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 --> 00:00:07,310 S1: podcast that looks at how best to thrive as humans 3 00:00:07,310 --> 00:00:11,540 S1: in a post AI world. It combines original ideas, analysis, 4 00:00:11,570 --> 00:00:14,780 S1: and mental models to bring not just the news, but 5 00:00:14,780 --> 00:00:22,340 S1: why it matters and how to respond. All right, welcome 6 00:00:22,340 --> 00:00:26,060 S1: to unsupervised learning. This is Daniel Miessler. All right. What 7 00:00:26,060 --> 00:00:29,480 S1: have we got here? Lots of content this week. Super 8 00:00:29,480 --> 00:00:32,840 S1: excited for this episode. Uh, did an all text episode 9 00:00:32,840 --> 00:00:34,940 S1: this time, which is kind of a callback to how 10 00:00:34,940 --> 00:00:37,520 S1: I used to do it. Got upcoming speaking at the 11 00:00:37,520 --> 00:00:43,400 S1: Snyk conference in October, Cyberstorm in Switzerland and Blackhat in Riyadh. 12 00:00:43,430 --> 00:00:47,000 S1: One tool in AI that you should be trying out, 13 00:00:47,000 --> 00:00:50,870 S1: that everyone is talking about, including Karpathy and a bunch 14 00:00:50,870 --> 00:00:54,800 S1: of other people, is called cursor AI. So it's cursor.com 15 00:00:54,800 --> 00:00:57,230 S1: is the domain. I thought it was cursor AI, but 16 00:00:57,230 --> 00:01:00,730 S1: it is not. But the big feature appears to be 17 00:01:00,730 --> 00:01:04,630 S1: that it basically it looks and feels like VSCode, but 18 00:01:04,630 --> 00:01:08,259 S1: what you do is you upload your entire repository into it. 19 00:01:08,290 --> 00:01:11,710 S1: I guess I guess technically VSCode could also see your 20 00:01:11,740 --> 00:01:14,259 S1: all your code if it's all in there as well. 21 00:01:14,260 --> 00:01:17,110 S1: But I think what cursor is supposedly doing well is 22 00:01:17,110 --> 00:01:19,900 S1: it's taking all of that content and kind of using 23 00:01:19,900 --> 00:01:22,900 S1: it in context to understand it better, not just the 24 00:01:22,900 --> 00:01:25,240 S1: current file that you're in. So I believe that's the 25 00:01:25,240 --> 00:01:28,360 S1: big feature. If I'm wrong about that, somebody correct me. Okay. 26 00:01:28,390 --> 00:01:33,280 S1: My work, a couple of massive episodes or essays that 27 00:01:33,280 --> 00:01:35,620 S1: I put out this week, and I actually sent them 28 00:01:35,620 --> 00:01:38,470 S1: out directly, which I only do for things that I 29 00:01:38,470 --> 00:01:43,000 S1: think are really decent and also evergreen. So one of 30 00:01:43,000 --> 00:01:46,600 S1: them is called We Were Lied To About Work or 31 00:01:46,600 --> 00:01:48,610 S1: the Real Problem with the job market. I had two 32 00:01:48,610 --> 00:01:52,000 S1: different titles actually, but it's basically why layoffs, hiring the 33 00:01:52,000 --> 00:01:54,850 S1: job market, and work in general just really sucks right now. 34 00:01:54,850 --> 00:01:56,559 S1: And I would say it's probably one of my top 35 00:01:56,590 --> 00:02:00,970 S1: 20 essays ever. So highly, highly recommend that one. And 36 00:02:00,970 --> 00:02:03,850 S1: I've got a new way to explain AI and specifically 37 00:02:03,880 --> 00:02:06,880 S1: Llms to people, and I think this one is short enough. 38 00:02:06,880 --> 00:02:08,920 S1: In fact, I'm just going to give the highlights of it. 39 00:02:08,919 --> 00:02:11,950 S1: So let me take you into this one. So here 40 00:02:11,980 --> 00:02:16,510 S1: is the basic concept five levels of LLM understanding. So 41 00:02:16,510 --> 00:02:19,630 S1: the first level at the bottom is it's just predicting 42 00:02:19,660 --> 00:02:23,440 S1: text like the next text token in a in a 43 00:02:23,440 --> 00:02:27,429 S1: text sequence. That's all it's doing. It's not magical. Right. 44 00:02:27,430 --> 00:02:30,010 S1: And this is kind of the most common argument for 45 00:02:30,010 --> 00:02:33,580 S1: why llms and AI is not all that special or 46 00:02:33,580 --> 00:02:37,180 S1: specifically AI based on Llms. People are just like, look, 47 00:02:37,210 --> 00:02:40,030 S1: it's just next token prediction. No big deal. If you 48 00:02:40,030 --> 00:02:43,090 S1: pull away from that one level, it's like, look, it's 49 00:02:43,090 --> 00:02:46,360 S1: just predicting the next item in a sequence, okay? So 50 00:02:46,360 --> 00:02:50,919 S1: it's just next token prediction okay. One abstraction away from there. 51 00:02:50,919 --> 00:02:54,880 S1: And this all comes from this thing from Eliezer Yudkowsky 52 00:02:54,910 --> 00:02:59,429 S1: who had this, this great kind of little statement on 53 00:02:59,430 --> 00:03:02,340 S1: X that made me think about this, but essentially what 54 00:03:02,340 --> 00:03:06,090 S1: it is, is for this level. Level three, it's predicting 55 00:03:06,090 --> 00:03:09,839 S1: the next token in the description of an answer. Okay. 56 00:03:09,870 --> 00:03:16,110 S1: So what Yudkowsky said was any predicting the next token 57 00:03:16,110 --> 00:03:20,850 S1: is isomorphic with predicting the next token of an answer. Okay. 58 00:03:20,880 --> 00:03:23,220 S1: And that is really, really powerful. And it wasn't the 59 00:03:23,220 --> 00:03:25,740 S1: exact quote, but that's essentially it. If you go one 60 00:03:25,740 --> 00:03:31,350 S1: level above that okay, it's predicting answers to insanely difficult questions. 61 00:03:31,350 --> 00:03:34,590 S1: So so here are the four levels so far. It's 62 00:03:34,590 --> 00:03:37,920 S1: predicting the next token in a piece of text. It's 63 00:03:37,920 --> 00:03:41,670 S1: predicting the next token third level. It's predicting the next 64 00:03:41,670 --> 00:03:46,080 S1: token of an answer. Second level it's predicting answers. And 65 00:03:46,080 --> 00:03:49,770 S1: the top level it seems to know everything okay. And 66 00:03:49,770 --> 00:03:53,100 S1: this all comes from this. Here it is right here 67 00:03:53,100 --> 00:03:57,390 S1: it just predicts the next token literally any well posed 68 00:03:57,410 --> 00:04:01,340 S1: problem is isomorphic to predict the next token of the answer. 69 00:04:01,340 --> 00:04:05,030 S1: And literally anyone with a grasp of undergraduate comp sci 70 00:04:05,060 --> 00:04:07,820 S1: is supposed to see that without being told. I don't 71 00:04:07,850 --> 00:04:10,160 S1: agree with that last part. This part here, I don't 72 00:04:10,190 --> 00:04:13,070 S1: agree with that. I don't think that's accurate. In fact, 73 00:04:13,070 --> 00:04:16,400 S1: I think the vast majority of people that's very much 74 00:04:16,400 --> 00:04:18,979 S1: not the case. Very much not the case. I would 75 00:04:18,980 --> 00:04:23,300 S1: say 95%. So I would say I disagree on multiple 76 00:04:23,300 --> 00:04:26,360 S1: levels for that second piece. The part that I like 77 00:04:26,390 --> 00:04:32,060 S1: is this here, which I highlighted and separated on purpose, 78 00:04:32,060 --> 00:04:36,950 S1: literally any well posed problem is isomorphic to predict the 79 00:04:36,950 --> 00:04:42,530 S1: next token of the answer. That is extraordinary. That is brilliant. 80 00:04:42,529 --> 00:04:46,310 S1: That is the answer to the it's just next token 81 00:04:46,310 --> 00:04:50,240 S1: prediction argument. That is, the immediate counter of that well 82 00:04:50,270 --> 00:04:54,500 S1: posed problem equals prediction of next token of answer. So 83 00:04:54,500 --> 00:04:58,010 S1: that's that one. And basically this essay is a full 84 00:04:58,040 --> 00:05:02,630 S1: breakdown of that. So it's a standalone evergreen essay about 85 00:05:02,630 --> 00:05:05,660 S1: why this is a powerful concept and why you shouldn't 86 00:05:05,660 --> 00:05:09,800 S1: trust that next token argument. Okay. Security. CrowdStrike did their 87 00:05:09,980 --> 00:05:15,680 S1: 2024 report talking about how North Koreans have infiltrated over 88 00:05:15,680 --> 00:05:19,339 S1: 100 US based companies in all sorts of important places 89 00:05:19,339 --> 00:05:22,250 S1: like aerospace, defense, retail and tech. They didn't mention much 90 00:05:22,250 --> 00:05:25,490 S1: about Blue Friday. Not sure why that was. State linked 91 00:05:25,490 --> 00:05:29,930 S1: Chinese entities are using cloud services from Amazon and other 92 00:05:29,930 --> 00:05:34,940 S1: competitors of Amazon to access advanced US chips and AI capabilities, 93 00:05:34,940 --> 00:05:38,000 S1: so basically they can't get the chips themselves. But you 94 00:05:38,000 --> 00:05:41,239 S1: can go to AWS. Anybody can, or most people can. 95 00:05:41,270 --> 00:05:44,029 S1: You go to AWS, you sign up and you just 96 00:05:44,029 --> 00:05:48,740 S1: attach Nvidia chips to that instance and you start doing workloads. 97 00:05:48,740 --> 00:05:51,890 S1: And so this is what China is doing. They're like, well, 98 00:05:51,920 --> 00:05:54,080 S1: we can't get the chips, but we can use a 99 00:05:54,080 --> 00:05:57,679 S1: service that leverages the chips. So I'm sure Amazon and 100 00:05:57,680 --> 00:06:00,590 S1: the government are in conversations about that trying to get 101 00:06:00,589 --> 00:06:04,190 S1: that fixed. Cisco has patched multiple vulnerabilities, including a high 102 00:06:04,190 --> 00:06:08,600 S1: severity bug in its Unified Communications Manager product. Thanks to 103 00:06:08,630 --> 00:06:12,860 S1: Threatlocker for sponsoring. Two US lawmakers are urging the Commerce 104 00:06:12,860 --> 00:06:18,260 S1: Department to investigate cybersecurity risks associated with TP-Link routers, citing 105 00:06:18,260 --> 00:06:22,219 S1: vulnerabilities and potential data sharing with the Chinese government. So 106 00:06:22,220 --> 00:06:25,820 S1: kind of a Huawei situation. A mini Huawei Quarks lab 107 00:06:25,820 --> 00:06:28,970 S1: found a major back door in RFID cards made by 108 00:06:28,970 --> 00:06:34,310 S1: Shanghai Fudan Microelectronics, one of China's top chip manufacturers. I 109 00:06:34,310 --> 00:06:37,339 S1: don't think this is a China influence story. I think 110 00:06:37,339 --> 00:06:40,489 S1: this is just vulnerabilities and chips, and the impact here 111 00:06:40,490 --> 00:06:43,730 S1: is the fact that it's smart cards for like office 112 00:06:43,730 --> 00:06:46,729 S1: stores and hotel rooms and whatever. And it's a big 113 00:06:46,730 --> 00:06:50,780 S1: company that does this. And lots of RFID chips and 114 00:06:50,779 --> 00:06:54,020 S1: cards come from China. So just an impact due to 115 00:06:54,050 --> 00:06:56,440 S1: the size of the market type situation. What is this 116 00:06:56,440 --> 00:07:03,370 S1: bouncing now? Now you're seeing my Diablo four chat messages. 117 00:07:03,370 --> 00:07:06,039 S1: It's really important stuff I'm working on in Diablo four. 118 00:07:06,070 --> 00:07:08,440 S1: All right. What are we talking about here? Yeah. Thanks 119 00:07:08,440 --> 00:07:12,460 S1: to defender five for sponsoring. Next one. Researchers found a 120 00:07:12,460 --> 00:07:15,670 S1: way to exfiltrate data from Slack's AI by using indirect 121 00:07:15,670 --> 00:07:18,910 S1: prompt injection. US Navy is rolling out Starlink on its 122 00:07:18,940 --> 00:07:23,380 S1: warships to provide high speed, reliable internet connections, improving operations 123 00:07:23,380 --> 00:07:27,790 S1: and crew morale. AI and Tech Anthropic has published the 124 00:07:27,790 --> 00:07:31,720 S1: system prompts for its latest AI models, including opus, Sonnet 125 00:07:31,720 --> 00:07:34,840 S1: and Haiku. AGI bot is a Chinese company and they 126 00:07:34,840 --> 00:07:38,920 S1: just unveiled unveiled a fleet of advanced humanoid robots to 127 00:07:38,950 --> 00:07:43,150 S1: compete directly with Optimus, which is the one from Tesla, 128 00:07:43,150 --> 00:07:46,900 S1: and they're designed for tasks ranging from household chores to 129 00:07:46,930 --> 00:07:50,410 S1: industrial operations. And they're going to start shipping supposedly by 130 00:07:50,410 --> 00:07:53,590 S1: the end of this year. So like immediately. And Optimus 131 00:07:53,620 --> 00:07:57,960 S1: is nowhere near ready for that kind of timeline. So 132 00:07:57,960 --> 00:08:02,910 S1: I'm basically anti-Chinese imports for both robotaxis and humanoid robots, 133 00:08:02,910 --> 00:08:07,740 S1: because China is too far ahead and they're too cheap 134 00:08:07,740 --> 00:08:10,800 S1: and I would say just too good. So I don't 135 00:08:10,800 --> 00:08:12,960 S1: want to give them a head start. And I don't 136 00:08:12,960 --> 00:08:16,740 S1: like being anti-competitive against any sort of country. I don't 137 00:08:16,740 --> 00:08:19,530 S1: like slowing pressure from the outside, but if this were 138 00:08:19,530 --> 00:08:22,560 S1: India or Ireland, I would actually be okay with them 139 00:08:22,560 --> 00:08:25,350 S1: applying pressure to the US but not China because they're 140 00:08:25,350 --> 00:08:28,830 S1: too obviously a malicious actor who actually just like, wants 141 00:08:28,830 --> 00:08:31,950 S1: to crush the United States in like all aspects, including 142 00:08:31,950 --> 00:08:34,920 S1: the United States going away as an economic power. So 143 00:08:34,920 --> 00:08:37,859 S1: it's not like friendly competition. So I think we should 144 00:08:37,890 --> 00:08:41,280 S1: actually just either tax the hell out of them or 145 00:08:41,280 --> 00:08:44,010 S1: not allow them to function until until we have proper 146 00:08:44,010 --> 00:08:47,280 S1: footing and like, can compete properly. And speaking of that, 147 00:08:47,280 --> 00:08:51,030 S1: Tesla is hiring people to train its Optimus humanoid robot 148 00:08:51,030 --> 00:08:55,100 S1: by wearing a motion capture suit and mimicking the different actions. 149 00:08:55,100 --> 00:08:57,679 S1: You get like $48 an hour, but you got to 150 00:08:57,710 --> 00:09:01,160 S1: walk over seven hours a day carrying £30 while wearing 151 00:09:01,160 --> 00:09:04,340 S1: a VR headset. That's a tough job. Waymo is looking 152 00:09:04,340 --> 00:09:08,150 S1: to launch a subscription service called Waymo Teen, so this 153 00:09:08,150 --> 00:09:11,900 S1: is basically to help parents not have to shuffle kids around. 154 00:09:11,929 --> 00:09:15,020 S1: Although I'm not sure, depending on the age of the teen, 155 00:09:15,020 --> 00:09:17,840 S1: should they have their own car, I'm not sure. But anyway, 156 00:09:17,900 --> 00:09:23,780 S1: cool idea. I scientists developed by the University of British Columbia, 157 00:09:23,809 --> 00:09:30,110 S1: Oxford and Sakana AI is creating its own machine learning experiments. Okay, 158 00:09:30,140 --> 00:09:34,309 S1: let's back up. An AI scientist is creating its own 159 00:09:34,309 --> 00:09:37,849 S1: machine learning experiments and running them autonomously. I think this 160 00:09:37,850 --> 00:09:40,880 S1: is where most innovation will come from. AI not just 161 00:09:40,880 --> 00:09:44,179 S1: implementing tasks, but in doing new research. And I talked 162 00:09:44,179 --> 00:09:48,260 S1: about it in a post there. Victor Miller, a mayoral 163 00:09:48,260 --> 00:09:51,350 S1: candidate in Wyoming's capital city, has vowed to let its 164 00:09:51,350 --> 00:09:56,679 S1: customized ChatGPT named Vic Virtual Integrated Citizen help run the 165 00:09:56,679 --> 00:09:59,950 S1: local government. I'm actually working on how to articulate a 166 00:09:59,950 --> 00:10:03,910 S1: political platform for any level of office using substrate. Basically, 167 00:10:03,910 --> 00:10:06,400 S1: define exactly what you want to do, how it branches 168 00:10:06,400 --> 00:10:10,510 S1: out with problem strategies, most importantly KPIs and promises. So 169 00:10:10,510 --> 00:10:12,670 S1: you could literally say, like I talked about it in 170 00:10:12,670 --> 00:10:15,610 S1: the substrate video, you could say, look, here's how I'm 171 00:10:15,610 --> 00:10:18,850 S1: measuring myself. Here are my assumptions. Here's what I believe 172 00:10:18,850 --> 00:10:23,080 S1: the problems are. Here are my strategies for attacking those problems. 173 00:10:23,080 --> 00:10:27,070 S1: And here are my specific projects that I'm going to 174 00:10:27,070 --> 00:10:30,400 S1: do to implement those strategies. And here's how much it's 175 00:10:30,400 --> 00:10:34,330 S1: going to cost. Here's how I'm measuring myself. And here's 176 00:10:34,330 --> 00:10:37,840 S1: the promise if I don't move these numbers by X amount, 177 00:10:37,840 --> 00:10:39,970 S1: you should fire me in two years or in four 178 00:10:39,970 --> 00:10:42,790 S1: years or whatever the term is. So I think this 179 00:10:42,790 --> 00:10:48,130 S1: is where leadership is heading. Transparent descriptions of vision, strategy, KPIs, 180 00:10:48,130 --> 00:10:54,280 S1: and promises. Sean Ammirati Arathi, a professor at Carnegie Mellon, 181 00:10:54,309 --> 00:10:58,210 S1: noticed a massive up leveling of progress in his entrepreneurship 182 00:10:58,210 --> 00:11:01,570 S1: class this year, thanks to generative AI tools like ChatGPT, 183 00:11:01,900 --> 00:11:06,520 S1: GitHub Copilot, and Flow wise, AI. So they basically the 184 00:11:06,520 --> 00:11:09,970 S1: students use these tools for marketing, coding, product development, and 185 00:11:09,970 --> 00:11:14,079 S1: recruiting early customers. This is what I've been talking about 186 00:11:14,080 --> 00:11:17,500 S1: with AI automation. If you were competing with a 95 187 00:11:17,530 --> 00:11:20,530 S1: out of 100 person before and they were a 95 188 00:11:20,559 --> 00:11:23,410 S1: because they went to CMU, well, now you're competing with 189 00:11:23,410 --> 00:11:26,350 S1: a 130 out of 100 because they went to CMU 190 00:11:26,350 --> 00:11:29,230 S1: and they're using AI for everything. And for me, in 191 00:11:29,230 --> 00:11:32,050 S1: my own life, I read better articles because of AI. 192 00:11:32,080 --> 00:11:35,110 S1: I get better ideas because of AI. Therefore I build 193 00:11:35,110 --> 00:11:37,990 S1: better stuff because of AI. And this is all feeding 194 00:11:37,990 --> 00:11:41,200 S1: on itself, just okay. And this all feeds itself and 195 00:11:41,200 --> 00:11:44,020 S1: just makes that whole top of funnel even better. And 196 00:11:44,020 --> 00:11:47,980 S1: I believe that your options are to upgrade or lose. Basically, 197 00:11:48,010 --> 00:11:51,540 S1: GM is cutting over 1000 software engineers to streamline its 198 00:11:51,540 --> 00:11:57,300 S1: software services organization, streamlining by cutting out 1000 devs. The 199 00:11:57,300 --> 00:12:00,630 S1: way I see this is they're actually just taking everyone 200 00:12:00,630 --> 00:12:03,210 S1: out because they're like, this is a huge waste of time. 201 00:12:03,210 --> 00:12:06,270 S1: We hired a bunch of duds. Let's start from scratch 202 00:12:06,270 --> 00:12:09,179 S1: and only hire the best possible people who are probably 203 00:12:09,179 --> 00:12:12,300 S1: also massively augmented with AI. Yeah, so I got a 204 00:12:12,300 --> 00:12:15,540 S1: whole bunch of other content killer cult members. This is 205 00:12:15,540 --> 00:12:19,650 S1: my new, um, way of framing. This whole thing is like, 206 00:12:19,650 --> 00:12:23,070 S1: killer cult members is what companies are looking for. It's 207 00:12:23,070 --> 00:12:25,980 S1: kind of obvious that this is what startups are looking for, 208 00:12:25,980 --> 00:12:28,140 S1: but I think corporations are going to look for this 209 00:12:28,140 --> 00:12:31,380 S1: as well. So people talk about, oh, toxic work culture. 210 00:12:31,410 --> 00:12:35,280 S1: Guess what? Companies only want employees who have a toxic 211 00:12:35,280 --> 00:12:38,610 S1: work culture. They want people who show up and are like, 212 00:12:38,610 --> 00:12:42,180 S1: I am religiously dedicated to this. I will sleep under 213 00:12:42,179 --> 00:12:44,550 S1: my desk. I will work as much time as I 214 00:12:44,550 --> 00:12:48,030 S1: need to. I am fully dedicated to this mission. I 215 00:12:48,059 --> 00:12:51,140 S1: wear all the swag in public. I talk about this. 216 00:12:51,140 --> 00:12:53,630 S1: I think about this all day long. That is what 217 00:12:53,630 --> 00:12:56,689 S1: people want. That is what hiring managers want. That is 218 00:12:56,690 --> 00:13:02,449 S1: what corporations want. Because that behavior, especially if you're on site, 219 00:13:02,450 --> 00:13:05,750 S1: which everyone is pulling everyone back to be on prem now, 220 00:13:05,750 --> 00:13:08,780 S1: if you are surrounded by other cult members with that 221 00:13:08,780 --> 00:13:11,900 S1: sort of energy and keep in mind, this also has 222 00:13:11,900 --> 00:13:16,100 S1: toxic aspects to it. There are massive downsides to this 223 00:13:16,100 --> 00:13:20,420 S1: type of culture, but the downsides are not to the company. Really. 224 00:13:20,420 --> 00:13:24,860 S1: The downsides are mostly to the employees themselves as a 225 00:13:24,860 --> 00:13:28,160 S1: trade off for doing this much risk to get much, 226 00:13:28,309 --> 00:13:32,090 S1: a lot of a reward, potentially in equity or pay 227 00:13:32,120 --> 00:13:36,620 S1: or whatever. So this culture is good for companies and 228 00:13:36,620 --> 00:13:39,140 S1: that's why they are fostering it. That's why they're saying 229 00:13:39,140 --> 00:13:42,590 S1: you must be on site. So for example, OpenAI OpenAI 230 00:13:42,620 --> 00:13:44,630 S1: requires you to be on site. You have to work 231 00:13:44,630 --> 00:13:48,290 S1: on site. There are some exceptions, very few. But in 232 00:13:48,290 --> 00:13:52,339 S1: general they make a cool office. In a cool city, 233 00:13:52,340 --> 00:13:55,160 S1: they require the best. They hire the best talent. They 234 00:13:55,160 --> 00:13:59,449 S1: basically tell you, or it's naturally implied that you must 235 00:13:59,450 --> 00:14:02,180 S1: be one of these killer cult members. And then that 236 00:14:02,179 --> 00:14:06,710 S1: is that energy which is synergistic with itself, builds and 237 00:14:06,710 --> 00:14:09,950 S1: builds and builds and it produces the best products that 238 00:14:09,950 --> 00:14:12,860 S1: people want to buy, and they get massive valuations. And 239 00:14:12,890 --> 00:14:18,110 S1: that's capitalism and that's what works. What doesn't work is, okay, 240 00:14:18,110 --> 00:14:21,980 S1: let's not do that. Let's have a company culture that's 241 00:14:21,980 --> 00:14:25,880 S1: good for employees, and let's make sure there's work life balance. 242 00:14:25,880 --> 00:14:29,180 S1: And let's make sure, like all these different policies and 243 00:14:29,180 --> 00:14:32,720 S1: it ends up with a lot of people not working, 244 00:14:32,720 --> 00:14:36,140 S1: not doing good work. It ends up with A-players hiring 245 00:14:36,170 --> 00:14:40,070 S1: B players. B players hire C players. You end up 246 00:14:40,070 --> 00:14:43,130 S1: with a bunch of C players, a few B's and 247 00:14:43,280 --> 00:14:46,790 S1: and some D players. And after a while, the corporation 248 00:14:46,790 --> 00:14:52,090 S1: looks at their entire headcount, spend their entire human resources spend, 249 00:14:52,090 --> 00:14:54,370 S1: and they're like, I'm not getting value from this. This 250 00:14:54,370 --> 00:14:59,140 S1: is not worth it. And they just fire everyone. And 251 00:14:59,140 --> 00:15:02,110 S1: so when I see somebody as firing a thousand software 252 00:15:02,110 --> 00:15:05,170 S1: engineers to streamline the software, I don't know exactly what's 253 00:15:05,170 --> 00:15:08,200 S1: happening here. I'm arguing that what I'm saying might be 254 00:15:08,230 --> 00:15:11,890 S1: happening here is definitely happening all over the place. And 255 00:15:11,890 --> 00:15:14,170 S1: what they'll do is they'll go to zero. Then they'll 256 00:15:14,170 --> 00:15:18,220 S1: find a players who are killer cult members and only 257 00:15:18,220 --> 00:15:20,860 S1: hire those from now on. And here's the crazy part. 258 00:15:20,890 --> 00:15:25,060 S1: 100 of those people might be worth 1000 or 10,000 259 00:15:25,090 --> 00:15:28,630 S1: C players who are interested in work life balance. Again, 260 00:15:28,630 --> 00:15:32,050 S1: I'm not saying anything about like the total value of 261 00:15:32,050 --> 00:15:35,350 S1: on society of work life balance or all of that. 262 00:15:35,350 --> 00:15:39,040 S1: I've got completely separate ideas about that. Actually, in the 263 00:15:39,040 --> 00:15:41,500 S1: essay We were Lied To About Work, you can see 264 00:15:41,500 --> 00:15:44,200 S1: what I really feel about that whole thing. The point is, 265 00:15:44,200 --> 00:15:46,690 S1: this is what companies are looking for. If you want 266 00:15:46,720 --> 00:15:49,330 S1: to get hired, you have the answer. Meta is using 267 00:15:49,360 --> 00:15:52,870 S1: AI to streamline system reliability investigations with a new root 268 00:15:52,870 --> 00:15:57,010 S1: cause analysis system. System combines heuristic based retrieval and large 269 00:15:57,010 --> 00:16:02,230 S1: language model based ranking, achieving 42% accuracy in identifying root 270 00:16:02,230 --> 00:16:06,310 S1: causes at the investigation start. I didn't look to see 271 00:16:06,340 --> 00:16:09,250 S1: or it wasn't in there, how that compares to humans 272 00:16:09,250 --> 00:16:12,760 S1: and how fast, because that's the trick. Accuracy and speed 273 00:16:12,760 --> 00:16:15,940 S1: and of course, cost. AI companies are shifting focus from 274 00:16:15,940 --> 00:16:20,410 S1: creating godlike AI to building practical products. Who knew? So 275 00:16:20,410 --> 00:16:22,150 S1: I don't think this is a bubble pop. I think 276 00:16:22,150 --> 00:16:25,240 S1: it's a natural maturity of brand new tech that just 277 00:16:25,240 --> 00:16:28,120 S1: came out. And because people are still figuring this stuff 278 00:16:28,120 --> 00:16:31,120 S1: out and it's basically day one, like AI hasn't even 279 00:16:31,120 --> 00:16:34,330 S1: gotten good yet, hasn't even started to get good. Canada 280 00:16:34,330 --> 00:16:40,690 S1: is slapping 100% import tariffs on Chinese electric vehicles starting 281 00:16:40,720 --> 00:16:44,290 S1: October 1st. We were just talking about that, former Google 282 00:16:44,320 --> 00:16:48,330 S1: CEO Eric Schmidt predicts rapid advancements in AI, the potential 283 00:16:48,330 --> 00:16:52,530 S1: to create significant apps like TikTok competitors in minutes within 284 00:16:52,530 --> 00:16:55,140 S1: the next few years. I know what he's saying, but 285 00:16:55,140 --> 00:16:57,840 S1: there's a difference between being able to run an app 286 00:16:57,840 --> 00:17:01,590 S1: at scale versus creating it. Of course, he knows that 287 00:17:01,590 --> 00:17:04,950 S1: better than I do or most people. But important point 288 00:17:04,950 --> 00:17:09,149 S1: to make. Claude 3.5 can now create icalendar files from images. 289 00:17:09,150 --> 00:17:12,750 S1: And this guy Greg's ramblings shows how you can use 290 00:17:12,750 --> 00:17:16,440 S1: this feature to generate calendar entries by snapping a photo 291 00:17:16,440 --> 00:17:20,940 S1: of a schedule or event flyer. AWS CEO Adam Selipsky 292 00:17:20,970 --> 00:17:25,110 S1: predicts that within the next 24 months, most developers might 293 00:17:25,109 --> 00:17:28,890 S1: not be coding anymore due to AI advancements. He says 294 00:17:28,890 --> 00:17:33,300 S1: the real skill shift will be towards innovation and understanding 295 00:17:33,300 --> 00:17:39,030 S1: customer needs, rather than writing code. 100% agree. Although most 296 00:17:39,030 --> 00:17:42,480 S1: developers in 24 months. This is the type of thing 297 00:17:42,480 --> 00:17:45,920 S1: where it happens way slower than you think and then 298 00:17:45,950 --> 00:17:48,800 S1: way faster than you think. At the same time, most 299 00:17:48,800 --> 00:17:52,520 S1: in 24 months will not be coding anymore. That is 300 00:17:52,520 --> 00:17:56,750 S1: wildly off, wildly off. But people who are like, oh, 301 00:17:56,780 --> 00:17:58,730 S1: it's going to take forever. It's going to take, you know, 302 00:17:58,760 --> 00:18:01,939 S1: three years, five years, seven years, ten years, 15 years. 303 00:18:01,940 --> 00:18:05,659 S1: They are also wildly off. Chinese companies have ramped up 304 00:18:05,660 --> 00:18:11,659 S1: imports of chip production equipment. $26 billion in the first 305 00:18:11,660 --> 00:18:15,530 S1: seven months of 24 on chip production equipment. They need 306 00:18:15,530 --> 00:18:20,240 S1: to equip 18 new fabs expected to start operations in 24. 307 00:18:20,570 --> 00:18:23,120 S1: There they are all in on this because they see 308 00:18:23,119 --> 00:18:26,359 S1: the world turning against them and shutting them down. So 309 00:18:26,359 --> 00:18:29,570 S1: they are all in on this. Question is, can they actually, 310 00:18:29,570 --> 00:18:33,560 S1: even with all the fabs and the equipment and you know, 311 00:18:33,590 --> 00:18:36,500 S1: the factories, do they have the know how to actually 312 00:18:36,500 --> 00:18:40,580 S1: do what TSMC is doing? My current understanding from current 313 00:18:40,609 --> 00:18:44,300 S1: knowledge combined with like the book chip. Wars know they 314 00:18:44,300 --> 00:18:47,230 S1: don't have the knowledge. So even with all that stuff, 315 00:18:47,230 --> 00:18:50,170 S1: it's not going to be enough humans. Cisco's laying off 7% 316 00:18:50,170 --> 00:18:54,040 S1: of its workforce, around 6000 employees, as it pivots towards 317 00:18:54,070 --> 00:18:58,209 S1: AI and cybersecurity. McKinsey's new study reveals that business leaders 318 00:18:58,210 --> 00:19:01,810 S1: are missing the mark on why employees are quitting. They 319 00:19:01,840 --> 00:19:06,970 S1: say companies are focusing on transactional perks like compensation and flexibility, 320 00:19:06,970 --> 00:19:12,909 S1: but employees are actually seeking meaning. Belonging, holistic care and 321 00:19:12,910 --> 00:19:17,230 S1: appreciation at work couldn't have been better timed with this 322 00:19:17,230 --> 00:19:22,660 S1: week's essay. 24 brain samples collected in early 2424 brain 323 00:19:22,660 --> 00:19:29,440 S1: samples in 2024 measured an average of 0.5% plastic by weight. 324 00:19:29,440 --> 00:19:33,490 S1: So a brain is multiple pounds. Is it £3 or £2? 325 00:19:33,490 --> 00:19:37,030 S1: I can't remember 3 or £4, something like that. It's heavy. 326 00:19:37,060 --> 00:19:39,430 S1: You can feel this thing right. You can feel your head. 327 00:19:39,430 --> 00:19:43,930 S1: It's heavy. Half a percent. Okay. Let's just type in here. 328 00:19:43,960 --> 00:19:46,120 S1: What's the half a percent of the weight of an 329 00:19:46,119 --> 00:19:49,990 S1: average brain? 1400 grams. Half a percent of that would 330 00:19:49,990 --> 00:19:56,230 S1: be seven grams. Okay, I put 25g of coffee to 331 00:19:56,260 --> 00:20:00,070 S1: brew when I brew coffee, so I roughly feel like 332 00:20:00,070 --> 00:20:02,950 S1: I know that's about a third, about a third of 333 00:20:02,950 --> 00:20:05,020 S1: the amount of coffee that I drink in the morning, 334 00:20:05,020 --> 00:20:08,260 S1: which when you brew that thing, I mean, it's significant. 335 00:20:08,260 --> 00:20:13,750 S1: This is a extremely non-trivial amount of plastic sitting inside 336 00:20:13,750 --> 00:20:16,480 S1: the brain. And there's a lot of speculation right now 337 00:20:16,480 --> 00:20:20,080 S1: which I don't put too much weight into it because, 338 00:20:20,109 --> 00:20:22,420 S1: you know, how we have these health scares and like, oh, 339 00:20:22,450 --> 00:20:25,690 S1: this is dangerous and that's dangerous or whatever, but seven 340 00:20:25,690 --> 00:20:31,510 S1: grams of plastic, seven grams of plastic sitting inside of 341 00:20:31,510 --> 00:20:34,869 S1: a brain. I feel like it can't be good unless 342 00:20:34,869 --> 00:20:38,199 S1: it's like alien plastic that is like nanobots or something. 343 00:20:38,200 --> 00:20:41,260 S1: But no, this is regular plastic. The question is, where 344 00:20:41,290 --> 00:20:43,210 S1: is it coming from? How is it getting in there? 345 00:20:43,210 --> 00:20:45,660 S1: Is it in all our foods? Is it in? Is 346 00:20:45,660 --> 00:20:48,600 S1: it from drinking bottles? I'm actually super concerned about this 347 00:20:48,600 --> 00:20:51,510 S1: because I drink energy drinks. That's why I switched to 348 00:20:51,540 --> 00:20:53,609 S1: these because I think they have less. But I saw 349 00:20:53,609 --> 00:20:58,140 S1: a crazy report about my favorite not energy drinks, protein drinks, 350 00:20:58,140 --> 00:21:01,740 S1: my favorite protein drink. It was the core power one, 351 00:21:01,740 --> 00:21:04,560 S1: and it was like off the charts in the amount 352 00:21:04,590 --> 00:21:07,350 S1: of plastic, according to this one lab that ran it. 353 00:21:07,380 --> 00:21:10,050 S1: Now who knows, maybe that lab was run by the 354 00:21:10,050 --> 00:21:12,930 S1: counter product, which I went out and bought, by the way, 355 00:21:12,930 --> 00:21:16,800 S1: which is Muscle Milk. Anyway, I'm concerned about this, but 356 00:21:16,800 --> 00:21:19,170 S1: I'm also cautious because you don't know if it's a 357 00:21:19,170 --> 00:21:23,130 S1: scare turn. That plastic could be completely inert and doesn't 358 00:21:23,130 --> 00:21:26,580 S1: even matter. And the thing that's causing all this cancer 359 00:21:26,580 --> 00:21:30,630 S1: and drop in testosterone and all this stuff is actually 360 00:21:30,630 --> 00:21:32,880 S1: something else could be our sun exposure, which I think 361 00:21:32,880 --> 00:21:36,120 S1: is a later. Yeah, it's two stories down. So anyway, 362 00:21:36,119 --> 00:21:39,450 S1: lots of plastic in our brains. Not sure what that 363 00:21:39,450 --> 00:21:43,820 S1: means yet. Gallup has released its 2023 Global Emotions Report, 364 00:21:43,820 --> 00:21:48,139 S1: which measures the world's emotional temperature through Positive Experience Index. 365 00:21:48,140 --> 00:21:50,959 S1: I'm opening this one because it is cool. Look at this. 366 00:21:50,990 --> 00:21:55,159 S1: Experienced anger. Look at this. You got these country breakdowns. 367 00:21:55,160 --> 00:21:57,650 S1: This is really cool. Then you got map views. And 368 00:21:57,650 --> 00:22:00,050 S1: look at this. The map view. You can click on 369 00:22:00,050 --> 00:22:04,340 S1: sadness stress worry pain. Look at that enjoyment. Okay. So 370 00:22:04,369 --> 00:22:10,460 S1: dark is better right? Um. China. Yes. Yeah. Dark. Yep. Okay, 371 00:22:10,490 --> 00:22:14,720 S1: so let's look at a light one. Yes, 100% dark 372 00:22:14,720 --> 00:22:19,369 S1: is better because that's Afghanistan. Two thirds. No. Oh, man, 373 00:22:19,400 --> 00:22:24,379 S1: that's so depressing. Afghanistan has two thirds. No enjoyment. However, 374 00:22:24,380 --> 00:22:28,040 S1: the question was asked another one with that similar color turkey. 375 00:22:28,070 --> 00:22:31,310 S1: I'm pretty sure that's turkey. Yeah that's Turkey. Uh, Ukraine. 376 00:22:31,310 --> 00:22:35,570 S1: About half, for obvious reasons. What's this one? Morocco half. 377 00:22:35,600 --> 00:22:40,340 S1: What's this one? Tunisia half. Roughly super happy over here. 378 00:22:40,340 --> 00:22:45,950 S1: Somalia Why is Somalia so happy? See, I love these visualizations. Uzbekistan. 379 00:22:45,950 --> 00:22:48,889 S1: Very happy. What have we got over here? Norway is happy. 380 00:22:48,920 --> 00:22:51,439 S1: Keep forgetting. Norway is the far left one. What is 381 00:22:51,440 --> 00:22:54,980 S1: this one? Over here. Estonia. Very happy. Iceland could have 382 00:22:54,980 --> 00:23:00,050 S1: guessed that one. Mexico! Kicking total ass over here. Ireland seems. Yeah. 383 00:23:00,080 --> 00:23:04,850 S1: Really happy. UK? Not so much. What are these? Over here. Indonesia. Indonesia. 384 00:23:04,880 --> 00:23:10,850 S1: Very happy. Malaysia. Very happy anyway. Really cool stuff. Russian Federation. 385 00:23:10,880 --> 00:23:14,000 S1: Pretty low score, actually. I think this is the actual 386 00:23:14,000 --> 00:23:18,199 S1: worst one. Afghanistan, which is totally explainable. I mean, so 387 00:23:18,200 --> 00:23:20,780 S1: this is one experience I've been having is I take 388 00:23:20,780 --> 00:23:23,149 S1: a decent number of Ubers and I always talk to 389 00:23:23,180 --> 00:23:26,750 S1: the person in the Bay area. My chances of getting 390 00:23:26,750 --> 00:23:31,190 S1: someone from Afghanistan are like 80, 90%, and I talk 391 00:23:31,190 --> 00:23:33,800 S1: to every single one of them for the entire duration 392 00:23:33,800 --> 00:23:38,270 S1: of the trip. Usually they are interpreters, don't say translator, 393 00:23:38,270 --> 00:23:41,800 S1: they are interpreters. And oftentimes they worked for the US government, 394 00:23:41,800 --> 00:23:43,900 S1: which is how they got their visa to come over 395 00:23:43,900 --> 00:23:47,710 S1: here and what they have back home if they brought 396 00:23:47,710 --> 00:23:51,100 S1: their family. Most likely they didn't bring most of their family. 397 00:23:51,100 --> 00:23:54,340 S1: Their family is in danger because they are here and 398 00:23:54,340 --> 00:23:58,000 S1: it is going from bad to worse every single week. 399 00:23:58,030 --> 00:24:01,960 S1: It just is so, so depressing. Anyway, I'm on a 400 00:24:01,960 --> 00:24:05,770 S1: tangent here, but talk to your Uber drivers. Experience life 401 00:24:05,770 --> 00:24:11,080 S1: through other people's eyes. Okay. Um. Data from surveys conducted 402 00:24:11,080 --> 00:24:14,530 S1: in 142 countries. Mix of telephone, face to face, and 403 00:24:14,530 --> 00:24:18,400 S1: some web stuff. About a thousand respondents per country. So 404 00:24:18,400 --> 00:24:20,950 S1: that's not great, but I'm sure they're doing it scientifically, 405 00:24:20,950 --> 00:24:25,270 S1: so it's a decent sample. Non-smokers who avoided the sun 406 00:24:25,300 --> 00:24:28,870 S1: had a life expectancy similar to smokers who got the 407 00:24:28,869 --> 00:24:33,639 S1: most sun. And this is nearly 30,000 Swedish women over 408 00:24:33,640 --> 00:24:37,960 S1: 20 years, by the way, Scandinavia likes to smoke. And Germany, 409 00:24:37,960 --> 00:24:40,169 S1: this is the worst thing about Europe and there are 410 00:24:40,170 --> 00:24:42,720 S1: many things competing for that right now. But one of 411 00:24:42,720 --> 00:24:45,240 S1: the worst things about Europe is smoking. I go there 412 00:24:45,240 --> 00:24:48,960 S1: and I'm just like, what is going on? Yeah, Switzerland. 413 00:24:48,990 --> 00:24:50,820 S1: I'm about to go back to Switzerland and everyone's going 414 00:24:50,850 --> 00:24:54,630 S1: to be smoking inside the restaurant. They're smoking inside the restaurant. 415 00:24:54,630 --> 00:24:56,820 S1: I'm just like, what is? I thought you guys were 416 00:24:56,820 --> 00:25:00,300 S1: the advanced group. Okay, the research suggests that avoiding sun 417 00:25:00,300 --> 00:25:04,919 S1: is as risky as smoking. So this needs more research, obviously. 418 00:25:04,920 --> 00:25:07,860 S1: But like I said, damn. For me, I get sun 419 00:25:07,859 --> 00:25:10,680 S1: in the morning. I got decent amount of sun this morning. 420 00:25:10,680 --> 00:25:14,550 S1: Massive boost for me. I put on the Waking Up app, 421 00:25:14,580 --> 00:25:17,550 S1: listen to Sam do a ten minute thing. I do 422 00:25:17,550 --> 00:25:19,860 S1: a ten minute walk out there, I do some breathing. 423 00:25:19,859 --> 00:25:22,170 S1: That's how I start my day. When I'm on a routine, 424 00:25:22,170 --> 00:25:27,270 S1: which I should be and often not enough. But I 425 00:25:27,270 --> 00:25:29,850 S1: did today and yesterday and the day before. All right. 426 00:25:29,880 --> 00:25:34,080 S1: Stanford researchers have found that blocking the whatever blocking this 427 00:25:34,080 --> 00:25:38,129 S1: pathway in the brain can reverse the metabolic disruptions caused 428 00:25:38,130 --> 00:25:43,770 S1: by Alzheimer's disease, providing cognitive improving cognitive functions in mice. 429 00:25:43,770 --> 00:25:46,920 S1: I'm starting to feel like we're about to make massive 430 00:25:46,920 --> 00:25:51,090 S1: progress on both Alzheimer's and cancer. And honestly, it's making 431 00:25:51,090 --> 00:25:52,980 S1: me want to invest in like, the top three drug 432 00:25:52,980 --> 00:25:56,250 S1: companies I think I'm already in, which one I'm in, 433 00:25:56,250 --> 00:25:58,620 S1: the one that does wegovy. I can't remember the name 434 00:25:58,619 --> 00:26:01,710 S1: of that one. I'll think of it. It's right there. 435 00:26:01,710 --> 00:26:04,440 S1: It's right there. Anyway, I'm in that one, but I'm 436 00:26:04,440 --> 00:26:07,139 S1: not in the Lilly one. I think they're called Eli Lilly, 437 00:26:07,140 --> 00:26:10,080 S1: but they're rebranding to Lilly. So I figure if I 438 00:26:10,080 --> 00:26:13,470 S1: get into the two top competitors, one or both of 439 00:26:13,470 --> 00:26:16,859 S1: them is going to do something with Alzheimer's and cancer, 440 00:26:16,859 --> 00:26:21,899 S1: and we're already doing it with obesity. Like, that's a trifecta. 441 00:26:21,930 --> 00:26:25,560 S1: All we need now is like balding and aging. Aging 442 00:26:25,560 --> 00:26:27,600 S1: is the big one. Aging and cancer I would say, 443 00:26:27,630 --> 00:26:31,800 S1: are the big ones. And then like obesity and hair loss, 444 00:26:31,800 --> 00:26:36,300 S1: that's amazing. So not investment advice, but I'm damn sure 445 00:26:36,330 --> 00:26:40,760 S1: getting in all right. Air purifiers and two Helsinki daycare 446 00:26:40,760 --> 00:26:45,440 S1: centers reduced sick kids days by 30%. And I don't 447 00:26:45,440 --> 00:26:48,590 S1: think this is Covid or flu or anything. I think 448 00:26:48,590 --> 00:26:51,290 S1: they're just talking about like, all cause based on the 449 00:26:51,290 --> 00:26:55,129 S1: the study parts that I saw, University of Missouri scientists 450 00:26:55,130 --> 00:26:59,240 S1: have developed a liquid based solution that removes over 98% 451 00:26:59,240 --> 00:27:02,930 S1: of nanoplastics from water. It uses water repelling solvents to 452 00:27:02,960 --> 00:27:07,189 S1: absorb the particles, which are easily separated and removed. I 453 00:27:07,190 --> 00:27:09,110 S1: assume you just run it through a filter and those 454 00:27:09,109 --> 00:27:12,500 S1: things will be big globs, and they get stopped behind 455 00:27:12,500 --> 00:27:14,990 S1: by the filter. I expect to see lots more of this. 456 00:27:15,020 --> 00:27:17,930 S1: Can't wait for the Huberman episode on this, because I 457 00:27:17,930 --> 00:27:20,930 S1: wanted to be able to do this like cheaply at home. 458 00:27:20,930 --> 00:27:25,070 S1: Eli Lilly's weight loss drug Tripeptide, found in Zepp bound 459 00:27:25,100 --> 00:27:29,359 S1: and Manjaro, reduced the risk of developing type two diabetes 460 00:27:29,359 --> 00:27:36,649 S1: by 94% in obese or overweight adults with pre-diabetes, 94%. 461 00:27:36,650 --> 00:27:39,790 S1: And Apple Podcasts is losing ground to YouTube and Spotify, 462 00:27:39,790 --> 00:27:46,600 S1: so recent study put YouTube at 31%, Spotify at 21%, 463 00:27:46,600 --> 00:27:50,050 S1: and Apple Podcasts at 12. I don't do Spotify, I 464 00:27:50,050 --> 00:27:54,699 S1: do YouTube and Apple Podcasts. But all right. Ideas thought 465 00:27:54,700 --> 00:27:58,360 S1: of a cool idea for fabric, Telos and substrate. Maintain 466 00:27:58,359 --> 00:28:01,210 S1: a list of everything I've been really, really wrong about, 467 00:28:01,210 --> 00:28:03,429 S1: which I'm already building that list, and then write a 468 00:28:03,430 --> 00:28:06,070 S1: fabric pattern that looks at that list. By the way, 469 00:28:06,070 --> 00:28:07,600 S1: I'm just going to have it all in my same 470 00:28:07,630 --> 00:28:10,600 S1: Telos file as I talked about in the augmented course, 471 00:28:10,600 --> 00:28:15,700 S1: but basically just have this list listed as all my 472 00:28:15,700 --> 00:28:20,200 S1: biggest mistakes, cognitive errors or whatever listed in there. Then 473 00:28:20,200 --> 00:28:23,290 S1: I also have another section inside that same telos file, 474 00:28:23,290 --> 00:28:25,869 S1: which is like all my current beliefs, my model of 475 00:28:25,869 --> 00:28:29,470 S1: the world. And then the fabric pattern basically says evaluate 476 00:28:29,470 --> 00:28:33,160 S1: all my current beliefs, look at all the previous mistakes 477 00:28:33,160 --> 00:28:36,670 S1: that I've made and look for patterns. Look for what 478 00:28:36,670 --> 00:28:40,390 S1: in my current beliefs might be broken in a similar 479 00:28:40,390 --> 00:28:43,030 S1: way as the way I was wrong about the other things. 480 00:28:43,030 --> 00:28:45,670 S1: First of all, it's going to help me diagnose why 481 00:28:45,670 --> 00:28:47,979 S1: I was wrong. Is it because I have a bias 482 00:28:47,980 --> 00:28:50,590 S1: towards this one thing? It's like, oh, you're so pro 483 00:28:50,590 --> 00:28:54,070 S1: I that you are wrong about this thing because you 484 00:28:54,070 --> 00:28:57,790 S1: thought technology was a solution. So turns out your bias 485 00:28:57,820 --> 00:29:00,670 S1: is thinking that tech can solve too many things in 486 00:29:00,670 --> 00:29:03,970 S1: the world. Good to know. Like I'm already actively defending 487 00:29:03,970 --> 00:29:07,270 S1: against that bias because I know it's there doesn't mean 488 00:29:07,270 --> 00:29:11,470 S1: I'm properly or adequately defending against that bias. Point is, 489 00:29:11,470 --> 00:29:13,150 S1: I want to see the bias. I want to see 490 00:29:13,150 --> 00:29:15,310 S1: it call me out on it, and look for other 491 00:29:15,310 --> 00:29:18,370 S1: evidence that my current beliefs are broken because of that problem. 492 00:29:18,370 --> 00:29:22,450 S1: Discovery Foof I uses fluff and I yeah, that's why 493 00:29:22,450 --> 00:29:25,720 S1: it's called that. To find more web hacking targets by 494 00:29:25,720 --> 00:29:30,700 S1: Joseph Thacker, go fuzzy recursively looks at JavaScript files and 495 00:29:30,700 --> 00:29:34,510 S1: finds endpoints that can be tested. Analyze interviewer techniques is 496 00:29:34,510 --> 00:29:37,660 S1: a new fabric pattern that will capture the je NE 497 00:29:37,660 --> 00:29:40,030 S1: sais quoi. By the way, this is spelled wrong. I 498 00:29:40,030 --> 00:29:42,550 S1: don't know why I didn't just use AI and spell 499 00:29:42,550 --> 00:29:46,480 S1: this correctly. I always spell this wrong. N e is 500 00:29:46,510 --> 00:29:49,690 S1: n a I s and I think the other one 501 00:29:49,690 --> 00:29:52,090 S1: might even be wrong too, but I think the quote 502 00:29:52,120 --> 00:29:55,660 S1: is right. But anyway, it's n I s I believe 503 00:29:55,690 --> 00:29:58,600 S1: is how you spell that. Anyway, someone French gave me 504 00:29:58,600 --> 00:30:02,290 S1: the proper spelling again, more I should be part of this. Um, 505 00:30:02,680 --> 00:30:05,590 S1: back to my bias of I being the problem, but 506 00:30:05,590 --> 00:30:10,780 S1: I've been using this on Dwarkesh and Tyler Cowen content 507 00:30:10,780 --> 00:30:14,680 S1: analyze interviewer techniques. It basically finds out, it figures out 508 00:30:14,680 --> 00:30:17,470 S1: and tells you why they're such a good interviewer. Harness 509 00:30:17,500 --> 00:30:19,420 S1: is a quick tool I put together to test the 510 00:30:19,420 --> 00:30:22,480 S1: efficacy of one prompt versus another. It runs both against 511 00:30:22,480 --> 00:30:25,450 S1: an input and then scores the output according to a 512 00:30:25,450 --> 00:30:28,989 S1: third objective prompt that rates how well they followed the 513 00:30:28,990 --> 00:30:33,880 S1: plot and actually executed the instructions of what the prompts 514 00:30:33,880 --> 00:30:37,340 S1: were trying to do in the first place. So super useful. 515 00:30:37,370 --> 00:30:40,700 S1: Stayed in time are the same thing by Hillel Wayne. 516 00:30:40,700 --> 00:30:43,190 S1: Don't Force yourself to become a bug bounty Hunter by 517 00:30:43,190 --> 00:30:47,540 S1: Sam Curry 67 years of RadioShack catalogs have been scanned 518 00:30:47,540 --> 00:30:51,290 S1: and are now available online. MD RSS is a go 519 00:30:51,320 --> 00:30:54,470 S1: based tool that converts markdown files to RSS feeds. You 520 00:30:54,470 --> 00:30:57,380 S1: can write articles in a local folder, and it automatically 521 00:30:57,380 --> 00:31:01,340 S1: formats them into an RSS compliant XML file. Super cool. 522 00:31:01,340 --> 00:31:05,480 S1: No hello, no quick call, no meetings without an agenda. 523 00:31:05,510 --> 00:31:08,870 S1: You already know that's good. Roger Penrose's book, The Emperor's 524 00:31:08,900 --> 00:31:12,560 S1: New Mind explores the relationship between the human mind and computers, 525 00:31:12,560 --> 00:31:16,880 S1: arguing that human consciousness cannot be replicated by machines. I 526 00:31:16,880 --> 00:31:18,740 S1: have the opposite view, which is why I'm going to 527 00:31:18,740 --> 00:31:21,650 S1: read this book collection of free public APIs that are 528 00:31:21,650 --> 00:31:24,020 S1: tested daily and the recommendation of the week. Take the 529 00:31:24,020 --> 00:31:26,930 S1: time to read this week's main essay. We've been lied 530 00:31:26,960 --> 00:31:28,970 S1: to about work, but more than just read it. Think 531 00:31:28,970 --> 00:31:31,040 S1: about what it means. If I am right, think about 532 00:31:31,040 --> 00:31:33,290 S1: what that means for you and your career, but also 533 00:31:33,290 --> 00:31:35,630 S1: all the young people you know and care about. So 534 00:31:35,630 --> 00:31:37,670 S1: I didn't talk about the solution in the piece, but 535 00:31:37,670 --> 00:31:40,460 S1: it's essentially human 3.0, and I'm going to be talking 536 00:31:40,460 --> 00:31:43,310 S1: a lot more about that, but start thinking about it now. 537 00:31:43,340 --> 00:31:46,640 S1: Definitely recommend that you read this piece. And the aphorism 538 00:31:46,640 --> 00:31:49,070 S1: of the week to fear love is to fear life, 539 00:31:49,070 --> 00:31:52,970 S1: and those who fear life are already three parts dead 540 00:31:53,000 --> 00:31:55,430 S1: to fear. Love is to fear life, and those who 541 00:31:55,430 --> 00:32:01,280 S1: fear life are already three parts dead. Bertrand Russell. Unsupervised 542 00:32:01,280 --> 00:32:03,800 S1: learning is produced and edited by Daniel Miessler on a 543 00:32:03,800 --> 00:32:08,660 S1: Neumann U87 AI microphone using Hindenburg. Intro and outro music 544 00:32:08,660 --> 00:32:11,750 S1: is by zombie with a Y. And to get the 545 00:32:11,750 --> 00:32:13,820 S1: text and links from this episode, sign up for the 546 00:32:13,820 --> 00:32:19,459 S1: newsletter version of the show at Daniel miessler.com/newsletter. We'll see 547 00:32:19,460 --> 00:32:20,240 S1: you next time.