Blacklisted facial recognition and surveillance companies

The United States has blacklisted several Chinese AI companies working in facial recognition and surveillance. Why? What are these companies doing exactly, and how does this fit into the international politics of AI? We dig into these questions and attempt to do some live fact finding in this episode. Sponsors: DigitalOcean https://do.co/changelog – DigitalOcean now offers three managed databases https://www.digitalocean.com/products/managed-databases/ — PostgreSQL, MySQL, and Redis. Get started for free with a $50 credit. Learn more at do.co/changelog https://do.co/changelog. The Brave Browser https://brave.com/cha913 – Browse the web up to 8x faster than Chrome and Safari, block ads and trackers by default, and reward your favorite creators with the built-in Basic Attention Token https://basicattentiontoken.org. Download Brave https://brave.com/cha913 for free and give tipping a try right here on changelog.com https://changelog.com. Fastly https://www.fastly.com/?utm_source=changelog&utm_medium=podcast&utm_campaign=changelog-sponsorship – Our bandwidth partner. Fastly powers fast, secure, and scalable digital experiences. Move beyond your content delivery network to their powerful edge cloud platform. Learn more at fastly.com https://www.fastly.com/?utm_source=changelog&utm_medium=podcast&utm_campaign=changelog-sponsorship. Featuring: Chris Benson – Website https://chrisbenson.com, GitHub https://github.com/chrisbenson, LinkedIn https://www.linkedin.com/in/chrisbenson, X https://x.com/chrisbensonDaniel Whitenack – Website https://www.datadan.io/, GitHub https://github.com/dwhitena, X https://x.com/dwhitenaShow Notes: 1 million ethnic Uyghurs in China have been held in “re-education camps” https://www.cfr.org/backgrounder/chinas-repression-uighurs-xinjiangSurge in Chinese research related to surveillance of Muslim minorities https://twitter.com/jacobeisenstein/status/1136742525257957377US AI Blacklist summary article from MIT Technology Review https://www.technologyreview.com/f/614494/china-ai-firms-blacklisted-xinjiang-ai-chips/Hikvision Their website https://www.hikvision.com/Here’s what you need to know about Hikvision, the camera maker behind China’s mass surveillance system https://www.scmp.com/tech/big-tech/article/2185123/heres-what-you-need-know-about-hikvision-camera-maker-behind-chinasChina’s Hikvision Has Probably Filmed You https://www.bloomberg.com/news/articles/2019-05-22/china-s-hikvision-weighed-for-u-s-ban-has-probably-filmed-youiFlytek Their website https://www.iflytek.com/Why 500 Million People in China Are Talking to This AI https://www.technologyreview.com/s/608841/why-500-million-people-in-china-are-talking-to-this-ai/Megvii Their website https://www.megvii.com/en/Behind the Rise of China’s Facial-Recognition Giants https://www.wired.com/story/behind-rise-chinas-facial-recognition-giants/Megvii, the Chinese startup unicorn known for facial recognition tech, files to go public in Hong Kong https://techcrunch.com/2019/08/26/megvii-the-chinese-startup-unicorn-known-for-facial-recognition-tech-files-to-go-public-in-hong-kong/Yitu Their website https://www.yitutech.com/enAI Startup Yitu Seeks IPO on China’s New Tech Board https://www.bloomberg.com/news/articles/2019-09-03/ai-startup-yitu-is-said-to-seek-ipo-on-china-s-new-tech-boardThis Chinese facial recognition start-up can identify a person in seconds https://www.cnbc.com/2019/05/16/this-chinese-facial-recognition-start-up-can-id-a-person-in-seconds.htmlLearning resources: TensorFlow 2.0 notebook https://colab.research.google.com/drive/1UCJt8EYjlzCs1H1d1X0iDGYJsHKwu-NOUpcoming Events: Register for upcoming webinars here https://practicalai.fm/webinars!

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Being with your change log is provided by Fastly. Learn more at fastly.com. We move fast and fix things here at ChangeLaw because of Rollbar. Check them out at rollbar.com. And we're hosted on Linode Cloud servers. Head to linode.com/changelog. This episode is brought to you by Digital Ocean, the simplest cloud platform out there. And we're excited to share they now offer dedicated virtual droplets. And unlike standard droplets, which use shared virtual CPU threads, their two performance plans, general purpose and CPU optimized. They have dedicated virtual CPU threads. This translates to higher performance and increased consistency during CPU intensive processes.

• 0:41Speaker 00

  So if you have build boxes, C I C D, video encoding, machine learning, ad serving, game servers, databases, batch processing, data mining, application servers, or active front-end web servers that need to be full-duty CPU all day, every day, then check out Digital Ocean's dedicated virtual CPU droplets. Pricing is very competitive, starting at forty bucks a month. Learn more, get started for free with a fifty dollar credit at do.co/changelog. Again, do.co/changelog.

• 1:18Speaker 00

  Welcome to Practical AI, a weekly podcast about making artificial intelligence practical, productive, and accessible to everyone. This is where conversations around AI, machine learning, and data science happen. Join the community and slay with us around various topics of the show at changelog.com/community. Follow us on Twitter, we're at Practical AI FM. And now into the show.

• 1:44Daniel Whitenack?

  Welcome to another fully connected episode of Practical AI, where my co-host Chris and I keep you fully connected with everything that's happening in the AI community. We'll take some time to discuss some of the latest AI news and dig into some learning resources to help you level up your machine learning game. I'm joined today, as usual, by my co-host Chris Benson, who is a Principal AI Strategist at Lockheed Martin. How are you doing, Chris? Welcome back from vacation.

• 2:17Chris Benson?

  Thank you very much. It's good to be back.

• 2:20Daniel Whitenack?

  Yeah, you, uh, so our listeners may not know, but you, you got to have a nice, uh, nice couple of weeks of, of vacation and, uh, hope, hope you, uh, had some, some good rest with that.

• 2:32Chris Benson?

  I, I did. Uh, I, I had started as, as we were recording some previous episodes, uh, in the UK, but my wife is British, so she and my daughter joined me there and we spent two weeks with, uh, family and friends and had a great time. Thanks, thanks. It was good. It was nice just to, to walk away for two weeks and, and recharge the batteries. So I'm back, ready to go.

• 2:51Daniel Whitenack?

  Awesome. Well, uh, you may or may not know, but during the last couple of weeks, while you were, uh, enjoying vacation, um, the international AI politics world kind of exploded a bit, uh, to, to say the least. So there's, there's been some developments related to, uh, China and, and AI. Um, and that's definitely been kind of a central point in the news that's intersected the AI community. So I thought that might be good to, to dig into a bit today. What do you think?

• 3:26Chris Benson?

  Absolutely. Let's jump into it. Um, and, and, and I've been largely sequestered from the news. So, uh, I'll be coming in with bright eyed and bushy tailed, uh, hearing it for the first time. So let's, uh, dive in with what you got, Daniel.

• 3:39Daniel Whitenack?

  Sounds, sounds good. And yeah, and I should say, I don't think either one of us consider ourselves, uh, complete, uh, you know, or, or even semi experts on, uh, international trade and politics and all sorts of things that, that factor into this. Um, but kind of the, the bottom line is that AI and tech and these sort of interactions that have been happening between the US and China, they're, they're all kind of intertwined in this really weird and interesting and sometimes disturbing way. Um, and so, yeah, maybe a good way to start out this discussion would just be to,

• 4:19Daniel Whitenack?

  in a, give a little bit of background on some of the elements that are coming to, uh, coming into this. Um, so we're eventually going to get to a point where we can talk about this AI blacklist that the US came out with that kind of is blacklisting, uh, uh, US companies from doing business with certain AI companies in China, which is why we're talking about it on this podcast. But there's a whole bunch of things factoring into that. One of those is, The Uighur population in China. So the Uyghurs in China are an ethnic minority. They're a Muslim minority group in China, in the western part of China, in Xinjiang.

• 5:02Daniel Whitenack?

  And it's pretty well documented at this point. So we're not speculating here, but it's pretty well documented by the United Nations and others that there's been about a million of these ethnic Uyghurs in China that have been detained by China in re-education camps, what they call re-education camps, essentially detained. Have you heard about this?

• 5:31Chris Benson?

  Totally have. That was, you know, before I went on vacation, that had been going on for quite some time. And I think it's tragic as any time that you have a government clamping down on ethnic groups. You know, that's just, it's a sad situation. So yeah, I think it's definitely something that we should all be aware of if we're not already.

• 5:52Daniel Whitenack?

  Yeah, definitely. I mean, there's of course a lot of human rights issues tied up in this, but it is interesting that the, that all of this sort of stuff that's happening with the Uyghurs in China is very connected actually to AI and the tech side of things. So of course China has been taking kind of a harder line approach to dealing with these Muslim minorities. And part of that has to do with like the sophisticated surveillance technology that they're, that they're developing and deploying across this region. I think you've probably mentioned a couple of times on the podcast, you know, things related to facial recognition, if I remember right,

• 6:38Daniel Whitenack?

  and some of the, some of the ethics things tied into that.

• 6:41Chris Benson?

  Yeah, I mean, China, in terms of surveillance of its own citizens, China, I mean, doesn't just lead the way globally. They are in a classification all by themselves. I recently, for, for a talk that I was giving that wasn't specifically about this, but I was looking... At different cities around the world with surveillance and you know that had different some tie into AI and some didn't and such but um so you know China just dominated the list and if you looked at the raw number of cameras in the regions um China just you know orders of magnitude more so um there's kind of China and then there's the rest of the world in terms of of surveillance of this type.

• 7:20Daniel Whitenack?

  And yeah, it's it's increased too. I think in the um I think it was around um like spring and June, I started seeing a pretty big surge in academic research papers that I mean you could imagine cases where maybe they wouldn't be used for surveilling Muslim minorities, but it was almost like completely obvious that these papers from Chinese researchers were were geared towards upping the surveillance um of these of these minority groups. I'm just pulling up a couple of those now that I that I had had written down. So this one, it's an article, it's like facial

• 8:03Daniel Whitenack?

  feature discovery for ethnicity recognition. So it's not like really like um hidden, you know, at all there. They even talk about constructing a data set, an ethnical group face data set, including Chinese, Uighur, Tibetan, and Korean. And and so these papers are not so subtle and it's pretty clear that there's a lot of AI research that's that's actually going into this, which is uh which is is pretty disturbing. I know there were a lot of people at that time um kind of calling for peer review journals to to up their, you know, ethics part of their review

• 8:48Daniel Whitenack?

  um with these things. And I guess that gets into some conversations around censorship and other things with a whole lot of things factored in here, but.

• 8:56Chris Benson?

  Yeah, it's kind of, you know, we've talked previously about, you know, the, the social credit system that is in place in certain parts of China and has been, you know, going through implementation over the last couple of years and continues to be implemented, you know, starting in, in, you know, major cities such as Beijing and then moving out from there. And this is, you know, what we're seeing with the Uyghurs here is essentially kind of the worst case scenario being realized, you know, where, where you're specifically targeting an ethnic group and you're using this, this advanced technology to enable that targeting. It's, it's the world that we, we definitely didn't want to see coming about, you know, as lovers of these, of these technologies that we talk about every week.

• 9:39Daniel Whitenack?

  Yep. And on top of that, so this is kind of a first slice of the complication pie. A second slice of that, of course, is the ongoing US-China trade negotiation slash trade war stuff that's happening. I'm sure even while you were on vacation, you probably could not avoid hearing every once in a while, someone talking about that at the pub or something like that.

• 10:05Chris Benson?

  Oh, constantly. So it was, I mean, in the UK, while I was there, I mean, everyone talks about obviously Brexit, but they also talk about the Americans and the Chinese and, and Trump and, and all of that. So, yeah, even, even though my wife largely tried to ban me from social media, news, you know, anything that was online to try to get me to focus on the family, which I, I tried to comply with as, as best as I possibly could. I think I succeeded. I still, you know, we would, we would meet up with friends and family and they would immediately say, Chris, what's the American take on this and that and the other? And I would try to explain. So, yeah, I mean, the US-China trade war is, is one

• 10:48Chris Benson?

  of those top things that everyone around the world is talking about right now.

• 10:52Daniel Whitenack?

  Yep, and it's kind of in a state of going back and forth like, oh, we're gonna put tariffs on soybeans and whatever, and then we're gonna put tariffs on cranberries, and we're gonna put tariffs on X, Y, and Z. And so you got this back and forth. I think even today, at least at the time of this recording, there's trade negotiations going on between Trump and his team and the Chinese delegation in Washington.

• 11:18Chris Benson?

  Correct.

• 11:19Daniel Whitenack?

  But yeah, so that's kind of pie slice complication too. Then you've got all of this other stuff that's happening in places like Hong Kong, especially Hong Kong, where there's been these pro-democracy protests going on forever. Of course, facial recognition and surveillance has been, you know, a topic in that as well. I've seen pictures of, you know, protesters taking down these surveillance like poles or like posts that have cameras on them and stuff. So there's, there's a lot of, so there's a pro-democracy protest

• 11:58Daniel Whitenack?

  element going on there tied in with, with China, but also an AI component as well.

• 12:05Chris Benson?

  There is. I remember even yesterday, I was reading an article that Apple in the, in the iPhone App Store had removed an app that told protesters where police were. And the, the, the Hong Kong government, their position was you're putting police lives in danger. I, I sense though that the reality is quite the opposite in terms of who is actually in danger. So, yeah.

• 12:32Daniel Whitenack?

  Yeah. And there's, I mean, there's really a lot of elements of this. There's even the MBA, the Basketball Association, which I'm not a sports person, but the MBA is huge in China. I do know that. So they're, they're involved in this now because of some of the things that a coach or manager had said. You know, the TV show South Park, they're, they're kind of embroiled in this. So there's like all of these different elements where like the US and China and AI and tech and trade are all sort of coalescing into this weirdness, I guess.

• 13:14Chris Benson?

  Yeah, I mean, it's definitely, you're definitely seeing, you know, we're seeing divided world in so many, so many areas, and this is becoming a rather extreme case of it.

• 13:26Daniel Whitenack?

  Yep. So that brings us to whenever it was last week, I guess, at this point, where we have this blacklist of AI companies coming out. So the US Commerce Department said that it's adding a bunch of Chinese organizations and businesses to this list, but it includes at least eight primarily AI-focused companies, adding to this list called the Entity List for Acting Against American Foreign Policy Interests. It's a long name, but essentially what this does, like the mechanism or what it results in,

• 14:08Daniel Whitenack?

  is that it bars US companies from selling technologies to these blacklisted entities, which obviously is a type of sort of sanctioned situation.

• 14:21Chris Benson?

  Sure, absolutely. And I think the concern there is probably, you know, first and foremost, security issues on whether or not information is collected and passed back to China for use by the government, or even by commercial entities that are maybe operating on behalf of the government. And then obviously, there's intellectual property issues tied to it as well.

• 15:00Speaker 03

  This episode is brought to you by Brave. The Brave team is on a mission to fix the web by building an open source, privacy-focused, and performance-oriented browser. Browse the web up to eight times faster than Chrome and Safari, block ads and trackers by default, and reward your favorite creators with the built-in basic attention token. Yes, you heard that right, a real-world use case for blockchain. Download Brave for free using the link in the show notes and give tipping a try on changelog.com.

• 15:42Daniel Whitenack?

  Okay, so there's these eight AI-focused companies that have been blacklisted by the US government. And I kind of started, as I was thinking through this, I started thinking, first of all, I mean, it's probably interesting technology in a variety of ways that these companies are developing, whether it's being used for bad or good, I'm not totally sure. I mean, they are kind of all tied up in this complicated situation. And so I thought, maybe, Chris, if you're up for it, maybe what we could do is just try to do a sort

• 16:22Daniel Whitenack?

  of blind taste testing of these companies and see essentially who they are, what they say they're developing, what news sources are telling us that they're developing, and kind of, I guess, get a pulse on, you know, a kind of state of the Chinese AI companies and what they're doing, and what is interesting tech-wise, maybe, what is cool and positive, what isn't cool and positive, what everybody's saying. I don't know, are you up for that game?

• 16:55Chris Benson?

  I'm up for the game. So essentially, here in front of all of our friends listening, we're gonna go to their websites and take a look at what they're saying?

• 17:05Daniel Whitenack?

  Yeah, definitely. So I'll kind of start us out. I thought maybe, so I'll go to a first one's website and kind of try to tell you what my impression is of what they're developing, what they have.

• 17:19Chris Benson?

  Okay.

• 17:20Daniel Whitenack?

  And then, and then maybe at the same time, you could look up a couple of news articles that you find, and we can see if they... Have the same impression of what they're doing, or if there's a gap there, or whatever, whatever we can learn. Sound good? It's a deal. What company you want to start with? All right. So the first one on my list that I had seen was called Hickvision or Heekvision. I'm not sure of the pronunciation. This demonstrates that I'm really going into this blind. Sorry if I'm mispronouncing your company name. I do apologize. But they, so their website

• 18:02Daniel Whitenack?

  that I'm at is us.hick or hikvision.com. And it looks like, like the big thing that I'm noticing when I go to this website is everything about cameras. So there's pages of like network cameras, like product selectors where you can select your camera. It's talking about apertures on cameras and advanced sensors and all sorts of things. But I see a kind of tab that's, that's machine vision. And so that seems probably to be what's most relevant for us. So it says Hickvision success in video surveillance market was established, blah, blah, blah.

• 18:49Daniel Whitenack?

  They have fast and accurate positioning guidance, dimension measurement and identification.

• 18:57Daniel Whitenack?

  And this is from the robotics division. They're kind of highlighting here an under-vehicle surveillance system and machine vision cameras. So it looks like they're kind of promoting this, the usage of this technology in a variety of ways, which do seem completely legitimate, right? Like if you're trying to see if there's bombs under your car, you could have this camera system under your car. I'm assuming that's what it's meaning to, to identify certain objects under a car. And they're also, they're also emphasizing the machine vision side of things in terms of like manufacturing. So installing these cameras in manufacturing

• 19:39Daniel Whitenack?

  places to trace certain objects through your, your factory, let's say, or to sort items in your, in your factory. And so that's kind of the sense that I get that they're, they're, they're really Leading in machine vision, but emphasizing a lot of these industrial applications of the of the machine vision. I do see one example. That's an unmanned aerial vehicle. So a drone and they're emphasizing use of that and managing transportation, infrastructure and equipment. So like highways.

• 20:16Chris Benson?

  Gotcha. So while you were doing that, I Googled and got a Bloomberg article called China's and I'm probably gonna butcher the name too, but heekvision has probably filmed you is what the article title is. Starts with a couple of bullets. It says cameras are installed at army bases, airports and schools. And then a second bullet says Trump administration concerned about Chinese spying tactics. And as I scanned down through this article while you were talking, there was one paragraph that kind of jumped out. It says heekvision, which is controlled by the Chinese government and Daho are leaders in the market for surveillance technology with cameras that can produce sharp full color images in fog and near total darkness. They also use artificial intelligence.

• 21:01Daniel Whitenack?

  I mean, that's pretty cool.

• 21:02Chris Benson?

  Yeah, they are. They also use artificial intelligence to power 3D people counting cameras and facial recognition systems on a vast scale. And so I certainly can understand why the US government, if if any of that is accurate, and if they were using it in certain venues, I could understand why there might be a concern over that. That doesn't mean obviously that that I have any insight into their activities or how they use it. But I could certainly understand why any government might be concerned about the potential for another government to have to be able to use these technologies to to gain insight from an intelligence standpoint.

• 21:45Daniel Whitenack?

  Yeah, I mean, the sense, it is interesting because the sense I get from the website is very much like a manufacturing focus. Even, you know, some of the headlines are about industrial area scanning cameras and, and of course, the traffic, transportation management stuff. And so, yeah, at least on the English version of the site, that's kind of what they're emphasizing, which, I mean, to be honest, that seems like a very legitimate

• 22:28Daniel Whitenack?

  use of the technology and, and probably something that Amazon and others are doing as well, right?

• 22:35Chris Benson?

  Yeah, and there's, it's interesting that you raise that. You know, a few months ago, there was a bit of a scandal with Amazon, specifically AWS, you know, and its services being used by, you know, for facial recognition, being used by law enforcement here in the US with our citizens. And it was a bit of an uproar. And Amazon turned around, I believe that the product, by the way, was, was the recognition with a K product, and they stopped selling it to law enforcement, is my understanding, if I recall correctly. It's been a few months.

• 23:08Daniel Whitenack?

  Right. But maybe they're using it in their warehouse for similar things that are being emphasized on this website, right? Like traceability or smart logistics is how they kind of term it.

• 23:19Chris Benson?

  Totally. It's funny, as we talk about this, I spend a lot of time on the topic of AI ethics as part of my job and other tangential interests to that. And, you know, as I look at this, you know, it's really all about what is your attention here? What, if you're looking at this company, if they're using it for some of the use cases that you outlined, then, you know, that is beneficial, that is something that, you know, that increases a capability potentially. If they're using it for nefarious purposes, you know, it really comes down to intent and use case, you know, in terms of, you know, whether or not they're being fairly called out or not.

• 24:02Daniel Whitenack?

  Yeah, and I think that, I mean, probably if they're advertising these things and they have like use case stories around them, I imagine they are using the technology for manufacturing and logistics and all sorts of things. I guess the the shadiness is probably the connection with the government, right? And, um, you know, what they what they potentially don't feature on the on the website.

• 24:25Chris Benson?

  I think and I think that's a fair statement. Um, the the concern, I think that, um, and it's it's different. It's not universal. Uh, it's different between the US and the UK, I've noticed. Um, the US concern is, uh, is is largely on, um, you know, kind of worrying about the nefarious issues and intent. Uh, the UK, on the other hand, has, uh, has, you know, largely said they are absolutely going to continue to do business with China. Um, and, you know, they have, uh, you know, outside China, you know, you know, London in general, London is one of the most surveilled cities in the world, uh, with their CCTV system. Um, and and they are a lot less worried about it

• 25:08Chris Benson?

  there. It's a it's an interesting perspective shift.

• 25:11Daniel Whitenack?

  For sure. Um, so let's go to candidate number two. Um, company on the blacklist number two that we're we're gonna look into. And this one's called iFly Tech. I'm pretty sure that's the correct pronunciation. I don't know how else you would you would say that. So that one's easier. Okay. Um, iFly Tech. Um, do you want to take a look at their website? Um, and maybe we can, maybe we can flip flop and then I'll kind of see what I can find elsewhere on the, uh, on the internet.

• 25:46Chris Benson?

  Okay, that is fine. Uh, so I have found their website. Uh, this is all in Chinese for the moment, all in, uh,

• 25:55Daniel Whitenack?

  Ah, you gotta get some, uh, Google Translate going.

• 25:58Chris Benson?

  Yeah.

• 25:59Daniel Whitenack?

  Open that in Chrome.

• 26:01Chris Benson?

  I did open it in Chrome, but it didn't automatically translate. So maybe we'll flip flop for the moment so we don't waste people's time. And then I'll take, I'll do the article Googling.

• 26:10Daniel Whitenack?

  Here we go. And, and I would like to take this time as a proud data scientist with S I L International to highlight the world's language problems.

• 26:21Chris Benson?

  Yes.

• 26:22Daniel Whitenack?

  That are still yet to be solved. Yeah, so I did get the translate option there. So I got the translate option and it's telling me, let's see, I'm scrolling down. There's a bit about education on the, on the front and I see this sort of odd looking robot thing, like a robot guy. They're talking about internet plus government affairs. Solve the roundup operation between civil affairs departments. They're talking about an AI lab.

• 27:03Daniel Whitenack?

  All around application of government and industry, intelligent business, smart, smart tube business. I'm assuming maybe that has to do with trains or something like that. But they have this, I don't know, they have this thing called Hyperbrain or Hyperbrain Project, which is a laboratory for speech and language information processing. That sounds pretty cool.

• 27:35Daniel Whitenack?

  Cognitive intelligence system based on a humanoid neural network based on the key projects of the 800 people. So I, this is all a translation, but it seems like they're at least doing something with like machine intelligence, language. Looks like they've got some partnerships with like malls and shopping. And some developer toolkit sort of stuff. So that's pretty much what I'm, I may be butchering all of that because I'm working off of a translation, but. Right at the top, it says empower the world with AI. So

• 28:15Chris Benson?

  Gotcha.

• 28:16Daniel Whitenack?

  I

• 28:16Chris Benson?

  Seems pretty ambitious. It does. I found an article here. It's technology review.com. Why 500 million people in China are talking to this AI? And it was funny when I pulled in, just as you were starting to get into looking over their webpage, when I heard iFly Tech, I just assumed it was a drone company.

• 28:39Daniel Whitenack?

  Yeah.

• 28:40Chris Benson?

  Not being familiar with it, you know, iFly Tech, I just came from the first Alpha Pilot race, you know, going back to a recent episode where we talked about Alpha Pilot. So I was thinking that. So I was rather surprised to suddenly realize it was language based. They talk in this article about a bunch of things, including that they have a developer platform called the iFly Tech Open Platform, which provides voice-based AI technologies to over 400,000 developers in various industries.

• 29:07Daniel Whitenack?

  Yeah, I see that developer tab on the website.

• 29:10Chris Benson?

  Yeah, they're they're valued in US dollars, they'd be valued as a $12 billion company. International operations. You know, looks interesting. Looks a lot like.

• 29:22Daniel Whitenack?

  Yeah, I mean, this honestly, a lot of the things that they've mentioned are personally interesting to me in terms of the voice and language technology, for sure.

• 29:32Chris Benson?

  Which reminds me, and I'll say this out loud so that all of our listeners here, we've got at some point to interview you about that because for those listeners who don't know, Daniel is quite the expert in that area. So upcoming episode, I've now committed you to in front of our entire audience.

• 29:49Daniel Whitenack?

  All right, we can, we can definitely arrange arrange that sometime. Yeah, I did you, did you see anything in those articles about this whole? Like, one of the things I see them keep emphasizing here is like education and like these child toy robot things.

• 30:07Chris Benson?

  Yeah, there's a, there's a picture at the top. Let me, let me delve into that here. I think that's what we're getting into. What, what do you know about that while I?

• 30:16Daniel Whitenack?

  I don't know. It just says, uh, children's intelligence and then I'm trying to open up the page, but it's not letting me go there. Hopefully I'm not, I'm not blacklisted.

• 30:26Chris Benson?

  No, it's, it's talking about a lot of different uses, but I haven't seen the, the children one yet.

• 30:32Daniel Whitenack?

  Interesting. Yeah. I see, uh, machine translation, smart office, um, intelligent transfer, which I'm not sure what that is. Audio, uh, children's intelligence.

• 30:47Daniel Whitenack?

  And, and learning or education.

• 30:50Chris Benson?

  They have a, a voice assistant for drivers, um, and I, I will not, I'm not going to try the Chi, the Mandarin word for this. I, I bet the English translation is little flying fish.

• 31:01Daniel Whitenack?

  Nice.

• 31:01Chris Benson?

  There you go. Very quaint there.

• 31:03Daniel Whitenack?

  Cool. Well, I mean, uh, I mean, it seems like they're involved in a lot of cool stuff. I don't know if they're, if, uh, they're, it seemed like one of the first things that came up with the last one, the, the, uh, Hickvision was the very close connection with the government.

• 31:19Chris Benson?

  Yes, it was. That, so, you know, just as speaking as someone who doesn't know these companies, I, I had, I could see the possibility for ominous overtones on that, on the, uh, first company we, we looked at. This one seems a lot friendlier, a lot more, you know, child and consumer oriented. Um, it, it doesn't, it certainly doesn't come off as, as, uh, with those ominous overtones. So, um, who knows?

• 31:47Daniel Whitenack?

  Well, let's see, um, let's see if we continue that trend into company number three, which is Meg V or Meg Vi. Uh, maybe one of those. It's M E G V I I.

• 32:00Chris Benson?

  Okay.

• 32:01Daniel Whitenack?

  So why don't you take the website this, this time around?

• 32:04Chris Benson?

  So long as it has the English, because I don't have the translate button. So we'll try it out here.

• 32:08Daniel Whitenack?

  See if, uh, see if you got that.

• 32:10Chris Benson?

  I do. I have an English version. Power humanity with AI. Okay.

• 32:16Daniel Whitenack?

  Ambitious.

• 32:17Chris Benson?

  There. Power humanity. I know. Let's not, let's not dream small. They have a proprietary deep learning framework. They gather top tier AI talent and they integrate. So far, this first page is very typical of what you see with companies.

• 32:34Daniel Whitenack?

  Almost sounds like, uh, like, uh, I mean, Google Brain, OpenAI sort of feel.

• 32:41Chris Benson?

  Yeah, it does. They describe themselves, uh, just as a world-class AI company with core competency in deep learning. Full stack solutions. It kind of feels IBM-ish, you know, kind of like a bunch of corporate talk. You know, that-

• 32:58Daniel Whitenack?

  Somewhat vague.

• 32:59Chris Benson?

  Yeah, somewhat vague. I'm so sorry to you IBMers. I really wasn't trying to insult you. I just apologized there.

• 33:05Daniel Whitenack?

  IBM is doing a lot of cool stuff. Hopefully we'll have someone from there on soon we're talking to, so.

• 33:11Chris Benson?

  Okay. Leadership team of young, uh, they're all men. They're all young men. They all could be, they look the age of my own grown children. So now I'm feeling really old. Okay.

• 33:30Daniel Whitenack?

  Yeah, well, I'm looking, so I'm looking at a couple of articles from Wired and TechCrunch. It looks like one of the big news recently was that the company had an IPO, public listing on the Hong Kong Stock Exchange. But both of these articles that I'm looking at definitely have the more ominous overtone, probably similar to the first one. There's a very, uh, very big focus in both articles, which we'll link in the show notes for sure, all of these things we're talking about, but much more focused on facial recognition,

• 34:11Daniel Whitenack?

  behind the rise of China's facial recognition giants. So, uh, this, this is saying that MegVee is, is one of the four Chinese AI startups specializing in facial recognition. So they specifically call that out as their specialty and valued at more than 1 billion, which is, uh, pretty crazy. So that would qualify them, this is saying qualifying them as a, as a unicorn in Silicon Valley speak. So a unicorn. Obviously promoting a lot of deep learning technology,

• 34:49Daniel Whitenack?

  but seems like that 1 billion and their IPO and most of their value is related to facial recognition and surveillance.

• 34:59Chris Benson?

  Yep.

• 34:59Daniel Whitenack?

  It says MegVee's investors include Alibaba, Ant Financial, and the Bank of China, which I think if, I mean, I'm no expert on those companies, but I think that's a fairly close tie to the Chinese government.

• 35:15Chris Benson?

  Yep, probably so. Interestingly, I don't think the website comes off quite as ominously as I've been scrolling around through it. It really does have that very corporate-y feel to it.

• 35:29Chris Benson?

  You know, we're, and a little bit, you know, we're, rather than, than talking about the applications, they talk about their amazing deep learning capabilities and they talk about, you know, smart cities and a lot of the normal buzzwords that you find with American Silicon Valley companies. So.

• 35:46Daniel Whitenack?

  Cool. It sounds, I mean, they're coming off definitely more research focused.

• 35:52Chris Benson?

  They are, definitely.

• 35:53Daniel Whitenack?

  It seems like the, maybe the other two kind of to be competitive with like a Google Brain or, or OpenAI or something like that.

• 36:03Chris Benson?

  Yep, I would agree.

• 36:05Daniel Whitenack?

  Cool. Well, let's, you know, say that MegVee definitely has some, some interesting, interesting things going on, but maybe there's, there's a sort of ominous overtone that's coming through. Are you getting that?

• 36:23Chris Benson?

  Possibly so, yeah, especially from what you read, from what you mentioned.

• 36:27Daniel Whitenack?

  All right. So that brings us to our last contestant today.

• 36:34Daniel Whitenack?

  I don't know if I should be, be phrasing this as a, as a game, because it's quite serious stuff.

• 36:41Chris Benson?

  It is. I don't know what else to do. Let's have fun with it.

• 36:44Daniel Whitenack?

  All right. Sounds good. I'll, I'll take the website this time around. This one is Yitu. Yitu, not, again, not sure on the pronunciation. Y I T U. Hopefully some of our, our listeners maybe correct us on some of these things that I'm sure we're getting wrong. But here, okay, so I'm at Y2. I actually like their website quite a bit. It's much more appealing design-wise to me than the previous ones I was looking at. So, great job there, Y2.

• 37:21Daniel Whitenack?

  But it looks like they're also, so, seems like a very similarly Google-brainy, research-y type website. They say that they're developing technologies that are driving change in the world. And those include proprietary, full-stack technology, in-depth explorations into fundamental AI technologies, including computer vision, speech recognition, natural language comprehension, and human-machine interactions.

• 37:59Daniel Whitenack?

  They talk about being champions in the face recognition vendor test, which is apparently a gold standard for global industrial applications. They talk about management technologies, 100-million-scale data traffic, unlocking multi-industry scenarios for intelligent cities, and having world-class technology talent. So, there's definitely a sort of "we're exploring all the AI things" feel,

• 38:42Daniel Whitenack?

  plus that sort of very clear call-out that they are the best in the world at facial recognition. There's no hiding that. These are the best.

• 38:54Chris Benson?

  And I see that. I saw two articles. One, it talks about that they are seeking an IPO. That's from Bloomberg. It was a recent one on September 3rd, just over a month ago as we record this. And then, but the other one is a CNBC article, and the title is "These Chinese Facial Recognition..."

• 39:15Chris Benson?

  This Chinese facial recognition startup can identify a person in seconds. And it starts off with key points.

• 39:22Daniel Whitenack?

  They're world champions, man.

• 39:23Chris Benson?

  Apparently so. It starts off with the points that China plans to be a global leader in AI by 2030, which is, we've all known for a while. And the market for facial recognition alone is expected to be 9.6 billion by 2022.

• 39:39Chris Benson?

  The next bullet is a little bit scary, but it's something we already know. It's China's facial recognition database includes nearly every one of China's 1.4 billion citizens.

• 39:51Chris Benson?

  And then it just talked about the fact that they had wide recognition for their facial scan platform. And so a little bit of an ominous overtone before you even get into the article right there. Now, this is CNBC and potentially it's an American take as such on that. But yeah, definitely feels like, you know, a little bit closer to supporting government aims, social currencies such as that.

• 40:22Daniel Whitenack?

  Yeah, one interesting thing that I'm noticing here on this one is, I mean, similar to some of the other ones, they're emphasizing some of the non-facial recognition stuff that they're doing, which sounds amazing. I mean, there's like a cancer detection tool, improving diagnosis of sick children, pushing the boundaries on Mandarin speech recognition. Really cool stuff. And one interesting thing is that some of this, like the sick children one, they talk about Chinese-US joint development. And so there's definitely an international flair to some of this research, which they're highlighting, which, yeah, a lot of that sounds really cool.

• 41:07Daniel Whitenack?

  And I don't doubt that, you know, it's not like we can't, I can look at the link to these papers, right? They're publishing papers on the cancer stuff and other things. So it's definitely not like they're just falsifying what they're doing, right? But facial recognition and that surveillance is a piece of it.

• 41:29Chris Benson?

  Yeah, and you know, that that once again comes back to what is their intent? You know, what are they trying to to do with it across their use case spectrum? Um, and you know, are in probably many are very reasonable use cases. Uh, you know, but are they, you know, it's there are some this this article certainly leaves that ominous overtone in terms of um how it might be supporting the the the social currency system in China, which which obviously, you know, there's a there's a value difference right there, uh, you know, between Western values and and Eastern values, at least there in China. That's the kind of thing that would scare most Westerners in in a variety of Western countries.

• 42:09Chris Benson?

  I think um even in even in places like like in the UK and London where people expect to be surveilled, I think they would expect that the way the information is being used is is not not so nefarious. So uh, it's it's interesting to see uh as we look at global trade, how we're going to reconcile some of these differences in terms of how we approach um society, how we approach uh uh business and trade. Um, and I think uh some of these these companies, you know, they they may be uh acting completely appropriately based on on a an objective standard.

• 42:51Chris Benson?

  I think it would, I think one of the scary things certainly for the uh the American perspective is just the ability to verify that, the ability to understand that they're dealing with a vendor that's that does not have an illachirer motive and stuff, uh, whether or not that put it would deserve to put them on the list here or not, I I have no insight into that. But um, it's a it's a tough thing, and I think I don't think we're anywhere near uh solving that.

• 43:16Daniel Whitenack?

  Yeah, I think, I mean, if I was to summarize kind of what I think I've seen in going through this exercise, which again is just a brief exercise, so you know, I'm sure there's a many more elements of this that that we don't know about. But if I was to kind of summarize my where my mind is at on it is, you know, on the positive side, these AI companies and researchers in China are without a doubt top notch.

• 43:47Chris Benson?

  Absolutely.

• 43:47Daniel Whitenack?

  And uh, are doing some amazing things.

• 43:52Daniel Whitenack?

  In in computer vision, but also outside of computer vision in places like language and and other areas like chat and dialogue and and voice. And so there's no question that they are producing some some amazing research findings and advancing those fields.

• 44:14Daniel Whitenack?

  But there's kind of always this undertone of like, well, how how much are these companies involved with the Chinese government? And how much of their funding is coming from these projects that are explicitly targeting and marginalizing these, you know, Muslim minority and other communities? So there's kind of always like, yes, we know you're innovating in all of these areas, but there's kind of this shadow cast on a lot of that, which is unfortunate in terms of, you know, how much of it is being used for for those purposes. So that's kind of where where my mind is at, I guess.

• 44:56Chris Benson?

  Yeah, I think there's a there's a real cultural difference in terms of knowing, you know, you know, as someone who is working in in the American defense industry, it is you you generally know whether or not a company in this industry has a direct government die or not. You may not know the specifics of the work that they do, but we don't tend to to leave that so ambiguous in terms of your understanding. You know, I work for Lockheed Martin. Everybody knows that Lockheed Martin does work with the with the US government and other governments. It's not hidden. It's in the news all the time. And so you kind of know what you're getting there.

• 45:36Chris Benson?

  I think the challenge in in certainly American minds and maybe maybe Western minds at large is the fact that the relationships are not are not so obvious in in with China and the companies between the government and the companies that are there. And in my very biased viewpoint, I would argue that that if they were able to establish more clearly in a transparent ...way what their business with their own government is, if any, that would help alleviate many of the concerns that other countries have because obviously, no nation state is going to want to subject itself to potential spying by any other country. And that's not

• 46:18Chris Benson?

  even specific to these cases. But I think this is one of those things where having that relationship, whether you have a relationship or don't have a relationship between government and business, being transparent about the existence or lack of one would certainly alleviate concerns around the world.

• 46:39Daniel Whitenack?

  Sure. Well, well put. And I think that this whole episode, it's been a learning experience for me. Hopefully, it has for our listeners as well. We will put all of the links that we accessed into our show notes so that you could recreate our experiment, if you like. Before we close out for the day, I just wanted to mention on a completely different subject, TensorFlow 2.0, that happened. And we always like to share a few practical learning resources. Hopefully, again, this episode has been a learning resource

• 47:19Daniel Whitenack?

  in some ways. But on the practical programming side, Francois Chalais created this really nice TensorFlow 2.0 and Keras overview code lab notebook. So we'll link that in the notes. If you're wanting to keep up with that TensorFlow 2.0 stuff and Keras, that's a great place to start. So I definitely want to take a look at that.

• 47:46Chris Benson?

  Absolutely. And I want to invite all of our listeners to engage us. So many of you already do. Engage us in our Slack community, engage us on Twitter, on LinkedIn. Today was an experiment as an episode, and we enjoyed ourselves. Let us know whether you liked it. If you didn't like it, let us know that, too. We're going to continue to experiment with the show and try different things. And a lot of the things that we try come from your comments. So don't hesitate to let us know what you think and make suggestions. And thank you for listening again.

• 48:18Daniel Whitenack?

  All right. We'll talk to you soon, Chris.

• 48:20Chris Benson?

  Take care, Daniel.

• 48:23Speaker 00

  Alright, thank you for tuning into this episode of Practical AI. If you enjoyed the show, do us a favor, go on iTunes, give us a rating, go in your podcast app and favorite it. If you are on Twitter or a social network, share a link with a friend, whatever you gotta do, share the show with a friend if you enjoyed it. And bandwidth for changelog is provided by Fastly, learn more at fastly.com. And we catch our airs before users do here at changelog because of Rollbar. Check them out at rollbar.com/changelog. And we're hosted on Linode Cloud Servers. Head to linode.com/changelog, check them out, support this show. This episode is hosted by Daniel Whitenack and Chris Benson. The music is by Breakmaster Cylinder. And you can find more shows just like this at changelog.com. When you go there, pop in your email address.

• 49:06Speaker 00

  Get our weekly email keeping you up to date with the news and podcasts for developers in your inbox every single week. Thanks for tuning in, we'll see you next week.