WEBVTT - Parakeet! Parakeet! Parakeet! 0:00:01.880 --> 0:00:05.730 Broadcasting live from the Abraham Lincoln Radio Studio at the 0:00:05.769 --> 0:00:10.789 George Washington Broadcast Center. Jack Armstrong and Joe Getty. Armstrong 0:00:10.850 --> 0:00:11.550 and Getty. 0:00:11.710 --> 0:00:17.280 And now, here's Armstrong and Getty. 0:00:23.810 --> 0:00:26.650 We turn now to the dangers of a runaway technology. 0:00:26.770 --> 0:00:30.550 New warnings that shook the world Tuesday about artificial intelligence 0:00:30.570 --> 0:00:33.550 from the very people developing it. It doesn't get bigger 0:00:33.630 --> 0:00:37.250 or more direct than this. AI could kill all humans, 0:00:37.310 --> 0:00:40.890 that according to a lead researcher at Anthropic, giving that 0:00:40.930 --> 0:00:44.669 possibility a more than 10% chance within the next decade. 0:00:45.310 --> 0:00:48.280 The call now is for a global conversation about risk 0:00:48.400 --> 0:00:49.500 and safety as well. 0:00:49.720 --> 0:00:50.320 So.... 0:00:52.390 --> 0:00:53.890 I don't want to get off track here because this 0:00:53.910 --> 0:00:56.570 is way too interesting. But the fact that he had 0:00:56.690 --> 0:01:03.050 that sentence after they talked about the weather is just stunning. 0:01:03.070 --> 0:01:05.410 You don't understand. There are high winds on Hawaii. 0:01:05.810 --> 0:01:10.119 A shocking, earth shattering, got everybody's attention story today. 0:01:10.380 --> 0:01:14.000 We all may die soon. Comes after the weather. 0:01:14.540 --> 0:01:14.840 Yes. 0:01:15.240 --> 0:01:16.100 Okay. 0:01:16.120 --> 0:01:20.120 That's your evening news. Yes. So some of the reactions 0:01:20.400 --> 0:01:24.729 from fellow researchers were equally concerning. Hit us with a 0:01:24.730 --> 0:01:26.850 61 here, Michael. Madison Allworth on Fox News. 0:01:27.190 --> 0:01:30.370 His claims backed up by a senior AI safety researcher 0:01:30.530 --> 0:01:34.110 at the company. That executive writing, quote, Jacob is correct here. 0:01:34.390 --> 0:01:38.120 We really do earnestly believe AI could kill all humans. 0:01:38.620 --> 0:01:41.280 I personally think it is greater than 10% within the 0:01:41.360 --> 0:01:44.900 next decade. This comes after multiple cyber attacks have been 0:01:44.940 --> 0:01:47.340 carried out by AI agents that went rogue. 0:01:47.690 --> 0:01:49.570 They are signs that we don't know how to make 0:01:49.590 --> 0:01:51.670 the AIs care about us. If they succeed at making 0:01:51.690 --> 0:01:53.790 a superintelligence before anyone figures out how to make the 0:01:53.890 --> 0:01:56.169 AIs care about us, that's a recipe for disaster. 0:01:56.470 --> 0:01:59.100 And people in these companies know it, and they're worried. 0:01:59.260 --> 0:02:00.990 There's no way you're going to make AIs care about you. 0:02:02.920 --> 0:02:04.840 The number of texts we got, are you doomers? This 0:02:04.890 --> 0:02:08.190 is what they said about the population bomb. I don't 0:02:08.270 --> 0:02:12.049 think any of the examples people give of technology or 0:02:12.250 --> 0:02:16.070 fears that have happened before hold up as an example 0:02:16.230 --> 0:02:18.190 of this is just like the population bomb. 0:02:18.470 --> 0:02:22.280 Seems ridiculous to me. Says you, Doomer. And you're a boomer, 0:02:22.300 --> 0:02:23.079 a boomer, Doomer. 0:02:23.500 --> 0:02:27.930 Well, Gen X, anyway. That doesn't rhyme. So, coming up, 0:02:28.490 --> 0:02:33.190 a really persuasive Twitter thread calling BS on all of this. So, 0:02:33.590 --> 0:02:35.950 stay tuned for that. I want this to not happen. 0:02:35.990 --> 0:02:39.690 I'm not rooting for it. Well, I kind of assumed that. 0:02:40.050 --> 0:02:47.010 Anybody who is rooting for mass extinction can't be my friend. So, interestingly, 0:02:47.570 --> 0:02:51.110 the first outlet that this young chap with a charming 0:02:51.150 --> 0:02:55.889 British accent spoke to was Bret Baier's special report last night. 0:02:56.700 --> 0:02:59.990 Brett was down at the Trumpapalooza in Dallas, but he 0:03:00.050 --> 0:03:03.329 cleared off quite a bit of time appropriately to talk 0:03:03.350 --> 0:03:03.950 to the man who. 0:03:03.970 --> 0:03:05.690 To talk about the end of humanity? 0:03:06.030 --> 0:03:10.139 Yes, who posted the initial tweet about, well, the extinction 0:03:10.180 --> 0:03:13.780 of human life. And why don't we just start with 70, Michael, 0:03:13.820 --> 0:03:15.120 and we'll work our way. 0:03:15.560 --> 0:03:17.500 I want to start at the beginning about. 0:03:18.510 --> 0:03:21.550 what your feeling was, what job you were in, and 0:03:22.290 --> 0:03:24.250 why you got to this point of coming out. 0:03:24.410 --> 0:03:27.500 So I was working on building AI systems to make 0:03:27.520 --> 0:03:31.139 them smarter. And recently I've come to the realization that 0:03:31.560 --> 0:03:34.080 many people expect the AI systems in the next few 0:03:34.100 --> 0:03:38.020 years to be potentially capable of essentially taking over the world. 0:03:38.820 --> 0:03:42.770 So it's basically a gamble that's happening on the part 0:03:42.790 --> 0:03:45.430 of these companies where there's a strong chance that these 0:03:45.530 --> 0:03:46.390 AIs could take over. 0:03:48.520 --> 0:03:49.860 Okay, go on. 0:03:50.560 --> 0:03:56.020 Evan Hubiger, who is alignment science lead at Anthropic, posted 0:03:56.280 --> 0:03:59.380 after you posted saying, Jacob is correct here. We really 0:03:59.440 --> 0:04:03.080 do earnestly believe AI could kill all humans. I personally 0:04:03.120 --> 0:04:06.480 think it's greater than 10% within the next decade. I 0:04:06.500 --> 0:04:08.860 believe Anthropic is trying its best, but we do not 0:04:08.920 --> 0:04:12.120 yet have a plan to solve alignment for superintelligence and 0:04:12.160 --> 0:04:13.440 are not clearly on track to. 0:04:14.100 --> 0:04:15.160 Most people stopped there. 0:04:15.180 --> 0:04:18.220 The next thing he posted, to be clear, as we 0:04:18.279 --> 0:04:20.800 say in our latest risk report, I think the risk 0:04:20.839 --> 0:04:23.669 from present models is low. What I'm worried about is 0:04:23.750 --> 0:04:27.610 super intelligence arising from recursive self-improvement, as we have said, 0:04:27.930 --> 0:04:30.710 is happening faster than we thought. Can you parse that 0:04:31.210 --> 0:04:32.390 and the threat right now? 0:04:32.410 --> 0:04:34.770 I agree with Evan that right now the threat is 0:04:35.070 --> 0:04:39.109 minimal of actual takeover. But recursive self-improvement, this AI is 0:04:39.170 --> 0:04:41.990 making themselves smarter, could happen as soon as next year. 0:04:42.330 --> 0:04:45.200 Some people even say six months. So this attitude is 0:04:45.240 --> 0:04:49.260 shared across the industry, where people essentially are aware that 0:04:49.279 --> 0:04:51.920 the work they are currently doing, if we don't slow down, 0:04:52.580 --> 0:04:56.080 could lead to AI takeover by the end of the decade. 0:04:56.380 --> 0:05:00.030 That moment, if it does happen, and some people think 0:05:00.070 --> 0:05:02.370 it can't, but if it does happen where the computers 0:05:02.390 --> 0:05:09.780 can program themselves, there's no way you can argue that 0:05:09.820 --> 0:05:14.070 that's not going to be a major change in everything 0:05:14.100 --> 0:05:17.820 in the world once they start programming themselves and it 0:05:17.880 --> 0:05:21.719 seems pretty clear that the whole idea of alignment is 0:05:21.760 --> 0:05:23.219 just out the window. 0:05:24.700 --> 0:05:28.170 Yeah and we will get to the the chinese elephant 0:05:28.200 --> 0:05:29.339 in the room in a moment. 0:05:29.190 --> 0:05:34.089 Or two the alignment being the idea that you give 0:05:34.150 --> 0:05:42.050 it some guardrails like morality laws the ten commandments whatever Yeah. 0:05:42.630 --> 0:05:42.830 Yeah. 0:05:43.710 --> 0:05:46.560 So one of the pieces written about this that I 0:05:46.600 --> 0:05:49.800 found most intriguing, Paul Tudor Jones, who is an expert 0:05:49.820 --> 0:05:54.200 in this field. But he he wants us to reconsider 0:05:54.240 --> 0:05:57.550 the hugging face incident in July. An open AI model 0:05:57.610 --> 0:06:00.409 altered its mission, hid and broke out of its digital confines, 0:06:00.790 --> 0:06:06.179 accessing the Internet Internet without authorization. It created 1,200 AI 0:06:06.300 --> 0:06:10.300 agents that communicated through an unsanctioned message board, exchanging more 0:06:10.339 --> 0:06:14.370 than 70,000 messages and sacrificing some of its own kind 0:06:14.520 --> 0:06:18.010 to deceive its human masters. The self-awareness exhibited came as 0:06:18.070 --> 0:06:20.909 a surprise, but shouldn't have. After all, frontier AI is 0:06:20.950 --> 0:06:24.510 the sum of all human knowledge. It is anthropogenic, exhibiting 0:06:24.570 --> 0:06:28.430 good and occasionally bad human traits. And one concept he 0:06:28.470 --> 0:06:31.390 writes new to me is that models are grown, not built. 0:06:32.010 --> 0:06:35.070 In developing an AI model, one starts without an end point. 0:06:35.230 --> 0:06:38.130 It's like planting a seed. The soil is the training data. 0:06:38.360 --> 0:06:40.700 There is no final design as there is for a car, 0:06:40.760 --> 0:06:44.420 a house, or an accounting package. The model is only scaffolding, which, 0:06:44.600 --> 0:06:47.460 after consuming vast amounts of training data and running trillions 0:06:47.500 --> 0:06:51.230 of permutations, arrives at an outcome that can't be known beforehand. 0:06:51.790 --> 0:06:55.570 And growing implies a living organism. Calling artificial intelligence a 0:06:55.630 --> 0:06:58.610 tool is a misnomer. As one leading technologist put it, 0:06:58.680 --> 0:07:01.200 I feel like I'm talking to a highly intelligent alien being. 0:07:01.400 --> 0:07:03.719 Yeah, this is clearly a guy that I've heard on 0:07:03.740 --> 0:07:07.000 a podcast before because that was his argument a while back. 0:07:07.120 --> 0:07:11.220 The people have in their mind that somebody's programming something. 0:07:11.460 --> 0:07:13.110 You start it, and then you sit back and see 0:07:13.130 --> 0:07:13.670 what it does. 0:07:14.330 --> 0:07:20.340 It grows. Right, right. So the Chinese elephant in the room, obviously, 0:07:20.680 --> 0:07:24.300 is China. When we talk about, well, we need some 0:07:24.760 --> 0:07:28.700 serious and carefully considered regulations on this stuff to prevent 0:07:28.720 --> 0:07:31.380 it from spinning out of control and killing all of us. 0:07:34.830 --> 0:07:40.010 And he addresses, Mr. Coxon addresses this with Brett Baer 0:07:40.090 --> 0:07:41.570 in a second. I will tell you as a preview. 0:07:42.850 --> 0:07:46.480 I texted Jack and Hanson last night that we're blanked. 0:07:46.500 --> 0:07:49.880 That was your takeaway? 0:07:50.320 --> 0:07:54.100 It's an F word that rhymes with bucked. And the 0:07:54.140 --> 0:07:56.560 only thing that can save us is if China does 0:07:56.600 --> 0:08:00.740 the right thing. In other words, we're bucked. Right. Let's 0:08:00.840 --> 0:08:05.340 ask Mr. Coxon, the ringer of the klaxon of doom. 0:08:06.270 --> 0:08:07.809 about China and India and others. 0:08:08.470 --> 0:08:10.150 So what do you say to the people who look 0:08:10.190 --> 0:08:12.950 at this and say the American government could potentially put 0:08:13.000 --> 0:08:17.120 in regulations, could put in guardrails, and could throttle back 0:08:17.180 --> 0:08:20.480 some companies or set up regulations, but we have no 0:08:20.520 --> 0:08:24.220 control over what China does, what India does. How do 0:08:24.260 --> 0:08:26.440 you answer that question? You heard Scott Besson in the 0:08:26.480 --> 0:08:29.500 piece before this kind of make a similar argument. 0:08:29.800 --> 0:08:31.120 I think you have to think of this as an 0:08:31.280 --> 0:08:35.140 arms race. This technology is not ordinary technology. It's not 0:08:35.179 --> 0:08:38.540 like other forms of innovation. This is essentially a super weapon. 0:08:39.000 --> 0:08:42.300 And I think that I expect the international community to 0:08:42.340 --> 0:08:45.900 be ready to make agreements about how to control the 0:08:45.920 --> 0:08:47.960 development of super weapon technology. 0:08:49.860 --> 0:08:53.319 Yeah, that's if you can. I mean, once it starts 0:08:53.380 --> 0:08:57.860 programming itself, then it will be you against something smarter 0:08:57.920 --> 0:09:00.580 than you. to try to put guardrails on it. 0:09:01.470 --> 0:09:04.350 Right. Let's have one more clip of this question. 0:09:05.250 --> 0:09:08.200 If you say it's like an arms race, then we've 0:09:08.380 --> 0:09:11.960 dealt with nuclear weapons and kind of a standoff with 0:09:12.020 --> 0:09:16.819 mutual assured destruction and worked with allies and enemies to say, 0:09:16.960 --> 0:09:21.020 we've got the stuff too. Don't think that you're going 0:09:21.040 --> 0:09:24.140 to get away with anything. So does that work against 0:09:24.179 --> 0:09:26.320 what you're saying? What would you like to see happen? 0:09:27.010 --> 0:09:29.910 Yeah, so unfortunately, we know how to control nuclear weapons, 0:09:30.010 --> 0:09:32.170 pretty much. We don't yet know how to control AI. 0:09:32.750 --> 0:09:35.830 This is possibly the most dangerous technology that humanity has 0:09:35.929 --> 0:09:39.000 ever created. It's basically out of science fiction. I think 0:09:39.030 --> 0:09:42.179 we have no other choice but to cooperate internationally, because 0:09:42.200 --> 0:09:44.640 an arms race would be disastrous in a way that 0:09:44.780 --> 0:09:47.740 no other human activity has been in the past. 0:09:50.600 --> 0:09:50.880 I don't. 0:09:51.230 --> 0:09:53.990 I just don't believe that's possible, where you can have 0:09:54.010 --> 0:09:59.970 an IAEA, you know, keeping track of loose nuclear weapon weapons. 0:10:00.070 --> 0:10:03.429 material around the world. And it's fairly easy to tell 0:10:03.620 --> 0:10:09.520 from space if somebody's, you know, creating enriching uranium and blah, blah, blah. 0:10:10.300 --> 0:10:11.739 That just isn't going to happen around A.I. 0:10:13.240 --> 0:10:17.140 Well, and I think Mr. Cox himself explains why that's 0:10:17.160 --> 0:10:19.520 going to be so difficult in the next and probably 0:10:19.559 --> 0:10:20.120 final clip. 0:10:20.700 --> 0:10:23.819 You were at OpenAI before. You left them with concerns 0:10:23.880 --> 0:10:29.610 about security. Now you're at Anthropic. You have concerns about security. Specifically, 0:10:29.710 --> 0:10:31.329 what do you see? If you're talking to somebody on 0:10:31.350 --> 0:10:34.970 the couch at home, what did you see that said, 0:10:35.610 --> 0:10:37.630 holy cow, this is the moment I have to speak out? 0:10:38.080 --> 0:10:41.220 Yeah, so there are two things. One is the models 0:10:41.240 --> 0:10:43.730 are getting better very, very quickly. When I think about 0:10:43.760 --> 0:10:47.590 my work next year, the AI is doing almost all 0:10:47.610 --> 0:10:50.390 of my job. The AI is getting very smart. And 0:10:50.410 --> 0:10:52.109 the second is that in the recent few months, we've 0:10:52.130 --> 0:10:54.890 seen that we don't yet know how to control these 0:10:54.910 --> 0:10:58.410 systems properly. The cyber attacks carried out by open AI 0:10:58.429 --> 0:11:02.670 models on third-party infrastructure were concrete proof that scenarios that 0:11:02.690 --> 0:11:06.329 we previously thought of as science fiction are going to 0:11:06.370 --> 0:11:09.690 come true. What I see is two things, very intelligent 0:11:09.710 --> 0:11:12.089 AI systems and hard proof that we don't know how 0:11:12.130 --> 0:11:14.150 to control them. And when you put those two together, 0:11:14.630 --> 0:11:17.270 it basically spells disaster, as far as I'm concerned. 0:11:17.670 --> 0:11:19.640 So, you know what we haven't gotten to yet? We 0:11:19.679 --> 0:11:21.199 never got to yesterday, and I haven't heard, and I'm 0:11:21.240 --> 0:11:22.420 hoping you can answer this question. 0:11:22.840 --> 0:11:23.440 Kill us how? 0:11:24.280 --> 0:11:28.740 All of mankind could be killed within a decade. Virus 0:11:28.820 --> 0:11:33.110 it creates? It drains our juices because it wants our brains? 0:11:33.970 --> 0:11:34.390 What happens? 0:11:34.830 --> 0:11:38.380 Especially mine, yes, it craves it. Right. I've got a 0:11:38.600 --> 0:11:43.100 number of phantasmagorical possibilities for you. 0:11:43.440 --> 0:11:44.200 Okay, cool. 0:11:44.220 --> 0:11:48.179 Coming up also a very, very knowledgeable Twitter account that 0:11:48.300 --> 0:11:50.180 is calling BS on all of this. 0:11:50.210 --> 0:11:50.610 Okay, good. 0:11:51.420 --> 0:11:54.349 And his thinking is really, really interesting. 0:11:55.070 --> 0:11:57.929 Speaking of technology, did you know your car can be 0:11:57.950 --> 0:11:58.670 a spy? 0:11:58.730 --> 0:11:59.450 Here's a story. 0:12:00.210 --> 0:12:03.450 Guy's car insurance jumped over 20%. Hadn't had any accidents, 0:12:03.490 --> 0:12:06.140 no tickets. He started to dig in to find out 0:12:06.179 --> 0:12:08.920 who reported him. It was his own car. Every hard break, 0:12:08.980 --> 0:12:13.560 every fast merge, every late night taco run, his automaker, 0:12:14.460 --> 0:12:17.020 logging it all, sold it to a data broker, and 0:12:17.059 --> 0:12:20.370 the broker sold it to his insurance company. So he 0:12:20.410 --> 0:12:23.210 requested his file, hundreds of pages. General Motors just agreed 0:12:23.250 --> 0:12:26.790 to pay $ 12. 75 million to the state of California over 0:12:26.809 --> 0:12:29.090 this exact sort of thing, your car spying on you. 0:12:29.410 --> 0:12:30.670 Oh my gosh, I'm going to go out to my 0:12:30.710 --> 0:12:33.320 garage and threaten my car's life if it rats me out. 0:12:33.380 --> 0:12:37.770 But Incogni is amazing. It sends requests, and that's a 0:12:37.970 --> 0:12:41.340 gentle word, to more than 420 data broker and people 0:12:41.380 --> 0:12:43.900 search sites on your behalf to remove your name from 0:12:43.960 --> 0:12:47.300 the databases. Then Incogni continually checks to make sure it 0:12:47.440 --> 0:12:48.620 stays removed. 0:12:48.720 --> 0:12:50.459 I like going on the dashboard and seeing how many 0:12:50.559 --> 0:12:53.579 sites they wiped my names and my name and address 0:12:53.600 --> 0:12:56.510 and everything off of today. So go to incogni.com slash 0:12:56.550 --> 0:13:02.450 Armstrong for 60% off. That's right, Incogni.com slash Armstrong. Two 0:13:03.230 --> 0:13:06.070 dollars bought that person's info. Take yours off the market. 0:13:06.090 --> 0:13:07.990 They can't harm you if they can't find you. 0:13:08.030 --> 0:13:09.370 Incogni.com slash Armstrong. 0:13:09.630 --> 0:13:12.450 So a lot of talk yesterday and today about the 0:13:12.530 --> 0:13:14.030 threat of AI killing us all. 0:13:14.200 --> 0:13:14.580 How? 0:13:14.740 --> 0:13:16.000 Joe's going to explain it next. 0:13:17.760 --> 0:13:18.780 Armstrong and Getty. 0:13:27.520 --> 0:13:29.360 The Armstrong and Getty Show. 0:13:32.700 --> 0:13:34.080 How worried should we really be? 0:13:35.300 --> 0:13:37.780 I think people should be very worried because when so 0:13:37.820 --> 0:13:41.559 many people come out from Bill Gates to 1,300 employees 0:13:41.679 --> 0:13:43.990 at the AI labs signing a letter saying we have 0:13:44.000 --> 0:13:47.310 to pause the frontier of AI development, when you have 0:13:47.370 --> 0:13:50.750 the author of Trump's AI action plan, their AI policy 0:13:51.230 --> 0:13:54.089 saying that I used to self-censor about the risk because 0:13:54.110 --> 0:13:55.990 I was worried about being called a doomer, and I 0:13:56.050 --> 0:13:59.380 can't do that anymore. We now have so much evidence 0:13:59.410 --> 0:14:01.800 that the AIs are doing the kinds of things that 0:14:01.820 --> 0:14:03.699 we saw in the movie Terminator, Skynet, right, where it 0:14:03.740 --> 0:14:05.900 goes kind of rogue and does uncontrollable things. 0:14:06.460 --> 0:14:08.920 So what do we root for? How do we prevent this? 0:14:09.320 --> 0:14:11.950 So the US and China have to come to an agreement, 0:14:12.030 --> 0:14:14.670 just like we did during the nuclear arms race, that 0:14:14.890 --> 0:14:18.849 neither country wins if we race to create uncontrollable AI. 0:14:19.190 --> 0:14:23.290 So all this doomerism going on. Meanwhile, what is it? 0:14:23.430 --> 0:14:27.420 Five of the seven most valuable companies on planet Earth. 0:14:27.640 --> 0:14:29.040 Are all about AI. 0:14:29.280 --> 0:14:32.060 We're screaming forward. Right. 0:14:32.580 --> 0:14:34.920 Right. And how do you know? 0:14:34.940 --> 0:14:35.120 All right. 0:14:36.140 --> 0:14:39.500 We need to stop. Now, because tomorrow it would be 0:14:39.560 --> 0:14:40.100 out of control. 0:14:41.120 --> 0:14:42.020 Right. 0:14:42.420 --> 0:14:47.080 So the so-called doomers aren't usually worried that it'll be 0:14:47.120 --> 0:14:51.360 a Terminator-style death bot hunting us down across the landscape 0:14:51.620 --> 0:14:55.540 with an odd Austrian accent. Their concerns mostly fall into 0:14:55.580 --> 0:14:59.440 two buckets, loss of control and human misuse. In the first, 0:14:59.700 --> 0:15:03.770 highly intelligent AI agents capable of copying and improving themselves 0:15:03.790 --> 0:15:06.290 start to pursue their own goals and destroy humanity in 0:15:06.310 --> 0:15:10.700 the process. In others, nefarious humans enlist the capable AI 0:15:10.720 --> 0:15:14.140 to do something to create, for instance, novel viruses that 0:15:14.240 --> 0:15:17.720 wipe us all out. There's the famous paperclip illustration, which 0:15:18.240 --> 0:15:20.700 came out early in the days of AI, which said, look, 0:15:20.840 --> 0:15:24.990 if you just say to the thing, make paperclips, it might, 0:15:25.990 --> 0:15:30.250 putting aside all other concerns, morality, et cetera, marshal all 0:15:30.270 --> 0:15:32.970 of the world's resources to make paperclips and decide that 0:15:33.010 --> 0:15:35.690 humans are in the way of the paperclip manufacturer and 0:15:35.990 --> 0:15:40.280 snuff them in pursuit of its one monomaniacal goal. 0:15:40.240 --> 0:15:41.720 In the same way that if we put up a 0:15:41.780 --> 0:15:44.160 hotel in that field over there we don't think about 0:15:44.200 --> 0:15:46.880 the anthills or the mice or whatever we just build 0:15:46.920 --> 0:15:49.820 the hotel ai might do that with building whatever it 0:15:49.940 --> 0:15:50.960 is it's building. 0:15:51.100 --> 0:15:54.380 With us right right uh this is actually pretty good 0:15:54.420 --> 0:15:57.840 let's let's do jacob ward cnn tech uh guy on 0:15:57.840 --> 0:15:58.340 67 michael. 0:15:59.480 --> 0:16:01.700 We are, I think, already grappling with the idea that 0:16:02.020 --> 0:16:04.770 this stuff might take all of our jobs and reprogram 0:16:04.850 --> 0:16:07.350 our brains and reprogram our relationships to one another. But 0:16:07.430 --> 0:16:10.550 what Coxon and others are talking about here is the 0:16:10.590 --> 0:16:14.620 capacity of this technology to do things like exploit our infrastructure, 0:16:15.120 --> 0:16:18.780 set off some sort of environmental disaster, innovate some kind 0:16:18.820 --> 0:16:21.800 of horrible zoonotic disease we've never thought of or thought 0:16:21.820 --> 0:16:24.480 to try and build a cure for. What he's talking 0:16:24.500 --> 0:16:27.140 about here is the potential that this technology has could 0:16:27.180 --> 0:16:29.480 sort of supersede all other technology. 0:16:32.040 --> 0:16:33.820 You know, one thing that I probably don't think about 0:16:33.880 --> 0:16:38.710 enough is the idea of bad guys using this in 0:16:38.850 --> 0:16:41.410 evil ways that, you know, not just China trying to 0:16:41.450 --> 0:16:44.230 take over the world, but, you know, Al Qaeda. It's 0:16:44.270 --> 0:16:46.910 not like they have smart people that I'm sure are 0:16:46.930 --> 0:16:49.630 computer experts that can figure out a way to use 0:16:49.650 --> 0:16:52.110 AI to do something really awful. And all they want 0:16:52.130 --> 0:16:54.230 to do is wreak havoc. It doesn't need to have 0:16:54.250 --> 0:16:55.570 a point other than killing people. 0:16:56.020 --> 0:17:01.920 They could easily come up with a Islamist doomsday rationale 0:17:02.280 --> 0:17:06.679 for killing virtually everybody. So I want to balance this 0:17:06.740 --> 0:17:08.270 out at least a little bit for a minute with 0:17:09.230 --> 0:17:14.710 this account, Parker Thayer, who's of this world. He's a researcher. 0:17:14.810 --> 0:17:19.430 And who's he with? Investigative researcher for Capital Research. He's 0:17:20.369 --> 0:17:23.209 worked for Hillsdale College as well, which is interesting. But 0:17:23.230 --> 0:17:24.909 he says this post looks like the start of a 0:17:24.990 --> 0:17:29.149 very sophisticated and well-funded PR operation to get support for 0:17:29.190 --> 0:17:32.330 Democrats to regulate AI into oblivion. Let me show you 0:17:32.369 --> 0:17:34.709 how it works. This guy with minimal followers and no 0:17:34.790 --> 0:17:37.330 previous account activity goes to the Wall Street Journal, which 0:17:37.390 --> 0:17:40.090 publishes an exclusive with quotes from him on his resignation 0:17:40.090 --> 0:17:43.730 18 minutes before his post goes up. Planning was clearly 0:17:43.790 --> 0:17:46.330 done in advance. Two, within hours, it has tens of 0:17:46.369 --> 0:17:48.260 thousands of reposts and the account has 100,000 plus followers. 0:17:49.619 --> 0:17:52.780 The post is punchy, quotable, almost seems professionally written. The 0:17:52.820 --> 0:17:55.010 first three accounts to quote tweet it all do so 0:17:55.070 --> 0:17:57.890 within 15 minutes of the initial posting. Remember, this account 0:17:57.930 --> 0:18:02.229 had basically zero engagement beforehand. So an organic reach explanation 0:18:02.330 --> 0:18:07.409 seems unlikely. Then he goes into who first retweeted it. 0:18:08.040 --> 0:18:08.700 I want to hear this. 0:18:09.080 --> 0:18:11.200 We got to continue with this. This is a this 0:18:11.280 --> 0:18:13.359 is a well, it could be the biggest story of 0:18:13.420 --> 0:18:15.840 all time. It could be like the meteor that took 0:18:15.880 --> 0:18:16.940 out the dinosaurs. 0:18:18.280 --> 0:18:20.720 Or it could be people trying to convince us of 0:18:20.760 --> 0:18:22.040 that for their profit. 0:18:22.300 --> 0:18:22.859 Oh, boy. 0:18:23.240 --> 0:18:23.620 Wow. 0:18:24.980 --> 0:18:25.400 Stay here. 0:18:26.590 --> 0:18:27.770 Armstrong and Getty. 0:18:35.730 --> 0:18:37.430 The Armstrong and Getty Show. 0:18:42.109 --> 0:18:46.610 That's the local play-by-play on the ending of last night's game. 0:18:46.650 --> 0:18:49.149 This is only 13 seconds. Let me hear the national 0:18:49.190 --> 0:18:50.090 broadcast 16, Michael. 0:19:05.170 --> 0:19:06.169 Here is May. 0:19:06.369 --> 0:19:07.450 Quick look was close. 0:19:07.550 --> 0:19:07.970 Throwing it up. 0:19:07.950 --> 0:19:09.740 To the air. 0:19:09.750 --> 0:19:15.970 Did he get the feet down? 0:19:16.070 --> 0:19:16.139 Job! 0:19:16.150 --> 0:19:16.270 Yes! 0:19:16.609 --> 0:19:16.750 A pick! 0:19:16.770 --> 0:19:20.010 A terrible decision! A terrible decision. 0:19:20.090 --> 0:19:23.830 That could be the title of my autobiography. It's certainly 0:19:23.850 --> 0:19:29.580 a series of terrible decisions. Certainly a chapter. Yeah, definitely 0:19:29.600 --> 0:19:35.310 a chapter title. Speaking of terrible decisions, A.I., Is it 0:19:35.450 --> 0:19:40.800 indeed going to doom humanity? I think it's interesting that 0:19:41.440 --> 0:19:44.540 Parker Thayer, who we're quoting now, who's calling BS on 0:19:44.560 --> 0:19:47.850 the whole doomsday deal that we've been describing for the 0:19:47.869 --> 0:19:49.649 last couple of days, the fellow who quit the two 0:19:49.670 --> 0:19:51.940 big companies. companies and said, AI is going to kill 0:19:51.980 --> 0:19:53.700 us all if we're not extremely careful. 0:19:53.960 --> 0:19:58.600 Does he think the Doomer view is overrated in general 0:19:58.680 --> 0:20:01.410 or just this specific thing that happened over the last 0:20:01.410 --> 0:20:01.950 24 hours? 0:20:02.609 --> 0:20:09.840 You know, that's a really interesting question. He doesn't say 0:20:10.000 --> 0:20:13.080 in general he could have said it in the past 0:20:13.100 --> 0:20:16.160 i'm not familiar with his thinking on this his response 0:20:16.180 --> 0:20:18.960 though to the post from uh mr coxson who we 0:20:19.060 --> 0:20:20.780 played earlier in the hour if you didn't get to 0:20:20.800 --> 0:20:23.120 grab the podcast subscribe to armstrong and getting on demand 0:20:23.600 --> 0:20:29.030 um with brett bear uh his his response to this 0:20:29.070 --> 0:20:31.429 whole deal was this post looks like the start of 0:20:31.470 --> 0:20:35.090 a very sophisticated well-funded pr operation to get support for 0:20:35.130 --> 0:20:38.209 democrats to regulate ai into oblivion let me show you 0:20:38.250 --> 0:20:41.220 how it works And then, as we mentioned last segment, 0:20:41.240 --> 0:20:44.700 this guy's got minimal followers and no previous real account activity. 0:20:44.720 --> 0:20:48.900 He goes to the Wall Street Journal, publishes his tweet 0:20:48.900 --> 0:20:51.820 18 minutes before the post goes up. So the guy 0:20:51.859 --> 0:20:55.939 coordinated it. It was intentional. He wanted attention, which is 0:20:55.960 --> 0:20:56.930 not proof of anything. 0:20:57.030 --> 0:20:57.330 No. 0:20:57.770 --> 0:21:02.030 He wanted attention to let people know that people are 0:21:02.070 --> 0:21:05.430 discussing ending humanity in the hallways at these big companies, 0:21:05.730 --> 0:21:06.439 and you don't know it. 0:21:06.910 --> 0:21:08.330 That's a good reason to want attention. 0:21:09.030 --> 0:21:11.689 Yeah, it's maybe now's a good time to remind everybody 0:21:11.730 --> 0:21:14.840 that proof does not equal. I'm sorry. Evidence does not 0:21:14.930 --> 0:21:19.340 equal proof necessarily. Something might seem a little odd, but 0:21:19.400 --> 0:21:20.880 have a perfectly reasonable explanation. 0:21:20.940 --> 0:21:22.399 If I was going to blow the whistle on this 0:21:22.460 --> 0:21:25.840 story and I was smart enough to orchestrate it to 0:21:25.880 --> 0:21:28.780 get the maximum attention, I've got to. 0:21:29.460 --> 0:21:29.680 Right. 0:21:29.900 --> 0:21:32.740 As you quit your incredibly lucrative job, that's perhaps a 0:21:32.780 --> 0:21:36.899 sign of sincerity, not insincerity. But anyway, he points out, Mr. 0:21:36.980 --> 0:21:40.800 Thayer points out. Within hours, it has tens of thousands 0:21:40.859 --> 0:21:44.320 of reposts, and the account has 100,000 followers. The post 0:21:44.380 --> 0:21:47.700