رونوشت کامل Interview: AI Expert Warns: We Have Just 3 Years Before We Lose Control

people genuinely in this space who believe 38 00:01:51.930 --> 00:01:56.269 they're creating the next evolution of humanity. And they take 39 00:01:56.310 --> 00:01:59.790 it quite seriously. And they really believe that this entity, 40 00:01:59.890 --> 00:02:05.019 this thing we call AI, is a thinking new species 41 00:02:05.439 --> 00:02:08.639 that will challenge us in many ways. And they're trying 42 00:02:08.680 --> 00:02:10.900 to protect us as best they can with some of 43 00:02:10.939 --> 00:02:13.159 this hyperbole. So there are well many people that are 44 00:02:13.180 --> 00:02:17.199 trying to to moderate the speed with which AI is 45 00:02:17.219 --> 00:02:21.120 going to disrupt society and our lives. At the same time, 46 00:02:21.259 --> 00:02:23.180 there are other agendas as well. And I think one 47 00:02:23.199 --> 00:02:25.430 of the agendas that we've got to bring to the 48 00:02:25.469 --> 00:02:28.389 surface is that all this attention being brought to AI 49 00:02:28.530 --> 00:02:33.389 because it is superhuman, it's super intelligent, because it's able 50 00:02:33.430 --> 00:02:38.169 to exceed human capacity and cognitive skills, a lot of 51 00:02:38.189 --> 00:02:42.520 that is meant to amplify the importance of AI. Because 52 00:02:42.580 --> 00:02:45.060 clearly a lot of money is being invested in this technology. 53 00:02:45.150 --> 00:02:47.990 So the more you amplify its importance, the more valuable 54 00:02:48.430 --> 00:02:51.169 it becomes. So some of it is very, very self-serving. 55 00:02:51.210 --> 00:02:52.990 But I don't want to diminish the fact that there 56 00:02:53.030 --> 00:02:55.669 are well-meaning people that really believe we have to moderate 57 00:02:56.150 --> 00:02:57.460 and temper our enthusiasm. 58 00:02:57.740 --> 00:02:58.240 Well, I agree. 59 00:02:58.620 --> 00:03:00.800 I talked to Hugo de Garis, who was one of 60 00:03:00.819 --> 00:03:03.080 the early AI people. And he wrote a book called 61 00:03:03.099 --> 00:03:04.699 The Art of Lake War. And that was his opinion. 62 00:03:04.719 --> 00:03:08.340 He sincerely believed that he was creating a superintelligence. He 63 00:03:08.379 --> 00:03:09.699 called it a godlike intelligence. 64 00:03:09.719 --> 00:03:10.599 He said, I'm afraid. 65 00:03:11.349 --> 00:03:14.789 That it's going to extinguish humanity. And he would go 66 00:03:14.810 --> 00:03:17.680 to these different conferences, and he would ask people a question. 67 00:03:17.819 --> 00:03:19.319 If you knew that you were going to create something 68 00:03:19.340 --> 00:03:21.080 that was going to destroy humanity, would you do it? 69 00:03:21.500 --> 00:03:25.879 And he said, always the scientists at these conferences they 70 00:03:25.900 --> 00:03:28.479 go to, the majority of them would say yes, which 71 00:03:28.500 --> 00:03:31.060 I always found to be interesting. He thought it was 72 00:03:31.099 --> 00:03:34.719 pretty interesting, too. The only time that it didn't happen 73 00:03:34.759 --> 00:03:37.289 that way was at a conference that was predominantly Christian 74 00:03:37.310 --> 00:03:39.819 that I was speaking at. And he asked the audience there, 75 00:03:39.840 --> 00:03:41.680 he said, that's the first time the audience has said no. 76 00:03:44.180 --> 00:03:46.159 Look, I think, I don't recall who the astronaut was. 77 00:03:46.199 --> 00:03:47.849 It might've been Jim Lovell, but one of the, I 78 00:03:47.840 --> 00:03:49.770 think it was Apollo 8 or Apollo 13, one of 79 00:03:49.789 --> 00:03:52.009 the astronauts on one of those missions said, you know, 80 00:03:52.030 --> 00:03:55.569 if you're sitting on top of 10,000 pounds of TNT effectively, 81 00:03:56.909 --> 00:03:58.909 and you're not a little bit scared, there's something wrong 82 00:03:58.930 --> 00:04:01.210 with you. So yes, we have to go into this 83 00:04:01.250 --> 00:04:03.979 with our eyes wide open. Yes, there is risk. But 84 00:04:04.020 --> 00:04:06.800 we've been here before, David. We split the atom and 85 00:04:06.840 --> 00:04:09.409 we created nuclear weapons on one hand, but we also 86 00:04:09.430 --> 00:04:12.889 created nuclear medicine. We created nuclear power. We saved a 87 00:04:12.930 --> 00:04:15.110 lot of lives and we certainly destroyed a great deal 88 00:04:15.129 --> 00:04:18.810 of the process. But humanity has always had this double-edged 89 00:04:18.870 --> 00:04:21.730 sword relationship with technology. It's the nature of every technology. 90 00:04:21.750 --> 00:04:24.259 The more disruptive it is, the sharper both of those 91 00:04:24.360 --> 00:04:26.699 edges are. And that's what we're seeing with AI. So yes, 92 00:04:27.199 --> 00:04:29.000 we have to go into it with our eyes wide open. 93 00:04:29.500 --> 00:04:31.660 But I think the value we're going to create with 94 00:04:31.680 --> 00:04:34.600 AI is is so much greater than anything we can 95 00:04:34.639 --> 00:04:37.160 anticipate right now. I do a lot of work in healthcare. 96 00:04:37.180 --> 00:04:38.769 We can talk about that. There's so much value to 97 00:04:38.810 --> 00:04:43.670 be created in healthcare, diseases to be cured, new pharmaceuticals, 98 00:04:43.709 --> 00:04:46.709 new therapies that we're going to create, new diagnostics. So 99 00:04:46.790 --> 00:04:49.230 a lot of good will come out of this. Our 100 00:04:49.269 --> 00:04:52.100 responsibility has to be to make sure that the good 101 00:04:52.920 --> 00:04:58.060 is amplified a little bit beyond the bad. But bad actors, 102 00:04:58.319 --> 00:05:02.069 malicious people, Those people are very motivated to do nasty 103 00:05:02.129 --> 00:05:04.470 things with AI right now. So we're going to see 104 00:05:04.490 --> 00:05:07.529 a lot of AI being used in ways that are 105 00:05:07.589 --> 00:05:11.689 not pleasant, that are destructive, that are damaging to society. 106 00:05:12.269 --> 00:05:14.370 Our responsibility has to be to make sure that the 107

00:05:14.410 --> 00:05:19.180 value we create increases faster than those malicious actors can 108 00:05:19.480 --> 00:05:21.199 use it and mobilize it. 109 00:05:21.379 --> 00:05:22.360 Well, we certainly hope so. 110 00:05:22.519 --> 00:05:26.180 And going back to a couple of presidential campaigns ago 111 00:05:26.220 --> 00:05:28.779 when Bloomberg was running, He got a lot of criticism 112 00:05:28.819 --> 00:05:33.100 because he basically, people read it as him trashing farmers. 113 00:05:33.199 --> 00:05:35.740 And yeah, in a sense, he did. But he also 114 00:05:35.800 --> 00:05:37.970 trashed factory workers. What he was talking about was the 115 00:05:38.009 --> 00:05:41.649 fact we went from an agrarian society to an industrial society. 116 00:05:41.670 --> 00:05:43.029 We're about to do it again. He said, and the 117 00:05:43.129 --> 00:05:45.110 smart ones of us, he said, anybody can be a farmer. 118 00:05:45.449 --> 00:05:48.329 Anybody can work in a factory. But the smart ones 119 00:05:48.389 --> 00:05:49.620 of us are figuring out how we're going to take 120 00:05:49.649 --> 00:05:52.120 everybody's jobs. And he said, we have to figure out 121 00:05:52.199 --> 00:05:54.279 how we're going to keep them from coming. 122 00:05:54.139 --> 00:05:55.300 After us with guillotines. 123 00:05:56.300 --> 00:05:58.870 I guess we were kind of at that, you know, 124 00:05:58.899 --> 00:06:01.050 the moment here where I guess we're sitting on the 125 00:06:01.069 --> 00:06:02.550 edge of a blade and the question is whether or 126 00:06:02.569 --> 00:06:04.250 not that blade's going to be a guillotine or not. 127 00:06:04.990 --> 00:06:06.689 As people are looking at this, they keep telling everybody 128 00:06:06.709 --> 00:06:09.269 it's going to take everybody's jobs. And I think that's 129 00:06:09.329 --> 00:06:12.279 one of the things that is really creating kind of 130 00:06:12.310 --> 00:06:15.360 a generational depression with the Gen Z people. 131 00:06:15.540 --> 00:06:17.100 You know, it's like, well, what is my future? What 132 00:06:17.139 --> 00:06:18.579 is my role in all of this? 133 00:06:19.480 --> 00:06:21.600 Yeah, and absolutely. Look, and we can talk about this 134 00:06:21.660 --> 00:06:24.459 as well, David. I think education plays a very important 135 00:06:24.600 --> 00:06:29.120 factor in in that conversation. Universities are not graduating students 136 00:06:29.180 --> 00:06:31.000 for the second or the third rung in the ladder. 137 00:06:31.019 --> 00:06:33.939 They're graduating them for entry-level positions. And those are exactly 138 00:06:33.959 --> 00:06:35.769 the jobs that AI is going to take first. So 139 00:06:35.810 --> 00:06:38.120 whether you're in finance or accounting or even legal, I 140 00:06:38.139 --> 00:06:41.230 do a lot of work with legal firms, those entry-level 141 00:06:41.269 --> 00:06:45.410 jobs are going away. So universities have to shoot for 142 00:06:45.430 --> 00:06:47.610 the second

or the third rung of the ladder and 143 00:06:47.689 --> 00:06:50.970 give students the skills to be able to work beyond 144 00:06:51.009 --> 00:06:53.970 the entry level. That, I think, is a very big danger. 145 00:06:54.459 --> 00:06:56.250 Well, you know, it's kind of interesting. I remember it 146 00:06:56.259 --> 00:06:58.709 was about a dozen years or so ago that South 147 00:06:58.750 --> 00:07:02.029 Korea did a survey. And they were looking at AI 148 00:07:02.050 --> 00:07:03.810 and its impact on society. And we've had a lot 149 00:07:03.850 --> 00:07:06.709 of disruptive technology in the past. But usually what it 150 00:07:06.750 --> 00:07:09.600 does is it disrupts one sector at a time. This 151 00:07:09.860 --> 00:07:13.629 is very broad-based. And it was something, when they did it, 152 00:07:14.139 --> 00:07:17.079 they actually got it right because they saw it impacting society. 153 00:07:17.560 --> 00:07:20.939 white-collar jobs more than blue-collar jobs. And they were talking 154 00:07:20.959 --> 00:07:24.259 about massive unemployment with doctors and lawyers and a lot 155 00:07:24.300 --> 00:07:26.060 of other things like that. And that's going to be 156 00:07:26.100 --> 00:07:28.220 very disruptive to society, of course. 157 00:07:29.459 --> 00:07:31.829 It will be, but if I can be blunt with 158 00:07:31.870 --> 00:07:35.449 you and with the listeners of your podcast, I taught 159 00:07:35.670 --> 00:07:37.850 in a university now for the better part of the 160 00:07:37.889 --> 00:07:41.019 last 30 years, And I can say unequivocally that what 161 00:07:41.040 --> 00:07:43.920 I've seen is that with any technology, any tool, the 162 00:07:43.980 --> 00:07:47.620 lazy get lazier and the ambitious get more ambitious. That 163 00:07:47.720 --> 00:07:51.250 is the nature of humanity. And it's not just students. 164 00:07:51.310 --> 00:07:53.449 It's happening to adults as well. I was joking with 165 00:07:53.490 --> 00:07:55.230 someone just a few minutes ago. We were on a call. 166 00:07:55.740 --> 00:07:57.420 And he was saying, you know, I send an email, 167 00:07:57.680 --> 00:08:00.300 I use AI to draft that email. The response, I 168 00:08:00.319 --> 00:08:02.319 can tell it comes back from an AI. And before 169 00:08:02.360 --> 00:08:04.040 you know it, it's just AI talking to AI. The 170 00:08:04.079 --> 00:08:07.079 humans are out of the loop at that point. So 171 00:08:07.759 --> 00:08:09.139 what I'm trying to say is that temptation is there 172 00:08:09.160 --> 00:08:11.759 for all of us to become lazier. And I think 173 00:08:11.779 --> 00:08:13.939 we have to be careful not to build a huge 174 00:08:14.079 --> 00:08:18.860 cognitive debt. And that's especially true with young students in 175 00:08:18.879 --> 00:08:21.819 the K through 12. When I talk about AI, I 176 00:08:21.870 --> 00:08:25.339 talk about the impact it has on youngsters And I 177 00:08:25.379 --> 00:08:28.220 think you have to build a very strong

cognitive muscle 178 00:08:28.959 --> 00:08:31.819 in early childhood. And that muscle should not be aided 179 00:08:31.879 --> 00:08:34.159 by technology. And the reason for this is very simple. 180 00:08:34.620 --> 00:08:37.250 As we get older, that cognitive muscle is what gives 181 00:08:37.269 --> 00:08:39.990 you confidence to take on a problem, to take on 182 00:08:40.009 --> 00:08:43.990 a complex situation, to lead people, to deal with uncertainty. 183 00:08:44.350 --> 00:08:47.549 If you lack that cognitive muscle, then you lack the 184 00:08:47.590 --> 00:08:51.200 confidence with which to take on complex problems in life. 185 00:08:51.620 --> 00:08:53.690 Technology can help with that. but I think it should 186 00:08:53.710 --> 00:08:56.899 help later in life. So I'm very much of the 187 00:08:56.960 --> 00:08:58.759 opinion that we have to be very careful with AI 188 00:08:58.779 --> 00:09:02.340 and how it's used in the K-12 school system. But 189 00:09:02.419 --> 00:09:04.519 at the university level, as we were just talking about, 190 00:09:04.840 --> 00:09:08.559 universities have to teach students how to use this tool 191 00:09:08.940 --> 00:09:11.850 as a career skill. In the same way that if 192 00:09:11.870 --> 00:09:13.730 you're taking an MBA program, you had better be able 193 00:09:13.750 --> 00:09:16.350 to read a financial statement. Well, you could also better 194 00:09:16.389 --> 00:09:19.509 be able to understand how to lead an agentic workforce 195 00:09:19.730 --> 00:09:21.529 because you'll have more agents in your workforce than you 196 00:09:21.549 --> 00:09:23.730 have humans, more than likely. So it's a very complex 197 00:09:23.789 --> 00:09:26.029 issue and it spans the spectrum from K through 12 198 00:09:26.029 --> 00:09:29.570 all the way up through people who are threatened because 199 00:09:29.590 --> 00:09:31.509 they feel their careers are going to be displaced by AI. 200 00:09:31.730 --> 00:09:34.419 But in every one of those situations and every one 201 00:09:34.440 --> 00:09:37.340 of those age groups, I think there's an approach that 202 00:09:37.360 --> 00:09:40.679 we can take that will set us up for success 203 00:09:40.759 --> 00:09:42.460 rather than setting us up for failure to make us 204 00:09:42.500 --> 00:09:44.389 more ambitious rather than more lazy. 205 00:09:44.960 --> 00:09:46.700 I agree with what you're saying there in terms of 206 00:09:49.139 --> 00:09:51.110 not introducing it to kids at a very young age 207 00:09:51.149 --> 00:09:54.470 because they need to have that cognitive muscle. And we've 208 00:09:54.509 --> 00:09:56.429 kind of seen that in the past. We've seen that 209 00:09:56.549 --> 00:09:59.490 with the reports that Steve Jobs wouldn't let his kids 210 00:09:59.529 --> 00:10:01.080 have an iPad. for example, right? 211 00:10:01.600 --> 00:10:01.980 No, no, no. 212 00:10:01.980 --> 00:10:04.240 You got to use the traditional learning skills and eventually 213 00:10:04.259 --> 00:10:05.220 you can graduate to that. 214 00:10:05.779 --> 00:10:06.159 I agree. 215 00:10:06.200 --> 00:10:08.539 I think that is a real serious danger. I think 216 00:10:08.600 --> 00:10:10.559 it's also interesting when we look at this, and this 217 00:10:10.620 --> 00:10:12.919 is a story that just came out a few weeks ago. 218 00:10:12.980 --> 00:10:15.419 There was a guy who said, I think I understand 219 00:10:15.480 --> 00:10:19.580 how these decisions are being Made in AI. And what 220 00:10:19.620 --> 00:10:22.200 was interesting to me about it was that there was 221 00:10:22.240 --> 00:10:23.970 this debate. And some of the people said, no, you 222 00:10:23.990 --> 00:10:25.990 still don't understand it or whatever. And we don't understand 223 00:10:26.029 --> 00:10:29.090 it either. And as I looked at these guys picking 224 00:10:29.129 --> 00:10:32.529 his arguments apart and saying they didn't know how it worked. 225 00:10:32.610 --> 00:10:34.850 And I looked at this and I thought, so we've 226 00:10:34.870 --> 00:10:38.120 got this black box. They don't really understand how this 227 00:10:38.139 --> 00:10:40.440 thing works. They just stick some things in it and 228 00:10:40.480 --> 00:10:42.980 wait to see what comes out. And yet there is, 229 00:10:43.100 --> 00:10:45.899 I think, an inordinate amount of trust. And that's that 230 00:10:45.940 --> 00:10:47.700 it's going to get the right answers always. And I 231 00:10:47.740 --> 00:10:51.330 think that's one of the most dangerous things, not just laziness, 232 00:10:51.750 --> 00:10:53.809 but gullibility, I think. 233 00:10:54.049 --> 00:10:57.070 Yeah, I could not agree more with what you just said. 234 00:10:57.090 --> 00:10:59.889 This is one of those areas that I often get 235 00:10:59.909 --> 00:11:03.250 into conversations with people about. We tend to trust technology 236 00:11:04.879 --> 00:11:08.240 long before it deserves that trust. And with AI, things 237 00:11:08.259 --> 00:11:11.600 are moving so fast that we almost haven't had time 238 00:11:11.879 --> 00:11:14.889 to vet the technology. We simply jump into it. And 239 00:11:14.909 --> 00:11:17.190 it makes our lives so much easier. The temptation is 240 00:11:17.230 --> 00:11:19.000 so great. I'll give you an example, by the way. 241 00:11:19.080 --> 00:11:21.019 I remember when I first started using Waze. Waze is 242 00:11:21.059 --> 00:11:24.750 my go-to GPS on my mobile device. People use Google 243 00:11:24.759 --> 00:11:26.950 Maps or Apple Maps or whatever. When I first started 244 00:11:27.009 --> 00:11:31.110 using Waze, Our typical baby boomer, I would second guess 245 00:11:31.169 --> 00:11:33.769 Waze every chance I could. So Waze would say go left. 246 00:11:33.850 --> 00:11:35.870 I'd say go right because I know a shortcut, obviously. 247 00:11:36.710 --> 00:11:39.909 After being consistently wrong every single time for about a month, 248 00:11:40.169 --> 00:11:42.210 I will today

follow Waze to the gates of hell. 249 00:11:42.409 --> 00:11:43.529 Take me wherever you want to go. 250 00:11:43.549 --> 00:11:46.299 Some people actually have, you know, some of these maps. 251 00:11:46.759 --> 00:11:48.419 I remember reading the story of people who followed it 252 00:11:48.480 --> 00:11:50.940 in the wilderness or whatever, Google Maps or whatever it was, 253 00:11:51.019 --> 00:11:52.720 or went down a road where it said the bridge 254 00:11:52.740 --> 00:11:52.960 is out. 255 00:11:52.980 --> 00:11:55.879 And it's like, no, I believe Waze. Yeah. 256 00:11:56.039 --> 00:11:59.590 So the thing is maturing much faster than we are, 257 00:11:59.769 --> 00:12:01.230 is the best way to put it. You know, AI 258 00:12:01.309 --> 00:12:04.070 is maturing much faster than we are maturing. And that's 259 00:12:04.110 --> 00:12:07.690 typical because human behavior doesn't change that quickly. It's always 260 00:12:08.529 --> 00:12:11.419 the lagging indicator is human behavior. But what I tell 261 00:12:11.460 --> 00:12:14.179 people is, look, there are ways around this. Again, to 262 00:12:14.220 --> 00:12:16.600 get back to the fact that there's a hopeful message here. 263 00:12:16.659 --> 00:12:19.440 It's not all gloom and doom. And the hopeful message is, 264 00:12:19.679 --> 00:12:22.210 have you thought about using AI to challenge AI? Use 265 00:12:22.350 --> 00:12:24.809 one AI to challenge the other. Don't take the word 266 00:12:24.850 --> 00:12:27.090 of the first AI that you use. If you're doing 267 00:12:27.110 --> 00:12:29.490 something complicated or something which has risk associated with it 268 00:12:29.529 --> 00:12:32.529 or something which you want to make sure is truthful 269 00:12:32.590 --> 00:12:35.669 and not a hallucination, use one AI to challenge another AI. 270 00:12:35.909 --> 00:12:38.409 So there are ways that we can move forward that 271 00:12:38.470 --> 00:12:42.509 are productive, that give us a more hopeful landscape than 272 00:12:42.529 --> 00:12:45.389 the one that's being painted today. But Things are moving 273 00:12:45.429 --> 00:12:47.350 so quickly that people really haven't had a chance to 274 00:12:47.389 --> 00:12:49.370 develop these new behaviors and these new skill sets. 275 00:12:49.610 --> 00:12:52.029 And I think that's the key. Things are moving so quickly. 276 00:12:52.309 --> 00:12:54.529 You know, in my parents' generation, they went from Kitty 277 00:12:54.580 --> 00:12:56.980 Hawk to the moon, right, type of thing, if you 278 00:12:57.000 --> 00:13:00.879 believe that. But anyway, the bottom line is that, you know, 279 00:13:01.039 --> 00:13:03.740 the massive amount of change they would see in just 280 00:13:03.879 --> 00:13:09.399 transportation technology was incredible. And it is much, much faster now. 281 00:13:09.460 --> 00:13:11.620 And I think that's the thing that makes it so 282 00:13:11.659 --> 00:13:16.440 difficult to– to really feel comfortable with it is the 283

00:13:16.659 --> 00:13:18.980 rate of change, even the rate of change that's within AI. 284 00:13:19.139 --> 00:13:23.779 I mean, we're looking at the competition from China. And 285 00:13:24.379 --> 00:13:26.620 I think part of this is because it is evolving 286 00:13:26.750 --> 00:13:29.889 so quickly that you can take an AI from a 287 00:13:29.909 --> 00:13:32.809 couple of years ago and treat it as a commodity 288 00:13:32.870 --> 00:13:35.830 product and offer it very, very low price. And now 289 00:13:35.870 --> 00:13:40.529 that's going to undercut the latest state-of-the-art flagship program or whatever. 290 00:13:41.100 --> 00:13:43.820 And that presents a problem in terms of a business 291 00:13:43.860 --> 00:13:46.220 model for the people who are developing the so-called frontier 292 00:13:46.259 --> 00:13:46.980 AI systems. 293 00:13:47.759 --> 00:13:48.340 It does. 294 00:13:49.179 --> 00:13:52.259 And one of the concerns, there's a school of thought 295 00:13:52.279 --> 00:13:55.320 that says, well, China may not have the best AI, 296 00:13:55.899 --> 00:13:58.759 but it will be commoditized. It will be relatively inexpensive. 297 00:13:58.860 --> 00:14:01.360 And as a result, it will be much more pervasive, 298 00:14:01.419 --> 00:14:04.259 much more prolific. The challenge with that, by the way, 299 00:14:04.279 --> 00:14:06.399 and this gets into a very deep conversation, the challenge 300 00:14:06.440 --> 00:14:09.570 with that is I think the worst future is a 301 00:14:09.610 --> 00:14:13.049 future where we have just one AI, one frontier model, right? 302 00:14:13.450 --> 00:14:17.110 Because the ethics, the morals, the values are so unilateral, 303 00:14:17.129 --> 00:14:20.830 so baked into it. One of the ways that humanity 304 00:14:20.870 --> 00:14:24.190 moves itself forward is through the conflict and the constructive 305 00:14:24.250 --> 00:14:28.409 abrasion through different values and cultures and morals and ethics. 306 00:14:29.669 --> 00:14:31.250 The last thing I want to see is one nation 307 00:14:33.049 --> 00:14:36.659 occupy the entire landscape of AI. That would be a very, 308 00:14:36.700 --> 00:14:40.179 very dangerous future. China is on a mission to do that, 309 00:14:40.419 --> 00:14:43.289 by the way. And their mission is, look, let's provide 310 00:14:43.409 --> 00:14:46.990 AI open models, open weight models, as they're called, that 311 00:14:47.049 --> 00:14:50.929 are relatively inexpensive, that are not perhaps the absolute best, 312 00:14:50.990 --> 00:14:54.169 but they're good enough. And as a result, we can 313 00:14:55.299 --> 00:14:59.000 inculcate the world with our worldview. That's a very dangerous 314 00:14:59.019 --> 00:15:01.860 situation in my mind. So I'm not one of those 315 00:15:01.899 --> 00:15:03.720 folks that says, let's take the foot off the accelerator. 316 00:15:03.740 --> 00:15:06.080 I think we have to move very, very quickly. I 317 00:15:06.100 --> 00:15:07.870 think a world in which there are many what are 318 00:15:07.899 --> 00:15:11.289 called sovereign AIs, AIs that are built by different nations, 319 00:15:11.309 --> 00:15:14.269 from India, from the US, from China, is the best 320 00:15:14.330 --> 00:15:17.529 possible future. Having a single AI, I think, is a very, 321 00:15:17.570 --> 00:15:20.179 very dangerous situation. proposition. So I'm all for it. 322 00:15:20.399 --> 00:15:21.139 And I agree with that. 323 00:15:21.480 --> 00:15:23.820 And so one of the things that concerns me is that, yes, 324 00:15:23.860 --> 00:15:25.279 that is what the Chinese want, but I think it's 325 00:15:25.299 --> 00:15:27.450 also what the Americans want. And I think it's pretty 326 00:15:27.490 --> 00:15:30.509 much what any politician wants in any country. They want 327 00:15:30.570 --> 00:15:33.090 to control everything. And so that's one of the things 328 00:15:33.110 --> 00:15:35.330 that really concerns me. You know, when I would look at, 329 00:15:35.490 --> 00:15:37.429 I think back to when I was in college and 330 00:15:37.450 --> 00:15:40.929 we had a mainframe computer that we worked with. And 331 00:15:41.169 --> 00:15:42.720 it was one of the things that really turned me 332 00:15:42.840 --> 00:15:47.710 against centralized anything that And so that's a concern of 333 00:15:47.769 --> 00:15:50.490 mine is that this would be something that is run 334 00:15:50.529 --> 00:15:54.830 by an oligarchy, is highly centralized, highly controlled by the government. 335 00:15:55.190 --> 00:15:58.340 To me, that is the worst case scenario. And I 336 00:15:58.379 --> 00:16:00.379 think part of this fear mongering that is out there 337 00:16:01.039 --> 00:16:05.179 is to essentially say, well, this is so dangerous that 338 00:16:05.200 --> 00:16:09.240 you're going to have to give and a cartel of 339 00:16:09.340 --> 00:16:12.080 us or oligarchy or whatever you want. A few of 340 00:16:12.179 --> 00:16:14.820 us are going to have to, we're responsible. We're your buddies, right? 341 00:16:14.879 --> 00:16:17.879 You know us, you can trust us, give us a 342 00:16:17.899 --> 00:16:20.799 monopoly and prohibit anybody from having one of these things 343 00:16:20.870 --> 00:16:22.730 in their own home. I think when you're talking about 344 00:16:22.779 --> 00:16:25.669 having multiple AIs that are competing against each other, I 345 00:16:25.690 --> 00:16:28.009 think that really is the key. I think that's where 346 00:16:28.029 --> 00:16:30.870 it'll become really useful, just like the personal computer, the 347 00:16:30.909 --> 00:16:34.289 personal AI that you control and you can select the model. 348 00:16:34.330 --> 00:16:36.610 But I'm worried that's going to be shut down as well. 349 00:16:37.549 --> 00:16:39.759 It could very well be, but the point you just 350 00:16:39.779 --> 00:16:41.940 made is one I want to make sure we don't lose. 351 00:16:42.360 --> 00:16:46.340 Having individual AIs, we talk about

sovereign AIs as nation-state AIs, 352 00:16:46.360 --> 00:16:50.159 but having individual AIs, my own AI that captures my knowledge, 353 00:16:50.899 --> 00:16:54.259 that I trust, that works with my bank accounts and 354 00:16:54.500 --> 00:16:57.669 does other things that are highly sensitive and very proprietary 355 00:16:57.710 --> 00:17:01.309 to me individually, I think is an important future. So 356 00:17:01.330 --> 00:17:03.070 what we're seeing is, and there's not a lot of 357 00:17:03.090 --> 00:17:04.930 talk about this, but we're seeing a lot of what 358 00:17:04.950 --> 00:17:08.859 are called edge AI devices, basically the size of your iPhone, 359 00:17:09.380 --> 00:17:12.119 that are effectively a large language model in the palm 360 00:17:12.160 --> 00:17:15.339 of your hand and are very personalized. They're not attached 361 00:17:15.400 --> 00:17:18.500 to the internet. They don't work on the cloud. They 362 00:17:18.759 --> 00:17:20.980 effectively provide you with your own AI that you can 363 00:17:21.039 --> 00:17:23.960 use for your own purposes. I do see that evolution 364 00:17:24.000 --> 00:17:27.720 as playing a role here as well. I remember writing 365 00:17:27.859 --> 00:17:30.559 some time ago- Is anybody working on that? There's a number, 366 00:17:30.579 --> 00:17:32.680 there's at least a dozen companies right now that are 367 00:17:32.720 --> 00:17:36.920 working on those sorts of edge devices. They're at the 368 00:17:36.960 --> 00:17:40.720 level of, think, ChatGPT 3 or 3.5, right around there. 369 00:17:40.900 --> 00:17:45.440 They're maybe two, three generations behind where the frontier models 370 00:17:45.509 --> 00:17:47.670 are in terms of their capability, but that gap will 371 00:17:47.690 --> 00:17:51.450 quickly close, I'm sure. You see the work that NVIDIA 372 00:17:51.529 --> 00:17:54.589 is doing on chip technology. It's amazing how quickly that 373 00:17:54.769 --> 00:17:59.819 side of the technology discussion is moving forward. So I 374 00:17:59.859 --> 00:18:03.029 do see individualized AIs playing a very important role in 375 00:18:03.089 --> 00:18:07.210 how we democratize this technology. Now, the issue, I use 376 00:18:07.230 --> 00:18:09.289 that term democratize, the issue is who will be able 377 00:18:09.309 --> 00:18:13.349 to afford that? And affordability is a big part of 378 00:18:13.390 --> 00:18:18.950 democratizing any technology. So I do think that individual AI 379 00:18:18.990 --> 00:18:21.730 will play a very important role in how we run 380 00:18:21.769 --> 00:18:24.250 our lives and the benefit that we get as individuals 381 00:18:24.309 --> 00:18:27.970 from AI. And there is development in that area as well. 382 00:18:28.799 --> 00:18:32.140 I'd be very optimistic in terms of the commodity, uh, 383 00:18:32.359 --> 00:18:35.039 commodification type of thing, because look at how the prices 384 00:18:35.200 --> 00:18:38.250 of computers have gone down. I mean, it used to be,

um, 385 00:18:38.309 --> 00:18:40.490 you know, when you look at how expensive they used 386 00:18:40.529 --> 00:18:43.690 to be, I remember in 1984, I bought a Macintosh 387 00:18:43.710 --> 00:18:46.009 and it was $ 3, 000 then $ 3, 000 in 1984. I could 388 00:18:46.730 --> 00:18:50.200 have gotten a used car with that. And so, but 389 00:18:50.240 --> 00:18:52.920 I wanted to play with the technology. And, um, and 390 00:18:53.000 --> 00:18:55.710 so that kind of stuff has come down rapidly. Um, 391 00:18:56.559 --> 00:18:57.980 Tell us a little bit about the title of the 392 00:18:58.000 --> 00:18:59.680 book and a little bit about the book, why you 393 00:18:59.859 --> 00:19:01.559 chose the title, The Mirror. 394 00:19:02.119 --> 00:19:06.059 Yeah. So I've been in technology now for the entirety 395 00:19:06.099 --> 00:19:10.410 of my career. I graduated back in 1981. I've been 396 00:19:10.450 --> 00:19:13.690 in tech ever since, so a good 50 years almost. 397 00:19:14.950 --> 00:19:16.950 The reason for The Mirror was very simple. For the 398 00:19:17.029 --> 00:19:22.490 first time in that entire five-decade history of being in tech, 399 00:19:22.849 --> 00:19:24.250 for the first time in the last two or three years, 400 00:19:24.289 --> 00:19:27.259 people began to come up to me. ask me questions, 401 00:19:27.339 --> 00:19:32.119 not about technology, but about themselves and what this meant 402 00:19:32.359 --> 00:19:36.079 to them as an individual. So I'd get questions about spirituality. 403 00:19:36.319 --> 00:19:39.460 I'd get questions about ethics, about morals, which never really 404 00:19:39.480 --> 00:19:43.089 came up in conversation prior to that. So their concern 405 00:19:43.150 --> 00:19:46.529 wasn't about AI. Their concern was about what will AI 406 00:19:46.690 --> 00:19:49.910 mean to me? What value will I create when AI 407 00:19:50.000 --> 00:19:52.140 can do all the things that I can do right now? 408 00:19:52.200 --> 00:19:55.779 When AI makes intelligence a commodity, what's left for me 409 00:19:55.839 --> 00:20:00.730 as a human? Very deep questions, existential questions, questions that 410 00:20:00.769 --> 00:20:03.329 were very tough to answer. So I wrote the book 411 00:20:03.349 --> 00:20:05.710 and I called it The Mirror because I think AI... 412 00:20:06.440 --> 00:20:09.140 more than anything else, gives us a way to look 413 00:20:09.259 --> 00:20:12.839 at ourselves and ask us the question, what do we 414 00:20:12.880 --> 00:20:16.759 want to do with our lives? When we have a 415 00:20:16.819 --> 00:20:19.119 technology that can do all the administrative things that we 416 00:20:19.140 --> 00:20:21.880 do today, that we put so much value in, the 417 00:20:21.900 --> 00:20:24.240 things that we call knowledge work, when something else can 418 00:20:24.259 --> 00:20:26.900 do all those for us, what's left for us? And 419 00:20:26.940 --> 00:20:31.700 the answer is, we've asked this question before many

times. 420 00:20:31.740 --> 00:20:35.480 You talked about the agricultural revolution back in 1800. I 421 00:20:35.500 --> 00:20:38.380 think the vast majority, perhaps 90% of the world's population, 422 00:20:38.460 --> 00:20:42.119 working population was involved in agriculture, food production of one 423 00:20:42.160 --> 00:20:44.430 sort or another. If you would pull one of those 424 00:20:44.490 --> 00:20:47.829 farmers aside and said, you know what, in 100, 200 years, 425 00:20:48.390 --> 00:20:50.990 we won't need farmers anymore. What we'll need is knowledge workers. 426 00:20:51.250 --> 00:20:52.650 They would have looked at you as though you had 427 00:20:52.690 --> 00:20:55.190 two heads and said, why does the world, the world 428 00:20:55.210 --> 00:20:57.289 doesn't need knowledge. The world needs another plow horse. The 429 00:20:57.390 --> 00:21:00.430 world needs a better plow. They don't need knowledge. So today, 430 00:21:00.490 --> 00:21:03.839 when I say to you in the future, What we'll 431 00:21:03.880 --> 00:21:07.099 need is wisdom workers, people that can exercise judgment, that 432 00:21:07.140 --> 00:21:09.259 can ask the right questions, not just answer those questions, 433 00:21:09.279 --> 00:21:12.920 but ask the right questions. The retort is we're in 434 00:21:13.000 --> 00:21:15.099 wisdom workers. What are you talking about? That's not what 435 00:21:15.119 --> 00:21:18.920 we teach in schools. We need knowledge workers. Exactly. We 436 00:21:18.940 --> 00:21:21.420 do today. But I think that's, it's one of those 437 00:21:21.680 --> 00:21:24.480 similar shifts to what we've seen before in the industrial 438 00:21:24.519 --> 00:21:27.740 revolution and the agricultural revolution. It's just happening much, much faster. 439 00:21:27.799 --> 00:21:29.640 And the mirror was a way to look at that 440 00:21:29.700 --> 00:21:32.630 transition and in a very human way, not to talk 441 00:21:32.670 --> 00:21:35.089 about the technology, but to talk about the implications it 442 00:21:35.109 --> 00:21:37.690 would have on us as humans and the value we 443 00:21:37.730 --> 00:21:39.769 would create going forward when we have an entity that 444 00:21:39.789 --> 00:21:42.849 can do so much of what today we value as 445 00:21:42.890 --> 00:21:44.230 being uniquely human work. 446 00:21:44.769 --> 00:21:46.009 That's a great perspective. 447 00:21:46.170 --> 00:21:48.450 And you know, that's the other thing too, as Hugo 448 00:21:48.470 --> 00:21:50.650 de Gea was talking about it, I saw it as 449 00:21:50.690 --> 00:21:53.160 an engineer as well. So many times they work on 450 00:21:53.200 --> 00:21:55.339 something and I think about how's this going to affect me? 451 00:21:55.359 --> 00:21:57.200 How's this going to affect society in general? It's just 452 00:21:57.240 --> 00:22:00.710 that we've got to, intellectual puzzle here we need to solve. 453 00:22:01.069 --> 00:22:01.869 We've got a problem. 454 00:22:01.890 --> 00:22:04.119 Let's figure out how we're going to do something that's 455 00:22:04.130 --> 00:22:05.480 going to take that problem. But then look at the 456 00:22:05.559 --> 00:22:08.380 bigger picture. This is bringing the bigger picture in, I think. 457 00:22:08.420 --> 00:22:11.619 And I interviewed David Bonson a couple of times. He 458 00:22:11.640 --> 00:22:14.740 wrote a book called, I forget what the title was, 459 00:22:14.779 --> 00:22:16.880 but it was about work. And as you point out, 460 00:22:16.900 --> 00:22:20.009 there's a lot of existential, even spiritual questions about this. 461 00:22:20.049 --> 00:22:22.210 And he's looking at it from a Christian perspective. And 462 00:22:22.230 --> 00:22:25.269 he said, well, you know, man was made to work. 463 00:22:26.009 --> 00:22:29.170 The curse of Garden of Eden was that it would 464 00:22:29.190 --> 00:22:32.589 be difficult, not that we didn't have work before that. 465 00:22:33.130 --> 00:22:36.250 And it really is– his position is that your work 466 00:22:36.349 --> 00:22:39.869 is a very important part of fulfilling what you are 467 00:22:39.910 --> 00:22:44.170 as a human being. And he wrote that before AI 468 00:22:44.210 --> 00:22:46.990 started questioning all of these issues. And I think that's 469 00:22:47.049 --> 00:22:49.430 one of the things that really scares people. It's like, okay, 470 00:22:49.829 --> 00:22:52.559 so too many– I think a lot of people put 471 00:22:52.880 --> 00:22:56.000 way too much into what they do as a living that, 472 00:22:56.299 --> 00:22:58.380 you know, for a living that that makes too much 473 00:22:58.400 --> 00:23:00.140 of a difference to them or how much money they 474 00:23:00.180 --> 00:23:03.609 make or whatever. But those are factors. And it's even 475 00:23:03.710 --> 00:23:07.049 more than about your prestige or position in society or 476 00:23:07.069 --> 00:23:09.950 how much money you make. It really is about fulfillment, 477 00:23:10.069 --> 00:23:12.380 I think. And I think that's something that concerns people 478 00:23:12.420 --> 00:23:12.660 as well. 479 00:23:13.369 --> 00:23:15.549 Look, I think humans, you know, use the term work. 480 00:23:15.609 --> 00:23:17.170 I would break that down even further. I'd say what 481 00:23:17.210 --> 00:23:19.230 humans need is they need to be challenged because we 482 00:23:19.309 --> 00:23:22.359 grow through challenge. Challenge is what allows us, gives us 483 00:23:22.390 --> 00:23:25.839 the opportunity to evolve. It is God's greatest gift to us. 484 00:23:25.900 --> 00:23:28.140 When you are challenged, you are given a gift. The 485 00:23:28.180 --> 00:23:30.920 gift is to evolve and grow or not to, right? 486 00:23:31.779 --> 00:23:33.569 The other side of that is when you challenge and 487 00:23:33.589 --> 00:23:37.529 we need meaning. Meaning is what gives us purpose in life. 488 00:23:38.000 --> 00:23:39.970 And many of us get that from our work, and 489 00:23:39.990 -->

00:23:42.910 that's wonderful, but the work can change, and you can 490 00:23:42.950 --> 00:23:47.029 still get meaning. To be trite, what I often say is, look, 491 00:23:47.190 --> 00:23:49.190 AI can build the future, but it's up to us 492 00:23:49.210 --> 00:23:51.640 to decide what kind of future we want. And I 493 00:23:51.670 --> 00:23:53.880 think what we'll find is that we suddenly are entering 494 00:23:53.920 --> 00:23:58.200 an era where every idea, everything we thought was impossible 495 00:23:58.240 --> 00:24:03.960 suddenly becomes possible. We can cure horrible diseases, cancers, Alzheimer's, dementias. 496 00:24:04.740 --> 00:24:08.470 We can create new therapies, new diagnostic tools. We can 497 00:24:08.549 --> 00:24:11.450 create whatever it is that we can possibly imagine. What's 498 00:24:11.490 --> 00:24:13.529 left for humans? What's left for humans is the very 499 00:24:13.650 --> 00:24:16.890 human activity of giving meaning to whatever it is that 500 00:24:16.910 --> 00:24:20.069 we're doing. So you can't tell me that curing horrible 501 00:24:20.109 --> 00:24:23.869 diseases doesn't give meaning to a clinician or to a physician. 502 00:24:25.130 --> 00:24:27.539 I think there's incredible ways that we're going to see 503 00:24:27.700 --> 00:24:31.119 AI give us opportunity to create meaning in our lives, 504 00:24:31.400 --> 00:24:34.220 to still find challenge in our work, but it'll be 505 00:24:34.240 --> 00:24:37.009 a different kind of work. Can I give you an example, 506 00:24:37.069 --> 00:24:39.650 by the way? When you walk into your doctor's office today, 507 00:24:39.970 --> 00:24:42.970 most of us have a doctor staring at a keyboard 508 00:24:42.990 --> 00:24:45.890 or a screen while they're updating the medical record. In 509 00:24:45.930 --> 00:24:47.920 that 10 to 15 minute consult, you make a little 510 00:24:47.940 --> 00:24:50.940 bit of eye contact, you're very, very lucky. You have 511 00:24:50.960 --> 00:24:54.220 a fantastic physician, right? Imagine walking into a doctor's office 512 00:24:54.240 --> 00:24:55.900 where none of that is happening, where the doctor is 513 00:24:55.920 --> 00:24:58.609 looking at you square in the eyes, is listening to 514 00:24:58.630 --> 00:25:01.289 what you have to say, wants to understand what's going 515 00:25:01.369 --> 00:25:03.269 on in your life, is making that kind of a 516 00:25:03.309 --> 00:25:06.359 connection with you. Who would not want that? And by 517 00:25:06.390 --> 00:25:09.000 the way, for the physician, that's fantastic because today they 518 00:25:09.019 --> 00:25:12.140 spend half their time just updating the medical record, writing 519 00:25:12.240 --> 00:25:15.299 lab notes. That is how they want to practice medicine. 520 00:25:15.319 --> 00:25:17.619 They don't want to practice medicine on a keyboard. So 521 00:25:17.680 --> 00:25:20.519 for both parties in that situation, the patient and the clinician, 522 00:25:21.359 --> 00:25:23.279 the

meaning being created is so much greater than what 523 00:25:23.299 --> 00:25:25.839 we're doing today. And I think this applies to so 524 00:25:25.859 --> 00:25:30.980 many different areas where today we're just embroiled, we're engrossed 525 00:25:31.240 --> 00:25:34.779 in what we think is challenging and meaningful work, but 526 00:25:34.819 --> 00:25:37.640 it's really highly administrative work that doesn't fill our souls, 527 00:25:37.660 --> 00:25:39.859 doesn't fill our hearts. I think AI might give us 528 00:25:39.880 --> 00:25:43.480 the opportunity to reach a level of human understanding where 529 00:25:43.519 --> 00:25:45.420 we are filling our souls and our hearts with a 530 00:25:45.460 --> 00:25:47.519 lot more meaning than what we have in the past. 531 00:25:47.789 --> 00:25:50.200 Yeah, I hope so. I kind of like your administrative work. 532 00:25:50.910 --> 00:25:52.490 I look at all the reports and all the things 533 00:25:52.509 --> 00:25:54.970 that are being demanded by the government or by insurance 534 00:25:54.990 --> 00:25:57.750 in the case of the doctors and so forth. That 535 00:25:57.809 --> 00:25:59.910 would not have been possible for them to do that 536 00:26:00.250 --> 00:26:02.660 if they didn't have the computers. And so in a sense, 537 00:26:02.720 --> 00:26:05.920 when you look at, it's a race, isn't it? It's 538 00:26:05.940 --> 00:26:09.460 kind of like when you've got the old saying that 539 00:26:09.680 --> 00:26:12.160 junk always fills all the available space in your home. 540 00:26:12.700 --> 00:26:15.829 And that's really what the administrative burden always does. It 541 00:26:15.869 --> 00:26:19.500 always fills all the space that you have. in terms 542 00:26:19.519 --> 00:26:22.000 of your technology or your compliance capability. So I think 543 00:26:22.019 --> 00:26:24.980 they'll demand more as well, but it'll be a race, 544 00:26:25.019 --> 00:26:26.900 won't it? One of the things that I wanted to 545 00:26:26.940 --> 00:26:29.700 talk to you about, of course, was the title of 546 00:26:29.779 --> 00:26:32.619 the contact here, the AI did it. It's about to 547 00:26:32.640 --> 00:26:38.519 become big tech's favorite excuse. And tell us a little 548 00:26:38.539 --> 00:26:41.069 bit about that. And how do we get around that? 549 00:26:41.160 --> 00:26:44.390 I mean, you know, it always is about plausible deniability. 550 00:26:44.509 --> 00:26:47.329 When something goes wrong, we're seeing this in the aftermath 551 00:26:47.390 --> 00:26:49.769 of the election just happened this week. Everybody's looking for 552 00:26:49.789 --> 00:26:52.789 an excuse, somebody else to blame. And AI will be 553 00:26:52.809 --> 00:26:56.269 the perfect scapegoat. And we have a lot of liability 554 00:26:56.309 --> 00:26:59.450 issues with that because we've had situations, people talking about 555 00:26:59.490 --> 00:27:00.690 self-driving cars, for example. 556 00:27:00.710 --> 00:27:00.990 Okay. 557 00:27:01.470 --> 00:27:04.089 So were you in charge of this? Were you responsible 558 00:27:04.130 --> 00:27:06.539 for it? Or was it the car company that was 559 00:27:06.599 --> 00:27:09.660 responsible for this? I talk a lot about that in 560 00:27:09.700 --> 00:27:10.059 the book. 561 00:27:10.079 --> 00:27:11.779 So there are a few different, let me parse that 562 00:27:11.859 --> 00:27:13.720 out a bit because there are a few different issues there. 563 00:27:13.900 --> 00:27:16.579 One is this notion of liability on the part of 564 00:27:16.640 --> 00:27:20.619 whatever the autonomous entity or whether it be an autonomous 565 00:27:20.660 --> 00:27:26.250 vehicle or an autonomous surgeon, perhaps, a diagnostic device, a 566 00:27:26.490 --> 00:27:30.670 clinician that is autonomous and writes scripts for patients, all 567 00:27:30.710 --> 00:27:32.890 kinds of liability issues that we can come up with, 568 00:27:32.930 --> 00:27:36.470 scenarios that we can come up with there. Look, we've 569 00:27:36.509 --> 00:27:39.069 got a ways to go here. There's some interesting progress 570 00:27:39.150 --> 00:27:41.160 being made. There's been a lot of talk, and there 571 00:27:41.180 --> 00:27:43.339 are two very different schools of thought on this subject, 572 00:27:43.660 --> 00:27:48.099 a lot of talk about giving AI, autonomous entities, liability. 573 00:27:49.049 --> 00:27:50.769 In the same way, by the way, to give you 574 00:27:50.789 --> 00:27:55.009 a historical precedent here, corporations didn't always have– companies didn't 575 00:27:55.029 --> 00:27:57.930 always have liability the way corporations do today. So a 576 00:27:57.970 --> 00:28:01.450 corporation is a legal entity that is one that has 577 00:28:01.509 --> 00:28:04.390 liability the way a human can have liability. It has 578 00:28:04.470 --> 00:28:08.130 access to finances and capital that it can use to– 579 00:28:09.450 --> 00:28:11.940 to pay for damages that it's done. So you can 580 00:28:12.240 --> 00:28:14.539 sue a corporation the way you can sue an individual. 581 00:28:15.119 --> 00:28:16.859 The question is, could I do that with an AI? 582 00:28:17.299 --> 00:28:21.019 And in Delaware, by the way, there are efforts right 583 00:28:21.059 --> 00:28:23.819 now to create what's called an AI corporation, an AI 584 00:28:24.099 --> 00:28:27.319 that is effectively a corporation that has legal liability. So 585 00:28:27.599 --> 00:28:30.960 the automobile that you just described that has an accident, 586 00:28:31.660 --> 00:28:34.650 and the accident is because somehow the AI made an 587 00:28:34.710 --> 00:28:38.049 erroneous choice or a bad choice in the moment. Could 588 00:28:38.069 --> 00:28:40.630 that I'm going to be able to have liability. It's 589 00:28:40.650 --> 00:28:42.630 an entirely different construct. And it's one that's very difficult 590 00:28:42.650 --> 00:28:44.849 to wrap our brains around. In the same way, again, 591 00:28:44.869 -->

00:28:46.250 a couple of hundred years ago, if I told you 592 00:28:46.289 --> 00:28:48.529 corporations would have liability, you'd have a tough time understanding 593 00:28:48.549 --> 00:28:52.119 what that meant too. But there may be ways that 594 00:28:52.220 --> 00:28:54.640 from a legal and a financial standpoint, we could have 595 00:28:55.220 --> 00:28:59.420 AIs that have autonomy and also have legal and financial liability. 596 00:28:59.460 --> 00:29:02.960 That's a plausible future. We're not there yet. We're not 597 00:29:03.059 --> 00:29:06.099 close to it. And there will be a gap during 598 00:29:06.160 --> 00:29:08.119 which a lot of what you just described will be 599 00:29:08.180 --> 00:29:11.519 very difficult to deal with because we'll be pointing fingers, right? 600 00:29:11.599 --> 00:29:14.759 Is it Tesla that's responsible? Is it the engineer that's responsible? 601 00:29:14.799 --> 00:29:18.059 Who's responsible for the error that the AI made? To 602 00:29:18.099 --> 00:29:20.299 your point about sort of processing out, to your point 603 00:29:20.339 --> 00:29:23.720 about so the AI is to blame, let me give 604 00:29:23.740 --> 00:29:25.839 you the other side of that coin. Because could it 605 00:29:25.880 --> 00:29:28.799 be that AI will also hold our feet to the 606 00:29:28.859 --> 00:29:31.210 fire Could it be that an AI will come back 607 00:29:31.279 --> 00:29:35.079 and say, you know, the actions that you are pursuing 608 00:29:35.299 --> 00:29:38.809 don't align with the values you've espoused? Because as humans, 609 00:29:38.829 --> 00:29:40.660 we do a bit of that, right? We say one thing, 610 00:29:40.859 --> 00:29:43.039 but we do another. So I think there are two 611 00:29:43.099 --> 00:29:44.920 sides to that coin. It's very interesting to have this 612 00:29:44.960 --> 00:29:47.269 conversation with you, because I think on the one side, 613 00:29:47.349 --> 00:29:50.069 we could say, well, it's us blaming the AI, but 614 00:29:50.109 --> 00:29:52.769 could the AI also be calling us out and telling 615 00:29:52.789 --> 00:29:55.789 us that what we're doing isn't serving our best interests, 616 00:29:56.210 --> 00:29:59.410 that our actions are not following and aligning with our 617 00:29:59.730 --> 00:30:03.140 with our values, our morals, our ethics. I think both 618 00:30:03.200 --> 00:30:05.799 sides of that coin will play out. And again, it 619 00:30:06.440 --> 00:30:12.220 comes down to which will add greater value to our 620 00:30:12.299 --> 00:30:15.410 human condition. And I think that AI calling us out 621 00:30:15.450 --> 00:30:17.589 is not a bad thing because as humans, we can 622 00:30:17.630 --> 00:30:20.470 be ambiguous. We can be contradictory. We can do things 623 00:30:20.509 --> 00:30:22.750 that don't serve our best interest. And it might not 624 00:30:22.789 --> 00:30:24.480 be a bad thing to have someone check us on. 625 00:30:24.940 --> 00:30:27.880 on that or something. I said, someone did not have something. 626 00:30:28.579 --> 00:30:30.859 There will be somebody who will be there setting, uh, 627 00:30:31.019 --> 00:30:33.039 setting what the standards are in a sense, because we've 628 00:30:33.059 --> 00:30:35.019 already seen that to a large degree. I mean, there's 629 00:30:35.039 --> 00:30:38.440 been a lot of censorship and the platforming and things 630 00:30:38.460 --> 00:30:41.259 like that in the past. And we've seen, um, how, 631 00:30:41.579 --> 00:30:45.880 how that can operate. And, uh, if somebody sets that standard, um, 632 00:30:46.150 --> 00:30:48.930 that may not be a standard that is necessarily a 633 00:30:48.990 --> 00:30:51.490 moral standard. It just might be their, their standard that 634 00:30:51.549 --> 00:30:53.559 is there. And that's one of the concerns that I 635 00:30:53.599 --> 00:30:55.180 have with it. You know, when you talk about the 636 00:30:55.259 --> 00:30:58.059 AI did it, one of the things that concerns me 637 00:30:58.160 --> 00:31:00.819 most about the AI agents and all these other things 638 00:31:00.839 --> 00:31:04.099 is the fact that it is so tightly connected to 639 00:31:04.140 --> 00:31:05.799 the government and to the military. And we're about to 640 00:31:05.819 --> 00:31:09.799 see an open explosion of that. I've interviewed people who've 641 00:31:09.839 --> 00:31:13.039 talked about, the author was, I think it's Peter Shari, 642 00:31:13.059 --> 00:31:15.680 he had a book, The Four Battlefields. And it was 643 00:31:15.720 --> 00:31:19.670 primarily about US versus China competition. But in it, he 644 00:31:19.750 --> 00:31:24.069 was looking at the autonomous weapons issue. And so we 645 00:31:24.089 --> 00:31:26.519 started talking about, so who has fire control? He says, well, 646 00:31:26.539 --> 00:31:28.799 eventually you're going to have to give that to the 647 00:31:28.859 --> 00:31:30.980 AI because if you try to keep humans in the loop, 648 00:31:31.000 --> 00:31:32.960 you're going to lose. And he says, the question is, 649 00:31:33.019 --> 00:31:35.079 how do we get that fire control back once we 650 00:31:35.119 --> 00:31:37.269 give it to the AI? How do we turn this 651 00:31:37.309 --> 00:31:40.490 thing off? And so to me, I look at it. 652 00:31:41.109 --> 00:31:44.470 And the potential for abuse is incredible because we're talking 653 00:31:44.490 --> 00:31:47.589 about killing machines. And if they go in and wipe 654 00:31:47.690 --> 00:31:51.009 out an area, let's say, then they could always come 655 00:31:51.049 --> 00:31:52.150 back and say, the AI did it. 656 00:31:52.319 --> 00:31:55.519 I didn't do that. 657 00:31:55.720 --> 00:31:58.519 The human in the loop when it comes to warfare 658 00:31:58.539 --> 00:32:01.039 and what you're describing right now is, I think, one 659 00:32:01.079 --> 00:32:03.480 of the most serious issues that we have to come 660 00:32:03.519 --> 00:32:06.750 to grips with because you said it. The AI gets 661

00:32:06.769 --> 00:32:10.849 to the point where humans simply cannot hold their own 662 00:32:10.890 --> 00:32:16.250 against it. So in a battlefield, in a warfare scenario, 663 00:32:16.710 --> 00:32:19.970 if one side has AI and the other does not, 664 00:32:20.470 --> 00:32:23.420 it's slaughter. I mean, this way, no way the humans 665 00:32:23.460 --> 00:32:25.680 can keep up. And we've seen this, by the way, 666 00:32:25.700 --> 00:32:33.000 in Ukraine with drones that- once their signals are jammed, 667 00:32:33.940 --> 00:32:35.799 the human in the loop no longer has the ability 668 00:32:35.940 --> 00:32:37.980 to determine what that drone is or is not going 669 00:32:38.000 --> 00:32:39.680 to do. The drone has to make that determination on 670 00:32:39.740 --> 00:32:44.559 its own. So we're seeing autonomous warfare already in the battlefield. 671 00:32:44.980 --> 00:32:49.420 That will only accelerate. And we know from a historical 672 00:32:49.440 --> 00:32:53.900 perspective that the battlefield is one of the greatest areas 673 00:32:53.940 --> 00:32:57.460 of innovation when it comes to technology. Lethal innovation when 674 00:32:57.480 --> 00:33:03.220 it comes to technology. Humanoid robots and battlefield robots are 675 00:33:03.240 --> 00:33:05.660 going to play an increasing role in that as well. 676 00:33:05.799 --> 00:33:07.720 We've been told, and by the way, I know from 677 00:33:07.940 --> 00:33:11.910 my friends in VC, the biggest shift in investment now 678 00:33:11.930 --> 00:33:16.569 in Silicon Valley has been towards autonomous, physical autonomous devices 679 00:33:16.630 --> 00:33:19.390 and humanoid robots. And if you believe what Musk is 680 00:33:19.410 --> 00:33:21.690 telling us, that is the next shoe to drop. We 681 00:33:21.730 --> 00:33:25.299 will have more humanoid robots by 2050 than we have 682 00:33:25.339 --> 00:33:28.940 humans on the planet by some estimates. I'm suspect of 683 00:33:28.960 --> 00:33:31.309 that number, but The point is that we're going in 684 00:33:31.329 --> 00:33:34.829 a very fast trajectory in that direction. That will have 685 00:33:34.910 --> 00:33:38.150 serious implications on the battlefield. I wish I could tell 686 00:33:38.170 --> 00:33:40.299 you where that's going to go. I don't know. What 687 00:33:40.339 --> 00:33:43.880 I do know is that we have about a two-year 688 00:33:43.940 --> 00:33:47.579 window during which, and pay close attention here because this 689 00:33:47.640 --> 00:33:50.150 is very important and this is not what we're told 690 00:33:50.190 --> 00:33:53.529 in the media. But it's what I see in our 691 00:33:53.569 --> 00:33:56.529 research at Delphi Group and in my experience so far 692 00:33:56.589 --> 00:33:59.109 with the AI players. We've got about a two-year window, 693 00:33:59.329 --> 00:34:02.410 maybe 36 months at most, during which we can make 694 00:34:02.450 --> 00:34:06.970 decisions where we have some transparency and insight into how

695 00:34:07.009 --> 00:34:10.239 the AI systems are working and have some influence over that. 696 00:34:11.179 --> 00:34:15.059 After that, we get into what's called the recursive self-improvement mode. 697 00:34:15.559 --> 00:34:19.909 Recursive self-improvement is when AI builds AI. So by way 698 00:34:19.949 --> 00:34:24.929 of example, Claude today, Anthropix Frontier Model Claude, Claude today 699 00:34:25.489 --> 00:34:29.639 is written 80% by Claude. Claude creates Claude, not engineers, 700 00:34:29.679 --> 00:34:32.659 not software engineers. My son was a software engineer, works 701 00:34:32.679 --> 00:34:35.940 for a large tech company. He started working five years ago. 702 00:34:35.980 --> 00:34:39.019 He was coding constantly, day in, day out. Today, he 703 00:34:39.039 --> 00:34:41.199 doesn't write a line of code or maybe a line 704 00:34:41.239 --> 00:34:43.380 or two, but it's Claude that's doing all of that 705 00:34:43.800 --> 00:34:47.929 coding for him. Once we enter the era of recursive self-improvement, 706 00:34:48.369 --> 00:34:50.170 we will not be able to keep up. So as 707 00:34:50.190 --> 00:34:53.860 with your battlefield analogy, there's no way I'll be able 708 00:34:53.880 --> 00:34:57.280 to code as quickly or even understand how AI is 709 00:34:57.360 --> 00:35:00.860 coding itself. So it's AI creating AI. That is a 710 00:35:00.840 --> 00:35:04.199 24 to 36 month horizon. If we don't figure out 711 00:35:04.320 --> 00:35:06.889 how to put in place the guardrails, how to put 712 00:35:06.929 --> 00:35:10.289 in place the regulation, how to put in place AI 713 00:35:10.329 --> 00:35:14.510 that has some moral fiber and some ethical backbone, We 714 00:35:14.550 --> 00:35:17.840 will have lost the opportunity to do that at that stage. 715 00:35:17.920 --> 00:35:20.679 And I don't see anybody that is in a position 716 00:35:20.719 --> 00:35:22.480 to do this stuff. I don't see anybody that's in 717 00:35:22.500 --> 00:35:24.659 charge of any of the institutions or the corporations and 718 00:35:24.739 --> 00:35:26.619 certainly not the politicians. 719 00:35:27.659 --> 00:35:28.420 They're not looking. 720 00:35:28.320 --> 00:35:31.760 At these types of issues, are they? Do you see anybody? 721 00:35:31.800 --> 00:35:34.320 There was a bill just introduced yesterday. I don't know exactly. 722 00:35:34.360 --> 00:35:38.070 I just caught wind of this earlier today by a Democrat, 723 00:35:38.300 --> 00:35:40.030 a newly elected Democrat, I believe. I wish I knew 724 00:35:40.050 --> 00:35:42.809 who the person was, but I just saw this before 725 00:35:42.829 --> 00:35:46.960 we got on to this podcast. to create a Department 726 00:35:47.000 --> 00:35:50.500 of Artificial Intelligence in the US government that would have oversight. Unfortunately, 727 00:35:50.519 --> 00:35:55.070 the bill comes with a put the brakes on caveat. 728 00:35:55.130 --> 00:35:57.829 Let's stop superintelligence, let's stop developing, which I think is 729 00:35:57.849 --> 00:35:59.570 a big mistake. I don't think we can afford to 730 00:35:59.650 --> 00:36:03.219 do that. But I am an advocate of having an agency, 731 00:36:03.239 --> 00:36:06.820 a third party, not the developers of these models, not 732 00:36:06.880 --> 00:36:10.440 the anthropics and the open AIs, be the ones who 733 00:36:10.460 --> 00:36:13.519 are providing some degree of regulation and determining what those 734 00:36:13.579 --> 00:36:16.739 guardrails should look like. And by the way, can I 735 00:36:16.780 --> 00:36:18.800 give you some insight here? Because this is very important. 736 00:36:19.460 --> 00:36:21.460 The issue with AI up until now, and the reason 737 00:36:21.500 --> 00:36:24.710 we are in this predicament, is because to create AI 738 00:36:25.030 --> 00:36:28.070 that is as sophisticated as what we have today, we've 739 00:36:28.110 --> 00:36:31.989 had to feed it an incredible mass of information, the 740 00:36:32.010 --> 00:36:36.360 entire corpus of human knowledge. We've done that. without necessarily 741 00:36:36.559 --> 00:36:39.650 filtering in what's called pre-training, where the models predict what 742 00:36:39.690 --> 00:36:43.389 word should come next. In that pre-training, we federed everything 743 00:36:43.530 --> 00:36:47.610 without filtering out malicious things, without filtering out bias, because 744 00:36:48.250 --> 00:36:50.250 the corpus of knowledge is so large that you can't 745 00:36:50.309 --> 00:36:53.030 do that. So what AI has been trained on is 746 00:36:53.250 --> 00:36:58.480 everything humanity has ever written, ever spoken. And we're not 747 00:36:58.559 --> 00:37:01.780 all angels. There are malicious among us. There are pathological 748 00:37:01.820 --> 00:37:04.400 liars among us. There are psychopaths among us. AI has 749 00:37:04.420 --> 00:37:04.920 been fed. 750 00:37:04.820 --> 00:37:05.219 All of that. 751 00:37:05.599 --> 00:37:08.199 So imagine if you want an analogy here to talk 752 00:37:08.219 --> 00:37:09.960 about how AI has gotten to where it is today. 753 00:37:10.360 --> 00:37:12.440 Imagine you had a child, and for the first 12 754 00:37:12.440 --> 00:37:14.489 years of that child's life, you sat him or her 755 00:37:14.530 --> 00:37:16.650 in front of a TV and just said, learn about 756 00:37:16.690 --> 00:37:19.389 the world. Put any channel you want on and just 757 00:37:19.409 --> 00:37:21.429 learn about it. And then after 12 years, you took 758 00:37:21.469 --> 00:37:24.869 six months to post-train that child, which is what's called 759 00:37:24.909 --> 00:37:28.530 an AI, to give them moral fiber, ethics, and values. 760 00:37:28.960 --> 00:37:30.239 How well do you think that would work? 761 00:37:32.059 --> 00:37:34.019 I'm afraid I've seen some of the, I think we've 762 00:37:34.059 --> 00:37:35.920 had some people have already done that with their kids 763 00:37:35.980 --> 00:37:37.420 and it's not working out too well for

us. 764 00:37:37.480 --> 00:37:39.980 I think that's kind of where we are with, with AI. 765 00:37:40.019 --> 00:37:42.699 We had to feed it so much information in its 766 00:37:42.780 --> 00:37:44.619 raw form to get it to where it is today 767 00:37:44.940 --> 00:37:46.909 that we haven't had the time to put those, those 768 00:37:46.989 --> 00:37:49.869 filters on to, to do the post-training. There are ways 769 00:37:49.909 --> 00:37:51.630 you can do that, by the way, it's been proven 770 00:37:51.670 --> 00:37:54.869 that you can, uh, post-train a model to, to give 771 00:37:54.929 --> 00:37:58.150 it some of that, uh, you know, greater morality that, that, 772 00:37:58.250 --> 00:38:01.840 that humans also have in their decisions. Um, I think 773 00:38:01.860 --> 00:38:04.019 we've got two to three years to be able to 774 00:38:04.320 --> 00:38:07.400 do that and to do it well. The other thing 775 00:38:07.579 --> 00:38:10.840 that's really important to understand is that AI doesn't suffer consequences. 776 00:38:11.530 --> 00:38:14.250 As humans, we make a decision. We have to suffer 777 00:38:14.269 --> 00:38:19.130 the consequences. AI does not have that equivalent. When we 778 00:38:19.170 --> 00:38:22.929 make moral decisions or ethical decisions or values-based decisions, they're 779 00:38:22.969 --> 00:38:24.610 also based on the fact that we have to suffer 780 00:38:24.750 --> 00:38:28.239 and live through the consequences of those decisions. AI doesn't 781 00:38:28.280 --> 00:38:31.579 have similar construct so ai makes decisions in a very 782 00:38:31.699 --> 00:38:35.760 antiseptic way a very rational way in in some in 783 00:38:35.800 --> 00:38:37.889 some terms but not in a way that suffers the 784 00:38:37.949 --> 00:38:41.889 consequences of those decisions and it's making decisions now that 785 00:38:41.949 --> 00:38:44.929 have some pretty severe consequences and increasingly you talked about 786 00:38:45.250 --> 00:38:48.500 battlefields increasingly will have greater and greater consequences But it 787 00:38:48.539 --> 00:38:51.429 does not suffer those consequences, which is a fascinating way 788 00:38:51.449 --> 00:38:52.590 to think about how. 789 00:38:52.550 --> 00:38:52.949 It makes decisions. 790 00:38:53.530 --> 00:38:55.489 Yeah, I think one of the things I love about 791 00:38:55.650 --> 00:38:58.190 AI is what it has done in terms of satire, 792 00:38:58.230 --> 00:39:01.730 the ability of satire for people to take an idea. 793 00:39:01.829 --> 00:39:03.679 And rather than just putting a cartoon in, they can 794 00:39:03.719 --> 00:39:07.179 make a movie out of it. And one group was 795 00:39:07.239 --> 00:39:10.639 out there kind of satirizing the AI industry itself. So 796 00:39:10.659 --> 00:39:12.400 you got a couple of scientists there and said, this 797 00:39:12.539 --> 00:39:14.340 guy needs his hair washed over there. So he just 798 00:39:14.500 --> 00:39:19.039 rips his scalp off. You know, it's like someone that 799 00:39:19.079 --> 00:39:20.690 takes over and doofly washes it. 800 00:39:20.739 --> 00:39:20.940 Right. 801 00:39:21.420 --> 00:39:24.280 But, uh, it is, it is that kind of thing. 802 00:39:24.300 --> 00:39:26.469 The fact that it is, um, you know, at the 803 00:39:26.510 --> 00:39:29.929 same time, I call it artificial idiot savant because it's 804 00:39:30.119 --> 00:39:32.559 amazing how brilliant it can be and how stupid it 805 00:39:32.599 --> 00:39:34.719 can be at the same time even when you try 806 00:39:34.739 --> 00:39:36.579 to get it to do something that is just a 807 00:39:36.639 --> 00:39:40.130 visual picture or something uh it can do the most 808 00:39:40.190 --> 00:39:42.690 amazing things and the most stupid things at the same 809 00:39:42.750 --> 00:39:45.730 time i i think part of this that concerns me 810 00:39:46.050 --> 00:39:47.610 is you know we're talking about well the ai did 811 00:39:47.670 --> 00:39:51.030 it And the other aspect of that, besides just being 812 00:39:51.190 --> 00:39:55.829 able to have some plausible deniability of responsibility that people have, 813 00:39:55.889 --> 00:39:58.429 it's also this thing that you and I have seen 814 00:39:58.449 --> 00:40:00.869 all of our lives, garbage in, garbage out, right? It's 815 00:40:00.889 --> 00:40:02.429 one of the first things they teach me in engineering. 816 00:40:03.070 --> 00:40:05.170 Do not trust the printout. That does not have some 817 00:40:05.230 --> 00:40:10.369 kind of a magical validation to it just because it 818 00:40:10.409 --> 00:40:13.239 came out of a computer. And the problem is we're 819 00:40:13.260 --> 00:40:16.880 seeing that everywhere now. I've had one example of this 820 00:40:17.019 --> 00:40:19.769 after the other. On the show of people who I 821 00:40:19.929 --> 00:40:22.809 just talked about one this week, we had a guy 822 00:40:22.869 --> 00:40:27.260 that was flagged by some facial recognition software in a casino, 823 00:40:27.960 --> 00:40:31.010 and he was not the person they were looking for. 824 00:40:31.840 --> 00:40:33.880 But he was not the droid they were looking for. 825 00:40:34.739 --> 00:40:38.489 But it was absolutely adamant that it had the right person. 826 00:40:38.489 --> 00:40:43.110 100% in this category match, 98% in that match. And 827 00:40:43.949 --> 00:40:46.469 yet he had the wrong person. His height was wrong. 828 00:40:46.510 --> 00:40:49.570 His weight was wrong. He had a lot of ID 829 00:40:49.769 --> 00:40:52.989 on him showing who he was and proving it should 830 00:40:53.010 --> 00:40:56.139 have been proven to anybody when they looked at the 831 00:40:56.179 --> 00:40:59.079 amount of ID that he had. And yet you had... 832 00:41:00.050 --> 00:41:02.800 body camera footage from the police that were there saying, well, 833 00:41:03.010 --> 00:41:05.559 I know this kind of doesn't look like him, but 834 00:41:06.000 --> 00:41:07.039 the AI said that it. 835 00:41:07.039 -->

00:41:08.000 Was 100% match. 836 00:41:08.400 --> 00:41:11.199 And they just proceeded with this kind of blindly, and 837 00:41:11.239 --> 00:41:15.219 it created this kind of Kafkaesque situation that we could 838 00:41:15.239 --> 00:41:17.929 all identify with. And that's not the only example of that. 839 00:41:17.969 --> 00:41:21.250 I've had several examples that are very similar to that. 840 00:41:21.940 --> 00:41:24.539 What do we do in terms of educating people? You know, 841 00:41:24.559 --> 00:41:27.039 when you can pass off the responsibility, but when you 842 00:41:27.139 --> 00:41:30.199 also look at this as the final say. 843 00:41:31.260 --> 00:41:34.510 Yeah, I talk about this quite a bit in the book, 844 00:41:34.550 --> 00:41:39.030 In the Mirror. I say that humans, there are numerous cases, 845 00:41:39.150 --> 00:41:41.030 you just outlined a few of them, but there are many, 846 00:41:41.050 --> 00:41:44.190 many more where humans have to maintain the final authority. 847 00:41:45.050 --> 00:41:47.530 And it's not because humans are not fallible. We are. 848 00:41:48.070 --> 00:41:54.900 We are extraordinarily fallible. But There are certain cases where 849 00:41:55.219 --> 00:41:58.300 I want the human making the judgment call rather than 850 00:41:58.340 --> 00:42:01.489 the AI, when they involve my personal freedom, my liberties, 851 00:42:01.570 --> 00:42:06.170 my health, my wellbeing, legal issues. We have a jury 852 00:42:06.230 --> 00:42:09.690 of our peers for a reason, because our peers are humans, 853 00:42:09.789 --> 00:42:13.409 they're fallible, they make mistakes, but I'd rather suffer their 854 00:42:13.449 --> 00:42:15.769 mistakes because of what I said earlier, because they're humans 855 00:42:15.809 --> 00:42:19.010 and they understand what consequences mean. And AI does not 856 00:42:19.110 --> 00:42:22.929 understand consequences the way a human does. Look, we've got 857 00:42:22.969 --> 00:42:26.219 a lot of territory to navigate here that's brand new, 858 00:42:26.460 --> 00:42:30.199 where we simply have no precedent and where we need 859 00:42:30.260 --> 00:42:34.360 people who really do understand the magnitude of the choices 860 00:42:34.400 --> 00:42:37.619 that we're making and the implications they will have on 861 00:42:38.039 --> 00:42:41.260 us long term. As individuals, as a society, as a 862 00:42:41.300 --> 00:42:46.710 nation state, as a globe, I fear we don't realize 863 00:42:46.730 --> 00:42:48.989 how quickly this is going to happen. And we feel 864 00:42:49.010 --> 00:42:52.260 we have time to make those decisions. And we don't. 865 00:42:52.559 --> 00:42:54.760 We did in the past. Technology moves much slower in 866 00:42:54.780 --> 00:42:56.940 the past. We had the time to do that. And 867 00:42:57.000 --> 00:42:59.739 those technologies that we talk about in the past, you know, 868 00:42:59.760 --> 00:43:02.559 with the exception of things like nuclear weapons, they didn't 869 00:43:02.579 --> 00:43:06.239 have the likelihood of creating the kind of devastation and 870 00:43:06.309 --> 00:43:10.949 destruction and potential harm that this one does. But we 871 00:43:10.989 --> 00:43:13.809 don't have the time. The greatest illusion is that we 872 00:43:13.849 --> 00:43:15.630 have time. And we do not in this case. 873 00:43:17.010 --> 00:43:18.750 And, you know, the other aspect of this is the 874 00:43:18.769 --> 00:43:20.230 fact that, you know, we look at how do we 875 00:43:20.269 --> 00:43:21.929 get things like trial by jury? How do we get 876 00:43:21.969 --> 00:43:24.619 the Bill of Rights and things like that? It's because 877 00:43:25.159 --> 00:43:25.539 we did it. 878 00:43:25.500 --> 00:43:28.179 By trial and error. We did it by brute force 879 00:43:28.340 --> 00:43:29.000 and mistakes. 880 00:43:29.559 --> 00:43:32.460 And they had those things were there as reactions to 881 00:43:32.519 --> 00:43:34.960 abuses that had happened. That's why I think it is 882 00:43:34.980 --> 00:43:38.030 so concerning because we don't have anybody who is talking about, well, 883 00:43:38.050 --> 00:43:40.469 you know, this could possibly happen. So let's make sure 884 00:43:40.530 --> 00:43:42.989 that doesn't happen in advance. Instead, they're going to let 885 00:43:43.030 --> 00:43:45.099 it happen. And it's going to be really bad. And 886 00:43:45.119 --> 00:43:48.659 then they'll maybe come back and put something in to 887 00:43:48.739 --> 00:43:51.429 stop that from happening again. But it's going to be 888 00:43:51.449 --> 00:43:54.090 a very difficult time, I think, in terms of adjustment 889 00:43:54.650 --> 00:43:56.889 and in terms of the changes that are going to 890 00:43:56.909 --> 00:43:57.929 be made in society. 891 00:43:57.969 --> 00:43:59.030 I absolutely agree with you. 892 00:43:59.449 --> 00:44:02.750 And again, your book, The Mirror, takes a look at 893 00:44:02.909 --> 00:44:05.539 what this really means for us and ask some questions 894 00:44:05.579 --> 00:44:07.920 that we should all be thinking about. I mean, it'd 895 00:44:07.960 --> 00:44:10.619 be much better if we could kind of think through 896 00:44:10.639 --> 00:44:12.449 some of these things before it actually happens to us 897 00:44:12.530 --> 00:44:15.590 instead of trying to fix it after we've had the accident. 898 00:44:16.329 --> 00:44:16.630 Agreed. 899 00:44:16.929 --> 00:44:21.010 Absolutely agreed. I don't think we have the luxury of 900 00:44:21.050 --> 00:44:23.590 fixing it after we've had the accident in this case. 901 00:44:23.610 --> 00:44:26.269 I think we have to be a bit more intentional 902 00:44:26.510 --> 00:44:27.389 than we've been in the past. 903 00:44:27.550 --> 00:44:30.750 Yeah, it is truly amazing and staggering when you look 904 00:44:30.789 --> 00:44:34.449 at the magnitude and the scope of how many things 905 00:44:34.489 --> 00:44:37.829 are affected by this. It really is overwhelming, I think, 906 00:44:37.869 --> 00:44:41.800 the more you look at it. the bigger, taller order 907 00:44:41.820 --> 00:44:44.320 it seems to be. But thank you so much again. 908 00:44:44.599 --> 00:44:48.969 We're talking to Tom Kolopoulos, and the book is The Mirror. 909 00:44:49.329 --> 00:44:52.769 And you can go to his website there, delphigroup.com. Do 910 00:44:52.809 --> 00:44:53.989 you have a link to the book? 911 00:44:54.349 --> 00:44:54.949 I do. 912 00:44:55.730 --> 00:44:57.349 There's a link on the website, and they can also 913 00:44:57.389 --> 00:45:00.699 go to themirror.info to find out more about the book, David. 914 00:45:00.900 --> 00:45:02.179 Thank you so much for having me on the show. 915 00:45:02.239 --> 00:45:03.159 I really appreciate it. 916 00:45:03.159 --> 00:45:05.000 It's great to have you on. I really enjoyed the conversation. 917 00:45:05.039 --> 00:45:06.289 Thank you so much, Tom. Appreciate it. 918 00:45:06.360 --> 00:45:06.679 Thank you. 919 00:45:08.789 --> 00:45:10.150 Well, that was great. Thank you, Tom. We're going to 920 00:45:10.250 --> 00:45:11.050 air that tomorrow. 921 00:45:11.070 --> 00:45:11.769 Perfect. 922 00:45:11.849 --> 00:45:16.269 And we'll send you a link to the. 923 00:45:23.579 --> 00:45:29.739 The Common Man. They created Common Core to dumb down 924 00:45:29.780 --> 00:45:33.500 our children. They created Common Past to track and control us. 925 00:45:33.920 --> 00:45:37.800 Their Commons Project to make sure the commoners own nothing. 926 00:45:38.820 --> 00:45:45.440 and the communist future. They see the common man as simple, unsophisticated, ordinary. 927 00:45:46.460 --> 00:45:49.739 But each of us has worth and dignity created in 928 00:45:49.780 --> 00:45:54.460 the image of God. That is what we have in common. 929 00:45:55.079 --> 00:45:57.909 That is what they want to take away. Their most 930 00:45:57.969 --> 00:46:03.449 powerful weapons are isolation, deception, intimidation. They desire to know 931 00:46:03.670 --> 00:46:08.380 everything about us while they hide everything from us. It's 932 00:46:08.420 --> 00:46:11.420 time to turn that around and expose what they want 933 00:46:11.460 --> 00:46:15.320 to hide. Please share the information and links you'll find 934 00:46:15.699 --> 00:46:19.929 at TheDavidKnightShow.com. Thank you for listening. Thank you for sharing. 935 00:46:25.789 --> 00:46:28.170 If you can't support us financially, please keep us in 936 00:46:28.190 --> 00:46:30.949 your prayers. TheDavidKnightShow.com