1 00:00:00,140 --> 00:00:03,710 S1: Whether you're starting or scaling your company's security program, demonstrating 2 00:00:03,740 --> 00:00:07,250 S1: top notch security practices and establishing trust is more important 3 00:00:07,250 --> 00:00:12,440 S1: than ever. Vanta automates compliance for Soc2, ISO 27,001 and more, 4 00:00:12,440 --> 00:00:16,430 S1: saving you time and money while helping you build customer trust. Plus, 5 00:00:16,430 --> 00:00:20,690 S1: you can streamline security reviews by automating questionnaires and demonstrating 6 00:00:20,690 --> 00:00:23,810 S1: your security posture with a customer facing trust center, all 7 00:00:23,810 --> 00:00:28,010 S1: powered by advanced AI. Over 7000 global companies like Atlassian, 8 00:00:28,010 --> 00:00:31,400 S1: Flow Health, and Quora use Vanta to manage risk, improve 9 00:00:31,400 --> 00:00:34,880 S1: security in real time. Get $1,000 off Vanta when you 10 00:00:34,880 --> 00:00:42,530 S1: go to Vanta comm slash unsupervised. That's vanta.com/supervised for $1,000 off. 11 00:00:44,000 --> 00:00:47,360 S1: Welcome to Unsupervised Learning, a security, AI and meaning focused 12 00:00:47,360 --> 00:00:50,240 S1: podcast that looks at how best to thrive as humans 13 00:00:50,240 --> 00:00:54,470 S1: in a post AI world. It combines original ideas, analysis, 14 00:00:54,470 --> 00:00:57,710 S1: and mental models to bring not just the news, but 15 00:00:57,710 --> 00:01:05,380 S1: why it matters and how to respond. All right. Welcome 16 00:01:05,380 --> 00:01:11,620 S1: to episode 438. This is Daniel Miessler. Okay. What do 17 00:01:11,620 --> 00:01:18,110 S1: we have here? So registration for augmented v2. July 26th 18 00:01:18,110 --> 00:01:21,590 S1: is now open. About half the slots are about gone. 19 00:01:22,110 --> 00:01:26,460 S1: So sign up by clicking down below in the newsletter. 20 00:01:26,459 --> 00:01:31,410 S1: Or you can just go to augmented dot Unsupervised learning.com 21 00:01:31,410 --> 00:01:37,040 S1: and sign up. And, uh, if you become a UL member, 22 00:01:37,040 --> 00:01:42,710 S1: you actually get $250 off. So inside of the chat, 23 00:01:42,709 --> 00:01:46,010 S1: you can get the discount code, which is $250 off, 24 00:01:46,010 --> 00:01:51,440 S1: and becoming a member is only $99 for the first year. So. 25 00:01:52,820 --> 00:01:56,940 S1: You could do the math there. Let's see here. Fabric 26 00:01:56,940 --> 00:02:01,980 S1: now supports Anthropic's sonnet 3.5, which is really, really good. 27 00:02:01,980 --> 00:02:05,820 S1: So if you update and then you do fabric list models, 28 00:02:06,030 --> 00:02:10,109 S1: it will show you that, uh, cloth 3.5 is in there. 29 00:02:10,110 --> 00:02:13,770 S1: And this is the actual tag for it. Claude dash three, 30 00:02:13,770 --> 00:02:20,010 S1: dash five. Sonnet dash 20240620. Uh, one thing to note, though, 31 00:02:20,010 --> 00:02:23,070 S1: is sonnet and haiku. They throw a lot of copyright 32 00:02:23,070 --> 00:02:27,000 S1: failures like it won't actually read podcast transcripts, which is 33 00:02:27,000 --> 00:02:29,250 S1: super annoying because it's one of my favorite things to 34 00:02:29,250 --> 00:02:33,000 S1: do is put long form conversations through these things and 35 00:02:33,000 --> 00:02:36,600 S1: like pull out like lists of references that they mentioned 36 00:02:36,600 --> 00:02:39,119 S1: if I want to order the books or whatever. So 37 00:02:39,120 --> 00:02:41,010 S1: that's kind of annoying that it doesn't allow me to 38 00:02:41,010 --> 00:02:44,490 S1: do that with one of the lower models. Opus does 39 00:02:44,490 --> 00:02:46,980 S1: allow me to do that, and GPT four does allow 40 00:02:46,980 --> 00:02:49,950 S1: me to do that, but I can't do it in 41 00:02:49,950 --> 00:02:54,170 S1: sonnet or haiku. So I'm really waiting for them to 42 00:02:54,169 --> 00:02:57,740 S1: either fix that with the lower models or to give 43 00:02:57,740 --> 00:03:02,630 S1: me opus 3.5. I cannot wait to use that. Definitely 44 00:03:02,630 --> 00:03:05,300 S1: gearing up for Vegas black Hat Defcon. Hope to see 45 00:03:05,300 --> 00:03:08,060 S1: you there! I'm going to have stickers, but, uh, don't 46 00:03:08,060 --> 00:03:12,110 S1: really have anything else other than stickers. Might eventually do 47 00:03:12,110 --> 00:03:15,410 S1: a t shirt or something with like a superhero logo 48 00:03:15,410 --> 00:03:19,669 S1: of just, uh, the UL logo, but no text or anything. 49 00:03:19,669 --> 00:03:22,579 S1: So I've been feeling quite grateful for my life lately 50 00:03:22,580 --> 00:03:26,570 S1: and somehow, like, super powered. Not really, because, uh, things 51 00:03:26,570 --> 00:03:30,260 S1: are going well, which, uh, they're definitely going pretty well 52 00:03:30,260 --> 00:03:33,169 S1: in terms of, like, the business and everything like that, but. 53 00:03:33,710 --> 00:03:35,870 S1: It's more so that I feel like I'm living an 54 00:03:35,870 --> 00:03:38,990 S1: authentic life. I feel like the stuff that I'm doing 55 00:03:38,990 --> 00:03:41,870 S1: is all oriented in the direction, and I'm working on 56 00:03:41,870 --> 00:03:45,380 S1: a like a essays and a course around this and 57 00:03:45,380 --> 00:03:48,440 S1: stuff like that, but it's essentially like, do you know 58 00:03:48,440 --> 00:03:50,870 S1: what you're trying to accomplish? And are you heading in 59 00:03:50,900 --> 00:03:54,140 S1: that direction? And what I've been trying to do is like, 60 00:03:54,140 --> 00:03:57,500 S1: hack together these systems where I say no to things 61 00:03:57,500 --> 00:04:00,380 S1: that are not in the direction of these goals. And 62 00:04:00,380 --> 00:04:02,480 S1: one thing that came up in a conversation with my 63 00:04:02,480 --> 00:04:06,170 S1: buddy recently is, um, the cool thing about this. Oh, 64 00:04:06,170 --> 00:04:09,710 S1: and I also got this from a David Perell analysis 65 00:04:09,710 --> 00:04:13,910 S1: of a bunch of Seinfeld stuff. It's like fail at 66 00:04:13,910 --> 00:04:17,710 S1: things that you wanted to do anyway. And it really 67 00:04:17,710 --> 00:04:20,950 S1: doesn't feel like failure, right? Because I'm doing a lot 68 00:04:20,950 --> 00:04:24,190 S1: of different stuff. Maybe too much sometimes, sometimes not enough. 69 00:04:24,190 --> 00:04:28,029 S1: But in general, I'm doing lots of different things. And 70 00:04:28,029 --> 00:04:31,360 S1: the question is, how many do you say no to? 71 00:04:31,360 --> 00:04:34,540 S1: And what is your criteria for saying yes or no? 72 00:04:34,570 --> 00:04:39,790 S1: And what I've come to basically figure out is that 73 00:04:39,970 --> 00:04:42,640 S1: if you look at that Seinfeld thing, he basically said, 74 00:04:42,640 --> 00:04:46,570 S1: fail doing it, doing what you want the way that 75 00:04:46,570 --> 00:04:51,159 S1: you want. And this really resonated with me, because I 76 00:04:51,160 --> 00:04:54,310 S1: think about people who are kind of chasing other people. 77 00:04:54,310 --> 00:04:57,099 S1: They're chasing like a high school bully, or they're chasing 78 00:04:57,100 --> 00:05:02,289 S1: like their boss or whatever, some, some competitor for some reason, 79 00:05:02,350 --> 00:05:06,609 S1: and they're chasing them, and they might actually not be 80 00:05:06,610 --> 00:05:11,529 S1: thinking about what really drives them, what purpose they are chasing, 81 00:05:11,529 --> 00:05:15,130 S1: the thing that they actually want to do deeply. That's 82 00:05:15,130 --> 00:05:18,700 S1: part of their own personal identity. Instead, they might be going, well, 83 00:05:18,700 --> 00:05:22,000 S1: the thing I'm doing isn't working. The thing they're doing 84 00:05:22,000 --> 00:05:24,580 S1: is working, so I'm going to copy them. Or if 85 00:05:24,580 --> 00:05:28,330 S1: someone gives them advice and they're like, hey! You know, 86 00:05:28,360 --> 00:05:31,060 S1: this thing that I'm doing works really, really well. You 87 00:05:31,060 --> 00:05:33,460 S1: should do that. And one thing that comes to mind 88 00:05:33,460 --> 00:05:37,359 S1: around this is Alex Hormozi. He is, like, obsessed with 89 00:05:37,360 --> 00:05:40,390 S1: making money. Him and his wife are obsessed with making money, 90 00:05:40,390 --> 00:05:43,870 S1: which I respect. They're also way younger than me. So 91 00:05:43,870 --> 00:05:47,260 S1: they're like in their late 20s, early 30s. Or maybe 92 00:05:47,260 --> 00:05:50,260 S1: he's like mid 30s or whatever, but he wants to 93 00:05:50,260 --> 00:05:53,410 S1: crush business. And guess what? He is crushing business. So 94 00:05:53,410 --> 00:05:54,760 S1: what I'm going to do is take some of that 95 00:05:54,760 --> 00:05:57,550 S1: DNA from crushing business and try to sprinkle it into 96 00:05:57,550 --> 00:05:59,710 S1: my stuff. But what I won't be doing is making 97 00:05:59,710 --> 00:06:03,370 S1: the mistake of trying to be Alex Hormozi and switching 98 00:06:03,370 --> 00:06:07,570 S1: my focus and my telescope to point towards this distant 99 00:06:07,570 --> 00:06:10,360 S1: thing of like, oh, I want to make $100 million, 100 00:06:10,360 --> 00:06:12,790 S1: and he doesn't want to make $100 million. He wants 101 00:06:12,790 --> 00:06:15,640 S1: to be a billionaire. I guarantee you that's his goal 102 00:06:15,640 --> 00:06:18,670 S1: is like $1 billion in revenue or wealth or something. 103 00:06:18,670 --> 00:06:23,440 S1: So the thing is, you have to know what direction 104 00:06:23,440 --> 00:06:26,830 S1: your stuff is pointing in, and then you can more 105 00:06:26,830 --> 00:06:30,460 S1: easily say no to things because are they going to 106 00:06:30,460 --> 00:06:33,099 S1: get you there sort of accidentally? And that's the way 107 00:06:33,100 --> 00:06:35,380 S1: I've sort of oriented things, is like, I want to 108 00:06:35,380 --> 00:06:38,890 S1: do things so that I can accidentally make money from them, 109 00:06:38,890 --> 00:06:41,650 S1: like do stuff that you would do anyway and hook 110 00:06:41,650 --> 00:06:43,540 S1: it up to a stripe page and hope it works out. 111 00:06:43,540 --> 00:06:46,060 S1: And I know that's not the way to make a 112 00:06:46,060 --> 00:06:48,370 S1: lot of money. And I know you should, like, go 113 00:06:48,370 --> 00:06:51,280 S1: all in on a thing that's like, got a giant 114 00:06:51,279 --> 00:06:54,100 S1: market and, you know, you spam the hell out of 115 00:06:54,100 --> 00:06:57,400 S1: the ads or whatever, and you just try to make 116 00:06:57,400 --> 00:06:59,680 S1: all your money from that. But I feel like it's 117 00:06:59,680 --> 00:07:03,490 S1: just I feel like it's a bad game. And another 118 00:07:03,490 --> 00:07:05,800 S1: one of my favorite things, which I've mentioned here before, 119 00:07:05,800 --> 00:07:09,130 S1: is like, the worst way to lose is to win 120 00:07:09,130 --> 00:07:11,440 S1: playing the wrong game. And so this is all sort 121 00:07:11,440 --> 00:07:14,530 S1: of like flowing into that same thing. And I've been 122 00:07:14,530 --> 00:07:17,020 S1: happy with the newsletter lately because I've just been more 123 00:07:17,020 --> 00:07:20,830 S1: of my authentic self, and I feel like that's something 124 00:07:20,830 --> 00:07:24,760 S1: I just want for everybody. I want everyone to be 125 00:07:24,760 --> 00:07:27,490 S1: in tune with who they are and what they want, 126 00:07:27,490 --> 00:07:31,870 S1: and have them actually chasing those things and talking about 127 00:07:31,870 --> 00:07:35,140 S1: those things, talking about it is super important. It's like 128 00:07:35,140 --> 00:07:39,910 S1: I'm really worried about people becoming invisible after AI, because 129 00:07:39,910 --> 00:07:42,460 S1: AI is going to be finding the coolest stuff and 130 00:07:42,460 --> 00:07:45,520 S1: the best stuff. Well, if you're not saying anything, AI 131 00:07:45,520 --> 00:07:47,860 S1: is not going to find you. And also, if you're 132 00:07:47,860 --> 00:07:51,130 S1: not saying anything, you might not be thinking the things enough, 133 00:07:51,130 --> 00:07:54,460 S1: or you might not be in practice enough with coming 134 00:07:54,460 --> 00:07:57,370 S1: up with the ideas, crystallizing them, and writing them down. 135 00:07:57,370 --> 00:08:01,750 S1: Because I'm not advocating that people try to become successful authors, 136 00:08:01,750 --> 00:08:05,050 S1: or they try to become successful streamers or YouTubers or 137 00:08:05,050 --> 00:08:07,900 S1: anything like that for the purpose of being successful. I'm 138 00:08:07,900 --> 00:08:11,020 S1: advocating that they do it because it will make them 139 00:08:11,020 --> 00:08:15,580 S1: feel good to know themselves and to articulate themselves and 140 00:08:15,580 --> 00:08:18,910 S1: to emote themselves. And that's why I'm obsessed with this 141 00:08:18,910 --> 00:08:22,240 S1: whole thing. That's why I'm obsessed with this human 3.0 142 00:08:22,240 --> 00:08:24,640 S1: and all the stuff that I'm talking about. So anyway, 143 00:08:24,640 --> 00:08:27,010 S1: I hope you've noticed the difference in terms of me. 144 00:08:27,010 --> 00:08:30,010 S1: But really, the thing I'm trying to do, the whole 145 00:08:30,010 --> 00:08:32,590 S1: purpose of doing this for myself is to like, battle 146 00:08:32,590 --> 00:08:34,689 S1: test it so I can help other people do it. 147 00:08:34,690 --> 00:08:38,920 S1: All right. Stream of consciousness, collection of thoughts on US politics. 148 00:08:38,920 --> 00:08:41,020 S1: Just skip it if you don't like politics. But I 149 00:08:41,020 --> 00:08:43,540 S1: basically lay down a whole bunch of stuff there. Oh, 150 00:08:43,540 --> 00:08:46,660 S1: this is just become a member. Yeah, yeah, become a 151 00:08:46,660 --> 00:08:49,570 S1: member if you can. I would appreciate it. All right. 152 00:08:49,570 --> 00:08:52,960 S1: CSA held its first. This is, uh, security stuff here 153 00:08:52,960 --> 00:08:56,110 S1: or it's the new stories section. But, yeah, I always 154 00:08:56,110 --> 00:09:00,940 S1: start with security. CIA held its first AI security tabletop exercise, 155 00:09:00,940 --> 00:09:05,170 S1: over 50 experts, and they're looking to simulate responses to 156 00:09:05,170 --> 00:09:08,590 S1: AI security incidents. I love that they're thinking about it. 157 00:09:08,590 --> 00:09:12,309 S1: Extremists in the US are using AI to spread hate speech, 158 00:09:12,309 --> 00:09:15,939 S1: recruit and radicalize faster than ever. And they got a 159 00:09:15,940 --> 00:09:20,050 S1: couple of examples here. President Biden using using racial slurs 160 00:09:20,050 --> 00:09:23,050 S1: and Emma Watson reading Mein Kampf. I kind of want 161 00:09:23,050 --> 00:09:25,810 S1: to go watch these, but I also don't want to 162 00:09:25,809 --> 00:09:29,800 S1: Google for them. So yeah, I do not need that 163 00:09:29,800 --> 00:09:32,950 S1: garbage in my YouTube feed. The US has banned the 164 00:09:32,950 --> 00:09:37,720 S1: sale of Kaspersky anti-virus software, citing security risks from Russia. 165 00:09:37,720 --> 00:09:40,420 S1: And they're basically using strong language here. They're like, do 166 00:09:40,420 --> 00:09:43,930 S1: not use this stuff, stop using it immediately, get off 167 00:09:43,929 --> 00:09:46,960 S1: of it. And I think they have a deadline in July, 168 00:09:46,960 --> 00:09:49,839 S1: and there's a new set of high security vulnerabilities in Asus. 169 00:09:49,840 --> 00:09:54,340 S1: Routers let you get complete control without any user interaction. 170 00:09:54,340 --> 00:09:56,949 S1: And you want to patch these immediately. And the Richter 171 00:09:56,950 --> 00:10:06,610 S1: scale is 9.8. Thank you to dropzone for sponsoring Dmitri Alperovitch. Yeah. 172 00:10:06,610 --> 00:10:11,920 S1: Dmitri Alperovitch why is that difficult? That's not difficult. Imagined 173 00:10:11,920 --> 00:10:16,390 S1: a detailed scenario of China invading Taiwan in 2028, focusing 174 00:10:16,390 --> 00:10:21,460 S1: on aerosol strategy through the rough Taiwan Strait waters. This 175 00:10:21,460 --> 00:10:24,970 S1: one is fascinating to me because I don't have strong 176 00:10:24,970 --> 00:10:29,140 S1: intuitions myself, I. I don't have strong expertise myself in 177 00:10:29,140 --> 00:10:33,610 S1: like this whole naval warfare island attack type situation. It's 178 00:10:33,610 --> 00:10:36,280 S1: just a it's not something I've ever studied. I don't 179 00:10:36,280 --> 00:10:39,190 S1: know the stuff, but what's really fascinating to me is 180 00:10:39,190 --> 00:10:43,089 S1: that the super, super experts on this, they disagree with 181 00:10:43,090 --> 00:10:45,520 S1: each other, which reminds me a lot of the AI 182 00:10:45,550 --> 00:10:51,250 S1: situation there. I do have strong intuitions and some expertise 183 00:10:51,250 --> 00:10:53,920 S1: and a lot of expertise in some areas, and some 184 00:10:53,920 --> 00:10:58,030 S1: expertise in others, but it's like I feel way more 185 00:10:58,030 --> 00:11:02,739 S1: grounded and solid and confident in at least my intuitions 186 00:11:02,740 --> 00:11:05,590 S1: on the AI side. But here I have no idea 187 00:11:05,590 --> 00:11:11,050 S1: what's going on. And in both cases, the absolute best people, 188 00:11:11,050 --> 00:11:13,090 S1: which we're supposed to be looking up to, they do 189 00:11:13,090 --> 00:11:16,599 S1: not agree with each other and they disagree massively with 190 00:11:16,600 --> 00:11:21,280 S1: each other. Tyler Cowen had someone, this woman from she 191 00:11:21,280 --> 00:11:26,050 S1: was Bulgarian on his podcast recently, and it was so 192 00:11:26,050 --> 00:11:29,260 S1: good and her view was completely different than so many 193 00:11:29,260 --> 00:11:33,370 S1: other people. So I'm watching like Peter Zeihan. I'm listening 194 00:11:33,400 --> 00:11:36,700 S1: to her. I'm listening to the Goodfellas, which is a 195 00:11:36,700 --> 00:11:40,360 S1: podcast of a bunch of security and like really smart people. 196 00:11:40,360 --> 00:11:44,080 S1: And I'm trying to triangulate between all these different people 197 00:11:44,080 --> 00:11:47,439 S1: and basically come up with my own thoughts and opinions 198 00:11:47,440 --> 00:11:50,110 S1: based on, you know, not being an expert on it. 199 00:11:50,110 --> 00:11:53,830 S1: I just I find it so interesting that they can't agree. Oh, 200 00:11:53,830 --> 00:11:59,199 S1: same with, um, same with Ukraine. Right. Vastly different predictions 201 00:11:59,200 --> 00:12:00,940 S1: on whether or not it was going to happen in 202 00:12:00,940 --> 00:12:05,679 S1: the first place, when it's going to end, vastly different predictions. 203 00:12:05,679 --> 00:12:10,090 S1: And that excites me that that massively excites me because 204 00:12:10,090 --> 00:12:13,750 S1: I'm like, there is signal in here somewhere. The question is, 205 00:12:13,750 --> 00:12:17,990 S1: is it available second? Are people not getting the signal 206 00:12:17,990 --> 00:12:24,410 S1: or are they misinterpreting it? How are the best experts wrong? 207 00:12:24,410 --> 00:12:27,439 S1: And it raises the question of like, what else are 208 00:12:27,440 --> 00:12:31,070 S1: the best experts wrong on? And I just like to 209 00:12:31,070 --> 00:12:36,110 S1: have that constant tension and questioning in my mind as 210 00:12:36,110 --> 00:12:39,260 S1: I think about everything, including the things that I. I 211 00:12:39,260 --> 00:12:42,560 S1: believe the strongest US is moving to limit investments in 212 00:12:42,559 --> 00:12:47,660 S1: China's AI, semiconductor and quantum computing sectors to curb China's 213 00:12:47,660 --> 00:12:52,460 S1: tech advancements. And now they're looking at private equity venture 214 00:12:52,460 --> 00:12:57,620 S1: capital funds. Great stuff. And then I say, yeah, cool. 215 00:12:57,620 --> 00:13:00,650 S1: But we need to be building a lot more energy production. 216 00:13:00,650 --> 00:13:04,610 S1: Look at this. Look at this graphic from Leopold. This 217 00:13:04,610 --> 00:13:07,699 S1: is from situational awareness, which I just printed out, by 218 00:13:07,700 --> 00:13:10,250 S1: the way. It's like, I don't know, two inches thick 219 00:13:10,250 --> 00:13:12,860 S1: front and back. It's it's a beast. And I'm just 220 00:13:12,860 --> 00:13:15,440 S1: going to keep it in my car and read it. Uh, 221 00:13:15,440 --> 00:13:20,240 S1: whenever I'm parked and whatever, listening to podcasts. I won't 222 00:13:20,240 --> 00:13:22,280 S1: do them both at the same time. But anyway, I'm 223 00:13:22,280 --> 00:13:24,320 S1: going to keep it in the car and read through it. 224 00:13:24,320 --> 00:13:26,630 S1: I've already read a whole bunch online and of course, 225 00:13:26,630 --> 00:13:32,900 S1: watch the video, but. Yeah. All right. National. Yeah. Nationwide 226 00:13:32,900 --> 00:13:36,260 S1: armed militia from this guy named Jake Lang, who is 227 00:13:36,260 --> 00:13:41,420 S1: a January 6th writer. And they're using telegram for coordination. 228 00:13:41,750 --> 00:13:45,199 S1: And this is a wired article study says staring directly 229 00:13:45,200 --> 00:13:49,069 S1: at the camera during online job interviews can significantly boost 230 00:13:49,070 --> 00:13:53,990 S1: your evaluation scores. And this is from University of Hiroshima 231 00:13:54,020 --> 00:13:57,830 S1: AI agents and the Raz Revolution results as a service 232 00:13:57,830 --> 00:14:03,470 S1: discussing the evolution from SAS to raz, emphasizing the role 233 00:14:03,470 --> 00:14:08,059 S1: of AI agents. Yeah, I like this a lot. It's 234 00:14:08,059 --> 00:14:10,609 S1: like abstract. This is exactly what I'm building with a 235 00:14:10,610 --> 00:14:12,829 S1: project called substrate, which you're going to hear a lot 236 00:14:12,830 --> 00:14:17,390 S1: about here before too long. But this is exactly what 237 00:14:17,390 --> 00:14:22,010 S1: I'm building. You make a query and agents and a 238 00:14:22,010 --> 00:14:24,860 S1: whole bunch of data work together in a whole bunch 239 00:14:24,860 --> 00:14:28,070 S1: of steps, and the output that you get is the 240 00:14:28,070 --> 00:14:32,870 S1: result of their work. So if you imagine one idea 241 00:14:32,870 --> 00:14:35,870 S1: I have here, which is super exciting, is I'm going 242 00:14:35,870 --> 00:14:39,890 S1: to build intelligence reports this way. So on one hand, 243 00:14:39,890 --> 00:14:43,880 S1: you've got a whole bunch of sources that are maybe 244 00:14:43,880 --> 00:14:47,300 S1: they're hard to find, maybe they're not, but there are 245 00:14:47,300 --> 00:14:50,270 S1: many of them. And it takes some effort and curation 246 00:14:50,270 --> 00:14:53,930 S1: ability to to pick the best ones. Right. You bring 247 00:14:53,930 --> 00:14:57,380 S1: that together, you turn that into a giant summary or 248 00:14:57,380 --> 00:15:01,370 S1: a giant context file, and you basically start producing intelligence 249 00:15:01,370 --> 00:15:04,400 S1: artifacts out of this thing. And that's what I like 250 00:15:04,400 --> 00:15:08,990 S1: about results as a service. Okay. You you ask the question, 251 00:15:08,990 --> 00:15:12,260 S1: what is going on in Ukraine? What are the latest 252 00:15:12,260 --> 00:15:15,470 S1: global trends? Where should I invest my money or whatever? 253 00:15:15,470 --> 00:15:20,000 S1: And there could be hundreds of operations that happen, multiple 254 00:15:20,000 --> 00:15:23,840 S1: different eyes doing multiple different things. But really, it's not 255 00:15:23,840 --> 00:15:26,840 S1: even so much the AI until the end where it's 256 00:15:26,840 --> 00:15:29,720 S1: like putting it together and telling the story. It starts 257 00:15:29,720 --> 00:15:32,540 S1: with more legacy tech. It starts with all this collection, 258 00:15:32,540 --> 00:15:38,090 S1: all this correlation and unification of this data and getting 259 00:15:38,090 --> 00:15:41,000 S1: it into a thing that's ready for the for the 260 00:15:41,000 --> 00:15:46,610 S1: actual AI. And having brushed up against intelligence a little 261 00:15:46,610 --> 00:15:49,100 S1: bit when I was in the Army, I was attached 262 00:15:49,100 --> 00:15:52,400 S1: to an intelligence group and got to see a little 263 00:15:52,400 --> 00:15:55,160 S1: bit of how things work. Very little. I don't want 264 00:15:55,160 --> 00:15:58,400 S1: to oversell it here, but I've also read a whole 265 00:15:58,400 --> 00:16:01,940 S1: bunch about it and seen a whole bunch of this 266 00:16:01,940 --> 00:16:06,020 S1: stuff in on the Intel side and also on the 267 00:16:06,020 --> 00:16:09,680 S1: military side, and it comes down. Oh, I also built 268 00:16:09,680 --> 00:16:15,440 S1: a program around this concept at Apple for the security org. Right. 269 00:16:15,440 --> 00:16:18,050 S1: And this is a product that's still being used today 270 00:16:18,050 --> 00:16:22,170 S1: inside the company, which I'm very proud of. But I 271 00:16:22,170 --> 00:16:28,680 S1: use these concepts there. It's like data information intelligence. And 272 00:16:28,680 --> 00:16:33,180 S1: the intelligence piece is the one where there's intelligence, human 273 00:16:33,180 --> 00:16:37,410 S1: intelligence going into that effort to turn this into a narrative, 274 00:16:37,410 --> 00:16:42,060 S1: into a story for decision support. So at the above intelligence, 275 00:16:42,060 --> 00:16:45,690 S1: you have decision, right? Or maybe maybe it's not above, 276 00:16:45,690 --> 00:16:48,479 S1: maybe it's, um, to the right or whatever. But it's 277 00:16:48,480 --> 00:16:51,960 S1: like the purpose of that intelligence is to distill down. 278 00:16:51,960 --> 00:16:56,130 S1: So the the general is not looking at log files. Right. 279 00:16:56,130 --> 00:17:00,330 S1: Which is like data streaming in. Okay. You have information. 280 00:17:00,330 --> 00:17:02,430 S1: It's like a summary of the log files. And this 281 00:17:02,430 --> 00:17:05,970 S1: might not be a perfect analogy, but it moves up 282 00:17:05,970 --> 00:17:10,830 S1: and up in layers of abstraction with the entire purpose 283 00:17:10,830 --> 00:17:14,070 S1: of a human making a decision. Right? So if you 284 00:17:14,070 --> 00:17:18,300 S1: look at results as a service, I'm thinking intelligence as 285 00:17:18,300 --> 00:17:21,270 S1: a service is what I'm thinking. That's the first thing 286 00:17:21,270 --> 00:17:22,800 S1: in my mind goes to there's a whole bunch of 287 00:17:22,800 --> 00:17:26,369 S1: different implementations of this. Like it could be a a 288 00:17:26,369 --> 00:17:29,700 S1: business decision. Do I cancel this customer? Do I refund 289 00:17:29,700 --> 00:17:32,880 S1: this customer? Do I do whatever? But either way, it's 290 00:17:32,880 --> 00:17:35,909 S1: a whole bunch of legacy things happening with collection and 291 00:17:35,910 --> 00:17:40,200 S1: stuff like that, and IT systems and data sources and everything. 292 00:17:40,200 --> 00:17:44,010 S1: And then there's this thing that traditionally human is done, 293 00:17:44,010 --> 00:17:49,500 S1: or like Deborah or Chris or Raj or whoever it is, 294 00:17:49,500 --> 00:17:51,659 S1: they look at all the data and they say, you 295 00:17:51,660 --> 00:17:55,500 S1: know what, we are going to issue a payout for 296 00:17:55,500 --> 00:17:59,720 S1: this insurance policy, and that is the decision part. And 297 00:17:59,720 --> 00:18:02,869 S1: the result is a yes or no. Thumbs up, thumbs down. 298 00:18:02,869 --> 00:18:06,410 S1: And that gets us to this results as a service. Right? 299 00:18:06,410 --> 00:18:08,960 S1: So ideally you could just take all your different data, 300 00:18:08,960 --> 00:18:12,320 S1: all your different decision points, throw that into context, throw 301 00:18:12,320 --> 00:18:16,070 S1: it up to one of these services, which is one 302 00:18:16,070 --> 00:18:19,130 S1: of these that I'm building as well. And in like 303 00:18:19,130 --> 00:18:21,500 S1: 3 or 4 years, everyone's going to know this, and 304 00:18:21,500 --> 00:18:23,420 S1: this is just going to be the way all software 305 00:18:23,420 --> 00:18:26,540 S1: is built. But I'm already building this and I've already 306 00:18:26,540 --> 00:18:30,980 S1: got pieces of this that work. Um, and it's it's 307 00:18:30,980 --> 00:18:33,590 S1: really exciting. I mean, this is essentially the future of 308 00:18:33,590 --> 00:18:37,460 S1: software is doing this. You ask a question, you include context, 309 00:18:37,460 --> 00:18:40,760 S1: and you get back a result, a very high quality result. 310 00:18:40,760 --> 00:18:44,330 S1: So very excited about this concept and very good, uh, 311 00:18:44,330 --> 00:18:47,359 S1: piece here on medium, which I wish nothing was on medium, 312 00:18:47,359 --> 00:18:51,470 S1: but whatever. All right. Sam Logan has a new post 313 00:18:51,470 --> 00:18:54,800 S1: on how to get structured output from Llms, covering various 314 00:18:54,800 --> 00:18:59,000 S1: frameworks and their pros and cons. Yeah, really enjoy this one. 315 00:18:59,060 --> 00:19:02,869 S1: Talked about converting English to JSON and a bunch of 316 00:19:02,869 --> 00:19:06,290 S1: other stuff, but yeah, enjoyed that. All right. Someone got 317 00:19:06,290 --> 00:19:09,560 S1: invited to a project called Rebind where I versions of 318 00:19:09,560 --> 00:19:13,250 S1: authors interact with readers about classic books. These are the 319 00:19:13,250 --> 00:19:16,970 S1: really cool ideas. I just love ideas like this. Coming 320 00:19:17,000 --> 00:19:20,990 S1: to life Fabian Boff from Octo Mind explains why they 321 00:19:20,990 --> 00:19:25,100 S1: stopped using laying chain for their agents, highlighting its rigid, 322 00:19:25,100 --> 00:19:28,820 S1: high level abstractions. So I'm going to say something extra 323 00:19:28,820 --> 00:19:34,430 S1: spicy right now. A lot of these low level libraries 324 00:19:34,430 --> 00:19:40,310 S1: and frameworks for doing really difficult things like Rag, uh, 325 00:19:40,310 --> 00:19:44,449 S1: or agents, a lot of these agent frameworks, things like 326 00:19:44,450 --> 00:19:50,840 S1: laying chain, they are extremely kludgy. If anything was like overhyped, 327 00:19:50,840 --> 00:19:55,820 S1: I would say these first generation efforts at these things 328 00:19:55,820 --> 00:19:59,540 S1: are massively overhyped. And it kind of goes back to 329 00:19:59,540 --> 00:20:02,840 S1: this results as a service thing. Results as a service 330 00:20:02,840 --> 00:20:05,960 S1: is kind of the future of this, because it's not 331 00:20:05,960 --> 00:20:11,720 S1: asking you to mesh stack together these Jenga blocks of 332 00:20:11,720 --> 00:20:16,040 S1: like really kludgy code. Because here's the problem. And it 333 00:20:16,040 --> 00:20:22,010 S1: kind of applies to, um, to, uh, what are those called, um, notebooks. 334 00:20:22,010 --> 00:20:25,850 S1: It kind of applies to notebooks as well, because notebooks 335 00:20:25,850 --> 00:20:30,860 S1: like notebooks and agents and in laying chain, they make 336 00:20:30,859 --> 00:20:34,520 S1: great demos, they make great demos. But when you try 337 00:20:34,520 --> 00:20:37,070 S1: to take this stuff and build a production app on it, 338 00:20:37,070 --> 00:20:40,460 S1: you throw real scenarios into it, you throw real data 339 00:20:40,460 --> 00:20:44,690 S1: at it and it just falls over. And what universally, 340 00:20:44,690 --> 00:20:48,260 S1: almost everyone I've seen, and this is definitely what I've done. 341 00:20:48,260 --> 00:20:52,639 S1: You have to rebuild the entire thing from scratch using 342 00:20:52,640 --> 00:20:55,970 S1: smaller parts, which I've always been obsessed with this smaller 343 00:20:55,970 --> 00:20:59,660 S1: parts connected together. So it's like it it's necessary for 344 00:20:59,660 --> 00:21:02,840 S1: AI as well, which is why I'm building my very own, 345 00:21:02,840 --> 00:21:06,950 S1: like complete stack on this, uh, which is called substrate. 346 00:21:06,950 --> 00:21:11,150 S1: And I'm basically using those components to build apps. And 347 00:21:11,150 --> 00:21:14,959 S1: each one of the components is highly trusted and highly 348 00:21:14,960 --> 00:21:19,610 S1: rigorous in what it takes in and processes and what 349 00:21:19,609 --> 00:21:23,510 S1: it outputs. So it's dependable all the way through the pipeline. 350 00:21:23,510 --> 00:21:27,320 S1: And if you don't do this and you just start 351 00:21:27,320 --> 00:21:32,840 S1: piecing together laying chain and agent frameworks. And don't get 352 00:21:32,840 --> 00:21:37,129 S1: me wrong, I am not like, uh, being negative about 353 00:21:37,130 --> 00:21:40,880 S1: these things. Laying chain has done amazing things for AI 354 00:21:40,880 --> 00:21:46,130 S1: and these agent frameworks Autogen crew AI. They're doing amazing 355 00:21:46,130 --> 00:21:50,330 S1: stuff too, like no question. But as someone who's actually 356 00:21:50,330 --> 00:21:53,330 S1: building this stuff, the reality is completely different when you 357 00:21:53,330 --> 00:21:57,169 S1: start throwing production data at this stuff because you need consistency, 358 00:21:57,170 --> 00:22:02,330 S1: you need trust in these systems. And, uh, pretty much 359 00:22:02,330 --> 00:22:06,800 S1: every one that I know, they started building with these 360 00:22:06,800 --> 00:22:11,510 S1: bigger blocks. And then they got a few months in 361 00:22:11,510 --> 00:22:14,120 S1: and they had to tear it all down and start over, 362 00:22:14,119 --> 00:22:18,200 S1: building their own individual components. And so I think that's 363 00:22:18,200 --> 00:22:21,680 S1: where this is all going. Um, and eventually that's going 364 00:22:21,680 --> 00:22:24,560 S1: to merge into results as a service, which is kind 365 00:22:24,560 --> 00:22:27,949 S1: of like the opposite, actually. It's more like somebody did 366 00:22:27,950 --> 00:22:32,480 S1: build that back end, uh, dependable infrastructure, and now you're 367 00:22:32,480 --> 00:22:36,230 S1: just getting the results from it. Um, and the way 368 00:22:36,230 --> 00:22:38,810 S1: to know if it works is does it work? Do 369 00:22:38,810 --> 00:22:41,090 S1: you keep getting the right answer? And if you do, 370 00:22:41,119 --> 00:22:44,120 S1: you'll we will, as a society, build trust in that 371 00:22:44,119 --> 00:22:48,379 S1: over time. And ideally the thing will also be it 372 00:22:48,380 --> 00:22:53,750 S1: will also be publishing information about its components and the 373 00:22:53,750 --> 00:22:57,560 S1: sources that it used. So there'll be like some transparency there. Uh, 374 00:22:57,560 --> 00:23:00,170 S1: that will be part of. The quality of the service 375 00:23:00,170 --> 00:23:03,290 S1: is if you can actually inspect it and inspect like, okay, 376 00:23:03,290 --> 00:23:07,070 S1: which models did you use, which data sources did you use? Right. 377 00:23:07,070 --> 00:23:09,290 S1: Because if you're just like, no, trust me, because I'm 378 00:23:09,290 --> 00:23:12,619 S1: awesome and you like the results and don't worry about it, 379 00:23:12,619 --> 00:23:15,020 S1: for a lot of people, that won't be enough. All right? 380 00:23:15,020 --> 00:23:19,310 S1: My buddy Evan Özcelik and long time Ulr, he argues 381 00:23:19,310 --> 00:23:22,820 S1: that our obsession with convenience is making us less human, 382 00:23:22,820 --> 00:23:25,490 S1: and he wrote this on his blog, Snake Eye Software. 383 00:23:25,490 --> 00:23:28,910 S1: Solar generated a fifth of global electricity at midday on 384 00:23:28,910 --> 00:23:34,490 S1: the summer solstice. Solstice? Yep, a fifth of global energy 385 00:23:34,910 --> 00:23:40,609 S1: at midday. On summer solstice. That is interesting. I feel like, 386 00:23:40,790 --> 00:23:43,850 S1: I don't know, I'm not an expert on energy either, 387 00:23:43,850 --> 00:23:48,200 S1: but I, I really just wish we would go way heavier, 388 00:23:48,200 --> 00:23:53,869 S1: way heavier on solar, build, 100 giant mega solar farms 389 00:23:54,200 --> 00:23:58,440 S1: or 500 of them. And just go head to head 390 00:23:58,440 --> 00:24:04,830 S1: against China on solar. Because nuclear reactors, I understand that 391 00:24:04,830 --> 00:24:07,500 S1: they're really good energy and I understand they're awesome and 392 00:24:07,500 --> 00:24:11,880 S1: I understand all that, but they're very complex. They are 393 00:24:11,880 --> 00:24:16,470 S1: fundamentally dangerous, even though I know they're way safer than 394 00:24:16,470 --> 00:24:19,530 S1: they are now. It's possible to make them safe. Here's 395 00:24:19,530 --> 00:24:22,590 S1: another reason they freak people out. So there's going to 396 00:24:22,590 --> 00:24:25,710 S1: be so much more friction to trying to get one 397 00:24:25,710 --> 00:24:29,550 S1: or 2 or 10 of them built. And we are 398 00:24:29,550 --> 00:24:34,320 S1: in a race where speed matters and time matters. So 399 00:24:34,320 --> 00:24:36,780 S1: China can just build a whole bunch of those because 400 00:24:36,780 --> 00:24:41,010 S1: they have nobody to ask. They're just like, move, move. 401 00:24:41,010 --> 00:24:44,910 S1: We're building stuff here and everyone's like, yay, China! Like 402 00:24:44,910 --> 00:24:49,080 S1: they have just this massive advantage. Well, we can we 403 00:24:49,080 --> 00:24:51,179 S1: can move fast too if we do solar and it's 404 00:24:51,180 --> 00:24:54,960 S1: not freaking people out. So let's just go build 500 405 00:24:54,960 --> 00:24:58,590 S1: mega solar plants like Elon has been talking about forever, 406 00:24:58,590 --> 00:25:02,460 S1: and just jump way ahead, way ahead of the game 407 00:25:02,460 --> 00:25:06,330 S1: and still do nuclear, right? And you know where that 408 00:25:06,330 --> 00:25:08,820 S1: makes sense and where we can. But don't let it 409 00:25:08,820 --> 00:25:11,310 S1: slow us down. So when I'm talking about using SSH 410 00:25:11,340 --> 00:25:15,240 S1: as a pseudo replacement, uh, basically you can use it 411 00:25:15,240 --> 00:25:20,129 S1: to execute commands as another user. So why not use 412 00:25:20,130 --> 00:25:24,990 S1: that permission model instead of pseudo? Interesting. Apple is renaming 413 00:25:24,990 --> 00:25:29,250 S1: Apple ID to Apple account I like it. A group 414 00:25:29,250 --> 00:25:34,919 S1: of 17 secondary schools in Southwark, London, are going smartphone 415 00:25:35,130 --> 00:25:39,840 S1: free to combat the negative effects of phone use on students, 416 00:25:40,680 --> 00:25:43,619 S1: and I really love that Jonathan Heights work here is 417 00:25:43,619 --> 00:25:46,080 S1: spreading so fast. He put out his new book. He's 418 00:25:46,080 --> 00:25:50,740 S1: been blasting it all over podcasts. And I just love 419 00:25:50,740 --> 00:25:53,980 S1: that his work is catching on. And Gavin Newsom is 420 00:25:53,980 --> 00:25:57,100 S1: now going to ban, uh, smartphones. Or he's trying he's 421 00:25:57,100 --> 00:26:00,310 S1: trying to ban them. And likely this is coming from 422 00:26:00,310 --> 00:26:04,090 S1: impact from height as well. House prices are surging again 423 00:26:04,090 --> 00:26:08,230 S1: with global index up over 3% year on year. Insufficient 424 00:26:08,230 --> 00:26:12,730 S1: sun exposure is now a serious public health issue, potentially 425 00:26:12,730 --> 00:26:15,669 S1: causing hundreds of thousands of deaths annually in the US 426 00:26:15,670 --> 00:26:18,369 S1: and Europe. Isn't this like cheese and eggs and milk 427 00:26:18,369 --> 00:26:21,910 S1: and I don't know, so many, so many things. It's like, 428 00:26:22,030 --> 00:26:25,240 S1: that'll kill you. Well, everyone stops and it's like, yeah, 429 00:26:25,240 --> 00:26:28,900 S1: if you don't do that, that'll kill you. So I 430 00:26:28,900 --> 00:26:32,470 S1: can't wait till Huberman figures this out for us. Uh, 431 00:26:32,470 --> 00:26:35,860 S1: reads all the papers, uh, invites all the experts on. 432 00:26:35,859 --> 00:26:37,630 S1: And I know a lot of people are starting to 433 00:26:37,630 --> 00:26:42,399 S1: doubt Huberman and stuff like that, and I'm. I'm getting 434 00:26:42,400 --> 00:26:45,190 S1: a little more cautious with him. I'm not sure why, 435 00:26:45,190 --> 00:26:47,110 S1: but I'm getting a little more cautious with him. But 436 00:26:47,109 --> 00:26:49,720 S1: I don't know anyone else who is looking at these 437 00:26:49,720 --> 00:26:53,410 S1: papers and doing this great analysis. And I don't fault 438 00:26:53,410 --> 00:26:57,880 S1: people for being wrong sometimes I don't, I fault people 439 00:26:57,880 --> 00:27:04,629 S1: for being overconfident, never looking at themselves, never going after 440 00:27:04,630 --> 00:27:08,320 S1: their own errors, never trying to self-correct, and not being 441 00:27:08,320 --> 00:27:12,220 S1: analytical and pointing that lens at themselves. And I feel 442 00:27:12,220 --> 00:27:14,920 S1: like Huberman is good at that. I would like to 443 00:27:14,920 --> 00:27:17,560 S1: him to come back and do some episodes on, like, 444 00:27:17,560 --> 00:27:19,989 S1: here's stuff I got wrong and why and why I'm 445 00:27:19,990 --> 00:27:23,200 S1: going to fix that in the future. But anyway, he's 446 00:27:23,200 --> 00:27:25,870 S1: the best there is right now. And what I would 447 00:27:25,869 --> 00:27:28,540 S1: like to know about the sun thing is, I feel 448 00:27:28,540 --> 00:27:30,730 S1: better when I go out in the sun in the morning, 449 00:27:30,730 --> 00:27:33,190 S1: and I'm not sure it's only because of the light 450 00:27:33,190 --> 00:27:37,030 S1: going in my eyes. I feel like there's something about 451 00:27:37,030 --> 00:27:40,750 S1: walking on the ground, absorbing sunlight, and a lot of 452 00:27:40,750 --> 00:27:43,840 S1: this feels very emotional. So that's why I actually want 453 00:27:43,840 --> 00:27:46,840 S1: to see data. Like, I don't want this to be 454 00:27:46,840 --> 00:27:50,740 S1: based on vibes, but meanwhile I'm going to do it 455 00:27:50,740 --> 00:27:54,460 S1: based on vibes. Seems like a net good. Oh, in 456 00:27:54,460 --> 00:27:59,770 S1: one caveat, there is, um, between 10 and 2 is 457 00:27:59,770 --> 00:28:02,980 S1: like the worst time because your UV index is the worst. 458 00:28:02,980 --> 00:28:07,000 S1: But like super early sun is supposedly fairly safe because 459 00:28:07,000 --> 00:28:11,350 S1: the angle of incidence like coming into the atmosphere. So 460 00:28:11,350 --> 00:28:15,909 S1: evidently the UV is pretty low, but the visible light 461 00:28:15,910 --> 00:28:18,879 S1: I think is still much higher. I'm not sure if 462 00:28:18,880 --> 00:28:22,510 S1: those are correlated, but visible light is going to, uh, 463 00:28:22,510 --> 00:28:26,740 S1: set your clock, your circadian rhythm. And I feel like 464 00:28:26,740 --> 00:28:30,940 S1: there's just other health benefits there, too. And anyway, what 465 00:28:30,940 --> 00:28:33,580 S1: that study was basically saying is that a lot of 466 00:28:33,580 --> 00:28:36,909 S1: people aren't getting any sun, they're not going outside, they're 467 00:28:36,910 --> 00:28:40,210 S1: not doing anything. And I wonder how mixed that is 468 00:28:40,210 --> 00:28:43,090 S1: with like other things like exercise or whatever. But obviously 469 00:28:43,090 --> 00:28:45,580 S1: they try to filter that out. But the the idea 470 00:28:45,580 --> 00:28:49,180 S1: is basically people aren't going outside and they're depressed. And 471 00:28:49,180 --> 00:28:53,110 S1: guess what? There's probably a connection. They're having more positive 472 00:28:53,110 --> 00:28:55,750 S1: experience in life is linked to lower odds of brain 473 00:28:55,750 --> 00:29:01,750 S1: disorders like Alzheimer's and lower cognitive decline. And, uh, got 474 00:29:01,750 --> 00:29:06,880 S1: this secret Ted Cruz funding documents slightly redacted. Cut this, um, 475 00:29:07,420 --> 00:29:11,230 S1: this guy over here called Capital Press, and he's basically 476 00:29:11,230 --> 00:29:16,990 S1: like a watchdog, like a one person, like journalist group monitoring, 477 00:29:16,990 --> 00:29:22,410 S1: like the goings on of Congress people. And it's not in, like, 478 00:29:22,410 --> 00:29:27,540 S1: a stalkery nasty way. Uh, or a legal way. I 479 00:29:27,540 --> 00:29:31,530 S1: think it's more like Osint. Like what? Things are being 480 00:29:31,530 --> 00:29:34,260 S1: dropped out there in the public, you know, little clues 481 00:29:34,260 --> 00:29:38,130 S1: that can be hoovered up and, uh, and checked out. 482 00:29:38,130 --> 00:29:40,380 S1: And this is one thing I'm really excited about. This 483 00:29:40,380 --> 00:29:43,739 S1: is another thing I'm building in substrate is like an 484 00:29:43,740 --> 00:29:47,250 S1: intelligence collector, uh, which will allow me to do intelligence 485 00:29:47,250 --> 00:29:49,500 S1: reports like the one that I did before. And like this, 486 00:29:49,500 --> 00:29:54,810 S1: this guy is doing so. Yeah. Before. Before basically, this 487 00:29:54,810 --> 00:29:58,980 S1: was spy agencies, right? Journalists and spies were able to 488 00:29:58,980 --> 00:30:01,650 S1: do this because they had the money and they had 489 00:30:01,650 --> 00:30:06,930 S1: the training. Right. Well, I, I is cheap. It has 490 00:30:06,930 --> 00:30:09,090 S1: a lot of knowledge about how to do things. It 491 00:30:09,090 --> 00:30:11,790 S1: also has knowledge about how to turn those things into 492 00:30:11,790 --> 00:30:15,510 S1: a story and see patterns between them. So I'm going 493 00:30:15,510 --> 00:30:19,140 S1: to build something really nasty here. Really, really cool. Nasty. 494 00:30:19,140 --> 00:30:21,540 S1: I mean awesome. Uh, so so I can have like 495 00:30:21,540 --> 00:30:24,810 S1: a daily intelligence report of like, here's what the smartest 496 00:30:24,810 --> 00:30:28,110 S1: people in the world think about this, and here's where 497 00:30:28,110 --> 00:30:32,940 S1: their ideas diverge from each other. Here's how their opinions 498 00:30:32,940 --> 00:30:37,680 S1: diverge from one another. Here's where their ideas point at 499 00:30:37,680 --> 00:30:40,080 S1: the exact same thing. And I can do like a 500 00:30:40,080 --> 00:30:43,050 S1: heat map of like, opinions and like, I can't wait 501 00:30:43,050 --> 00:30:45,930 S1: to mess with this thing. All right, in our solar 502 00:30:45,930 --> 00:30:50,340 S1: eclipse photo competition. Look at this thing. This thing is insane. 503 00:30:51,120 --> 00:30:55,110 S1: Look at that. Look at this. That is so beautiful. Purple. 504 00:30:55,110 --> 00:30:57,870 S1: It's got the solar halo. And you got a plane. Oh. 505 00:30:57,870 --> 00:31:01,650 S1: Second place. Yeah. The plane one's way better. Sorry. Yeah. 506 00:31:01,650 --> 00:31:06,360 S1: Look at this. Purple. Oh, man. Got some, uh, vapor 507 00:31:06,360 --> 00:31:12,660 S1: trails there. Yeah. Very cool. All right. Ideas. Toddler intelligence 508 00:31:12,660 --> 00:31:17,470 S1: to PhD in around four years. That's what, uh, Mira 509 00:31:17,470 --> 00:31:21,880 S1: Murati just talked about on a podcast. Very interesting. I've 510 00:31:21,880 --> 00:31:24,790 S1: got this thing. Uh, just don't call it AI if 511 00:31:24,790 --> 00:31:26,920 S1: you're tired of hearing the word, I have a solution 512 00:31:26,920 --> 00:31:30,010 S1: for you. Forget. I forget the word. Forget everything about it. 513 00:31:30,010 --> 00:31:32,770 S1: And now just imagine the addition of hundreds of billions 514 00:31:32,770 --> 00:31:37,060 S1: of little helpers that people and companies can use to 515 00:31:37,060 --> 00:31:40,480 S1: do things like white collar work and writing and editing documents, 516 00:31:40,480 --> 00:31:44,800 S1: responding to emails, doing customer service, making sales calls, etc. 517 00:31:45,310 --> 00:31:48,850 S1: and taking ideas and turning those into full presentations, books 518 00:31:48,850 --> 00:31:51,820 S1: or films or whatever, and millions of other tasks that 519 00:31:51,820 --> 00:31:54,610 S1: otherwise would have not been done at all or would 520 00:31:54,610 --> 00:31:59,740 S1: have been done slowly or poorly by humans. And we're 521 00:31:59,740 --> 00:32:01,690 S1: about to have that. That's what we're about to have. 522 00:32:01,690 --> 00:32:05,620 S1: We're about to have billions of those things, doing tons 523 00:32:05,620 --> 00:32:10,690 S1: of stuff. I mean, just mountain loads of stuff that 524 00:32:10,690 --> 00:32:13,630 S1: has never been done before or was being done worse 525 00:32:13,630 --> 00:32:17,200 S1: by humans. Maybe we should call these things helpers, or 526 00:32:17,200 --> 00:32:20,410 S1: maybe we should call them whatever, but they're definitely going 527 00:32:20,410 --> 00:32:23,080 S1: to be very helpful, especially to employees who want to 528 00:32:23,080 --> 00:32:25,600 S1: do ten x the amount of work, or 100 x 529 00:32:25,600 --> 00:32:28,270 S1: the amount of work, or 1000 times the amount of 530 00:32:28,270 --> 00:32:31,150 S1: work that they were doing before, and have it be 531 00:32:31,150 --> 00:32:35,200 S1: done in a more consistent way or with higher quality. 532 00:32:35,200 --> 00:32:38,380 S1: And remember, the bar for a human work is low. 533 00:32:38,380 --> 00:32:40,930 S1: It is very low. If you've ever worked anywhere, if 534 00:32:40,930 --> 00:32:45,190 S1: you've ever worked with anyone. We are fallible. We make mistakes. 535 00:32:45,190 --> 00:32:47,860 S1: We are inconsistent. And if you ask a whole bunch 536 00:32:47,860 --> 00:32:50,080 S1: of experts how to do something, or you have them 537 00:32:50,080 --> 00:32:53,440 S1: actually do something, the quality level will be vastly different 538 00:32:53,440 --> 00:32:57,490 S1: and the average worker is just not that good at 539 00:32:57,490 --> 00:33:01,090 S1: their job and they don't really care that much. I mean, 540 00:33:01,090 --> 00:33:04,390 S1: that's just a fact. Or I would say maybe not 541 00:33:04,390 --> 00:33:07,570 S1: the average one, but like right below the average and 542 00:33:07,570 --> 00:33:10,630 S1: there are millions upon millions of those workers. So we're 543 00:33:10,630 --> 00:33:15,670 S1: talking about replacing that with better versions, more scalable, more 544 00:33:15,670 --> 00:33:21,010 S1: consistent and more upgradable, like instantly upgradable. So that's what 545 00:33:21,010 --> 00:33:24,070 S1: we're about to have billions of those. But don't call 546 00:33:24,070 --> 00:33:28,270 S1: it AI. Nope, nope, don't call it AI. Because AI 547 00:33:28,270 --> 00:33:30,850 S1: is annoying. It's an annoying word, so just don't use 548 00:33:30,850 --> 00:33:33,670 S1: it and you'll be safe. All right. Discovery web check. 549 00:33:33,670 --> 00:33:36,820 S1: Free and open source tool. Oh, this thing is so cool. 550 00:33:37,090 --> 00:33:39,910 S1: This thing is so cool. Yeah, I was playing with 551 00:33:39,910 --> 00:33:42,820 S1: this earlier. Watch this. All right, so first of all, 552 00:33:42,820 --> 00:33:44,770 S1: you can download it and run it yourself. But they 553 00:33:44,770 --> 00:33:48,610 S1: also have a web version. So look at the website websites. Cool. 554 00:33:49,030 --> 00:33:54,000 S1: All right. Is this cool. This is cool. I mean, 555 00:33:54,000 --> 00:33:57,910 S1: is it valuable? I don't know. But I love it. 556 00:33:57,910 --> 00:34:00,280 S1: I mean, it's got so much value in here. It's 557 00:34:00,280 --> 00:34:02,709 S1: just I'm not sure how I'm going to use that value. 558 00:34:03,410 --> 00:34:07,710 S1: Probably parse the whole thing and ask AI to. Tell 559 00:34:07,710 --> 00:34:11,280 S1: me what I need to pay attention to and prioritize 560 00:34:11,280 --> 00:34:15,520 S1: it for me. Give me a turn into a task list. 561 00:34:15,520 --> 00:34:19,300 S1: That's what I probably do anyway. It's really cool. Hayes 562 00:34:19,300 --> 00:34:25,420 S1: Labs automated red teaming against Llms Agentic LLM Vulnerability Scanner 563 00:34:25,420 --> 00:34:28,720 S1: open source tool for fuzzing and stress testing Llms with 564 00:34:28,719 --> 00:34:33,370 S1: customizable rule sets. Incogni I use another one I think 565 00:34:33,370 --> 00:34:37,239 S1: I use delete me is very similar to this service 566 00:34:37,239 --> 00:34:39,729 S1: that removes your personal info from the web to block 567 00:34:39,730 --> 00:34:45,300 S1: spam calls and protect privacy. Confusion is a muse. This 568 00:34:45,300 --> 00:34:49,700 S1: is a really cool little article. Talk about. Yeah. You 569 00:34:49,700 --> 00:34:52,760 S1: don't want to get rid of confusion because it it 570 00:34:52,760 --> 00:34:57,109 S1: speaks to you, makes you smarter. Complexity expands to fill 571 00:34:57,110 --> 00:35:01,820 S1: the space available. Mhm. Love it. The winter of content 572 00:35:01,820 --> 00:35:08,000 S1: how Game of Thrones changed media driving traffic and homogenizing journalism. 573 00:35:08,000 --> 00:35:12,500 S1: A new salary database created from viral videos. The internet 574 00:35:12,500 --> 00:35:16,520 S1: is now for bots, not humans, and an impossibly thin 575 00:35:16,520 --> 00:35:21,020 S1: fabric can cool you down over 16 degrees. I hesitate 576 00:35:21,020 --> 00:35:24,020 S1: to show these things unless I can buy them, but 577 00:35:24,020 --> 00:35:26,780 S1: this one was cool enough. I made an exception and 578 00:35:26,780 --> 00:35:29,719 S1: the recommendation of the week. Get a copy of Meditations 579 00:35:29,719 --> 00:35:32,960 S1: by Marcus Aurelius and keep it by your bed and 580 00:35:32,960 --> 00:35:37,009 S1: read some before you go to sleep. Two major benefits one. 581 00:35:37,010 --> 00:35:40,850 S1: You'll fall asleep faster. Two you will fall asleep feeling 582 00:35:40,850 --> 00:35:45,350 S1: grateful for your life. I think that's what stoicism does 583 00:35:45,350 --> 00:35:47,210 S1: for me, and I hope it does it for you 584 00:35:47,210 --> 00:35:50,690 S1: as well. And the aphorism of the week inspired by love, 585 00:35:50,719 --> 00:35:57,140 S1: guided by knowledge, inspired by love, guided by knowledge Bertrand Russell. 586 00:36:00,100 --> 00:36:03,250 S1: Unsupervised Learning is produced and edited by Daniel Miller on 587 00:36:03,250 --> 00:36:07,870 S1: a Neumann U87 AI microphone using Hindenburg. Intro and outro 588 00:36:07,870 --> 00:36:11,170 S1: music is by Zomby with the Y, and to get 589 00:36:11,170 --> 00:36:13,270 S1: the text and links from this episode, sign up for 590 00:36:13,270 --> 00:36:18,930 S1: the newsletter version of the show at Daniel miessler.com/newsletter. We'll 591 00:36:18,930 --> 00:36:19,739 S1: see you next time.