WEBVTT - Do Robots Write of Electric Sheep? 0:00:00.160 --> 0:00:07.400 Brought to you by Toyota. Let's go places. Welcome to 0:00:07.560 --> 0:00:15.280 Forward Thinking. Hey there, and welcome to Forward Thinking, the 0:00:15.280 --> 0:00:17.320 podcast that looks at the future and says it's like 0:00:17.360 --> 0:00:19.520 a nine and shining armor from a long time ago. 0:00:19.880 --> 0:00:23.200 I'm Johnvin Strickland and Lauren and I'm Joe McCormick. And 0:00:23.239 --> 0:00:26.720 today we want to talk more about about stories and storytelling. 0:00:26.760 --> 0:00:28.960 But you know, in our last podcast we talked about 0:00:29.040 --> 0:00:33.400 interactive storytelling and this idea collaboratively creating a story in 0:00:33.520 --> 0:00:37.599 some way or participating within a story in some way. Now, 0:00:37.640 --> 0:00:40.000 let's talk a little bit about, um, what the stories 0:00:40.040 --> 0:00:43.160 that we wish existed but don't like. Can you imagine 0:00:43.240 --> 0:00:45.760 some of the greatest authors of all time, who who 0:00:45.800 --> 0:00:49.680 maybe died before you felt like they had really produced 0:00:49.880 --> 0:00:52.480 everything they could have produced? Like, do you have any 0:00:52.479 --> 0:00:55.440 favorite authors that you wish had lived longer to tell 0:00:55.480 --> 0:00:59.200 at least one more amazing story? Oh of course. Well, 0:00:59.240 --> 0:01:02.360 actually what first popped into my mind was poets. I'd 0:01:02.400 --> 0:01:04.679 love to have more of Uncle Walton Anti m when 0:01:04.680 --> 0:01:08.440 I'm feeling when I'm feeling patriotic, you know, Walt Whitman, 0:01:08.480 --> 0:01:12.839 Emily Dickinson, I gotta feel the America, you know, coming 0:01:12.880 --> 0:01:16.040 up right bubbling up inside you. But they've been dead 0:01:16.080 --> 0:01:19.040 a long time, right, So, uh well, Lauren, are there 0:01:19.040 --> 0:01:20.840 any authors that you could think of off the top 0:01:20.880 --> 0:01:23.360 of your head where you're thinking, Man, I really wish 0:01:23.520 --> 0:01:27.600 I could have had one more book or story or 0:01:27.680 --> 0:01:30.800 play or poem from this person. I'm a completely terrible person. 0:01:30.840 --> 0:01:32.319 The first thing that just popped into my head was 0:01:32.360 --> 0:01:35.080 George RR. Martin, who is not actually dead as of 0:01:37.560 --> 0:01:40.480 So you want you want an insurance policy to guarantee 0:01:40.640 --> 0:01:43.560 that a Song of Ice and Fire is complete at 0:01:43.560 --> 0:01:46.959 some point, and you want it in George RR. Martin's voice. 0:01:47.000 --> 0:01:50.720 That's really important. Obviously someone else could pick it up, 0:01:50.720 --> 0:01:52.480 but that wouldn't be the same. Yeah, no, no, no 0:01:52.520 --> 0:01:55.040 one else would write. And I'm being actually very earnest 0:01:55.080 --> 0:01:57.840 here the kinds of descriptions of food and dresses into 0:01:58.080 --> 0:02:02.000 my political fantasy that he writes. Well, you guys have 0:02:02.120 --> 0:02:06.960 both made my choice of Shakespeare seemed pedestrian at this point. I'm, 0:02:07.000 --> 0:02:11.559 of course, I am a Shakespearean nut. I love Shakespeare's 0:02:11.600 --> 0:02:14.600 plays and I love his poetry as well. And uh 0:02:14.840 --> 0:02:17.760 we they're already some plays that may or may not 0:02:18.200 --> 0:02:21.200 belong to Shakespeare. We'll talk more about some of those 0:02:21.240 --> 0:02:23.720 in a little bit, but uh, you know, there were 0:02:23.720 --> 0:02:27.520 there's some lost plays that perhaps we might actually have 0:02:27.760 --> 0:02:30.440 if he had lived longer to get to the point 0:02:30.480 --> 0:02:33.160 where he was committing these so that they could be published. 0:02:33.200 --> 0:02:34.959 You know, in Shakespeare's day, he wasn't It wasn't what 0:02:35.080 --> 0:02:40.079 he was writing them down necessarily any anymore than too. Yeah, 0:02:40.080 --> 0:02:42.800 he would get the sides to the actors, but the 0:02:42.800 --> 0:02:46.520 whole purpose of the play was for performance, not for reading. 0:02:46.680 --> 0:02:49.320 So in Shakespeare's day, very few of his plays ever 0:02:49.360 --> 0:02:54.080 were published. It was only after his death when his contemporaries, Yeah, 0:02:54.160 --> 0:02:56.520 his contemporaries began to gather them all together. People who 0:02:56.520 --> 0:02:59.760 worked in the King's Men production company began to gather 0:02:59.800 --> 0:03:02.640 as stuff together. And then even then there were points 0:03:02.680 --> 0:03:06.680 where scholars have argued whether other plays should have been included. 0:03:06.720 --> 0:03:08.960 Perhaps there were plays that Shakespeare wrote when he wasn't 0:03:09.360 --> 0:03:12.120 writing for the King's Men that weren't included. And if 0:03:12.120 --> 0:03:14.760 you have been attributed to to other authors. Sure now 0:03:15.320 --> 0:03:18.280 people think Shakespeare didn't write any of those plays, They're wrong. 0:03:18.560 --> 0:03:21.880 So I would love it if if Shakespeare could turn 0:03:21.919 --> 0:03:24.160 out a few books. I would love it if Dickens 0:03:24.440 --> 0:03:28.000 could finish the Mystery of Edwin Drewd because there's nothing 0:03:28.040 --> 0:03:31.359 more irritating than a mystery that ends halfway through. He 0:03:31.560 --> 0:03:35.480 killed him? Or did he even die? Oh? Yeah, he 0:03:35.560 --> 0:03:37.840 may have just disappeared. There's a there's a science fiction 0:03:37.840 --> 0:03:39.960 book that C. S. Lewis never finished before his death 0:03:40.000 --> 0:03:41.880 that I would very much like to read the ending too. 0:03:41.920 --> 0:03:44.280 I would have liked to have read The Simarillian as 0:03:44.320 --> 0:03:47.240 if Tolkien had actually written it in a form that 0:03:47.320 --> 0:03:49.560 was meant to be read by a human being, as 0:03:49.600 --> 0:03:52.120 opposed to a collection of things that his son pulled 0:03:52.160 --> 0:03:55.440 together from various works and notes and then put into 0:03:55.840 --> 0:03:58.920 some kind of bizarre encyclopedia. Yeah. I mean, don't get 0:03:58.960 --> 0:04:02.080 me wrong, I nuinely love The Simarillian, but I wonder 0:04:02.120 --> 0:04:04.720 what it would have been like had Tolkien actually himself 0:04:05.080 --> 0:04:07.840 put it all out together. So, I mean there are 0:04:07.880 --> 0:04:12.760 lots of different examples, obviously, uh Hunter S. Thompson, Kurt. 0:04:13.320 --> 0:04:14.840 I mean, there's some authors out there I would love 0:04:14.880 --> 0:04:17.080 to have been able to read some more stuff. And 0:04:17.160 --> 0:04:20.760 Charles Dickens, Oh, I already mentioned him. My anglophile traits 0:04:20.800 --> 0:04:23.920 are starting to show, and there is always fan fiction. 0:04:24.040 --> 0:04:27.520 I mean, you know, some some fan authors can write 0:04:27.640 --> 0:04:31.400 very convincingly in an author's voice. Well, what if we 0:04:31.440 --> 0:04:36.559 could create a fan fiction style author that wasn't really 0:04:36.600 --> 0:04:39.640 a person. You know. Before I get into that, we're 0:04:39.640 --> 0:04:41.640 gonna have to talk a little bit about some some 0:04:41.800 --> 0:04:45.840 terms here, like artificial intelligence. Artificial intelligence, What does that 0:04:45.920 --> 0:04:49.560 have to do with literature. Well, we're gonna get there. 0:04:49.160 --> 0:04:53.720 We'll take some steps first. So artificial intelligence or machine intelligence, 0:04:53.720 --> 0:04:56.280 this is a concept that's been around for more than 0:04:56.400 --> 0:04:59.240 half a century at this point. Uh, And it was 0:04:59.279 --> 0:05:02.000 something that was kind of of played with early early 0:05:02.040 --> 0:05:05.040 early days in in the earliest days of computer science, 0:05:05.080 --> 0:05:08.640 in fact, before we even called it computer science. One 0:05:08.680 --> 0:05:12.400 of the fathers of the concept of artificial intelligence, you 0:05:12.440 --> 0:05:16.279 could argue, is Alan Turing. Now Touring was a computer 0:05:16.400 --> 0:05:19.320 scientist before there were was a word for such a thing, 0:05:19.360 --> 0:05:22.320 who worked for the British government. He was working within 0:05:22.360 --> 0:05:25.839 the War Department to help decode messages that were sent 0:05:26.080 --> 0:05:29.680 by Germans using the Enigma machine you guys familiar with. 0:05:30.960 --> 0:05:33.159 He was one of he was on the team, and yes, 0:05:33.200 --> 0:05:35.560 he was highly responsible for cracking the code of the 0:05:35.640 --> 0:05:40.279 Enigma machine. The Enigma machine was a physical mechanical device 0:05:40.320 --> 0:05:43.000 that the Germans were using that you would set to 0:05:43.040 --> 0:05:46.280 a particular setting, had certain reels, and then what you 0:05:46.480 --> 0:05:48.159 do is I had like a typewriter and as you 0:05:48.160 --> 0:05:52.120 would type, the reels would channel a an electric signal 0:05:52.240 --> 0:05:55.800 through so that a particular lightbulb would light up on 0:05:55.839 --> 0:05:58.320 the other side. But it was designed in such a 0:05:58.320 --> 0:06:00.800 way so that would encode eat letters. So if you 0:06:00.880 --> 0:06:03.040 typed a G, one thing you could be sure of 0:06:03.360 --> 0:06:05.680 is that the one bulb that's not gonna light up 0:06:05.680 --> 0:06:07.880 on that board is G. And in fact, that was 0:06:07.920 --> 0:06:10.040 one of the few things that Touring was able to 0:06:10.080 --> 0:06:12.919 figure out that that broke the code was knowing that 0:06:12.960 --> 0:06:15.160 whatever the letter was that you were looking at at 0:06:15.160 --> 0:06:19.520 an encoded message, it was definitely not exactly it'll and 0:06:19.560 --> 0:06:22.440 you would think eliminating one choice would not give you 0:06:22.480 --> 0:06:24.960 that big of an advantage, but that was kind of 0:06:24.960 --> 0:06:28.120 the crack that really helped break it open. Okay, So 0:06:28.200 --> 0:06:31.480 what did Alan Turing have to do with the idea 0:06:31.480 --> 0:06:35.280 of artificial intelligence? So he was very much interested in 0:06:35.320 --> 0:06:38.840 this idea of what is what's the capability of the 0:06:38.880 --> 0:06:41.839 machine world and are there limits? Can it go as 0:06:41.880 --> 0:06:45.640 far as humans can go? Can machines think? Can machines think? 0:06:45.800 --> 0:06:48.200 He thought about that in such a way It's a 0:06:48.200 --> 0:06:52.240 little different than the way you might initially imagine a 0:06:52.320 --> 0:06:55.479 machine thinking. Uh, Touring said, it wasn't so much. It 0:06:55.520 --> 0:06:58.159 wasn't really important if machines thought the same way that 0:06:58.200 --> 0:07:03.000 people do. In other words, what was necessary from Touring's 0:07:03.000 --> 0:07:05.160 point of view, was that the output of a machine 0:07:05.560 --> 0:07:09.080 would be indistinguishable to what a human would be capable 0:07:09.080 --> 0:07:13.880 of doing. But the process doesn't. Yeah. So in other words, 0:07:13.920 --> 0:07:16.840 if the machine was able to produce something that you 0:07:17.000 --> 0:07:20.400 seem the same as the way a human would produce it, 0:07:20.400 --> 0:07:22.400 it didn't matter how the machine did it. It just 0:07:22.520 --> 0:07:26.000 meant that that machine possessed some form of intelligence. Uh, 0:07:26.040 --> 0:07:28.560 And I'll explain that in a second to In nineteen 0:07:28.640 --> 0:07:32.720 fifty he published a paper Computer Machinery and Intelligence and 0:07:32.800 --> 0:07:35.320 propose the idea of the touring test. Now, in general, today, 0:07:35.360 --> 0:07:38.480 the way we talk about the touring test is, uh. You. 0:07:39.200 --> 0:07:42.200 A simple version would be you have a computer in 0:07:42.240 --> 0:07:44.440 front of your computer terminal. All you get is text, 0:07:44.960 --> 0:07:48.800 and your text is going to someone that's in another room, 0:07:48.840 --> 0:07:50.480 and that might be a person or it might be 0:07:50.520 --> 0:07:53.920 a machine. So you're you're like in a chat set up. Yeah, yeah, 0:07:54.000 --> 0:07:56.160 and you cannot see the other person at all all 0:07:56.280 --> 0:07:58.320 or the machine at all. All you see is whatever 0:07:58.360 --> 0:08:00.440 shows up on your screen and you type in a 0:08:00.480 --> 0:08:04.760 message and a message returns back to you that's generated 0:08:04.760 --> 0:08:07.080 either by a person or a machine. The Touring test 0:08:07.120 --> 0:08:12.000 said that if you were unable to determine to a 0:08:12.040 --> 0:08:16.240 certain degree of certainty whether or not the the the 0:08:16.440 --> 0:08:18.360 thing on the other end was a person or a machine, 0:08:18.720 --> 0:08:21.320 that machine was said to pass the Touring test. So 0:08:21.600 --> 0:08:25.760 if it could imitate human speech through text, yes, So 0:08:25.840 --> 0:08:28.960 if the machine was able to carry on a conversation 0:08:29.000 --> 0:08:30.640 in such a way that it would convince you that 0:08:30.640 --> 0:08:33.200 it's another person, then it could pass the Touring test. 0:08:33.240 --> 0:08:35.640 It would at least seem to possess intelligence. And here's 0:08:35.679 --> 0:08:37.280 how Touring kind of thought about this. It's kind of 0:08:37.280 --> 0:08:42.240 a philosophical way of thinking about intelligence. So I, Jonathan Strickland, 0:08:42.280 --> 0:08:46.120 I know that I possess at least some level of 0:08:46.160 --> 0:08:50.080 intelligence because I have that personal experience. I know what 0:08:50.120 --> 0:08:53.920 it means in an abstract way to myself. When I 0:08:53.960 --> 0:08:58.080 talked to you, Lauren, and you start to display several 0:08:58.120 --> 0:09:00.880 of the same things I associate with an intelligence. I 0:09:00.960 --> 0:09:05.840 then assume you too, possess intelligence of some level, and 0:09:05.960 --> 0:09:09.600 that that is just something that I am granting to Lauren. 0:09:09.679 --> 0:09:13.199 From my perspective, because I cannot inhabit Lauren, I can't 0:09:13.720 --> 0:09:17.080 know what her experience really is. I can only base 0:09:17.120 --> 0:09:21.320 that upon the observations I make of her behavior. Touring says, well, 0:09:21.360 --> 0:09:24.240 if we do that, if you assume that every person 0:09:24.280 --> 0:09:27.920 you encounter has some level of intelligence, why would you 0:09:27.960 --> 0:09:31.840 not offer a machine the same courtesy If it seemed 0:09:31.840 --> 0:09:35.000 to display the same sort of behaviors, You might as 0:09:35.000 --> 0:09:37.120 well say it has intelligence, because there's no way that 0:09:37.160 --> 0:09:39.520 you can inhabit that machine, just as there's no way 0:09:39.559 --> 0:09:42.760 you can inhabit another human being. And in fact, saying 0:09:42.840 --> 0:09:46.040 that we know the machine is different because we programmed 0:09:46.080 --> 0:09:49.120 it might be an example of the genetic fallacy, right 0:09:49.240 --> 0:09:52.439 that the idea that because you know where a phenomenon 0:09:52.520 --> 0:09:57.360 came from, that proves it's not genuine. Sure. Yeah, and 0:09:57.600 --> 0:10:00.600 so when we get to the actual touring test. We've 0:10:00.640 --> 0:10:06.320 had plenty of people design various kinds of of software. 0:10:06.400 --> 0:10:09.200 Usually they're called chatbots. These are just programs that are 0:10:09.240 --> 0:10:12.800 designed to interact with people and respond in a conversational 0:10:12.840 --> 0:10:17.160 way through some context. And uh, and some of them 0:10:17.320 --> 0:10:21.720 have been pretty successful. Surely they could never really trick somebody, right, 0:10:21.760 --> 0:10:24.760 I mean, I've talked a smarter child on instant Messenger. 0:10:25.360 --> 0:10:28.560 I don't think that's a person. It all depends on again, 0:10:28.640 --> 0:10:31.880 the context and how well designed the program was. So 0:10:31.920 --> 0:10:35.520 for example, one of the earliest that was designed was Eliza, 0:10:35.679 --> 0:10:40.360 which was created by Joseph Wisenbaum, and Eliza was effective 0:10:40.640 --> 0:10:42.920 within a certain context, and that it was it was 0:10:43.040 --> 0:10:46.760 presenting a point of view of almost kind of a 0:10:46.880 --> 0:10:50.479 naive person, someone who has very little real world experience, 0:10:50.920 --> 0:10:55.200 and so as long as the real human being had 0:10:55.240 --> 0:11:01.080 that same kind of perspective, it was fairly effective. There 0:11:01.120 --> 0:11:04.920 was another one called Perry p a r R Y 0:11:04.960 --> 0:11:09.200 that was made by Kenneth Colby, and that one simulated 0:11:09.440 --> 0:11:14.120 a paranoid schizophrenic responding to questions. And what they did 0:11:14.160 --> 0:11:17.160 was they ran a whole bunch of conversations between an 0:11:17.200 --> 0:11:21.880 interviewer and actual patients who suffered from paranoid schizophrenia and 0:11:22.280 --> 0:11:26.000 conversations with Perry, and then printed out all those conversations 0:11:26.320 --> 0:11:28.720 handed that to a panel of judges who were made 0:11:28.760 --> 0:11:31.439 up of psychologists, and they had a success rate of 0:11:31.480 --> 0:11:34.240 forty eight percent of saying which one was real in 0:11:34.280 --> 0:11:36.959 which one. Now, in that case, you're talking about a 0:11:37.080 --> 0:11:40.560 specific context that has its own limiting factors, right sure, yeah, yeah, 0:11:40.559 --> 0:11:43.840 you're you're you're talking about two different, um, you know, 0:11:44.000 --> 0:11:46.440 hypotheticals of a human person that aren't the kind of 0:11:46.480 --> 0:11:50.800 conversation that you would normally have straight And in the 0:11:50.800 --> 0:11:53.560 case of Perry, you're talking about a conversation that was 0:11:53.600 --> 0:11:57.040 not conducted by the actual panel of judges. Right, So 0:11:57.400 --> 0:12:00.160 their experience is very different from someone who say, it's 0:12:00.160 --> 0:12:04.000 down at a terminal and it's actually having the conversational moment. 0:12:04.320 --> 0:12:07.880 They're reviewing something that's already happened. That's different too, I 0:12:07.880 --> 0:12:10.200 mean that experience is different. But there are other examples too. 0:12:10.280 --> 0:12:14.320 There's the artificial linguistic Internet computer entity also known as Alice, 0:12:15.120 --> 0:12:18.120 and there's a one that was made by Rollo Carpenter 0:12:18.240 --> 0:12:22.200 called Jabberwocky. Was brilliant, the slightly topes uh that one 0:12:22.240 --> 0:12:24.760 was meant to simulate human chat in a very kind 0:12:24.760 --> 0:12:27.640 of humorous way. I actually tried this one out before 0:12:27.679 --> 0:12:29.720 we came to the podcast. I wish I had been 0:12:29.760 --> 0:12:32.520 able to print out our conversation because I mean, you 0:12:32.520 --> 0:12:35.000 could tell that it was or I could tell it 0:12:35.040 --> 0:12:37.240 was a chat bot. But for one thing, I already knew, 0:12:37.320 --> 0:12:40.560 So that that's problematic, right, I mean I already know 0:12:40.840 --> 0:12:44.720 going into it, So yeah, very scientific, right. I would 0:12:44.760 --> 0:12:46.960 need to have like a double blind test if I 0:12:46.960 --> 0:12:50.000 really wanted to do this properly. But also there were 0:12:50.000 --> 0:12:53.040 other giveaways, like you would it would ask you a 0:12:53.160 --> 0:12:55.640 question and I would answer because it was being snarky. 0:12:55.679 --> 0:12:57.680 I was being snarky. I answered in a snarky way, 0:12:57.720 --> 0:13:00.440 and then it didn't know how to deal with that, 0:13:00.480 --> 0:13:02.800 like it didn't have enough keywords to work off of, 0:13:03.200 --> 0:13:06.720 so it would go with a uh, like a stock 0:13:06.840 --> 0:13:09.400 response that was just a generic response. And in fact, 0:13:09.480 --> 0:13:11.839 that's the way a lot of these chatbots work is 0:13:11.880 --> 0:13:14.440 that they search for keywords and the things that the 0:13:14.480 --> 0:13:18.480 person the interviewer is typing in, and then it generates 0:13:18.480 --> 0:13:22.440 a response based upon those keywords. And if none of 0:13:22.440 --> 0:13:25.680 the keywords that it normally quote unquote knows as in 0:13:25.760 --> 0:13:28.680 it keywords that are in its database, then it will 0:13:28.760 --> 0:13:33.119 generate some other form of response, either a generic response, 0:13:33.400 --> 0:13:36.080 or it will repeat something that it had already said 0:13:36.080 --> 0:13:39.680 previously in the conversation. So what are some of the 0:13:39.720 --> 0:13:43.520 ways that we think a chat bought could get really convincing? 0:13:43.600 --> 0:13:47.200 Because I'd imagine that we're not there yet, right, but 0:13:47.440 --> 0:13:51.640 we're getting better. We are getting better, um um. And 0:13:51.760 --> 0:13:54.400 obviously there's some kind of hurdle we'd have to get 0:13:54.440 --> 0:13:59.000 over there. There's some like strategy for supplying these things 0:13:59.080 --> 0:14:02.600 with conversation asian rules that we haven't quite achieved yet. 0:14:02.800 --> 0:14:05.920 I mean, would uh, would machine learning have anything to 0:14:05.960 --> 0:14:09.520 do with it? Sure, like the idea of maybe mining 0:14:09.760 --> 0:14:14.160 thousands of other conversations to establish rules about how real 0:14:14.280 --> 0:14:18.560 people interact, or even having every time the machine has 0:14:18.600 --> 0:14:22.480 a conversation that it ends up reviewing that afterward and 0:14:22.600 --> 0:14:25.120 learning from its own experiences. I mean, these are right 0:14:25.160 --> 0:14:28.520 because you know, it's human. Human vocabulary and human understanding. 0:14:28.600 --> 0:14:31.920 Don't stop when you learn a language. We're continually learning 0:14:31.920 --> 0:14:34.360 our own languages. We're learning new rules for it, we're 0:14:34.560 --> 0:14:38.880 learning new vocabulary for it, we're creating things like metaphor. 0:14:38.920 --> 0:14:43.040 And similarly, we're creating these ideas, these abstract ideas that 0:14:43.120 --> 0:14:46.200 makes sense to people who are native speakers of that language, 0:14:46.200 --> 0:14:48.600 who have had exposure to this sort of thing, But 0:14:48.720 --> 0:14:52.000 to something that is just creating, you know, working out 0:14:52.040 --> 0:14:55.440 of say, a very strict dictionary of words, it would 0:14:55.760 --> 0:14:58.960 be meaningless or confusing. This reminds me of that time 0:14:58.960 --> 0:15:01.480 that that IBM t um. Remind me to remind me 0:15:01.520 --> 0:15:04.800 the name of the computer. Um. They yes, they taught 0:15:04.840 --> 0:15:08.720 Watson Urban dictionary. Oh no, yeah, oh no. Yeah. They 0:15:08.960 --> 0:15:11.840 then then after after Watson started talking back at them 0:15:11.840 --> 0:15:13.400 a little bit too much, they were just like, let's 0:15:13.440 --> 0:15:16.360 just nuke that portion of Watson from more. Yeah, it's 0:15:16.360 --> 0:15:17.880 the only way to be sure. Yeah. It turned out 0:15:17.880 --> 0:15:22.400 that Watson Watson developed a bit of a potty mouth, like, Um, 0:15:22.440 --> 0:15:24.960 are you familiar with the show Breaking Bad? Do you 0:15:25.000 --> 0:15:28.360 know how Jesse ends most of his sentences. That's pretty 0:15:28.400 --> 0:15:32.640 much how Watson was ending his sentences. Um. So yeah, 0:15:33.240 --> 0:15:35.600 they found out it was I think Urban Dictionary, and 0:15:35.640 --> 0:15:37.760 it was one other thing too that I think it 0:15:37.840 --> 0:15:40.200 might have been. It may have been Wikipedia, but I'm 0:15:40.200 --> 0:15:44.240 not sure. But they fed that into how to talk. Yeah, 0:15:44.880 --> 0:15:50.240 well then it just becomes unintelligible, right, hacks or yeah, okay, 0:15:50.560 --> 0:15:53.280 so uh I think I've seen I could be wrong. 0:15:53.280 --> 0:15:55.320 About this. What I think I remember seeing is that 0:15:55.680 --> 0:15:59.000 most recently, Ray Kurzwild said he thought that chat bots 0:15:59.040 --> 0:16:01.720 would reliable, you be able to beat the Turing test. 0:16:03.400 --> 0:16:06.040 And to be fair, when we say beat the Turing tests, 0:16:06.160 --> 0:16:10.520 there's no hard and fast rule. I think what that means, 0:16:11.840 --> 0:16:16.160 reliably fool the judges, right, yeah, because there's it all 0:16:16.160 --> 0:16:19.040 depends upon whom you ask, Right. Some people say you 0:16:19.120 --> 0:16:21.720 need to fool people at least thirty percent of the 0:16:21.760 --> 0:16:24.480 time and then you beat the Turing test. Other people say, well, 0:16:24.520 --> 0:16:27.120 Touring had suggested that you show it to a panel, 0:16:27.200 --> 0:16:29.280 and if the majority of the panel thinks that it's 0:16:29.280 --> 0:16:31.680 a real person, when in fact it's a computer, then 0:16:31.720 --> 0:16:34.160 the computer beats the Turing test. It all depends on 0:16:34.160 --> 0:16:36.560 how you frame it. There is not like a magical 0:16:36.920 --> 0:16:38.880 This is the way the test should be administered, and 0:16:38.920 --> 0:16:40.200 this is the only way you can tell if you 0:16:40.240 --> 0:16:42.240 pass or fail. I just want to make that clear 0:16:42.320 --> 0:16:45.400 because when we say passing the Turing test, that's so fuzzy. 0:16:45.600 --> 0:16:50.160 We really are talking about if you the listener, you 0:16:50.160 --> 0:16:51.960 you who are listening right now, if you were to 0:16:52.000 --> 0:16:54.160 have a conversation with one of these machines, you would 0:16:54.200 --> 0:16:56.840 not be completely certain whether or not that was a 0:16:56.880 --> 0:16:59.240 machine or a person. We see this actually used a 0:16:59.280 --> 0:17:03.520 lot in corporate settings for things like UM customer service. 0:17:03.880 --> 0:17:06.320 So if you ever have one of those customer service 0:17:06.359 --> 0:17:08.639 things pop up when you are on a website and 0:17:08.680 --> 0:17:11.360 it says do you need some help? Often this ends 0:17:11.400 --> 0:17:13.520 up being a chatbot that actually just has a very 0:17:13.640 --> 0:17:16.640 deep series of keywords that relate back to the products 0:17:16.680 --> 0:17:19.400 and services of that company, so that when you start 0:17:19.440 --> 0:17:22.679 typing things in, it can start sending you two links 0:17:22.720 --> 0:17:26.840 that possibly can solve your problem, but more likely will 0:17:26.880 --> 0:17:29.600 cause you to go into a red hot rage, burning 0:17:29.600 --> 0:17:32.439 brighter than a thousand exploding suns. Of course, they're the 0:17:32.520 --> 0:17:35.960 goal is to get you to the information or or 0:17:36.040 --> 0:17:38.439 service that you need. It's not really to trick you 0:17:38.480 --> 0:17:41.360 into thinking like, wow, I really just formed an emotional 0:17:41.359 --> 0:17:44.000 connection with this person, or the goal is to cut 0:17:44.040 --> 0:17:46.679 down on the cost of manning actual human beings and 0:17:46.720 --> 0:17:51.119 customer support roles. But that your your point stands either way. UM, 0:17:51.160 --> 0:17:55.040 so I would like to imagine a future where we're 0:17:55.080 --> 0:17:58.720 proposing the next Turing test. Okay, so we've gotten to 0:17:58.720 --> 0:18:00.760 the point where we can have a versation and not 0:18:00.880 --> 0:18:03.080 be sure if it's a person or a machine. Yes, so, 0:18:03.160 --> 0:18:05.280 either because the machines have become really smart or people 0:18:05.280 --> 0:18:08.120 aren't just dumb as bricks. Imagine we get there. Let's 0:18:08.119 --> 0:18:13.000 say it's nine kurs wild was right. Um now we've 0:18:13.040 --> 0:18:17.280 got chat bots that people cannot tell the chat bought 0:18:17.280 --> 0:18:20.160 from a human that every single time, you know, it's 0:18:20.160 --> 0:18:26.439 statistically and significant. Um uh. Once we're there, what's the 0:18:26.480 --> 0:18:30.120 next step. What's the next hardest thing in that same 0:18:30.160 --> 0:18:34.280 sort of realm of imitating human intelligence? Well, Joe, since 0:18:34.320 --> 0:18:37.160 this whole podcast is about storytelling, I'm going to take 0:18:37.200 --> 0:18:39.520 a wild stab in the dark and say it's storytelling. 0:18:39.640 --> 0:18:42.320 That's my guess. I think that'd be really interesting. I mean, 0:18:42.400 --> 0:18:46.200 so we can sort of imagine, like the chat bots 0:18:46.240 --> 0:18:49.560 we've interacted with today, they can't really do conversation, but 0:18:49.600 --> 0:18:52.920 we can imagine it. We should see it storytelling. Now, 0:18:52.960 --> 0:18:56.800 on one hand, I can imagine how an AI story 0:18:56.880 --> 0:18:59.240 bought would work. But on the other hand, it seems 0:18:59.320 --> 0:19:03.159 so alien into me. Surely they couldn't really create a 0:19:03.160 --> 0:19:06.800 piece of AI that could tell stories that seemed as 0:19:06.800 --> 0:19:11.679 good as human stories, right, I think it would be 0:19:11.720 --> 0:19:14.600 a huge challenge, But I don't see anything that's fundamentally 0:19:14.840 --> 0:19:17.440 impossible about it. Now, well, I mean if you take 0:19:17.440 --> 0:19:19.680 the fact that there are programs these days that can 0:19:19.720 --> 0:19:23.360 create music or or create works of visual art, then 0:19:23.520 --> 0:19:25.439 you know, and those are based on a few thousand 0:19:25.560 --> 0:19:30.520 rules about what makes a good piece of whatever media. Right. Okay, Well, 0:19:30.560 --> 0:19:33.159 well let's let's set it up. Now, let's describe the 0:19:33.200 --> 0:19:36.919 literary touring test. Okay that what what kind of test 0:19:37.320 --> 0:19:40.840 would a story bought have to pass? Well, you'd have 0:19:40.920 --> 0:19:44.119 to be able, I would say, to read the story 0:19:44.280 --> 0:19:48.720 and feel like it was saying satisfied, for lack of 0:19:48.720 --> 0:19:51.600 a better word, organic that that it did not it 0:19:52.200 --> 0:19:56.040 was not just a series of choppy sentences where a 0:19:56.040 --> 0:19:58.639 person goes through some mundane task. It would need to 0:19:58.720 --> 0:20:02.600 be uh, it would need to be engaging. I know 0:20:02.640 --> 0:20:04.760 you love that word, Joe, I keep using it because 0:20:04.760 --> 0:20:06.479 of that. But it would need to It would need 0:20:06.520 --> 0:20:09.920 to captivate an audience in some way, either because of 0:20:10.080 --> 0:20:13.560 the actual plot or the characterization or some combination thereof. 0:20:14.040 --> 0:20:15.520 And it would need to It would need to have 0:20:15.720 --> 0:20:19.000 a narrative flow. It would need to have internal consistency. So, 0:20:19.040 --> 0:20:21.600 in other words, things that happen earlier in the story 0:20:21.720 --> 0:20:25.280 could not be contradicted later in the story. Let's say 0:20:25.320 --> 0:20:29.120 that you have a story about a father and uh, 0:20:29.160 --> 0:20:33.560 the father experiences the heartbreak of a child dying, and 0:20:33.560 --> 0:20:35.480 that's a terrible thing. This is something that we would 0:20:35.520 --> 0:20:38.400 encounter in a story on novel whatever. But then three 0:20:38.480 --> 0:20:41.040 chapters later, the child's there and nothing has gone wrong, 0:20:41.080 --> 0:20:43.520 and there's it's not like it's mystery or I think 0:20:43.520 --> 0:20:46.400 it's just that the child had never died. And and 0:20:46.480 --> 0:20:49.440 either you know, either adhering to those rules, or if 0:20:49.440 --> 0:20:52.479 it breaks those rules, to be breaking them in a 0:20:52.520 --> 0:20:57.040 way that seems conscious and purposeful. Exactly, it couldn't just 0:20:57.080 --> 0:21:01.879 be breaking rules indiscriminately. Here's my blanket rule for the 0:21:02.119 --> 0:21:06.520 literary touring test I've imagined. Um, it doesn't have to 0:21:06.560 --> 0:21:10.200 create stories that are as good as your favorite author. 0:21:11.000 --> 0:21:13.359 It just has to do in the same way that 0:21:13.400 --> 0:21:16.280 the original Turing test didn't have to be as great 0:21:16.280 --> 0:21:20.840 a conversationalist as your most interesting friend. It just has 0:21:20.880 --> 0:21:23.720 to be good enough that you think this is probably 0:21:23.720 --> 0:21:27.760 written by a human being. Sure, I've read some awful stories, 0:21:27.800 --> 0:21:31.240 so the bar is set fairly low. Um, we're going 0:21:31.280 --> 0:21:34.200 back to that fan fiction thing now, But you've read 0:21:34.200 --> 0:21:37.120 some awful stories. But even people who are bad writers 0:21:37.600 --> 0:21:41.119 write better stories than than we could generate now with 0:21:41.240 --> 0:21:47.160 a I some really bad stories. I've read some really 0:21:47.160 --> 0:21:49.040 bad things. Well, let's let's just play this way. The 0:21:49.040 --> 0:21:51.320 way the way a computer would generate a story right now, 0:21:52.119 --> 0:21:55.040 the only way I think you could have one where 0:21:55.160 --> 0:21:59.360 it would have any sort of of actual sense to it, 0:22:00.520 --> 0:22:03.080