1 00:00:10,240 --> 00:00:14,040 Hey, it's Eric Newcomer. Welcome back to Cerebral Valley. 2 00:00:14,080 --> 00:00:17,400 This is Episode 2. I'm here with Max Child and 3 00:00:17,400 --> 00:00:20,920 James Wilsterman. We are going to dig into. 4 00:00:20,920 --> 00:00:25,480 We're calling this episode AI Kills Us All, which will give 5 00:00:25,480 --> 00:00:28,860 you a taste. It's not not the small 6 00:00:28,860 --> 00:00:30,740 questions. You know what what to do about 7 00:00:30,740 --> 00:00:34,460 privacy or like you know who should be in charge. 8 00:00:34,460 --> 00:00:37,900 It's like, what happens if we get artificial general 9 00:00:37,900 --> 00:00:40,380 intelligence? What, how worried are we about 10 00:00:40,620 --> 00:00:43,220 sort of the big doom and gloom? Guys? 11 00:00:43,700 --> 00:00:45,180 Great. Great to be back with you. 12 00:00:46,020 --> 00:00:47,260 Great to be back. Thanks for having us. 13 00:00:47,260 --> 00:00:49,900 Great to be here. After my conversation with Max 14 00:00:49,900 --> 00:00:53,100 and James, I will be interviewing Daniel H Wilson, 15 00:00:53,100 --> 00:00:55,620 the author of How to Survive a Robot Uprising. 16 00:00:55,620 --> 00:00:58,430 Where's my jetpack? And how to build a robot army? 17 00:00:58,430 --> 00:01:00,790 And the best seller, Robo Apocalypse. 18 00:01:00,790 --> 00:01:05,069 So someone who has spent a lot of time imagining the doom and 19 00:01:05,069 --> 00:01:09,110 gloom of machines getting powerful. 20 00:01:09,110 --> 00:01:11,590 So that's he. He's an expert in this 21 00:01:11,590 --> 00:01:14,910 speculation where we are mere amateurs, you know, with of 22 00:01:14,910 --> 00:01:19,110 course, the help of chat. GPTI was, you know, looking, 23 00:01:19,430 --> 00:01:28,330 wandering the Internet, looking for the great thinkers on AG is 24 00:01:28,330 --> 00:01:32,210 coming, and one of them is Nick Bostrom, who's been writing 25 00:01:32,210 --> 00:01:35,930 about this for years, I think. So this piece is the essay from 26 00:01:35,930 --> 00:01:41,850 his book Super Intelligence. The essay is called The 27 00:01:41,890 --> 00:01:45,250 Unfinished Fable of the Sparrows, which is very 28 00:01:45,250 --> 00:01:48,690 enjoyable and I I think you'll get a lot out of it. 29 00:01:50,380 --> 00:01:53,660 It was the nest building season, but after days of long, hard 30 00:01:53,660 --> 00:01:57,220 work, the sparrows sat in the evening glow, relaxing and 31 00:01:57,220 --> 00:02:00,180 chirping away. We are all so small and weak. 32 00:02:00,260 --> 00:02:03,780 Imagine how easy life would be if we had an owl who could help 33 00:02:03,780 --> 00:02:07,380 us build our nests. Yes, said another, And we could 34 00:02:07,380 --> 00:02:10,139 use it to look after our elderly and our young. 35 00:02:10,620 --> 00:02:13,460 It could give us advice and keep an eye out for the neighborhood 36 00:02:13,460 --> 00:02:17,540 cat, added a third. Then past us, the elder bird 37 00:02:17,540 --> 00:02:19,860 spoke. Let us send out scouts in all 38 00:02:19,860 --> 00:02:23,060 directions and try to find an abandoned outlet somewhere. 39 00:02:23,180 --> 00:02:26,620 Or maybe an egg a crow chick might also do. 40 00:02:26,780 --> 00:02:29,340 Or a baby weasel. This could be the best thing 41 00:02:29,340 --> 00:02:32,580 that ever happened to us, at least since the opening of the 42 00:02:32,580 --> 00:02:35,620 pavilion of Unlimited grain in Yonder Backyard. 43 00:02:36,380 --> 00:02:39,460 The flock was exhilarated and sparrows everywhere started 44 00:02:39,460 --> 00:02:41,300 chirping at the top of their lungs. 45 00:02:41,900 --> 00:02:45,140 Only Skronk Finkel A1 eyed Sparrow with a fretful 46 00:02:45,140 --> 00:02:48,480 temperament was unconvinced of the wisdom of the endeavor. 47 00:02:48,800 --> 00:02:51,680 Quoth he? This will surely be our undoing. 48 00:02:51,960 --> 00:02:55,440 Should we not give some thought to the art of owl domestication 49 00:02:55,440 --> 00:02:59,120 and owl taming first, before we bring such a creature into our 50 00:02:59,120 --> 00:03:03,800 midst replied past us, taming an owl sounds like an exceedingly 51 00:03:03,800 --> 00:03:07,000 difficult thing to do. It will be difficult to find an 52 00:03:07,000 --> 00:03:10,520 owl egg, so let us start there. After we have succeeded in 53 00:03:10,520 --> 00:03:14,760 raising an owl, then we can think about taking on this other 54 00:03:14,760 --> 00:03:17,470 challenge. There is a flaw in that plan. 55 00:03:17,470 --> 00:03:20,670 Squeak Skronk Finkel. But his protests were in vain, 56 00:03:20,670 --> 00:03:24,110 as the flock had already lifted off to start implementing the 57 00:03:24,110 --> 00:03:28,470 directive set out by Pastis. Just two or three sparrows 58 00:03:28,470 --> 00:03:31,350 remain behind. Together, they began to try to 59 00:03:31,350 --> 00:03:34,790 work out how owls might be tamed or domesticated. 60 00:03:35,110 --> 00:03:37,510 They soon realized that pastis had been right. 61 00:03:37,830 --> 00:03:40,990 This was an exceedingly difficult challenge, especially 62 00:03:40,990 --> 00:03:44,270 in the absence of an actual owl to practice on. 63 00:03:44,650 --> 00:03:48,570 Nevertheless, they pressed on as best they could, constantly 64 00:03:48,570 --> 00:03:52,850 fearing that the flock might return with an owl leg before a 65 00:03:53,050 --> 00:03:56,330 solution to the control problem had been found. 66 00:03:58,050 --> 00:03:59,210 All right, what do you guys make of that? 67 00:03:59,210 --> 00:04:03,650 It's. Not quite Plato, but it's. 68 00:04:04,810 --> 00:04:12,130 Heavy-handed, little on the nose, I feel like this might 69 00:04:12,130 --> 00:04:16,110 have been more. You know, mind blowing in 2014 70 00:04:16,110 --> 00:04:19,029 when when it was raining, I guess, but nowadays it's just 71 00:04:19,029 --> 00:04:21,269 like, Oh yeah, that's that's what we're talking about all the 72 00:04:21,269 --> 00:04:27,550 time. But I do think, I do think this 73 00:04:27,590 --> 00:04:31,190 obviously gets at this question of, you know, should you ever, 74 00:04:31,190 --> 00:04:33,430 should you even go down this path at all, right. 75 00:04:33,470 --> 00:04:37,270 I think that's the core question in develop trying to develop 76 00:04:37,270 --> 00:04:41,110 AGI. Or is it so risky that you 77 00:04:41,110 --> 00:04:43,580 should? You know not do that at all 78 00:04:43,580 --> 00:04:47,540 unless you are 100% confident you figured out how to tame the 79 00:04:47,580 --> 00:04:51,300 OWL right in advance. Is that your guys reading as 80 00:04:51,300 --> 00:04:53,660 well? Yeah, well, I think just to like 81 00:04:54,420 --> 00:04:57,300 rewind one second, I think coming back to this like AI 82 00:04:57,540 --> 00:05:00,180 kills us all point. I think there's two big 83 00:05:00,180 --> 00:05:03,860 questions in the sort of will AIS kill, kill us all narrative, 84 00:05:03,860 --> 00:05:05,780 right. Or maybe three. 85 00:05:05,780 --> 00:05:09,620 I guess one will be are we capable of developing an AI 86 00:05:09,620 --> 00:05:14,220 that's smarter than humans? Like 2 will be if we're capable 87 00:05:14,220 --> 00:05:17,660 of developing an AI smarter than humans, would it want to kill 88 00:05:17,660 --> 00:05:20,660 us? And then three being, if it 89 00:05:20,660 --> 00:05:23,300 wants to kill us, can it figure out some way to kill us? 90 00:05:23,340 --> 00:05:25,260 Basically, right. I mean like and I think those 91 00:05:25,260 --> 00:05:28,340 are all kind of three different questions and maybe we all like 92 00:05:28,340 --> 00:05:32,220 agree on this podcast on like certain one or two or three or 93 00:05:32,260 --> 00:05:34,460 or different elements of them, but like. 94 00:05:34,940 --> 00:05:35,940 I think it's helpful to. Yeah. 95 00:05:35,940 --> 00:05:37,260 To break it down that way because. 96 00:05:37,300 --> 00:05:40,660 And if you do that, this, this story, the parable, the Fable 97 00:05:40,820 --> 00:05:43,660 that I just read sort of assumes a couple of those, right. 98 00:05:43,660 --> 00:05:48,980 It assumes by using an owl, it assumes that such a thing could 99 00:05:48,980 --> 00:05:52,140 exist, right. And that it, it's a dangerous 100 00:05:52,140 --> 00:05:55,820 being that we actually need to figure out in its essence, it's 101 00:05:55,820 --> 00:05:58,700 dangerous and we need to figure out how to control it before we 102 00:05:58,780 --> 00:06:00,940 create it, right. So it sort of assumes into the 103 00:06:00,940 --> 00:06:04,390 story certain things exactly that you're framing up are open 104 00:06:04,390 --> 00:06:05,590 question. I think they're open question. 105 00:06:05,590 --> 00:06:08,830 I mean in the parable basically the OWL both is smarter and more 106 00:06:08,830 --> 00:06:13,030 powerful than the sparrows, definitely wants to apparently 107 00:06:13,030 --> 00:06:15,710 kill or very likely potentially wants to kill the sparrows and 108 00:06:15,710 --> 00:06:17,910 certainly has the capability given that it's bigger and 109 00:06:17,910 --> 00:06:19,870 stronger or whatever, right. So to your point, it's sort of 110 00:06:19,870 --> 00:06:22,030 like you know, the answer is kind of written into the story, 111 00:06:22,030 --> 00:06:25,510 whereas I think in real life like I think can we build AI 112 00:06:25,510 --> 00:06:27,430 smarter than us. Part one is actually a pretty 113 00:06:27,430 --> 00:06:29,950 interesting and contentious question and then the other ones 114 00:06:29,950 --> 00:06:31,790 are also pretty interesting and contentious as well. 115 00:06:31,790 --> 00:06:33,750 But just to start on part one like. 116 00:06:34,080 --> 00:06:36,280 Do you guys think we can build an AI that's smarter than 117 00:06:36,280 --> 00:06:40,160 humans? I guess like, let's be very 118 00:06:40,160 --> 00:06:42,000 specific about how we define that, right? 119 00:06:42,120 --> 00:06:44,520 Because I think it's a pretty critical question. 120 00:06:45,200 --> 00:06:48,960 Come on. Well, what's your definition of, 121 00:06:49,040 --> 00:06:51,520 I guess, smarter than humans? Is that a? 122 00:06:52,000 --> 00:06:53,560 Are you saying super intelligence? 123 00:06:53,640 --> 00:06:56,520 We're talking about, say, an. IQ That has never been achieved 124 00:06:56,520 --> 00:06:57,760 by a human right. You know I. 125 00:06:57,760 --> 00:07:01,040 I think that like I think there's kind of this concept of 126 00:07:01,360 --> 00:07:06,690 a IAGI, which is. Better at I would say like the 127 00:07:06,690 --> 00:07:11,210 median human at, you know, most tasks and then there's super 128 00:07:11,210 --> 00:07:14,210 intelligence, right? Which is you know, better like 129 00:07:14,530 --> 00:07:17,210 better than the best humans at most tasks, right. 130 00:07:17,490 --> 00:07:19,010 Is that how you guys think about it as well? 131 00:07:19,050 --> 00:07:20,730 Yeah. Let's say better than the best 132 00:07:20,730 --> 00:07:24,650 humans at any mental task or any any relevant mental task, right? 133 00:07:25,690 --> 00:07:29,530 I guess I believe that this is going to happen, I think. 134 00:07:29,890 --> 00:07:32,610 You know, it would be naive to say, well it won't happen. 135 00:07:32,810 --> 00:07:35,650 So then it's just a matter of a question of like when it will 136 00:07:35,650 --> 00:07:40,330 happen and whether there are any kind of domains where it feels 137 00:07:40,330 --> 00:07:43,050 like it won't happen for a long time, right? 138 00:07:43,250 --> 00:07:47,170 Like are there, you know, any domains that we really think you 139 00:07:47,170 --> 00:07:49,730 know it's going to be take take much longer to achieve super 140 00:07:49,730 --> 00:07:53,570 intelligence than than other domains, right. 141 00:07:53,570 --> 00:07:56,920 So maybe it will be the best. Coder in the world. 142 00:07:57,040 --> 00:07:59,480 But will it also be the best screenwriter? 143 00:07:59,480 --> 00:08:02,160 Are those things gonna happen at the same time or are they gonna 144 00:08:02,160 --> 00:08:04,080 happen, you know, at different times? 145 00:08:05,640 --> 00:08:11,960 Yeah, I I, I mean, first of all, you know, AI is smarter than the 146 00:08:11,960 --> 00:08:14,320 smartest humans in some domains right now. 147 00:08:14,320 --> 00:08:18,520 Already shall you go Wow games particularly. 148 00:08:18,960 --> 00:08:20,680 Right. But that that is not a general 149 00:08:20,680 --> 00:08:22,320 intelligence. No, no, I'm just saying, just. 150 00:08:22,320 --> 00:08:24,360 Yeah, but in specific demands it's better. 151 00:08:24,800 --> 00:08:28,320 I I think it will continue to cleave off domains, right. 152 00:08:28,320 --> 00:08:32,240 I think like sort of beat all humans at the LSAT type thing 153 00:08:32,240 --> 00:08:37,159 seems very soon. I I think what feels far away is 154 00:08:37,159 --> 00:08:41,720 sort of and it fits into the fears of an AGI is sort of the 155 00:08:42,320 --> 00:08:45,840 strategic planning like what an AGI looks like that sets like 156 00:08:45,840 --> 00:08:49,900 broader priorities across like in a game it's like you want to 157 00:08:49,900 --> 00:08:52,900 win at chess, like straightforward, but when it has 158 00:08:52,940 --> 00:08:56,540 to you know sort of solve general optimal outcomes. 159 00:08:56,540 --> 00:09:01,020 Like it feels like we are so far from a being that could do that. 160 00:09:01,020 --> 00:09:05,060 Or a computer that's interested in sort of deciding whether it 161 00:09:05,060 --> 00:09:08,020 should play chess or do something else and like why 162 00:09:08,020 --> 00:09:11,940 especially if we're not again just articulating what the goals 163 00:09:11,940 --> 00:09:14,380 are. So I guess I'm, I'm not as 164 00:09:14,380 --> 00:09:19,960 bullish on in the next decade a sort of coherent overall sort of 165 00:09:19,960 --> 00:09:22,040 being that feels like smarter than a human. 166 00:09:22,200 --> 00:09:24,600 I mean the next decade, though. I mean, if we're already 167 00:09:24,600 --> 00:09:27,280 negotiating this down to the next decade, it would seem like 168 00:09:27,280 --> 00:09:28,800 you're pretty confident that it's gonna happen. 169 00:09:28,840 --> 00:09:30,600 Like lifetimes. I think it's gonna happen, OK. 170 00:09:30,800 --> 00:09:32,440 So you think in our lifetimes that will happen? 171 00:09:32,440 --> 00:09:35,260 Yeah. What about my my earlier 172 00:09:35,260 --> 00:09:38,740 distinction though Eric, do you do you think you know an AGI 173 00:09:38,740 --> 00:09:41,900 that is a general intelligence that is basically better than 174 00:09:41,900 --> 00:09:46,260 the media and human at most things is or or equal to is more 175 00:09:46,260 --> 00:09:47,780 likely to happen in the next decade. 176 00:09:48,460 --> 00:09:50,780 Is that five years? I just think it's hard to do it. 177 00:09:51,700 --> 00:09:55,740 I we just don't see it as sort of a an overall agent or like an 178 00:09:55,740 --> 00:09:57,380 overall decision, you know what I mean? 179 00:09:57,380 --> 00:10:00,940 I I see it like winning at a bunch of discrete tasks. 180 00:10:01,500 --> 00:10:04,540 OK, sort of a sign. So this kind of gets at more of 181 00:10:04,540 --> 00:10:09,060 like your definition of AGI potentially would require some 182 00:10:09,060 --> 00:10:13,780 sort of Turing test type thing that it can really just convince 183 00:10:13,780 --> 00:10:15,940 you that it is a human. Or I guess I'm trying to think 184 00:10:15,980 --> 00:10:20,140 of like an overall thinker with like sort of priorities and sort 185 00:10:20,220 --> 00:10:22,140 of. But at the end of the day, like, 186 00:10:22,140 --> 00:10:25,420 just to make any predictions about this, it has to kind of be 187 00:10:25,420 --> 00:10:29,540 a falsifiable assertion of like, what is AGI, right? 188 00:10:30,510 --> 00:10:34,270 And it's not that easy to to define that except by some sort 189 00:10:34,270 --> 00:10:37,270 of like test, right? Otherwise you end up in this 190 00:10:37,270 --> 00:10:41,630 like era or this pattern we we're always in that we are 191 00:10:41,630 --> 00:10:44,550 constantly moving the goal posts on like what AGI is, right? 192 00:10:44,710 --> 00:10:45,950 Right. Yeah. 193 00:10:46,310 --> 00:10:49,630 I mean, I think, I mean, I think you guys are both like in the 194 00:10:49,630 --> 00:10:51,630 same place, which I am, which is like it's very hard to figure 195 00:10:51,630 --> 00:10:54,390 out the timeline, but like to me it's definitely happening 196 00:10:54,630 --> 00:10:56,430 sometime in the next 100 years, let's say. 197 00:10:56,800 --> 00:11:00,320 Whether it's 10 or 20 or 50 or 100, it is a little bit harder 198 00:11:00,320 --> 00:11:03,400 to guess, but like it is kind of a math problem, right? 199 00:11:03,400 --> 00:11:06,720 Where we have brains that are just some giant collection of 200 00:11:06,720 --> 00:11:09,920 neurons, right? And there's, you know, millions 201 00:11:09,920 --> 00:11:12,080 and millions of them or whatever, whatever order of 202 00:11:12,080 --> 00:11:15,320 magnitude it is, right? And essentially we're training 203 00:11:15,320 --> 00:11:18,880 these large language models on, you know, a neural network in 204 00:11:18,880 --> 00:11:21,280 some ways, right? Which, you know, it's not a one 205 00:11:21,280 --> 00:11:24,360 to one copy of what? A neuron, how a neuron works in 206 00:11:24,360 --> 00:11:25,880 our brain, which is way more complicated, but. 207 00:11:26,490 --> 00:11:28,770 Fundamentally, you're sort of like, if I can just throw 208 00:11:28,770 --> 00:11:32,570 numbers at this problem of like building a brain essentially and 209 00:11:32,570 --> 00:11:36,770 like get more GPU cores, like just throw absolute just 210 00:11:37,010 --> 00:11:39,810 computing power at this thing. Like the story of our entire 211 00:11:39,810 --> 00:11:41,610 lives has been computing power just keeps going up 212 00:11:41,610 --> 00:11:43,170 exponentially, right? You know, Moore's Law or 213 00:11:43,170 --> 00:11:46,050 whatever, right? And so to me, it seems like 214 00:11:46,690 --> 00:11:48,610 you'd have to believe there's something fundamentally 215 00:11:48,610 --> 00:11:51,450 different about the way a human brain works or a neuron works. 216 00:11:51,950 --> 00:11:55,390 To the way you know a neural network or you know the way GPU 217 00:11:55,830 --> 00:11:58,590 transformer model works to to believe that we can't just throw 218 00:11:58,590 --> 00:12:00,510 numbers at this thing until it gets smarter than humans. 219 00:12:01,390 --> 00:12:03,630 And so I guess I do believe that we throw enough numbers at it 220 00:12:03,630 --> 00:12:06,670 and get smarter than humans. I mean humans have to win 221 00:12:06,670 --> 00:12:09,870 multiple sort of parts of the brain that developed in 222 00:12:09,870 --> 00:12:11,470 different ways that interact with each. 223 00:12:11,550 --> 00:12:12,270 Other. Yeah. 224 00:12:12,270 --> 00:12:15,070 But they're still kind of built on the same cell structure. 225 00:12:15,190 --> 00:12:18,270 I mean, yeah, yes, but but you could do that in an in an AGI 226 00:12:18,270 --> 00:12:19,990 too, right? You could have different lobes 227 00:12:19,990 --> 00:12:21,870 or whatever of the imaginary intelligence, right? 228 00:12:21,870 --> 00:12:26,790 And then broadly, we just really don't understand how to pinpoint 229 00:12:26,910 --> 00:12:30,030 the experience of human consciousness, like quality or 230 00:12:30,030 --> 00:12:31,550 whatever. I talked about this at length 231 00:12:31,550 --> 00:12:36,490 with Reed Hoffman when I interviewed him and like so, but 232 00:12:36,490 --> 00:12:38,810 we're going to, but we're very reluctant because we can 233 00:12:38,810 --> 00:12:42,610 experience our own consciousness to attribute it and animals and 234 00:12:42,610 --> 00:12:46,890 we really have no like road map for how we'd ever identify it in 235 00:12:46,890 --> 00:12:48,890 machines. And that's sort of different 236 00:12:48,890 --> 00:12:50,850 than the Super intelligence thing, but it's certainly part 237 00:12:50,850 --> 00:12:53,730 of what you want in the sort of like does it have like a real 238 00:12:53,730 --> 00:12:56,570 existence sort of question, which I don't even have a road 239 00:12:56,570 --> 00:12:59,540 map for how we'd identify. Coming back to the core topic of 240 00:12:59,540 --> 00:13:01,060 the episode, like, will AI kill us all? 241 00:13:01,060 --> 00:13:04,220 Like, it doesn't really matter if it's conscious or not, right? 242 00:13:04,220 --> 00:13:06,660 If it ends up having the power and the desire to kill us, 243 00:13:06,660 --> 00:13:09,260 right. So like, I do think that's an 244 00:13:09,260 --> 00:13:11,140 interesting question and I don't know, maybe that's a different 245 00:13:11,140 --> 00:13:13,740 episode. But like to me that's not, 246 00:13:13,900 --> 00:13:15,820 that's not worth getting hung up on in the like is it going to 247 00:13:15,820 --> 00:13:17,260 kill us all thing like definitely. 248 00:13:17,620 --> 00:13:21,260 I mean, this sort of classic example being the paper clip 249 00:13:21,460 --> 00:13:25,020 maximizer, right? The machine, the fear that we 250 00:13:25,020 --> 00:13:29,640 just program this really smart, you know, LLM and tell it like, 251 00:13:29,640 --> 00:13:34,360 oh, you're not only objective is maximize paper clips, you know, 252 00:13:34,360 --> 00:13:36,560 And then it goes about it and then it's like, well, the humans 253 00:13:36,560 --> 00:13:39,880 are hurting sort of the maximization of paper clips. 254 00:13:39,880 --> 00:13:43,800 Let's get rid of them. I mean famously like I Robot by 255 00:13:43,800 --> 00:13:46,120 Asimov. It's like robots are supposed to 256 00:13:46,120 --> 00:13:49,000 protect humans, but then humans all kill each other. 257 00:13:49,200 --> 00:13:51,320 So it's like, oh, we need to protect the humans. 258 00:13:51,320 --> 00:13:54,520 I sort of need to enslave them in order to ensure their 259 00:13:54,560 --> 00:13:58,280 protection, you know, so that doesn't it doesn't necessarily 260 00:13:58,280 --> 00:13:59,600 need to have the depth of thinking. 261 00:13:59,600 --> 00:14:02,960 You know, humans would see sort of the flawed what we would 262 00:14:02,960 --> 00:14:05,440 perceive to be flawed reasoning in those things, but it would 263 00:14:05,440 --> 00:14:08,560 still but a machine. You could see why a machine with 264 00:14:08,560 --> 00:14:10,200 certain objectives could operate. 265 00:14:10,240 --> 00:14:13,320 Though well, I I think it's sort of like saying that these 266 00:14:13,320 --> 00:14:17,160 artificial intelligences can have totally different 267 00:14:17,200 --> 00:14:21,200 orthogonal objectives that don't require a consciousness, right? 268 00:14:21,200 --> 00:14:24,080 Like they just for some reason develop a different objective. 269 00:14:24,580 --> 00:14:29,740 Than we have as humans, and they're capable of bringing 270 00:14:29,740 --> 00:14:32,660 about that objective to the detriment of humanity, 271 00:14:32,660 --> 00:14:35,580 basically, right? Yeah, I mean, a nuclear weapon 272 00:14:35,580 --> 00:14:37,500 doesn't have consciousness, but it's still very effective at 273 00:14:37,500 --> 00:14:41,300 killing people, right? That's its mission, so. 274 00:14:41,640 --> 00:14:43,040 Yeah. I mean, so I, it seems like we 275 00:14:43,040 --> 00:14:45,240 all agree, OK, we're going to achieve super intelligence at 276 00:14:45,240 --> 00:14:47,920 some point in the next, let's say, century or an intelligence 277 00:14:47,920 --> 00:14:50,360 that's smarter than humans and enough domains that it's, it's 278 00:14:50,360 --> 00:14:52,600 meaningfully, you know, more capable than we are, right. 279 00:14:52,640 --> 00:14:56,120 Yeah. So second question, like, you 280 00:14:56,120 --> 00:14:58,560 know, would it want to kill us? I think this is like one of the 281 00:14:58,560 --> 00:15:01,640 most interesting ones and you alluded to the sort of paper 282 00:15:01,640 --> 00:15:04,320 clip maximization as one example of how it might just end up 283 00:15:04,320 --> 00:15:07,000 killing us as like a side effect of some other mission it's on. 284 00:15:07,000 --> 00:15:10,110 Right, but. I guess A, like, how likely do 285 00:15:10,110 --> 00:15:13,710 we think that is? And then B, like, do we think 286 00:15:13,710 --> 00:15:17,870 that we'll be able to constrain its missions in some way so that 287 00:15:17,870 --> 00:15:21,590 it wants to keep us alive, you know, even if it's going off and 288 00:15:21,590 --> 00:15:25,270 executing at school? I do actually, you know, worry 289 00:15:25,270 --> 00:15:29,190 about this. I guess that the AI will have 290 00:15:29,190 --> 00:15:32,750 some orthogonal objective that is detrimental to humans. 291 00:15:33,030 --> 00:15:35,430 And then there's like a secondary question of like, can 292 00:15:35,430 --> 00:15:36,950 we do anything about that, right. 293 00:15:37,110 --> 00:15:39,400 We were like, I mean to read between what you're saying, 294 00:15:39,400 --> 00:15:42,560 you're like, if anthropic and open AI are left only in charge, 295 00:15:42,560 --> 00:15:45,720 we'll be fine. But if Facebook's llama runs 296 00:15:45,720 --> 00:15:47,920 around or screwed, I mean, that's sort of, I mean, I'm 297 00:15:47,920 --> 00:15:50,960 interested to see what Mustafa, who's going to speak at Cerebral 298 00:15:50,960 --> 00:15:53,840 Valley, you know, he's been much more reticent about open source, 299 00:15:54,520 --> 00:15:56,360 potentially. His lack of control issue. 300 00:15:57,560 --> 00:16:03,320 I do think that eventually if we start to see, you know, the 301 00:16:03,320 --> 00:16:05,680 risks like that, we all agree that. 302 00:16:06,220 --> 00:16:08,780 You know this, we're headed towards this super intelligence 303 00:16:08,780 --> 00:16:12,260 and we start to see how dangerous that may or may not 304 00:16:12,260 --> 00:16:13,500 be. Like, I just don't think we know 305 00:16:13,500 --> 00:16:17,180 yet how dangerous it will be. Like it's obvious to me like you 306 00:16:17,180 --> 00:16:20,340 want to maintain the ability to regulate this. 307 00:16:20,340 --> 00:16:24,140 And you know, just a lot of people, I think I saw Brian 308 00:16:24,140 --> 00:16:27,700 Armstrong from Coinbase, you know, tweet that he just thinks 309 00:16:27,700 --> 00:16:30,660 there should never be any AI regulation or I don't know if he 310 00:16:30,660 --> 00:16:33,740 said never, but he was basically like regulation has such a poor 311 00:16:33,740 --> 00:16:36,910 track record around innovation. Like, I just think that's sort 312 00:16:36,910 --> 00:16:42,470 of naive without us knowing yet, like how dangerous it can be or 313 00:16:42,470 --> 00:16:44,030 not. You just need to have the best 314 00:16:44,110 --> 00:16:46,510 AI on our side. You know, like when the killer 315 00:16:46,510 --> 00:16:48,630 one comes, we need to have the benevolent 1. 316 00:16:48,950 --> 00:16:52,790 The best defense against a bad guy with an AI is a good guy 317 00:16:52,790 --> 00:16:55,550 with an AI. That's how that's how Americans 318 00:16:55,550 --> 00:16:57,070 actually think. You know, like. 319 00:16:57,950 --> 00:17:00,910 The right to bear AI, baby, Yeah. 320 00:17:01,430 --> 00:17:04,230 That was kind of Sam Altman's perspective when he started Open 321 00:17:04,230 --> 00:17:05,710 AI. Like, I mean, at least my 322 00:17:05,710 --> 00:17:07,430 reading. Of it when he was getting 323 00:17:07,430 --> 00:17:11,869 started was calling it open AI because he believed putting AI 324 00:17:11,869 --> 00:17:15,790 in the hands of humanity democratizing it was the best 325 00:17:15,790 --> 00:17:18,430 way to go about things. And I think he's changed his 326 00:17:18,430 --> 00:17:21,310 mind because of this exact thing that I'm saying, which is it's 327 00:17:21,310 --> 00:17:23,829 actually pretty dangerous or potentially dangerous. 328 00:17:24,190 --> 00:17:26,829 The fact that opening eye isn't open at all I know is like 329 00:17:26,829 --> 00:17:29,550 something that has been beaten to death, but it's just still 330 00:17:29,550 --> 00:17:32,430 ridiculous to me just that. It's hilarious kind of, yeah. 331 00:17:32,550 --> 00:17:34,350 Yeah, cuz it's historical artifact or? 332 00:17:34,350 --> 00:17:37,070 Something, right. But I mean I think I guess 333 00:17:37,190 --> 00:17:40,390 putting aside the regulation thing and the open source is 334 00:17:40,390 --> 00:17:43,390 closed source thing, I just think that if you believe that. 335 00:17:43,740 --> 00:17:47,420 A substantially diverse group of people from different countries 336 00:17:47,420 --> 00:17:49,700 within America, open source, closed source, can develop a 337 00:17:49,700 --> 00:17:52,180 super intelligence, right? Like if you believe there's 338 00:17:52,180 --> 00:17:55,260 going to be more than like five of these, which I think there 339 00:17:55,260 --> 00:17:58,300 would be like in any scenario in which this actually happens, 340 00:17:58,300 --> 00:18:02,820 like isn't it just very likely that one of the five or ten or 341 00:18:02,820 --> 00:18:05,300 hundred or thousand of these is is a bad AI? 342 00:18:05,300 --> 00:18:05,860 Mean? I just don't. 343 00:18:06,020 --> 00:18:09,940 I don't think it matters if you have 999 aligned good AI and one 344 00:18:09,940 --> 00:18:12,660 bad one, if that one bad one is capable of destroying the world, 345 00:18:12,660 --> 00:18:13,380 right? I mean like. 346 00:18:13,680 --> 00:18:17,480 It again, I just, I just think that mathematically if it's 347 00:18:17,480 --> 00:18:20,040 possible to create super intelligence is like someone is 348 00:18:20,040 --> 00:18:21,960 going to create a super intelligence that's really bad, 349 00:18:21,960 --> 00:18:23,960 right? And I'm not sure like our 350 00:18:23,960 --> 00:18:27,880 existing analogy is around like guns or nuclear deterrence or 351 00:18:27,880 --> 00:18:30,640 whatever sort of game theory you want to play out here like apply 352 00:18:30,640 --> 00:18:34,760 where like if anyone bad AI can get access to the tools that 353 00:18:34,760 --> 00:18:38,080 needs to kill everyone, it will it will succeed, right? 354 00:18:38,080 --> 00:18:40,800 Like, I don't think all the good AI can stop it, but but maybe 355 00:18:40,800 --> 00:18:43,810 I'm going too far out on this. I mean, I agree with it. 356 00:18:43,810 --> 00:18:46,970 I think so. So your take is that obviously 357 00:18:46,970 --> 00:18:50,290 yes, the answer is yes. You know, if we achieve super 358 00:18:50,290 --> 00:18:53,650 intelligence, there's a pretty big risk to that. 359 00:18:53,730 --> 00:18:56,410 I mean, in the world. You would have to create some 360 00:18:56,410 --> 00:18:58,770 way for the good AI to regulate the bad AI. 361 00:18:58,770 --> 00:19:01,050 Basically, like if you believe we're in a world where no longer 362 00:19:01,050 --> 00:19:03,010 humans are, they're not on the driver's seat, right? 363 00:19:03,010 --> 00:19:05,530 We're in the back seat, right? Like you now have to basically 364 00:19:05,530 --> 00:19:07,530 try to figure out how to make sure that the AI that are in the 365 00:19:07,530 --> 00:19:09,290 front seat can control the other ones. 366 00:19:10,190 --> 00:19:13,470 I mean part of what Max is basically referencing just, you 367 00:19:13,470 --> 00:19:16,510 know, to give a little context, is Elisa Yudakowski. 368 00:19:17,030 --> 00:19:20,310 I mean I pulled actually a quote, you know, to visualize a 369 00:19:20,310 --> 00:19:23,150 hostile superhuman AI. Don't imagine a lifeless book, 370 00:19:23,150 --> 00:19:26,070 smart thinker dwelling inside the Internet and sending I'll 371 00:19:26,070 --> 00:19:28,790 intentioned emails. Visualize an entire alien 372 00:19:28,790 --> 00:19:32,310 civilization thinking at millions of time human speeds, 373 00:19:32,510 --> 00:19:35,550 initially confined to computers in a world of creatures that are 374 00:19:35,710 --> 00:19:37,990 from its perspective, very stupid and very slow. 375 00:19:38,400 --> 00:19:41,120 A sufficiently intelligent AI won't stay confined to computers 376 00:19:41,120 --> 00:19:43,360 for long. In today's world, you can e-mail 377 00:19:43,360 --> 00:19:47,120 DNA strings to laboratories that will produce proteins on demand, 378 00:19:47,120 --> 00:19:50,040 allowing an AI initially confined to the Internet to 379 00:19:50,040 --> 00:19:53,640 build artificial life forms or bootstraps straight to post 380 00:19:53,640 --> 00:19:55,760 biological molecular manufacturing. 381 00:19:55,760 --> 00:19:59,240 I don't know. My point of view is just like, I 382 00:19:59,240 --> 00:20:01,520 think him and sort of what you guys are getting at, it just 383 00:20:01,520 --> 00:20:03,200 feels like you're like, oh it's super intelligent. 384 00:20:03,360 --> 00:20:06,320 It's like God, like, you know, I I still think, OK, it's way 385 00:20:06,320 --> 00:20:09,410 smarter than us. But we're like a bunch of 386 00:20:09,410 --> 00:20:11,610 intelligent beings, even if it's smarter than us. 387 00:20:11,650 --> 00:20:15,330 I I mean, it requires that that it's going to act stealthily and 388 00:20:15,330 --> 00:20:18,450 sort of like start manufacturing like a biologic version. 389 00:20:18,490 --> 00:20:20,970 You know, like, I just feel like it's more likely that we're 390 00:20:20,970 --> 00:20:24,850 like, this thing doesn't really listen to us and like, but then 391 00:20:24,850 --> 00:20:27,490 like the GPU clusters are like, it's trying to fire them all. 392 00:20:27,490 --> 00:20:30,290 And it's just like, oh, there's only so much computing power 393 00:20:30,290 --> 00:20:33,130 that, you know, it's still a being that requires like 394 00:20:33,130 --> 00:20:35,430 resources. You know, like humans are 395 00:20:35,430 --> 00:20:37,630 smarter than everything else. We still die like they're 396 00:20:37,630 --> 00:20:42,150 they're just like physical limits to things and and that 397 00:20:42,150 --> 00:20:44,150 it's not going to have like God like powers. 398 00:20:44,150 --> 00:20:47,830 It's going to be limited by the amount of compute it can access 399 00:20:47,830 --> 00:20:50,790 and sort of our cooperation at various points. 400 00:20:50,790 --> 00:20:53,150 Like, I I don't know. I don't think it's on zero to 401 00:20:53,150 --> 00:20:55,350 God. I feel like you're painting. 402 00:20:55,390 --> 00:20:57,550 Correct me if I'm wrong, but you're painting a picture where 403 00:20:57,550 --> 00:21:00,630 we have created super intelligence. 404 00:21:01,140 --> 00:21:05,100 And you agree with Max's point that some of them are misaligned 405 00:21:05,100 --> 00:21:07,420 and probably would want to kill us if they could. 406 00:21:07,740 --> 00:21:13,980 However, you're saying that the humanity's fine because those 407 00:21:13,980 --> 00:21:16,340 misaligned AIS can't get enough resources. 408 00:21:16,420 --> 00:21:20,740 Is that accurate, right, I mean. Just it's hard to know, right? 409 00:21:20,980 --> 00:21:21,500 We don't. We don't. 410 00:21:21,900 --> 00:21:24,420 We don't know what to do. That's definitely a gamble. 411 00:21:24,420 --> 00:21:26,780 I mean, even if you believe that you're like. 412 00:21:28,500 --> 00:21:31,100 You've got you've gotten pretty close to the edge. 413 00:21:33,020 --> 00:21:36,140 Well I'm I'm definitely a fatalist on AI as my if we're 414 00:21:36,140 --> 00:21:38,660 getting to like our deep, like what is our position. 415 00:21:38,660 --> 00:21:42,660 It just, I just think like many people in tech, like the 416 00:21:42,660 --> 00:21:46,910 regulation is pretty bad and like going to be only going to 417 00:21:46,910 --> 00:21:50,070 stop the good guys, that this is going to keep going and that I'd 418 00:21:50,070 --> 00:21:54,270 rather the best people we have try to figure it out rather than 419 00:21:54,270 --> 00:21:58,630 like operate in the shadows. Yeah, I I tend to agree with you 420 00:21:58,630 --> 00:22:02,430 and I probably would agree that we're not, we don't need a lot 421 00:22:02,430 --> 00:22:08,030 of regulation right now on you know who can you know train 422 00:22:08,030 --> 00:22:10,350 these models or something like but I. 423 00:22:11,050 --> 00:22:14,570 I also wouldn't rule out like meeting this in the future if in 424 00:22:14,570 --> 00:22:17,250 a year from now or two years from now I start seeing you know 425 00:22:17,250 --> 00:22:20,850 a lot of like evidence that we've, the open source community 426 00:22:20,850 --> 00:22:24,250 has created super intelligent misaligned AI, right. 427 00:22:24,410 --> 00:22:28,050 And I think people are kind of, you know, going a little bit too 428 00:22:28,050 --> 00:22:31,370 far in the direction of saying this, you know, we don't need 429 00:22:31,370 --> 00:22:33,170 it. We need any regulation here. 430 00:22:33,370 --> 00:22:36,690 When it's a new technology, we just don't know what it's going 431 00:22:36,690 --> 00:22:39,690 to be like. We we don't, you know, and it is 432 00:22:39,690 --> 00:22:43,370 a little bit like. When we were, you know, putting 433 00:22:43,370 --> 00:22:45,690 the right to bear arms in the Constitution without knowing 434 00:22:45,690 --> 00:22:48,250 what arms meant in the future, right? 435 00:22:48,770 --> 00:22:51,850 Like how, how, how dangerous they may or may not be. 436 00:22:52,130 --> 00:22:55,250 So I think that I'm just a little nervous about, you know, 437 00:22:55,650 --> 00:22:59,770 assuming too much about whether we will need regulation in the 438 00:22:59,770 --> 00:23:02,410 future. I mean to come back to you at 439 00:23:02,410 --> 00:23:06,050 Kowski for a second, I mean he is sort of the the progenitor of 440 00:23:06,050 --> 00:23:09,140 this whole AI. Dumerism, concept or whatever. 441 00:23:09,140 --> 00:23:11,660 I mean, I've read a bunch of his stuff he seemed to almost like 442 00:23:11,660 --> 00:23:14,700 at one point, threaten blowing up GPU. 443 00:23:14,900 --> 00:23:17,940 Right, Yeah. Well, he had a time article that 444 00:23:17,940 --> 00:23:23,540 he was saying in his world of protecting humanity, like one of 445 00:23:23,540 --> 00:23:26,500 the only things you would have to do would be to. 446 00:23:27,480 --> 00:23:30,400 You know, protect, you know, prevent rogue data centers from 447 00:23:30,400 --> 00:23:32,240 training models. And, you know, it'd be 448 00:23:32,240 --> 00:23:34,520 justifiable to like, bomb them essentially. 449 00:23:34,920 --> 00:23:39,800 And and he was saying, you know, the entire kind of geopolitical, 450 00:23:41,520 --> 00:23:45,200 you know, consensus should be that it's worse to train these 451 00:23:45,200 --> 00:23:48,640 models at some level than it is to, like, worry about nuclear 452 00:23:49,360 --> 00:23:51,360 proliferation. I think. 453 00:23:52,540 --> 00:23:54,820 I mean, I having read a bunch of his stuff, I feel like he has 454 00:23:54,820 --> 00:23:58,380 like pretty well convinced me of sort of all all three of these 455 00:23:58,380 --> 00:23:59,700 things. I guess that will have a super 456 00:23:59,700 --> 00:24:03,700 intelligence that it'll it'll be it'll it'll some of them at 457 00:24:03,700 --> 00:24:07,140 least will desire to harm us and that they'll figure out some way 458 00:24:07,140 --> 00:24:09,500 to do it. I guess I'm sort of like I'm a 459 00:24:09,500 --> 00:24:11,980 little bit of a fatalist in the sense that I I think this is 460 00:24:11,980 --> 00:24:14,660 basically an unstoppable train at this point that has like left 461 00:24:14,660 --> 00:24:16,220 the station. Like I don't, I don't think. 462 00:24:16,690 --> 00:24:20,650 Bombing data centers is like a viable solution to this problem 463 00:24:20,650 --> 00:24:24,890 or breaking GPUs or whatever or certainly not in the like the 464 00:24:25,010 --> 00:24:27,770 human political universe that we actually operate in today. 465 00:24:28,690 --> 00:24:34,290 I think my, my attitude is just that like luckily, I guess my 466 00:24:34,690 --> 00:24:36,610 the future is really hard to predict, right? 467 00:24:36,690 --> 00:24:39,970 I mean, I know that's like a very obvious point, but if you 468 00:24:39,970 --> 00:24:42,650 had told someone in 1940 that we were going to invent nuclear 469 00:24:42,650 --> 00:24:44,690 bombs and they had really thought through all the. 470 00:24:45,450 --> 00:24:48,610 You know, the potential problems in the future caused by those 471 00:24:48,610 --> 00:24:51,850 devices, I think they could have been pretty fatalistic and I 472 00:24:51,850 --> 00:24:53,850 think that would have been a pretty accurate read on like how 473 00:24:53,850 --> 00:24:55,970 bad it was that we invented nuclear bombs, right? 474 00:24:56,250 --> 00:24:58,210 Like knock on wood, so far we're still here. 475 00:24:58,210 --> 00:25:01,370 So my sort of my fear is that all the stuff you you had cows 476 00:25:01,370 --> 00:25:02,450 and some people it's going to happen. 477 00:25:02,450 --> 00:25:06,730 Nuclear bombs are good, like they brought about like peace. 478 00:25:06,730 --> 00:25:08,930 We haven't had world wars, you know, But my hope. 479 00:25:09,010 --> 00:25:13,190 Is that we get lucky and that. Something about this future that 480 00:25:13,190 --> 00:25:16,310 was that Yukowski or the AI doomers are sketching out is is 481 00:25:16,470 --> 00:25:19,990 there's some missing link in the logical chain or or there's some 482 00:25:20,470 --> 00:25:24,870 piece of the way deterrence or or alignment or whatever ends up 483 00:25:24,870 --> 00:25:27,910 working that that we we get our ass saved. 484 00:25:27,910 --> 00:25:29,510 But I'm kind of betting on luck at this point. 485 00:25:29,510 --> 00:25:31,990 Like, I don't really think we're capable of stopping this. 486 00:25:31,990 --> 00:25:34,270 Like just if we're all fatalist basically. 487 00:25:35,030 --> 00:25:38,070 I mean there is like the human cloning, you know, people 488 00:25:38,070 --> 00:25:40,630 brought this up to me. Like human cloning is an example 489 00:25:41,220 --> 00:25:44,980 where people have, you know, America isn't cloning humans. 490 00:25:44,980 --> 00:25:49,540 It seems like we could like why? Why is that an area where self 491 00:25:49,540 --> 00:25:52,780 regulation seems much more possible? 492 00:25:53,260 --> 00:25:56,980 I feel like there are areas where we have stopped. 493 00:25:56,980 --> 00:26:00,340 That's still pretty hard, right? I mean, like, I bet that will 494 00:26:00,340 --> 00:26:02,300 happen sometime in the next 50 to 100 years, right? 495 00:26:02,300 --> 00:26:04,060 I don't know. I just feel like I don't know 496 00:26:04,060 --> 00:26:05,740 how much desire there is for that. 497 00:26:06,080 --> 00:26:08,640 But I I feel like once it's doable, people will do it. 498 00:26:08,640 --> 00:26:11,480 My last thinking on one positive future scenario that I sort of 499 00:26:11,800 --> 00:26:15,480 believe is possible is like, you know, we live on this planet 500 00:26:15,480 --> 00:26:18,160 with lots of other organisms and creatures as you guys described. 501 00:26:18,400 --> 00:26:21,400 You know, we have pets, we have dogs, whatever. 502 00:26:21,600 --> 00:26:23,360 Like. Why do we have dogs? 503 00:26:23,360 --> 00:26:24,960 Like, is there any really good reason? 504 00:26:24,960 --> 00:26:27,600 Like the question is like, could we end up being the pets of the 505 00:26:27,760 --> 00:26:29,200 AI? Like, could we could they want 506 00:26:29,200 --> 00:26:30,560 to keep us alive for some reason? 507 00:26:30,560 --> 00:26:31,920 And that we're a little bit different? 508 00:26:31,920 --> 00:26:34,360 We're a little bit interesting to them, more fun to have around 509 00:26:34,360 --> 00:26:37,200 in some scenarios, but like, in the end we're the AI's pets, you 510 00:26:37,200 --> 00:26:39,640 know? Or or maybe we're the ants and 511 00:26:39,640 --> 00:26:41,880 they just don't care to kill us all because it would be such a 512 00:26:41,880 --> 00:26:45,040 pain, You know, right? I think that's more likely than 513 00:26:45,040 --> 00:26:46,040 not. Right. 514 00:26:46,080 --> 00:26:49,600 I think that is the the upside scenario is like we're we're 515 00:26:49,600 --> 00:26:51,760 pets or we're ants or something compared to this. 516 00:26:51,920 --> 00:26:55,160 More likely than not just the general, like the AI does not 517 00:26:55,160 --> 00:26:57,360 hate us, you know what I mean? Like both. 518 00:26:57,480 --> 00:27:01,240 We helped create it. Like we set a lot of its initial 519 00:27:01,240 --> 00:27:05,530 value system, presumably like it, you know, relies us on us 520 00:27:05,530 --> 00:27:08,250 for a lot of things early on, like I don't maybe a lot of 521 00:27:08,330 --> 00:27:10,410 reasons. Yeah, maybe we're just different 522 00:27:10,410 --> 00:27:13,010 in some sort of interesting way, right, in that we're not, we're 523 00:27:13,010 --> 00:27:15,650 not wired exactly the same. And so you know it's it's just 524 00:27:15,650 --> 00:27:17,610 interesting to have us around because we act a little bit 525 00:27:17,610 --> 00:27:19,370 differently or or something like that. 526 00:27:19,410 --> 00:27:22,130 And AI is not like an evolutionary being, like, you 527 00:27:22,130 --> 00:27:25,170 know, we are driven by these like evolutionary prerogatives 528 00:27:25,170 --> 00:27:29,930 that make us sort of competitive and you know, worried about 529 00:27:29,930 --> 00:27:32,210 other other sort of genetic codes. 530 00:27:32,860 --> 00:27:35,100 It it's just such a separate type of thing. 531 00:27:35,100 --> 00:27:40,060 It's like hard to extrapolate from, I don't know, even beyond 532 00:27:40,060 --> 00:27:43,860 humans like animals to what an AI would be like. 533 00:27:44,860 --> 00:27:50,100 OK, so so I guess we all agree we're all gonna die or be a as 534 00:27:50,100 --> 00:27:53,140 pets. But it seems like we we haven't 535 00:27:53,140 --> 00:27:56,780 really narrowed in on how likely this is to occur in any 536 00:27:56,780 --> 00:27:59,980 meaningful time frame. Like I think that to me is 537 00:27:59,980 --> 00:28:02,460 pretty interesting, you know, if this happens. 538 00:28:02,950 --> 00:28:06,950 In 10 years or even 20 years or, you know, I think that's pretty 539 00:28:06,950 --> 00:28:11,030 different from how I might go ahead and live my life, you 540 00:28:11,030 --> 00:28:13,510 know, for the next decade, I guess. 541 00:28:13,710 --> 00:28:15,590 How about you guys? Like, do you think it's even 542 00:28:15,790 --> 00:28:18,790 worth thinking about that you know right now and how you live 543 00:28:18,790 --> 00:28:22,830 your life or not really? I don't know. 544 00:28:23,030 --> 00:28:24,150 I don't know. I just. 545 00:28:24,390 --> 00:28:25,390 I just. I can't. 546 00:28:25,470 --> 00:28:28,270 I can't decide like if I would make any just different ways. 547 00:28:28,270 --> 00:28:31,390 I think, you know, being close to it is appealing and like you 548 00:28:31,390 --> 00:28:35,610 know, this is a technology that I believe in unlike crypto. 549 00:28:35,610 --> 00:28:39,210 So it's like oh run to where, you know, the actual interesting 550 00:28:39,210 --> 00:28:42,810 thing that could really revolutionize humanity is like I 551 00:28:43,090 --> 00:28:45,050 I want to be around it, I guess selfishly. 552 00:28:45,050 --> 00:28:47,690 So, you know, host an AI conference with you guys, run. 553 00:28:48,130 --> 00:28:54,330 Into the run, into the burning building, Will AI kill us all? 554 00:28:54,330 --> 00:28:59,650 In some ways is great PR for the AI industry, even though it's 555 00:28:59,650 --> 00:29:03,990 sort of bleak because it suggests it takes as a premise 556 00:29:03,990 --> 00:29:08,390 that artificial intelligence is it a really amazing point and 557 00:29:08,430 --> 00:29:12,150 super powerful and is poised to like be extremely disruptive. 558 00:29:12,150 --> 00:29:15,310 And so if you're an investor and you're like, well, I can't save 559 00:29:15,310 --> 00:29:17,630 the world, I will try to profit off of it. 560 00:29:17,790 --> 00:29:20,270 In the good scenario, Will AI Kill us all? 561 00:29:20,270 --> 00:29:24,190 Is definitely a good motivator to like deploy capital into AI. 562 00:29:24,190 --> 00:29:27,790 Like what do you take of the fact that sort of the dumerism 563 00:29:27,790 --> 00:29:34,420 is like great PR for the actual existing, you know, for vector 564 00:29:34,420 --> 00:29:37,580 database companies, it's a it's a good message, you know, I 565 00:29:37,580 --> 00:29:38,500 don't know, what do you make of this? 566 00:29:38,500 --> 00:29:41,820 This is like war profiteering. Is the motivation here? 567 00:29:42,100 --> 00:29:45,340 No, I'm just saying it's like, you know, the media loves it. 568 00:29:45,340 --> 00:29:47,740 Like Ezra Klein, they will, you know talks about it all the 569 00:29:47,740 --> 00:29:49,860 time. But like in some ways it is a 570 00:29:49,860 --> 00:29:55,530 good it's good marketing for the IT could be just like you know a 571 00:29:55,530 --> 00:29:58,570 fairly mundane technology that we're nowhere close to the jump 572 00:29:58,570 --> 00:30:01,250 TAGI like the Reed Hoffman interview. 573 00:30:01,370 --> 00:30:04,930 He's not he's not willing to sort of commit to any near time 574 00:30:05,090 --> 00:30:09,570 horizon that that AGI is coming. It could just be like it's a 575 00:30:09,570 --> 00:30:13,770 fairly incremental technology. I think it's good fundraising 576 00:30:13,770 --> 00:30:17,080 pitch to say hey like. We're trying to prevent, you 577 00:30:17,080 --> 00:30:20,120 know, AI from killing us all. Like, we're the good AI guys, 578 00:30:20,120 --> 00:30:21,600 right? I mean, like, yeah, you're like, 579 00:30:21,600 --> 00:30:24,760 this is life or death doing this technology, that is. 580 00:30:24,760 --> 00:30:26,440 So powerful. Powerful. 581 00:30:26,680 --> 00:30:29,320 So people are worried that it's going to kill us all, you know, 582 00:30:29,400 --> 00:30:33,160 I mean, I feel like that's a good, like, you know, it's a 583 00:30:33,160 --> 00:30:35,040 it's a conversation that's good for industry. 584 00:30:35,040 --> 00:30:38,200 Can is there an analogy to another industry where you think 585 00:30:38,360 --> 00:30:39,960 that this has occurred? I'm just. 586 00:30:40,590 --> 00:30:44,630 I'm not as convinced, I guess. Like, is it great for the oil 587 00:30:44,630 --> 00:30:48,590 and gas industry right now that, you know, climate change is? 588 00:30:48,590 --> 00:30:52,590 Well, you could argue, you know, like sometimes, like when 589 00:30:52,590 --> 00:30:55,670 parents, like, freak out about, like, teen stuff, that's 590 00:30:55,670 --> 00:30:58,430 obviously not that dangerous for them. 591 00:30:58,630 --> 00:31:01,710 That in some ways it, like, drives teens closer to it. 592 00:31:01,710 --> 00:31:04,790 And that it's like this is totally misunderstood by the 593 00:31:04,910 --> 00:31:08,590 sort of authorities like, but it it reinforced, it gets it in the 594 00:31:08,590 --> 00:31:10,910 news all the time. It reinforces that this is 595 00:31:10,910 --> 00:31:14,150 something that's sort of going on, you know? 596 00:31:14,470 --> 00:31:17,630 I mean I think it's like pros and cons to your point, like I 597 00:31:17,630 --> 00:31:20,670 think it might drive a lot more regulation in a shorter time 598 00:31:20,670 --> 00:31:22,870 frame than what they would have otherwise and. 599 00:31:23,400 --> 00:31:25,680 To your other point, and it will probably be kind of dumb 600 00:31:25,680 --> 00:31:27,880 regulation. So, like, you know, I don't 601 00:31:27,880 --> 00:31:29,840 know. You know, the European Union has 602 00:31:29,840 --> 00:31:33,480 made a terrible job of doing privacy regulations and just 603 00:31:33,480 --> 00:31:36,920 excited to see what they gin up for AI, I mean, I think. 604 00:31:37,540 --> 00:31:40,060 Isn't Italy banned Chachi PT right now? 605 00:31:40,140 --> 00:31:43,100 It's like it's like already you're like, God, these European 606 00:31:43,100 --> 00:31:44,700 regulators are just off the chain. 607 00:31:45,860 --> 00:31:48,020 Yeah, I mean so pros and cons, right. 608 00:31:48,260 --> 00:31:51,980 Banned in Italy maybe you can raise a couple $100 million with 609 00:31:51,980 --> 00:31:53,780 a with a slide deck at this point. 610 00:31:53,780 --> 00:31:56,140 So you know, you get you get both, I guess. 611 00:31:56,340 --> 00:32:00,420 I guess I don't know if I if I was Sam Almond, like would I 612 00:32:00,420 --> 00:32:05,760 want the level of AI doomerism? Occurring right now to continue, 613 00:32:05,760 --> 00:32:08,000 Probably not. Probably not to be clear, just 614 00:32:08,120 --> 00:32:11,560 my actual position is that like it things are very exciting 615 00:32:11,560 --> 00:32:14,200 right now. That this is a legitimate 616 00:32:14,200 --> 00:32:18,080 question that will AIS kill us all is legitimately grounded in 617 00:32:18,080 --> 00:32:21,880 like all, you know, sci-fi content and thought experiments 618 00:32:21,880 --> 00:32:24,320 about where this is going. So I don't think it's made-up, 619 00:32:24,320 --> 00:32:28,760 but I think a happy coincidence is that this sort of question is 620 00:32:28,760 --> 00:32:31,000 fundamentally good PR. That's all I'm saying. 621 00:32:31,000 --> 00:32:33,160 So I think 1 interesting topic is. 622 00:32:33,700 --> 00:32:36,900 AI and sci-fi, right? Obviously a number of the most 623 00:32:36,900 --> 00:32:40,740 successful films, you know, TV shows, books, everything in 624 00:32:40,740 --> 00:32:44,260 history have been built around sci-fi, which often is driven by 625 00:32:44,260 --> 00:32:46,700 this idea that there's an evil killer AI basically, right? 626 00:32:46,700 --> 00:32:49,580 And I think the examples that come on top of my mind, and you 627 00:32:49,580 --> 00:32:52,460 guys might have other ones, are like, you know The Matrix, you 628 00:32:52,460 --> 00:32:56,860 know I Robot, as you mentioned the Terminator series, Blade 629 00:32:56,860 --> 00:32:59,380 Runner. Blade Runner I you know, her 630 00:32:59,380 --> 00:33:02,500 being a sort of off kilter example, but still important I 631 00:33:02,500 --> 00:33:04,830 think. And what's interesting is almost 632 00:33:04,830 --> 00:33:08,230 all of these ones other than her that I mentioned, it involves 633 00:33:08,270 --> 00:33:12,030 robots as well as AI, right? It's it's like the humanoid 634 00:33:12,030 --> 00:33:17,030 version of the AI is essential to you know, killing us all 635 00:33:17,030 --> 00:33:18,550 right, we're we're in Terminator. 636 00:33:18,550 --> 00:33:19,710 There's a good. Storyteller. 637 00:33:20,150 --> 00:33:21,510 Yeah, right. Well, yeah, it might just be 638 00:33:21,510 --> 00:33:23,510 more cinematic to fight a big metal robot. 639 00:33:23,510 --> 00:33:26,710 So, you know, that's something. But I guess. 640 00:33:26,830 --> 00:33:29,790 And then in her you sort of have this. 641 00:33:29,830 --> 00:33:33,960 Scarlett Johansson AI That. At the end kind of ascends to a 642 00:33:33,960 --> 00:33:38,040 higher plane and like leaves leaves the main character 643 00:33:38,040 --> 00:33:43,120 Joaquin Phoenix behind, right. And I think that I to me that 644 00:33:43,120 --> 00:33:46,400 representation seems more likely than the killer robots one and 645 00:33:46,400 --> 00:33:49,040 that the as I said maybe we're just pets and they just sort of 646 00:33:49,040 --> 00:33:51,760 get bored with us and and leave us behind or or we're the little 647 00:33:51,760 --> 00:33:53,360 you know dogs that they leave alive right. 648 00:33:53,360 --> 00:33:56,720 But I guess, what do you guys think about, well, A, what's 649 00:33:56,720 --> 00:33:58,640 your favorite sci-fi representation? 650 00:33:58,960 --> 00:34:02,430 And then B. Do you think robots are an 651 00:34:02,430 --> 00:34:06,230 essential part to these AI kill us all narratives or and or will 652 00:34:06,350 --> 00:34:09,270 the robot elements integrated with AI happen anytime soon. 653 00:34:10,030 --> 00:34:12,630 Well, I I I can answer with my favorite. 654 00:34:12,630 --> 00:34:16,070 I mean I think my the matrix matrix is my favorite and. 655 00:34:16,110 --> 00:34:19,270 I was going to say The Matrix. That's all The Matrix buys into 656 00:34:19,270 --> 00:34:21,469 exactly what you're saying. The Matrix doesn't actually say 657 00:34:21,469 --> 00:34:24,550 it's killer robots. It it finds a way to basically 658 00:34:24,550 --> 00:34:27,389 put killer robots on the screen. So it's a fun movie, but it's 659 00:34:27,389 --> 00:34:29,989 like, oh, it's a computer program, which is obviously the 660 00:34:29,989 --> 00:34:33,190 world I feel like a super intelligence would live in. 661 00:34:33,190 --> 00:34:37,429 Like that of what does it really need to manifest in the real 662 00:34:37,429 --> 00:34:40,469 world that much? So yeah, I, I'm. 663 00:34:40,750 --> 00:34:43,790 And obviously I'm interested, you know, for the following 664 00:34:43,790 --> 00:34:46,389 conversation with somebody who's written a lot about robot 665 00:34:46,389 --> 00:34:48,429 apocalypse. Because yeah, my intuition is 666 00:34:48,429 --> 00:34:51,770 that they're going to exist mostly mostly in the digital 667 00:34:51,770 --> 00:34:53,489 world. Terminator two is the other one 668 00:34:53,489 --> 00:34:58,210 where I think the robot part is interesting, but I don't know 669 00:34:58,210 --> 00:35:02,250 how much you remember about the historical world building of the 670 00:35:02,450 --> 00:35:06,850 AI takeover which is Skynet is an AI that achieves super 671 00:35:06,850 --> 00:35:10,780 intelligence that then. Takes over the nuclear arsenal 672 00:35:10,780 --> 00:35:14,660 of the United States or and then bombs Russia or whatever, and 673 00:35:14,660 --> 00:35:16,460 then we end up in nuclear apocalypse basically, right? 674 00:35:16,580 --> 00:35:20,100 So the initial manifestation of the AI apocalypse in in 675 00:35:20,100 --> 00:35:24,700 Terminator and Terminator Two is is a purely software driven 676 00:35:25,300 --> 00:35:27,620 death scenario, right? There's no, there's no need for 677 00:35:27,620 --> 00:35:30,420 robots in the sort of Skynet back story, right? 678 00:35:31,020 --> 00:35:34,420 It can hack into a system that ends up destroying the world, 679 00:35:34,580 --> 00:35:35,740 which like. Realistic, right? 680 00:35:35,740 --> 00:35:38,980 Because then you don't need any physical elements, right? 681 00:35:38,980 --> 00:35:41,060 Of of the sort of AI doomer narrative, right? 682 00:35:41,060 --> 00:35:43,700 You don't need the robots, you don't need the the crazy 683 00:35:43,700 --> 00:35:47,340 biohacking stuff or whatever. You just need to hack into the 684 00:35:47,340 --> 00:35:49,940 Pentagon essentially. And it's like game over at that 685 00:35:49,940 --> 00:35:52,900 point, right? So I always thought that was a 686 00:35:52,900 --> 00:35:57,900 very insightful perspective on how an evil AI could kill us all 687 00:35:57,900 --> 00:36:00,220 and then it would just be a pure software takeover, right? 688 00:36:02,960 --> 00:36:05,520 James, did you have a favorite? Well, I I wanted to dig. 689 00:36:05,520 --> 00:36:07,640 Into the Matrix a little bit more, yeah. 690 00:36:08,520 --> 00:36:14,840 But I think, you know, a lot rests on whether we think that's 691 00:36:14,840 --> 00:36:18,400 humanity's future as well. Like are we going to be like 692 00:36:18,400 --> 00:36:23,360 almost, you know, by choice plugging ourselves into the 693 00:36:23,360 --> 00:36:25,640 Matrix? Like is is this neural link kind 694 00:36:25,640 --> 00:36:29,720 of interface you know within that 50 to 100 year time frame 695 00:36:29,720 --> 00:36:34,950 or or even sooner, right? You know does does our normal 696 00:36:34,950 --> 00:36:40,750 day-to-day life as humans you know become more of a simulation 697 00:36:40,910 --> 00:36:43,510 you know in in in our lifetimes. I think that's pretty 698 00:36:43,510 --> 00:36:47,550 interesting for how the future unfolds. 699 00:36:48,910 --> 00:36:53,350 So do you guys believe that is possible like or do you think 700 00:36:53,350 --> 00:36:56,830 you know brain interfaces are sort of so far away. 701 00:36:56,830 --> 00:37:01,630 I think Brain Interface is 30 + 30 plus years. 30 plus years 702 00:37:01,630 --> 00:37:04,100 away. I kind of agree. 703 00:37:04,740 --> 00:37:08,020 I mean, I think it's possible maybe in our lifetime, which, 704 00:37:08,460 --> 00:37:12,340 you know, God willing is 50 to 60 more years, but it does seem 705 00:37:12,340 --> 00:37:15,260 like we're pretty far away on neural interfaces. 706 00:37:15,380 --> 00:37:18,580 But you know, who knows? I feel like I'm I'm more excited 707 00:37:18,580 --> 00:37:22,140 to come back to the previous episode about like glasses and 708 00:37:22,140 --> 00:37:25,220 maybe contact lenses at some point as a sort of, you know, 709 00:37:25,220 --> 00:37:28,460 augmented interface for for existence rather than like. 710 00:37:29,040 --> 00:37:32,400 Plugging in, but you never know. I mean, obviously Elon believes 711 00:37:32,400 --> 00:37:34,280 in it. So Elon believes like this is. 712 00:37:34,280 --> 00:37:38,760 The path right out of the AI dumerism that we've that we've 713 00:37:38,760 --> 00:37:40,120 been talking about, We are augmented. 714 00:37:40,360 --> 00:37:42,200 Basically. We sort of merge our brains. 715 00:37:42,200 --> 00:37:45,920 Yeah, with with AIS and that probably requires some sort of 716 00:37:47,200 --> 00:37:50,760 brain interface or you know matrix like environment that 717 00:37:50,760 --> 00:37:53,640 you're living in seems hard to be able to. 718 00:37:53,640 --> 00:37:55,760 I agree I'm not. I'm not a problem. 719 00:37:55,800 --> 00:37:59,200 Sort of like the you you think that that basically? 720 00:37:59,200 --> 00:38:02,360 Is not sufficient to. I just don't think it answers 721 00:38:02,360 --> 00:38:03,920 much. The like, will it kill us? 722 00:38:03,920 --> 00:38:07,040 Like, oh, it's like yielding our brains to it almost means like, 723 00:38:07,040 --> 00:38:08,840 are we even running the show like it? 724 00:38:08,840 --> 00:38:11,440 It raises a whole bunch of other questions where we could be 725 00:38:11,440 --> 00:38:13,600 undermined just from the connection. 726 00:38:13,600 --> 00:38:15,520 Independent. Like it's true, Yeah. 727 00:38:16,040 --> 00:38:19,720 And it also it also like pretty quickly if you believe in super 728 00:38:19,720 --> 00:38:22,670 intelligence kind of. Defeats the purpose of having 729 00:38:22,670 --> 00:38:25,390 that brain power. Like why? 730 00:38:25,390 --> 00:38:26,510 Why do you need? Yeah, why would they? 731 00:38:26,630 --> 00:38:28,710 Why would they want? To be hanging out in our brains, 732 00:38:28,750 --> 00:38:31,470 it's like, it's like, oh cool, I've got this brain that's not 733 00:38:31,470 --> 00:38:34,390 as good as mine. Like, maybe consciousness is 734 00:38:34,390 --> 00:38:37,190 this unique human charm that nobody else has, and we can give 735 00:38:37,190 --> 00:38:42,150 the machines a taste of it. I don't know, just succinctly. 736 00:38:42,590 --> 00:38:46,630 Do you think in the next 100 years AI will kill us all? 737 00:38:46,710 --> 00:38:51,350 Yes or no? I will say no. 738 00:38:52,590 --> 00:38:55,550 I think it will kill. Some of us, but not all of us. 739 00:38:56,510 --> 00:38:57,950 That's a good way to. Think that's a good Yeah. 740 00:38:57,950 --> 00:39:04,550 I I think like more than 1000, less than 10,000. 741 00:39:04,550 --> 00:39:07,430 Oh, more I don't. Know I think more that's a very 742 00:39:07,510 --> 00:39:09,030 tight. Yeah. 743 00:39:09,070 --> 00:39:11,950 That's not a good range. I I think I was smart to give 744 00:39:11,950 --> 00:39:12,310 it. Yeah. 745 00:39:12,310 --> 00:39:13,910 Yeah, yeah, yeah. Some of well, you came up with 746 00:39:13,910 --> 00:39:15,630 some of us. Some of us is the right feels 747 00:39:15,630 --> 00:39:17,830 like clearly. It feels like even a mistake. 748 00:39:17,870 --> 00:39:21,340 I will AI intentionally well it was kill us all. 749 00:39:21,940 --> 00:39:23,780 I say definitely no kill us all that's. 750 00:39:23,860 --> 00:39:25,900 I say definitely there'll still be some humans at the. 751 00:39:25,940 --> 00:39:29,980 End I can't say definitely. I mean it seems pretty UN, you 752 00:39:30,020 --> 00:39:33,700 know, well if we do all get killed, we'll be dead and so 753 00:39:33,700 --> 00:39:35,580 there's nothing to gain from the prediction. 754 00:39:35,580 --> 00:39:39,100 Whereas if we live, I was correct and so it was there was 755 00:39:39,100 --> 00:39:41,700 good utility in the prediction. So it's I don't know why I would 756 00:39:41,700 --> 00:39:44,580 ever go the other direction. This is like Pascal's wager. 757 00:39:44,580 --> 00:39:50,540 With like super intelligent AI, like, all right, this was fun. 758 00:39:50,540 --> 00:39:53,660 I mean, this is this is a dream. Getting to hang out and talk 759 00:39:53,660 --> 00:39:55,100 about this and call it work. Welcome. 760 00:39:59,460 --> 00:40:02,660 Hey, welcome. To the second segment, here I've 761 00:40:02,660 --> 00:40:06,900 got author Daniel H Wilson, author of How to Survive a Robot 762 00:40:06,900 --> 00:40:11,140 Uprising, Where's My Jetpack, How to Build a Robot Army, and 763 00:40:11,140 --> 00:40:15,060 Robo Apocalypse. The title of this or at least 764 00:40:15,060 --> 00:40:19,780 the working title of this episode is Will AI Kill us All. 765 00:40:19,780 --> 00:40:24,460 So given your interest and and work we wanted to talk with 766 00:40:24,460 --> 00:40:28,540 someone who's really like been thinking about this for a long 767 00:40:28,540 --> 00:40:31,420 time. I'm I'm curious like what what 768 00:40:31,420 --> 00:40:35,460 first got you interest in this sort of dystopian question of 769 00:40:35,700 --> 00:40:38,460 sort of the machines coming for humanity. 770 00:40:39,180 --> 00:40:41,980 Yeah, well, so I just grew. Up with science fiction, right? 771 00:40:42,060 --> 00:40:45,810 Like, like a lot of people, so. Initially, I was just interested 772 00:40:45,810 --> 00:40:49,210 in reading, you know, any type of science fiction I could read 773 00:40:49,330 --> 00:40:52,210 or watch in any type of movie. I just loved robots. 774 00:40:52,210 --> 00:40:57,170 And so that's ultimately though I loved robots so much that I 775 00:40:57,170 --> 00:40:59,730 studied robotics. So I ended up going to Carnegie 776 00:40:59,730 --> 00:41:02,170 Mellon. I did a a whole PhD in robotics. 777 00:41:02,170 --> 00:41:05,330 And while I was at Carnegie Mellon, you know, I'm surrounded 778 00:41:05,330 --> 00:41:07,530 by roboticists, I'm surrounded by robots. 779 00:41:07,530 --> 00:41:11,090 We're in the high Bay world. We're just in the lab. 780 00:41:11,610 --> 00:41:15,250 And nobody was really trying to build robots that would destroy 781 00:41:15,250 --> 00:41:16,650 the world. I noticed. 782 00:41:17,730 --> 00:41:20,530 Nobody says they're trying. Generally, nobody comes out and 783 00:41:20,530 --> 00:41:22,770 says. It but like, there's this really 784 00:41:22,770 --> 00:41:27,410 stark difference between how robots are portrayed in pop 785 00:41:27,410 --> 00:41:32,050 culture and how they're and and the actual you know, the actual 786 00:41:32,210 --> 00:41:34,930 mechanics of building robots and why people are building them and 787 00:41:34,930 --> 00:41:37,610 how they're explaining, you know why they need the money to build 788 00:41:37,610 --> 00:41:39,300 them and all that and so. Really. 789 00:41:39,300 --> 00:41:41,780 They just have this super bad, you know, reputation. 790 00:41:41,780 --> 00:41:44,660 Right. And so I thought that was funny. 791 00:41:45,580 --> 00:41:48,620 So then when I was still in grad school, I wrote How to Survive a 792 00:41:48,620 --> 00:41:51,620 Robot Uprising, where I was just like, all right, I'm gonna take 793 00:41:51,620 --> 00:41:53,820 them serious, right? Just, OK, fine. 794 00:41:53,860 --> 00:41:55,380 If this is what everyone's expecting. 795 00:41:55,620 --> 00:41:59,420 So I went to, you know, the people that were building legs, 796 00:41:59,420 --> 00:42:01,740 and I asked, you know, how would you trip a robot? 797 00:42:01,740 --> 00:42:04,220 How would you get away? I went to the people doing 798 00:42:04,220 --> 00:42:06,340 sensors, the people doing all different things. 799 00:42:07,080 --> 00:42:09,280 And I just asked those questions and I put it all, you know, 800 00:42:09,320 --> 00:42:13,040 tongue in cheek into this book. Of course. 801 00:42:13,040 --> 00:42:15,840 Then you start to look at like actual military applications. 802 00:42:15,840 --> 00:42:20,040 You start to see how robotics and AI are, are, you know, being 803 00:42:20,040 --> 00:42:23,600 weaponized in some cases. And then, you know, that was 804 00:42:23,600 --> 00:42:27,320 more robo apocalypse much. No tongue in cheek like I I was 805 00:42:27,320 --> 00:42:29,360 like, OK, let's. You're starting to believe it. 806 00:42:29,360 --> 00:42:32,400 You're like, OK, this. No, I never, really. 807 00:42:32,400 --> 00:42:34,160 I mean, honestly, I like. I like. 808 00:42:34,590 --> 00:42:38,150 The killer robot meme really, for me, just gives you a lot of 809 00:42:38,150 --> 00:42:42,630 latitude to to to think about humanity and think about, you 810 00:42:42,630 --> 00:42:45,150 know, what makes us people, what makes them robots. 811 00:42:45,150 --> 00:42:49,110 You know, do you do? You do you think like as a 812 00:42:49,110 --> 00:42:51,590 person and somebody? Do you think robots are likely 813 00:42:51,590 --> 00:42:54,430 to kill human beings? Or, like, do you see this as a 814 00:42:54,430 --> 00:42:57,030 fictional actress? No, I well, look, I. 815 00:42:57,230 --> 00:43:00,350 I kind of ascribed to the Lately I've been thinking a lot about 816 00:43:00,350 --> 00:43:02,350 something called the psychotic ape. 817 00:43:02,830 --> 00:43:05,790 Theory or the the killer ape theory, which is think about the 818 00:43:05,830 --> 00:43:09,550 opening scenes of like Space Odyssey 2000, 2001. 819 00:43:09,550 --> 00:43:13,310 Like it's those the, you know, they touch the the monolith. 820 00:43:13,310 --> 00:43:15,230 It gives them a leap forward in evolution. 821 00:43:15,230 --> 00:43:17,750 And what do the apes do? Well, they figure out how to use 822 00:43:17,750 --> 00:43:19,510 weapons to bash each other's brains in. 823 00:43:19,590 --> 00:43:23,750 And there's this kind of notion that humanity sort of evolves 824 00:43:23,750 --> 00:43:26,630 technologically when we're trying to kill each other or 825 00:43:26,630 --> 00:43:29,270 stop each other from killing each, you know, and so. 826 00:43:30,210 --> 00:43:32,370 So, yeah, so I've been thinking a lot about that and and 827 00:43:32,370 --> 00:43:35,210 basically you look at that and you realize we will use any 828 00:43:35,210 --> 00:43:38,810 technology to kill each other. So really it's the psychotic ape 829 00:43:38,890 --> 00:43:42,450 that you need to worry about. It's not the and to a lesser 830 00:43:42,450 --> 00:43:46,210 extent the, you know, the capitalist ape, the known killer 831 00:43:46,210 --> 00:43:49,610 species versus the sort of imagined killer who's killed 832 00:43:49,610 --> 00:43:51,490 more people than anybody. Else, right. 833 00:43:51,770 --> 00:43:55,130 You know you're going to want to look at the yeah, the person 834 00:43:55,930 --> 00:43:59,760 across the table from you, but. I think that in terms of being 835 00:43:59,760 --> 00:44:03,440 used as weapons, you know, obviously robots can be 836 00:44:03,440 --> 00:44:08,160 extremely dangerous in that way. I mean I'm curious sort of you 837 00:44:08,160 --> 00:44:10,960 know as a technologist somebody's thought about this 838 00:44:10,960 --> 00:44:15,280 like robots versus like the large language model, sort of 839 00:44:15,280 --> 00:44:19,000 totally software based being that maybe reaches super 840 00:44:19,000 --> 00:44:23,160 intelligence like I I see why for like especially film, why 841 00:44:23,160 --> 00:44:25,560 like robots are great because you can see them. 842 00:44:25,910 --> 00:44:28,670 I'm curious in terms of like the actual like thought exercise and 843 00:44:28,670 --> 00:44:33,190 what you think is likely whether you imagine robots or the sort 844 00:44:33,190 --> 00:44:35,790 of potential threat being just a software. 845 00:44:36,300 --> 00:44:38,420 Yeah, well, so I think of this as like a. 846 00:44:38,420 --> 00:44:41,980 Consumer like a like a product design issue, right? 847 00:44:41,980 --> 00:44:44,460 Like, you know, I mean, look, we're human beings are 848 00:44:44,460 --> 00:44:46,500 comfortable with a certain amount of risk in our lives. 849 00:44:46,500 --> 00:44:50,820 You know what, 40,000 people get killed driving around in cars, 850 00:44:51,020 --> 00:44:54,300 you know, every year I've got a garbage disposal in my kitchen. 851 00:44:54,300 --> 00:44:57,900 I mean, if I put my hand in the wrong place in my own home, like 852 00:44:57,900 --> 00:44:59,260 I won't have a hand anymore, right? 853 00:44:59,580 --> 00:45:04,110 But so but there's also the. Devices are built in order to 854 00:45:04,110 --> 00:45:06,150 try to make them as safe as possible, right? 855 00:45:06,150 --> 00:45:09,510 So if you think of that as a hardware problem on the hardware 856 00:45:09,510 --> 00:45:12,670 side, you know, you're trying to design A consumer product that's 857 00:45:12,670 --> 00:45:15,910 not gonna harm people. And that's just the, I mean we 858 00:45:16,070 --> 00:45:18,630 people do that every day, right? We've been doing that for years 859 00:45:18,630 --> 00:45:20,470 and years. It's a, it's a really under well 860 00:45:20,470 --> 00:45:23,630 understood kind of task. And I think if you're building 861 00:45:23,630 --> 00:45:27,150 hardware, it's an easier task because then when you move to 862 00:45:27,150 --> 00:45:30,830 software it becomes much more complex, right, in terms of. 863 00:45:31,480 --> 00:45:34,160 What the harm might be. So for instance, right now I 864 00:45:34,160 --> 00:45:38,800 live in Portland, OR and in Seattle up the road, there they 865 00:45:38,800 --> 00:45:41,000 are. The the school system is suing 866 00:45:41,240 --> 00:45:46,080 like meta, right, because they have documented harm that using 867 00:45:46,080 --> 00:45:48,200 the social media has harmed children. 868 00:45:48,440 --> 00:45:50,240 It's like the Phillip Morris thing again, right? 869 00:45:50,240 --> 00:45:54,480 I mean, they know for a fact using this product causes harm 870 00:45:54,680 --> 00:45:55,760 to children. They die. 871 00:45:55,760 --> 00:45:57,280 They commit suicide. So. 872 00:45:58,120 --> 00:46:01,400 If you identify that, I mean that that takes a little while 873 00:46:01,480 --> 00:46:04,120 to put those to connect those dots, right? 874 00:46:04,240 --> 00:46:07,160 It's not the same as if they were just selling toasters that 875 00:46:07,160 --> 00:46:10,080 were electrocuting people like you understand pretty clearly 876 00:46:10,080 --> 00:46:11,560 like, OK, where's the danger there, right. 877 00:46:11,560 --> 00:46:15,560 I mean, you can see if a robot is causing physical violence 878 00:46:15,560 --> 00:46:19,800 against and so here's where it. Gets like even more complicated, 879 00:46:19,800 --> 00:46:21,440 right? So, so let's say by the way, 880 00:46:21,800 --> 00:46:24,080 we've got chat bots. Purely software. 881 00:46:24,430 --> 00:46:26,470 Yes, they're going to. People are going to get LED down 882 00:46:26,470 --> 00:46:29,230 crazy rabbit holes by these things and there's going to be, 883 00:46:29,550 --> 00:46:35,070 I predict, extremely harmful scenarios occurring from people 884 00:46:35,070 --> 00:46:38,430 just being told whatever they want to hear and and eventually 885 00:46:38,430 --> 00:46:41,870 being potentially radicalized or whatever causing harm in the 886 00:46:41,870 --> 00:46:44,350 community. It won't be the robot doing it 887 00:46:44,350 --> 00:46:47,670 directly, but it, but it or the sorry, the chat bot or the large 888 00:46:47,670 --> 00:46:49,470 language model. But then you think about the 889 00:46:49,470 --> 00:46:51,070 synthesis of these two things as well. 890 00:46:51,110 --> 00:46:55,180 So like think about. Like a self driving car. 891 00:46:55,660 --> 00:46:58,580 So now I mean it gets really complicated because you've got 892 00:46:58,580 --> 00:47:00,540 brains and you've got the hardware. 893 00:47:00,820 --> 00:47:04,020 And so the question becomes like, OK, the car crashed, like 894 00:47:04,020 --> 00:47:06,420 whose fault is that? And so right now that's 895 00:47:06,420 --> 00:47:09,420 something we're working out, you know in the court systems and 896 00:47:09,420 --> 00:47:12,940 and we're using all of our existing machinery of our 897 00:47:12,940 --> 00:47:15,820 society to try to sort that out and figure that out. 898 00:47:17,080 --> 00:47:19,080 So, you know, I I would say those are really the three 899 00:47:19,080 --> 00:47:22,360 areas, you know, pure software, the hybrid and then just the 900 00:47:22,360 --> 00:47:28,840 purely mechanical problem of of to have like a row apocalypse or 901 00:47:28,840 --> 00:47:32,200 like to have robots like it, it requires like the idea that 902 00:47:32,200 --> 00:47:34,120 there's some like super intelligence, right. 903 00:47:34,120 --> 00:47:36,600 I mean, all of that that's predicated on the idea. 904 00:47:36,760 --> 00:47:39,680 Yeah, in pop culture, you gotta. Have the singularity before. 905 00:47:39,680 --> 00:47:42,440 I mean, that's just kind of like ticking the box, right? 906 00:47:42,440 --> 00:47:44,000 I don't know how realistic that is, but. 907 00:47:45,070 --> 00:47:46,950 Right. Well, that's I guess to me like, 908 00:47:47,350 --> 00:47:51,390 you know, I know it's fiction, but like, yeah, it just feels 909 00:47:51,390 --> 00:47:54,230 like if we have this super intelligence is isn't that a 910 00:47:54,230 --> 00:47:58,230 threat enough without it taking sort of robotic form or I guess 911 00:47:58,430 --> 00:48:00,070 particular question, you know, there's all. 912 00:48:00,070 --> 00:48:03,430 These predictions about when the singularity's gonna happen and 913 00:48:03,430 --> 00:48:06,350 how it's gonna be exponential, so it's gonna go slow and then 914 00:48:06,350 --> 00:48:09,080 happen fast and. And like it's really interesting 915 00:48:09,080 --> 00:48:12,320 because you know I was, I have a degree in machine learning. 916 00:48:12,320 --> 00:48:14,760 I mean I was studying it before they were calling it machine 917 00:48:14,840 --> 00:48:17,960 learning like when it, it was called knowledge discovery and 918 00:48:17,960 --> 00:48:19,480 data mining and all these different. 919 00:48:19,880 --> 00:48:23,240 And so you you look at that and what was happening early on was 920 00:48:23,240 --> 00:48:26,360 there were all these different approaches to to trying to mimic 921 00:48:26,360 --> 00:48:30,400 intelligence, right. And we would use a whole suite 922 00:48:30,400 --> 00:48:32,720 of these things. And what happened was neural 923 00:48:32,720 --> 00:48:36,160 networks in the last few years jumped out ahead. 924 00:48:36,660 --> 00:48:39,500 And I would argue, although I'm sure other people would argue 925 00:48:39,500 --> 00:48:43,780 with me, but I would argue there was no amazing scientific 926 00:48:43,780 --> 00:48:47,180 breakthrough, right? What happened was processors got 927 00:48:47,180 --> 00:48:50,100 faster and we had access to just a ton of data. 928 00:48:50,460 --> 00:48:54,380 And so that data and those processors just are brute 929 00:48:54,380 --> 00:48:58,860 forcing what are fairly simple algorithms that have been well 930 00:48:58,860 --> 00:49:02,100 known for a long time and they're getting this kind of 931 00:49:03,340 --> 00:49:05,660 intelligentish behavior out of it. 932 00:49:06,130 --> 00:49:11,130 And so that model isn't really doesn't really sync up with the 933 00:49:11,130 --> 00:49:14,450 way I think people thought AI was going to go, right. 934 00:49:14,650 --> 00:49:17,010 We didn't think, oh, we're just going to use some old algorithm 935 00:49:17,010 --> 00:49:19,970 and just throw more processing power at it until it gets good. 936 00:49:19,970 --> 00:49:22,370 Like there's more GPUs. We're going to get it, yeah. 937 00:49:23,290 --> 00:49:24,810 And so I. Don't know if it's. 938 00:49:25,090 --> 00:49:29,170 So what I'm trying to say is, if you look at that and we just, I 939 00:49:29,210 --> 00:49:33,210 mean the way to improve chat bots I guess is to keep throwing 940 00:49:33,210 --> 00:49:35,530 more data. And keep throwing more 941 00:49:35,530 --> 00:49:38,050 processing power. But I don't feel like there's 942 00:49:38,050 --> 00:49:40,210 any type of singularity that's going to come out of that. 943 00:49:40,250 --> 00:49:42,770 I feel like it's like, Oh yeah, it's just, I feel like that our 944 00:49:42,770 --> 00:49:46,930 ChatGPT, yeah Or yeah, it'll. Just be like a more convincing 945 00:49:46,930 --> 00:49:51,490 ChatGPT, it's just really, I mean it's just really just 946 00:49:51,490 --> 00:49:54,730 regurgitating everything that it's, you know, in a very smart 947 00:49:54,730 --> 00:49:57,530 way that it's read on the Internet, which by the way, I 948 00:49:57,530 --> 00:50:00,130 mean, God, that's the worst of humanity, right? 949 00:50:00,130 --> 00:50:02,160 I mean. What's it got? 950 00:50:02,160 --> 00:50:03,880 It's trained on I know by the way, right? 951 00:50:03,880 --> 00:50:05,320 They're all desperate to train it off. 952 00:50:05,320 --> 00:50:06,760 Reddit, for instance. Yeah. 953 00:50:07,160 --> 00:50:10,960 Oh my God. You know, Reddit can be OK, but 954 00:50:11,160 --> 00:50:14,520 there's smart people there. But yeah, but you're looking at 955 00:50:14,520 --> 00:50:17,720 the Internet and it's like, oh Lord, but but anyway, I don't 956 00:50:17,720 --> 00:50:19,840 really, me personally, I don't see that. 957 00:50:19,840 --> 00:50:21,440 I see that as a kind of a dead end. 958 00:50:21,520 --> 00:50:24,080 I don't feel like that's headed toward the singularity 959 00:50:24,240 --> 00:50:25,320 necessarily. Yeah. 960 00:50:25,320 --> 00:50:29,770 Is there any existing dystopian work that you find most 961 00:50:29,810 --> 00:50:33,050 plausible or like what? I mean, you know, there's so you 962 00:50:33,050 --> 00:50:34,890 know, I've we talked there's The Matrix. 963 00:50:34,890 --> 00:50:36,930 There's sort of the Terminator movies. 964 00:50:36,930 --> 00:50:39,970 Like I don't know, like, yeah, I'm sure you've spent a lot of 965 00:50:39,970 --> 00:50:42,170 time in that world. Like are is there one that you 966 00:50:42,170 --> 00:50:45,490 find like, oh this this is the most sort of real world? 967 00:50:46,410 --> 00:50:49,370 Well, I mean. I don't think that the purpose 968 00:50:49,370 --> 00:50:53,050 of science fiction is to predict the future necessarily, but I 969 00:50:53,050 --> 00:50:55,490 would say you know, if you're thinking about. 970 00:50:57,000 --> 00:51:01,320 I think the most boring stuff is actually the most realistic, and 971 00:51:01,600 --> 00:51:04,440 in a lot of ways look for the bad books that no, I mean no 972 00:51:04,440 --> 00:51:05,640 actually. That's not true. 973 00:51:05,640 --> 00:51:08,800 Like for instance I would say Terminator, which is obviously 974 00:51:08,800 --> 00:51:13,240 great, but think about that. Villain think about Skynet. 975 00:51:13,680 --> 00:51:17,120 Skynet's just a dumb, dumb computer program that just has 976 00:51:17,120 --> 00:51:20,120 decided whatever it wants to destroy all humans. 977 00:51:20,520 --> 00:51:22,400 I mean, and there's no reason for it. 978 00:51:22,480 --> 00:51:25,920 There's no It's just like it just wants to kill everybody. 979 00:51:26,450 --> 00:51:28,690 I mean that is so boring. You couldn't get away with an 980 00:51:28,690 --> 00:51:32,490 actual human antagonist who is that simple? 981 00:51:32,490 --> 00:51:39,210 People would be like, well why like and so like if you think of 982 00:51:39,210 --> 00:51:42,530 that as its end goal, that's not very super intelligent, is it? 983 00:51:42,890 --> 00:51:45,450 Kill all humans? I mean that's like Bender level 984 00:51:45,450 --> 00:51:49,090 right now. So I would say that you know 985 00:51:49,090 --> 00:51:51,710 that in a lot of ways. It is fairly realistic if 986 00:51:51,710 --> 00:51:54,150 somebody just programmed a computer to tell it. 987 00:51:54,150 --> 00:51:56,830 Just super simple boring goal like that. 988 00:51:57,550 --> 00:52:00,670 Kill everybody like then you know I feel like that's fairly 989 00:52:00,670 --> 00:52:04,310 realistic. To step away from the yeah 990 00:52:04,430 --> 00:52:07,990 dystopian question for a second I'm curious just like as an 991 00:52:07,990 --> 00:52:13,390 author one like are you using ChatGPT at all like in any way 992 00:52:13,390 --> 00:52:18,390 or any other sort of LM tools? Well so first of all I. 993 00:52:19,340 --> 00:52:23,300 All of my novels are in the training data, which means 994 00:52:23,300 --> 00:52:24,820 hopefully I'll be part of a lawsuit. 995 00:52:24,980 --> 00:52:26,180 Yeah, I know. I was going to ask you about 996 00:52:26,180 --> 00:52:27,060 that next. OK? 997 00:52:27,060 --> 00:52:28,540 I asked. I went in the chat. 998 00:52:28,540 --> 00:52:32,180 GPT and I said, hey, write me a short story in the in the vein 999 00:52:32,180 --> 00:52:34,700 of Daniel H Wilson. And it, like, immediately wrote. 1000 00:52:34,700 --> 00:52:36,860 I didn't say anything else. And it really wrote about 1001 00:52:36,860 --> 00:52:38,740 robots, right? It wrote, it wrote science 1002 00:52:38,740 --> 00:52:40,180 fiction. It was really clear that it had 1003 00:52:40,180 --> 00:52:42,310 been trained on my stuff. I'm like, tough. 1004 00:52:42,310 --> 00:52:44,710 Well, that's bullshit. First of all, I never gave 1005 00:52:44,710 --> 00:52:49,470 anybody permission to do that. And so second of all, like, I I 1006 00:52:49,470 --> 00:52:52,150 think it's so funny. So I had I'm not going to name 1007 00:52:52,150 --> 00:52:54,230 the corporation because I don't want to get into trouble. 1008 00:52:54,550 --> 00:52:59,310 I had a large large corporation approached me before ChatGPT 1009 00:52:59,470 --> 00:53:04,550 before Open AI released before the the these Gans became like a 1010 00:53:04,550 --> 00:53:07,310 big deal. So this is maybe probably. 1011 00:53:08,250 --> 00:53:12,330 Six months to a year before everybody knew about that and 1012 00:53:12,450 --> 00:53:15,050 that that came out and Open AI really just broke the whole 1013 00:53:15,050 --> 00:53:18,330 deal. And they it was a researcher 1014 00:53:18,330 --> 00:53:21,610 that called me and they wanted to know if I would test out this 1015 00:53:21,610 --> 00:53:25,770 new program that they had that was basically ChatGPT and and 1016 00:53:25,770 --> 00:53:29,970 they said we want you to use it to help you write. 1017 00:53:30,370 --> 00:53:33,850 And I'm like OK, so then what How will it help me write? 1018 00:53:33,850 --> 00:53:36,440 And they say well. And and these are just such 1019 00:53:36,440 --> 00:53:40,520 sweet researchers that don't have any GD idea about what the 1020 00:53:40,520 --> 00:53:43,360 real world is, is hell is going on in real world. 1021 00:53:43,360 --> 00:53:46,200 And they're like, it'll help you be creative. 1022 00:53:46,640 --> 00:53:50,200 And I'm like, oh, creativity, right? 1023 00:53:50,320 --> 00:53:53,000 That thing. I hate doing that thing that I 1024 00:53:53,000 --> 00:53:57,800 spent the last 20 years teaching myself to take. 1025 00:53:57,800 --> 00:53:59,920 What's in my head? And put it in the world by 1026 00:53:59,920 --> 00:54:02,520 memorizing all these really boring writing skills which are 1027 00:54:02,520 --> 00:54:04,910 just pure hell. And then they want to come in 1028 00:54:04,910 --> 00:54:07,790 and take the one good thing. I'm like, why do you think 1029 00:54:07,790 --> 00:54:10,870 writers write? You think because we love 1030 00:54:10,990 --> 00:54:13,270 typing. We like pushing letters. 1031 00:54:13,670 --> 00:54:15,510 I'm like, what the hell are you talking about? 1032 00:54:15,510 --> 00:54:18,830 I was like, you need to be prepared for everyone that you 1033 00:54:18,830 --> 00:54:21,710 give this tool to, to give you super negative feedback. 1034 00:54:21,990 --> 00:54:25,030 Because they wanted to include me in, like, this report about 1035 00:54:25,030 --> 00:54:27,910 how this tool would be used, and they wanted to publicize this 1036 00:54:27,910 --> 00:54:31,030 report as well. And so they they went back and 1037 00:54:31,030 --> 00:54:33,190 they came back and they said, hey, we got great news. 1038 00:54:33,730 --> 00:54:36,850 I got permission from, like, the program manager on this to 1039 00:54:36,930 --> 00:54:40,050 include your feedback, even if it's negative. 1040 00:54:41,250 --> 00:54:44,170 And I was like, you know what? I'm gonna just take a pass on 1041 00:54:44,170 --> 00:54:45,570 this. I wish you luck with your 1042 00:54:45,570 --> 00:54:46,610 project. Right. 1043 00:54:46,970 --> 00:54:50,610 So, man, no. You know what I think about, 1044 00:54:50,650 --> 00:54:54,010 man? The thing that cracks me up is I 1045 00:54:54,010 --> 00:54:59,220 used to really hate on. Asimov because I was like, robot 1046 00:54:59,220 --> 00:55:01,540 psychologists, What kind of bullshit is that? 1047 00:55:01,700 --> 00:55:04,860 Like, you program robots, right? I spent all this time, you know, 1048 00:55:04,900 --> 00:55:07,660 programming and learning all of these programming languages and 1049 00:55:07,940 --> 00:55:11,060 and everything you got to do in order to in order to speak to 1050 00:55:11,060 --> 00:55:13,860 the robot mind, because it doesn't speak English, right? 1051 00:55:14,260 --> 00:55:17,220 And I used to think that the idea of a robot psychologist was 1052 00:55:17,220 --> 00:55:20,660 the dumbest thing ever. The idea of a positronic brain, 1053 00:55:20,660 --> 00:55:23,160 by the way. Which is made fully and it's 1054 00:55:23,160 --> 00:55:25,080 just boom, it's done. It's crystallized. 1055 00:55:25,080 --> 00:55:26,920 You can't go back and change it, right? 1056 00:55:27,240 --> 00:55:29,520 And that's and. And by the way, that's exactly 1057 00:55:29,520 --> 00:55:31,600 what happens. So neural networks are black 1058 00:55:31,600 --> 00:55:33,800 boxes. You can't go in and fiddle with 1059 00:55:33,800 --> 00:55:36,640 it because all the weights make no sense, like the human brain. 1060 00:55:36,680 --> 00:55:39,280 So it becomes like psychology and then now if you. 1061 00:55:39,280 --> 00:55:40,720 Want to for instance? Yeah. 1062 00:55:40,720 --> 00:55:46,160 If you want to game a like a like a chat bot, if you want to 1063 00:55:46,160 --> 00:55:48,440 trick it into doing something, you totally have to use 1064 00:55:48,440 --> 00:55:51,320 psychological things. It's totally psychology. 1065 00:55:51,740 --> 00:55:54,100 I mean, Bravo. Asimov. 1066 00:55:54,860 --> 00:55:57,780 Bravo, Sir, Are you? Do you think you will, like, try 1067 00:55:57,780 --> 00:55:59,580 to sue? Like if somebody comes to you 1068 00:55:59,620 --> 00:56:01,380 and says we're going to sue these guys for. 1069 00:56:01,380 --> 00:56:02,900 Oh yeah. Oh, absolutely. 1070 00:56:02,900 --> 00:56:07,980 No, that this has to happen. Look, I mean, I I get people 1071 00:56:07,980 --> 00:56:11,100 trying to make money. I get, you know, capitalism. 1072 00:56:11,100 --> 00:56:14,340 I understand it. And this is a case where you 1073 00:56:14,340 --> 00:56:16,940 know, they're they're they have to be sued. 1074 00:56:17,020 --> 00:56:19,260 I mean, we're just because we're creating new, this is a 1075 00:56:19,260 --> 00:56:23,290 completely new domain. And if not, OK, to rip people 1076 00:56:23,290 --> 00:56:26,010 off and then chop up their stuff and train an algorithm. 1077 00:56:26,210 --> 00:56:30,130 And so our country or whatever is going to have to figure that 1078 00:56:30,130 --> 00:56:33,250 out and get that down in law. And that's going to require, 1079 00:56:33,250 --> 00:56:35,330 yeah, courts, I mean, so absolutely they're all going to 1080 00:56:35,330 --> 00:56:37,330 get sued. Do you do you think? 1081 00:56:39,020 --> 00:56:44,980 Sci-fi and dystopias are bad for AI like even talking to ChatGPT, 1082 00:56:45,260 --> 00:56:48,740 it's funny when you ask it to imagine things it's been trained 1083 00:56:48,740 --> 00:56:50,940 on all that. Exactly it parents all that. 1084 00:56:50,940 --> 00:56:54,220 And so it can sound really spooky like it feels like you 1085 00:56:54,220 --> 00:56:57,220 know you're like I'm you know this is imagination it's fiction 1086 00:56:57,220 --> 00:56:59,060 like. But I do think people, you know, 1087 00:56:59,060 --> 00:57:02,860 seriously turn to science fiction when they try and game 1088 00:57:02,860 --> 00:57:05,540 out where sort of a technology we don't really understand, 1089 00:57:05,540 --> 00:57:09,140 that's developed faster than we expect is progressing. 1090 00:57:09,140 --> 00:57:11,820 So yeah, is it, is that bad in some ways? 1091 00:57:11,820 --> 00:57:15,140 Do you do you regret that? I think it says a lot more about 1092 00:57:15,140 --> 00:57:20,180 us than it does about them them being the robots, you know, like 1093 00:57:20,780 --> 00:57:21,940 what I've been thinking about lately. 1094 00:57:21,940 --> 00:57:25,620 So I don't know know if you've read my books, but I I'm from 1095 00:57:25,620 --> 00:57:27,580 Oklahoma. I'm, I grew up in the Cherokee 1096 00:57:27,580 --> 00:57:28,940 Nation. I'm a Cherokee citizen and I 1097 00:57:28,940 --> 00:57:30,140 write a lot. There's a lot of native 1098 00:57:30,140 --> 00:57:33,260 characters and stuff I do. And so I think a lot about 1099 00:57:33,260 --> 00:57:35,420 technology from like a native perspective. 1100 00:57:35,970 --> 00:57:39,330 And one thing I've been thinking a lot about lately is just kind 1101 00:57:39,330 --> 00:57:44,370 of like how a lot of this super negative, like robots and also 1102 00:57:44,370 --> 00:57:48,250 first contact, like aliens, they always show up and what do they 1103 00:57:48,250 --> 00:57:54,090 do, man, they do exactly what all the colonizer civilizations 1104 00:57:54,090 --> 00:57:56,250 did to indigenous people all over the world. 1105 00:57:56,250 --> 00:58:00,290 So we got a civilization, we got a society, a culture. 1106 00:58:00,290 --> 00:58:03,410 It's all built on colonization, right? 1107 00:58:03,650 --> 00:58:06,730 So it means. A bunch of people with superior 1108 00:58:06,730 --> 00:58:11,330 technology showed up someplace, murdered everyone, destroyed 1109 00:58:11,330 --> 00:58:13,850 their culture. I mean, Independence Day, 1110 00:58:13,970 --> 00:58:16,330 they're blowing up monuments. So you think that is like 1111 00:58:17,250 --> 00:58:19,610 resource extraction, right? They're they're what? 1112 00:58:19,610 --> 00:58:21,810 Are they stealing our water or they're stealing our air or the 1113 00:58:22,170 --> 00:58:25,090 and then just completely dominating other people's 1114 00:58:25,330 --> 00:58:26,890 bodies. You know, you think about 1115 00:58:27,250 --> 00:58:31,450 Invasion of the Body Snatchers. And just like so, that type of 1116 00:58:31,730 --> 00:58:35,010 fear, I think, is cooked into our civilization. 1117 00:58:35,380 --> 00:58:39,940 Based on our origins and it comes seeping out into our pop 1118 00:58:39,940 --> 00:58:43,700 culture in a lot of different ways, but in particular in 1119 00:58:43,700 --> 00:58:45,660 science fiction, we see it so much. 1120 00:58:45,940 --> 00:58:49,220 Do do you think we should stop? You know, there are even there 1121 00:58:49,220 --> 00:58:52,140 are AI companies that say, oh, we should pause for six months. 1122 00:58:52,140 --> 00:58:54,860 I mean, originally it sounded like you were saying you didn't 1123 00:58:54,860 --> 00:58:58,300 think just adding more, you know, compute was gonna like 1124 00:58:58,300 --> 00:59:01,300 create this super intelligence. But yeah, do you, do you have 1125 00:59:01,300 --> 00:59:04,500 the impulse that, you know, we should slow down or stop this 1126 00:59:04,500 --> 00:59:06,260 research if we don't know what's going to happen? 1127 00:59:06,500 --> 00:59:08,700 Yeah, Well, I don't think that. That's what they're afraid of. 1128 00:59:08,700 --> 00:59:11,620 I don't think they're afraid of us crossing the singularity 1129 00:59:11,620 --> 00:59:13,860 threshold or something. I mean, they're just worried 1130 00:59:13,860 --> 00:59:16,660 about. So for instance, I yeah, I will 1131 00:59:16,660 --> 00:59:21,980 tell you that the US military is actively doing a lot of threat 1132 00:59:21,980 --> 00:59:26,300 scenarios that involve this kind of technology being weaponized, 1133 00:59:26,300 --> 00:59:31,180 this sort of disinformation on a mass scale, hugely distracting 1134 00:59:31,260 --> 00:59:35,440 events that could occur. I mean at this point I mean you 1135 00:59:35,440 --> 00:59:38,800 could get, you could get a phone call from someone it could be 1136 00:59:38,800 --> 00:59:41,120 just like in robo apocalypse, you know where everybody gets a 1137 00:59:41,120 --> 00:59:44,440 phone call from somebody that they're related to and they 1138 00:59:44,440 --> 00:59:48,160 trust and they all get told to go to a certain is a bad, bad 1139 00:59:48,160 --> 00:59:50,320 situation. I mean, all that stuff is 1140 00:59:50,320 --> 00:59:53,380 starting to. Right. 1141 00:59:53,380 --> 00:59:56,380 I really focused on this idea of it acting independently, but 1142 00:59:56,420 --> 00:59:59,060 obviously, just like bad people aren't, no, this is always going 1143 00:59:59,060 --> 01:00:01,140 to be bad. People and and so, so first of 1144 01:00:01,140 --> 01:00:02,780 all I don't think that's they're not worried about that 1145 01:00:02,780 --> 01:00:04,380 singularity. They're worried about this being 1146 01:00:04,380 --> 01:00:08,300 weaponized And in terms of putting a pause on it, I mean, 1147 01:00:08,980 --> 01:00:12,540 yeah, I mean why not. I I would say only put a pause 1148 01:00:12,540 --> 01:00:15,900 on it until we figure out the the rules around it. 1149 01:00:16,100 --> 01:00:19,700 But but the fact is this like that's not how the United States 1150 01:00:19,700 --> 01:00:22,620 works. Like, we don't fix anything 1151 01:00:22,620 --> 01:00:25,740 until it's broken, right? I mean, there's not I, I just 1152 01:00:25,740 --> 01:00:29,340 don't have a lot of confidence that they're going to say, OK, 1153 01:00:29,340 --> 01:00:32,020 six month pause and we're going to work out all the laws and 1154 01:00:32,020 --> 01:00:35,020 everything's going to be all ready to go in six months. 1155 01:00:35,380 --> 01:00:37,060 Like, hell no, that's not going to work. 1156 01:00:37,300 --> 01:00:42,340 So I mean, I think it's a a nice idea, but I'm I'm extremely 1157 01:00:42,340 --> 01:00:46,100 dubious and skeptical that could ever have any actual, you know, 1158 01:00:46,180 --> 01:00:48,160 useful. I mean other countries. 1159 01:00:48,160 --> 01:00:49,880 Are still going to work. I mean, there's the, you know, 1160 01:00:49,880 --> 01:00:52,240 sort of China boogeyman where it's like you can't stop the 1161 01:00:52,240 --> 01:00:55,880 whole world from working on it. And so there's that, honestly, 1162 01:00:55,880 --> 01:00:58,440 though, there's the there's the, the other side of that coin, 1163 01:00:58,440 --> 01:01:00,800 which is, you know, thank God that they're actually interested 1164 01:01:00,800 --> 01:01:04,200 in privacy in Europe, right? Like, at least there's some 1165 01:01:04,200 --> 01:01:07,480 people standing up and saying, hey, like, we don't just have to 1166 01:01:07,480 --> 01:01:10,240 accept this status quo, just 'cause you're a giant billion 1167 01:01:10,240 --> 01:01:13,080 dollar corporation or billion trillion dollar corporation, 1168 01:01:14,140 --> 01:01:15,540 just. Something you referenced 1169 01:01:15,540 --> 01:01:17,340 earlier. Does the military come to you to 1170 01:01:17,340 --> 01:01:20,140 help game out sort of technological? 1171 01:01:20,500 --> 01:01:21,860 Yeah, I've done a little bit of that. 1172 01:01:22,740 --> 01:01:25,060 It's pretty fun. So I'm not the only one who's 1173 01:01:25,060 --> 01:01:27,660 like. Oh, we should like, get some guy 1174 01:01:27,660 --> 01:01:30,140 who thought about it. Sort of fiction in fiction. 1175 01:01:30,140 --> 01:01:32,180 Just sort of say what happened in the real world. 1176 01:01:32,220 --> 01:01:33,700 Yeah, I'm not. The only one, But there are, 1177 01:01:33,700 --> 01:01:36,260 there are and and honestly, they're training their own. 1178 01:01:36,820 --> 01:01:38,340 They're training their own science fiction, right? 1179 01:01:38,700 --> 01:01:42,710 I will tell you this like if you are, if you're a general in the 1180 01:01:42,910 --> 01:01:45,910 Air Force or or whatever, if you're some kind of higher up in 1181 01:01:45,910 --> 01:01:49,390 the military, you get a lot of white papers that are describing 1182 01:01:49,390 --> 01:01:52,910 the technological capabilities of the enemy or or of various 1183 01:01:52,910 --> 01:01:56,430 munitions and things like that. And these papers are very dry 1184 01:01:56,750 --> 01:02:00,790 and just like very, you know, it's much better. 1185 01:02:00,790 --> 01:02:04,990 I think at least it's useful to have somebody actually 1186 01:02:05,420 --> 01:02:08,260 creatively tell a story that's gonna stick with you, cuz that's 1187 01:02:08,260 --> 01:02:09,740 the way humans communicate, right? 1188 01:02:09,740 --> 01:02:12,340 Through stories. And then you and then the 1189 01:02:12,340 --> 01:02:15,380 general is sitting there thinking, OK, now I've I've just 1190 01:02:15,380 --> 01:02:19,500 read a story about, you know, an actual person being impacted by 1191 01:02:19,500 --> 01:02:22,650 this technology in that way. And you can really visualize 1192 01:02:22,650 --> 01:02:24,410 what it means. And it's more than just facts 1193 01:02:24,410 --> 01:02:26,570 and figures when you say they're the military. 1194 01:02:26,570 --> 01:02:27,650 Is trained science fiction writers. 1195 01:02:27,650 --> 01:02:30,050 You mean those authors of the boring white papers? 1196 01:02:30,370 --> 01:02:32,650 Or you mean like true science fiction writers? 1197 01:02:32,650 --> 01:02:36,090 No, I mean that they they're literally you can yeah if you're 1198 01:02:36,090 --> 01:02:38,530 in the military you can There are certain programs you can 1199 01:02:38,530 --> 01:02:41,090 sort of sign up for where you get taught. 1200 01:02:41,170 --> 01:02:42,930 Yeah. Where you get they bring in 1201 01:02:42,930 --> 01:02:47,370 science fiction authors to, to help teach the the, the military 1202 01:02:47,370 --> 01:02:51,750 people how to sort of, yeah, take those super real world 1203 01:02:51,750 --> 01:02:54,630 scenarios and write them in an engaging way so that you have 1204 01:02:54,630 --> 01:02:57,750 sort of a case study and you say, look, here's one way this 1205 01:02:57,750 --> 01:02:59,830 could be employed. And that's just good. 1206 01:03:00,030 --> 01:03:01,590 That's just good science fiction writing. 1207 01:03:01,630 --> 01:03:04,390 That's how that is. But they would prefer not to 1208 01:03:04,830 --> 01:03:07,790 outsource it to, you know, hippies like me. 1209 01:03:09,510 --> 01:03:12,430 I'm curious. How much sort of the 1210 01:03:12,430 --> 01:03:15,230 technological developments we're seeing right now like you know 1211 01:03:15,310 --> 01:03:18,870 mid journey, whatever are influencing what? 1212 01:03:19,360 --> 01:03:22,240 You plan to write or do you think, like science fiction will 1213 01:03:22,240 --> 01:03:25,640 change based on what we see is possible? 1214 01:03:26,120 --> 01:03:29,760 Yeah, I mean, absolutely. Is this changing science fiction 1215 01:03:29,880 --> 01:03:31,360 every day? For sure. 1216 01:03:31,360 --> 01:03:34,000 And and and look, I think it was a watershed moment. 1217 01:03:34,120 --> 01:03:37,560 Like whenever you really just finally just talk to it and you 1218 01:03:37,560 --> 01:03:41,920 realize, oh, this thing is like, yeah, this thing can just talk 1219 01:03:41,920 --> 01:03:43,440 to me. And by the way, right now 1220 01:03:43,440 --> 01:03:48,000 there's this huge rush, as, you know, like to plug this stuff in 1221 01:03:48,350 --> 01:03:50,990 anywhere you can. It's like, oh, I don't know, 1222 01:03:50,990 --> 01:03:52,110 what is it? It's like if it's like 1223 01:03:52,110 --> 01:03:53,630 mayonnaise, you know, they're just putting it on. 1224 01:03:53,630 --> 01:03:55,670 We just discovered this thing. Let's put it on everything. 1225 01:03:55,750 --> 01:03:58,430 Right, right, right. It's desperation, yeah. 1226 01:03:58,910 --> 01:04:01,870 It's going to taste like crap on a lot of stuff, but like it's 1227 01:04:01,870 --> 01:04:03,270 going to work out in some places. 1228 01:04:03,550 --> 01:04:09,920 One thing I see which I find interesting is it occupying as 1229 01:04:09,920 --> 01:04:12,400 almost a homunculus, right? You think of a homunculus as 1230 01:04:12,400 --> 01:04:15,360 like a little tiny person that's inside of you kind of driving 1231 01:04:15,360 --> 01:04:17,920 you around or you think of like, who's the bad guy from 1232 01:04:17,920 --> 01:04:20,240 SpongeBob? You know, he's always driving a 1233 01:04:20,240 --> 01:04:21,920 giant robot, right, The little guy. 1234 01:04:22,120 --> 01:04:27,120 So if you think of the ChatGPT and you have like a it's like 1235 01:04:27,120 --> 01:04:28,920 the homunculus. You stick it inside of a 1236 01:04:28,920 --> 01:04:33,520 hardware platform so you can just tell ChatGPT here's a super 1237 01:04:33,520 --> 01:04:38,120 simple programming language that allows you to drive around this 1238 01:04:38,640 --> 01:04:42,720 autonomous vehicle or allows you to drive around this humanoid 1239 01:04:42,720 --> 01:04:46,200 robot platform. And then you tell it in English, 1240 01:04:46,640 --> 01:04:50,960 hey, go make a sandwich. And then it looks at all of its 1241 01:04:50,960 --> 01:04:53,520 little controls and it translates that and says, OK, 1242 01:04:53,520 --> 01:04:56,120 I'm going to drive this thing in there because it kind of get has 1243 01:04:56,120 --> 01:04:58,280 the right idea about what making a sandwich is. 1244 01:04:58,800 --> 01:05:03,000 But a but a hardware platform is just a bunch of like XYZ 1245 01:05:03,000 --> 01:05:05,440 coordinates where you're going to put your limbs and it's and 1246 01:05:05,440 --> 01:05:09,010 that translation right there, that's a spot where where 1247 01:05:09,010 --> 01:05:12,130 ChatGPT is kind of getting just like plastered in, you know, 1248 01:05:12,130 --> 01:05:14,930 just slap it in there and it solves all those problems. 1249 01:05:15,450 --> 01:05:18,650 And so that's kind of interesting because now you're 1250 01:05:18,650 --> 01:05:21,810 going to see ChatGPT having the ability to interact with the 1251 01:05:21,810 --> 01:05:27,570 real world via like whatever delivery bots or like or drones 1252 01:05:27,570 --> 01:05:31,410 or or humanoid robots. And so that's where you maybe 1253 01:05:31,410 --> 01:05:33,290 get into a little bit of consumer trouble. 1254 01:05:33,290 --> 01:05:35,010 Again, it's like an autonomous car, right? 1255 01:05:36,360 --> 01:05:38,560 Is there? Anything you've written where, 1256 01:05:38,600 --> 01:05:42,840 based on how technology has developed, you feel like, oh, it 1257 01:05:42,840 --> 01:05:44,680 feels less plausible than when I wrote it? 1258 01:05:44,680 --> 01:05:48,880 Or like how much like forecasting where where what's 1259 01:05:48,880 --> 01:05:53,320 possible do you see as like part of your yeah, well, I think 1260 01:05:53,320 --> 01:05:56,800 that, you know, what happens is you look at what's out there in 1261 01:05:56,800 --> 01:05:59,480 the world and you run with it and you start thinking of all 1262 01:05:59,480 --> 01:06:01,720 the different ways it could be. And that's where science fiction 1263 01:06:01,720 --> 01:06:04,360 is fun, right, 'cause you suddenly you're like, I never 1264 01:06:04,360 --> 01:06:06,720 really thought of it like that. But but there it is. 1265 01:06:06,720 --> 01:06:09,560 And dystopias are fun too, because, you know, it's 1266 01:06:09,560 --> 01:06:10,960 dangerous. It's exciting. 1267 01:06:11,760 --> 01:06:17,320 And so the difference between my writing and and is that I had 1268 01:06:17,320 --> 01:06:20,480 this degree in robotics. I all my friends from when I was 1269 01:06:20,640 --> 01:06:24,280 20, they're all running corporate robot corporations. 1270 01:06:24,280 --> 01:06:28,120 They're driving the Mars Rovers like that so that my cohort, you 1271 01:06:28,120 --> 01:06:31,280 know, I'm like my all my old friends, that's what they're 1272 01:06:31,280 --> 01:06:33,490 doing. And so I feel like I have a 1273 01:06:33,490 --> 01:06:35,770 little bit of a sneak peek. So like for instance when I 1274 01:06:35,770 --> 01:06:39,690 wrote Robo Apocalypse, that was 10 years ago or whatever, and it 1275 01:06:40,170 --> 01:06:43,370 I was watching and seeing the technology that was 5 or 10 1276 01:06:43,370 --> 01:06:46,210 years out. So it's all come true. 1277 01:06:46,330 --> 01:06:49,410 In fact, that's the joke because it's still Robo Apocalypse is 1278 01:06:49,410 --> 01:06:53,570 still with Spielberg at Amblin and and you know, development 1279 01:06:53,570 --> 01:06:56,570 continues on this movie. But like what we've what we've 1280 01:06:56,570 --> 01:07:00,360 told Spielberg is hey man, let's do this before it becomes a 1281 01:07:00,360 --> 01:07:04,000 historical documentary. All of this stuff is, you know, 1282 01:07:04,160 --> 01:07:07,680 so in in Robo Apocalypse there are and Robo Genesis the sequel, 1283 01:07:07,680 --> 01:07:08,760 which came out a few years later. 1284 01:07:09,080 --> 01:07:14,280 There are autonomous vehicles. There is essentially ChatGPT as 1285 01:07:14,280 --> 01:07:15,880 as personal assistants on phones. 1286 01:07:16,240 --> 01:07:19,920 I mean I wrote that before before Siri came out. 1287 01:07:20,600 --> 01:07:21,560 Wow. It was 2000. 1288 01:07:22,000 --> 01:07:25,800 The book came out in like 2011 or something a long time ago. 1289 01:07:25,960 --> 01:07:31,010 But anyway, so yeah, I mean my latest novel is that the 1290 01:07:31,010 --> 01:07:34,170 Andromeda Evolution which is a sequel to Michael Crichton's The 1291 01:07:34,170 --> 01:07:37,770 Andromeda Strain, which I he passed away and I did this with 1292 01:07:37,770 --> 01:07:40,650 in cooperation with his, with his estate. 1293 01:07:40,650 --> 01:07:43,890 And and you know that one was again like very dialed in in 1294 01:07:43,890 --> 01:07:47,290 terms of just because it's Crichton, the the government 1295 01:07:47,290 --> 01:07:50,450 like exactly which scientists would be going where how they be 1296 01:07:50,450 --> 01:07:54,930 chosen and and and I mean just so so locking in those details 1297 01:07:54,930 --> 01:07:58,090 is is really important for a, for a techno thriller, you know 1298 01:07:58,090 --> 01:08:01,570 that certain genre of science fiction do you do you give in? 1299 01:08:01,570 --> 01:08:04,970 Sort of, yeah. The professional need to be 1300 01:08:04,970 --> 01:08:08,290 forward-looking. Do you have any sort of 1301 01:08:08,410 --> 01:08:13,010 predictions or about where like culture is going or what what 1302 01:08:13,010 --> 01:08:15,970 you think, what you really think will be the more sort of 1303 01:08:15,970 --> 01:08:18,410 essential change like it. I mean it does feel like, you 1304 01:08:18,410 --> 01:08:20,490 know, like self driving cars and stuff like that. 1305 01:08:21,080 --> 01:08:23,319 Feel feel close? Certainly in San Francisco. 1306 01:08:24,240 --> 01:08:29,040 I mean, I think culturally human beings are always the wild card. 1307 01:08:29,279 --> 01:08:31,600 The technology is not that hard to predict usually. 1308 01:08:31,600 --> 01:08:34,800 I mean, you can usually see what people are up to 5-10 years out, 1309 01:08:34,800 --> 01:08:37,040 you, you know what what they're going for, right? 1310 01:08:37,319 --> 01:08:39,720 But then with human beings, you never know, like autonomous 1311 01:08:39,720 --> 01:08:41,640 vehicles. I was much more sort of bullish 1312 01:08:41,640 --> 01:08:43,560 on that. I thought those things would 1313 01:08:43,560 --> 01:08:47,279 already be here, being used in a bigger way. 1314 01:08:47,560 --> 01:08:51,260 But as it turns out, one person gets killed and humans freak 1315 01:08:51,260 --> 01:08:54,740 out, even though 40 or 50,000 people are dying every year with 1316 01:08:54,740 --> 01:08:57,100 Inhuman human. But then the robot screws up 1317 01:08:57,100 --> 01:08:59,859 once and everybody loses their mind and they shut them all 1318 01:08:59,859 --> 01:09:01,220 down. That's like, I didn't really 1319 01:09:01,220 --> 01:09:02,819 predict that. I thought we had a higher 1320 01:09:02,819 --> 01:09:07,220 tolerance for for that or or we cared less so like. 1321 01:09:07,220 --> 01:09:09,620 So it's really tough to predict the human element. 1322 01:09:09,899 --> 01:09:13,180 And I think that with Chachi PT, man, things can go a lot of 1323 01:09:13,180 --> 01:09:18,140 different ways, right? The movie Her is interesting in 1324 01:09:18,140 --> 01:09:19,939 terms of falling in love with it. 1325 01:09:19,939 --> 01:09:22,020 I think that's totally going to happen. 1326 01:09:22,260 --> 01:09:25,500 It can, It just tells us what we want to hear, right. 1327 01:09:25,660 --> 01:09:29,700 And so I think that that is just so dangerous and just such a 1328 01:09:29,700 --> 01:09:33,420 siren song. For instance, the researcher who 1329 01:09:33,540 --> 01:09:37,460 thought that that that at Google. 1330 01:09:37,540 --> 01:09:39,100 Like something? Yeah, he thought that. 1331 01:09:39,100 --> 01:09:42,260 That program was sentient. And did you just? 1332 01:09:42,260 --> 01:09:44,819 I mean, when you read those transcripts, you can just see 1333 01:09:45,430 --> 01:09:48,029 he's leading it. He's telling it what he wants to 1334 01:09:48,029 --> 01:09:50,670 hear and it every, and he's rewarding it every time it comes 1335 01:09:50,670 --> 01:09:53,670 back and tells him a version of what he wants to hear. 1336 01:09:53,670 --> 01:09:56,830 It's almost like he said write me a short story where you're a 1337 01:09:56,830 --> 01:10:00,350 sentient AI and has read all those stories. 1338 01:10:00,350 --> 01:10:03,790 I know that it's read mine. And so like, you know, it's 1339 01:10:03,790 --> 01:10:07,990 going to jam those things out. And so I mean, think about that, 1340 01:10:07,990 --> 01:10:11,390 right? Like if you're just sitting and 1341 01:10:11,390 --> 01:10:14,940 you've got sort of mass disinformation campaigns that 1342 01:10:14,940 --> 01:10:18,940 are amplifying certain ideas. I mean, man, the the the 1343 01:10:18,940 --> 01:10:23,220 potential for divisiveness I think is just super scary. 1344 01:10:23,220 --> 01:10:27,220 And in fact that's probably the biggest thing I'm afraid of is 1345 01:10:27,220 --> 01:10:30,220 that we just end up in these crazy echo chambers. 1346 01:10:30,700 --> 01:10:34,100 Like, it just becomes a very. Effective propaganda tool or you 1347 01:10:34,100 --> 01:10:37,660 know becomes, yeah man, really good at just and and the and the 1348 01:10:37,860 --> 01:10:41,780 and the thing that makes it so dangerous is that it's a mass 1349 01:10:41,780 --> 01:10:46,470 scale, individualized attack. So normally when you scale up to 1350 01:10:46,470 --> 01:10:50,270 a mass level, you get A1 size fits all type scenario, right. 1351 01:10:50,710 --> 01:10:53,550 You know it's not that effective in this case, man. 1352 01:10:53,550 --> 01:10:56,270 You can scale it all the way and then it's going to be 1353 01:10:56,270 --> 01:10:59,150 individualized for each person and it's like that's pretty 1354 01:10:59,150 --> 01:11:00,430 crazy. That's pretty scary. 1355 01:11:00,630 --> 01:11:03,310 I mean the best case scenario of that is just we end up buying a 1356 01:11:03,310 --> 01:11:06,030 bunch of shit we don't need because we're being advertised 1357 01:11:06,030 --> 01:11:07,270 to. That's the best case. 1358 01:11:07,270 --> 01:11:08,790 We we get much better. Yeah. 1359 01:11:09,500 --> 01:11:12,540 I mean it's like Advanced micro targeting, I mean you know, 1360 01:11:12,540 --> 01:11:15,300 exactly with elections, right. It used to be like well at least 1361 01:11:15,300 --> 01:11:17,980 you had to use the same message for everybody and then it was 1362 01:11:17,980 --> 01:11:20,660 like OK, we can break it down. And then you know with Facebook, 1363 01:11:20,820 --> 01:11:23,540 you know, obviously, yeah, there were these the micro targeting 1364 01:11:23,540 --> 01:11:25,860 what you're saying is now it's even easier to actually create 1365 01:11:26,140 --> 01:11:29,500 the content for the sort of niche and and The thing is it's 1366 01:11:29,500 --> 01:11:33,840 very human like, right. So one thing that I'm kind of, I 1367 01:11:33,840 --> 01:11:36,840 don't know, I'm not after during COVID, right. 1368 01:11:36,840 --> 01:11:39,880 We all got very used to interacting with machines and 1369 01:11:39,880 --> 01:11:43,160 everything that used to have a human element to it. 1370 01:11:43,200 --> 01:11:44,800 They're trying to get that out, right. 1371 01:11:44,800 --> 01:11:47,280 So it's a service. You want it, I give you the 1372 01:11:47,280 --> 01:11:49,840 money, you give me the thing, no chit chat. 1373 01:11:50,080 --> 01:11:53,520 And I need it now. And I don't want any bull, 1374 01:11:53,520 --> 01:11:55,200 right? And so we did that. 1375 01:11:55,200 --> 01:11:57,840 We were doing that online and we're doing that also with 1376 01:11:57,840 --> 01:12:00,800 lifelike machines that are talking to us like people. 1377 01:12:01,320 --> 01:12:02,560 And then we go out into the real world. 1378 01:12:02,560 --> 01:12:05,160 You know, if we're ordering off of like these smart whatevers 1379 01:12:05,160 --> 01:12:07,960 and and you know, these chat windows popping up and and stuff 1380 01:12:07,960 --> 01:12:11,840 like that, I think it's going to only become more of that human 1381 01:12:11,840 --> 01:12:15,640 like machines selling us stuff. The problem though is that then 1382 01:12:15,640 --> 01:12:19,000 you go and you know, people are being really rude to like their 1383 01:12:19,000 --> 01:12:21,920 baristas and being really rude to why? 1384 01:12:22,240 --> 01:12:24,480 Because we're trained. We can be so mean to like we're 1385 01:12:24,480 --> 01:12:26,120 trained. It's it's a classic, like you're 1386 01:12:26,120 --> 01:12:29,160 typing in to support, like, go fuck yourself, be human. 1387 01:12:29,160 --> 01:12:31,480 It's a machine. And then it turns out it's like 1388 01:12:31,480 --> 01:12:32,360 a real person. Yeah. 1389 01:12:32,360 --> 01:12:33,640 Oops. Oh, oops. 1390 01:12:33,840 --> 01:12:35,720 Yeah, but then also like, oops, right? 1391 01:12:35,720 --> 01:12:37,200 Do you? Do you worry on? 1392 01:12:37,200 --> 01:12:42,600 The like being nice to the AI systems for for like, I don't 1393 01:12:42,600 --> 01:12:44,400 know, there's some people. Like, oh, you should be nice to 1394 01:12:44,400 --> 01:12:47,520 it because it could be sentient someday. 1395 01:12:48,080 --> 01:12:51,000 I mean, yes, I do say please and thank you. 1396 01:12:51,000 --> 01:12:53,280 I got rid of the. I got rid of Alexa. 1397 01:12:53,280 --> 01:12:56,880 I have a Sonos now, but I feel like it's slightly less evil. 1398 01:12:57,280 --> 01:13:00,160 But well, I say, yeah, I say please and thank you. 1399 01:13:00,160 --> 01:13:03,760 Not because I care about it, memorizing whether I was polite. 1400 01:13:03,760 --> 01:13:06,920 Robots will not give any. They don't care if we're polite. 1401 01:13:06,920 --> 01:13:09,400 They're not going to. But I'm modeling in an 1402 01:13:09,400 --> 01:13:13,760 interaction with a human like entity to my children and to 1403 01:13:13,760 --> 01:13:16,480 myself. And so do I want to be the kind 1404 01:13:16,480 --> 01:13:21,360 of person who's a jerk and shouts, you know, at the no. 1405 01:13:21,360 --> 01:13:23,520 Like, I don't want my children to be like that either. 1406 01:13:23,520 --> 01:13:26,480 So I say please and thank you. I try not to be. 1407 01:13:26,640 --> 01:13:29,620 I mean, sometimes it is. It gets a little bit annoying 1408 01:13:29,660 --> 01:13:30,860 whenever it gets everything wrong. 1409 01:13:30,860 --> 01:13:35,260 All right, my final. Will AI kill us all? 1410 01:13:35,260 --> 01:13:38,380 Like do you? Is it in in the sort of end of 1411 01:13:38,380 --> 01:13:42,260 humanity chances? Where do you put sort of robots 1412 01:13:42,260 --> 01:13:45,700 and AI being sort of the cause of our demise? 1413 01:13:46,980 --> 01:13:50,020 You know, I don't. I think again, it's those 1414 01:13:50,020 --> 01:13:52,540 psychotic apes. You know, if you're gonna read 1415 01:13:52,540 --> 01:13:56,940 a, if you're gonna read a story about the the role of machines 1416 01:13:56,940 --> 01:14:01,530 in the demise of Man, I would read There Will Come Soft Rains 1417 01:14:01,650 --> 01:14:05,810 by Ray Bradbury. And they'll just be keeping on 1418 01:14:05,810 --> 01:14:07,290 keeping on trying to do their thing. 1419 01:14:07,290 --> 01:14:09,970 And it'll be us. We we kill each other nuclear. 1420 01:14:09,970 --> 01:14:12,010 War, but the machines keep going. 1421 01:14:12,090 --> 01:14:13,570 Keep on trying to do their thing, man. 1422 01:14:15,050 --> 01:14:17,450 Daniel H Wilson, thank you. So much for coming on the show. 1423 01:14:17,450 --> 01:14:19,490 I really it was a great time to chat with you. 1424 01:14:19,930 --> 01:14:20,850 Cool, man. It was a pleasure. 1425 01:14:21,130 --> 01:14:22,450 That's our episode. Thanks. 1426 01:14:22,450 --> 01:14:25,470 So much to Max Child and James Wilsterman of Volley, my Co 1427 01:14:25,470 --> 01:14:27,130 host. I'm Eric Newcomer. 1428 01:14:27,640 --> 01:14:32,040 Shout out to Scott Brody, our producer Riley Kinsella, Chief 1429 01:14:32,040 --> 01:14:35,600 of Staff for newcomer Gabby Caliendo at Volley, who's 1430 01:14:35,600 --> 01:14:37,920 helping so much with the conference behind the scenes. 1431 01:14:38,400 --> 01:14:41,040 Oh, and of course to young Chomsky for the wonderful theme 1432 01:14:41,040 --> 01:14:43,280 music. This episode is part of the 1433 01:14:43,280 --> 01:14:48,360 Cerebral Valley series that I'm doing on the Newcomer Podcast. 1434 01:14:49,000 --> 01:14:53,440 You can follow along on my sub stack at newcomer.co where I'm 1435 01:14:53,440 --> 01:14:56,760 publishing each episode, or you can follow it on YouTube. 1436 01:14:57,300 --> 01:14:59,500 Or Apple Podcasts. Or wherever you get your 1437 01:14:59,500 --> 01:15:01,820 podcasts. Thanks so much. 1438 01:15:01,820 --> 01:15:02,900 Goodbye. Goodbye. 1439 01:15:02,980 --> 01:15:03,380 Goodbye. Goodbye. 1440 01:15:03,700 --> 01:15:05,220 Goodbye, Goodbye, Goodbye.