platforms that are considered essential with Nate Langson and Len Peralta draw on the top tech stories. Talk to you then. The DTNS family of podcasts helping each other understand Diamond Club hopes you have enjoyed this program.