WEBVTT - The AI Arms Race Part One 0:00:00.200 --> 0:00:07.280 Brought to you by Toyota. Let's go places. Welcome to 0:00:07.440 --> 0:00:15.159 Forward Thinking. Hey there, and welcome to Forward Thinking, the 0:00:15.360 --> 0:00:18.160 podcast that looks at the future and says I'm a 0:00:18.160 --> 0:00:22.480 little intellectual someone who knows it all. I'm Jonathan Strickland, 0:00:22.600 --> 0:00:25.439 and I'm Joe McCormick, and our other host, Lauren voc 0:00:25.480 --> 0:00:28.280 obamb is not with us today. She is in the 0:00:28.320 --> 0:00:31.000 city of New York, within the state of New York. Yes, 0:00:31.080 --> 0:00:34.360 the city of Apples. You were sell what it is 0:00:34.400 --> 0:00:38.479 now according to the I T crowd, you were there 0:00:38.520 --> 0:00:42.080 not long ago. To Joe, yeah, that's right. Uh So 0:00:42.640 --> 0:00:45.559 my co hosts on the other podcast I do here 0:00:45.560 --> 0:00:48.760 Step to blow your mind, Robert Lamb, Christian Saga, and 0:00:48.800 --> 0:00:53.320 I did a live podcast at the Star Trek Mission 0:00:53.360 --> 0:00:59.560 New York convention that was held at the Javits Center. 0:00:59.800 --> 0:01:03.160 That fantastic. I'm so envious that you guys got to 0:01:03.240 --> 0:01:06.000 do that. And then of course Lauren is up there 0:01:06.040 --> 0:01:08.039 with a lot of our crew just kind of exploring 0:01:08.080 --> 0:01:11.319 New York and also taking part in a bunch of 0:01:11.360 --> 0:01:14.720 different activities, including a trivia night. She's having a blast. 0:01:15.160 --> 0:01:18.399 We are having a blast here talking about something that 0:01:18.400 --> 0:01:20.720 could potentially blast us off the face of the planet. 0:01:20.760 --> 0:01:22.360 I knew you were going to make a blast punt 0:01:22.680 --> 0:01:25.920 today to get there. Today we're gonna be talking about 0:01:25.959 --> 0:01:30.280 the potential scenario of an AI arms race. Yeah, and 0:01:30.280 --> 0:01:32.440 this is really going to be a two part series. 0:01:32.520 --> 0:01:35.880 So this first part we're talking mainly about just the 0:01:35.959 --> 0:01:40.080 concept of artificial intelligence, the various definitions that we have 0:01:40.200 --> 0:01:43.600 created to talk about AI. And in our next episode, 0:01:43.760 --> 0:01:46.440 we're really going to dive into some of these various 0:01:46.440 --> 0:01:49.800 scenarios people have proposed that we could see as a 0:01:49.840 --> 0:01:53.280 result of this emerging discipline. And it's funny to call 0:01:53.320 --> 0:01:56.040 an emerging discipline because really, if you look back at 0:01:56.040 --> 0:01:59.120 the history of AI, it's it's more than a century old. 0:01:59.800 --> 0:02:03.640 But I think most people would argue that what what 0:02:03.720 --> 0:02:06.920 the general public would consider artificial intelligence. That's the kind 0:02:06.920 --> 0:02:10.320 of stuff we're just now starting to see kind of 0:02:11.240 --> 0:02:14.000 kind of peek around the corner a little bit. So 0:02:14.960 --> 0:02:21.840 starting that all off, let's start defining artificial intelligence, because 0:02:21.880 --> 0:02:24.720 this is a term that gets used all the time. 0:02:25.240 --> 0:02:32.080 Sometimes it is misused, sometimes liberally misused or misrepresented, and uh, 0:02:32.680 --> 0:02:34.960 to me, it's sort of like virtual reality. It's one 0:02:35.000 --> 0:02:37.400 of those terms that people have heard it, and they 0:02:37.560 --> 0:02:41.080 generally know what it means, but sometimes they have a 0:02:41.080 --> 0:02:44.519 different vision than what was necessarily intended. It means a 0:02:44.560 --> 0:02:48.200 stock image of a sexy robot with a gun, yeah yeah, 0:02:48.720 --> 0:02:51.320 or sometimes a cyborg lady with an ear of corn. 0:02:52.160 --> 0:02:54.600 Ear of corn. You've never seen those, I'm not sure 0:02:55.120 --> 0:02:57.400 the ear of corn, the ear of corn, cyborg lady? 0:02:57.520 --> 0:03:00.480 I uh, I am ashamed to admit I have used 0:03:00.600 --> 0:03:03.320 that image in an in an article for how Stuff 0:03:03.320 --> 0:03:06.280 Works dot com. Uh tangent, And I know we're going 0:03:06.320 --> 0:03:10.640 on a tangent already. Dylan who works here, he actually 0:03:10.680 --> 0:03:13.200 he edits a lot of the podcasts, but he also 0:03:13.400 --> 0:03:15.680 does a lot of photo work, and he held a 0:03:15.720 --> 0:03:18.640 workshop here in the office to talk about the kinds 0:03:18.680 --> 0:03:22.079 of photos that you should and should not use and 0:03:22.160 --> 0:03:25.639 when he starts from stock photo websites in order to 0:03:25.720 --> 0:03:29.480 illustrate your work. And I had to cringe as some 0:03:29.600 --> 0:03:31.079 of the ones he was picking up as ones you 0:03:31.120 --> 0:03:33.760 should not use. I know for a fact I have 0:03:33.919 --> 0:03:36.080 used at least one of the images from that series, 0:03:36.120 --> 0:03:39.040 if not that specific one, then one from that same 0:03:39.120 --> 0:03:42.320 photography series that generated that particular image that he used 0:03:42.320 --> 0:03:44.680 as an example yeah. So if you're ever looking for 0:03:44.720 --> 0:03:47.240 a good belly laugh, go to a stock photo website 0:03:47.240 --> 0:03:52.400 and look up pictures of artificial intelligence or pictures of hacker. Yes, 0:03:52.680 --> 0:03:55.080 and so this, this actually tells me that we don't 0:03:55.080 --> 0:03:57.280 need to worry about artificial intelligence. We need to worry 0:03:57.320 --> 0:04:01.400 about just human intelligence first, at least in my case. Well, 0:04:01.480 --> 0:04:04.360 that's not necessarily all that off the money, according to 0:04:04.440 --> 0:04:08.400 some people. So yeah, so how would we define artificial intelligence? 0:04:08.720 --> 0:04:11.600 One really simple way to characterize it would be intelligence 0:04:11.680 --> 0:04:15.760 possessed or performed by machines. But this is already complicated, 0:04:15.880 --> 0:04:20.200 right because you used a word intelligence to define artificial intelligence. Yeah, 0:04:20.240 --> 0:04:24.000 and it's not that simple because it's there in the 0:04:24.120 --> 0:04:28.960 name artificial intelligence. Now, if it were really just about 0:04:29.120 --> 0:04:34.839 machines meeting some objectively defined criterion of intelligence, there would 0:04:34.880 --> 0:04:38.080 be nothing artificial about it, right. You might just refer 0:04:38.120 --> 0:04:43.040 to as machine intelligence then. Yeah. So that and machine intelligence, 0:04:43.040 --> 0:04:45.919 in fact, is a terminology that many experts writing this 0:04:46.000 --> 0:04:49.000 area seem to prefer. They like it better. But when 0:04:49.000 --> 0:04:52.279 some people use the term artificial intelligence, I think they're 0:04:52.320 --> 0:04:57.640 implying that machines are merely performing some kind of simulation. Generally, 0:04:57.680 --> 0:05:02.120 the simulation of the behavior you're of a human brain, right, 0:05:02.200 --> 0:05:06.280 like the science fiction depiction of artificial intelligence, is a 0:05:06.400 --> 0:05:10.839 robot or computer that thinks, at least on some level, 0:05:11.320 --> 0:05:14.160 similar to the way a human being thinks. Yeah, And 0:05:14.200 --> 0:05:16.679 this does come through in many of the public facing 0:05:16.720 --> 0:05:22.400 criteria for determining what constitutes a successful example of artificial intelligence, 0:05:22.640 --> 0:05:25.080 for example, the so called Turing tests, which we will 0:05:25.120 --> 0:05:28.360 refer to numerous times throughout this and we've talked about 0:05:28.360 --> 0:05:30.160 it on the show before, you've probably heard about it 0:05:30.480 --> 0:05:33.680 from us before. But essentially, a computer program, as it's 0:05:33.720 --> 0:05:37.680 understood today, a computer program is designed to chat back 0:05:37.720 --> 0:05:40.760 and forth with a human via text messages on a 0:05:40.760 --> 0:05:44.000 computer terminal, and ultimately its goal is to trick the 0:05:44.080 --> 0:05:47.120 human into thinking it's not a computer program but another 0:05:47.240 --> 0:05:50.320 live human, right. Essentially the idea of being that you 0:05:50.360 --> 0:05:54.200 are communicating with a series of entities, some of which 0:05:54.480 --> 0:05:56.880 are humans, some of which are computers, and you are 0:05:56.960 --> 0:06:00.560 unable to reliably tell the difference between the two. And 0:06:00.640 --> 0:06:03.800 so the test as conceived today arises out of a 0:06:03.839 --> 0:06:09.200 philosophical concern originally voiced by Alan Turing over essentially our 0:06:09.279 --> 0:06:12.919 inability to tell the difference between a machine that truly 0:06:13.000 --> 0:06:16.920 possesses intelligence and a machine that merely gives the appearance 0:06:16.960 --> 0:06:21.039 of intelligence. Right, we we give people the benefit of 0:06:21.040 --> 0:06:23.799 the doubt that they too are intelligent when we interact 0:06:23.800 --> 0:06:26.240 with them. Correct. Like Joe and I are sitting across 0:06:26.279 --> 0:06:28.599 from one another at a table. We both have microphones 0:06:28.640 --> 0:06:31.120 in front of us. We are interacting, We're having this conversation, 0:06:31.200 --> 0:06:34.560 and Jonathan has no idea that I'm not conscious. Right 0:06:35.120 --> 0:06:37.640 to me, you know you are behaving in a way 0:06:37.680 --> 0:06:41.000 that I would associate with being intelligent. Correct, Like you 0:06:41.040 --> 0:06:44.279 are hearing what I say, you are processing it, you 0:06:44.320 --> 0:06:48.040 are responding with your own thoughts. So to me, that 0:06:48.200 --> 0:06:50.720 is the appearance of intelligence. I know that I am 0:06:50.760 --> 0:06:54.919 intelligent because I have that personal experience. Therefore, I assume 0:06:55.120 --> 0:06:57.720 you also have a personal experience that is similar, if 0:06:57.760 --> 0:07:01.960 not identical, to my own, and you must possess intelligence 0:07:02.040 --> 0:07:04.520 if a computer is able to behave in such a 0:07:04.560 --> 0:07:10.040 way that it also appears to have these these faculties, 0:07:10.640 --> 0:07:12.440 even if you don't know what's going on in the 0:07:12.480 --> 0:07:15.880 back end, Touring would say you should at least extend 0:07:15.920 --> 0:07:18.680 the courtesy to say the machine is intelligent, because you 0:07:18.720 --> 0:07:21.320 would have done the same thing for another human being. Yes, 0:07:21.400 --> 0:07:24.400 but I think this all gives a very human gloss 0:07:24.480 --> 0:07:28.440 to the definition of intelligence. Like interpreted like this, intelligence 0:07:28.480 --> 0:07:31.440 sort of seems to mean that property of a conscious 0:07:31.480 --> 0:07:34.440 mind that is possessed by a normally functioning brain of 0:07:34.480 --> 0:07:38.280 an adult Homo sapiens. Yes. In other words, we have 0:07:38.600 --> 0:07:44.120 couched the word intelligence to be a distinctly human type 0:07:44.120 --> 0:07:47.400 of experience, right like that, Like there is if we're 0:07:47.440 --> 0:07:50.080 talking about artificial intelligence and we're framing it in the 0:07:50.160 --> 0:07:54.120 human experience than machine intelligence almost as meaningless. Yeah, and 0:07:54.200 --> 0:07:56.320 so I what I would be on the search for 0:07:56.640 --> 0:08:00.120 is a definition of intelligence that's more kind of an 0:08:00.160 --> 0:08:05.600 objective definition that could apply to any object, including humans 0:08:05.760 --> 0:08:09.880 or machines, depending on what what they could do. So, 0:08:10.120 --> 0:08:13.720 uh so, do we have anything that's like an objective, 0:08:13.800 --> 0:08:17.200 universal definition of intelligence that doesn't just mean acting like 0:08:17.280 --> 0:08:20.360 a human? Might? I came across one that I like 0:08:20.440 --> 0:08:22.680 a lot. I'll see what you think of this, Jonathan. 0:08:23.280 --> 0:08:27.360 So this comes from the systems theorist David Krakauer, and 0:08:27.640 --> 0:08:31.360 he's got definitions of intelligence and of stupidity that I 0:08:31.360 --> 0:08:35.280 think are actually very illuminating. So, according to krak our 0:08:35.800 --> 0:08:41.400 intelligence is finding very simple solutions to complex problems. And 0:08:41.400 --> 0:08:43.439 he gives an example I'd like to read from a 0:08:44.559 --> 0:08:48.160 piece that's published in Nautilus that that he did. So 0:08:48.320 --> 0:08:52.360 he says, quote, let's take a very simple example, and 0:08:52.400 --> 0:08:55.080 the example I'd like to choose is the Rubik's cube. 0:08:55.480 --> 0:08:57.720 If I give you a Rubik's cube and you try 0:08:57.760 --> 0:09:00.880 to solve it just randomly, so is imagining you're just 0:09:00.960 --> 0:09:04.360 turning in all directions until you hope, by random chance, 0:09:04.400 --> 0:09:09.360 you get solved. He continues, it will take many, many lifetimes. 0:09:09.400 --> 0:09:12.640 There are a billion billion solutions to the Rubik's cube. 0:09:12.880 --> 0:09:16.480 That's several lifetimes. That would be ignorance. That would be 0:09:16.480 --> 0:09:18.880 where you just don't know what to do, and so 0:09:18.960 --> 0:09:23.640 you perform essentially at a random level. Stupidity for the 0:09:23.679 --> 0:09:27.959 Rubik's cube is if you just consistently moved and manipulated 0:09:28.120 --> 0:09:32.320 one face. Maybe if I just rotate this one face forever, 0:09:32.480 --> 0:09:35.840 eventually the cube will be solved, and it will never 0:09:35.920 --> 0:09:39.760 be solved, even in infinite time, unlike the random case, 0:09:39.840 --> 0:09:44.480 which it will. Intelligence is a series of rules manipulations 0:09:44.559 --> 0:09:47.400 that will guarantee that you reach a solution in in 0:09:47.679 --> 0:09:51.080 steps or less. So my way of translating this is 0:09:51.120 --> 0:09:54.200 that he's saying intelligence is the ability to come up 0:09:54.200 --> 0:09:57.680 with strategies for improving one's chance of success at a 0:09:57.679 --> 0:10:01.360 given goal. So, in other word, uh, let's say that 0:10:01.400 --> 0:10:04.400 you to even generalize this, to go beyond like the 0:10:04.440 --> 0:10:07.160 Rubek's Cuban say, just a general problem. Let's say that 0:10:07.240 --> 0:10:10.360 you are faced with a problem and you try to 0:10:10.400 --> 0:10:12.600 come up with a solution. It turns out your solution 0:10:12.720 --> 0:10:15.560 doesn't work, so then you adjust, You try a different 0:10:15.600 --> 0:10:18.920 solution to see if that perhaps is more applicable, and 0:10:19.280 --> 0:10:21.600 eventually you hit on it. You get to a point 0:10:21.600 --> 0:10:23.760 it may not be the most elegant, it may not 0:10:23.840 --> 0:10:26.640 be the most efficient, but you finally find a way 0:10:26.679 --> 0:10:29.880 to solve that problem. Whereas the stupidity issue is you 0:10:29.920 --> 0:10:33.360 go with that first attempt, it doesn't work, and you 0:10:33.480 --> 0:10:35.959 just keep going with that same attempt, you do worse 0:10:36.000 --> 0:10:39.960 than random chance. Yes, it ends up almost falling into 0:10:40.000 --> 0:10:44.280 that same definition. As as the idea of insanity, you know, 0:10:44.400 --> 0:10:47.760 doing the same same thing over and over but expecting 0:10:47.800 --> 0:10:51.040 a different result. Uh, same kind of argument. There, the 0:10:51.080 --> 0:10:55.080 idea that that and you could see this easily with robotics, right, 0:10:55.160 --> 0:10:58.040 a robot that all it can do is pick up 0:10:58.080 --> 0:11:01.120 an object and turn it and set it down. And 0:11:01.160 --> 0:11:03.720 that's all I can do. Uh, it's never going to 0:11:03.760 --> 0:11:06.280 be able to solve a problem that doesn't involve picking 0:11:06.320 --> 0:11:08.400 up an object, turning it, and then setting it down. 0:11:08.480 --> 0:11:13.640 And we do see stupidity in robotics. Sometimes sometimes it's necessary, 0:11:13.760 --> 0:11:16.920 right because because the stupid solution, if you're talking about 0:11:16.920 --> 0:11:18.920 something that is always going to be the same, you 0:11:18.920 --> 0:11:23.520 are never going to have any variance from the approach. 0:11:24.000 --> 0:11:27.079 You want a stupid robot, right right? You just wanted 0:11:27.160 --> 0:11:31.120 to do this this three set are rather the series 0:11:31.120 --> 0:11:34.160 of three steps and that's all it does. Oh well, 0:11:34.200 --> 0:11:37.280 I mean if it's that would be if I'm hearing 0:11:37.320 --> 0:11:39.160 you correctly, that I mean, that's a different thing. Like 0:11:39.200 --> 0:11:43.080 a robot can not have any innovative solutions, but it 0:11:43.120 --> 0:11:46.960 can still do its job correctly because it's been programmed 0:11:46.960 --> 0:11:49.319 to do. You could even have a what I mean 0:11:49.440 --> 0:11:54.320 is a stupidly programmed robot hour since where where it 0:11:54.400 --> 0:11:59.400 performs worse than random performance at a task in like 0:11:59.440 --> 0:12:01.920 the example would be the robot that is trying to 0:12:02.080 --> 0:12:06.560 hand you something but instead stabs you. That would sort 0:12:06.559 --> 0:12:10.160 of be stupid robotics. Yes, yes, I would definitely feel 0:12:10.160 --> 0:12:14.440 stupid for using that robot. Uh. Now, moving beyond just 0:12:14.800 --> 0:12:19.120 trying to define what is intelligence, even beyond what is 0:12:19.240 --> 0:12:26.440 artificial intelligence, you start encountering subsets of artificial intelligence. And 0:12:26.480 --> 0:12:28.560 this is where a lot of other confusion comes in, 0:12:28.640 --> 0:12:31.880 and we're gonna be talking about some of these subsets 0:12:31.920 --> 0:12:35.640 in detail. One thing you may have encountered is the 0:12:35.679 --> 0:12:39.800 concept of weak AI versus strong AI. Yeah, and so 0:12:40.040 --> 0:12:43.840 this distinction is sometimes invoked as a philosophical one rather 0:12:43.880 --> 0:12:47.199 than a technological one, and for good reason. I mean, 0:12:47.240 --> 0:12:50.560 for one thing, we don't have anything that I think 0:12:50.760 --> 0:12:55.080 approaches the strong AI definition, and so we we can't 0:12:55.280 --> 0:13:00.200 say what the technological distinction is because we haven't achieved it, right, 0:13:00.240 --> 0:13:03.440 But let me explain. So, in the philosophical sense, week 0:13:03.480 --> 0:13:07.600 AI is a machine that can simulate or generate the 0:13:07.640 --> 0:13:12.880 outward appearance of intelligence or quote thinking, and strong AI 0:13:13.040 --> 0:13:17.599 is a machine that is quote actually thinking. So I 0:13:17.640 --> 0:13:20.800 would say this distinction is it's sometimes extended to sort 0:13:20.840 --> 0:13:24.160 of assert that strong AI is conscious, you know, like 0:13:24.200 --> 0:13:27.320 that there's something that's like to be that robot. And 0:13:27.360 --> 0:13:29.760 I do find the debate about whether machines can be 0:13:29.800 --> 0:13:32.800 conscious and whether machines can really think. I do find 0:13:32.840 --> 0:13:35.719 that interesting, but I also think that it's sort of 0:13:35.760 --> 0:13:40.559 a different question than the question of implementing AI technology 0:13:40.600 --> 0:13:44.480 in the real world, because you can't know whether another 0:13:44.559 --> 0:13:48.040 person is conscious. So it's just different than talking about 0:13:48.040 --> 0:13:51.719 strategies for how to create an intelligent machine, right, And 0:13:51.840 --> 0:13:54.520 I see your point, and I agree with it mostly. 0:13:54.559 --> 0:13:58.040 I think that that it's important only in the sense 0:13:58.120 --> 0:14:00.480 that you have a lot of people talking in terms 0:14:00.880 --> 0:14:03.640 of a I that, at least from the way they 0:14:03.640 --> 0:14:07.360 are talking about it, seems to indicate strong AI, this 0:14:07.480 --> 0:14:11.600 idea of machines that are quote unquote actually thinking. Uh. 0:14:11.600 --> 0:14:13.720 And there's a lot of debate on whether or not 0:14:13.880 --> 0:14:16.240 we will ever achieve that. There's some people who don't 0:14:16.240 --> 0:14:18.720 think it's achievable at all. There are others who say, well, 0:14:18.720 --> 0:14:22.880 that seems that seems like it's pretty egotistical to suggest 0:14:23.000 --> 0:14:26.160 that that only are gray matter in our heads is 0:14:26.200 --> 0:14:28.160 capable of this, and that we would never be able 0:14:28.240 --> 0:14:32.560 to achieve this in a machine environment. It may may 0:14:32.600 --> 0:14:35.560 be that it's impractical. Some would say, how would you 0:14:35.600 --> 0:14:39.040 know the difference? And in fact, there are lots of 0:14:39.080 --> 0:14:42.120 different philosophical you know, like thought experiment type things. We 0:14:42.160 --> 0:14:44.200 I think we've even mentioned one in a previous episode. 0:14:44.200 --> 0:14:46.120 Have we ever talked about the Chinese Room. Yeah, we 0:14:46.120 --> 0:14:48.640 did a whole episode on the Chinese Room experiments, all 0:14:48.680 --> 0:14:51.440 about whether machines can think right or are they just 0:14:51.640 --> 0:14:54.760 responding in a way that they've been programmed to respond 0:14:55.000 --> 0:14:57.560 so that it seems like they are thinking, that the 0:14:57.640 --> 0:15:00.800 illusion of thought is there, but there's no actual comprehension 0:15:00.840 --> 0:15:04.840 going on. But I agree that overall that conversation is 0:15:04.880 --> 0:15:08.960 not really relevant to most of what we're concentrating on, 0:15:09.360 --> 0:15:12.480 other than the fact that I think at least some 0:15:12.640 --> 0:15:17.200 of the people will chat about appear to have that 0:15:17.320 --> 0:15:20.960 kind of concept of AI in mind when they're talking 0:15:21.000 --> 0:15:22.960 about well, and to be fair to them, I think 0:15:23.000 --> 0:15:26.160 some people would make a not stupid case like it, 0:15:26.320 --> 0:15:30.760 like an interesting case that these these things we're talking about, consciousness, 0:15:30.840 --> 0:15:35.520 something deep about the very human, distinctly human things we 0:15:35.560 --> 0:15:39.240 think of as thinking and consciousness are actually in some 0:15:39.280 --> 0:15:41.720 way crucial to what we're about to talk about, which 0:15:41.720 --> 0:15:46.480 is general intelligence. But another way we could address the 0:15:46.520 --> 0:15:50.480 idea of week AI. This is sort of different, but 0:15:50.720 --> 0:15:54.240 the same terms are used sometimes UM also known as 0:15:54.360 --> 0:15:57.280 narrow AI. It's the AI that you're already familiar with 0:15:57.400 --> 0:16:02.520 any device or computer program that's mulates some mental ability 0:16:02.560 --> 0:16:06.600 of human beings. So you could argue that facial recognition 0:16:06.640 --> 0:16:10.240 in photographs on a computer is a very specific form 0:16:10.320 --> 0:16:13.720 of narrow AI, or those turing chat bots, you know, 0:16:13.800 --> 0:16:17.120 the inner the programs, or another form of narrow AI. 0:16:17.240 --> 0:16:21.560 We're already surrounded by examples of this kind of thing. Yeah. 0:16:21.560 --> 0:16:24.960 In fact, uh, we see advances in this all the 0:16:25.000 --> 0:16:29.880 time in interesting ways that maybe you weren't necessarily associating 0:16:29.880 --> 0:16:33.640 with artificial intelligence. One great example of this is the 0:16:33.720 --> 0:16:39.480 capture series, where you know captures that that that step 0:16:39.520 --> 0:16:41.120 you need to go through in order to prove you're 0:16:41.120 --> 0:16:45.480 not a robot to various online services. Typically, you know, 0:16:45.520 --> 0:16:48.080 early on it was just you read a word and 0:16:48.120 --> 0:16:50.360 then you had to type in what the word was, 0:16:50.960 --> 0:16:52.920 and or maybe it was just a string of letters 0:16:52.920 --> 0:16:56.200 and numbers, whatever it might be, and then the computer 0:16:56.320 --> 0:16:58.680 would verify that you are, in fact the human being 0:16:58.760 --> 0:17:01.080 and not some sort of autumn aided bought that's just 0:17:01.120 --> 0:17:04.920 trying to spam a service. Then there was a point 0:17:04.960 --> 0:17:10.399 where the object recognition or the text recognition capabilities of 0:17:10.480 --> 0:17:13.560 computers began to improve, and that is one aspect of 0:17:13.640 --> 0:17:17.560 artificial intelligence, the ability to recognize characters and understand what 0:17:17.640 --> 0:17:21.040 they correspond to. And so the captions began to get 0:17:21.040 --> 0:17:25.200 more difficult. They were distorted in weird ways or somewhat 0:17:25.400 --> 0:17:28.480 um covered up by other images in order to fool 0:17:28.960 --> 0:17:33.480 these image recognition programs, which then made the people designing 0:17:33.520 --> 0:17:36.280 the image recognition software go a step further so that 0:17:36.359 --> 0:17:39.680 they could recognize those examples too, And it became back 0:17:39.720 --> 0:17:44.160 and forth between security systems and artificial intelligence. And if 0:17:44.160 --> 0:17:48.359 you read up on the folks behind capture, they said, yeah, 0:17:48.440 --> 0:17:52.200 we were kind of thinking like this, this will help 0:17:52.480 --> 0:17:54.760 services figure out that, you know, the people who are 0:17:54.760 --> 0:17:56.800 trying to use their services are in fact human beings. 0:17:57.119 --> 0:18:00.320 But ultimately we wanted to drive the develop element of 0:18:00.400 --> 0:18:04.120 artificial intelligence. It seems so weird because we often think 0:18:04.119 --> 0:18:08.160 of AI as being this much more broad concept, nothing 0:18:08.280 --> 0:18:11.720 so specific as being able to recognize that this string 0:18:12.000 --> 0:18:15.960 of weird images is in fact a series of letters 0:18:16.000 --> 0:18:19.240 and numbers, but that is wrapped up in the concept 0:18:19.240 --> 0:18:22.440 of artificial intelligence and would be very much an example 0:18:22.440 --> 0:18:25.680 of narrow AI. My favorite example, are the ones that say, 0:18:25.680 --> 0:18:29.040 which of these pictures is not a clown? Right? Five 0:18:29.119 --> 0:18:33.280 clowns and then a pocket watch and a tiger right right, 0:18:33.359 --> 0:18:35.760 I've I've seen those two. Yeah. Where you you get 0:18:35.880 --> 0:18:38.639 like a bank of like nine images and about five 0:18:38.680 --> 0:18:41.320 of them tend to be whatever it is that you're 0:18:41.440 --> 0:18:44.680 supposed to be looking for, or or you get one 0:18:44.720 --> 0:18:47.800 that's really kind of makes you feel stupid, like it says, 0:18:48.640 --> 0:18:51.800 pick the images that have a river in them, and 0:18:51.840 --> 0:18:53.959 you're looking at one you see like a house on 0:18:54.000 --> 0:18:56.720 a shore and you're thinking, all right, is that a river? 0:18:56.840 --> 0:19:02.240 Is that a lake? And does this know the difference? Well, 0:19:02.320 --> 0:19:05.200 regular caption makes me feel stupid. Sometimes I can't see 0:19:05.240 --> 0:19:07.719 what those letters are. Yeah, I I sometimes will go there. 0:19:07.720 --> 0:19:09.919 You know, there's audio ones too, and in fact audio 0:19:10.080 --> 0:19:13.600 capture the same thing, like the idea of getting uh, 0:19:13.720 --> 0:19:17.320 the the voice recognition, the audio recognition that plays into 0:19:17.359 --> 0:19:21.640 artificial intelligence, natural language processing plays into artificial intelligence. All 0:19:21.680 --> 0:19:25.600 of these different things. So really, uh, you know, these 0:19:25.600 --> 0:19:29.520 are all aspects of narrow AI. These are all or 0:19:29.640 --> 0:19:33.760 or rather you know, manifestations of narrow AI. And some 0:19:33.880 --> 0:19:36.080 of them are are better than others. Some of them 0:19:36.119 --> 0:19:39.200 are more accurate than others. Um, some of them are 0:19:39.240 --> 0:19:44.520 further along than others, but it shows how diverse this 0:19:44.720 --> 0:19:47.200 discipline is. Yeah, but at the other end of the 0:19:47.240 --> 0:19:50.240 spectrum from narrow AI, you would get to the main 0:19:50.359 --> 0:19:53.000 concept that we're gonna be talking about in these two episodes, 0:19:53.359 --> 0:19:58.679 which is general AI. Artificial general intelligence often abbreviated a 0:19:58.760 --> 0:20:03.480 g I, and so under some usage, I think sometimes 0:20:03.520 --> 0:20:07.479 some people use the term strong AI to mean this, Right, 0:20:07.520 --> 0:20:10.160 they interchange the two terms to mean the same thing, 0:20:10.200 --> 0:20:13.200 but more accurately. Yeah, what we're trying to talk about 0:20:13.240 --> 0:20:17.000 is artificial general intelligence, and it's the property possessed by 0:20:17.000 --> 0:20:20.240 a machine that can apply intelligence to many or even 0:20:20.280 --> 0:20:24.119 all problems, rather than just one problem or some small 0:20:24.200 --> 0:20:27.840 number of them. And it's more like the diverse and 0:20:27.920 --> 0:20:33.480 adaptable problem solving engine within the complex animal brain, except 0:20:33.640 --> 0:20:38.119 it's going to presumably be free from biological and psychological 0:20:38.200 --> 0:20:41.879 limitations that we have in our human brains and probably 0:20:42.119 --> 0:20:45.520 be able to outperform the human brain most if not all, 0:20:45.640 --> 0:20:48.880 problem solving tasks. So this would be the difference between 0:20:49.560 --> 0:20:53.760 an AI implementation that can recognize a face and go 0:20:53.920 --> 0:20:56.359 beyond to one that not only can recognize a face, 0:20:56.480 --> 0:20:59.840 but also know the context of what's going on, also 0:21:00.119 --> 0:21:04.600 even respond to questions that are related to the the 0:21:04.800 --> 0:21:08.719 picture and ones that go beyond like tangential to the pictures. 0:21:08.760 --> 0:21:12.480 So you might say, uh, what is this a picture 0:21:12.480 --> 0:21:14.280 of it? It's a crowd of people, And the answer 0:21:14.359 --> 0:21:16.399 might be, well, that is as a group of human beings, 0:21:16.920 --> 0:21:19.920 all right, well how many how many of them are 0:21:20.040 --> 0:21:22.240 wearing blue T shirts? And it could tell you, and 0:21:22.240 --> 0:21:26.280 then you could start asking more questions that that start 0:21:26.320 --> 0:21:29.439 building off of that. It would continue to answer those, 0:21:29.960 --> 0:21:34.560 and it goes beyond just a single uh, a single 0:21:34.640 --> 0:21:38.440 task or a single um course of action, right. Like 0:21:40.119 --> 0:21:42.520 another great example is that a lot of the robots 0:21:42.560 --> 0:21:47.720 we depend upon today are functional because they focus on 0:21:47.720 --> 0:21:50.959 a very specific task. They don't they don't branch beyond it. 0:21:51.359 --> 0:21:54.439 And we've seen that the there are a lot of 0:21:54.520 --> 0:21:58.040 challenges to create robots that are able to tackle numerous 0:21:58.119 --> 0:22:03.399 challenges and uh, Thus we come to the conclusion that 0:22:03.560 --> 0:22:07.560 coming coming up with like an artificial general intelligence is 0:22:07.600 --> 0:22:11.959 going to be incredibly challenging just from the the baby 0:22:12.040 --> 0:22:15.840 steps that we've seen so far in that arena. And 0:22:15.840 --> 0:22:18.800 and even then like with the darker robotics challenge, which 0:22:18.800 --> 0:22:22.239 we will mention again later on. Uh. Even with the 0:22:22.320 --> 0:22:25.240 advancements that we've seen in that realm, we see the 0:22:25.280 --> 0:22:29.160 limitations that are there, and that's only a small, a tiny, 0:22:29.200 --> 0:22:36.480 tiny slice of everything. Whereas general intelligence we we tend 0:22:36.560 --> 0:22:40.800 to say could deal with any topic or any task. 0:22:41.920 --> 0:22:44.520 Anything humans can do, it should be able to do too, 0:22:44.600 --> 0:22:48.120 and probably some other stuff too. Yes, yes, so yeah. 0:22:48.160 --> 0:22:50.120 I think I'm going to suggest that it makes more 0:22:50.119 --> 0:22:52.960 sense for us to speak of narrow AI and general 0:22:53.080 --> 0:22:55.879 AI for the purpose of these episodes than we k 0:22:55.920 --> 0:22:58.159 AI and strong AI, just so we can avoid these 0:22:58.200 --> 0:23:01.480 side questions about you know, the nature of consciousness in 0:23:01.520 --> 0:23:05.320 the phenomenal logical nature of thinking and stuff. I agree, 0:23:05.520 --> 0:23:08.120 and uh. I also want to point out that general 0:23:08.920 --> 0:23:13.919 artificial intelligence can still refer to a machine that is 0:23:14.040 --> 0:23:18.760 dedicated to a specific set of tasks. It doesn't have 0:23:18.880 --> 0:23:21.639 to mean that you've got, you know, the deep thought 0:23:21.680 --> 0:23:24.280 computer where you can ask any question and it will 0:23:24.280 --> 0:23:27.480 give you the answer. It may be that we look 0:23:27.600 --> 0:23:31.439 at examples like a machine that could exist within the 0:23:31.480 --> 0:23:34.560 medical field is the example I picked, where it helps 0:23:34.640 --> 0:23:39.640 doctors diagnose and treat patients, and IBM s. Watson computer 0:23:40.080 --> 0:23:43.120 is meant to do those sort of things. But I'm 0:23:43.160 --> 0:23:45.840 talking about going a step beyond what Watson can do 0:23:46.400 --> 0:23:50.400