广告
摩尔定律与GaN、面部识别与隐私、贸易战肥皂剧
00:00
00:00

 BRIAN SANTO: I’m Brian Santo, EE Times Editor in Chief, and you're listening to EE Times on Air. This is your Briefing for the week ending July 5th.

BRIAN SANTO:我是EE Times主编Brian Santo,你正在的是EE Times全球联播以下是截至75日的一周新闻要点。

This week…

本周…

Facial recognition is being deployed more frequently, but is the technology ready? And are we ready for it?

面部识别正在更加频繁地被采用,但这一技术准备就绪了吗?我们做好准备了吗?

President Trump said he's lifting official restrictions on doing business with Huawei, an important supplier to communications companies around the world. What does that mean for the global electronics industry moving forward?

特朗普总统表示,他正在取消对与华为公司开展业务的官方限制,华为是全球通信公司的一个重要供应商。这对全球电子行业的未来发展意味着什么?

And we talk with Silicon Valley entrepreneur Alex Lidow about Moore’s Law, gallium nitride, and easing at least one of the indignities of aging.

接下来我们与硅谷企业家Alex Lidow谈论摩尔定律、氮化镓,以及可以避免年长人常常遭遇的身体检测尴尬的技术。

First on today’s program – facial recognition. It’s already being used to keep your phone secure, and to help fight crime. But just like any other tool, it can be mishandled; it’s very easy to violate privacy with facial recognition. Also, just like any other tool, it can be deliberately misused. There are documented instances of governments using facial recognition to intimidate their citizens, not protect them.

首先开始今天的第一个话题--面部识别。它已被用于保护您的手机安全,并帮助打击犯罪。但就像任何其他工具一样,它可能会被滥用,因为面部识别很容易侵犯个人隐私。此外,它也可能被故意误用。已经有记录显示政府使用面部识别来恐吓其公民,而不是保护他们。

EE Times and its sister publications recently published a series that examined technology at the intersection of privacy and security. Editor Sally Ward-Foxton’s contribution was an article on facial recognition. International editor Junko Yoshida caught up with Sally to learn more.

EE Times及其姐妹刊物最近发表了一系列文章,探讨了隐私与安全交叉的技术。我们的编辑同事Sally Ward-Foxton为此写了一篇关于面部识别的文章,我们的国际主编Junko YoshidaSally进行了深入访谈

以下为英文原稿,Enjoy!

JUNKO YOSHIDA: I was really fascinated by your latest story about Facial Recognition: The Ugly Truth. It was interesting to be because I thought this story, this particular story, offered the perfect nexus of technology and regulations.

SALLY WARD-FOXTON: Yes, that's absolutely right. Certain the technology's coming along leaps and bounds, but I think we are still playing catch-up a little bit with the law and the regulatory framework.

JUNKO YOSHIDA: Yeah. Now, what struck me reading your article was that it sort of begged the question, What's the state of accuracy in face recognition technology today? Tell me.

SALLY WARD-FOXTON: So it's still pretty heavily application-dependent, I think. In something like a passport gate where you're walking through, you look directly at the camera, you're close to the camera, there's good lighting. Then it can be very accurate. But for surveillance footage, results can definitely vary.

In the field trials that the British police did, they misidentified 96%. It wasn't very accurate at all. So there are a lot of false positives, let's put it that way.

JUNKO YOSHIDA: Wow. Ninety-six percent kind of tells us either the technology is not ready at all, or something is fundamentally wrong about the technology. What's your take on this?

SALLY WARD-FOXTON: I think it is a very difficult environment. So the police had cameras on top of a van, literally just looking at people's faces as they were walking by. It is a very difficult environment, and I think we do need to do more tests and figure out exactly where we're going wrong.

JUNKO YOSHIDA: So as for these trials in the UK you just mentioned on face recognition, how did the society-- I'm talking about the media and public-- respond?

SALLY WARD-FOXTON: So there was certainly a lot of coverage of it in the media. Some of the civil liberties organizations over here made a lot of noise about it, particularly because there isn't really a legal framework for the use of this technology. We don't know exactly what the police are going to do with this data, how long will they store it for. And in particular there's been one kind of landmark legal case, which is a guy in South Wales who is suing the police, effectively saying that his civil liberties have been infringed.

JUNKO YOSHIDA: Good for him, actually. The reason why I say it, it's that the low accuracy of the technology itself is clearly a concern, but even more worrying is, as you said, the lack of framework, right? Lack of clear guidance or regulations in the application of face recognition or face analysis technologies.

So walk us through. What's the regulatory landscape in the UK and maybe in Europe at large?

SALLY WARD-FOXTON: So I think it's still unclear. Currently in the UK I believe there are some very comprehensive rules for when the police can take your fingerprints or your DNA. But UK law doesn't really cover the biometric data. It's all about fingerprints, DNA, how long can the police store that data, what can they do with it, that kind of thing. We do have the, under EU law, the GDPR, the General Data Protection Regulation. This kind of broadly covers personal data. Biometrics is mentioned, but I think it's looser or broader than some people would like.

JUNKO YOSHIDA: I see. Okay. So this is really a hard question to a technology journalist, but do you think it's time to call for a moratorium on this technology in your opinion?

SALLY WARD-FOXTON: Yes. Absolutely. I definitely think it would be time to hit pause on this a little now that we've had the results from the British trials. It wasn't so accurate.

JUNKO YOSHIDA: Yeah.

SALLY WARD-FOXTON: I mean, we know from... We've talked about this in previous projects, previous articles. We know AI definitely is more likely to misidentify women and people of color.

JUNKO YOSHIDA: Right.

SALLY WARD-FOXTON: So it's not accurate. It misproportionately misidentifies certain groups of people. It certainly made me think twice about whether, if I saw the police camera van in the street, would I turn around and walk the other way. I'm not a person of color, but I am a woman. I would be more likely to be misidentified than other people. So yeah, it's something I would be concerned about for sure.

BRIAN SANTO: Global leaders gathered at the G-20 Summit in Japan last weekend. US President Donald Trump and China’s President Xi Jinping met, and unexpectedly came to an agreement that was vague on details, but seems to reset the trade war back to its status from several months ago: the US would remove trade restrictions-- notably those on Huawei-- and China pledged to buy more US agricultural products.

Trump’s assault on Huawei has been incredibly disruptive to the global electronics industry. I got on the phone with international editors Junko Yoshida and Bolaji Ojo to discuss what might happen next. How does the electronics industry proceed? Of course, all of that is contingent on where we are right now. I asked Junko and Bola about the de-escalation in the trade war.

JUNKO YOSHIDA: Well, I must inject myself here, Brian. I don't think anything is de-escalated. It's just put on pause I think. I don't think we are in a position to say, Things have gotten any better. Maybe you could say that, but it's not much of a de-escalation in my opinion.

BRIAN SANTO: Well that's the point, isn't it?

JUNKO YOSHIDA: Yeah.

BRIAN SANTO: Bolaji has been talking about the uncertainty about what's going on here, and that hasn't alleviated, has it, Bolaji?

BOLAJI OJO: No, it hasn't. I think what's happening has to do more with just a reassessment of the situation. If I was a business executive right now, I would not be pulling back from whatever I was drawing on before. I wouldn't even think that anything has changed in order for me to say, Stop; don't do anything else again.

JUNKO YOSHIDA: Right. But here's the thing though: What were you doing until now? Were you stopping exporting your goods to China? I guess it depends on the company. But when you say that they're not changing strategy, what does it mean? I don't get that.

BOLAJI OJO: I guess the point is that, if you had a best-case scenario option and you had a worst-case scenario option, and mostly likely to have an option you still continue to put in place every single action and step that will help you execute any one of them that comes up. Because there isn't a sign that you can rule out any one of these. It's still there. All of it is driven by the unknown. And that unknown remains a huge presence in the room. You still got to deal with that.

JUNKO YOSHIDA: Right. Okay.

BRIAN SANTO: So next steps are going to be difficult for everybody. Before they were considering changing all their supply chains and whether they're going to continue to do business with the partners that they've had so far. It may be worry or fear that the ban from China will be reinstituted and then where are you? Or maybe they're just going to have to rethink what their relationships are. Everybody's afraid of what's going to happen next still I think.

JUNKO YOSHIDA: I guess we have to think about everybody. I mean, every company on earth outside China must be thinking about Plan B. And that Plan B is still-- I think to Bolaji's point-- is that Plan B still can't totally ignore China. Because China is the biggest market in terms of consumption. So what do we do now? I mean, what kind of strategy do we develop then?

BRIAN SANTO: I think that you continue to try to sell into China as best you can. I think that's going to be tougher moving forward for US-based companies. The trust is gone. If there was any trust there before, it's been shattered and obliterated. If I were somebody who had a business in China, if I were a Chinese company, I'd start looking for alternate sources of technology. Or, again, redoubling my efforts to develop my own IP, intellectual property.

JUNKO YOSHIDA: That's true. That Plan B goes both ways, doesn't it? I mean, Chinese companies also already are developing Plan B in many ways. That's a good point, Brian.

BOLAJI OJO: What I would like to add is very simply this: Let's go back to what is driving this. What drove China into determining that it wants to dominate a particular segment of the industry? What's driving the US into saying, Look, we don't want such a huge Chinese presence in this part or whatever. It's fear. The fear of the unknown. The fear of what the other party might do.

Now, today it's between the US and China. Who's to say it's not going to be between the US and Germany, the US and France, France and the UK, the UK versus Russia or whoever tomorrow? That fear has to do with what's happening in the technology world.

In my opinion, technology has become such a huge part of what everyone does now that we don't even know what kind of trajectory is going happen. We don't know what kind of technology is going to come out tomorrow that's going to up-end the current environment, the current political environment. What's leading? Who's likely to drop off? I know that fear is a factor. It's driving government, it's driving businesses. And there's no way that you can prepare for that.

BRIAN SANTO: And it's two kinds of fear. It's economic fear: Are we going to maintain...?

Earlier you mentioned that China has a plan to develop its own technology. And you used the word "dominance." But I'm not sure that China ever did. I think China had always talked about self-sufficiency. It's a fine distinction, but the trade war was about dominance. And I think that has increased the fear you're talking about.

If somebody says, I think we all ought to get together and increase the pie, and then everybody's pie slice gets bigger, too. The alternative is we're going to make the pie bigger and I'm going to take most of the pie, that creates fear and uncertainty. More fear, more uncertainty. And I think that's kind of what the spat kind of revolved around. Dominance versus just creating a bigger pie.

So there's an economic worry now, moving forward. Who's going to try to eat my lunch? As opposed to who's going to share a bigger lunch and bring more stuff to the picnic.

And the other fear is, as you said, political. We're worried that the government of China might exploit its relationship with Huawei, whatever that relationship is, and use that for espionage or disruption of some sort. And the fear is probably legitimate on the US side because the US has been doing that.

BOLAJI OJO: With every new technology, with every new advancement comes this tug of war between what was, what is and what we all hope will become. This war is about who's going to dominate, who's going to control the future in the political sphere and in the economic sphere.

Technology has now become so pervasive that there isn't a way by which you can excise it from the economy, you can excise from the political world or anything like that. So if the guys that we report about-- the Qualcomm, the Intel, the Huawei and all of them-- if they are wondering now how they are being sidelined by political leaders, well maybe they should look in the mirror. They created the technology that has now made them bit players in the same tournament that they're supposed to kind of... I mean, Huawei's supposed to go against Ericksson! Huawei's not going against Ericksson right now. It's going against the US government.

So there you go. This is what we have now, and I don't know how that's going to be resolved.

BRIAN SANTO: And whatever agreement came out of China and the US last weekend doesn't point to any solution either.

JUNKO YOSHIDA: Well exactly. It's a Band-Aid.

BOLAJI OJO: And here's the thing, Brian. And I'll just kind of wrap this up on my side with this: Whatever agreement finally comes out of this, two governments, two political leaders are going to sign it, and then they're going to hand it over to the technology world to go and implement. They're going to say, Well, the US government and the Chinese government have agreed this is the new playing field that we've leveled for you guys in the technology world. And you know what? None of these companies, even on the Western side, or on the Asian side, is going to be able to say, We don't like it. Go and rewrite it. We have to accept what they've been given, and we have to deal with it. That's the playing field you've got it. That's it.

BRIAN SANTO: Alex Lidow has deep roots in the electronics industry. His father and grandfather founded International Rectifier in 1947. Alex eventually ran the company himself for 12 years. He is currently the CEO of EPC, a company that manufactures gallium nitride based power transistors and integrated circuits. These products are now found in lidar systems for autonomous vehicles, in 4G/LTE base stations, in DC-DC converters for servers and satellites, and in a wide variety of medical products.

There’s been a lot of consternation about the end of Moore’s Law. Now I’d spoken with Lidow before, and we’d discussed how Moore’s Law had ended for some circuitry a long time ago.

Everybody's in a tizzy about Moore's Law coming to an end. Now for a certain class of semiconductors, integrated circuits, Moore's Law came to an end a while ago, right?

ALEX LIDOW: Yeah. And I think that's true also for power semiconductors. And they are for different reasons.

I think that Moore's Law, in it's conception by Gordon Moore, it was transistors, number of transistors on a chip doubling every year. But I think that it became a social compact between producers of integrated circuits and consumers of integrated circuits. And the consumer expected every year we're going to see a doubling of performance and it's going to come at a lower price. And as a result, consumers-- you know, computer manufacturers and designers or all sorts of gadgetry-- they would actually design in parallel with integrated circuit development in anticipation of higher performance. You know, I'll design by laptop computer, but I'll make sure that it can use twice as much memory when it comes out a year from now.

Now when Moore's Law started slowing down and then stopping-- and I would contend that it is actually no longer doubling every couple of years and lowering the price-- it's now OR instead of an AND. And so now we see product development that is in series with new product launches from the semiconductor industry. So there's a general slow down in that whole chain of innovation that we see.

And power semiconductors followed the same path, although not because of a doubling of transistors, but rather of a doubling of current density over a few years. And that stopped approximately the beginning of this century.

BRIAN SANTO: Now part of the story is silicon as a semiconductor running out of head room. But there are other ways to improve performance in semiconductors: alternate semiconductor materials, other constructions. And you're a big proponent of using gallium nitride.

ALEX LIDOW: Yeah. I think that in the world of digital electronics they can go to QBits and all sorts of things like that. In the world of power electronics, I think that the solution to maintain this vibrancy in performance really comes from new materials. And there are two materials that I think are the next generation past silicon, and that's silicon carbide and gallium nitride. Of course, my company, Efficient Power Conversion, does everything with gallium nitride grown as a very thin layer on top of a standard piece of silicon.

BRIAN SANTO: And the two different materials excel in two different ends of the performance spectrum, yeah?

ALEX LIDOW: Yeah. Look, I think that silicon carbide-- which is a marvelous material, and there are many companies making good products with that-- is really good at higher voltage ranges. Say above 7, 800 volts up to thousands of volts. Areas where silicon could never reach. And gallium nitride is really great at 600 volts and down, making higher power density devices than silicon, and also at a lower cost. So that's dividing up the future markets that silicon otherwise has dominated.

BRIAN SANTO: We've been talking about power ICs. You also have suggested that it might be possible to use gallium nitride in almost a C-MOS type of a construction. Not there yet. Explain what you're talking about and where gallium nitride might be able to actually do some really interesting things in the future.

ALEX LIDOW: Gallium nitride is a wonderful material for integrated. In silicon, silicon's great for digital integration, but terrible for power integration. And with gallium nitride, you don't have that problem. So you can integrate a bunch of power devices. So you can now build systems on a chip in the world of power that you couldn't do in silicon. And we are doing that. And I would say that we're redefining what a power semiconductor is, and it's more of a combination of power devices along with analog circuitry and digital circuitry.

Now in the world of digital circuitry, gallium nitride is great at N-MOS. We can do anything that silicon can do in N-MOS, and we can do it better. But where the limitation is, is C-MOS. And with gallium nitride, those holes-- you know, the positive carriers or p-type carriers-- just don't go very well in gallium nitride.

Now there are ways that we can probably improve on that, but until we do, we're not going to get C-MOS or complimentary MOS like in silicon.

BRIAN SANTO: Are improvements possible?

ALEX LIDOW: I believe it's possible. But we have three different paths that we think each could lead to a success. When that happens, then there's no real good advantage of silicon over gallium nitride in most logic applications.

BRIAN SANTO: Let's talk about where gallium nitride is excelling right now, perhaps the marquis application is lidar.

ALEX LIDOW: Yeah. I think it's a great application, because in order to measure the time of flight of light-- lidar is Light Detection and Ranging-- in order to measure the time of flight of light, you have to be really fast. And gallium nitride is maybe a hundred times faster than silicon. So we can resolve to say millimeter precision objects that are hundreds of meters away using gallium nitride, when if you used silicon you'd just see this fuzzy sphere about ten feet in diameter. And that's going not just on autonomous cars now, but we see them putting them on robot vacuum cleaners and drones and just about anything now gets a lidar system on it.

BRIAN SANTO: And you were even talking about maybe hand-hands, mobile phones, or augmented reality.

ALEX LIDOW: Exactly. Absolutely for facial recognition, augmented reality glasses where the way you figure out what reality is, you create a quick digital map of your surroundings, and you do that with lidar. It's a very simple, fast way to do it.

BRIAN SANTO: That's fascinating. Also, there's a bunch of other really fascinating, fun things that people have been doing with gallium nitride. It's space-hardened, radiation-hardened, so space applications.

ALEX LIDOW: Yeah.

BRIAN SANTO: And some medical applications.

ALEX LIDOW: Yeah, we want to fly to Europa and Titan. I guess there was a newly announced thing to go to Titan in space because very, very radiation hard.

But some of the really cool applications that are more down to earth, shall we say, are... and because you and I are of an age and we're male, we need to have colonoscopies periodically, and one of the things gn's allowed to do as enabled is miniaturizing an x-ray machine into the size of a pill. And you swallow this thing and it actually takes a very, very ultra-low dose x-ray of your colon. So you don't even need to purge ahead of time. You don't need to get rid of the food matter, because it can tell what's food and what's tissue. And it wirelessly sends it to a patch that you wear on your back. So basically to have a colonoscopy, you take a pill and three days later you peel this patch off your back and you send it in a prepaid package. And that's your colonoscopy for now. And it's a whole lot better than a stick up the hoo-hah...

BRIAN SANTO: Now that you mention it, yeah.

ALEX LIDOW: And it's approved in Europe. It's approved in Europe.

BRIAN SANTO: And the US is close, right?

ALEX LIDOW: Maybe a year away.

BRIAN SANTO: Okay. Fantastic. Alex, thank you so much for spending time with us. We hope to have you back sometime.

ALEX LIDOW: All right, Brian. I'd love it. And thank you so much.

BRIAN SANTO: And now some recent anniversaries of notable technological events in the past:

On June 30th in 1947, Bell Labs held a press conference to announce the creation of the first point-contact transistor. It was barely noted at the time, but the discovery eventually went on to earn its inventors a Nobel Prize, and every electronic device in the world is now based on it.

On July 1st, 1979, the Sony Walkman first went on sale in Japan. There wasn’t any new technology involved to speak of, but the form-factor was unique; it’s one of the most iconic consumer electronics devices ever. True fact: It got developed because Sony founder Masaru Ibuka wanted to listen to opera on long flights without bothering his seatmates.

In honor of Ibuka, we're going to take you out with my favorite operatic piece, Rossini’s The Barber of Seville.

And that’s your Weekly Briefing for the week ending July 5th.

This podcast is Produced by AspenCore Studio. It was Engineered by Taylor Marvin and Greg McRae at Coupe Studios. The Segment Producer was Kaitie Huss.

The transcript of this podcast can be found on EETimes.com, complete with links to the articles we refer to.

Be sure to join us next week for your July 12th Weekly Briefing on EE Times On Air. I’m Brian Santo.

感谢收听本期推送,全球联播 (EE|Times On Air) 现已同期在喜马拉雅以及蜻蜓FM上线,欢迎订阅收听!
广告