WEBVTT - Facial Recognition Machinery, Part 3 0:00:00.120 --> 0:00:04.760 My REGI welcome Stuff to Blow Your Mind, a production 0:00:04.800 --> 0:00:13.680 of I Heart Radios How Stuff Works. Hey, welcome to 0:00:13.720 --> 0:00:15.960 Stuff to Blow Your Mind. My name is Robert Lamb 0:00:16.280 --> 0:00:19.520 and I'm Joe McCormick, and we're back with part three 0:00:19.680 --> 0:00:23.200 of our Journey through Facial Recognition. In the first episode, 0:00:23.200 --> 0:00:25.640 we talked a lot about the sort of the current 0:00:25.680 --> 0:00:29.520 tech landscape of a company focusing on facial recognition, some 0:00:29.640 --> 0:00:32.080 issues with that. In the last episode we focused primarily 0:00:32.080 --> 0:00:35.800 on the biological domain of facial recognition, and now we're 0:00:35.840 --> 0:00:39.120 bringing it back to technology to uh to finish up today. 0:00:39.880 --> 0:00:42.200 So one of the things that we were looking at 0:00:42.600 --> 0:00:45.360 coming into today's episode was an article that I thought 0:00:45.400 --> 0:00:48.520 was really good in Wired magazine again from this month 0:00:48.520 --> 0:00:52.559 from January of twenty by Seaun Revieve called the Secret 0:00:52.680 --> 0:00:56.040 History of Facial Recognition, And I will say I was 0:00:56.160 --> 0:01:00.800 surprised to find out how far back facial recognition projects go. 0:01:01.080 --> 0:01:05.160 This this goes into the nineteen sixties. Yeah, yeah, this, 0:01:05.160 --> 0:01:07.759 this this is a really good read. This article. I mean, 0:01:07.800 --> 0:01:10.480 it's it's extremely well written. It almost has I would 0:01:10.480 --> 0:01:13.600 say it has a very narrative flow to a beginning 0:01:13.840 --> 0:01:18.320 with this scene in which Uh, an elderly researcher who 0:01:18.440 --> 0:01:19.959 is you know at this point, I believe you can 0:01:20.000 --> 0:01:24.440 find to a wheelchair is instructing his son to to 0:01:24.680 --> 0:01:28.920 unlock some old, rotting files from the sixties and burn 0:01:29.000 --> 0:01:31.280 them in front of him and the garbage can in 0:01:31.319 --> 0:01:34.000 the garage or something. Yeah, with you know, with and 0:01:34.080 --> 0:01:36.319 you can you can tell that there are things about 0:01:36.600 --> 0:01:41.319 classified information, top secret or what have you on the documents. 0:01:41.600 --> 0:01:45.399 So it's a wonder wonderful, um you know, ominous start 0:01:45.560 --> 0:01:48.960 to this article, which of course deals predominantly with the 0:01:48.960 --> 0:01:52.720 the origins of facial recognition and touches on other at 0:01:52.800 --> 0:01:59.280 times inspiring and other times creepy and devastating scientific programs 0:01:59.400 --> 0:02:02.520 that were going on or in like in full swing 0:02:02.720 --> 0:02:05.000 during the sixties. Well. Yeah, one of the things that 0:02:05.080 --> 0:02:07.720 really comes home in this article is that the creepiness 0:02:07.800 --> 0:02:11.040 of facial recognition technology is not new. That's sort of 0:02:11.080 --> 0:02:13.120 been there since the very beginning. It's not one of 0:02:13.160 --> 0:02:16.760 these things where um, like even the author here mentions like, 0:02:16.800 --> 0:02:19.200 you know, social media, where it seems great at first, 0:02:19.200 --> 0:02:21.600 and it's not until it's quote in the wild that 0:02:21.720 --> 0:02:24.919 we begin to realize, oh yeah, this is um civilization 0:02:25.000 --> 0:02:28.080 wrecking awfulness. Uh. And not just a fun way to 0:02:28.120 --> 0:02:33.000 share photos. No. At the time, there was a realization 0:02:33.200 --> 0:02:37.840 that this was potentially problematic. Yeah. And so it might 0:02:37.880 --> 0:02:40.040 not come as a big surprise, especially given the story 0:02:40.080 --> 0:02:43.359 we mentioned about burning documents that some of the earliest 0:02:43.400 --> 0:02:47.720 funding for facial recognition technology research clearly came from the 0:02:47.760 --> 0:02:50.760 c i A and front companies set up to funnel 0:02:50.800 --> 0:02:54.600 c i A money. Yeah, this was this was super interesting, 0:02:54.880 --> 0:02:58.840 the CIA funding through these various phony companies, and and 0:02:59.120 --> 0:03:01.360 due to the CI funding, some of the you know, 0:03:01.480 --> 0:03:04.760 stuff was was secret. Some of the image material has 0:03:04.800 --> 0:03:07.040 only come out, you know, due to Freedom of Information 0:03:07.480 --> 0:03:10.720 Act filings, and and some of the work was just 0:03:10.800 --> 0:03:13.959 never published. A lot of the work was never published. Uh. 0:03:14.280 --> 0:03:16.840 To to drive home the creepiness though, one of the 0:03:16.840 --> 0:03:21.480 companies involved was this company, Panoramic, which which was also 0:03:21.520 --> 0:03:25.480 tied to other programs including UH the Project mk Ultra. Uh. 0:03:25.880 --> 0:03:28.959 It was one of eighty organizations that worked on Project 0:03:29.120 --> 0:03:33.600 mk Ultra in particular quote subprojects and ninety four on 0:03:33.639 --> 0:03:36.800 the study of bacterial and fungal toxins and the remote 0:03:36.880 --> 0:03:42.880 directional control of activities of selected species of animals Animal control. Yeah, 0:03:42.960 --> 0:03:44.320 I was like, if I can, if I want to 0:03:44.320 --> 0:03:47.440 sick tigers at you from the other side of the world. Yeah. 0:03:47.680 --> 0:03:50.120 MK Ultra, just to remind everybody, was a c I 0:03:50.360 --> 0:03:57.040 project that explored the potential for mind control using psychedelics 0:03:57.400 --> 0:04:00.480 UH and other tactics like basically looking at ways to 0:04:00.600 --> 0:04:07.040 take these mind expanding um UH agents and use them 0:04:07.040 --> 0:04:10.960 to break down the human psyche and then inevitably built 0:04:11.040 --> 0:04:13.800 something back up that it could be tightly control. Now, 0:04:13.800 --> 0:04:16.640 the evidence today is that the mind control experiments of 0:04:16.760 --> 0:04:19.200 m K culture didn't really work, but they were really 0:04:19.200 --> 0:04:22.560 good at destroying the human mind. I mean because basically 0:04:22.720 --> 0:04:27.120 the project was responsible for psychological torture. UH. Just a 0:04:27.279 --> 0:04:32.360 horrible program in a real blight on the scientific history 0:04:32.480 --> 0:04:36.200 of the United States. UM. You know, not the only blight, 0:04:36.279 --> 0:04:39.440 but but I think an appalling one. They were not good. 0:04:39.480 --> 0:04:41.080 They did not figure out a way how to read 0:04:41.160 --> 0:04:43.919 and you know, to rebuild say an ideal sleeper agent 0:04:44.040 --> 0:04:47.279 out of the psychological destruction that they wrought. Yeah. So 0:04:47.360 --> 0:04:49.880 a lot of this article focuses on this one main 0:04:50.000 --> 0:04:53.919 figure named Woody Bledso, who was a leader at this company. 0:04:54.160 --> 0:04:57.840 A founder and leader at this company, Panoramic Research, Incorporated, 0:04:57.839 --> 0:05:00.960 which you mentioned earlier, which got a lot of business 0:05:01.080 --> 0:05:04.279 from c I A and c I A front organization 0:05:04.400 --> 0:05:08.640 funding in the sixties to study things like facial recognition. 0:05:09.080 --> 0:05:11.640 But if you put yourself back in the context to 0:05:11.640 --> 0:05:13.720 the nineteen sixties, I think one thing that's kind of 0:05:13.760 --> 0:05:17.640 funny is people then might not yet have realized how 0:05:17.680 --> 0:05:21.159 difficult of a project recognizing a face would be for 0:05:21.240 --> 0:05:24.560 a machine, because we'd have tons of sci fi goes 0:05:24.600 --> 0:05:28.120 back decades before that, where of course robots, computers, whatever, 0:05:28.120 --> 0:05:31.000 just recognize people easily. Yeah, And I think that's mainly 0:05:31.040 --> 0:05:33.200 because we would just we generally would just have a 0:05:33.279 --> 0:05:35.880 rough idea that a robot can do everything a person 0:05:36.080 --> 0:05:38.680 can do. A robot is a mechanical person, and you 0:05:38.720 --> 0:05:41.120 didn't have to think too hard about all the complexities 0:05:41.160 --> 0:05:45.159 involved there. I mean, even like the best example of 0:05:45.240 --> 0:05:48.839 nineteen sixties science fiction and really one of the twenty 0:05:48.920 --> 0:05:51.640 and twenty one centuries best examples of science fiction two 0:05:51.680 --> 0:05:56.160 thou one of Space Odyssey, it does reference facial recognition 0:05:56.200 --> 0:05:59.719 capabilities for how but it doesn't I wouldn't say that 0:05:59.760 --> 0:06:02.080 it really it goes in depth about what that means. 0:06:03.120 --> 0:06:06.360 But but how does is able to recognize faces, and 0:06:06.400 --> 0:06:09.040 even like can recognize faces when it is a sketch 0:06:09.360 --> 0:06:12.080 as opposed to video feed or a photo. Yeah, and 0:06:12.279 --> 0:06:15.480 I think if you're not well versed in the computer 0:06:15.560 --> 0:06:18.760 technology world, it might not be immediately apparent what's so 0:06:18.839 --> 0:06:22.440 difficult about making a computer recognize a face? But you know, 0:06:22.520 --> 0:06:25.240 our facial recognition systems, the things going on in our 0:06:25.279 --> 0:06:28.400 brain have amazing capabilities and their analog right. You know, 0:06:28.440 --> 0:06:31.360 a face, face has all kinds of variables that move 0:06:31.400 --> 0:06:34.719 around all the time. It can be extremely difficult to 0:06:35.120 --> 0:06:38.520 reduce a face to a set of nu miracle values, 0:06:38.600 --> 0:06:40.479 which are what you need to do in order to 0:06:40.560 --> 0:06:44.320 have a computer recognize a face. Yeah, as as I 0:06:44.360 --> 0:06:47.960 think we've explored in the previous two episodes, it's um yeah, 0:06:48.000 --> 0:06:50.640 there's there's a lot going on in facial recognition. There 0:06:50.640 --> 0:06:52.480 are a number of challenges to it, and it is 0:06:52.520 --> 0:06:55.880 still not a perfected technology by any means. Yeah, that's true, 0:06:55.880 --> 0:06:58.560 And so I think it's reasonable to think of what 0:06:58.680 --> 0:07:02.240 he bled So and his colleague as legitimate AI pioneers, 0:07:02.279 --> 0:07:05.320 even with their work in the nineteen sixties here. Oh absolutely, 0:07:05.400 --> 0:07:08.000 but I'm Bledso was working with a number of you know, 0:07:08.080 --> 0:07:14.000 highly talented individuals, sometimes on projects like involving atomic weaponry, 0:07:14.040 --> 0:07:18.679 for instance, before he became more focused on AI. Another 0:07:18.720 --> 0:07:22.680 individual that he worked on concerning facial recognition was a 0:07:22.720 --> 0:07:27.120 Helen shan Wolf. Um Wolf was involved in the development 0:07:27.120 --> 0:07:30.440 of Shaky, which is a robot which DARPA describes as 0:07:30.560 --> 0:07:33.880 quote the first mobile robot with enough artificial intelligence to 0:07:33.960 --> 0:07:36.640 navigate on its own through a set of rooms. I 0:07:36.680 --> 0:07:39.120 think somebody at the company also worked on a robot 0:07:39.160 --> 0:07:43.160 called Mobot, which mode lawns and a random and unattended pattern. 0:07:44.480 --> 0:07:46.080 I'm not joking, by the way, where that is the 0:07:46.160 --> 0:07:49.240 thing Revieve mentioned. Yeah, it just always that That makes 0:07:49.240 --> 0:07:52.040 me think of the the old Gumby short where the 0:07:52.080 --> 0:07:55.120 Gumby family have the robots that are doing lawn care 0:07:55.200 --> 0:07:58.040 and home repair and they just go perserk and the 0:07:58.120 --> 0:08:00.720 Gumby family has to as to at them down. I 0:08:00.720 --> 0:08:02.280 don't think I know that one. Oh, it's good. There 0:08:02.320 --> 0:08:04.480 was an MST three k riff five eight years back. 0:08:04.720 --> 0:08:06.960 That sounds horrible. It is. It's horrifying. At the end, 0:08:06.960 --> 0:08:09.560 there's like a robot's head on the wall. It's it's 0:08:10.400 --> 0:08:14.200 terrifying material. But one of the things that ended up 0:08:14.240 --> 0:08:16.520 being the case if if you're trying to think, okay, 0:08:16.560 --> 0:08:19.320 in the sixties, how would you even begin to get 0:08:19.320 --> 0:08:24.000 a computer to recognize a face across different images? Uh. 0:08:24.360 --> 0:08:26.840 One of the problems is, of course, the lack of 0:08:26.920 --> 0:08:30.880 existing digitized imagery at the time, right. I mean, we 0:08:30.920 --> 0:08:34.080 live in a world where digital imagery is ubiquitous. It 0:08:34.120 --> 0:08:35.880 was not at all back then. There was almost none 0:08:35.920 --> 0:08:38.440 of it. Yeah. Yeah, Today when you read about official 0:08:38.520 --> 0:08:41.840 recognition research that are often working off of digital databases 0:08:41.880 --> 0:08:46.199 containing hundreds and not thousands of photographs. Uh. And at 0:08:46.200 --> 0:08:50.319 the time, yeah, how many digital photographs were there in 0:08:50.360 --> 0:08:52.600 the world, right, So they actually had to have some 0:08:52.720 --> 0:08:55.960 kind of digitization process, Like they had to be able 0:08:56.000 --> 0:08:59.640 to take a photo and turn that photo into some 0:08:59.800 --> 0:09:03.280 no numbers that could then be interpreted by the machine 0:09:03.800 --> 0:09:06.760 to try to recognize a person or you know, say that, yeah, 0:09:06.760 --> 0:09:09.000 that's the same person or not. And so a lot 0:09:09.000 --> 0:09:12.480 of these methods involved number one, recognition rules based on 0:09:12.640 --> 0:09:16.240 explicit measures, not machine learning. They didn't have the machine 0:09:16.320 --> 0:09:18.840 learning methods we have today back then. They would have 0:09:18.920 --> 0:09:23.439 to have explicit rules like the distance between randomly selected 0:09:23.520 --> 0:09:26.600 features of the face. So in this image, what's the 0:09:26.640 --> 0:09:29.920 difference between the left eyebrow and the right ear and 0:09:29.960 --> 0:09:31.959 the you know, left eye and the corner the right 0:09:32.000 --> 0:09:35.720 corner of the mouth or something. And furthermore, to get 0:09:35.760 --> 0:09:38.319 those measures, they often had to resort to what they 0:09:38.320 --> 0:09:42.439 called a man machine approach, which was it would pair 0:09:42.480 --> 0:09:47.000 initial measurements made by humans. Uh. It would take those measurements, 0:09:47.000 --> 0:09:50.480 turn them into numerical values, and then train the computer 0:09:50.559 --> 0:09:55.760 to match based on those values, which is quite laborious. UH. 0:09:55.760 --> 0:09:59.000 And the man machine approach was apparently necessary to input 0:09:59.000 --> 0:10:03.120 the measures based really until the seventies, until um Review 0:10:03.160 --> 0:10:06.520 writes quote. In nineteen seventy three, a Japanese computer scientists 0:10:06.600 --> 0:10:11.040 named Takeo Kannate made a major leap in facial recognition technology. 0:10:11.160 --> 0:10:14.040 Using what was then a very rare commodity, a database 0:10:14.240 --> 0:10:19.120 of eight hundred and fifty digitized photographs eight hundred and 0:10:19.160 --> 0:10:24.200 fifty taken from the nineteen seventy World's Fair in sweet To, Japan, 0:10:24.880 --> 0:10:28.280 Kannada developed a program that could extract facial features such 0:10:28.320 --> 0:10:31.920 as the nose, mouth, and eyes without human input, and 0:10:31.960 --> 0:10:34.800 so by doing that, Cannadi was able to finally eliminate 0:10:35.160 --> 0:10:38.480 the man part of the man machine approach the human 0:10:38.520 --> 0:10:41.600 input of the measuring and input of the values. But 0:10:41.760 --> 0:10:44.000 throughout all this period, I mean there were still huge 0:10:44.160 --> 0:10:48.200 problems with machine recognition of faces. Like they would sometimes 0:10:48.200 --> 0:10:51.719 get to the point where the system developed by by 0:10:51.760 --> 0:10:55.760 Panoramic would be it would be more efficient than humans 0:10:55.800 --> 0:10:59.000 at matching faces under ideal conditions. So if you could 0:10:59.000 --> 0:11:02.320 get all the faces like oriented the same way, looking 0:11:02.440 --> 0:11:05.400 right into the camera and all that, the machine would 0:11:05.400 --> 0:11:08.160 be better than humans trying to match photos. But if 0:11:08.200 --> 0:11:10.719 you just pollute the imagery a little bit and make 0:11:10.760 --> 0:11:14.920 people like turn their heads or change the lighting, etcetera. Yeah, 0:11:14.960 --> 0:11:17.800 any kind of problems like that, suddenly the machine loses 0:11:17.840 --> 0:11:20.440 all its advantages and the humans are better again. Right, 0:11:20.440 --> 0:11:22.920 And some of the other problems that are still issues 0:11:23.000 --> 0:11:25.960 today we're problems at the time, like like depending too 0:11:26.040 --> 0:11:31.040 much on say, uh an all lighte male database, you know, 0:11:31.280 --> 0:11:33.080 or something to that effect. You know, where you just 0:11:33.160 --> 0:11:36.400 do not have you don't have a broad enough sample 0:11:36.640 --> 0:11:40.360 of of human appearances to to really have a robust 0:11:40.720 --> 0:11:44.400 facial recognition system. Right. If it's not trained on humans generally, 0:11:44.400 --> 0:11:47.240 it's not gonna work on humans generally, right, Uh. And 0:11:47.320 --> 0:11:50.079 so yeah, like the racial bias problems that show up 0:11:50.080 --> 0:11:53.960 an existing facial recognition technologies today, we're basically they're right 0:11:54.000 --> 0:11:56.319 from the beginning. But towards the end of his article, 0:11:56.360 --> 0:11:59.560 Revieve writes, quote only in the past ten years or so, 0:11:59.600 --> 0:12:03.000 as face show recognition started to become capable of dealing 0:12:03.040 --> 0:12:07.040 with real world imperfection, says A. Neil K. Jane, a 0:12:07.080 --> 0:12:10.320 computer scientist at a Michigan State University and co editor 0:12:10.440 --> 0:12:14.000 of handbook on Face Recognition, nearly all of the obstacles 0:12:14.000 --> 0:12:17.200 that would he encountered, in fact, have fallen away. And 0:12:17.240 --> 0:12:19.800 one of the big ones they point to is not 0:12:19.880 --> 0:12:22.920 just the fact that you can get some digital images now, 0:12:23.160 --> 0:12:25.520 but you can get so many of them that you know, 0:12:25.679 --> 0:12:29.360 it provides these vast databases for for machine learning and 0:12:29.679 --> 0:12:32.920 data sets for training of neural networks and stuff. That's right, 0:12:32.960 --> 0:12:35.679 you can just basically crawl, um, you know, any given 0:12:35.679 --> 0:12:37.960 social media side. And well, we mentioned a few already. 0:12:37.960 --> 0:12:40.880 We'll mention a few different ones as we proceed here. Well, 0:12:40.920 --> 0:12:43.480 maybe we should pivot to talk about the ways that 0:12:43.559 --> 0:12:48.400 facial recognition technology is being used today. Uh And there 0:12:48.480 --> 0:12:50.640 was one thing that I was looking at, uh So 0:12:50.679 --> 0:12:52.200 it was a piece in the New York Times. It 0:12:52.240 --> 0:12:55.560 was an opinion piece a guy named Bruce Schneier, who 0:12:55.640 --> 0:12:58.480 is a fellow at the Harvard Kennedy School, that was 0:12:58.559 --> 0:13:01.200 making a point about facial recognition that I'm not sure 0:13:01.200 --> 0:13:05.000 I fully agree with the framing of. But within making this, 0:13:05.320 --> 0:13:08.400 within writing this article, he articulates something that I think 0:13:08.520 --> 0:13:11.000 is important and clarifying for us to remember as we 0:13:11.040 --> 0:13:14.040 go on. So to first acknowledge his main point, Uh, 0:13:14.040 --> 0:13:17.360 he writes, quote, facial recognition bands are the wrong way 0:13:17.360 --> 0:13:22.000 to fight against modern surveillance. Focusing on one particular identification 0:13:22.080 --> 0:13:25.880 method misconstrues the nature of the surveillance society we're in 0:13:25.920 --> 0:13:29.840 the process of building. So he's arguing facial recognition, you know, 0:13:29.880 --> 0:13:33.640 it's just one technology among many, and banning it doesn't 0:13:33.640 --> 0:13:37.600 necessarily stop all of these other surveillance technologies from doing 0:13:37.640 --> 0:13:40.680 effectively the same thing. Uh. And we can talk later 0:13:40.720 --> 0:13:43.960 about why. I'm probably not really convinced by this argument, 0:13:44.080 --> 0:13:47.240 But Uh, the point that I think is clarifying is 0:13:47.280 --> 0:13:50.080 that he, you know, he says what we call facial 0:13:50.160 --> 0:13:53.560 recognition is not just one act, but it's at least 0:13:53.600 --> 0:13:57.880 three different acts, each coming with their own challenges. Quote 0:13:58.040 --> 0:14:04.640 in all cases. Modern mass r valance has three broad components, identification, correlation, 0:14:04.880 --> 0:14:09.000 and discrimination. So of course identification is just the first step. 0:14:09.080 --> 0:14:13.840 It recognizes who you are. Correlation then associates that ide 0:14:14.080 --> 0:14:18.760 with other information about you, and then finally discrimination treats 0:14:18.760 --> 0:14:21.600 you differently because of that I D or because of 0:14:21.640 --> 0:14:25.280 the associated information. Now, this in and of itself, of course, 0:14:25.360 --> 0:14:28.040 is something that that humans are. Humans are perfectly capable 0:14:28.040 --> 0:14:29.760 of carrying out all three of these tasks without the 0:14:29.800 --> 0:14:32.800 aigive machines. But what we're talking about here is that 0:14:33.040 --> 0:14:35.640 it would be automated. It would be something that would 0:14:35.640 --> 0:14:40.200 be happening by default to everybody, as opposed to something 0:14:40.200 --> 0:14:44.040 that might take place uh and perhaps even laboriously uh 0:14:44.240 --> 0:14:47.760 in specific scenarios such as, you know, once you've been 0:14:47.800 --> 0:14:51.520 flagged at a security checkpoint or something like that, they 0:14:51.800 --> 0:14:53.440 they ask you to pull out your I D and 0:14:53.440 --> 0:14:56.240 then look you up, to correlate you with other information 0:14:56.360 --> 0:14:58.600 and then maybe treat you differently based on whatever they 0:14:58.640 --> 0:15:01.320 find out. That we're dealing with the scenario here where 0:15:01.360 --> 0:15:03.880 that this would all be happening in real time and 0:15:04.000 --> 0:15:06.720 would happen perhaps with very little delay. And I think 0:15:06.760 --> 0:15:10.200 most importantly would be the scale and pervasiveness. It would 0:15:10.200 --> 0:15:14.480 be happening everywhere all the time. We know from experience 0:15:14.480 --> 0:15:19.680 how quickly new digital technology pervades all spaces, so basically, 0:15:20.200 --> 0:15:22.280 you know the one of his arguments is that these 0:15:22.320 --> 0:15:26.240 are three broad components that it might be easier to 0:15:26.320 --> 0:15:31.280 regulate individually, as opposed to saying, let's not do facial recognition. Well, instead, 0:15:31.360 --> 0:15:33.760 let's maybe we go after each of these three things. 0:15:34.000 --> 0:15:36.920 I think that's not necessarily a bad idea to to 0:15:37.000 --> 0:15:40.400 think about a framework for overall regulation of surveillance and 0:15:40.760 --> 0:15:44.360 preservation of privacy. I mean, ultimately, I think I agree 0:15:44.440 --> 0:15:48.200 that it's important to better understand and regulate the entire 0:15:48.320 --> 0:15:52.920 process recognizing the three different components of identification, correlation, and 0:15:52.960 --> 0:15:56.840 discrimination individually. I just don't think it makes a lot 0:15:56.920 --> 0:15:59.920 of sense to frame this is an argument against banning 0:16:00.000 --> 0:16:03.080 a regulating facial recognition, because to me, that sounds kind 0:16:03.080 --> 0:16:06.400 of like saying, well, international treaties banning the use of 0:16:06.440 --> 0:16:09.440 smallpox virus as a weapon of war miss the point 0:16:09.480 --> 0:16:11.960 that we need to rethink our entire concept of war 0:16:12.080 --> 0:16:15.280 and defense, and we need to regulate international conflict in 0:16:15.320 --> 0:16:19.280 a more comprehensive way. Um. I mean, like, yes, that 0:16:19.360 --> 0:16:21.800 would be true. But if you don't know when and 0:16:21.840 --> 0:16:24.920 whether that full comprehensive thing is going to be accomplished, 0:16:25.120 --> 0:16:27.560 and you do currently have a consensus to ban the 0:16:27.640 --> 0:16:31.440 use of germ warfare, why wouldn't you do it? Absolutely? Um, 0:16:31.520 --> 0:16:33.640 here's another quote from that that article though that I 0:16:33.720 --> 0:16:36.240 that I was really taken by quote. The point is 0:16:36.280 --> 0:16:38.920 that it doesn't matter which technology is used to identify 0:16:39.040 --> 0:16:42.640 people that they're currently is. No comprehensive database of heartbeats 0:16:42.720 --> 0:16:45.600 or gates doesn't make the technologies that gather them any 0:16:45.680 --> 0:16:48.080 less effective, And most of the time it doesn't matter 0:16:48.120 --> 0:16:51.040 if identification isn't tied to a real name. What's important 0:16:51.280 --> 0:16:54.960 is that we can be consistently identified over time. And 0:16:55.000 --> 0:16:57.360 then on on top of that though, you know, once 0:16:57.400 --> 0:17:00.560 your real name is then attached to that information, there's 0:17:00.600 --> 0:17:03.160 no turning back, right. The system is building a picture 0:17:03.200 --> 0:17:06.639 of you, move by move, purchased by purchase, search by search, 0:17:06.920 --> 0:17:10.800 over the course of years, if not decades. Yeah. Uh. 0:17:10.840 --> 0:17:13.119 And of course, because it's the Internet, what kind of 0:17:13.200 --> 0:17:15.919 system are we talking about? I mean, I would argue 0:17:15.960 --> 0:17:18.320 that it is a dystopia A little bit less like 0:17:18.480 --> 0:17:22.200 nineteen eighty four and more like Terry Gilliam's Brazil, where 0:17:22.320 --> 0:17:24.560 you know, there's no one person in charge that you 0:17:24.600 --> 0:17:28.199 can rebel against. No one person seems to understand how 0:17:28.240 --> 0:17:30.720 it all works or call all the shots. It is 0:17:30.800 --> 0:17:35.199 a terrible will emanating from the void that is expressed 0:17:35.200 --> 0:17:39.320 through millions of antlike functionaries, each doing it's bidding without 0:17:39.400 --> 0:17:43.240 being bidden. I mean, isn't the oppression scarier when there's 0:17:43.280 --> 0:17:45.159 no boss in charge of it all that you can 0:17:45.200 --> 0:17:48.719 rebel against? Absolutely, And especially when it's uh when it's 0:17:48.760 --> 0:17:51.399 a case like facial recognition where a lot of the 0:17:52.440 --> 0:17:54.320 a lot of the the advancements that have been made, 0:17:54.320 --> 0:17:56.960 and then more specifically a lot of the cases where 0:17:57.320 --> 0:18:02.080 it has been or is being um rolled out. You know, 0:18:02.119 --> 0:18:04.720 it's often it's it's not as a situation where people 0:18:04.720 --> 0:18:07.240 are getting to vote on it or even necessarily having 0:18:07.320 --> 0:18:11.600 any kind of really broad discussion about it before it 0:18:11.600 --> 0:18:14.080 takes place. It just sort of happens and then here 0:18:14.119 --> 0:18:16.280 we are. You know, So not only is there no 0:18:16.440 --> 0:18:20.479 key individual in place that you can blame and rebell against, 0:18:20.520 --> 0:18:22.600 there's like they're not even necessarily a set point in 0:18:22.680 --> 0:18:24.520 time where you could say, we have to go back 0:18:24.560 --> 0:18:27.359 and change this, you know, getting a time machine and 0:18:27.400 --> 0:18:30.840 try and prevent facial recognition from being rolled out. And 0:18:30.840 --> 0:18:33.320 where do you go, Who do you try and stop? 0:18:33.359 --> 0:18:36.040 Who do you try and speak to except the whole world? 0:18:36.640 --> 0:18:38.080 You know? Another thing I would say is to go 0:18:38.119 --> 0:18:41.560 back to my germ warfare analogy. I think it's possible 0:18:41.720 --> 0:18:44.960 that that he's not quite right, that the different methods 0:18:45.000 --> 0:18:48.280 of ideaing you and tracking you are indistinguishable from each other. 0:18:48.600 --> 0:18:52.160 I think it's possible that facial recognition is an especially 0:18:52.320 --> 0:18:57.159 insidious and corrupting type of automated identification compared to some 0:18:57.200 --> 0:18:59.640 other methods, like you know, ideas of credit card numbers 0:18:59.680 --> 0:19:03.720 or math addresses on your phone. Because our lives are 0:19:03.800 --> 0:19:07.400 built around faces, our social existence and our cultures are 0:19:07.440 --> 0:19:11.120 built around interfacing with faces. I mean, it seems like 0:19:11.200 --> 0:19:15.560 somehow a more dangerous kind of well of information to 0:19:15.800 --> 0:19:19.080 poison in the culture than to say, like, well, you know, 0:19:19.160 --> 0:19:21.040 your phone knows where you are. I mean, you could 0:19:21.119 --> 0:19:25.280 potentially smash your phone with a hammer. Yeah, and again, 0:19:25.720 --> 0:19:27.879 I think I've mentioned this already. So many of the 0:19:27.880 --> 0:19:31.480 things involving facial recognition on the surface, it doesn't seem threatening. 0:19:31.480 --> 0:19:33.520 It seems like, well, we wanted machines to do what 0:19:33.520 --> 0:19:35.840 people could do, so we we've created a way for 0:19:35.920 --> 0:19:38.199 them to do it. That's what we always wanted. How 0:19:38.280 --> 0:19:42.120 is that that terrifying? And I think part of it 0:19:42.160 --> 0:19:44.399 is haven't you always wanted a person who is a 0:19:44.440 --> 0:19:47.399 boss who can watch you every minute of the day. Well, 0:19:47.440 --> 0:19:49.679 I mean, yes, when you get do specific examples of it. 0:19:49.680 --> 0:19:51.399 But but I think I think one of the keys 0:19:51.520 --> 0:19:54.560 is that when we're talking about automating this sort of thing, 0:19:54.800 --> 0:19:57.600 and when we're talking about the machine version of it, 0:19:57.800 --> 0:19:59.879 we're talking about a version of it that exists on 0:20:00.119 --> 0:20:03.600 scale and on a scope that is beyond human. And 0:20:03.680 --> 0:20:05.920 that's that's the thing. It's like, we're not talking about 0:20:05.920 --> 0:20:10.040 a human capability anymore. We're talking about an inhuman capability 0:20:10.480 --> 0:20:12.000 the same way that we've you know, we've talked about 0:20:12.000 --> 0:20:14.440 consciousness before and we sort of post the question, well, 0:20:15.160 --> 0:20:18.520 is part of consciousness? The limitation of what we can 0:20:18.560 --> 0:20:21.760 focus our attention on? Is there is part of the 0:20:21.840 --> 0:20:25.159 key to being human? Uh? Is part of it? The 0:20:25.200 --> 0:20:27.520 limitations on what our brains can do, on what our 0:20:27.560 --> 0:20:31.200 senses can do. And we're talking about models that are 0:20:31.280 --> 0:20:34.000 not limited by those senses or those uh, you know, 0:20:34.040 --> 0:20:37.200 those computational resources as beautifully put. And I think you're 0:20:37.240 --> 0:20:40.400 exactly right. I mean, if you could be conscious of everything, 0:20:40.440 --> 0:20:42.800 I'm not sure you would be conscious anymore. I'm not 0:20:42.840 --> 0:20:45.359 sure you'd be a person. Yeah, yeah, I don't think. 0:20:45.880 --> 0:20:50.000 You know, this is something that that far more specialized 0:20:50.280 --> 0:20:53.560 people than myself probably have better arguments about. But yeah, 0:20:53.560 --> 0:20:56.760 to to what extents extent would a super consciousness not 0:20:56.880 --> 0:20:59.600 be a consciousness? It would be beyond what we think 0:20:59.640 --> 0:21:02.720 of a consciousness. Okay, we need to take a quick break, 0:21:02.760 --> 0:21:05.080 but when we come back, we will discuss how facial 0:21:05.080 --> 0:21:08.520 recognition technology is already being used in several countries around 0:21:08.520 --> 0:21:14.680 the world. Alright, we're back, all right, So let's let's 0:21:14.680 --> 0:21:16.360 look at just this is just going to be kind 0:21:16.359 --> 0:21:19.040 of a snapshot at a few different a few different 0:21:19.040 --> 0:21:22.560 stories covering facial recognition as it is being rolled out 0:21:22.680 --> 0:21:25.400 right now as of this recording in at the tail 0:21:25.520 --> 0:21:28.960 end of January. So the first a couple of sources 0:21:28.960 --> 0:21:32.040 of looking at uh. There's a an opinion piece by 0:21:32.240 --> 0:21:37.400 Frederic califooner UH in The Guardian titled Facial Recognition cameras 0:21:37.400 --> 0:21:41.119 will put us all in an identity parade. And this 0:21:41.160 --> 0:21:44.359 piece also refers in places to reporting from the same 0:21:44.400 --> 0:21:48.000 month and publication by Vicram Dodd, a police and crime 0:21:48.040 --> 0:21:53.080 correspondent for The Guardian. So basically, London's Metro Police announced 0:21:53.119 --> 0:21:57.119 their intention to launch live facial recognition cameras in London, 0:21:57.480 --> 0:22:00.239 a city already known for a mass surveillance survey ence 0:22:00.320 --> 0:22:03.960 via their CCTV system. It's a move condemned by civil 0:22:04.000 --> 0:22:08.439 liberty groups as as a quote breathtaking assault on human rights. Uh. 0:22:08.520 --> 0:22:11.880 The police, however, claimed that eight of people surveyed back 0:22:12.000 --> 0:22:13.880 the move and they would only be used to catch 0:22:13.960 --> 0:22:17.119 violent criminals and find missing people. Uh. They also stressed 0:22:17.160 --> 0:22:19.480 that it will be properly posted and it will be 0:22:19.600 --> 0:22:22.840