 Coming up on DTNS, Facebook's helping make open street maps better. Ford demos its all-electric truck and why cities are banning facial recognition. This is the Daily Tech News for Tuesday, July 23, 2019 in Los Angeles. I'm Tom Merritt. And from Studio Feline, I'm Sarah Lane. And from Studio Threat Wire, I'm Shannon Morse. And I'm the show's producer, Roger Chang. We were just having a lovely time speculating on weird town names and former session movements in the United States on good day internet. If that sounds interesting to you, become a member and subscribe at patreon.com slash DTNS. Let's start this show with a few tech things you should know. Sources tell 9 to 5 Mac that Apple will release three new iPhone 11. That's what they probably be called models this fall, all featuring the upcoming A13 chip and keeping a lightning port rather than switching to a rumored USB-C port. The three new models would reportedly replace the iPhone XS, XS Max, and XR. The two XS replacements may feature a 3X OLED retina display with the XR successor keeping a liquid retina display, which it has now. A new Taptic engine is said to replace 3D Touch and allow new features that have been introduced in iOS 13. A Korean hydrogen fluoride factory from SoulBrain will begin operations in September, hoping to replace imports from Japan, which is curbing exports to Korea of some chemicals needed in the chip making process. South Korea sourced about 44% of its hydrogen fluoride from Japan in the first five months of this year alone. The gas is used in chip etching. And a mobile app called GAPO, which lets users create profiles and share posts to a news feed has launched in Vietnam following government calls last September to create alternatives to Facebook and Google. GAPO is partnering with Sony Music and plans to reach 3 million users in 2019 and 20 million by January 2021. That would be better than previous Vietnamese social platforms Vietnam Ta and Ha Ha Lolo, which have not found large user bases yet. Still time. Let's talk a little bit more about what's going on with Apple buying something from Intel, All right, sources tell the Wall Street Journal that Apple is within weeks of closing a deal to buy Intel's cellular modem business for around $1 billion. The purchase would give Apple access to a variety of patents and skilled workers it would need to eventually design its own modem chips for the iPhone. Apple and Qualcomm recently settled a legal dispute and at the time Intel said it would cancel planned 5G modem chips and reevaluate the business. Intel got into the wireless modem business when it purchased Infineon in 2011. Yeah, this makes sense. Apple's got the cash. Apple loves to be able to own parts of its process so that it can control them and combine them and make them more efficient. Intel doesn't want to be in this business anymore. And Apple wants some leverage against Qualcomm. So I see no reason not to believe this Wall Street Journal's got good track record with sources like this. I imagine if this deal can happen we'll hear about it in a couple weeks. I wonder who would be another suitor besides Apple because it doesn't sound like a company besides Apple is interested in Intel's business. And Apple could obviously build something like this from scratch, but it would take more work. It sounds like being able to take a patent portfolio and a workforce from Intel that's already in place makes a lot more sense. Yeah, the patent portfolio particularly in the battle with Qualcomm is important, the workers, the talent important in design. I imagine there's probably not many other bidders at the billion dollar level. I'm sure Intel could find somebody to buy that stack of patents at some point, but maybe not quite for that amount of money. It's a good point. Facebook's Messenger Kids app let users add contacts in group chats even if other members of the group were not approved to talk to them, which goes against what Facebook Messenger Kids app is supposed to do. Messenger Kids is only supposed to let users chat with approved contacts that's been approved by parents or legal guardians. A notice sent to parents by Facebook and was obtained by The Verge and published explains that the company turned off a group chat feature where this was the case and directs questions to its help center and Messenger Kids parental control. So it sounds like the one-on-one chats worked as advertised. The problem was that if two kids got into a group chat and there were other kids in the group chat, then technically you didn't get the parental controls for that fourth kid who might not have been vetted yet. So it sounds like this was basically an oversight on Facebook's part. Yeah, and it doesn't sound like there was a whole lot of damage done. Facebook caught it. That's a good thing. You want companies to catch these problems, not outsiders or malicious actors. And it wasn't that bad. It meant that a kid could be invited into a chat with a person that they were approved to talk to, but other people in that chat might not have been approved to talk to. It doesn't mean they would have been horrible people or some predatory ring or anything. In fact, that's less likely because the one person inviting you in had to be approved to bring that in. And that kid who is inviting people into chat could only invite people into chat that they had been approved to talk to. All of that said, this is, I think Roger called it in our pre-show, meaning sloppy. Facebook should have realized that this was going on earlier and caught it. Thankfully, they did catch it and they fixed it now. Yeah, the fact that there's a standalone app just for this purpose to keep kids safe and has restrictions in place in order to make sure that kids aren't chatting with the wrong kids, you know, something as simple as some kid is bullying another kid and they would never have a one-on-one chat that would never be approved, but somehow it gets them into a group chat. I can see where a parent would say, well, screw you, Facebook. This is, you know, I don't want to use your product anymore. This completely goes against what you advertised to me. Yeah. And to be clear, it doesn't sound like that has happened yet. Or I bet we'll hear about it shortly if it did. But yeah, I mean, this is, in the world of outrage at Facebook, certainly you have a reason to be outraged that they didn't catch this. But if you step back from that, I think this is Facebook locking down something before it could be misused because they noticed that they had missed something. And, you know, I always hesitate to punish a company for catching its own mistakes. Yeah. Well, Ford plans to sell a hybrid F-150 truck next year and has announced in 2017, it will follow that at some point with an all-electric F-150. And Tuesday, Ford showed off the first prototype that had shown in public of an all-electric F-150 by having it pull 10 loaded double-decker rail cars weighing 1.25 million pounds for a thousand feet. Now, that's like the traditional like, look what our truck can do because it's, you know, rails and stuff, but it looks impressive. And, you know, 1.25 million pounds, even on rails. That shows that this all-electric F-150 can do what other trucks at size can do, even the ones with gas engines. The Verge notes that a Tesla Model X towed a 287,000-pound 787-9 Dreamliner nearly a thousand feet on a taxiway at Melbourne Airport in Australia last year. So Tesla's still winning the who could pull the heaviest thing the farthest of electric cars right now. But I'm overly excited about the idea of an all-electric truck because I feel like that's a milestone that means, okay, now we've hit the mainstream if an F-150 that's all-electric is just as good as a gas-powered one. I always laugh whenever I read these stories because I'm just like, who's actually pulling, you know, Boeing 787 Dreamliners or double-decker rail cars? I know it's just like them showing how powerful these are, but it's always funny because I don't think there's any kind of, or maybe there is some kind of standard of measurement to see how powerful they are. I am curious more about, you know, how far they go and how this increases the abilities of somebody that would purchase one of these, one of these trucks because it is really cool. And I know a lot of people, especially like some of my friends from the Midwest would love something like this. Yeah, I took a road trip last week and, you know, you get out of the city and you start seeing a lot of big trucks and a lot of those trucks are carrying a lot of stuff. And that would be my first question as well, Shannon, is okay, well, if you like the all-electric factor of it, it's got the power, it's got the power that you need, does it have the distance that you need? Because once you get into rural areas, most things aren't right around the corner. And Roger, you were telling me that one of the advantages of the electric engine is the torque. Yeah, I mean, with any electric motor, you get a flat torque curve. So, you know, when you wind up a gas car or even a diesel, you slowly build up where you get the sweet spot where the RPMs give the most amount of power. On the get-go, electric motor gives you everything off the line. That's why locomotives are all diesel-electric, right? They have diesel engines that power electric traction motors. And that's a huge advantage, especially if you're towing uphill because now you don't need to stand on it to get any distance. The issue with distance, basically, how far you can get on the charge is still relevant for a lot of people who buy F-150s as more of a lifestyle vehicle and not necessarily, I have a job that requires me to have a truck so I can load my tools and stuff in the bed. But if they... If you're an urban truck cowboy. Yeah. Well, you know, and one of the, you know, one of the more interesting clients for these would be fleet sales to companies that need to pick a truck, but they definitely, you know, they're not averaging over 100 miles or 200 miles in a day because they drive in-city or they drive in a specified... But we don't know what the range of these things is. I mean, the range of electric vehicles is getting in the 200 to 300 mile or more. I'm not saying that an F-150 would be able to meet that, but it certainly should be more than 100 and probably around 200. Yeah. Well, you know, and the 100 is like a safety, so you can, you know, go around all day, whether it's a delivery vehicle or if it's something that's carrying, you know, whatever, like a plumber or an electrician. Yeah, I guess what I'm saying is I know a lot of people are going to jump to the conclusion, yeah, I won't have the range because, you know, 15 years ago, electric cars, that was the ding, but range is becoming less and less of an issue. It's more about can you find the charge when you need it? Yeah. Do you have the time to recharge? And pickup trucks definitely have the space to add more batteries, unlike a passenger car. Yeah, right. Anyway, we'll find out more of those details from Ford when they get around to doing more than showing us pulling 1.25 million pounds of a gas-powered F-150s on a drinker. Well, moving on, Facebook announced it's opening its map with AI tool to the OpenStreetMap community to help identify rural areas in Afghanistan, Bangladesh, Indonesia, Mexico, Nigeria, Tanzania, and Uganda with plans to expand to other areas and eventually the whole planet. Mass with AI spots patterns in satellite imagery to identify possible roads that may not be paved or named and highlights them in OSM's platform. Facebook built a neural network model to recognize roads on satellite images with a resolution of two square feet per pixel. Then OSM volunteers provide final verification. Facebook also announced its RAPID editor to auto detect roads from satellite imagery and include additional data integrity checks will improve OSM's current ID editor. Facebook tested its tools in Thailand and Indonesia in an 18 months mapped all of Thailand and 90% of Indonesia. So cool. Yeah, I like the idea that Facebook has been supporting OpenStreetMap. I don't know if a lot of people realize that. It's not the only way OpenStreetMap is supported. But if you wondered like who uses it, well, Facebook does. They power a lot of their location-based mapping with OpenStreetMap and are a big supporter of it. This is a great way to do something altruistic that is in Facebook's benefit. One of their ways of expanding and hedging against future growth is to be able to provide the conditions for people in developing areas and rural areas to want to use its services. Having the most accurate maps and areas that are previously unmapped is a great way to do that. On the other hand, that's good for everybody because they're doing it with OpenStreetMap. This isn't proprietary Facebook map that nobody else can do. Everybody can take advantage of OpenStreetMap. So it is benefiting the public. And I hope they add more of these kinds of countries and even do get to the point of helping to map the whole planet with the most accurate maps. Wouldn't it be amazing if at some point Apple or Google just give up and switch to OpenStreetMaps because their maps are more accurate and easier to maintain than keeping it in-house and keeping it under lock and key? Well, there's also a community in place already with OpenStreetMaps. So satellite imagery that Facebook has access to, there are cases in extremely rural areas where it's like, is that a dry riverbed? Is that a road? Are people using this to get from one place to another? It's somewhat unclear and the community from OpenStreetMaps can go in and help edit that and flesh out the information. And I did want to point out the two square feet for pixel thing. I know that it doesn't sound like much, but that's a really, really good resolution, especially for that kind of satellite imagery. So being able to bring in that much information, especially for rural areas, is going to be so helpful not only for the people that live there, but also for travelers because a lot of times you might find yourself going through these areas and the only thing you might have in these rural areas might be an actual paper map, which who uses those anymore? So it'll be really nice to have this available in some kind of digital format. Yeah. And some of these were never even paper mapped. So we're actually getting to the point where digital maps are going to be more accurate than a lot of paper maps, just because it wasn't in any paper publishers' interest to keep these kind of rural roads updated because it was too hard to get to. And the combination of satellite imagery with AI that says pretty sure that's a road, not a riverbed, with someone in Tanzania who can say, oh yeah, no, I actually drove that road the other day. That's definitely a road. I'll confirm that. Seems to be a really good procedure for this. AI researchers from UC Berkeley and the universities of Washington and Chicago have released a database called ImageNetA of 7,500 images that AI systems struggle to identify. Among the examples are butterflies labeled as washing machines, alligators as hummingbirds, and dragonflies identified as bananas. AI, tough stuff. The images were all four online and have not been digitally altered. This is these are these are real mistakes. The hope is making the dataset public and will help other researchers improve image classification algorithms. All the images in the database are online. They haven't been digitally altered. The point being that ImageNet was a database that has been used for years to help train AI. And ImageNetA will be, these are the ones that everybody gets wrong, that all the AIs are getting wrong. So if we can fix AI to be able to recognize these accurately, we'll have made a big advance. And I think that's interesting. It is funny to look at these and go, how does an AI dragonfly is a banana? That's because we have this low watt, really high efficiency computer running in our heads that can do this kind of stuff that is really hard. I do wonder. There's so many, so many variables too. Is it color? Is it overall shape, shadows, all that stuff? Go ahead, Shannon. I was just going to say the same exact thing. Like it could be anything. It could be like, I don't know, pixel density or whatever. Like the color, the density of the thing that they see in front of them, the size could be the only factor. But of course, a butterfly in a washing machine, very different sizes. So I don't think that that's probably not the thing that makes them get confused. And I think this is a really good instruction of what is the difference between AI and us. We have a whole different way of pattern matching that looks for a ton of signals that we are not even aware we're looking for. AI has been trained on a dataset of perfect looking dragonflies. And so when that dragonfly is at an odd angle with a shadow and some sun, it's like, well, this actually matches the points of data of a washing machine. It doesn't think it's a washing machine. It's just like this shadow with this line, with this color. We also do a lot of contextual analysis. For example, the bullfrog photo, it's in the grass. You're not going to assume it's going to be something else. You could probably think it's something alive first as you go through a pattern matching system in your head. Well, but the other thing I was going to add to that too, because the context thing is a really good point, is if every picture of a washing machine had a shadow next to it in the training set, the AI may identify the shadow as the washing machine, right? It may not be identifying the washing machine. It's like, oh, every time I see this square looking shadow, that's a washing machine. We think it's looking at the washing machine. And then when we get out in this other dataset, that dragonfly happens to have a, or I guess the butterfly happens to have a square looking shadow. And the AI goes, oh, look, square shadow. That's a washing machine. I mean, it's stuff like that too. Yeah, we don't have. We know better because of that context. Hey, folks, if you want to get all the tech headlines each day in about five minutes, be sure to subscribe to DailyTechHeadlines.com. Now you think AI is having a hard time keeping butterflies from washing machines. Now let's have it try to recognize people's faces. This is an issue. On July 16th, an ordinance banning municipal use of facial recognition technology passed its first vote by the Oakland, California City Council. A final vote to make the ordinance law is scheduled on September 17th. It's expected to pass that as well. San Francisco previously has banned its police officers from using facial recognition technology. They did that back on May 14th. And a Boston area city, Somerville, Massachusetts banned police and city government from using facial recognition tech back on June 27th. This is definitely a trend. It's certainly not sweeping the nation, but the bigger trend is departments using facial recognition to supplement law enforcement. And in some cases, you can see where this is a really good thing. Hey, tell us if this criminal that we have a search warrant for is in this crowd and alert us so we can go get him. I don't think if that works perfectly, anyone objects to that. The way that I've described it, you want law enforcement to do that. But there's all these other things about what it could be used for and how well it works that start to muddy the issue, right, Shannon? Yeah, exactly. And that's the major thing that I really wanted to discuss with this is how many limitations that facial recognition currently has and how it misidentifies to a much broader scale, especially minorities, which could impact them, especially when it comes to law enforcement. We've seen stories again and again about minorities and law enforcement, and I won't get into the details about that, but you do have some serious concerns when it comes to facial recognition technology and how this might impact especially them. I believe there was a study done a while back that determined that misidentification was far better at identifying men than women and lighter skin over darker skinned faces. However, in June, Microsoft improved 20 times for dark skin tones and nine times for women overall. So we are seeing an improvement. And I think bringing it back to what you said, like, hey, if it was perfect, this would be awesome. I do agree with you, even though I have a lot of privacy concerns with facial recognition technology, I think if it worked perfectly and it did get it right all of the time, then it would be very, very helpful for catching criminals and catching people that are actually doing bad things. But right now, if it's misidentifying you, you could get thrown in jail for no reason, just because you were misidentified by facial recognition. And that could screw up, you know, whatever you might be trying to do that day. So I know that that's a, yeah, just that day, right? Just that day. Well, and even if they don't throw you in jail, which I'm not discounting as a possibility, but harassment, if you are consistently misidentified as close to an actual criminal by a system that is more likely to identify you wrong than other people, that that's going to end up becoming a problem. And that we shouldn't be relying on a system that has that kind of error prone built aspect built in yet. Or if we do, there should be some very strict rules about how this technology is used, kind of like a second factor, like, okay, facial recognition says it might be this person, but don't apprehend them until you have another factor that makes you think it is. I'm glad you mentioned having some kind of rules, maybe regulations, maybe laws should be enforced to kind of regulate what people can do with facial recognition technology, especially because currently, there are no like consent permissions set for facial recognition if law enforcement is using this for, for example, for cameras on the street or anything like that. So they could be picking up information about you without you even knowing and without you giving any kind of consent, just by the fact that you were on that sidewalk at that time. When you compare this to a private company, say like a Vegas casino who currently do use facial recognition technology and some of them, when you enter that private business, you are kind of inherently giving your consent by entering that private business and doing your business there, especially if you're staying at that hotel at a casino, you're probably signing something upon entering that says that you probably will be captured by facial recognition technology. So there's a lot of, I think, regulation that needs to happen, maybe on a government sector to make this viable for law enforcement in the future. Yeah, whether it's government regulation, whether it's standards that say, look, you, if you're implementing facial recognition as a private company, these are the standards that you have to abide by. We don't have that yet. And that needs to happen because companies are using this. Carnegie Mellon did a study using facial recognition to identify people on a dating website where real names weren't used. So that's not even out in public. That's saying, hey, we got a database of people's names attached to images. We have this website where people have pictures. Yes, conceivably, I could do that on my own, but I'd have to look through all of the Facebook posts and compare it to the dating apps. Whereas this computer program can just do it like that. I'm not saying that technology should be illegal, but we should be aware that facial recognition can do that before we start talking about when and where it should be applied. That same study also identified people walking on campus by looking at public Facebook profiles. And a lot of the concerns around facial recognition aren't about what you do with the facial recognition. It's about being able to attach data sets to say, ah, okay, that's the person on Facebook with this name. And I have all of this data on that Facebook profile from an ad agency. And now I see them walking across campus so I can mark it to them exactly where they are, which is kind of almost a minority report situation potentially. Do we want that? Do we want to allow that right now? We don't get a choice. And as we have seen, not having a choice in how you're marketed to and whether your data is collected has become something people are not okay with these days. Excellent, excellent point. The fact that we don't know anything about how the facial recognition technology might be stored, who's to say that it's stored in a factor that is very, very secure? What if it was mishandled or misused? What if there's a vulnerability for whichever company is using the facial recognition technology? That information could be stored in a very bad way that would allow somebody to steal it and then create a whole subset of factors that could identify you for anything, for stocking and harassment like we had mentioned previously, like for identity theft. There's so many different potentials of misuse when it comes to facial recognition that it really needs to be taken seriously. So I'm definitely in the camp of I agree with the bands. I don't think that we need to be implementing facial recognition technology publicly in cities yet, especially for law enforcement. I just don't think it's there yet. And I think that we really need to study the technology and really get it down and maybe enforce some kind of regulations. Yeah. And there are good uses of facial recognition. One of the potential uses that I've seen to write it is identifying emotions like distress, because the assumption is, well, then they'll try to market, they'll try to exploit you because they're like, oh, they're sad. Let's market them with things to cheer them up. That kind of thing could also be used to diagnose you. It could be used to improve your life. It could be used to help you. So I hesitate to say facial recognition should be banned. I do think that it needs to be used very careful and in very limited ways. City of Oakland wants to ban it because they don't think that their setup can use it properly fine. But there's a nearby county in the Bay area that's using it and Oakland police cooperate with that county. And so even if it can't be used in Alameda County or in the city of Oakland, it could be used by Oakland police in conjunction with that other police department. So it's out there. I think the sensible way to approach it is let's take control of it. Let's study it, like you said, Shannon, which means using it, but using it in a way that isn't used maybe even for apprehensions, for arrests, but as a test to be like, how good is it yet? What can it be relied on? Let's create some rules of the road in these different departments that say, okay, facial recognition is very much in its early infancy. It should only be used in test situations and should never be involved in a case. I think the worry people have when I say something like that is, yeah, that sounds great, Tom, but we know that all these departments overstep the boundaries if you give them a tool and that is a potential problem. You're right. Well, thanks to everybody who participates in our subreddit, Facial Recognition Technology is definitely a story that shows up there often. You can submit your own stories and vote on others at DailyTechNewShow.Reddit.com. Join our Facebook group if you haven't already, Facebook.com slash groups slash DailyTechNewShow. Let's check out the mailbag. So we had a conversation yesterday about the idea of DoorDash and other services, not necessarily giving full tips to the courier if you tip within the app and think that that's where it's going. Bernard wrote in and said, I only recently started using DoorDash on a regular basis. I've never tipped because of the previous March discussion he's talking about hours. If the delivery person gets paid the same amount, where is the benefit in me tipping? Tom had actually written back and forth email and said, well, you could tip in cash. That could be a workaround. Bernard said, well, part of the problem is that I don't carry cash. I don't disagree with Tom's premise though. DoorDash is already shady with the practice. It really isn't a tip in my eyes if the driver isn't getting a bonus. Yeah. And to me, that makes me say, and I actually responded to Bernard with this too, is like, but then I probably wouldn't use that company. It may be the only company that delivers the thing you want, in which case, well, okay, I understand if you really need that particular brand of food, it's your favorite, and you don't carry cash, then you might make that decision. But it would make me not want to support that company if they're doing it that way. And honestly, I don't use DoorDash not because of this particularly, but because they haven't really been very good with my deliveries. Whereas Postmates, 70%, they're okay most of the time. Uber Eats seems really reliable. So those are the ones I tend to go to. That's why I don't use them either, just because of reliability. I've got my four restaurants down. You know, sweet green ramen, that's all Sarah eats. I mean, I eat other things, but that's all I really get delivered these days. You talk about ramen a lot, yeah. Yeah, it's a thing. I mean, it's just feel good food. You know what else is feel good having Shannon Morse on the show. Shannon, thanks for being with us today and let folks know where they can keep up with your work. Hey, thank you so much. And I apologize for the construction noise if you heard it in the background. You can find me, I'm doing Threat Wire, which is over at youtube.com slash hack five. That show is currently going really strong. I'm really excited about it. I do have a special offer available on Patreon for patrons. I will send you a personalized video. Thank you note directly to you if you sign up right now. So if you sign up, super excited about those videos. My most recent episode was all about Google's new incognito mode and how they're closing up a loophole that was in there that allowed people to identify if you're using incognito mode. So that's pretty cool. And I also covered all that face app debacle. It's not as serious as everybody thinks. So check out the newest video if you're interested in hearing all about it. And a big thanks to Darren Polvoy, who just became a patron of DTNS today and joins John Willie, who's been a patron for five years. We take you at any point in your life cycle to support the show at patreon.com slash DTNS. If you've got feedback for us, well, guess what? We have an email address. Feedback at dailytechnewshow.com is where to point those emails. We're waiting. We're also live Monday through Friday at 4 30 p.m. Eastern. That's 2030 UTC and you can find out more at daily technewshow.com slash live. Back tomorrow with Scott Johnson. Talk to you then.