 Daily Tech News Show is made possible by its listeners, thanks to all of you, including Paley Glendale, Dr. X-17, and Dustin Campbell. Coming up on DTNS, Nika Monford talks to us about some things it takes to make an autonomous vehicle plus how CEO and inventor Jessica Matthews powers the world and why AI and radiologists together are better at identifying cancer than either one is alone. This is the Daily Tech News for Monday, July 11th, 2022 in Los Angeles. I'm Tom Merritt. And from Studio Redwood, I'm Sarah Lane. And Nika Monford, live from Atlanta, Georgia. And I'm the show's producer, Roger Chen. We have got some excellent stuff on the docket today as you just heard, so let's start with a few tech things you should know. The Guardian obtained a trove of over 124,000 internal documents from Uber during the tenure of CEO Travis Kalanek dating from 2013 through 2017. These detail lobbying efforts, growth strategies, and include texts and emails from executives. The Guardian shared access to these documents through the International Consortium of Investigative Journalists with media outlets like LeMond, The Washington Post, and the BBC publishing further investigative pieces. Uber said it made mistakes and missteps, but the company has been transformed since 2017 under its current CEO, Dara Khazver Shahi. Uber's former head of European lobbying, Mark McGann, came forward as the whistleblower, saying, quote, it is my duty to speak up and help governments and parliamentarians right some fundamental wrongs. Morally, I had no choice in the matter. Okay. FOSS patents reports that Erickson won an injunction against Apple in Columbia on a patent dispute and has begun trying to enforce a sales and import ban on 5G iPhones and iPads in Columbia. The current patent dispute isn't about whether Erickson's 5G patents are valid. It's about whether they are charging a reasonable royalty for them. In response, Apple brought an emergency motion to a court in the Eastern District of Texas asking Erickson to, quote, indemnify Apple from any fines, fees, penalties, and costs that it incurs as a result of the Colombian injunction. In a letter to Twitter's legal counsel disclosed in a Securities and Exchange Commission filing, Elon Musk's attorney, Mike Ringler, said Musk would not proceed with an acquisition of Twitter, citing materially inaccurate representations and that Twitter failed or refused to provide relevant business information regarding spam accounts. Twitter board chair, Brett Taylor, says the company remains committed to closing the deal at the agreed price. Bloomberg sources say that Twitter hired the firm, Wachell, Lipton, Rosen and Katz, whose founder, Michael Lipton, is credited with inventing the poison pill defense against hostile takeovers. You might recall, we talked about the fact that that is what Twitter has been doing as of late. But they tried to anyway, yeah, yeah. Yeah, exactly. Twitter plans to sue Musk as early as this week for backing out of the deal in the Delaware chancellery court. Musk waived due diligence when he agreed to buy Twitter, but is pursuing the idea of material adverse effect, which requires him to prove that Twitter's actual business differs dramatically from what he agreed to buy. I don't think the story is going to end anytime soon. Somebody wants a cheaper price. That's what it feels like to me. Consumer Intelligence Research Partners reports that Amazon had roughly 172 million U.S. Prime members on June 30th, which is about the same as six months earlier. That's no growth, if that's true. This comes after Amazon added 60 million U.S. Prime members in 2020 and in 2021. Back in February, Amazon raised the annual price of Prime from 20 bucks to, or raised it by 20 bucks to 139 bucks. Meta open sourced a tool called Sphere designed to provide a knowledge base for AI and other complex platforms using information from the open web. Sphere was trained on 134 million public web pages. Wikipedia will use Sphere to automatically scan entries to identify how strong a citation is supported. It's unclear if use of Sphere is part of a paid deal between the two organizations. Some people think that is probably the case, but neither company has said such. Yeah. If it works well, seems like it might be a decent use of AI. A good deal. We need more good uses of AI. Yeah, we do. So we talk a lot about the best uses of AI on the show, obviously. If you listen, you know that to be true. Many folks are worried about giving it control. Too much AI control can be a little scary, what's going to happen, especially given that it's only as good as its data and data is often biased. And we've had lots of stories about AI being used as a diagnostic assistant, not making decisions necessarily that humans would make, but helping medical personnel identify them. But as we say often on the show, it's always better to find out if things actually work than just implement policy based on what sounds like it makes sense at the time before knowing exactly how it's going to work. That's why we're pleased to see a study in the medical journal, the land set this month called Combining the Strengths of Radiologists and AI for Breast Cancer Screening, a retrospective analysis. So, Tom, let's talk a little bit about what this report says. Yeah, the authors of the study used software from a German startup called VARA. VARA is in use in around a quarter of Germany's Breast Cancer Screening Centers. It's also in use in Mexico and Greece. They worked with radiologists on the study as well, from Essen University Hospital in Germany and Memorial Sloan Kettering Cancer Center in New York. VARA has been trained on 367,000 mammograms. The radiologist notes and assessments of those mammograms and the eventual patient results. Did they end up having cancer? Did they survive? Were they treated? Etc. VARA learned to classify confident normal, not confident and confident cancer. So they had three categories. One was pretty sure there's cancer there. One was pretty sure there's not. The other was not enough to tell. Yeah. So the study focused on that, but also focused on how well the AI did identifying cancer, but there's a little bit more to it, right? Yeah, yeah. They compared two approaches. In one, VARA alone analyzed the mammograms and they assessed how well it did there. In the other, radiologists reviewed scans without VARA's help. So they reviewed them on their own. And then if the radiologist cleared a scan that VARA had said was concerning, VARA would send an alert to the radiologist to review it a second time. So you got three numbers out of those two approaches. You got how VARA did alone. You got how the radiologist did alone without knowing anything about what VARA would say. And then how well they did once the team was combined. And you should know that this was a study. So nobody's actual like prognosis rested on this. This was to see how well the system worked. The radiologist had better accuracy than VARA alone. So the humans outperformed VARA in this case. But the combined team, when VARA said, take a second look just to be sure, the combined team was 3.6 percent better than the radiologist alone and reduced the number of false alarms where they thought there was something there and turned out there wasn't. In addition, VARA classified 63 percent of scans as normal, where it was like confident that it was normal, which sped up the process of review. Humans still looked at them, but they got pre-filled reports, meaning if the radiologist agreed they could just press enter and move on. So that helped speed things up. That can be especially useful in regions that don't have enough radiologists. India, for example, has one radiologist for every 100,000 people. So you need things that can allow those radiologists to review more scans. The next step here, of course, is to judge how well VARA performs over time in actual clinics, not just in the study. Well, as somebody who is pretty intimately familiar with this sort of thing, breast cancer specifically, and I won't go on and on about it, but I was diagnosed with breast cancer last year, but before I was, I had a mammogram because I had a little bit of a concern, and the mammogram and the human who was the lovely person, they said, you're okay, we think you're good, I wasn't. And that is no fault of anybody's, these things happen. So personally, I'm like, yeah, AI, 100 percent do it. I don't know why this would ever be a bad thing, even if it was a false positive, even if it was that, just to have some sort of machine learning help to say, maybe take another look at this. I don't think my outcome would have been different personally, but I think a lot of other people's might be. And I think from this study, they really got two things right. And one of those things is that it's the notion that AI along with human intervention is the best course of how we think about AI. I think we have this idealized view that AI is here, machine, go out into the world and do all these things for us so that we don't have to do it. I think we need to shift our focus into thinking more along the lines of what this study proved is that machines along with humans makes for the most ideal outcome. Because again, you are still teaching a machine to do human things, but it still does lack that human aspect. The second thing that I thought that was important, that they got right, was the data. It all really starts with the data. If you have a bad data set, if you're manipulating the data to get the outcome that you want, it's not going to give you anything you can actually use in the long term. Looking at the notes from the Lancet website, what really stood out to me is that for the data that they used to train it and validate it, the test data set didn't include any of those people. So it's not one of those things where they can tweak the data to get the outcome that they want. It's more along the lines of, this is what we started with. This is what we're training on. This is what we're validating on. If we just inject this new set of data that's not linked in any way to what we use to validate and to train, let's test it against this new set. And I think those are the two biggest things that they got right in the study. Yeah. I'm excited about this study because I have suspected that AI's best use is to help us because there are so many things that you just need a more intuitive sense of knowledge to understand them, especially if you're a professional like a radiologist that are hard to manipulate, but humans get tired, you know, humans make mistakes. And even the most highly trained, most professional human will make a mistake. And so having a backstop like that seemed to me like the best way to do this. And I love that this study says, yeah, in this particular case, we found that to be true. We still need to do more work, but it points to the direction of don't look at AI as a replacement for humans. Look at AI as a useful tool that in certain situations like this can be helpful. Absolutely. A companion, so to speak. Yeah. Yeah. Well, folks, it's time again for, you know, Nika Bonford's on the show. We're going to do tech and while black, shining the light on a technology leader you might not have heard of, but probably should have Nika, who are we highlighting this episode? So this week we're highlighting someone who comes from a little bit of an untraditional background. We're highlighting Jessica Matthews. She is an inventor, a CEO and a founder. And she is the current founder and CEO of Uncharted Power, which is a global company that specializes in building and operating renewable power infrastructure. And she created this company at the age of 22. Oh man. I think it's probably even more impressive is that her first invention is, it's called Socket and it's an energy generating soccer ball. And she did this at 19. Jessica is both a dual citizen of the United States and of Nigeria. And as we know, soccer is one of the biggest sports, if not the biggest sports in the world. And to be able to use something as, I won't say common, but as easily accessible as a soccer ball to be able to generate energy. And we know in some parts of the world, energy is very hard to come by. It can be very scarce. It can be hard to keep on a consistent basis. And that's one of the things that she said was that her goal was to inspire and to find ways. She says, quote, I asked myself, how do I design solutions that inspire them to try? So it was one of those things where you want to, her goal was to find ways to not only help her native country, which she, again, like I said, has a dual citizenship of, but how can we do this for the whole world? And one of the visions of Uncharted Power is to aid cities in becoming smart cities. We all know climate change, the things that we're experiencing. If we can find ways, which this company seems to be doing, to find a way to have an alternative power source, a renewable power source, that helps all nations across the world. And for this company, she has backers. I think Maddie Johnson recently joined the board of directors. She has backing from large companies like Disney. So it's one of those things where this is a huge step forward in how we power the country, this country and nations all around, in addition to her young age of creating this, her offering, her CSA offering for her company, initially when it was established back in 2016, it broke the record for funding by a black female founder. So to me, that says, hey, this is a very important aspect that we are definitely trying to, trying to fulfill and meet. And she has the backing to continue to do that. I mean, how many times have you looked at a bunch of kids playing something like soccer and think, man, if I could harness that energy, and Jessica Matthews was like, I think I can harness that energy. I can do that. That was sort of my question is like, I mean, is this something where, you know, if kids are playing soccer in the backyard type thing, it could power the house? I think so. That's what it is. That's the idea, right? Yeah. The idea behind it is to be able to generate energy that can be actually used by people playing something that most children play. Are you in anyway? It's just the open move for Jessica Matthews. Right. At 19. Yeah. That's crazy. Yeah. It's pretty wild. I say an untraditional tech background because when we think of scientists and technology, we think of folks in maybe computer science or chemists or environmentalists, those type of individuals, but she has her undergraduate degrees from Harvard in psychology and economics and an MBA from Harvard Business School. So she has a little bit of an untraditional tech background. But when you think of the human aspects of something as trying to find ways to generate renewable power, you need the aspects of how people think, how people act. And also, where's the money coming from? How are we going to get this done? And also, in addition to inventing the socket, which was her first invention, she has over 12 patents and patents pending. So she didn't stop there. She's still going. And there's more to come. Some of the honors and accolades, which are tremendous, I kind of narrowed it down. Harvard Foundation Science of the Year in 2012, Fortune's 10 Most Powerful Women Entrepreneurs. Fortune's one of the most promising women entrepreneurs. She's actually an ambassador for entrepreneurship for the country of Nigeria. So it's one of those things where it's something that's continuing to build and she's continuing to build because she's still relatively young. So it was definitely something I think definitely needs to be highlighted, especially where we are in and what we're seeing with our with our global economies and our global infrastructures. So shouts out definitely to Jessica Matthews. You can find information about the company at Uncharted.city and you can find her over on Twitter at Jessomatt. Yeah, go, go, go say kudos to Jessomatt on Twitter. Yeah, good stuff. Thank you, Nika. And it's part of DTNS's mission to sort of help us understand the world by by meeting different people and and finding out about different people that are doing different things next week is special guest week. New guests we haven't had on the show before, including Jack Risider from Dark Net Diaries, Will Smith from Tested.com, Quinn Nelson from Snazzy Labs, Joel Telling, the 3D printing nerd. Don't miss it. And most importantly, tell your friends, hey, new folks, interesting folks coming on to Daily Tech News Show next week, starting on July 18th, Daily Tech News Show dot com. Well, much like hope, reports about Apple's car project, spring eternal, folks. Recently, we have gotten updates on the project, mostly when there are staffing changes, executive left is the Apple car not happening anymore. And nobody really knows. But the information published a wider ranging report covering the last eight years of the so-called Project Titan. Yes, the report claimed Apple's Senior Vice President of Software Engineering Craig Federighi is skeptical of the project, at least according to those sources. CEO Tim Cook, quote, rarely visits its offices and has not yet committed to producing a vehicle. If you haven't followed the frequent management changes, retired senior hardware engineer Bob Mansfield ran the project from 2016 to 2018, followed by former Tesla SVP Doug Field. He just left in September for Ford. The project is now overseen by Kevin Lynch, who was the person who led Apple Watch development. Yeah. And as far as, you know, capabilities, the Project Titan team showed executives a prototype self-driving vehicle that was able to travel through a known 40 mile route in Montana successfully. Sounds good. However, in real world testing, prototypes have shown lane issues, hitting curbs. In one instance, a human safety driver had to intervene when the car didn't stop for a jogger, but only slightly adjusted its path forward. So, you know, it's not perfect. Apple concluded that the safety driver hadn't intervened. If the safety driver hadn't intervened, the car would likely have struck the person. Not good. Want to get that stuff ironed out before you start selling cars to people? Yeah. And honestly, that kind of thing is not super unusual where the safety driver intervenes. That's why the safety driver is there. But it does indicate that Apple is not as far down the road as some of the others. That's what Apple is reportedly doing. Nika, you're working on these type of systems every day. So can you shed a little light into what goes into navigating routes, so to speak, in this kind of testing? So it is quite difficult. And one of the things is this is a long haul effort. This is something you can kind of just churn out over a couple of years. You have to have a genuine commitment to, you know, getting into this and following it through. As you did say, Tom, I do work in this every day. I work on the features engineering team for General Motors Ultra Cruise, which was announced, I believe, October of last year. And what the system does is it's not fully autonomous, it's semi-autonomous, but it does offer hands-free driving and it does. It is able to navigate, you know, different types of avenues. And one of the things is when you're doing this type of work, you have to make sure you take into consideration the terrain that you're on. Are you going to be traveling on highways? Are you going to be traveling on residential areas? Are you going to be traveling on streets, interstates? How does this work? And you think about interchanges. If you're going around a steep curve to get on and off an exit, those types of things have to be taken into account. And it's not just something that is easily done because you have. When you think about it, if you're if you're having a vehicle drive for your hands free, hands off, it has to know things like, where's the dotted line? Is it a straight line? Can I cross? Can I not cross? Is this am I going on to the the median? You know, those type of things, what's the speed limit? Those different types of features have to all come into play and it has to work seamlessly because you are dealing with humans. And one of the things that I have encountered is you have to be able to make sure that you're looking at just not the basic things being on a straight road. You have to look at different edge cases as well, because, again, as we mentioned earlier with the story regarding breast cancer, you this is still a machine and it still needs human intervention. You can't just be completely hands off. Well, I guess in some vehicles, you can. But particularly on what I work on, it's not intended to be, you know, completely just checked out. It has safety protocols in place to make sure that the driver is still engaged. Should the need arise where you have to kind of take over. But it's one of those things where you have a route. I think the article mentioned that Apple had a 40 mile route. We as well have a specific number mile route that we test on. We've tested on it, you know, in the prototype. And we've tested on it in more real world scenarios. And those things are imperative when you're thinking about human life, when you're thinking about how you engage with other vehicles, how you engage with someone who may be crossing the street. Those things are of the utmost importance. And what this says to me as a relation to the Apple car is that this is not something you kind of just jump in for the fun of it because this novel is something you truly have to plan out. I think it was just announced in October of last year for Ultra Cruise, but it was worked on significantly before and is still a significant amount of work being done before it's it's ready to, you know, be out in the wild. And there are different aspects that you definitely have to take into account because again, human interaction and machines, they need to work cohesively together. Yeah, because I feel like there's a few levels you're describing there. One is like there's a test track, we know exactly what it looks like. We're pretty much, you know, except for maybe a bird or something, we're pretty much can control what's what it's going to encounter. Then there's what you're dealing with when you develop, which is, well, we can handle most situations, but we've always got the human because it's not an fully autonomous system. So that makes it a little easier because as long as you keep the human in the loop, you can fall back to that. But that just shows how much more complicated it is to make a fully autonomous car that can handle every single thing without a human involved, which it sounds like from this story from the information is what Apple is targeting. They want to make a car where you all just sit around and, you know, play UNO while you're headed wherever you're headed. Right. And like GM started with SuperCruise, that was the first foray into this senior autonomous driving. I think it was of like a level two. And so this next iteration of UltraCruise kind of kicks it up a notch because now we are on, you know, all types of roads. And we are expected to be able to cover, I think, over 2 million miles of road. And this includes all roads. This includes interstates. It includes neighborhoods. It includes, you know, side streets. And the other thing you have to take into account as well is where are you sourcing your data from? Are you going to buy Google's map data or are you going to create your own map data that adds in a whole another level to it if you want to use some existing map data? Sure, that helps. But that has to be pretty expensive, I would presume. So if you're talking about generating your own map data that adds a whole another level, how are you going to map the whole country to be able to find every kind of nook and cranny road? So it's one of those things where you it's it's a long haul. And, you know, the fact that Apple announced this so early, it was exciting because, I mean, it sounds cool. But the thing is how how ready are you for this? Did you really think how convoluted this could possibly get? Did they announce it, though? I think that's the thing that I always try to remember is they didn't announce anything. This has always just been like, well, we hear they're doing a thing. So, you know, like Apple's Apple's been spending a lot of time and money on it. But it's been mostly rumors about what they're doing. But don't you think some of this stuff is probably leaked by some people like strategically? Maybe. I don't know. Yeah, I don't know. It's a good point. Yeah. But yeah, so it's a lot to take in into consideration when when doing something or taking something like this on. Yeah, well, Nika, thank you so much for sharing your hands out, like literally hands on information in this. That's super invaluable. Indeed, indeed. Speaking of cars. Have you ever looked at a car and said, oh, you know, it's a car painted black, but it could be blacker. The team over at Pit One Customs in Japan has just the thing for you. I was going to do a Stefan voice for this, but it will just keep it pushing. It painted a Porsche 911 in co-pros Muso Black acrylic paint, which absorbs acclaimed 99.6 percent of inbound light. The result is kind of a shapeless shadow of a Porsche 911 because the paint makes it hard to make out contours of the vehicle. Some people might like that. It also costs about one hundred thirty dollars per liter. So it's not going to be the cheapest paint job out there. But boy, does it look nice. The the the video is will be in our show notes. Now, there aren't many ways to get a darker car to indulge your nihilistic dreams if you have them. Vantablack coatings. Some people might be familiar with that made from randomly aligned carbon nanotubes can absorb up to ninety nine point nine six five percent of visible light, but can't be supplied to private individuals unless your name is Anish Kapoor, because Anish Kapoor holds the exclusive license for this particular artistic work. If you follow the art world at all, you might know many other artists are not pleased about anybody having exclusive rights to a particular color, even in the name of art. Little context here, another artist named Stuart Semple developed a new pigment to he deemed the pinkest pink. It's very hot back in 2016. He posted it for sale with this legal ease. And I quote by adding this product to your card, you confirm that you are not Anish Kapoor, you are in no way affiliated with Anish Kapoor. You are not purchasing this item on behalf of Anish Kapoor or an associate of Anish Kapoor to the best of your knowledge, information and belief. This paint will not make its way into the hands of Anish Kapoor. Yeah, the fact that he tied it up and won't let anybody else license it is what got everybody said Anish Kapoor. Yeah. And that even that Porsche 911 with the GoPro Musil Black sounds hot. I don't I need my car to not absorb sunlight right now in the middle of July. I have I mean, it's not cool like this, but my my Volvo was black. And, you know, I was like, oh, yeah, I'm black car. I never had one before when you live in the country. Terrible idea. I mean, it's you know, I come back from the car wash 30 minutes later. I'm like, oh, dirty car. So, you know, whatever, but it looks pretty. It does. It does. Do you want that? My thing goes my thought automatically goes to when you're driving at night, like, don't do you have to have like extra light to have like put in it? Yeah, like it's so dark. Running lights become extremely important in that situation. Yeah. All right, let's check out the mail bag before we get out of here. Yeah, let's do it. This one comes in from Tyson, who wanted to share. We're extremely excited to have a star link option for ships. Because Tyson says my job is to build a cloud on a ship where students travel around for a semester and take college courses on the ship. Oh, cool. Like semester at sea. I guess kind of our internet was about 35 K to install about 30 K per month for 10 megabit down and four megabit up connection. Sharing that with 500 to 600 students and staff painfully slow. Having star link would enable us to lower cost and use some cloud services that are currently extremely painful to build local versions of. Tyson says, hopefully this helps understand just how expensive current maritime solutions are and why even at that price, star link is a game changer on ships. Ah, thank you, Tyson. That yeah, that's amazing. Patrick and I were suspecting that it would feel cheaper to those who had to pay and Tyson confirmed that. So we really appreciate that. Absolutely. Yeah. Thank, thanks, thanks for the on the ground report, Tyson. And if you have any reports on anything we talk about, anything we might talk about in the future, feedback at daily tech news show.com is where to send an email because we really love your feedback. We also love having Nika Monford on the show with us. Always a delight, Nika. Let folks know where they can keep up with what you do. I am pretty much on Twitter most of the time at tech savvy diva. You can also follow me over on a snob a Westcast. That is podcast that I host with Terence Gaines, where we talk about a whole bunch of apple stubs because we are Apple snobs. So you can definitely find information about that. We are at snob west cast on all the social media outlets and our website is snobwestcast.com. It's a good show. Uh, subscribe early and often, everybody. Also, we wanted to thank some brand new bosses. We got over the weekend, Alex, Yaneve and Eric all just started back in us on Patreon. Thank you, Alex. Thank you, Yaneve. And thank you. We're one for each day. Indeed. Yeah. Saturday, Sunday and Monday. Thanks. Thanks for that. Perfect. Just need one for tomorrow. Could be. Yes. Just, you know, just, just keep it rolling, everybody. Uh, just a reminder, there's a longer version of the show called Good Day Internet available at patreon.com slash D T N S, which we roll into right after we wrap up D T N S. But just a reminder, we do the show live. We're live Monday through Friday, 4 p.m. Eastern, 20 hundred UTC. Find out more at dailytechnewshow.com slash live. And we'll be back to it all again tomorrow with our old co-host. I, as actor, join us. Don't miss it. Talk to you soon.