WEBVTT - Halt! Bot or Not? 0:00:04.519 --> 0:00:12.319 Technology with tech Stuff from works dot com. Hey there, 0:00:12.360 --> 0:00:15.680 and welcome to tech Stuff. I'm your host, Jonathan Strickland. 0:00:15.720 --> 0:00:18.880 I'm a senior writer for how stuff works dot com 0:00:19.040 --> 0:00:24.320 focusing on all things technological, and recently I did an 0:00:24.320 --> 0:00:28.880 episode about artificial intelligence and how Mark Zuckerberg and Elon 0:00:28.960 --> 0:00:33.760 Musk had kind of a public disagreement about the direction 0:00:33.800 --> 0:00:36.239 of AI, and how other people have weighed in. Some 0:00:36.280 --> 0:00:40.240 people have said that perhaps Musk and Zuckerberg are arguing 0:00:40.280 --> 0:00:44.360 about something that isn't really relevant right now, and that 0:00:44.440 --> 0:00:47.000 there are in fact other elements of artificial intelligence that 0:00:47.040 --> 0:00:50.000 we should be focusing on instead of whether or not 0:00:50.800 --> 0:00:53.320 it is certain to make our lives better or worse 0:00:53.560 --> 0:00:56.400 or rule over us. But it got me to thinking 0:00:56.440 --> 0:00:59.960 about a related topic, and I touched on it all 0:01:00.040 --> 0:01:03.640 little bit in that episode, and that was all about 0:01:04.560 --> 0:01:10.840 how do you tell when a a an entity that 0:01:10.959 --> 0:01:16.080 is communicating with you is in fact a person or 0:01:16.160 --> 0:01:19.679 it is a computer program that is mimicking a person. 0:01:20.200 --> 0:01:22.840 So we're going to look at that, And honestly, I 0:01:22.880 --> 0:01:25.360 was inspired a lot also by the fact that we've 0:01:25.360 --> 0:01:28.000 got a new Blade Runner movie coming out. It's Blade 0:01:28.080 --> 0:01:31.560 Runner two thousand, forty nine, which has no connection to 0:01:31.640 --> 0:01:34.399 this show. By the way, they are not sponsoring us. 0:01:35.000 --> 0:01:37.640 I'm pretty sure no one connected to Blade Runner two 0:01:37.640 --> 0:01:41.280 thousand forty nine even is aware that I exist. But 0:01:41.800 --> 0:01:44.240 I'm a fan of the original Blade Runner film and 0:01:44.240 --> 0:01:48.480 I'm looking forward to seeing what happens in Blade Runner 0:01:48.520 --> 0:01:52.320 two thousand forty nine. I'm a little hesitant because it 0:01:52.360 --> 0:01:56.360 depends upon which interpretation of the original film they decided 0:01:56.400 --> 0:01:59.560 to ultimately go with. If they went with the director's vision, 0:02:00.320 --> 0:02:02.680 might not want to see two thousand forty nine, but 0:02:03.320 --> 0:02:07.440 I wanted to kind of talk about the difference between 0:02:07.440 --> 0:02:10.040 communicating with a person and a synthetic being. Now in 0:02:10.080 --> 0:02:13.959 Blade Runner, the synthetic beings are called replicants, and they 0:02:14.000 --> 0:02:17.760 are not exactly robots. They're often referred to as androids, 0:02:18.240 --> 0:02:21.280 but I don't really think that's terribly accurate either. They're 0:02:21.280 --> 0:02:28.520 more like genetically engineered human simulations. Like they're not fully human. Uh. 0:02:28.560 --> 0:02:33.960 They have other elements that either augmented abilities and intelligence, 0:02:34.560 --> 0:02:37.440 but a lower lifespan that sort of stuffy. They tend 0:02:37.480 --> 0:02:41.519 to be born in the adult stage of their lives 0:02:41.520 --> 0:02:45.440 and implanted with false memories, but they're meant to do 0:02:45.560 --> 0:02:48.120 jobs that humans can't or won't do, and they do 0:02:48.240 --> 0:02:50.720 have a tendency to resent their lot in life, seeing 0:02:50.760 --> 0:02:53.880 as how in the original film they did just have 0:02:54.000 --> 0:02:56.200 that built an expiration data just a few years. They 0:02:56.240 --> 0:02:58.280 don't live for a few years, and then they would 0:02:58.720 --> 0:03:03.440 their bodies would break down and blade Runner the story 0:03:03.520 --> 0:03:07.760 follows an investigator who is seeking out specific replicants that 0:03:07.800 --> 0:03:10.640 are on the run in order to quote unquote retire 0:03:10.760 --> 0:03:14.640 them with extreme prejudice. So this is all set up 0:03:14.639 --> 0:03:17.959 at the beginning of the movie. Now, one thing those 0:03:18.000 --> 0:03:23.280 investigators or blade Runners do is ask questions of suspects, 0:03:23.320 --> 0:03:25.600 suspected replicants. You know, they find someone they think that 0:03:25.720 --> 0:03:29.000 might be a replicant, and then they interview that person 0:03:29.080 --> 0:03:32.560 and they look for signs that that is not actually 0:03:32.600 --> 0:03:36.880 a real human being, because replicants are not exactly human. 0:03:36.920 --> 0:03:40.720 They're human like, but they do not process emotions the 0:03:40.840 --> 0:03:44.080 same way that humans do. So blade Runners can look 0:03:44.120 --> 0:03:47.880 for indications that the suspect is actually a replicant and 0:03:47.920 --> 0:03:50.840 they use what is called the void comp test in 0:03:50.880 --> 0:03:54.040 the movie. This is a test that includes the hypothetical 0:03:54.080 --> 0:03:57.440 situation you're in a desert walking along in the sand 0:03:57.480 --> 0:03:59.640 when all of a sudden, you look down and see 0:03:59.680 --> 0:04:02.200 it towards it. This you reach down and flip the 0:04:02.200 --> 0:04:05.440 tortoise on its back. The tortoise lays on its back. 0:04:05.520 --> 0:04:08.480 It's belly baking in the hot sun, beating its legs, 0:04:08.480 --> 0:04:12.120 trying to turn itself over, but it can't, not without 0:04:12.200 --> 0:04:16.800 your help. But you're not helping. Why is that? Now? 0:04:16.800 --> 0:04:19.720 I kind of paraphrase that scene because it actually happens 0:04:19.720 --> 0:04:23.560 as dialogue between two characters. But that was the attempt 0:04:23.760 --> 0:04:27.120 of an interrogator to figure out whether or not the 0:04:27.160 --> 0:04:29.960 person they were talking to is actually a human being, 0:04:30.400 --> 0:04:34.800 because the emotional responses would indicate whether or not it 0:04:34.880 --> 0:04:38.039 was a human response, or if there was a lack 0:04:38.080 --> 0:04:40.520 of that, that it was perhaps a replicant. Now, that's 0:04:40.520 --> 0:04:42.560 all science fiction, but in the real world there are 0:04:42.680 --> 0:04:45.960 times when we encounter bots or AI constructs and we 0:04:46.040 --> 0:04:49.040 might not know at first, at least they were not 0:04:49.080 --> 0:04:52.880 communicating with a real life person. In fact, the Interactive 0:04:52.960 --> 0:04:57.200 Advertising Bureau reported in two thousand fourteen that thirty six 0:04:57.279 --> 0:05:00.880 per cent of all web traffic is general to buy bots, 0:05:01.320 --> 0:05:05.080 not people, and the security firm Imperva reported in early 0:05:05.120 --> 0:05:09.239 two thousand seventeen that today that figure is now closer 0:05:09.279 --> 0:05:14.600 to fifty, which means that right now there's more traffic 0:05:14.640 --> 0:05:19.680 on the web being generated from bots than actual human beings, 0:05:20.240 --> 0:05:23.320 And that's not exactly great. Much of the web depends 0:05:23.400 --> 0:05:27.120 upon advertising for monetization. But how do you figure out 0:05:27.200 --> 0:05:31.080 what the value of traffic to your website is when 0:05:31.120 --> 0:05:33.240 you know there's a good chance that more than half 0:05:33.440 --> 0:05:36.960 of all those page views were generated from algorithms, not 0:05:37.040 --> 0:05:40.479 from human beings. Now, much of the bot traffic isn't 0:05:40.480 --> 0:05:43.520 meant to be outright malicious. There might be bots that 0:05:43.600 --> 0:05:47.760 are essentially trying to to scour the Internet for data 0:05:47.839 --> 0:05:50.560 for nefarious purposes, but a lot of them are just 0:05:50.680 --> 0:05:57.680 they're gathering information for, you know, completely innocent purposes. Really, 0:05:58.240 --> 0:06:01.760 gathering information on its own is not necessarily a bad thing. 0:06:02.080 --> 0:06:04.640 It's how we use the information that makes it good 0:06:04.720 --> 0:06:09.039 or bad. It's kind of paraphrasing Shakespeare there. But there 0:06:09.040 --> 0:06:12.680 are the various bots on social platforms and websites that 0:06:12.800 --> 0:06:16.160 also interact with people, and some of them are again benign. 0:06:16.200 --> 0:06:18.960 They're meant to be helpful, such as bots that can 0:06:19.000 --> 0:06:23.160 answer basic customer service questions for companies. You've probably encountered 0:06:23.160 --> 0:06:25.920 one of these where you were looking for some information 0:06:25.920 --> 0:06:29.320 about a particular product or service and then a little 0:06:29.360 --> 0:06:31.880 chat window pops up, and you get the feeling that 0:06:31.920 --> 0:06:35.400 the entity you're talking to is not exactly another human 0:06:35.400 --> 0:06:37.320 being on the other end. It may just be a bot. 0:06:38.360 --> 0:06:42.120 Sometimes that's fine. Sometimes it's more frustrating than helpful, because 0:06:42.440 --> 0:06:44.240 you find that you have to word things in a 0:06:44.320 --> 0:06:47.279 very particular way for the bot to comprehend what you mean, 0:06:47.480 --> 0:06:49.719 whereas a human would probably pick it up much faster. 0:06:50.480 --> 0:06:53.280 But you get the idea of why that was employed, right. 0:06:53.360 --> 0:06:56.880 That was meant to make things a little more smooth 0:06:57.080 --> 0:07:00.200 and to remove the necessity of putting a human being 0:07:00.240 --> 0:07:03.240 in charge of that at all hours of the day. 0:07:03.320 --> 0:07:06.359 You can also find these sorts of automated services on 0:07:06.440 --> 0:07:09.920 phone lines, including bots that call you, which is always fun. 0:07:10.400 --> 0:07:12.480 There's nothing like having a conversation with a bot for 0:07:12.560 --> 0:07:14.880 half a minute before you figure out something fishy is 0:07:14.920 --> 0:07:18.200 going on. But other bots are meant to serve the 0:07:18.240 --> 0:07:23.640 purposes of some third party, sometimes with malicious intent, such 0:07:23.680 --> 0:07:26.120 as convincing you to click on a link that leads 0:07:26.120 --> 0:07:30.480 to malware, and that's where we really run into obvious problems. 0:07:30.520 --> 0:07:34.080 Some of the benign ones can run into problems too often. 0:07:34.120 --> 0:07:38.640 There are unintended consequences if you're scouring the web for data. 0:07:38.760 --> 0:07:42.360 Data is valuable and sometimes people will want to get 0:07:42.440 --> 0:07:45.840 hold of it for bad reasons, even if the initial 0:07:46.000 --> 0:07:50.080 approach wasn't to do anything nefarious. Now, some hackers have 0:07:50.320 --> 0:07:53.160 used bots to flood a platform with complaints in an 0:07:53.160 --> 0:07:55.760 effort to silence people that the hackers do not like. 0:07:56.200 --> 0:08:00.240 So let's say there's this jerk face hacker who thinks 0:08:00.280 --> 0:08:03.640 a Facebook page devoted to promoting women in STEM education 0:08:03.680 --> 0:08:07.480 and careers is dumb. So this jerk face then creates 0:08:07.600 --> 0:08:11.840 or more likely purchases bots to flood Facebook with complaint 0:08:11.920 --> 0:08:15.960 reports about that specific page in an attempt to get 0:08:16.000 --> 0:08:19.280 Facebook to shut the page down. Now that's a pretty 0:08:19.280 --> 0:08:21.520 lousy thing to do. And to be clear, some of 0:08:21.520 --> 0:08:23.960 the jerk faces are aiming at pages that the average 0:08:23.960 --> 0:08:26.000 person would say is a bad one. It doesn't have 0:08:26.120 --> 0:08:29.600 to be something that like I feel strongly about and 0:08:29.600 --> 0:08:32.720 and in favor for. I think STEM education and careers 0:08:32.760 --> 0:08:35.240 for women is amazing, and I would be very upset 0:08:35.280 --> 0:08:37.120 to hear about a page that was shut down because 0:08:37.120 --> 0:08:39.600 of one of these attacks. On the other hand, let's 0:08:39.600 --> 0:08:42.000 say that there was a page that was promoting something 0:08:42.040 --> 0:08:44.480 I really do not like. Maybe it was a page 0:08:44.480 --> 0:08:48.679 that was promoting, uh, you know, racial discrimination. I would 0:08:48.760 --> 0:08:51.600 think that was terrible. If someone else were to take 0:08:51.679 --> 0:08:54.880 bots and direct them to that page in order to 0:08:54.920 --> 0:08:56.960 shut it down, I would also think that that's not 0:08:57.080 --> 0:09:01.760 so great. I don't think that a page about racial 0:09:01.800 --> 0:09:05.960 discrimination should be promoted or exist on Facebook. I don't 0:09:06.000 --> 0:09:08.440 think that's appropriate. But at the same time, I don't 0:09:08.440 --> 0:09:14.240 think it's appropriate to use automated systems to bring that down. 0:09:14.559 --> 0:09:18.960 I would rather see an actual ground swell of human 0:09:19.120 --> 0:09:24.360 support for that, not to you know, boost it with 0:09:24.400 --> 0:09:27.839 a bunch of automated scripts. I don't want to give 0:09:27.840 --> 0:09:31.200 the indication that the only people who ever use bots 0:09:31.240 --> 0:09:35.120 are those who want to silence vulnerable or underrepresented populations. 0:09:35.160 --> 0:09:37.520 There are some who use them to attempt to silence 0:09:37.600 --> 0:09:41.120 voices of hate. In either case, it's dirty pool. I 0:09:41.160 --> 0:09:45.760 don't think it's really a legitimate strategy. Uh, it ends 0:09:45.840 --> 0:09:48.840 up hurting everyone in the long run to use bots 0:09:48.840 --> 0:09:52.320 in that specific way. Butts in general, I'm not against. 0:09:53.080 --> 0:09:56.400 I do think there are times when they are incredibly useful, 0:09:56.400 --> 0:09:59.960 but to use them specifically to fool people into think 0:10:00.040 --> 0:10:03.200 ging their actual human beings in order to achieve an 0:10:03.280 --> 0:10:07.600 ulterior motive that sets me on edge that I can't 0:10:07.640 --> 0:10:10.200 really see an upside to that. I can definitely see 0:10:10.240 --> 0:10:13.600 it from the side of customer service or answering general questions, 0:10:13.920 --> 0:10:18.080 maybe even just trying to funnel out people who have 0:10:18.200 --> 0:10:21.640 a very simple issue to resolve versus those who need 0:10:22.000 --> 0:10:25.880 more attention, whereas you know those people would get directed 0:10:25.920 --> 0:10:27.800 towards a pathway that would lead to speaking to an 0:10:27.800 --> 0:10:31.760 actual human being. I get it from that perspective. Now, 0:10:31.760 --> 0:10:34.679 in a recent episode, I explained in brief what the 0:10:34.679 --> 0:10:37.720 Turing test was, or at least how we interpret it. 0:10:37.840 --> 0:10:40.520 The Turing test is sort of the inspiration for the 0:10:40.600 --> 0:10:44.400 Void comp test and Blade Runner. Alan Turing, one of 0:10:44.400 --> 0:10:47.960 the fathers of computer science, proposed the test back in 0:10:48.080 --> 0:10:51.560 nineteen fifty and in the actual thought experiment that he 0:10:51.600 --> 0:10:55.040 was proposing, it was a variation on a parlor game 0:10:55.080 --> 0:10:58.400 called the imitation game. Now, the imitation game is one 0:10:58.440 --> 0:11:01.600 where you have an interrogator that's player, and the player 0:11:01.720 --> 0:11:05.199 is presented with two subjects, neither of whom the interrogator 0:11:05.240 --> 0:11:08.600 can see or talk to directly. One of the two 0:11:08.679 --> 0:11:13.240 subjects is a woman, the other is a man. Both 0:11:13.400 --> 0:11:15.960 of the subjects can communicate with the interrogator in a 0:11:16.000 --> 0:11:18.600 way that does not require face to face contact or 0:11:18.720 --> 0:11:21.520 voice or anything like that. Typically it would be through 0:11:21.600 --> 0:11:25.160 something like typewritten letters, because that would help disguise handwriting 0:11:25.200 --> 0:11:28.160 as well. And the two subjects have the same task. 0:11:28.559 --> 0:11:32.400 They have to convince the interrogator that they are female. 0:11:32.720 --> 0:11:34.920 So the woman will be telling the truth, the man 0:11:35.080 --> 0:11:38.520 will be lying, and it's the interrogator's job to figure 0:11:38.559 --> 0:11:41.360 out who is imitating a woman and who actually is 0:11:41.440 --> 0:11:46.480 a woman. Touring then suggested taking this game a step 0:11:46.520 --> 0:11:51.040 further by replacing the male subject in this thought experiment 0:11:51.240 --> 0:11:54.920 with a computer. The computer would also attempt to convince 0:11:54.920 --> 0:11:58.480 the interrogator that the computer was in fact a woman. 0:11:59.520 --> 0:12:02.520 Now would the interrogator be able to detect the computer's 0:12:02.640 --> 0:12:06.760 ruse if not? Touring suggested that this would indicate some 0:12:06.840 --> 0:12:10.880 form of intelligence, though not necessarily human intelligence. But you 0:12:10.920 --> 0:12:16.720 could say the machine is capable of fooling a human being, 0:12:16.920 --> 0:12:20.520 of of practicing deception, which I think most of us 0:12:20.559 --> 0:12:24.280 would argue. The ability to practice deception does indicate at 0:12:24.360 --> 0:12:28.520 least some form of intelligence. Maybe not the type of 0:12:28.559 --> 0:12:31.320 intelligence that's gonna go out and teach a class on 0:12:31.440 --> 0:12:35.439 quantum mechanics, but the type of intelligence that does understand 0:12:36.240 --> 0:12:39.719 the concept of manipulation at least or at least is 0:12:39.760 --> 0:12:43.600 able to employ the concept of manipulation, if not understand 0:12:43.679 --> 0:12:47.640 it from a truly cognitive point of view. Now, the 0:12:47.679 --> 0:12:50.959 other variations and refinements to the Touring test followed after 0:12:51.080 --> 0:12:55.000 Touring's death in nineteen fifty four, and Touring's life was 0:12:55.200 --> 0:12:58.720 very tragic. We've done an episode on Alan Touring, so 0:12:58.720 --> 0:13:00.280 if you want to go back and find that in 0:13:00.320 --> 0:13:03.880 our archives, you can learn all about his his death 0:13:03.960 --> 0:13:07.320 and why some people rule it a suicide. I think 0:13:07.320 --> 0:13:10.359 most people do, and some people say it was accidental. 0:13:11.520 --> 0:13:15.880 But it is an interesting and tragic tale. Today, the 0:13:15.920 --> 0:13:18.480 general interpretation of the Turing test is that if a 0:13:18.520 --> 0:13:22.040 certain threshold is met, such as a greater than thirty 0:13:22.120 --> 0:13:25.959 percent success rate of a computer convincing interrogators that's actually 0:13:25.960 --> 0:13:30.120 a human, it has passed the Turing test. So, in 0:13:30.200 --> 0:13:33.160 other words, if you're an interrogator and you've got a 0:13:33.240 --> 0:13:35.840 computer terminal in front of you, and you're typing messages 0:13:36.400 --> 0:13:40.160 and the the response is coming back to you. And 0:13:40.320 --> 0:13:42.800 if more than thirty percent of the time you cannot 0:13:42.880 --> 0:13:45.280 tell if that actually is a computer or a person, 0:13:46.080 --> 0:13:49.320 maybe you misidentified as a person more than thirty percent 0:13:49.360 --> 0:13:52.320 of the time, and it's actually the computer. That computer 0:13:52.400 --> 0:13:54.360 is said to pass the Turing test, and that it 0:13:54.480 --> 0:13:57.160 is capable of fooling you into thinking it's an actual 0:13:57.240 --> 0:13:59.640 human being. Now, there was a case in two thousand 0:13:59.679 --> 0:14:03.080 and four in which a chat bought called Eugene seemed 0:14:03.240 --> 0:14:07.400 to accomplish this. Eugene's persona was that of a thirteen 0:14:07.480 --> 0:14:11.560 year old Ukrainian boy. Critics pointed out that Eugene's limitations 0:14:11.640 --> 0:14:15.160 as a non native English speaker with a limited knowledge 0:14:15.160 --> 0:14:17.520 of the world due to his age and the fact 0:14:17.520 --> 0:14:20.400 that he was from the Ukraine, meant that people were 0:14:20.480 --> 0:14:24.120 lowering their expectations on his performance when they were chatting 0:14:24.120 --> 0:14:27.120 with him over a computer. In other words, critics were 0:14:27.120 --> 0:14:33.600 saying that Eugene was gaming the system by making people think, oh, well, 0:14:34.000 --> 0:14:38.000 non native English speaker, so if the responses come back 0:14:38.040 --> 0:14:43.080 a little weird, that explains that. And being young means 0:14:43.120 --> 0:14:46.440 that they this kid doesn't have that much knowledge about 0:14:46.880 --> 0:14:49.440 a lot of things in the world, pop culture, politics, 0:14:49.680 --> 0:14:53.720 lots of stuff, so your expectations are set low, and 0:14:53.760 --> 0:14:56.600 then you just think, all right, well, are the messages 0:14:56.640 --> 0:14:59.320 I'm getting Are those in line with what I would 0:14:59.360 --> 0:15:02.840 expect a thirteen year old non native English speaker to 0:15:02.960 --> 0:15:06.240 say to me, or do they stand out as being artificial? 0:15:07.360 --> 0:15:10.040 And a lot of this ends up being deflection as well, 0:15:10.160 --> 0:15:14.720 where if you ask somebody a question and the computer 0:15:14.800 --> 0:15:17.400 program doesn't have a way of responding, it will try 0:15:17.400 --> 0:15:21.480 to deflect the question so that it doesn't indicate that 0:15:21.520 --> 0:15:27.360 and in fact is a computer program. Well, Eugene managed 0:15:27.400 --> 0:15:30.440 to to fool a lot of people, But again the 0:15:30.480 --> 0:15:35.880 critics were saying, well, Eugene was kind of an outlier 0:15:35.920 --> 0:15:39.160 in the sense that you didn't really think of Eugene 0:15:39.160 --> 0:15:44.520 as being a native speaker with a lifetime of experience 0:15:45.080 --> 0:15:48.560 where you could really quiz the the entity and find out, Okay, 0:15:48.640 --> 0:15:51.760 is this actually a person or is it a computer program. 0:15:51.960 --> 0:15:53.680 It's sort of beside the point. I'm not here to 0:15:53.760 --> 0:15:57.920 argue about whether or not machines possess intelligence if they 0:15:57.920 --> 0:16:01.480 passed the Turing test, because I did that recently already. Instead, 0:16:01.560 --> 0:16:04.080 let's focus on the flip side of the scenario, we're 0:16:04.160 --> 0:16:08.160 human at least I'm assuming you're a human. You might 0:16:08.160 --> 0:16:11.520 be a bot who subscribed to text stuff. Apparently of 0:16:11.520 --> 0:16:14.280 you out there are in that case. Thanks. I hope 0:16:14.320 --> 0:16:17.240 you like the show. But this is all for the 0:16:17.280 --> 0:16:19.680 humans here, this bit out here, so you bots out 0:16:19.720 --> 0:16:23.080 there can take a break. How can we humans tell 0:16:23.120 --> 0:16:25.320 if we're dealing with an actual person or if it 0:16:25.400 --> 0:16:29.240 is a bot. Well, one of the ways that we 0:16:29.360 --> 0:16:34.440 have created a means of separating bots from humans is capture. 0:16:35.320 --> 0:16:38.640 Capture is an acronym that stands for a completely automated 0:16:38.720 --> 0:16:43.200 public touring test to tell computers and humans apart. That 0:16:43.240 --> 0:16:44.960 pretty much sums it up when you break it down. 0:16:45.120 --> 0:16:48.560 It's completely automated, meaning there's no human oversight necessary for 0:16:48.600 --> 0:16:53.120 any given implementation of the technology. It's public. It's pretty 0:16:53.120 --> 0:16:55.880 self explanatory. It's a test that's out there in the public. 0:16:56.040 --> 0:16:58.080 I guess I explained it even though it wasn't necessary. 0:16:58.560 --> 0:17:01.160 That's my bad, y'all. Now, it's said to be a 0:17:01.160 --> 0:17:04.560 turing test because it's meant to detect human versus automated 0:17:04.640 --> 0:17:08.480 agents operating on a given web page. We talked about 0:17:08.600 --> 0:17:11.320 the touring test just now. But h so we're not 0:17:11.359 --> 0:17:13.520 gonna go over that again. But you know, again, it's 0:17:13.560 --> 0:17:17.560 just just this indicator. Is there something there that implicates 0:17:17.680 --> 0:17:20.200 this as being a computer agent not a human being? 0:17:20.640 --> 0:17:22.680 And if it is in fact a computer agent, then 0:17:22.720 --> 0:17:24.960 you have a gate up saying all right, you don't 0:17:25.000 --> 0:17:28.280 get to participate in this because it's not meant for you. 0:17:28.520 --> 0:17:31.680 When you have of your web traffic out there generated 0:17:31.680 --> 0:17:35.359 by bots and you're trying to collect meaningful data about 0:17:35.400 --> 0:17:38.879 real human being users, you have to have a way 0:17:39.359 --> 0:17:41.960 to separate the two. Right. So, if I'm a web 0:17:42.000 --> 0:17:45.400 administrator and let's say that I've got let's just say 0:17:45.400 --> 0:17:50.200 that I'm running a sweepstakes, have created an online entry form. 0:17:50.240 --> 0:17:53.080 I don't want someone flooding my sweepstakes with bots in 0:17:53.119 --> 0:17:56.040 an effort to try and game the system and win 0:17:56.359 --> 0:17:59.800 by submitting more entries than anybody else. I want to 0:17:59.800 --> 0:18:02.400 be able to control that. So I want to have 0:18:02.600 --> 0:18:04.800 some sort of element on there that can weed out 0:18:05.400 --> 0:18:11.040 the automated agents out there versus the actual human beings. Now, 0:18:11.080 --> 0:18:13.879 that last bit and capture to tell computers and humans 0:18:13.880 --> 0:18:16.119 apart is the key to all of this. Capture is 0:18:16.160 --> 0:18:18.720 a Guardian right, Like I was just saying, it's meant 0:18:18.760 --> 0:18:20.680 to keep people from just writing a script to fill 0:18:20.680 --> 0:18:24.080 out a form or make a comment on forums, really 0:18:24.600 --> 0:18:27.560 complete any interaction on the web in an automated way. 0:18:28.160 --> 0:18:33.480 As someone who creates content online and I get lots 0:18:33.480 --> 0:18:36.480 of comments on various platforms, I don't want a whole 0:18:36.520 --> 0:18:42.960 bunch of automated gobbledygook showing up under my various podcasts 0:18:42.960 --> 0:18:47.120 and videos because then I can't tell where the actual 0:18:47.600 --> 0:18:51.240 signal is. All I'm seeing is noise. So you want 0:18:51.280 --> 0:18:54.880 to have some way of controlling that, and you might 0:18:55.080 --> 0:18:57.439 use it to limit spam in the message board, or 0:18:57.480 --> 0:19:00.520 to stop people from abusing the format of an online mistakes, 0:19:00.640 --> 0:19:04.399 or or again to stop people from harassing others on 0:19:04.480 --> 0:19:07.760 social platforms. Now, the necessity for cap chub is due 0:19:07.840 --> 0:19:10.800 to a fundamental flaw of the Internet, and that flaw 0:19:10.960 --> 0:19:14.600 is this, it doesn't take very many people to make 0:19:14.760 --> 0:19:18.960 using the Internet a total drag. You don't want some 0:19:19.160 --> 0:19:21.840 jerk face to use a script to create thousands of 0:19:21.880 --> 0:19:25.200 email addresses from a web based email provider and then 0:19:25.359 --> 0:19:29.359 use those email addresses for spam purposes or for someone 0:19:29.400 --> 0:19:31.560 to gain the system. In other ways a single person 0:19:31.600 --> 0:19:35.840 has the potential to impact lots of other people. So 0:19:35.960 --> 0:19:39.320 everything's out of balance, and the force demands a Jedi 0:19:39.400 --> 0:19:44.080 to right the wrongs or something. Now, the ideal application 0:19:44.119 --> 0:19:48.760 of capture is some sort of test that is very 0:19:48.840 --> 0:19:53.560 easy for humans to complete, but very difficult for computers 0:19:53.640 --> 0:19:56.639 to complete. And that requires some creative thinking. So what 0:19:56.720 --> 0:19:59.119 are some things that people are really good at but 0:19:59.200 --> 0:20:03.800 computers are aren't so great at. Over time this changes. 0:20:04.160 --> 0:20:07.840 Computer programmers get better at designing software that allows computers 0:20:07.840 --> 0:20:11.359 to simulate more of what humans can do. And that's 0:20:11.400 --> 0:20:14.320 not a bad thing necessarily because it pushes our development 0:20:14.320 --> 0:20:18.880 of artificial intelligence. But for the purposes of gate keeping, 0:20:19.040 --> 0:20:22.240 it does make it more tricky. You've gotta figure out 0:20:22.240 --> 0:20:24.560 a new way to be able to prevent people from 0:20:24.560 --> 0:20:28.840 abusing the system. Now, the idea for capture came from 0:20:28.880 --> 0:20:32.040 a couple of different teams. One team was at Alta Vista, 0:20:32.280 --> 0:20:34.320 which started to work on ways to cut down on 0:20:34.320 --> 0:20:38.480 online abuse way back in the Ulta Vista team was 0:20:38.480 --> 0:20:41.439 trying to find a way to prevent bots or scripts 0:20:41.480 --> 0:20:44.000 from adding active u r l s to the search 0:20:44.080 --> 0:20:49.000 engine platform. Meanwhile, the other team was at Carnegie Mellon University, 0:20:49.160 --> 0:20:51.680 And actually this happened a couple of years after Alta 0:20:51.800 --> 0:20:55.560 Vista's work, and they included some researchers who were really 0:20:56.040 --> 0:20:58.360 eager to try and find a solution to this problem, 0:20:58.359 --> 0:21:03.600 and they included Louis on On, Manuel Bloom, Nicholas Hopper, 0:21:03.760 --> 0:21:07.520 and John Langford. It was the Carnegie Melon team that 0:21:07.600 --> 0:21:11.119 coined the term capture back in two thousand three, and 0:21:11.200 --> 0:21:14.320 it worked pretty well. Humans could get a capture right 0:21:14.400 --> 0:21:17.480 more often than not, and computers weren't nearly as good 0:21:17.520 --> 0:21:21.000 at it, at least not at first. Now we'll talk 0:21:21.040 --> 0:21:23.400 a lot about captures in just a minute and get 0:21:23.400 --> 0:21:26.399 into some more elements about telling the difference between butts 0:21:26.400 --> 0:21:29.359 and humans, but right now let's take a quick break 0:21:29.680 --> 0:21:41.280 to thank our sponsor. So with early capture implementations, things 0:21:41.520 --> 0:21:45.879 were pretty simple. The capture would take on a pretty 0:21:45.960 --> 0:21:49.119 universal form. You'd have a little box and inside that 0:21:49.160 --> 0:21:51.840 box you would see a couple of different words or 0:21:52.000 --> 0:21:57.200 collections of letters or other characters, often distorted in some way, 0:21:57.320 --> 0:21:59.960 and a little field beneath it telling you, hey, tie 0:22:00.080 --> 0:22:03.160 been what you see here? And it was your job, 0:22:03.200 --> 0:22:05.480 as a human being type person to type in the 0:22:05.520 --> 0:22:08.000 correct characters, and that would allow you to gain access 0:22:08.000 --> 0:22:10.800 to whatever it was that the capture was guarding. And 0:22:10.800 --> 0:22:13.040 the thought was that computers just weren't as good at 0:22:13.080 --> 0:22:17.280 recognizing those characters as humans are. That if you distort them, 0:22:17.480 --> 0:22:22.560 then the character recognition software couldn't put piece that altogether. 0:22:22.760 --> 0:22:27.760 The weird shapes would be too far outside the norm 0:22:28.160 --> 0:22:31.399 for the computer model. So if you had a one, 0:22:31.800 --> 0:22:35.320 but that number one, the numeral one, it was all 0:22:35.440 --> 0:22:39.159 wavy and staticky or something like you were, uh, you 0:22:39.240 --> 0:22:41.800 were breaking up the shape a bit by changing it. 0:22:42.240 --> 0:22:45.680 Computers can't really see that and conceptualize that's a one, 0:22:45.760 --> 0:22:48.480 or at least not in the early days, so it 0:22:48.480 --> 0:22:50.399 would just look like a weird squiggle to them, and 0:22:50.400 --> 0:22:53.360 they wouldn't be able to complete the capture. Whereas we 0:22:53.520 --> 0:22:56.000 human being type people, we'd look and think, that's the 0:22:56.080 --> 0:22:58.520 worst number one I've ever seen. Some kid must have 0:22:58.560 --> 0:23:01.240 drawn that, but we understand it is, we recognize it, 0:23:01.280 --> 0:23:03.879 so we would type that in. That was the basis 0:23:03.960 --> 0:23:07.720 for capture. Create a test that's relatively easy for humans, 0:23:08.240 --> 0:23:12.639 very difficult for computers. Now, not everyone was capable of 0:23:12.680 --> 0:23:17.440 seeing these captures. Clearly, some people have visual impair impairment, 0:23:17.680 --> 0:23:20.320 and so they need to have some other element to 0:23:20.480 --> 0:23:24.960 captures in order to be able to access that same content. 0:23:25.280 --> 0:23:29.280 So there are also audible captures, which is pretty important 0:23:29.400 --> 0:23:32.880 option to get around those visual impairments that some people have. 0:23:33.640 --> 0:23:41.040 And uh, you might get a distorted voice being repeating 0:23:41.080 --> 0:23:44.359 out the same sort of letters and numbers that you 0:23:44.359 --> 0:23:47.679 would encounter with a capture. There might also be some 0:23:47.760 --> 0:23:52.240 background noise that would include some other elements that would 0:23:52.280 --> 0:23:56.280 make it hard for a computer program to analyze the 0:23:56.280 --> 0:23:59.160 audio and figure out what was being said, but hopefully 0:23:59.400 --> 0:24:02.840