 This 10th year of Daily Tech News show is made possible by you. Thank you for being here. Thanks to all of you, including Jeffrey Zilx, Tony Glass, and Philip Less. Coming up on DTNS, can tech learn from bats? Yes, and Blair Basdrich from This Week in Science will tell us how. Plus, Microsoft is getting all 90s bossy again, and why Generative AI is affecting software developers the most. This is the Daily Tech News for Monday, March 27th, 2023 in Los Angeles. I'm Tom Merritt. I'm from lovely Cleveland, Ohio. I'm Rich Frappolino. I'm the show's producer, Roger Chang. And joining us, Blair Basdrich, co-host of This Week in Science. Welcome back, Blair. Thanks so much. Happy to be here. Thank you for bringing the bats. You got it any time. Baseball season is opening soon. We needed bats. Let's start with the quick hits, shall we? Last Friday, GitHub complied with a DMCA infringement notice from Twitter that sought to remove proprietary source code and tools that had been on GitHub for months at that point. Twitter now wants a subpoena to search for those responsible for leaking and downloading its code. The DMCA notice says someone with the handle free speech enthusiast is behind the leak. And Twitter has indicated that GitHub should provide info about that accounts access history as a solution to the copyright infringement case. First Citizens Bank shares has agreed to buy Silicon Valley Bank. The FDIC says that First Citizens is buying around $72 billion of SVB's assets at a discount of a mere $16.5 billion, which still leaves the FDIC in control of roughly $90 billion in securities and other assets. Keep track of all those billions. $20 billion of the purchase price will go to the Deposit Insurance Fund with the FDIC getting equity appreciation rights in First Citizens. Alibaba's co-founder Jack Ma, once the face of tech in China, made some comments about China's financial regulators and was called in for a stern talking to back in 2020, after which he dropped off the radar. In fact, he spent the entire last year outside of China, mainly in Japan. He bumped around to a few other places. So it's worth noting that he has now returned to China. In fact, visit a school in Hangzhou where Alibaba is based. He spoke to students there about AI saying we are going to use AI to solve problems instead of being controlled by AI. China has cracked down hard on tech entrepreneurs in the software and services sector since 2020, but Ma's return could indicate that that crackdown is finally easing. Microsoft launched a public preview of its latest version of Teams. Microsoft calls this a reimagining of Teams from the ground up, featuring twice the performance while using half the memory as before. Joining meaning should now be two times faster. Toggling between chat should be 1.7 times faster. It's not a clean number, but it's still impressive. Mac support is due later this year as well. Microsoft also announced Teams has 280 million monthly active users up from 270 million in January 2022. So they're killing slack slowly, is what I'm hearing. Following up on the lawsuit between the Internet Archive and book publisher Heshet et al. Judge John G. Keltel ruled Friday that the archive created derivative works and did not have a fair use of them and therefore should not have lent out copies of books digitally without permission from the publisher. If you recall from our previous conversation, the Internet Archive had departed from controlled digital lending during lockdowns in 2020 to eliminate wait lists for book borrowing. If you don't remember, controlled digital lending means only lending out one electronic copy for every physical copy that the library holds. The Internet Archive says it will repeal the ruling. Let's talk licenses. Ooh, it's one of my favorite things to talk about that. And Microsoft has really joined my two favorite. Microsoft licenses its Bing search index to other providers on top of obviously operating Bing itself. Companies like DuckDuckGo, Yahoo, Neva, and U.com all are licensees. And Bing has become a lot more interesting to people since it integrated OpenAI into a chat sidebar. Lots of people are trying it again or given to try for the first time. Lots of other search engines have been racing to add similar functionality, which means they need to train their chatbots on a search index. Maybe you see where I'm going here. If you're Google, of course, you just use your own. But if you're U.com or DuckDuckGo, well, you're getting a search index from Bing's... Oh, nope, just checked Bloomberg sources say Microsoft informed at least two customers that it will terminate their license to use the Bing search index if they use it to train AI tools. Hey, what happened to the kinder, kindler, gentler Microsoft under, you know, kind Uncle Sacha? Does that only apply when Microsoft isn't winning, Tom? It does look that way of like, oh, if we're the most popular chatbot in town, we want to protect it. I also think there might be some liability issues here. If you train off the Bing index, there is a certain bias to any index you train off of that Microsoft knows best how to adapt. So it's possible that they are saying, hey, don't train off our index unless you talk to us first. That way we can give you guidance on how to adapt to, you know, the SKU that's in our particular index because every training set has one. You know, Blair, I know you probably don't follow this super closely, but have you played with any of these chatbots yet? A little bit. Isn't the Bing chatbot the one that declared love for somebody and asked them to leave their wife? They all have done something similar in Bing's defense. Yeah, like as soon as these things come out, somebody tries to figure out how to make it do something silly sounding and succeeds because you can make them do lots of stuff. But yes, I think you're right. The Bing is the one that did that. Yeah, and isn't DuckDuckGo's whole thing that they don't track your data? So they would have to use Microsoft's data to be able to even use an AI chatbot. Right, that's an interesting question. If they're getting the search index, they wouldn't have any personal data. But could that, the fact that Microsoft has personal data somehow, you know, work its way into the search index. A search index is public. And if DuckDuckGo is using the search index to feed their search results, then I guess training off it isn't any different than serving up the search results. I'm gonna do something that goes against my former Linux hippie heart. And I'm going to defend Microsoft because... This is gonna hurt, go for it. Here's everything that we've talked about with the kindler general of Microsoft is that they're not forcing everybody to like buy a Windows license, right? They're like, they're happy to like sell you stuff other like in almost every other situation, right? But it's not like they're giving anything away. Sure, they contribute more to open source. Great, congratulations guys. But it's like they still are like just a massively popular services company now as opposed to a massively popular OS company. So like when I see that they're saying, hey, you can't use a search index. It's because you signed up to use the search index so that you can pay us amount of money so that you can make money serving ads to people making a search engine, right? Like fundamentally a chatbot is a different product, right? So I don't think this is Microsoft saying you can't use the Bing search index for training chatbots. You have to pay us a license to use it for that purpose versus operating a search engine. I honestly think that's a perfectly reasonable like business decision in that this is an asset. If I'm paying for access to an index, why shouldn't I be able to do whatever I want with it? Because if you see subsection three of clause 32 because you totally not do it really. Yes, I mean, now if they are wholesale just saying like we're changing the terms of it and I don't know how often these license agreements were new or anything like that. Maybe that feels a little bit more, hey, all of a sudden we have a lead in something as opposed to being an afterthought in the search engine market. And so we're going to put our foot on your throat or something like that, but this just feels more like we want to create a service offering so that we can appropriately profit off of this new hot category versus just like pure where we're cutting out everybody else. I think it's probably a combo, right? There's probably liability issues and guidance issues that are legitimate that they want to use. I also think that they will charge you for offering those guidance and why wouldn't they? So yeah, I don't know. Or even just optics like Microsoft has, with even with all the headlines of a chat bot says it loves somebody or left. I made me leave my wife or something like that. You know, all of the National Enquirer headlines that come out of this like Microsoft at some point is a big enough organization that they knew they were kind of going into this road. So even if it's not like surely liability, like someone's going to sue us, I think they would do want to be in control of the optics of this for a while so that a company, you know, that DuckDuckGo or Neva not that they're not unprepared for this kind of thing, but they're not prepared for it on Microsoft's terms. So even if it's just optics, not even we're worried about X amount of financial obligation as a result of that, I totally buy that. Well, one of the things you can use generative AI for with abandon is creating software. It's apparently really good at coding. Scott Rosenberg at Axios has an article called AI Changes the Software Making Game, and he notes that software developers seem to be feeling the biggest effect of generative AI. He's quoting a couple of venture investors in this article, Paul Kedroski and Eric Norlin who wrote, such technologies are terrific to the point of dark magic at producing, debugging and accelerating software production quickly and almost costlessly. That GPT can write code based on human language instructions from non-programmers and explain how that code works. So you can just say like, write me a script that will show a block bouncing back and forth between two paddles and it can do stuff like that. The code still might need debugging. It isn't always bug free, but it's pretty good. And in fact, you could get GPT to debug its own code. Humans are still going to be needed though to check the machine's work to develop new kinds of programming systems. But Rosenberg wrote, there may be much less demand for the routine labor of taking existing software systems and wiring them up to work together. That's a lot of what developers do today. We've talked about no code in the past, meaning like, oh, you can create programs without having to know how to code, but those have been fairly limited. This feels closer to limit less. So Blair, if we said you can just create your own apps, use chat GPT, just tell it, describe what kind of software you want. Could you use that in your job? Oh, absolutely. I mean, in my day-to-day job, I don't know, but just for fun in like my animal research, loving, birding world, absolutely. The thing that comes to mind immediately is I saw this animal. I can describe it verbally. What was it? So it was a bird. It had a long neck. It had a curly, pointy beak. It was like reddish-brownish and black and white. Oh, that was an appa set. Okay, that's what you saw, right? So I would absolutely love to have that because there's things like seek, but you have to get so close and take a photo and even then it's not that accurate. So it would be great. They could do that. Of course, the thing that I worry about though is the big AI question, which is truthfulness. So if it wasn't sure, would it make up an answer, right? And I guess that would be down to the code. I guess what I would say is if I were in that situation, I would tell ChatGPT, write me a program that can draw on a database to match my descriptions with actual birds and then you provide the database like that, you know, and that way you're not asking ChatGPT to identify the birds at that point. You're saying, you know, use this reliable information. Right, yeah, here's a wildlife guide for my area. Yeah, yeah. Yeah, and then if it's going to hallucinate, it'll just have broken code and then you won't have an app, but at least it won't be giving you, like that's the best case scenario. But it's interesting with all the ChatGPT plugin talk that I noticed Zapier was one of the companies that they were testing with and this seems to me like, I don't know, like a Zapier killer to be quite honest. I mean, I'm sure that's why they're on top of it, right, because they see this as the next evolution of their services as well. But there are so many services like that where it's like, I know that there are APIs that can do specific functions and that I would like those to tie into other services. I don't know how to code, so I really can't do that. But if I know that an API is the capability to do this, all of a sudden I'm not reliant on if or Zapier or something like that or as reliable to do that. I could imagine Zapier actually training ChatGPT or a Chatbot specifically to like kind of answer queries of like, whenever I get this email, I want you to save it to this place or I want you to perform this action on it in a way that is real language. I mean, a lot of those services, if does a better job of giving you this extremely narrowed down sandbox of functions, that'll kind of all plug in and talk to each other, Zapier can get a little more of a hairy sometimes depending on how deep in the weeds you get into that. But if you can have something that's just like pure text put over that, Tom, you mentioned about no code. I mean, that to me feels much more functional than here is this stripped down baby version of a coding interface that'll let you do like these four things that we kind of have mapped out all the possible things that it can do. And yeah, you can create some like very specific business functions or something like that. We've seen Salesforce and Google buy up companies in the past couple of years to that effect because there is a business use case for that. But this to me feels so much more expansive and useful from a non-coder. But we've also seen studies from GitHub and from others that this kind of stuff like makes real world differences for people that are in the weeds for developers for sure. Yeah, War Singer in our own Discord wrote a couple of days ago. I learned that chat GPT is pretty good at writing shell scripts way better than I am even having to Google everything. If that's part of your job or life, give it a try. It's also good at modifying them with new requirements or when it doesn't get things quite right. It's less good but still better than me at the AWS CLI commands but probably saved me about 50% of the work and created a more robust solution than I would have come up with without a lot more work. What I took away from this is you're still going to need to know a little bit about programming about what's possible and what's not possible to make this effective. But for a developer, they do. So this is going to be incredibly helpful for someone who is unfamiliar with coding. They may still need help. They may still need some instructions on how to write those prompts. Probably need someone to double check the code before you implement it just to make sure there's no bugs in there. But it's certainly going to speed stuff up, right? Yeah, it seems like and make like quality of life improvements like, you know, there's with every job developers included. There's a lot of drudgery in these kind of stuffs and there's a reason, right? Like Microsoft launched a lot of their AI initiatives with GitHub's co-pilot because they very clearly saw something, one that it would be good at and two has a business use case. Obviously they can also sell service based on that. No surprise there. But like, you know, we were seeing reports from GitHub that, you know, it allowed developers like 73% of developers in a survey, they said help them to kind of stay in their flow state. You know, when they're working on there, having co-pilot to just, okay, it's going to auto suggest stuff. I'm not having to go to stack overflow and look up something for a half hour or something like that. And really helped them also 87% help with repetitive tasks as well. Again, like just taking out of that drudgery and hopefully opening the door to, you know, more creative stuff as well. Developer tested, developer approved, chat GPT. Go check it out. But if only Rich, if only for those of us who are still like, wait, but what is chat GPT actually? Like what does it do? If only there were like five simple ways to explain that. Well, then you should check out Tom's new show, Tom's top five. This is where Tom, you may know him, will break down the top five things you need to know about technology. Everything from AI, chat GPT, to online services, security, even tech legislation. There will be five things. You can watch it at youtube.com slash daily technician. Bats use echolocation to quote unquote see objects in areas with low or no light, you know, like bat caves. They send ultrasonic pulses from their mouths and nose and receive the echoes with their ears to create a 3D map of their space. It's like a map made of tiny screams. It sounds fun, right? Well, bio engineers from Hiroshima University in Japan have studied the movements of bat ears to determine how they receive echolocation signals and then create a mathematical model that could amplify direction detection and replicate the bats abilities in human technology, thus making us all bat people. Blair, what did they learn? Well, it's not surprising. It's that the kind of the angle and vector that the ears are pointing in help hone in on locations of sounds and locations of things in the distance, be it a wall, another bat or some prey. And they found that when they were opposed, it did the best job when they were kind of pointing in opposite directions and that you can actually apply that information and the model that they created into potentially creating sonar systems artificially. And so of course that could be used in drones in all sorts of kind of actual helpful technology but also you could potentially make a real batman cowl with ears that move that you could actually use sonar in, which would be pretty cool. So we're talking, they were able to reverse engineer the system then. Yes, so basically they looked at bats, they noticed the way that they work. Sometimes bats have this kind of constant frequency that doesn't change. Sometimes they modulate their frequency as they get closer or farther away. They might change how high or low the screeching is. And with that in tandem, they have this ear movement that's helping them completely match their space. And before we go any further, I will debunk the myth that bats are blind. They are not blind, they can see, but the hearing and the sonar system really, really helps in that it is still pretty dark where they live. And so even though they see, they might not be able to see well. So with all that in mind, they can kind of map their surroundings and track moving targets with the movements of their ears. And just like with binocular vision, like how we see our two eyes in the front of our face, we can hone in on something really close with animals with monocular vision, that's like a chameleon or a rabbit with eyes on the sides. You can get a 360 view of your space. And so they can actually use their ears to do both. They can hone in on a thing in front of them or they can turn their ears sideways and map their space in terms of the soundscape to see where they are in space. Eyes in the back of their head. Exactly, yes. All right, except with ears. Like we do this in a very clunky way, right? If you're trying to listen or hear something, you cock your head or you turn to hear. It just sounds like they have a lot more control and a lot more precision than we have. Yes, absolutely. And when I used to do overnight tours at the zoo, I would have kids in the dark stand by a fountain and then I'd have them cup their ears in one direction and then the other and they could actually pinpoint where the water was coming from based on kind of the cupping of their ears so they could hone in on the location. And that's exactly what the ear flap, the structural ear on the bat is doing. But so they took kind of this observation, put it into a mathematical model so that they could manipulate it on three axes and the X, the Y and the Z axis to figure out exactly how to move these ears to best map a space. So that's something that they can now apply to other technology. Now, I do want ears on my smartphone. You mentioned us having batcalls. Yes. Obviously those will be things that happened, of course. What else? Just the next step, for sure. Yeah, what else would they be able to do with this? So they just basically, the next step, it's still baby steps. So this study was just creating the mathematical model. Their next step is to create a practical sensing system with pseudo-moving ears so that it creates a simple sonar for 3D navigation. And so they, like I said, this could go on drones. This could go, I don't know, on a remote control cockroach. That's like always where my brain goes. Yeah, remote control cockroach is a good one. That is part of it. But I also think this is something that you can apply to other animals. There are lots of animals that use sound data, the way that they behave. And so you could actually track visually through, like, high-speed cameras. You could see how other animals' rabbits, for example, how they move their ears to see what they're thinking, where they're listening. There's a lot of things you could do with this model. There's a use of Wi-Fi to map and detect motion in a space. This could potentially improve that, too, I guess. Yeah, it's just basically another data point. It's we use five senses to get around our space. And this is just another way that we can do that artificially. It's another sense. It's another tool. Yeah, so those cute, fuzzy ear attachments for your phone will soon be very useful. There you go. Maybe not soon, but it's okay. Well, something that might also be useful in the future is a Dyson sphere. Now, that may sound like a fancy, if overly expensive vacuum. But it's actually a concept outlined in the 1960s by physicist Freeman Dyson, who imagined that an advanced civilization could eventually enclose its home star to capture as much solar energy as possible. Make a big sphere around a star. Sounds good in theory, but how long would it take a galactic civilization to kind of break even from the energy cost of building such a massive sphere? Well, theoretical cosmologist Paul Sutter wrote a piece for Artist Tech of the Gun. He did the back of the envelope map for us. So put away your calculators. He's done some of the hard work here. If we convert all of the Earth into a series of flat panels to better collect solar radiation in this scenario, I believe there were one kilometer thick, assuming a 10% efficiency in collection like we all do, it would take about 60,000 years to recoup the energy investment. Of course, if we move the Earth's orbit closer to the sun, that could cut it down to a mere 10,000 years. I don't know what that would do the ecology of life, but 50,000 years, we could figure it out. So it might require some long-term planning and maybe a place to stay while it's being built. Yeah, just a place for humanity to live for a good 10,000 to 60,000 years, but worth it. Well, and my favorite part about this is Sutter also didn't just look at Earth. He was like, well, if we use Mercury, it's a smaller mass, but it's way closer to the sun. And then he was like, and Jupiter way further away, totally not. Hold on, hold on, Mercury, Venus. Why we convert Venus into the flat panels? It's a closer already. True, although I don't think it's, well, no, it's the sister planets Earth. It's almost the same size. There we go. Man, we just need to get a little better, more efficient in the collection of the radiation. So let's start a change.org petition right now. Build the Dyson Sphere out of Venus. And he's just building the panels with this. He's using all of Earth to build the panels. That's just for power. And yeah, and really it's not a Dyson Sphere. It's like a Dyson Collect 0.004. We're going to need to take Mars to make the surface of it if we don't want to use Earth for any of this. So yeah, it's getting complicated. For the whole sphere, we'd have to bring in other planetary systems, deconstruct them in like an Independence Day situation and then move them over to here to build our giant sphere. So you're saying there's a chance. Give me 60,000 years. You never know what's going to happen. I appreciate that this one was a lattice structure as opposed to in sci-fi. You usually see it's kind of an enclosed cave switch. It's not practical for so many reasons and it also cuts off the entire rest of the solar system from energy source. So if we use Venus, we can collect everything and then just a big wire to bring the power into Earth. Right. And then, you know, who knows what the loss of the mass of Venus, where it's supposed to be, will then do to the solar system. What I'm hearing is details, okay? We need to think. Work the problem. Yeah. We need to make this fit into the orbit for this. Using nothing. All right, all right. Let's check out the mailbag in response to our discussion about Framework's new expansion base system for their modular laptops. Damon wrote in to add, I believe this is perfect for the education sector. Students that are interested in computer science could be issued a device like this in their freshman year of high school and they could keep it and easily upgrade it until they graduate. After the student graduates, they would be given the device. This computer could follow them to college and beyond. This would get a generation used to a modular design and build a customer base for life. So Damon's like, steal a page from Google Chromebook for Framework and you actually have a more sustainable solution because they're not going to throw it away. Well, and I like as a student chooses to, hey, I wanted to go after graphic design or I want to go after this, like I can slap on that GPU or slapping more storage or upgrade my CPU. I think that's, yeah, I mean, I'm sure Framework would love to get in there too. Yeah, yeah. Framework's like, please tell us who to call at the biggest schools. We love this big contract with your school district, please. A really good point in the chat. Do you know how beat up textbooks get? Are you going to make them cover it in a paper bag cover? Well, that kids have laptops now. They have Chromebooks now. So you just got to, you know, and if the outside gets beat up, the internal components are fine. So you could just buy a new framework and closure and transfer all the components over. That's right. Plus if you're worried about them losing it, you just put a couple of bat ears on it and you're good to go. There you go. You can always find it. Well, thank you so much Blair Bazrich for filling us in on all of the fabulous bat tech that's in our near future. I absolutely love it. I'm just thinking of weird simulated ears while I was researching for today's show. It was very bizarre. Where can people find more of your great stuff if they're so inclined? Go to twist.org. That's TWIS as in this week in science. We broadcast weekly at 8 p.m. Pacific Time on YouTube, Facebook, Twitch, all the places. Just go to twist.org. You'll find out all about it. And we are also, of course, available as a podcast on Stitcher, Spreaker, the podcast app, all of the places podcasts exist. We're there. And you can also support us on patreon.com slash this week in science. Do all of those things, everybody. And speaking of Patreon, you can support us as well. In fact, thanks to our brand new boss, Colin, just started backing us on Patreon. Welcome, Colin, all the other patrons. Give Colin a big warm DTNS. Welcome. Help him out in the Discord. Help him out on the Patreon comments. And Colin, you get to stick around for the extended show now. Yeah, good day. Internet's going to be happening right after our stream here. And we'll be talking about the viral fame of swag Pope and how we're reacting to plausible AI-generated imagery. Oh, you don't want to miss that. You can also catch us live. If that's your preferred way Monday through Friday, 4 p.m. Eastern, 200 UTC. Find out more at DailyTechNewShow.com slash live back tomorrow talking about AI book narration with Erin Carson. She wrote that up for CNET. Talk to you then.