1 00:00:01,129 --> 00:00:04,460 S1: Welcome to Unsupervised Learning, a security, AI and meaning focused 2 00:00:04,460 --> 00:00:07,340 S1: podcast that looks at how best to thrive as humans 3 00:00:07,340 --> 00:00:11,570 S1: in a post AI world. It combines original ideas, analysis, 4 00:00:11,570 --> 00:00:14,810 S1: and mental models to bring not just the news, but 5 00:00:14,810 --> 00:00:22,250 S1: why it matters and how to respond. All right, welcome 6 00:00:22,250 --> 00:00:25,460 S1: to unsupervised learning. This is Daniel Meisler. Okay. Lee Honeywell 7 00:00:25,460 --> 00:00:29,590 S1: had probably the best or coolest little quote here about, uh, 8 00:00:29,590 --> 00:00:34,060 S1: the CrowdStrike incident. Any sufficiently bad software update is indistinguishable 9 00:00:34,060 --> 00:00:36,880 S1: from a cyber attack. Love that. Had a wonderful couple 10 00:00:36,880 --> 00:00:40,480 S1: of days celebrating my buddy Jason's birthday, and it's actually 11 00:00:40,479 --> 00:00:43,690 S1: coming up on the 27th and mine is on the 26th, 12 00:00:43,690 --> 00:00:46,330 S1: so it's a little bit in the future. But we 13 00:00:46,330 --> 00:00:48,610 S1: celebrated last weekend and it was super fun. Did a 14 00:00:48,610 --> 00:00:51,190 S1: presentation for a UN group on the future of AI 15 00:00:51,190 --> 00:00:55,020 S1: and employability, and that should be coming out soon to YouTube. 16 00:00:55,020 --> 00:00:58,500 S1: So look for that soon. Doing another UL dinner in Vegas. 17 00:00:58,500 --> 00:01:00,660 S1: This is going to be cool. Mad props to all 18 00:01:00,660 --> 00:01:03,180 S1: the people who had to hustle and grind this weekend 19 00:01:03,180 --> 00:01:06,360 S1: as a result of Blue Friday, and let's get into it. 20 00:01:06,360 --> 00:01:08,430 S1: So I've been mostly heads down on the class, so 21 00:01:08,430 --> 00:01:13,319 S1: I haven't written any new essays. Security, CrowdStrike. Outage I 22 00:01:13,319 --> 00:01:16,770 S1: think we all know the deal at this point. Banks, airlines, hospitals, 23 00:01:16,770 --> 00:01:21,840 S1: media companies. It was probably the largest IT outage. I 24 00:01:21,840 --> 00:01:24,720 S1: think it's probably up for debate and people are probably 25 00:01:24,720 --> 00:01:27,060 S1: still talking about it. So I'm trying to come up 26 00:01:27,060 --> 00:01:29,310 S1: with that. Lessons learned here. But perhaps the biggest one 27 00:01:29,310 --> 00:01:33,360 S1: is around PR. So at first the CEO came out 28 00:01:33,360 --> 00:01:37,410 S1: and basically said something like, don't worry, this isn't a 29 00:01:37,410 --> 00:01:41,710 S1: security problem and that's not the way to go because 30 00:01:41,709 --> 00:01:44,500 S1: the internet's basically down as a result of you. So 31 00:01:44,500 --> 00:01:46,570 S1: I don't think that was the right stance to take. 32 00:01:46,569 --> 00:01:49,660 S1: He did much better later on. I think his subsequent 33 00:01:49,660 --> 00:01:53,559 S1: posts were much better, and where he eventually landed was great. But, um, 34 00:01:53,560 --> 00:01:55,690 S1: I like what my buddy Chris Hoff said about it, 35 00:01:55,690 --> 00:01:58,600 S1: which was, uh, this was not a security attack against 36 00:01:58,600 --> 00:02:01,750 S1: CrowdStrike or its customers, but an outage caused by a 37 00:02:01,750 --> 00:02:05,640 S1: bad software update. And this was just he was proposing 38 00:02:05,640 --> 00:02:09,899 S1: a possible option to to what was actually said. And 39 00:02:09,900 --> 00:02:12,690 S1: I like this take by Chris. And Chris has been 40 00:02:12,690 --> 00:02:16,170 S1: a CISO, uh, through some pretty rough times. So he 41 00:02:16,169 --> 00:02:18,840 S1: knows the deal. And another thought I had is that 42 00:02:18,840 --> 00:02:22,560 S1: this would be probably much less likely to happen if, say, 43 00:02:22,560 --> 00:02:26,760 S1: everything was Microsoft, right? If the EDR was just part 44 00:02:26,760 --> 00:02:30,980 S1: of Microsoft. Kind of similar to the way that defender 45 00:02:30,980 --> 00:02:33,500 S1: is getting so good that it's almost as good as 46 00:02:33,500 --> 00:02:36,410 S1: any third party option. And I think this is the 47 00:02:36,410 --> 00:02:40,520 S1: natural flow, right? You start off with a cool app idea. 48 00:02:40,520 --> 00:02:44,210 S1: It eventually becomes like a big company and eventually that 49 00:02:44,210 --> 00:02:46,340 S1: moves into the platform. And you see that with all 50 00:02:46,340 --> 00:02:51,110 S1: sorts of things like location services, antivirus and stuff like that. 51 00:02:51,110 --> 00:02:53,680 S1: So I think that's likely to just happen with EDR 52 00:02:53,710 --> 00:02:56,860 S1: as well. And the whole point is maybe Microsoft would 53 00:02:56,860 --> 00:03:00,250 S1: have handled that better or avoided it if the testing 54 00:03:00,250 --> 00:03:03,550 S1: and everything like that was more deeply integrated with the 55 00:03:03,550 --> 00:03:06,040 S1: OS itself, because they are the OS and they are 56 00:03:06,040 --> 00:03:09,430 S1: the other one. Now with this, you actually get a 57 00:03:09,430 --> 00:03:13,989 S1: different problem, which is more eggs in fewer baskets, right? 58 00:03:13,990 --> 00:03:17,520 S1: So there's a trade off there. But I feel like 59 00:03:17,520 --> 00:03:20,550 S1: this is the natural trend to basically have everything that's 60 00:03:20,550 --> 00:03:23,340 S1: awesome eventually end up in the platform itself. Got a 61 00:03:23,340 --> 00:03:26,880 S1: new threat actor called Crystal Ray using open source tool 62 00:03:26,880 --> 00:03:32,340 S1: called SSH snake to move laterally across networks, exfiltrate credentials 63 00:03:32,340 --> 00:03:36,000 S1: and deploy crypto mining software. GitHub has warned developers about 64 00:03:36,000 --> 00:03:41,270 S1: a social engineering campaign by Lazarus, which is North Korean group, 65 00:03:41,270 --> 00:03:46,310 S1: and they're targeting developers in cryptocurrency gambling and cybersecurity. And 66 00:03:46,310 --> 00:03:48,320 S1: essentially what they do is kind of what happened with 67 00:03:48,320 --> 00:03:51,890 S1: the Linux attack recently. They gained trust over time, and 68 00:03:51,890 --> 00:03:54,860 S1: then they start submitting malware once you trust them to 69 00:03:54,860 --> 00:03:58,910 S1: submit things. Special thanks to sponsor dropzone. And I wrote 70 00:03:58,910 --> 00:04:01,880 S1: this copy myself here, which I'm not going to read here, 71 00:04:01,880 --> 00:04:06,020 S1: but definitely check out dropzone AI Zonai and check out 72 00:04:06,020 --> 00:04:08,780 S1: a demo there. So not only are they sponsoring here, 73 00:04:08,780 --> 00:04:12,020 S1: but I'm actually an advisor for them as well. That's 74 00:04:12,020 --> 00:04:14,690 S1: how much I believe in their stuff. Palmer Luckey, the 75 00:04:14,690 --> 00:04:17,630 S1: guy who created Oculus, is now making AI weapons for 76 00:04:17,630 --> 00:04:21,919 S1: Ukraine through his company Anduril think it's Anduril that looks 77 00:04:21,920 --> 00:04:24,080 S1: like a Lord of the rings or some kind of 78 00:04:24,080 --> 00:04:28,760 S1: fantasy thing. See if this identifies it. Open source? Nope. 79 00:04:28,760 --> 00:04:31,450 S1: Can't identify it. All right. Yeah. So we started Android 80 00:04:31,480 --> 00:04:34,840 S1: to build AI driven weapons like drones and submarines, but 81 00:04:34,839 --> 00:04:38,620 S1: they're now being used by the Pentagon and sent to Ukraine. 82 00:04:38,620 --> 00:04:42,099 S1: China is installing massive amounts of solar and wind energy, 83 00:04:42,100 --> 00:04:47,830 S1: adding ten gigawatts of wind and solar capacity every two weeks, 84 00:04:48,100 --> 00:04:52,659 S1: which is like building five large nuclear plants every week. 85 00:04:52,660 --> 00:04:55,950 S1: And this really, really makes me mad because I, the 86 00:04:55,950 --> 00:04:58,320 S1: US needs to be doing this. Every time I fly. 87 00:04:58,320 --> 00:05:01,799 S1: I fly over wide open country that is getting hit 88 00:05:01,800 --> 00:05:05,279 S1: by the sun and it's windy down there and it's like, oh, 89 00:05:05,279 --> 00:05:07,920 S1: we don't have enough jobs for people and we don't 90 00:05:07,920 --> 00:05:10,590 S1: have enough energy, and we have way too much open 91 00:05:10,589 --> 00:05:13,799 S1: land being hit by the sun and wind. Why can 92 00:05:13,800 --> 00:05:17,010 S1: we not do this, especially when we know that AI 93 00:05:17,010 --> 00:05:19,310 S1: is going to need all this energy and we just 94 00:05:19,310 --> 00:05:21,589 S1: we need to put people to work and we need 95 00:05:21,589 --> 00:05:25,010 S1: the energy, and we have the wide open space and 96 00:05:25,010 --> 00:05:27,950 S1: we need to build more housing. So it's like this 97 00:05:27,950 --> 00:05:33,320 S1: is the perfect opportunity for a massive, like FDR type event, 98 00:05:33,320 --> 00:05:37,909 S1: which hopefully the next president gets on this because yeah, 99 00:05:37,910 --> 00:05:41,410 S1: I do not like the idea of China having all 100 00:05:41,410 --> 00:05:44,109 S1: this extra energy to do whatever with and us not 101 00:05:44,110 --> 00:05:47,440 S1: having it. Iran and China are increasing their foreign influence efforts, 102 00:05:47,440 --> 00:05:51,669 S1: using social media to stoke discord and promote anti-U.S. narratives. 103 00:05:51,670 --> 00:05:53,500 S1: And this is coming out of Google. They blocked over 104 00:05:53,500 --> 00:05:58,599 S1: 10,000 instances of Chinese influence activity in Q1 alone, thanks 105 00:05:58,600 --> 00:06:01,690 S1: to Nudge Security for sponsoring the US Department of Justice, 106 00:06:01,690 --> 00:06:05,910 S1: seize two domains and search nearly 1000 social media accounts 107 00:06:05,910 --> 00:06:11,339 S1: used by Russian actors to spread pro-Kremlin disinformation. Cloudflare says 108 00:06:11,339 --> 00:06:14,940 S1: nearly 7% of all internet traffic is malicious, with DDoS 109 00:06:14,940 --> 00:06:19,980 S1: attacks making up over 37% of all mitigated traffic. And 110 00:06:19,980 --> 00:06:22,950 S1: UK police arrested a 17 year old suspect of being 111 00:06:22,950 --> 00:06:26,610 S1: part of the scattered spider hacking group involved in the 112 00:06:26,610 --> 00:06:31,790 S1: 2023 MGM resorts ransomware attack, also known as This is 113 00:06:31,790 --> 00:06:36,080 S1: the person responsible or the group responsible for making Defcon 114 00:06:36,080 --> 00:06:39,830 S1: way farther north this year. It's now in the convention center, 115 00:06:39,830 --> 00:06:42,500 S1: pretty far north of the city, which it looks like 116 00:06:42,500 --> 00:06:45,650 S1: it's going to be pretty cool, especially for the different villages, 117 00:06:45,650 --> 00:06:47,810 S1: because all the villages are going to be together. Looking 118 00:06:47,810 --> 00:06:52,219 S1: forward to that real time video transcription with timestamps. Whisper 119 00:06:52,220 --> 00:06:57,640 S1: Diarization Diarization. Yeah, that's right, Beijing support has seen China 120 00:06:57,640 --> 00:07:00,340 S1: make up ground in the AI race, but it still 121 00:07:00,339 --> 00:07:05,050 S1: has handcuffed a bunch of AI companies with really tight restrictions. 122 00:07:05,050 --> 00:07:08,680 S1: And a lot of these are just purely political. So 123 00:07:08,680 --> 00:07:11,500 S1: they're basically saying this is going to be good for 124 00:07:11,500 --> 00:07:15,460 S1: control and bad for speed. So I think unless they 125 00:07:15,460 --> 00:07:18,480 S1: steal some sort of pinnacle AI tech, which they probably 126 00:07:18,480 --> 00:07:21,270 S1: are going to, as per Leopold, and as per a 127 00:07:21,270 --> 00:07:23,910 S1: lot of people talking about this, I think ultimately, if 128 00:07:23,910 --> 00:07:28,110 S1: that doesn't happen, this closed nature of their approach to 129 00:07:28,110 --> 00:07:31,590 S1: AI is going to hurt them, especially relative to the US, 130 00:07:31,590 --> 00:07:34,920 S1: which is just running with scissors, which you can see because, um, 131 00:07:34,920 --> 00:07:39,210 S1: meta just released llama three, uh, 405 B today and 132 00:07:39,210 --> 00:07:41,190 S1: is basically giving it out to the world. And by 133 00:07:41,190 --> 00:07:44,420 S1: the way, it's not fully open source, but it's much 134 00:07:44,420 --> 00:07:48,740 S1: more open source than other pinnacle models. But anyway, that 135 00:07:48,740 --> 00:07:51,560 S1: stuff is coming out quickly. And so the US is 136 00:07:51,560 --> 00:07:55,100 S1: innovating very, very fast. And you can't do that when 137 00:07:55,100 --> 00:07:57,440 S1: you're China and you have to be very careful what 138 00:07:57,440 --> 00:07:59,510 S1: you put in the model. Because if you train the 139 00:07:59,510 --> 00:08:03,230 S1: model on open knowledge, the model is going to figure 140 00:08:03,230 --> 00:08:06,190 S1: out how reality actually is and China doesn't want to 141 00:08:06,190 --> 00:08:09,010 S1: present the world as it actually is. They want to 142 00:08:09,010 --> 00:08:12,310 S1: present the filtered version of the world, which means they 143 00:08:12,310 --> 00:08:15,070 S1: have to do so much more protection and, you know, 144 00:08:15,070 --> 00:08:20,530 S1: post-training tricks or, um, nuking the knowledge inside of the, 145 00:08:20,530 --> 00:08:24,190 S1: the training data to not have the real knowledge of 146 00:08:24,190 --> 00:08:26,350 S1: the world. That's another way to do it, to train 147 00:08:26,350 --> 00:08:28,750 S1: the model and have it just be worse. It'll just 148 00:08:28,750 --> 00:08:31,920 S1: be worse. It'll be a worse representation of actual reality. 149 00:08:31,920 --> 00:08:35,250 S1: It'll be like, yeah, Tiananmen never happened or whatever. So, 150 00:08:35,250 --> 00:08:39,300 S1: you know, they're not oppressive. The US is bad. You know, 151 00:08:39,300 --> 00:08:42,270 S1: they are the true free society or whatever kind of 152 00:08:42,270 --> 00:08:44,939 S1: garbage they want to, you know, put in there. But 153 00:08:44,940 --> 00:08:48,780 S1: that's extra work for them. That's extra friction for them, 154 00:08:48,780 --> 00:08:51,809 S1: that's going to slow them down. And so that's one thing. 155 00:08:51,809 --> 00:08:53,840 S1: It's going to be slower. The second thing is the 156 00:08:53,840 --> 00:08:55,850 S1: people who want to go fast, who are in China. 157 00:08:55,850 --> 00:08:57,650 S1: So they're super smart and they want to go fast 158 00:08:57,650 --> 00:09:01,040 S1: and they're more like Western oriented. They're just going to leave. 159 00:09:01,040 --> 00:09:03,470 S1: They're going to go go to the US, to Canada, 160 00:09:03,470 --> 00:09:05,930 S1: to EU. Not all of them, of course, but a 161 00:09:05,929 --> 00:09:09,650 S1: lot of them. So ultimately this makes you slower as China, 162 00:09:09,650 --> 00:09:13,220 S1: the country. And it also produces more brain drain because 163 00:09:13,220 --> 00:09:16,100 S1: people don't want to be there. Kaiser Permanente is using AI, 164 00:09:16,130 --> 00:09:19,780 S1: wearables and other tech to bring health care directly to patients. 165 00:09:19,780 --> 00:09:23,230 S1: Very AI forward approach from them, which I really appreciate. 166 00:09:23,230 --> 00:09:25,210 S1: It's all in the implementation, of course, so we'll see 167 00:09:25,210 --> 00:09:28,120 S1: how that goes. But I like the open mind. Sam 168 00:09:28,120 --> 00:09:31,510 S1: Altman revealed that OpenAI's Voice Alpha mode release is coming 169 00:09:31,510 --> 00:09:34,750 S1: later this month, and llama three one just came out today, 170 00:09:34,750 --> 00:09:37,390 S1: so you know they really need to speed that up. 171 00:09:37,390 --> 00:09:40,270 S1: And I love what my buddy Matthew Berman said about this. 172 00:09:40,270 --> 00:09:44,520 S1: He's a YouTube AI guy and he says, let's denormalize 173 00:09:44,520 --> 00:09:48,540 S1: companies demoing products earlier than three months before release. So 174 00:09:48,540 --> 00:09:53,190 S1: Microsoft Recall did this. Apple intelligence did this definitely because 175 00:09:53,190 --> 00:09:56,310 S1: it's not coming out until like after September, maybe even 176 00:09:56,309 --> 00:10:01,350 S1: not until beginning of 2025. And OpenAI did this big 177 00:10:01,350 --> 00:10:05,489 S1: demo on Sora and GPT for voice, and it's still 178 00:10:05,520 --> 00:10:07,380 S1: not out yet. And that was a long time ago. 179 00:10:07,410 --> 00:10:10,110 S1: I mean, that was ancient history at this point. So yeah, 180 00:10:10,110 --> 00:10:12,179 S1: three months, I think is like the max. You should 181 00:10:12,179 --> 00:10:13,920 S1: do that. Don't even talk about it if it's not 182 00:10:13,920 --> 00:10:18,540 S1: coming out. Andrej Karpathy is launching Eureka Labs to create 183 00:10:18,540 --> 00:10:22,080 S1: AI teaching assistants for education. And yeah, this is going 184 00:10:22,080 --> 00:10:24,540 S1: to be really cool. They're basically going to have AI teachers, 185 00:10:24,540 --> 00:10:28,709 S1: which is just fantastic. Google has launched Project Oscar, open 186 00:10:28,710 --> 00:10:31,990 S1: source platform that enables development teams to create AI agents 187 00:10:31,990 --> 00:10:35,530 S1: that monitor issues, manage bugs and handle various aspects of 188 00:10:35,530 --> 00:10:39,969 S1: the software lifecycle. Omega's AI will map how Olympic athletes win. 189 00:10:39,970 --> 00:10:42,910 S1: They're using AI to map how Olympic athletes win by 190 00:10:42,910 --> 00:10:47,140 S1: analyzing their full performance, not just start and finish times. 191 00:10:47,140 --> 00:10:50,440 S1: That includes motion sensors on athletes clothing to capture every 192 00:10:50,440 --> 00:10:54,300 S1: detail of their movements. Compelling US is thinking about new 193 00:10:54,300 --> 00:10:59,280 S1: trade restrictions that can stop Nvidia from selling its HG, H20, 194 00:10:59,309 --> 00:11:02,580 S1: AI GPUs to China, and people are worried. Investors are 195 00:11:02,580 --> 00:11:05,640 S1: worried that it could cost Nvidia like $12 billion in 196 00:11:05,640 --> 00:11:08,580 S1: revenue because, you know, China is big and they buy 197 00:11:08,580 --> 00:11:10,620 S1: a lot of stuff if they're allowed to. But I 198 00:11:10,620 --> 00:11:14,100 S1: don't think that'll ultimately matter for that long. So not 199 00:11:14,100 --> 00:11:18,199 S1: too worried. As an Nvidia stockholder, Beijing scientists have developed 200 00:11:18,200 --> 00:11:24,950 S1: the world's smallest and lightest solar powered drone just 4.21g 201 00:11:24,950 --> 00:11:29,450 S1: with a 200 millimeter wingspan. I want one of these. 202 00:11:29,750 --> 00:11:32,780 S1: It can fly nonstop during the day because the motor 203 00:11:32,780 --> 00:11:37,160 S1: is lighter and more efficient, and because the thing is 204 00:11:37,160 --> 00:11:40,819 S1: so light itself, the actual craft is so light itself. 205 00:11:40,820 --> 00:11:44,140 S1: I would love to mess with that a Florida. Yeah, 206 00:11:44,140 --> 00:11:49,930 S1: it's usually either Florida man. Yeah, a Florida man. It's 207 00:11:49,929 --> 00:11:52,930 S1: it's either a Florida man or DNS. They're responsible for 208 00:11:52,929 --> 00:11:57,160 S1: all problems. Uh, this person got arrested for shooting down 209 00:11:57,160 --> 00:12:00,849 S1: a Walmart delivery drone claiming it was spying on him. 210 00:12:00,850 --> 00:12:03,849 S1: Typical of Florida man. Shooting at drones is treated as 211 00:12:03,850 --> 00:12:07,420 S1: a felony, similar to firing at a passenger aircraft. That's 212 00:12:07,420 --> 00:12:11,290 S1: probably because the bullets keep going and might hit other 213 00:12:11,290 --> 00:12:14,500 S1: things in the air. Unlikely, but turns out they they 214 00:12:14,500 --> 00:12:17,920 S1: don't launch off into orbit or move towards the sun. 215 00:12:17,920 --> 00:12:20,559 S1: They just fall down, potentially on people's heads or their 216 00:12:20,559 --> 00:12:24,550 S1: pets or whatever. Waymo wants to bring robotaxis to SFO. 217 00:12:24,580 --> 00:12:27,100 S1: They already have approval in the city, but they need 218 00:12:27,100 --> 00:12:30,179 S1: separate approval for the airport and Microsoft is laid off 219 00:12:30,179 --> 00:12:34,560 S1: their Di team, probably one of them. I assume there's many, many, 220 00:12:34,559 --> 00:12:38,460 S1: but laid off its Di team saying it's no longer 221 00:12:38,460 --> 00:12:41,189 S1: business critical. I'm not sure if this no longer business 222 00:12:41,190 --> 00:12:44,970 S1: critical is an actual quote from Microsoft leadership. It must 223 00:12:44,970 --> 00:12:49,740 S1: be because this article, which is a Business Insider article, 224 00:12:49,740 --> 00:12:53,610 S1: probably quoted that correctly. But okay. I mean, I think 225 00:12:53,610 --> 00:12:57,080 S1: the overall sentiment is worth noting, which is a lot 226 00:12:57,080 --> 00:12:59,090 S1: of people are doing this. A lot of people are 227 00:12:59,090 --> 00:13:03,860 S1: rolling back Dei as like a prime directive and realizing 228 00:13:03,860 --> 00:13:06,439 S1: it should be about merit, which I think is a 229 00:13:06,440 --> 00:13:09,319 S1: good thing. I just don't want them to go back 230 00:13:09,320 --> 00:13:13,280 S1: to just hiring people that look like you and sound 231 00:13:13,280 --> 00:13:16,700 S1: like you, and respond to questions like you, because then 232 00:13:16,700 --> 00:13:20,740 S1: you get rid of diversity, which is actually bad because 233 00:13:20,740 --> 00:13:24,100 S1: you end up with like these, these blinders. And you 234 00:13:24,100 --> 00:13:27,309 S1: have to make sure that your meritocracy is has a 235 00:13:27,309 --> 00:13:30,970 S1: wide enough top of filter, right top of funnel. So 236 00:13:30,970 --> 00:13:33,580 S1: you want to make sure you're getting everyone there into 237 00:13:33,580 --> 00:13:37,510 S1: this meritocracy filter and not just looking for certain types 238 00:13:37,510 --> 00:13:40,209 S1: of people because you think they, um, they come out 239 00:13:40,210 --> 00:13:42,960 S1: of the bottom of the filter more often. Right? So 240 00:13:42,960 --> 00:13:47,190 S1: I think the concept of Dei is 100% correct, and 241 00:13:47,190 --> 00:13:51,000 S1: we need to 100% keep doing it. The implementation that 242 00:13:51,000 --> 00:13:53,550 S1: we've done over the last few years has been bad, 243 00:13:53,550 --> 00:13:57,180 S1: and it's good to see that rolling back. Andreessen Horowitz 244 00:13:57,179 --> 00:14:00,060 S1: argues that bad government policies are now the biggest threat 245 00:14:00,059 --> 00:14:03,000 S1: to tech startups, which they call little tech. And they 246 00:14:03,000 --> 00:14:06,530 S1: basically launched a manifesto about Little tech. And I did 247 00:14:06,530 --> 00:14:09,229 S1: an analysis of it in fabric. Probably should have included that, 248 00:14:09,230 --> 00:14:14,089 S1: but it's basically pro startup stuff that's bottom line, which 249 00:14:14,090 --> 00:14:17,090 S1: makes sense because they're a VC. Essentially, Google is shutting 250 00:14:17,090 --> 00:14:21,170 S1: down its URL shortening service, so any links created with 251 00:14:21,170 --> 00:14:23,630 S1: it will stop working. If you have any important links 252 00:14:23,630 --> 00:14:26,210 S1: using the service, you want to migrate them. I'm pretty 253 00:14:26,210 --> 00:14:29,570 S1: sure Google will soon sell YouTube to like Johnson and 254 00:14:29,590 --> 00:14:32,740 S1: Johnson and Gmail to Luxottica, and then go full speed 255 00:14:32,740 --> 00:14:37,330 S1: into the what the f are we doing business? Whatever 256 00:14:37,330 --> 00:14:41,680 S1: business that is, it's the single most perplexing company like 257 00:14:41,680 --> 00:14:44,050 S1: I've ever seen. They were first on j'en ai. They 258 00:14:44,050 --> 00:14:47,560 S1: wrote the paper for Transformers, and now they're completely lapped 259 00:14:47,560 --> 00:14:51,190 S1: by not just OpenAI, but anthropic as well. And now 260 00:14:51,190 --> 00:14:56,100 S1: they're being lapped by meta by meta in an open 261 00:14:56,100 --> 00:14:58,860 S1: source model. How are you like fifth place when you 262 00:14:58,860 --> 00:15:01,170 S1: have all the people and all the money and you 263 00:15:01,170 --> 00:15:04,320 S1: actually invented the tech? They're like the opposite of Cloudflare, 264 00:15:04,320 --> 00:15:07,590 S1: which does small things really well that add up over 265 00:15:07,590 --> 00:15:11,490 S1: time because Google is like slowly dismantling everything. It's getting 266 00:15:11,490 --> 00:15:13,260 S1: rid of all the best things that it has. The 267 00:15:13,260 --> 00:15:16,650 S1: main thing Google is growing is the Google graveyard. Such 268 00:15:16,650 --> 00:15:20,180 S1: a colossal waste of money and talent. And these failures 269 00:15:20,180 --> 00:15:22,940 S1: should be studied for centuries as an example of what 270 00:15:22,940 --> 00:15:26,870 S1: happens when you don't lead with UX focused product management. 271 00:15:26,870 --> 00:15:30,500 S1: Product management, like being in charge, like thinking about the 272 00:15:30,500 --> 00:15:33,830 S1: customer and the total package and the total experience of 273 00:15:33,830 --> 00:15:36,740 S1: this product or service that you're launching, as opposed to 274 00:15:36,770 --> 00:15:40,190 S1: throw shit at the wall. Focused engineering and engineering is 275 00:15:40,190 --> 00:15:43,690 S1: in charge of product. It makes no sense whatsoever. Like 276 00:15:43,690 --> 00:15:47,350 S1: it is the most inefficient, like large organization I can 277 00:15:47,350 --> 00:15:50,170 S1: imagine that I've ever seen. All right, and Google rant, 278 00:15:50,170 --> 00:15:53,260 S1: I need a section just for Google rant. Humans. Iran 279 00:15:53,260 --> 00:15:56,590 S1: backed Houthi rebels said they were behind a drone attack 280 00:15:56,590 --> 00:16:00,640 S1: on Tel Aviv, killed one person and injured several others. 281 00:16:00,670 --> 00:16:04,330 S1: USA household income distribution by state A Reddit user shared 282 00:16:04,330 --> 00:16:08,220 S1: a detailed visualization of household income distribution across different states 283 00:16:08,220 --> 00:16:11,640 S1: in the US. New meta analysis shows that tooth brushing 284 00:16:11,640 --> 00:16:15,450 S1: can tooth brushing not hurt it like that kind of verb. 285 00:16:15,450 --> 00:16:19,260 S1: Is that tooth brushing? It's a gerund. Anyway. Brushing your 286 00:16:19,260 --> 00:16:24,300 S1: teeth can significantly reduce hospital acquired pneumonia in ICU patients. 287 00:16:24,300 --> 00:16:28,650 S1: Simple intervention could lead to 17,000 fewer deaths each year. 288 00:16:28,650 --> 00:16:31,550 S1: This is a this is a big thing I've realized recently, um, 289 00:16:31,550 --> 00:16:34,850 S1: or just gotten the religion on, which is all the 290 00:16:34,850 --> 00:16:38,660 S1: bacteria like in your teeth and gums, that's related to 291 00:16:38,660 --> 00:16:41,960 S1: inflammation in the body. It's also related to the heart 292 00:16:41,960 --> 00:16:45,560 S1: being in really bad shape. So, uh, really important that 293 00:16:45,560 --> 00:16:49,100 S1: you kind of treat that infection in your mouth or 294 00:16:49,100 --> 00:16:52,130 S1: the potential for infection in your mouth, like having a 295 00:16:52,130 --> 00:16:54,790 S1: disease throughout your entire body. And they're talking about it 296 00:16:54,790 --> 00:16:57,970 S1: going into the the brain as well, and just being 297 00:16:57,970 --> 00:17:02,500 S1: this state of disease in your body and really, really 298 00:17:02,500 --> 00:17:05,920 S1: focused on the mouth. So I've upped my game around 299 00:17:05,920 --> 00:17:09,640 S1: that significantly as a result. That was probably like a 300 00:17:09,640 --> 00:17:12,850 S1: year ago or whatever. Young adulthood is no longer one 301 00:17:12,850 --> 00:17:16,570 S1: of life's happiest times. Research shows that young adulthood is 302 00:17:16,570 --> 00:17:19,080 S1: now one of the most unhappy times in life, with 303 00:17:19,080 --> 00:17:24,150 S1: a significant rise in despair among people, especially women from 304 00:17:24,150 --> 00:17:27,540 S1: 18 to 25. And if you've been following my stuff, 305 00:17:27,540 --> 00:17:30,180 S1: I have lots of reasons for that which we will 306 00:17:30,180 --> 00:17:33,540 S1: not go into. Most of Gen Z is using TikTok 307 00:17:33,540 --> 00:17:38,250 S1: for health advice. 56% are using TikTok for a wellness 308 00:17:38,250 --> 00:17:42,139 S1: diet and fitness, 34% are relying it as their main 309 00:17:42,140 --> 00:17:46,370 S1: source of health information. About a third ask HN, which 310 00:17:46,369 --> 00:17:50,600 S1: is Hacker News everyday feels like prison. A mid-thirties guy 311 00:17:50,600 --> 00:17:52,609 S1: in tech feels trapped in a 9 to 5 job 312 00:17:52,609 --> 00:17:55,340 S1: he no longer cares about, and is struggling to build 313 00:17:55,340 --> 00:17:58,310 S1: a business on the side. This is the reason I 314 00:17:58,310 --> 00:18:01,850 S1: do unsupervised learning. This is the mission of unsupervised learning 315 00:18:01,850 --> 00:18:06,129 S1: is giving people mental models and frames and ways to 316 00:18:06,130 --> 00:18:11,500 S1: think about the world and actual methodologies for getting true 317 00:18:11,500 --> 00:18:15,250 S1: meaning in their lives. This problem right here is pretty 318 00:18:15,250 --> 00:18:18,130 S1: much the whole game, and the way I plays into 319 00:18:18,130 --> 00:18:20,679 S1: that is I think AI is going to make it 320 00:18:20,680 --> 00:18:24,070 S1: worse as it removes jobs before we have the ability 321 00:18:24,070 --> 00:18:27,520 S1: to have it increase jobs. There's going to be that that, uh, 322 00:18:27,520 --> 00:18:30,270 S1: chasm in the middle when that happens, when the jobs 323 00:18:30,270 --> 00:18:33,990 S1: come away, this meaning crisis is going to get way worse. 324 00:18:33,990 --> 00:18:36,660 S1: And that's what we are here for. All right. Get 325 00:18:36,660 --> 00:18:39,810 S1: a cool button here. Read the full newsletter online. Appreciate 326 00:18:39,810 --> 00:18:44,010 S1: the team doing that. Ideas. Sam Altman is simultaneously building 327 00:18:44,010 --> 00:18:47,670 S1: AGI and doing big studies on UBI. Super obvious what 328 00:18:47,670 --> 00:18:50,160 S1: he's doing in fact. Oh, this is kind of hot 329 00:18:50,160 --> 00:18:52,859 S1: off the press here because I just did this analysis 330 00:18:52,859 --> 00:18:56,030 S1: with fabric and I saw a tweet thread about this 331 00:18:56,030 --> 00:18:59,149 S1: as well, the UPI study that came out. This is 332 00:18:59,150 --> 00:19:02,180 S1: not in the newsletter. This just happened this morning, this 333 00:19:02,180 --> 00:19:06,860 S1: UPI study that OpenAI put out, which was funded by essentially, Altman. 334 00:19:06,859 --> 00:19:10,250 S1: It turns out at least one of the interpretations and 335 00:19:10,250 --> 00:19:12,650 S1: some of the manual work that I just did doing 336 00:19:12,650 --> 00:19:16,400 S1: AI analysis on the results, it did not increase a 337 00:19:16,400 --> 00:19:19,060 S1: lot of the things that they were hoping their spinning 338 00:19:19,060 --> 00:19:22,510 S1: it as a very positive outcome, and I'm sure they 339 00:19:22,510 --> 00:19:26,260 S1: said there is some minor positive outcomes, but in general 340 00:19:26,260 --> 00:19:28,959 S1: it did not encourage people to be more ambitious. It 341 00:19:28,960 --> 00:19:32,169 S1: did not encourage people to go and get more work. 342 00:19:32,170 --> 00:19:35,920 S1: It didn't do that. So it looks a bit bad. Now. 343 00:19:35,920 --> 00:19:38,380 S1: I think we need to do more analysis to say 344 00:19:38,380 --> 00:19:42,840 S1: for sure that UBI is not going to massively help. Um, 345 00:19:42,840 --> 00:19:45,240 S1: or it's not going to help people be more ambitious 346 00:19:45,240 --> 00:19:49,350 S1: or more creative or more productive, which which I think 347 00:19:49,350 --> 00:19:51,810 S1: is the hope. The hope is that or the idea 348 00:19:51,810 --> 00:19:55,140 S1: here is that ambitious people are lucky people who have 349 00:19:55,140 --> 00:19:57,660 S1: the time and energy to go and be ambitious, and 350 00:19:57,660 --> 00:20:00,449 S1: that when you look at people who are struggling and 351 00:20:00,450 --> 00:20:04,710 S1: they're not ambitious, it's likely because they have too many jobs, 352 00:20:04,710 --> 00:20:06,810 S1: you know, they got too much drama in their lives. 353 00:20:06,810 --> 00:20:09,390 S1: There's too much going on. And if that were to 354 00:20:09,390 --> 00:20:11,609 S1: not be the case, if they were, we were able 355 00:20:11,609 --> 00:20:14,910 S1: to reduce their stress. They would become just like the 356 00:20:14,940 --> 00:20:17,879 S1: go getters and just like the ambitious and smart people. 357 00:20:17,880 --> 00:20:22,560 S1: And this study, being so large, has the ability to 358 00:20:22,560 --> 00:20:25,830 S1: undo that narrative a bit and basically say that there's 359 00:20:25,830 --> 00:20:29,070 S1: a chance here, we need to do full analysis of this, 360 00:20:29,070 --> 00:20:32,360 S1: but there's a chance here that what it uncovers is 361 00:20:32,359 --> 00:20:36,320 S1: that no ambitious people have something else. They have this 362 00:20:36,320 --> 00:20:41,000 S1: ambition and drive that lets them thrive almost despite whatever 363 00:20:41,000 --> 00:20:44,000 S1: is going on. So if they have lots of burdens 364 00:20:44,000 --> 00:20:46,070 S1: on them, they're going to find a way through. Unless 365 00:20:46,070 --> 00:20:49,190 S1: they're overwhelming, of course. But given the same amount of burden, 366 00:20:49,190 --> 00:20:53,420 S1: people with this attribute of like drive and self-discipline and ambition, 367 00:20:53,420 --> 00:20:55,570 S1: they're just going to crush it. They're going to find 368 00:20:55,570 --> 00:20:57,790 S1: a way to crush it. And the whole narrative for 369 00:20:57,790 --> 00:21:01,390 S1: this UBI thing was that, no, everyone has that. And 370 00:21:01,390 --> 00:21:04,060 S1: I think I'm old enough at this point to realize, no, 371 00:21:04,060 --> 00:21:08,320 S1: not everyone has special beans or special magic. It really 372 00:21:08,320 --> 00:21:11,800 S1: is a small portion of, you know, society who is 373 00:21:11,800 --> 00:21:14,950 S1: in the top 10% of a given thing. Um, I mean, 374 00:21:14,950 --> 00:21:18,629 S1: just by the numbers, like only 10% can be in 375 00:21:18,630 --> 00:21:22,409 S1: the 10%, right? So I think first we need to 376 00:21:22,410 --> 00:21:25,200 S1: confirm that this is what it's saying. But even if 377 00:21:25,200 --> 00:21:27,930 S1: it is saying the worst possible thing, which is no, 378 00:21:27,930 --> 00:21:31,800 S1: this doesn't actually make people more ambitious. It doesn't really matter. 379 00:21:31,800 --> 00:21:34,590 S1: We still have to keep looking for the thing that 380 00:21:34,590 --> 00:21:37,050 S1: is the ambition. We still have to find the source 381 00:21:37,050 --> 00:21:40,109 S1: of this thing that is so valuable to us, which 382 00:21:40,109 --> 00:21:44,840 S1: is people who are driven and ambitious. Now, if it's genetic, well, 383 00:21:44,840 --> 00:21:48,560 S1: then I don't care. Right. We're not we're not going 384 00:21:48,560 --> 00:21:52,010 S1: to be changing genetics, right. Any time soon anyway. So 385 00:21:52,010 --> 00:21:54,770 S1: it's like, I don't care if we find that there's 386 00:21:54,770 --> 00:21:58,250 S1: more ambition in this group or less ambition in that group, 387 00:21:58,250 --> 00:22:01,850 S1: because those numbers are likely to be so tiny compared 388 00:22:01,850 --> 00:22:04,100 S1: to what we can control. You know, even if they 389 00:22:04,100 --> 00:22:06,280 S1: exist and they might not even exist, it might be 390 00:22:06,280 --> 00:22:08,710 S1: all the environment. But, I mean, this is a three 391 00:22:08,710 --> 00:22:11,650 S1: year study. This doesn't control, like what went into their 392 00:22:11,680 --> 00:22:15,220 S1: their upbringing and their childhood and everything. Um, because if 393 00:22:15,220 --> 00:22:18,639 S1: you look at cultures where they have this really, really 394 00:22:18,640 --> 00:22:22,090 S1: strong drive in ambition and everything, it's part of their 395 00:22:22,090 --> 00:22:24,970 S1: actual culture. This is like they pop out of the 396 00:22:24,970 --> 00:22:27,580 S1: womb and it's like, pick your college, right? You are 397 00:22:27,580 --> 00:22:31,680 S1: told from a very early age that you must be producing. 398 00:22:31,680 --> 00:22:34,740 S1: It's your responsibility to give back to community. All your, 399 00:22:34,770 --> 00:22:37,560 S1: you know, family members are looking up to you. You 400 00:22:37,560 --> 00:22:42,390 S1: know you owe us, you owe society quality outputs, and 401 00:22:42,390 --> 00:22:46,350 S1: you must work hard to create those that cultural influence 402 00:22:46,350 --> 00:22:48,660 S1: and power, plus all the peers that come with it. 403 00:22:48,660 --> 00:22:51,120 S1: Plus then you go to nice colleges and they all 404 00:22:51,119 --> 00:22:54,680 S1: believe the same thing. That entire mechanism is what we 405 00:22:54,680 --> 00:22:57,770 S1: need to give to the entire world. Okay, that that 406 00:22:57,770 --> 00:23:01,340 S1: is the thing that I believe is creating ambitious people. 407 00:23:01,340 --> 00:23:04,220 S1: So you do a three year study on UBI, you 408 00:23:04,220 --> 00:23:08,660 S1: find out that doesn't make people into those people. Fine, fine. 409 00:23:08,660 --> 00:23:12,740 S1: Something is and I believe it's that entire culture as 410 00:23:12,740 --> 00:23:15,650 S1: a package. And that's the thing that we need to 411 00:23:15,650 --> 00:23:18,459 S1: give to everyone. That's our responsibility. And if the world 412 00:23:18,460 --> 00:23:21,310 S1: were more like that, the world would would be better. 413 00:23:21,310 --> 00:23:24,580 S1: First of all, that automatically fixes crime. If you're trying 414 00:23:24,580 --> 00:23:27,790 S1: to build things and doing things and you're busy, that 415 00:23:27,790 --> 00:23:33,070 S1: automatically fixes literacy crime, like so many different things that 416 00:23:33,070 --> 00:23:36,220 S1: are a problem. So I just I don't see this 417 00:23:36,220 --> 00:23:38,920 S1: as too negative. I see it as a little bit negative, 418 00:23:38,920 --> 00:23:42,240 S1: but positive in the sense that it's going to uncover 419 00:23:42,240 --> 00:23:48,330 S1: the real issues, which is how to propagate successful mental 420 00:23:48,330 --> 00:23:53,220 S1: models and frames, um, in a wider scope across wider 421 00:23:53,220 --> 00:23:56,490 S1: groups of people across the entire planet. And I think 422 00:23:56,490 --> 00:23:59,340 S1: that's the problem we need to focus on and move 423 00:23:59,340 --> 00:24:04,890 S1: towards solving. Okay, some standalone ideas here. And I just 424 00:24:04,890 --> 00:24:08,150 S1: put the tweets in here because basically when I have ideas, I, 425 00:24:08,150 --> 00:24:12,380 S1: I um, post them on X and then come back 426 00:24:12,380 --> 00:24:15,800 S1: to them to distill into the idea section here. So 427 00:24:15,800 --> 00:24:21,200 S1: first one, what if cannabis? Cannabis, not cannabis is soma 428 00:24:21,200 --> 00:24:24,500 S1: from the Brave New World. So it makes people comfortable 429 00:24:24,500 --> 00:24:28,550 S1: with mediocrity, makes people more accepting of whatever they're handed, 430 00:24:28,550 --> 00:24:31,870 S1: and makes people less likely to change their situation. And 431 00:24:31,869 --> 00:24:36,280 S1: legalization is actually happening coincident with the rise of AI, 432 00:24:36,310 --> 00:24:38,650 S1: which is pretty cool. And I don't actually think that's 433 00:24:38,650 --> 00:24:41,590 S1: a conspiracy. Just to keep in mind, I think they're 434 00:24:41,590 --> 00:24:44,439 S1: just both happening at the same time. Conspiracy culture. Speaking 435 00:24:44,440 --> 00:24:47,140 S1: of that is getting stupid at this point. Um, yeah, 436 00:24:47,170 --> 00:24:50,470 S1: bad thing happens. It must be deep state. Um, Biden 437 00:24:50,470 --> 00:24:54,359 S1: pulls out of the race, must be deep state. I 438 00:24:54,359 --> 00:24:57,450 S1: saw a reply to my my post here. Somebody said 439 00:24:57,450 --> 00:25:00,780 S1: the number of people believing the world is like that, 440 00:25:00,780 --> 00:25:04,380 S1: where they see deep state and everything is inversely proportional 441 00:25:04,380 --> 00:25:07,199 S1: to the number of people who have worked in a 442 00:25:07,200 --> 00:25:10,110 S1: very large organization. And I think that is so smart, 443 00:25:10,109 --> 00:25:13,080 S1: because when you work in these very large organizations or 444 00:25:13,080 --> 00:25:16,980 S1: in the government or whatever, you just see constant stupidity, 445 00:25:16,980 --> 00:25:22,129 S1: just constant stupidity performed by really smart people. And the 446 00:25:22,130 --> 00:25:24,409 S1: fact that they're in these large groups with lots of 447 00:25:24,410 --> 00:25:28,610 S1: bureaucracy just makes the whole thing stupid and inefficient. So 448 00:25:28,609 --> 00:25:31,040 S1: you see enough of that. And then when you see 449 00:25:31,040 --> 00:25:35,480 S1: a botched Secret Service thing or whatever, you're like, yeah, 450 00:25:35,810 --> 00:25:38,690 S1: the world is stupid. People are stupid. Like, where's I'm 451 00:25:38,690 --> 00:25:42,520 S1: surprised anything works, right? That's the eventual attitude you get 452 00:25:42,520 --> 00:25:45,490 S1: after you work at enough of these large organizations. And 453 00:25:45,490 --> 00:25:47,650 S1: this one here, I'm going to save, because this one 454 00:25:47,650 --> 00:25:51,100 S1: is like, I'm going to do a whole essay and 455 00:25:51,100 --> 00:25:54,970 S1: video on this one. Essentially, it is the use of 456 00:25:54,970 --> 00:25:59,470 S1: AI for security on a whole bunch of intractable problems. 457 00:25:59,470 --> 00:26:02,170 S1: And I believe this is the future of AI and security, 458 00:26:02,170 --> 00:26:03,639 S1: which is why I'm going to turn it into its 459 00:26:03,640 --> 00:26:06,129 S1: own thing. And the future of security and risk management 460 00:26:06,130 --> 00:26:09,420 S1: is to have them disappear into SOPs. This one, I'm 461 00:26:09,420 --> 00:26:11,790 S1: also going to break out into a separate thing and 462 00:26:11,790 --> 00:26:15,720 S1: got someone posted here. A great tribe of Silicon Valley 463 00:26:15,720 --> 00:26:18,330 S1: is making a bid to take over the US. Vance 464 00:26:18,359 --> 00:26:22,170 S1: is Thiel's man. Musk and Andreessen Horowitz are backing him. 465 00:26:22,170 --> 00:26:25,050 S1: He will be the nominee in four years as an admin, 466 00:26:25,050 --> 00:26:28,830 S1: steered by Thiel and Musk. Right when AGI is due 467 00:26:28,830 --> 00:26:32,570 S1: and Starship goes to Mars. Incredible timeline. And basically I 468 00:26:32,570 --> 00:26:35,899 S1: just said, hmm, interesting. I'll be watching this closely. I 469 00:26:35,900 --> 00:26:41,540 S1: love looking at movements like this. Like, what is Silicon 470 00:26:41,540 --> 00:26:45,350 S1: Valley thinking? What are the top tier people? What are 471 00:26:45,350 --> 00:26:49,010 S1: their political views? Are they ambitious? Are they pessimistic? Are 472 00:26:49,010 --> 00:26:52,220 S1: they optimistic? What are they doing? What are they thinking? 473 00:26:52,220 --> 00:26:55,420 S1: Who are they backing? What are their arguments? I want 474 00:26:55,420 --> 00:26:57,639 S1: to know all this stuff. And I've got a buddy 475 00:26:57,640 --> 00:27:00,550 S1: who is so good at tracking this stuff and just 476 00:27:00,550 --> 00:27:04,240 S1: knows everything about this. And I go walking with him, um, 477 00:27:04,240 --> 00:27:07,450 S1: very often and have dinner and just get into these 478 00:27:07,450 --> 00:27:11,140 S1: great discussions about like, all these different zeitgeists happening. And 479 00:27:11,140 --> 00:27:13,959 S1: it's just it's just fantastic. And it doesn't mean I 480 00:27:13,960 --> 00:27:16,570 S1: have to agree with any of them. But hopefully if 481 00:27:16,570 --> 00:27:19,500 S1: I capture enough of them and I understand them enough, 482 00:27:19,500 --> 00:27:22,170 S1: it helps me triangulate and build my own view. All right. 483 00:27:22,170 --> 00:27:25,950 S1: Discovery lemma new recon security tool that runs via Lambda 484 00:27:25,950 --> 00:27:31,409 S1: in your browser. Responder honeypot for responder tricks attackers into 485 00:27:31,410 --> 00:27:34,709 S1: revealing their presence. Exo. Run your own AI cluster at 486 00:27:34,710 --> 00:27:39,150 S1: home on everyday devices. Why aren't we using SSH for everything? 487 00:27:39,150 --> 00:27:43,880 S1: By Shaso Grayson AI specializes in AI safety and security 488 00:27:43,880 --> 00:27:49,250 S1: tools to assess and safeguard AI deployments and Costco's Apocalypse Bucket. 489 00:27:49,250 --> 00:27:52,490 S1: They're selling a 25 year shelf life emergency food kit 490 00:27:52,490 --> 00:27:58,010 S1: called the Apocalypse Bucket for $80. Includes 150 freeze dried 491 00:27:58,010 --> 00:28:02,240 S1: and dehydrated meal servings ranging from teriyaki rice to apple 492 00:28:02,240 --> 00:28:05,359 S1: cinnamon cereal. I bought three of these just in case 493 00:28:05,359 --> 00:28:09,470 S1: things get crazy end of this year and into next year. 494 00:28:09,470 --> 00:28:12,650 S1: Plus I'm in California, lots of earthquakes and we're waiting 495 00:28:12,650 --> 00:28:15,260 S1: for the big one. All right, recommendation of the week. 496 00:28:15,260 --> 00:28:19,729 S1: Don't ask what someone's politics are because that's likely just 497 00:28:19,730 --> 00:28:22,550 S1: to get gross. They'll just respond with a bunch of 498 00:28:22,550 --> 00:28:25,760 S1: emotional things. They will trigger you emotionally and it will 499 00:28:25,760 --> 00:28:28,969 S1: be a complete mess. Instead, ask them what their ideal 500 00:28:28,970 --> 00:28:33,110 S1: world looks like, including questions like these. Are there multiple religions? 501 00:28:33,109 --> 00:28:36,379 S1: Are there multiple ethnic groups? Are people free to love 502 00:28:36,380 --> 00:28:38,840 S1: whoever they want? Do we all live together? What are 503 00:28:38,840 --> 00:28:41,360 S1: the most famous people in the world? Who gets paid 504 00:28:41,360 --> 00:28:43,969 S1: the most? Who gets paid the least? What happens to 505 00:28:43,970 --> 00:28:47,180 S1: someone if they're truly disabled and they cannot work? What 506 00:28:47,180 --> 00:28:49,970 S1: happens to someone if they're too lazy to work? What 507 00:28:49,970 --> 00:28:53,090 S1: happens to someone who is addicted to drugs? I think 508 00:28:53,090 --> 00:28:56,640 S1: many of our disagreements are about how and not what, 509 00:28:56,640 --> 00:28:59,490 S1: because I know a lot of people who support Trump, 510 00:28:59,490 --> 00:29:03,960 S1: for example, who would say, yeah, you can absolutely be gay. Yeah, 511 00:29:03,960 --> 00:29:08,760 S1: you can absolutely be transgender. No problem. Love to everyone. Yes, 512 00:29:08,760 --> 00:29:12,330 S1: there absolutely can be other religions. Yes, all the ethnic 513 00:29:12,330 --> 00:29:15,630 S1: groups should live together. Yes, there should be a safety net. 514 00:29:15,630 --> 00:29:20,540 S1: I literally know multiple Trump supporters who voted for Trump twice, 515 00:29:20,540 --> 00:29:23,360 S1: who believe all four of these. You can be gay, 516 00:29:23,360 --> 00:29:26,870 S1: you can be transgender, you can be other religions. We 517 00:29:26,870 --> 00:29:29,510 S1: should all live together. We should all get along, and 518 00:29:29,510 --> 00:29:32,540 S1: there should be a social safety net. Okay, so if 519 00:29:32,540 --> 00:29:34,190 S1: you're on the left and you hear someone on the 520 00:29:34,190 --> 00:29:36,200 S1: right say those things, or you're on the right and 521 00:29:36,200 --> 00:29:38,540 S1: you hear the left say, you know some other stuff, 522 00:29:38,540 --> 00:29:42,320 S1: this is an opportunity for a real conversation. Both sides 523 00:29:42,320 --> 00:29:46,300 S1: should describe what they view as an ideal version of 524 00:29:46,300 --> 00:29:48,940 S1: the world. And I would argue to you that if 525 00:29:48,940 --> 00:29:52,630 S1: you both start with that, it's going to look very similar. 526 00:29:52,630 --> 00:29:57,370 S1: I think roughly the center 80 or 70% of the 527 00:29:57,370 --> 00:30:00,460 S1: country is going to build that model, and it's mostly 528 00:30:00,460 --> 00:30:03,010 S1: going to look the same, and it's mostly going to 529 00:30:03,010 --> 00:30:06,070 S1: include those 4 or 5 things that I just mentioned. 530 00:30:06,070 --> 00:30:09,450 S1: So if you had a smart moderator in between, you 531 00:30:09,450 --> 00:30:11,580 S1: could look at these two people and be like, okay, 532 00:30:11,580 --> 00:30:13,830 S1: let me get this straight. You think there should be 533 00:30:13,830 --> 00:30:18,900 S1: multiple ethnic groups and religions and gender identities, and we 534 00:30:18,900 --> 00:30:22,080 S1: should all work together and all live together and be 535 00:30:22,080 --> 00:30:24,480 S1: happy together and not infringe on each other. Is that 536 00:30:24,480 --> 00:30:27,270 S1: is that correct? And both of them say yes. Do 537 00:30:27,270 --> 00:30:29,460 S1: you realize how much progress that is? Do you realize 538 00:30:29,460 --> 00:30:32,780 S1: how much we need that conversation to take place right now, 539 00:30:32,780 --> 00:30:36,440 S1: in the middle of 2024, going into this crazy election cycle? 540 00:30:36,440 --> 00:30:39,020 S1: We absolutely need that. So that's where you should start 541 00:30:39,020 --> 00:30:42,080 S1: the conversation. Don't start with like, hey, what are your politics? 542 00:30:42,080 --> 00:30:44,360 S1: Because they're they're going to go to an emotional hot 543 00:30:44,360 --> 00:30:46,700 S1: button that's going to trigger both of you. Instead, talk 544 00:30:46,700 --> 00:30:49,160 S1: about the ideal. And when you realize you agree on 545 00:30:49,160 --> 00:30:52,190 S1: the ideal, now we have a how problem okay, now 546 00:30:52,190 --> 00:30:56,020 S1: we have a transition problem. Cool. That's much more tractable 547 00:30:56,020 --> 00:30:59,860 S1: and much more discussable than disagreeing on where we're supposed 548 00:30:59,860 --> 00:31:02,620 S1: to go. All right. In the aphorism of the week, 549 00:31:02,620 --> 00:31:06,580 S1: silence is a fence around wisdom. Silence is a fence 550 00:31:06,580 --> 00:31:12,070 S1: around wisdom. German proverb. Unsupervised learning is produced and edited 551 00:31:12,070 --> 00:31:16,510 S1: by Daniel Miessler on a Neumann U87 AI microphone using Hindenburg. 552 00:31:17,020 --> 00:31:19,830 S1: Intro and outro music is by Zomby with the Y, 553 00:31:20,550 --> 00:31:22,800 S1: and to get the text and links from this episode, 554 00:31:22,800 --> 00:31:25,050 S1: sign up for the newsletter version of the show at 555 00:31:25,050 --> 00:31:29,760 S1: Daniel miessler.com/newsletter. We'll see you next time.