 Coming up on DTNS, Intel doubles down on AI chips, LinkedIn snuffs out the fakes and the human eye faces some stiff competition. This is for Wednesday, August 21st, 2019 from Studio Feline. I'm Sarah Lane. From Salt Lake City, Utah. I'm Scott Johnson. And I'm the show's producer, Roger Chan. Before the show, Roger and Amos and Scott and I were all talking about backernims. Yes. That's an acronym that goes in reverse, so to speak. If you're confused and want to know a little bit more about how we put this together, you can get the wider conversation on our expanded show. That's good day internet. We do it before and after DTNS. And you can join us and hear all the good stuff like backernims and more by becoming a member at patreon.com slash DTNS. All right, let's start with a few tech things you should know. Google started rolling out an update for G Suite that uses AI for making real time spell check suggestions while detecting potential grammar issues, including the addition of as you type auto correction for common spelling errors. The feature is already available for rapid release domains and is coming to scheduled release domains across all G Suite additions by September 12th. It's like they know me. Google and Mozilla are both blocking a route certificate issued by the Kazakhstan government, which previously had been mandatory to access the internet. Kazakhstan halted deployment earlier this month following legal challenges. The certificate let the government mitigate HTTPS encryption with a man in the middle attack on 37 domains, including Facebook, Twitter and Google. Google also added the certificate to the block list in Chromium source code. Speaking of Chromium, Microsoft released the first beta of its Chromium based edge browser. According to VP of Windows Experience, Joan Belfiore, the company hopes to have a stable release out by late 2019 or early 2020. Previously, Chromium based edge builds had only been available as less stable developer versions or nightly canary builds. The beta is available for Windows 7, Windows 10 and macOS. Very nice minds installing right now. Hey, check this out. Apple announced the Shazam Discovery Top 50 playlist for Apple Music. This will use global Shazam song identification data to create a list of trending songs based either on rapid growth, steady growth or geographic growth. Apple stated that Shazam app is used 20 million times a day to identify a song. That's a really interesting metric, isn't it? It's a bunch of people who are like, oh, this song, who is this? Rather than I know who this is, that's why I like this song. Yeah, it's very interesting. All right, let's talk a little bit more about neural network processors, shall we? At the Hot Chips Conference in Palo Alto, Intel, which is going on this week, Intel unveiled two Nirvana neural network processors designed to accelerate training and inferencing on AI models. Developed in partnership with Baidu, the Nirvana neural network processor for training is designed to handle data from a wide variety of deep learning models like Baidu's Paddle Paddle, Facebook's PyTorch, and TensorFlow, Google's TensorFlow, with 24 TensorFlow processors per chip capable of up to 119 Tera operations per second. The Nirvana neural network processor for inference is designed to make getting insights from train data sets more efficient and faster, offering scalable power from 10 to 50 watts per chip and up to 4.8 Tera operations per second per watt. That translates to capable capability of processing 360 images per second per watt in an image recognition model. So it's big stuff. Reuters reports that Facebook is also already using these new processors. So this, as far as I'm concerned, is next step stuff. Because what we're talking about here is the acceleration of testing of our ability to throw scenarios at AI learning and quickly learn what's sticking and what isn't. Like this is the stage. It reminds me of like, this is the test driving part of the brand new car you're building up till now it's been in the shop. You've seen some cool stuff. You were told it would be cool, but you weren't really sure until you hit the open road. Now you're on the oval and you're just screaming and that's going to just keep growing. This is super exciting. If you're into this stuff and you've been following it all or at all, this is a huge jump forward. I'm kind of one of those people who if I have, let's say I'm buying a new computer and I've got like base model versus fully loaded model. I try to find reasons to get the fully loaded model even if I know I don't really need to access all that power. This is one of the things where I'm like, I want one of these for my AI collection to process things really quickly. We'll get there. Yeah, one way or the other we'll get there. But I do, I predict we're on the fast track now. Things are going to get real interesting real fast. Both good and bad. I'm excited. Tuck in everybody. Also in the news, LinkedIn announced a blog in a blog post rather that it blocked or remove 21.6 million. It's a lot fake accounts from January to June of this year. That's 2019. LinkedIn's head of trust and safety, Paul Rockwell wrote that 19.5 million were auto blocked at the registration at registration stage and never went live. So they were never there to begin with. LinkedIn took down 2 million live fake accounts and members flag 67,000 more compared to Facebook has removed 3 billion fake accounts between October 2018 and March of 2019. LinkedIn fake accounts problems could be worse. Yeah. It's funny. Just just over a week ago, we we New York Times ran a story about how LinkedIn seemed to be faring very well in the age of social networks having to battle a lot of a lot of this, you know, privacy issues, trust issues, fake accounts, fake news. And it's almost like LinkedIn is like, well, we get it to everybody. We're just pretty good at it. You know, between us, our automated system, flagging stuff pretty quickly and then having the community handle the rest. We're mostly on top of it, but it did get me thinking. I had never considered because I'm a I'm a trustworthy person. Please hire me. I had never considered making a fake LinkedIn account. Now that's not necessarily what they're talking about making a fake account where you lie about where you went to college. But I'm sure that happens all the time. A lot of the examples that LinkedIn point to where he just kind of spammy stuff, people getting messages from bots and just stuff that was was adding to noise that wasn't really part of the LinkedIn experience or shouldn't have been. But yeah, you know, that's something that LinkedIn has to be on top of right. You have to make sure, well, Microsoft at this point has to make sure that it's a trustworthy place where you're not going to get a bunch of, you know, fake job information, you know, and and fake accolades for previous employees and that sort of thing. Sure. Plus the services, you know, supposed to be a bit of a social network for the working world, right? Like this is where you're networking and finding people and making recommendations and relying on each other for good information. So they're no different than anything else. Anything of this scale, I would assume at this point, I think I used to be kind of naive to this. But I assume at this point anything with any sort of front facing, free account creation type service, whatever it may be aimed at, whether it's just Twitter or something like LinkedIn is probably just infested with bots at this point. And I suspect that there are hundreds of thousands, perhaps maybe more millions of LinkedIn accounts that are purely just bought accounts and they just haven't discovered them yet or figured out that they are bots yet because that stuff just gets smarter and smarter and better and better. I finally, I feel like this is more of a personal anecdote, but I feel like we finally got into a place where I'm starting to be kind of okay with it, because I'm accepting it for what it is. Like, I just don't think it's possible to have a successful service that isn't a large number of people who are who are bought accounts. And I'm not even blaming the companies. I'm not saying all I'm saying they're getting better at fooling people better at fooling algorithms certainly better than fooling or better at feeling, you know, fooling the people that are supposed to be watching this stuff. And they also get better at it's like viruses or whatever same idea except bots are just so proliferated they're everywhere. So, you know, Well, do you think you feel okay with it because you're able to discern what's real and what isn't because I think that's a that's a big. That's that that's a that's a that's a side of the fence or that you know if you feel like hey I don't know what's real or not this is horrible this is bad for my experience because I can't trust anything on on a platform. That's different. Or I agree I mean in this particular case. That's weird because my relationship with LinkedIn is kind of lame. Like I barely use it even when I was working in sort of a corporate world. I had to count but I never really did anything with it. So never really found much use for me but I get fake messages constantly from LinkedIn. I get a message all the time from some account that's got I am absolutely 100% sure is not real. And so I don't know if that's just I'm my experience is indicative of what's happening lately and that's why they've had to make some of these moves to remove so many accounts. But it just is a reminder I think to everybody out there that it doesn't matter what service you're on right now just know that have your have your own base level of education about the sort of thing know what signs to look for. And then it'll be easier for you to sleep at night because most of the stuff is fake it turns out. I had to tell a friend of mine to take down an Instagram post that was one of those I hear by Forbid Instagram from using my likeness in the future Channel 13 told me to you know I was like please this is like the Internet's longest running hoax take that down. Anyway, yeah LinkedIn probably happening on LinkedIn as well. Moving on to Amazon the company launched a new campus in Hyderabad India, which is now its largest campus in the world. The new digs are spread out over 9.5 acres and will house more than 15,000 employees. The new Hyderabad campus will host employees across divisions including Amazon Web Services and others. Amazon already has 62,000 employees in India and a third of those are already based in Hyderabad which is down in the south of the country big country. Amazon founder and really rich person Jeff Bezos has committed an investment of over $5 billion for India operations going forward. Big market for Amazon, a lot of competition for Amazon in India at this point. But yeah, this is a big commitment. Amazon is saying we are here. We will hire many of you in a place that is already known to have a lot of Amazon employees. And it sounds like, yeah, Washington State, you've got competition. Also Hyderabad campus is just a really rad thing to say out loud. I like how it sounds. There's a lot of people 15,000 employees. They already have 62,000 employees. Like that's talking about an investment in a place that isn't here. And I know a lot of people would say, well, why don't you expand in the States? I think there's still this overriding reality that right now it's just a lot less expensive to do the kinds of work that they need to do over there overseas. So if you're going to do support work, if you're going to do certain kinds of fulfillment work like Roger, the place to do it is India and parts of China and the east. That's where we go. I think it also cements the fact that Amazon considers India one of its, if not currently a tentpole market, a future tentpole market where they need to have a presence physically in the country in order to be a successful player. And this, I mean, partly it might be, you know, off-shoring work, but I think a lot of it is also they want to develop Amazon as a key growth area for their business. I mean, China is operating a lot of successful Chinese businesses already operating in that market and Amazon wants to be able to compete. And you can't really do that unless you have a good presence in that country. Yeah, like all of them. Like I would argue, I mean, you see the same kind of movement from Google and Apple and Samsung and everybody else, everybody knows that it's already a big market and a huge potential market. Like there's parts of India that nobody's gotten to in any major way. And so it's the next big place to really kill it. So why not Amazon? Why not Bezos? Why not 60,000 employees? Why not? I say Bloomberg reports that YouTube is finalizing plans to stop targeted advertising on videos aimed at children. According to sources, the US Federal Trade Commission recently reached an undisclosed settlement with YouTube over an investigation into violations of the Children's Online Privacy Act. Although it's unknown if these changes are part of that settlement. But anyway, going forward, videos would be used or sorry, videos would use contextual ads based on the content of the video rather than based on a digital identifier of the viewer. It is unclear how YouTube will define if a video is directed at children sort of quote unquote there. This is super interesting because my kids are all getting older now. My youngest is 19. But they, you know, my kids were all junior high high school when YouTube started to be this huge thing and everything. But it really isn't until recent years that you started to see content directed at children. And by that, I mean as young as toddlers who can't talk yet. My niece was over. She's barely learning how to talk herself. She's got a tablet. She's sitting on the couch. I went and sat by her and said, what are we watching? And I looked at it and it was this succession of Japanese people and Japanese subtitles and no English running around throwing pies at each other. And it was just pie after pie in everybody's face. Everybody's getting hit with a pie and laughing and smiling and pulling the pie out of their eyes. And then it would end and then she would start it up again and watch that pie commercial or that pie video all over again. So children's programming today, even by like teletubby standards is pretty weird. And is that even, I mean, I guess it's something that a kid would probably laugh at him and want to watch over and over. But it's not really children's programming either. No. And that's like what kind of ad gets served up after, you know, around that video based on the content of the video. Yeah. No matter who's watching the video of the pies. Exactly. In TV, we have rules. If I'm watching Super Friends, the rules kind of narrowed it down. We were talking pre-show about this, but narrowed it down to like a lot of serial commercials and toy commercials. That's kind of what I was fed as a kid. And that's TV for you in YouTube's case or in the case of online video. It's a much weirder space of sort of Wild West and nobody really knows what is defined as children's programming. It can be anything from this pie throwing stuff to weird cartoons where they just hum the whole time. I've seen that. There's some just weird stuff going on out there that is not like Sesame Street and the things we're used to. I'm willing to say I may not understand the latest and greatest and what kids really glom onto and want to see. But now we got to rely on computers to sit around and go, well, I think these are the ads we should serve these kids that really like Japanese pie throwing contests. It's a very odd thing to even see them try to address the talk. It is. It seems like a better idea than saying, well, who's watching this video? Who cares whatever reason they're watching this video? But let's target an ad based on the human attached to those eyeballs that are watching this video. At least the content of the video seems to be going in the right direction. But yeah, without a lot of explanation, it is unclear to me if this is going to turn into a good experience for a lot of kids. On Monday, Gizmodo posted a story detailing an email exchange between a Gizmodo writer and the UN police department in New Jersey showing a quote from that town's chief of police that Amazon had edited by removing the sentence quote, security cameras have been proven to be essential in deterring crime and surveillance systems have assisted in closing cases. That was taken out of a statement that was then released to local news outlets. Now rings contracts with police do give them final say over statements that police can issue regarding the product. So they didn't do anything wrong. But you in PD announced that it was adopting rings neighbor neighbors app back in March allowing police to request footage captured from doorbell cameras. We talked about that when it happened back in March and the fact that it was making some folks uncomfortable. And before we kind of kick this around, Tom had a really good interview with SMR podcast Chris Ashley gave us a very interesting consumer perspective and somebody who uses ring and had a lot of really great thoughts daily tech news show dot com definitely check out that interview because it does reference a lot of a lot of probably echoes what some of our thoughts are here. But Scott Roger Amos Amos I know you've you've recently put put some ring products into your home. Is this this is unsettling. Well, personally, I to me it's an opt in opt out thing I should be able to opt in or opt out on whether or not it's actually used by local law enforcement. I would personally opt in because I enjoy a safe neighborhood and I don't have anything here to hide. That's probably a controversial point of view, but it should still be something that as the user, I can allow or disallow police to use my footage because it is my product. I think what's really unsettling is that Amazon or ring gets final say on any kind of statements or publicity regarding the ring unit from police statements right. So they removed they wanted them to remove the word surveillance and security cameras because that would send I guess ring with felt that it would send the wrong message to what their product was. And it's, I don't know, it seems a little off like you know the police are just referring to it the way they see it and the way it's being used. Well, I look at the whole edit security cameras have been proven to be essential in deterring crime. And surveillance systems have often assisted in closing cases but the first part security cameras have been proven to this essential in deterring crime. I mean ring from a legality standpoint and whatever fleet of lawyers they had look at this may just not have liked that line and said well nothing proven where what proof. Yeah, but it's but it's a police statement issued by the police it's not what Amazon is or ring is pushing out. That's my point they edited ring said take that out because we don't agree with it. I don't know it feels a little bit like OCP from Robocop where, you know, they run the police department and they get to issue out all the statements they get a vet all the time. And I'm not to say that, you know, they were wrong or anything but it is it seems a little unsettled unsettling since it's a contract to use their service, and you can only reference. Now and Amazon's, you know, argument is hey this just you know when you say things like security and surveillance, we don't want people to think that they need to use a certain product and that's not what we're saying so we're trying to reduce confusion in the public eye. And there may be some truth to that, but it does come down as so Amazon gets to tell the police what to say. Yeah, it that's the weird juxtaposition here. It does feel funny, but I think the stuff that they removed is more about a marketing thing and them saying well that a that's not proven so don't say it's proven. And also be that's not really the focus of the product we're trying to sell this as a different thing. Like, I don't know if they. It seems like they want it both ways they want to be real picky about what the police say but at the same time they're the ones that kind of wanted to get into bed with this whole thing. So, so I don't feel too bad for them getting the scrutiny, but also we're in weird times this is like folding at home for security in kind of a weird way, you know where we're all contributing a little if we want if we opt in like Amos says we're all contributing a bit or two to the cause potentially and it's very new thing it's weird. It's not really at home, but for people who might break into your house right felons at home feloning at home wait. I don't know what it was an acronym for this somewhere, we'll find it eventually. Well researchers at Harvard john a pulse and school of engineering and applied scientists that's the SES have created a new lens that outperforms the human outperforms rather the human eye by focusing in real time thanks to an elastometer This is important to note that the human eye is right now anyway the fastest focusing device on the planet happens to be a biological one and jammed on our heads but it nobody's been able to do better than that. Until now says here but also adjusting for astigmatism and image shift both of which lead to blurry vision for humans. Only one centimeter in diameter metal lens is flat rather than spherical. That is new using tiny nanostructures to focus light by focusing the entire visible light spectrum in one single point a metal lens is much less bulky than a traditional lens which uses multiple elements to get the same result. Bottom line. This thing will perform better than your human eye although it's not meant to be an eye replacement. Let's not get this confused. We're probably going to go into you know some cool VR headsets right. Point I hadn't even thought of that like VR benefits from VR is number one area of growth comes from speed, not just processing speed and rendering speed but how quickly the image is rendered how how much the eye is is is convinced that what that is just as real as if it was looking at something with its with its natural vision. So sure yeah absolutely this could feed into that but also eventually the sort of stuff makes it into our phones it makes it into our SLR cameras. It'll be expensive at first it'll come down and then pretty soon will be commonplace. But the idea that we're finally getting to a place where optics are outperforming at least in a speed in a speedy way the human eye is pretty intense. Yeah, and even I mean even if you really took the technology which again is it's some years away still, but thought of the next gen of bifocals that are really thin, you know, and you take a lot of bulk off of something that you might be wearing on your face that works a lot better than. I mean, we have variety of prescription glasses that do the trick, but there's, you know, you have to you have to give up some stuff you got to wear something it's got to look a certain way. And this opens up a lot of avenues to the future of better eyesight. Better, smaller, faster. Better, smaller, faster. Yes, thank you, meta lens. Thank you in advance anyway. To get all the tech headlines each day in about five minutes you can subscribe or right now or in a few to daily tech headlines.com. Got a couple of mail bags today. Let's start with one from Sean. Sean has a little tip for anybody who might have the same product he does. He says, for a while now I've been contemplating the inevitable demise of the Harman Kardon Invoke Smart Speaker with Cortana's voice assistant I purchased it last year. Then I had an epiphany, and he actually shared a tweet which then he shared with us. Tweet says, don't kick your Harman Kardon Invoke Smart Speaker with Cortana down the road just yet. Give it new life by simply connecting an Amazon Echo input device to your Cortana Invoke Speaker using Bluetooth. Now you can either use Cortana or Amazon's assistant voice commands and experience the great sound quality of the Invoke Speaker with both. That's awesome. I didn't know any actual, this shows how little I follow the Cortana stuff, but I didn't know actual Cortana functioning assisted devices went out. I didn't know that too many speakers did or anything else. There were some, yeah, it's like a couple CES's ago. And Harman Kardon, I've got, well, no, I have a different car now. I used to have Harman Kardon speakers in my old car. They're nice. I mean, good speakers. Yeah, they're good brand. Nothing wrong with these guys. I just, I don't know, I always felt bad for Cortana feeling like it was the one that was going to get left out the most. I felt like Alexa, sorry, Echo, I almost said the A1. Amazon's assistant. Yes, the Amazon assistant. And that combined with like Google's assistant and maybe Siri a little bit, but obviously a little more Apple control. But I always felt like those were going to be the ones that got out there more. And sure enough, they kind of are. They're the third party integrations that you see. You see cars with Apple Play. You see home thermostat systems with Echo capability, that sort of thing. And the Cortana stuff, I just never seem to hear about anything. Well, Microsoft has shifted, the winds have shifted a bit, putting a lot more kind of built-in Cortana stuff into the Windows experience. So it hasn't died. It has merely pivoted. Nice. Yeah, new life. But yeah, thanks, Sean. Also, Kimberly, the Texas teacher weighed in on our story. Unsurprisingly, Kimberly, I knew you would. Yesterday we talked about AI exam grading and the fact that sometimes humans aren't involved. Is a robot as good as a human teacher when you're talking about grading stuff that could be really important for students in schools in the future. Kimberly said, you seem to assume that people marking the tests were even teachers. In 2013, a report found an ad from Pearson on Craigslist recruiting anybody to mark fourth and seventh grade level state writing tests. And the high school English, one and English to end of course exam. She links us to a Washington Post article about it. She says this was corrected according to the state, so they're not doing anymore, but they still don't tell us who's grading the essays. Kimberly thinks it's probably not trained teachers. She says, unless your state explicitly says the teachers are grading writing samples, it might be a consultant making minimum wage hired off the streets. Especially essay stuff. No, I guarantee you my teachers didn't sit down and read my essays. Actually, part of the reason I know why, this is terrible to admit, but I used to insert a bunch of fakery in the middle of my essays. Mostly to see if I could get away with it. There were still long well thought out essays, but in the middle, I'd say things like, then the incredible Hulk ate an entire car. How is this possible? Like stupid stuff that it would be obvious if you read it, that I am dorking around and they missed it every time. I'd still get A's on these things. Wow. I didn't notice. So I was convinced in high school that no teacher was actually grading anything, that it was all a ruse and I'm still not totally unconvinced that that's not true. Man, I wish I had some of your teachers. My teachers were red inking every little thing, you know, I was always good at spelling, but, you know, they didn't like my sentence structure. Everything was always too long using too many commas. I think it just depends on where you're at, what you're dealing with district wise or whatever. But yeah, like I'm with this email that they're probably not teachers as much as you think they are. There's some, I'm sure, but they're, they aren't sitting up all night reading 150 essays. Well, and I think, I think Kimberly was also moved to write in because Patrick, who was Patrick Beja, who was, who was our co-host yesterday was like, this is crazy, you know, our schools really just like for, for going human grading of exams that are written exams and, you know, require thoughtful analysis. And Kimberly saying, well, you know, that's not really necessarily happening either, at least in certain areas. So yeah, depends on humans have bias too. That all be a lesson to us. All right. Scott Johnson, this was a fun show. Thanks for being with us on this fine hump day of ours. Let folks know where they can keep up with the rest of your work. Well, they want to follow me around and see the kind of stuff I'm pumping out. They can check me out over at frogpants.com. In particular, I've been putting up a bunch of new artwork. You can find that over on the comics and art tab on that site over at frogpants.com. It's always fun to dig through there and look at that stuff. So check that out if you are interested. There's associated prints and things in the store. If you want them, that's at frogpants.com. All the shows are there. You can also find me on Twitter. I'm over there at Scott Johnson. We mentioned Tom's interview with Chris Ashley regarding the ring that you all should check out. You know what? If you're a patron, you would have gotten that interview earlier than everybody else. Patrons get all sorts of perks, not only stuff in advance but cool behind the scenes, special episodes from us, newsletters, all sorts of information on how we do the show. Sometimes we look at the tech of Christmas past, not really Christmas, but tech from a few years ago and how things have changed. It's a good time to be a DTNS patron. So head on over to patreon.com. DTNS and join our little clan. Our email address is feedback at dailytechnewshow.com. We are live Monday through Friday. Join us if you can. It's 4.30 p.m. Eastern. That's 2030 UTC. And find out more at dailytechnewshow.com. We're back tomorrow with Justin Rubber Young. Talk to you then. This show is part of the Frog Pants Network. Get more shows like this at frogpants.com. Diamond Club hopes you have enjoyed this program.