WEBVTT 00:01.740 --> 00:05.040 [speaker_0] Hello, my name is Samuel. I graduated from MIT in 00:05.160 --> 00:08.900 2023. I did ML Safety Scholars under Dr. Dan Hendrycks at UC Berkeley. 00:08.940 --> 00:12.060 It's a full-time course. I got convinced by some of these 00:12.100 --> 00:15.940 why the superintelligence might be coming soon, why it could be dangerous, and I 00:15.980 --> 00:17.700 realized actually my life is in danger. 00:17.740 --> 00:19.480 [speaker_1] Yes, I think there's an actual chance I might 00:19.960 --> 00:23.840 [speaker_2] So your motive why you are doing so, I know because, uh, when we 00:23.980 --> 00:27.460 reach a certain level, the AI will, uh, going to impact us. 00:27.860 --> 00:28.240 [speaker_1] Something. 00:28.560 --> 00:28.610 [speaker_0] Yeah. 00:28.680 --> 00:30.360 [speaker_1] Okay? It will fail. 00:31.500 --> 00:34.120 No matter what you do, you're wasting your time. 00:35.780 --> 00:39.580 [speaker_0] Uh, yeah, I think in the next five to 10 years, 00:39.600 --> 00:41.220 will become better than humans at everything. 00:42.220 --> 00:46.120 I'm against the idea of like, uh, like I'm pro-technology, pro-capitalism, 00:46.160 --> 00:47.680 but I'm specifically against superintelligence. 00:47.750 --> 00:50.089 An AI which is so much better than humans is do everything. 00:50.900 --> 00:54.620 And then you can run the entire business, politics, research. 00:54.660 --> 00:57.760 There's no thing you can run. When you're in that sort of scenario, 00:57.800 --> 01:01.140 basically hundred percent unemployment. It'll be like everybody will be useless. 01:01.580 --> 01:05.259 [speaker_2] See, brother, you see the human race take over. 01:05.300 --> 01:05.440 [speaker_0] Yeah. 01:05.880 --> 01:09.310 [speaker_2] We, human race, you know, we work without technology 01:09.560 --> 01:10.140 also. 01:10.180 --> 01:10.630 [speaker_0] Yeah. 01:10.640 --> 01:12.000 [speaker_2] Without science also. 01:12.030 --> 01:12.030 [speaker_0] Yeah. 01:12.040 --> 01:15.730 [speaker_2] Without, we work. Science and technology and 01:15.780 --> 01:16.260 machinery. 01:17.160 --> 01:17.320 [speaker_0] Yeah. 01:17.330 --> 01:18.420 [speaker_2] We are surviving. 01:18.460 --> 01:18.660 [speaker_0] Yeah. 01:18.740 --> 01:19.360 [speaker_2] There is a way. 01:20.560 --> 01:23.450 [speaker_1] No, instead of companies, can we 01:24.000 --> 01:27.920 have some kind of system where we democratize. 01:28.240 --> 01:32.080 [speaker_0] Hey, I'm completely in support with you on, 01:32.150 --> 01:35.280 had an institution that could democratically control what to 01:35.300 --> 01:35.840 superintelligence- 01:35.860 --> 01:35.990 [speaker_1] Yeah 01:36.040 --> 01:40.030 [speaker_0] ... I would be in support of that. Uh, 01:40.180 --> 01:41.640 I haven't built it, you haven't built it. 01:41.740 --> 01:44.400 If nobody builds it in the next five to 10 years, I'm saying this thing can be 01:44.420 --> 01:44.780 happening. 01:45.200 --> 01:48.720 [speaker_3] I think this is, so I believe like it's way ahead of a 01:48.900 --> 01:51.180 Like, I don't think it's coming any year to us. 01:51.580 --> 01:55.470 It, it being exponential because, uh, priorly what the, 01:55.780 --> 01:59.720 the major thing which was solved, all the attention mechanism we got earlier 01:59.820 --> 02:02.300 also, we have got the second going to I believe. 02:03.620 --> 02:07.380 [speaker_0] Like I, like there are some AI researchers in the AI research 02:07.440 --> 02:10.420 think we can build an AI that's better than humans in everything. 02:10.740 --> 02:13.140 Like we might be able to build this in like next five years, 10, you know. 02:13.320 --> 02:15.660 [speaker_2] Thinking probably, yeah. I'm 02:15.700 --> 02:19.660 afraid of this. I'm afraid of this kind of 02:19.700 --> 02:21.100 scenario. 02:21.109 --> 02:21.130 [speaker_0] Yes. 02:21.620 --> 02:22.520 [speaker_2] Then what we will do? 02:23.400 --> 02:27.380 [speaker_0] Right. Now I'm basically, uh, trying to raise awareness to put, like, 02:27.420 --> 02:30.630 so that the public puts pressure on the US government, on the Chinese government. 02:31.240 --> 02:33.410 [speaker_1] So the AI for, for 02:35.180 --> 02:36.440 peace. For peace. 02:36.540 --> 02:36.640 [speaker_0] Yeah. Yeah. 02:37.680 --> 02:39.140 That could be the winning something. 02:39.280 --> 02:41.800 [speaker_1] Are you making this just to become famous? 02:42.060 --> 02:45.280 [speaker_0] Uh, no. I actually think my life is in danger. 02:45.340 --> 02:47.430 I could have had, you know, career in, you know, software research. 02:47.500 --> 02:48.920 I actually think my life is in danger, that's all. 02:48.940 --> 02:52.880 [speaker_1] The actual, uh, good life kind of things 02:52.940 --> 02:56.800 not reached the Earth. Main thing is whatsoever is virtual 02:56.840 --> 02:57.609 we must implement. 02:57.960 --> 02:58.200 [speaker_0] I 02:58.520 --> 03:03.230 don't 03:03.280 --> 03:06.060 want to discourage anyone working on it, but also when I looked at it, what I 03:06.080 --> 03:08.300 realized was like this is not getting solved in five years. 03:08.360 --> 03:10.020 Like this is going to take more time than that. 03:10.040 --> 03:10.120 Like 03:10.980 --> 03:13.460 unless, you know, some genius comes in here, 03:13.500 --> 03:16.470 like this will take time and you know, if we are actually on track, you know, 03:16.540 --> 03:19.010 superintelligence in five years, I feel like this is not getting solved in 03:19.020 --> 03:19.390 years, and... 03:19.560 --> 03:23.540 [speaker_1] But I get where are you coming. 03:23.620 --> 03:27.380 to like compete with this frontier. They are claiming that it is their own. 03:28.300 --> 03:29.500 Doesn't make any sense to me. 03:29.900 --> 03:31.580 [speaker_2] No, no, no. System don't know anything. 03:31.980 --> 03:35.800 System only know what the algorithms you are going to design based 03:35.860 --> 03:37.600 on those data points it's going to- 03:37.880 --> 03:39.680 [speaker_0] I think this is where actually we're disagreeing. 03:39.740 --> 03:42.780 Like I'm saying even right now we are feeding the models with so much junk 03:42.820 --> 03:43.320 much false data. 03:43.440 --> 03:44.360 [speaker_2] Has to, of course. 03:44.400 --> 03:47.560 [speaker_0] Like it's fine on that, like without us telling it, and we 03:47.600 --> 03:49.019 data also, false data also both. 03:49.060 --> 03:49.400 [speaker_2] True. 03:49.520 --> 03:51.520 [speaker_0] Because internet has all sorts of correct data also. 03:51.660 --> 03:53.660 But right now still when you ask you putting- 03:56.120 --> 03:58.660 [speaker_2] Whatever data has been there 03:59.460 --> 04:02.250 for, okay, thirty or forty years, just reuse the data and call up the 04:02.660 --> 04:04.440 algorithm. 04:04.960 --> 04:07.280 [speaker_1] I feel out of place as well, so yeah. 04:07.340 --> 04:09.120 [speaker_0] You must have then considered also leaving India. 04:09.740 --> 04:12.560 [speaker_1] Mm, I am planning to, but not as of now. 04:13.320 --> 04:13.640 [speaker_0] Okay. 04:13.920 --> 04:17.720 [speaker_1] Like not trying to like leave the field un-until unless I have like 04:17.760 --> 04:20.380 very concrete facts, but yeah. 04:20.399 --> 04:23.200 [speaker_2] We're building our core infrastructure. Government is talking. 04:23.260 --> 04:26.180 I don't know what is happening. Maybe we are doing very good. 04:26.240 --> 04:28.610 I'm happy to hear, uh, that people are building. 04:28.920 --> 04:32.670 But maybe we have lost one software industry, uh, uh, 04:33.160 --> 04:36.850 software industry. In the same way, we are becoming service industry, but at this 04:36.920 --> 04:39.080 time we should infrastructure should be created. 04:39.440 --> 04:43.410 So point is that, uh, uh, this is good for our 04:43.440 --> 04:46.460 country, but you have got an opportunity to be a 04:46.840 --> 04:50.740 And being in ecosystem, you have to really be top security 04:50.800 --> 04:51.390 person. 04:51.740 --> 04:55.640 [speaker_0] Search for, you know, a superintelligence. But I'm, again, 04:55.700 --> 04:58.320 I feel like this is a hard problem. It's not getting cracked in five years. 04:58.480 --> 05:01.920 And if we get, you know, superintelligence like this soon, 05:01.960 --> 05:04.860 five years, I feel like we're not prepared. You know, 05:04.900 --> 05:08.480 That's my first point. My second point is like, okay, even if we 05:08.500 --> 05:12.140 successfully keep a superintelligence under our control, like under our control 05:12.200 --> 05:14.930 means whose control? It'll be, you know, the people running, you know, the, the 05:15.000 --> 05:17.560 CEOs of, you know, OpenAI, DeepMind, Anthropic, you know, 05:17.600 --> 05:20.100 the race. It'll be, you know, the US president, you know, 05:20.180 --> 05:22.680 Like basically the people who are actually at the top, 05:22.700 --> 05:24.680 will run the entire society how they want, and- 05:25.400 --> 05:27.800 [speaker_1] It would may not injure a human being. 05:27.840 --> 05:29.550 [speaker_0] Yeah. This Asimov's laws, right? 05:29.550 --> 05:29.820 [speaker_1] Yeah. Yeah. 05:29.900 --> 05:31.010 [speaker_0] Yeah, yeah. I've read about it. Yeah. 05:31.040 --> 05:34.250 [speaker_1] Okay. Sorry, what is that? All I heard, what is that name? 05:34.880 --> 05:34.890 [speaker_0] Yeah. 05:34.900 --> 05:35.640 [speaker_1] Asimov's laws. 05:35.660 --> 05:36.200 [speaker_0] Asimov. Okay. 05:36.260 --> 05:37.920 [speaker_1] As- Okay. These are the laws. 05:38.020 --> 05:38.740 [speaker_0] Asimov. Okay. 05:38.750 --> 05:41.890 [speaker_1] We cannot even create any AI model without having these laws. 05:42.390 --> 05:45.100 AI can never like give us a problem. 05:45.160 --> 05:45.340 [speaker_0] Yes. 05:45.400 --> 05:49.130 [speaker_1] Until, unless it's, you know, some robbing jobs of the tunga people, 05:49.140 --> 05:49.500 people, you know. 05:49.580 --> 05:49.740 [speaker_0] Yeah. 05:50.340 --> 05:54.130 [speaker_1] At that time, the movie was created in the '90s that, uh, 05:54.140 --> 05:54.280 not. 05:54.880 --> 05:55.000 [speaker_0] Yeah. 05:55.260 --> 05:58.659 [speaker_1] Now you tell me buses are more viable or the palgadi 05:58.790 --> 05:58.960 [speaker_0] Yeah. 05:59.490 --> 06:00.420 [speaker_1] What would you, what do you think? 06:00.640 --> 06:03.440 [speaker_0] Yeah, like I'm pro-technology. I'm even pro, I am specifically against 06:03.460 --> 06:06.039 superintelligence. I think superintelligence 06:06.100 --> 06:06.560 technology. 06:06.880 --> 06:08.090 [speaker_1] Solving the problem 06:08.764 --> 06:11.804 [speaker_4] Uh, because Indian has one particular problem- 06:11.864 --> 06:11.874 [speaker_0] Okay 06:11.874 --> 06:12.524 [speaker_4] ... again. 06:12.544 --> 06:12.624 [speaker_0] Mm. 06:12.754 --> 06:14.624 [speaker_4] Now you will see that problem in everything. 06:14.884 --> 06:15.084 [speaker_0] Okay. 06:15.284 --> 06:19.184 [speaker_4] This is it, uh, our makeup of how we think, and 06:19.244 --> 06:21.084 spirituality or, uh, education. 06:21.424 --> 06:21.454 [speaker_0] Mm. 06:21.494 --> 06:23.724 [speaker_4] I do not see much of education, but- 06:23.944 --> 06:24.464 [speaker_0] Like, uh, like- 06:24.964 --> 06:24.964 [speaker_4] ... 06:24.984 --> 06:27.564 [speaker_0] No, no, no. There are too many points I want to first, 06:27.604 --> 06:27.854 [speaker_4] Okay, you want to- 06:27.854 --> 06:30.884 [speaker_0] Yeah, yeah, yeah, yeah. Yeah, you were saying, you know, yeah, 06:30.904 --> 06:33.844 of robotics, like, uh, there is this field of alignment 06:33.924 --> 06:36.814 Even people at OpenAI, Anthropic, there are a few people at these companies 06:36.904 --> 06:40.264 to, you know, ensure their, you know, AI follows, you know, certain laws. 06:40.324 --> 06:41.804 You can use those laws as an example. 06:41.864 --> 06:44.384 I'm saying even they are, like right now failing, like this 06:44.424 --> 06:45.444 They're not actually able to do it. 06:45.664 --> 06:47.564 [speaker_4] Anytime a new technology is being created- 06:47.644 --> 06:47.654 [speaker_0] Yes 06:47.654 --> 06:51.644 [speaker_4] ... there are a lot of dummy, uh, like even when iPhone 06:51.664 --> 06:52.244 phones failed. 06:52.524 --> 06:52.764 [speaker_0] Yes. 06:53.304 --> 06:54.504 [speaker_4] N- not because they were, they were wrong. 06:54.784 --> 06:54.844 [speaker_0] Yeah. 06:54.884 --> 06:58.684 [speaker_4] Technology. Just look how the technology is being, be- 06:58.764 --> 07:02.064 like made, how, how they've been creating, and how you can 07:02.144 --> 07:04.864 actually help that technology being created in a safer way. 07:05.244 --> 07:08.124 Not every, every perspective of, of profession. 07:08.504 --> 07:08.624 [speaker_0] Yeah. 07:08.654 --> 07:09.164 [speaker_4] And from design- 07:09.294 --> 07:09.294 [speaker_0] Okay 07:09.384 --> 07:10.174 [speaker_4] ... and from marketing. 07:10.464 --> 07:10.704 [speaker_0] Yeah. 07:10.844 --> 07:12.844 [speaker_4] I know what it, how to write and all. 07:17.164 --> 07:20.764 Right. You can always put the plug inside. Runs on electricity, right? 07:21.064 --> 07:23.863 [speaker_0] Okay, yeah. So there are multiple problems with that proposal. 07:24.084 --> 07:24.094 Uh- 07:24.144 --> 07:27.504 [speaker_4] It, it, it has, it, it runs on certain things and 07:27.724 --> 07:31.664 always autonomy is given by humans to the extent. 07:31.804 --> 07:31.864 [speaker_0] Yeah. 07:31.904 --> 07:35.344 [speaker_4] At the end of the day, if it... I'm not saying complete autonomy, 07:35.424 --> 07:38.764 autonomy is, is a, is the basis of what AI has to have. 07:39.404 --> 07:42.943 No AI can have, uh, complete autonomy without guardrails of 07:43.024 --> 07:43.654 universe also. 07:44.544 --> 07:47.644 Yeah, I'm only saying we have to trust the 07:48.164 --> 07:51.944 When human beings have done damages to, to the, to the 07:52.024 --> 07:55.984 world to a degree as well. Not to say that, okay, 07:56.064 --> 07:59.564 against of course. Consider of saying, uh, 07:59.964 --> 08:03.604 for causing research you could say start research 08:04.104 --> 08:07.624 for guardrails, okay? Uh, towards, uh, 08:07.684 --> 08:08.844 artificial superintelligence. 08:09.204 --> 08:09.744 [speaker_0] I, I understood what you're saying. 08:09.804 --> 08:11.884 I appreciate you put this effort into thinking about it. 08:11.904 --> 08:15.704 But yeah, I mean, coming back to pausing, I feel, yeah, I mean, unless, 08:15.764 --> 08:18.934 like public actually puts pressure on governments, on companies to do something 08:19.004 --> 08:19.393 about it, but- 08:19.484 --> 08:21.744 [speaker_4] No, but I think you need to self-evolve. 08:21.804 --> 08:24.864 That the point is that only self-actualized individuals can 08:25.424 --> 08:28.673 I am not fully self-actualized. I'm on my way to self-actualization. 08:29.624 --> 08:29.684 [speaker_0] Yeah. 08:29.764 --> 08:32.544 [speaker_4] I have more coherence in terms of others. 08:33.024 --> 08:36.764 So I'm saying that if you have coherence, you can build anything in the 08:36.804 --> 08:39.263 world with guardrails. That's the whole point. 08:40.264 --> 08:43.444 [speaker_0] Yeah. Uh, you're saying, you're saying if you like, 08:43.484 --> 08:46.764 you know, full automated autonomy, it can run, you know, everything, uh, 08:46.824 --> 08:49.324 we will pull the plug. I'm saying like- 08:49.484 --> 08:50.364 [speaker_4] Structured autonomy. 08:50.394 --> 08:50.854 [speaker_0] ... first we can appreciate- 08:50.854 --> 08:54.784 [speaker_4] Structured... Even right now, the AI that is there, it is not without 08:54.864 --> 08:56.764 structured autonomy. Don't think that it has- 08:57.104 --> 08:57.714 [speaker_0] Sorry. Yeah, please answer first. 08:57.724 --> 08:57.864 [speaker_4] Huh? 08:58.184 --> 08:59.104 [speaker_0] Do you, I'd react to what you said. 08:59.164 --> 09:00.934 [speaker_4] Yeah, yeah, yeah. Yeah. Okay. Okay. 09:01.044 --> 09:04.264 [speaker_0] Uh, yeah, I mean, first problem with that is it 09:04.284 --> 09:08.184 between US and China. Even if a country or a company knows 09:08.204 --> 09:10.964 that, you know, their model could cause dangers, 09:11.004 --> 09:12.814 with it. That is, like, the first problem. 09:12.924 --> 09:16.124 The second problem is they might not know in advance, like 09:16.134 --> 09:18.524 sufficiently intelligent and be aware this is happening. 09:18.544 --> 09:22.524 Third problem, if the takeoff is very fast, like if the jump from, 09:22.544 --> 09:25.894 human to like way beyond human, if the jump happens too quickly, again, 09:25.964 --> 09:27.084 see it coming. 09:27.864 --> 09:29.964 [speaker_4] Okay, your blind spot and bottleneck question. 09:30.004 --> 09:32.084 [speaker_5] The initiative is good, man, but like I have just one question. 09:32.164 --> 09:34.304 Seeing AI to AI conversation that's happening- 09:35.144 --> 09:35.804 [speaker_4] So let me- 09:35.854 --> 09:35.854 [speaker_5] Yeah 09:35.854 --> 09:36.724 [speaker_4] ... let me, let me just- 09:36.824 --> 09:39.124 [speaker_5] Instead of pressurizing the government that has no rights. 09:39.604 --> 09:40.444 India has no, 09:41.284 --> 09:42.594 no, how do you say, no say 09:43.464 --> 09:45.444 compared to the US and compared to China. 09:45.564 --> 09:49.453 So I don't think you can achieve anything, uh, tangible, but, uh, 09:49.924 --> 09:51.424 it's a good initiative. I appreciate it. 09:52.064 --> 09:52.184 [speaker_4] Yeah. 09:52.224 --> 09:53.484 [speaker_5] Many people don't know about this stuff. 09:53.864 --> 09:55.064 [speaker_0] Yeah. That is true. 09:55.224 --> 09:59.144 [speaker_4] My only point is that, you know, if you're just going to say 09:59.204 --> 10:01.864 be a linear thing, life is non-linear as well. 10:01.944 --> 10:04.364 You never know what surprises sprung up. 10:04.764 --> 10:08.504 So instead of trying to control AI like a 10:08.524 --> 10:12.464 dictator who does that and respect, you know, myopia, dystopia or 10:12.544 --> 10:16.424 utopia, build frameworks and structures and systems around that, 10:16.864 --> 10:19.984 that will implement and, you know, it will give it direction. 10:20.344 --> 10:24.024 It needs struct- systems, structure, clarity and direction. 10:24.124 --> 10:27.744 Powerful direction. Your internet, a lot of individuals 10:28.064 --> 10:31.584 did say that, you know, there would be this catastrophe, 10:31.974 --> 10:35.164 Now the big media houses are controlled by the big vested 10:35.644 --> 10:37.804 This is democratizing every individual. 10:38.264 --> 10:41.804 You know, you could basically post this using AI 10:42.184 --> 10:44.744 but instead of a negative or a pause, 10:46.124 --> 10:49.764 look at the positive impact of it and take this 10:50.084 --> 10:54.064 as, as real, real time systemic feedback. Okay? 10:54.424 --> 10:54.764 You need- 10:54.884 --> 10:56.814 [speaker_0] So think I'm agreeing on you on some part, 10:56.884 --> 10:57.153 part. I think that's- 10:57.184 --> 10:58.964 [speaker_4] But you ... I'm here to put up my point. 10:59.104 --> 10:59.314 [speaker_0] Yeah, yeah. 10:59.324 --> 11:00.444 [speaker_4] I have to go. 11:00.584 --> 11:00.594 [speaker_0] Sorry. 11:00.624 --> 11:04.224 [speaker_6] I partially agree with you because I believe the answer will be lie 11:04.264 --> 11:08.004 between. Suppose there is a community where we 11:08.284 --> 11:09.564 It is very fine, very much fine. 11:09.944 --> 11:10.264 [speaker_4] Something. 11:10.694 --> 11:10.694 [speaker_0] Yeah. 11:10.764 --> 11:12.904 [speaker_4] Okay? It will fail. 11:12.924 --> 11:13.494 [speaker_6] Because I believe- 11:13.544 --> 11:17.224 [speaker_4] No matter what you do, you're wasting your time by just by 11:17.244 --> 11:21.184 here instead of doing something constructive, productive 11:21.784 --> 11:22.904 a clarity mission. 11:22.984 --> 11:26.184 [speaker_0] An answer, I mean, I am supportive of people working on 11:26.224 --> 11:29.964 research for a superintelligence to be built. However, I think that 11:30.004 --> 11:32.584 [speaker_4] But you can't basically control, uh, US and China. 11:32.724 --> 11:34.984 You are basically expecting a utopia here. 11:35.024 --> 11:35.314 [speaker_0] Yeah. I- 11:35.314 --> 11:36.374 [speaker_4] And it's not going to work 11:36.404 --> 11:39.764 [speaker_0] ... I personally can't do this. Only 11:39.824 --> 11:40.504 public is convinced- 11:40.624 --> 11:43.844 [speaker_4] No, but, uh, see everybody's going to AI India Impact 11:43.874 --> 11:43.874 [speaker_0] Yes. 11:43.944 --> 11:45.804 [speaker_4] I don't have time for this. This is not... 11:46.324 --> 11:50.264 You are basically trying to con- contain something 11:50.284 --> 11:54.224 has a huge capacity. It's beyond the realistic 11:54.444 --> 11:58.344 idea. There's you, if you really want, you invent something that will 11:58.424 --> 12:00.124 ensure it in proper direction. 12:00.204 --> 12:00.614 [speaker_0] Okay. I, I- 12:00.644 --> 12:00.784 [speaker_4] You- 12:01.624 --> 12:03.624 [speaker_0] I think we disagree, but I understood your point of view. Yeah. 12:04.264 --> 12:06.764 [speaker_4] It's not whether you disagree or I'm not here to convince you. 12:06.824 --> 12:08.724 I'm here to ensure that you don't waste your time. 12:08.764 --> 12:08.924 [speaker_0] Okay. 12:09.044 --> 12:10.044 [speaker_4] And nor, nor do others. 12:10.764 --> 12:10.924 [speaker_0] No. 12:11.344 --> 12:14.944 [speaker_4] First thing is, the more you stand up against something 12:14.964 --> 12:17.404 reason, a proper cause, you are not standing up for a real cause. 12:17.424 --> 12:17.934 [speaker_0] Yeah. I think it- 12:18.224 --> 12:19.724 [speaker_4] I think this is not the right cause 12:19.734 --> 12:20.974 [speaker_0] ... you should explain, explain to my reasons why- 12:20.974 --> 12:24.804 [speaker_4] No, no, I don't want to argue. You are basically, you are a- you 12:24.844 --> 12:27.664 imposing your dictatorship on something that you don't, you don't have control 12:27.784 --> 12:29.734 over. You want to contain it. And that's- 12:29.734 --> 12:31.934 [speaker_0] I said the public should contain it, 12:31.944 --> 12:33.824 [speaker_4] No, no. Whether you or not the public 12:33.864 --> 12:37.314 [speaker_6] I think we need to make sure superintelligence is in 12:37.744 --> 12:38.274 correct hands. 12:38.324 --> 12:41.204 [speaker_4] Su- superintelligence is different than a general intelligence. 12:41.244 --> 12:44.884 [speaker_6] Whenever it comes, it should be in the correct hands rather 12:45.824 --> 12:46.694 [speaker_4] I think you're, you're- 12:46.804 --> 12:47.104 [speaker_0] So, so yeah, so the, uh- 12:47.524 --> 12:50.884 [speaker_4] This is dystopia and utopia and myopia.