WEBVTT - UL NO. 433: China's Flawed Strategy 0:00:01.100 --> 0:00:04.490 Welcome to Unsupervised Learning, a security, AI and meaning focused 0:00:04.490 --> 0:00:07.370 podcast that looks at how best to thrive as humans 0:00:07.370 --> 0:00:11.570 in a post AI world. It combines original ideas, analysis, 0:00:11.570 --> 0:00:14.810 and mental models to bring not just the news, but 0:00:14.810 --> 0:00:21.709 why it matters and how to respond. All right. Welcome 0:00:21.710 --> 0:00:25.040 to unsupervised learning. This is Daniel. Hope you're doing well. 0:00:25.040 --> 0:00:28.550 So we added a new fabric pattern called explained Term's 0:00:28.550 --> 0:00:31.730 which takes an input, finds the difficult terms and gives 0:00:31.730 --> 0:00:35.059 a definition. And I got this put together because I 0:00:35.060 --> 0:00:40.400 was watching a really complex machine learning AI video with Daksh, 0:00:40.400 --> 0:00:43.610 and I forget who the people were. Oh, it was 0:00:43.610 --> 0:00:46.610 two different guests, one from anthropic and maybe another one 0:00:46.610 --> 0:00:49.010 from DeepMind. Not sure, but they went into some pretty 0:00:49.010 --> 0:00:51.140 heavy stuff, and there was a bunch of terms that 0:00:51.140 --> 0:00:53.150 I just heard them talking about, and I wished I 0:00:53.150 --> 0:00:55.100 could just go and look them up. So I went 0:00:55.100 --> 0:00:58.790 and created this pattern for that. Had a fantastic weekend 0:00:58.790 --> 0:01:02.450 at EDC, the annual pilgrimage for a bunch of friends 0:01:02.450 --> 0:01:05.450 and I. And yeah, it was fabulous. Just great music, 0:01:05.450 --> 0:01:09.080 great companionship, just amazing. And I would say if you 0:01:09.080 --> 0:01:12.020 don't have friends that you go and do something with, 0:01:12.020 --> 0:01:15.560 like every year, you want to definitely do that. It's, uh, 0:01:15.560 --> 0:01:19.069 it's really valuable to have something to look forward to 0:01:19.069 --> 0:01:21.560 and just have that time with friends. And it's like, 0:01:21.560 --> 0:01:23.900 it helps you get through the rest of the year, 0:01:23.900 --> 0:01:27.260 which I haven't been having that much trouble doing that lately, 0:01:27.260 --> 0:01:30.950 but it's, uh, it's something that everyone should be doing. Like, 0:01:30.950 --> 0:01:33.530 Tim Ferriss talked about this and it got me thinking 0:01:33.530 --> 0:01:36.500 about it. And luckily I already have a couple of these. Basically, 0:01:36.500 --> 0:01:38.390 one is like Defcon black hat, and the other one 0:01:38.390 --> 0:01:41.929 is EDC, and we're thinking about sprinkling a few smaller 0:01:41.930 --> 0:01:44.570 ones in there as well. But something to do if 0:01:44.569 --> 0:01:46.700 you don't have these already and it could be local. 0:01:46.700 --> 0:01:50.120 It could just be like LAN parties or whatever, just 0:01:50.120 --> 0:01:52.790 small things. Thinking about writing a short book about what 0:01:52.790 --> 0:01:56.270 people will need to become antifragile. In the world of AI, 0:01:56.300 --> 0:01:58.310 a lot of people are asking me this like, what 0:01:58.310 --> 0:02:00.080 do I need to learn? What are the skills I 0:02:00.080 --> 0:02:03.800 have to master to be resilient to AI? And a 0:02:03.800 --> 0:02:07.700 working title for this is something like human 3.0 Skills 0:02:07.700 --> 0:02:11.179 and Mental Frames Required to thrive in a post AI world. 0:02:11.180 --> 0:02:15.320 And I've actually forgot to look at this. I haven't 0:02:15.320 --> 0:02:18.440 looked at that, but we'll take a look at it afterwards. So, uh, 0:02:18.440 --> 0:02:22.130 got a post here that, uh, was from a while back, 0:02:22.130 --> 0:02:24.890 like May of 2023, where I talked about the attack 0:02:24.889 --> 0:02:28.310 surface map. Not sure why this is not linking. There 0:02:28.310 --> 0:02:31.040 it goes. Yeah. So this is the attack surface map 0:02:31.040 --> 0:02:35.150 I put together in May of 23, I think so 0:02:35.150 --> 0:02:38.150 about a year ago, and it turns out to be 0:02:38.150 --> 0:02:42.830 pretty useful. Still, I think it's still spot on. I'm 0:02:42.830 --> 0:02:45.320 not sure what I would modify from this, but worth 0:02:45.320 --> 0:02:48.200 taking a look at if you haven't seen it. Okay, security, 0:02:48.200 --> 0:02:51.650 see what we got here? So the FBI seized a 0:02:51.650 --> 0:02:55.339 bunch of hacking, uh, breach forms again, Chinese brain drain. 0:02:55.340 --> 0:02:58.550 So Microsoft is moving nearly 10% of its China based 0:02:58.550 --> 0:03:01.730 tech talent out of the country. Yeah, they're talking about 0:03:01.730 --> 0:03:05.419 brain drain here. And I personally am pretty happy about this. 0:03:05.419 --> 0:03:07.820 I think I feel like the US is in a 0:03:07.820 --> 0:03:10.700 good state right now. It's in a good place to 0:03:10.700 --> 0:03:14.600 win because Europe is suffering from all sorts of missing 0:03:14.600 --> 0:03:19.310 innovation because of regulations. They just love regulating things, regulating 0:03:19.310 --> 0:03:22.460 things too much. And that's a problem. So I think 0:03:22.460 --> 0:03:25.400 that's like taking them out of the game. China is 0:03:25.400 --> 0:03:29.210 being just super aggressive to everyone all the time, hacking everyone, 0:03:29.210 --> 0:03:32.150 trying to control everyone. Everyone's trying to control everyone and some, 0:03:32.150 --> 0:03:35.690 some sense. And definitely in the US is not innocent 0:03:35.690 --> 0:03:39.260 of that. But I would say that our means of 0:03:39.260 --> 0:03:44.000 doing that in our goal of doing that is more benign. 0:03:44.000 --> 0:03:48.650 And this really leaves the US and Canada to benefit 0:03:48.650 --> 0:03:52.010 from this boom in talent and people that are going 0:03:52.010 --> 0:03:56.000 to be leaving China. China has so many people, and 0:03:56.000 --> 0:03:59.240 a small percentage of those people have a high IQ, 0:03:59.240 --> 0:04:01.610 and a small percentage of people in the US have 0:04:01.610 --> 0:04:04.520 a high IQ. And guess what? When you have three 0:04:04.520 --> 0:04:06.830 times the people, or is it four times the people, 0:04:06.830 --> 0:04:09.200 I can't remember. But when you have that many more 0:04:09.200 --> 0:04:12.830 people and plus they have a better education system, like 0:04:12.830 --> 0:04:17.180 they have so much talent and to have a lot 0:04:17.180 --> 0:04:19.909 of that talent go to the US and Canada and 0:04:19.910 --> 0:04:22.790 to some degree to Europe. That is a major advantage 0:04:22.790 --> 0:04:26.270 for the West. And the other disadvantage that China has 0:04:26.270 --> 0:04:29.060 around AI is that I think they don't want to 0:04:29.060 --> 0:04:32.330 just turn it on and enable it, because it encourages 0:04:32.330 --> 0:04:36.560 people to ask questions. And China doesn't like people asking questions. 0:04:36.560 --> 0:04:38.929 They want to feed them a narrative and have that 0:04:38.930 --> 0:04:42.890 narrative be the only one and basically control that. And 0:04:42.890 --> 0:04:45.230 it's just not a game that they're going to win. 0:04:45.230 --> 0:04:48.620 I don't think, well, they could win, they could still win. 0:04:48.620 --> 0:04:52.219 But I feel like they're setting themselves up to fail 0:04:52.220 --> 0:04:56.420 and for the West to benefit. So what's essentially going 0:04:56.420 --> 0:04:58.940 to happen as a result is the world will pull 0:04:58.940 --> 0:05:02.120 away from them, and their best minds will brain drain 0:05:02.120 --> 0:05:04.190 out of the country. They will go to Europe, they 0:05:04.190 --> 0:05:06.230 will go to Canada, and they will go to the US. 0:05:06.230 --> 0:05:10.250 Air Force Secretary Frank Kendall flew in an AI controlled 0:05:10.250 --> 0:05:14.089 F-16 and basically said, it's just about as good as 0:05:14.089 --> 0:05:18.950 experienced human pilots, technology. Everything covered at Google I o got. 0:05:19.080 --> 0:05:22.109 A link here, and unfortunately, it's impossible for me to 0:05:22.110 --> 0:05:25.080 get excited about this because Google has turned into like 0:05:25.080 --> 0:05:27.539 a cool idea lab because they're not really good at 0:05:27.540 --> 0:05:32.099 making products. Why? Because they're simply not good at managing products. 0:05:32.100 --> 0:05:34.710 They have so many smart engineers, and they have lots 0:05:34.710 --> 0:05:37.799 of talent and lots of cool potential ideas, but they 0:05:37.800 --> 0:05:40.350 don't have vision and they don't have product management, and 0:05:40.350 --> 0:05:42.930 they don't have vision. That's led by product management and 0:05:42.930 --> 0:05:45.510 by the leaders. They have basically a throw it at 0:05:45.510 --> 0:05:48.720 the wall technique and it's really hurting them. And I 0:05:48.720 --> 0:05:51.240 think the only thing that's going to fix it is 0:05:51.240 --> 0:05:55.260 like a dramatic overhaul with replacement of the CEO. I 0:05:55.260 --> 0:05:57.780 think that's looking more and more likely. Pretty cool idea 0:05:57.810 --> 0:06:02.160 about how to avoid the mediocre success trap, because you 0:06:02.160 --> 0:06:05.279 basically want a home run or strike out. And this 0:06:05.279 --> 0:06:08.550 is talking about not getting a strong signal from an experiment. 0:06:08.550 --> 0:06:11.700 This is the context that they're talking about here. Ilya 0:06:11.700 --> 0:06:17.610 Sutskever suggests mastering these 30 ML papers to grasp 90% 0:06:17.610 --> 0:06:21.479 of crucial machine learning knowledge. I've started going through the list. 0:06:21.480 --> 0:06:25.140 I'm not all the way through by any means. MIT 0:06:25.140 --> 0:06:29.520 researchers came across three ancient stars, and they're going around 0:06:29.520 --> 0:06:32.489 the Milky Way for some reason, and they appear to 0:06:32.490 --> 0:06:35.729 be some of the oldest in the entire universe. And 0:06:35.730 --> 0:06:38.820 I'm like, how do they randomly end up near us? 0:06:38.820 --> 0:06:42.630 A strange and the article didn't really explain it that well, 0:06:42.630 --> 0:06:45.630 scientists just found out that whales have their own alphabet 0:06:45.630 --> 0:06:48.570 in their songs, and my buddy Mark has been working 0:06:48.570 --> 0:06:51.479 on deciphering this stuff for years. He's got some real 0:06:51.480 --> 0:06:54.720 cool research that he's working on, hopes to be able 0:06:54.720 --> 0:06:57.479 to present on it at some sort of marine conference soon. 0:06:57.480 --> 0:07:00.960 And this one's about how Swiss parenting styles are so 0:07:00.960 --> 0:07:03.539 different from American ones, and they talk about it in 0:07:03.540 --> 0:07:06.089 this article. Basically, four year olds being able to walk 0:07:06.089 --> 0:07:09.300 to school by themselves in Switzerland. And some will say, well, 0:07:09.300 --> 0:07:11.760 that's because it's more safe there. It's not really that 0:07:11.760 --> 0:07:14.970 unsafe here either in the US for kids to walk 0:07:14.970 --> 0:07:17.730 to school, but people think it is and that's that's 0:07:17.730 --> 0:07:20.460 their frame. That's their reality. All right. This one here, 0:07:20.460 --> 0:07:23.580 I think the magic amount of therapy for most people, 0:07:23.610 --> 0:07:26.970 of course, not all might be some relatively small amount 0:07:26.970 --> 0:07:29.760 of therapy that untangles your knots, but doesn't point your 0:07:29.760 --> 0:07:32.910 lens permanently on yourself in your troubles. Because I feel 0:07:32.910 --> 0:07:36.990 like too much therapy has the approach of like bad chiropractic, 0:07:36.990 --> 0:07:39.420 where the goal is they want you to come back 0:07:39.420 --> 0:07:43.050 and have more chiropractic visits, right? The goal is not 0:07:43.050 --> 0:07:45.300 to fix you where you don't come back. And I 0:07:45.300 --> 0:07:48.330 think just so much focus on yourself and your problems. 0:07:48.330 --> 0:07:52.770 It's essentially rumination. And rumination is a bad thing that actually, ironically, 0:07:52.770 --> 0:07:55.260 you go to therapy to figure out how not to do. 0:07:55.260 --> 0:08:00.270 And I'm worried that the therapy industry is basically encouraging rumination. 0:08:00.270 --> 0:08:02.380 Recommendation of the week. Make sure you have 1 to 0:08:02.380 --> 0:08:04.920 2 friend trips planned per year, even if it's not 0:08:04.920 --> 0:08:09.060 everyone and it's just local. Just anything camping, hiking, whatever, 0:08:09.060 --> 0:08:12.420 role playing games, video games, whatever. And maybe you have 0:08:12.420 --> 0:08:14.820 a rule of like no tech during that time just 0:08:14.820 --> 0:08:19.140 so you can have interactions, live interactions, eat food together, 0:08:19.140 --> 0:08:23.940 hang out, even watching TV, just talking, whatever. Super replenishing, 0:08:23.940 --> 0:08:26.280 as I just found out from this last trip. And 0:08:26.280 --> 0:08:28.890 the aphorism of the week every minute you are angry, 0:08:28.890 --> 0:08:33.240 you lose 60s of happiness. Every minute you are angry, 0:08:33.240 --> 0:08:38.670 you lose 60s of happiness. Ralph Waldo Emerson. Unsupervised Learning 0:08:38.670 --> 0:08:41.309 is produced and edited by Daniel Meisler on a Neumann 0:08:41.309 --> 0:08:45.900 U87 AI microphone using Hindenburg. Intro and outro music is 0:08:45.900 --> 0:08:49.140 by zombie with the why and to get the text 0:08:49.140 --> 0:08:51.300 and links from this episode, sign up for the newsletter 0:08:51.300 --> 0:08:56.660 version of the show at Daniel meisler.com/newsletter. We'll see you 0:08:56.660 --> 0:08:57.320 next time.