1 00:00:17,347 --> 00:00:21,067 S1: All right. Welcome to episode 483. Just a quick reminder. 2 00:00:21,067 --> 00:00:23,667 S1: I'm trying to get a lot more content out, and 3 00:00:23,707 --> 00:00:27,387 S1: I don't want to miss regular newsletter episodes. And I 4 00:00:27,387 --> 00:00:30,227 S1: want to actually get them out more closer to Monday 5 00:00:30,227 --> 00:00:35,387 S1: or Tuesday, ideally Monday, so that the podcast and any 6 00:00:35,387 --> 00:00:39,507 S1: videos and basically this newsletter and the podcast version of 7 00:00:39,507 --> 00:00:42,027 S1: the newsletter comes out as close to Monday as possible. 8 00:00:42,627 --> 00:00:44,507 S1: So I'm going to be doing a number of things 9 00:00:44,507 --> 00:00:46,147 S1: to try to speed that up. But one of those 10 00:00:46,147 --> 00:00:51,346 S1: things is actually having less editing on this podcast, especially 11 00:00:51,346 --> 00:00:53,627 S1: when I'm doing the newsletter. I mean, I'm talking straight 12 00:00:53,626 --> 00:00:56,147 S1: for 20, 30 minutes, and if I give this to 13 00:00:56,187 --> 00:00:58,107 S1: an editor, it's going to take them, you know, a 14 00:00:58,107 --> 00:01:01,387 S1: day or two to turn that around. Now, eventually, hopefully soon, 15 00:01:01,387 --> 00:01:04,307 S1: I will be able to find all my mistakes and 16 00:01:04,627 --> 00:01:07,867 S1: clean out the ums and ahs and the restatements of 17 00:01:07,867 --> 00:01:10,107 S1: a sentence, because what I'll often do is like, make 18 00:01:10,107 --> 00:01:12,027 S1: a mistake in a sentence and then I'll say the 19 00:01:12,027 --> 00:01:15,627 S1: word edit, and then I will say the sentence over again. 20 00:01:15,747 --> 00:01:18,816 S1: So if you hear any of those or you hear me, 21 00:01:18,817 --> 00:01:22,697 S1: just restate the sentence. I know it's annoying, I apologize, 22 00:01:22,697 --> 00:01:24,817 S1: I do not like to hear that in podcasts I 23 00:01:24,857 --> 00:01:28,857 S1: listen to, but if the podcast is high enough signal, 24 00:01:28,857 --> 00:01:31,617 S1: then I'm okay. You know, taking a little bit of 25 00:01:31,657 --> 00:01:34,937 S1: that and I'm saying, just bear with me, because I 26 00:01:34,937 --> 00:01:36,817 S1: think it's more important that we get the content out 27 00:01:36,817 --> 00:01:38,817 S1: and it's more important that we get the ideas out, 28 00:01:39,177 --> 00:01:41,337 S1: and we are just waiting for I to be able 29 00:01:41,337 --> 00:01:43,257 S1: to automatically clean that up. So I will just drag 30 00:01:43,257 --> 00:01:45,777 S1: and drop. It'll go clean all of that up and 31 00:01:45,777 --> 00:01:48,217 S1: we ship the thing and there's no delay whatsoever. So 32 00:01:48,417 --> 00:01:51,657 S1: just wanted to mention that and sort of preemptively apologize. 33 00:01:52,897 --> 00:01:55,977 S1: All right. First piece here. Release my new video on 34 00:01:55,977 --> 00:01:58,177 S1: where I think hacking is going. It's called The Future 35 00:01:58,177 --> 00:02:00,857 S1: of Hacking is context, and it is on YouTube. And 36 00:02:00,857 --> 00:02:02,937 S1: I recommend you go check it out. Got a new 37 00:02:02,937 --> 00:02:05,857 S1: essay on how I see AI affecting education that's on 38 00:02:05,857 --> 00:02:10,857 S1: the website. Uh, new essay on AI job replacement timelines, 39 00:02:11,737 --> 00:02:13,897 S1: and a new essay on my two groups of cyber 40 00:02:13,897 --> 00:02:17,596 S1: AI friends where basically it's crazy. I've got one group 41 00:02:17,597 --> 00:02:20,556 S1: of friends who are like, I can't believe this whole 42 00:02:20,556 --> 00:02:24,117 S1: thing is a fad. This is nothing but billionaires trying 43 00:02:24,117 --> 00:02:27,716 S1: to get rich like it's an NFT crypto scheme, and 44 00:02:27,716 --> 00:02:30,476 S1: it's just complete and absolute garbage. And you know, I've 45 00:02:30,477 --> 00:02:33,876 S1: got one friend, Marcus Hutchins, who doesn't even believe that 46 00:02:33,876 --> 00:02:38,637 S1: AI is intelligence. He just believes it's a complete scam. 47 00:02:38,677 --> 00:02:43,196 S1: The entire industry is broken. Um, it's not real intelligence. 48 00:02:43,197 --> 00:02:48,477 S1: It's not real AI. And it just completely down on it. 49 00:02:48,477 --> 00:02:51,677 S1: And he published a really cool essay, I think, which 50 00:02:51,677 --> 00:02:55,276 S1: I linked to, which talks about why it's not real intelligence. 51 00:02:55,276 --> 00:02:57,596 S1: And in the short answer or the short summary of it, 52 00:02:57,596 --> 00:03:00,677 S1: is the fact that it knows so much, but it 53 00:03:00,677 --> 00:03:07,237 S1: still can't invent new things. And my concise answer to this, 54 00:03:07,276 --> 00:03:09,636 S1: and by the way, I'm about to debate him, uh, 55 00:03:09,637 --> 00:03:14,026 S1: in public, in a recorded podcast on both of our shows, 56 00:03:14,667 --> 00:03:16,947 S1: and we're going to go through this in depth. But 57 00:03:17,387 --> 00:03:22,707 S1: the short answer to that is your rights. It's not 58 00:03:23,147 --> 00:03:27,786 S1: producing real innovation yet. But my argument is that real innovation, 59 00:03:27,827 --> 00:03:32,626 S1: real net new ideas is actually a tiny fraction of 60 00:03:32,626 --> 00:03:37,826 S1: the intelligence that humans use every day. You know, 99.99% 61 00:03:37,827 --> 00:03:42,387 S1: of intelligence that we use every day is not inventing 62 00:03:42,387 --> 00:03:46,427 S1: net new things. It's applying existing knowledge that we have 63 00:03:46,467 --> 00:03:49,907 S1: existing experience that we have, using our intelligence to apply 64 00:03:49,907 --> 00:03:52,266 S1: that to a new problem. And the new problem is 65 00:03:52,267 --> 00:03:57,547 S1: also not a net new 99.99% problem. It's just a 66 00:03:57,547 --> 00:04:00,307 S1: slightly different version of that problem, right? If you think 67 00:04:00,307 --> 00:04:03,666 S1: about an average white collar worker, if you think about 68 00:04:03,667 --> 00:04:09,867 S1: an average admin or secretary person or a an average 69 00:04:09,867 --> 00:04:13,657 S1: coder even, right. Usually the problems are like very similar 70 00:04:13,657 --> 00:04:17,217 S1: to problems we've solved before, right? And there are some 71 00:04:17,217 --> 00:04:20,657 S1: exceptions with coding and with high end engineering and of 72 00:04:20,657 --> 00:04:24,897 S1: course with, you know, pinnacle level science. There's lots of exceptions, 73 00:04:24,897 --> 00:04:27,897 S1: but the vast majority of the intelligence that we use 74 00:04:27,897 --> 00:04:30,817 S1: every single day, it's all the same stuff with different 75 00:04:30,817 --> 00:04:35,017 S1: names and slightly different configurations. It's all the same stuff, right? 76 00:04:35,057 --> 00:04:39,977 S1: And more importantly, teaching kids answering their questions, being a tutor, 77 00:04:40,297 --> 00:04:44,177 S1: being a teacher, being a helper, you know, being able 78 00:04:44,177 --> 00:04:47,497 S1: to help out inside of companies and be an assistant and, 79 00:04:47,817 --> 00:04:51,457 S1: you know, make meetings happen and coordinate and summarize things 80 00:04:51,657 --> 00:04:54,657 S1: and keep everyone up to date. This is what the 81 00:04:54,657 --> 00:04:59,497 S1: vast majority of work is, right? And I would argue 82 00:04:59,537 --> 00:05:04,577 S1: that's absolutely intelligence. It's absolutely human intelligence. And we know 83 00:05:04,577 --> 00:05:08,177 S1: this because we can't automate it. That's the reason it's 84 00:05:08,177 --> 00:05:13,287 S1: AI is because. If if we could automate it away, 85 00:05:13,287 --> 00:05:15,327 S1: it would have already be scripted and we wouldn't actually 86 00:05:15,327 --> 00:05:18,807 S1: have human employees doing that work. So if you think 87 00:05:18,807 --> 00:05:21,887 S1: about the average job that someone does in an intellectual 88 00:05:21,887 --> 00:05:27,327 S1: or cognitive role, like knowledge work, 99.9% of it, and 89 00:05:27,327 --> 00:05:30,167 S1: I'm just making up that number to, you know, to 90 00:05:30,207 --> 00:05:35,166 S1: be a little bit hyperbolic there, 99.99% of that work 91 00:05:35,327 --> 00:05:40,287 S1: is extremely repetitive. It's just slightly changing the work that 92 00:05:40,287 --> 00:05:44,127 S1: we do, right? It's slightly different scenarios, slightly different users, 93 00:05:44,127 --> 00:05:50,567 S1: slightly different use cases. But ultimately what this comes down 94 00:05:50,567 --> 00:05:54,287 S1: to is can I replace that? And I think we're 95 00:05:54,287 --> 00:05:56,687 S1: already seeing that. So I think that's empirically being shown 96 00:05:56,687 --> 00:05:59,887 S1: to be the case. Now Marcus's argument here is well, 97 00:05:59,887 --> 00:06:04,127 S1: that's not real intelligence. That's not real AI because it's 98 00:06:04,127 --> 00:06:06,367 S1: just doing the same thing over and over. And my 99 00:06:06,367 --> 00:06:09,427 S1: point is, that's what millions of jobs are actually doing. 100 00:06:09,747 --> 00:06:12,587 S1: That's why millions of jobs are actually going to get replaced, 101 00:06:12,907 --> 00:06:15,187 S1: is because AI is going to be able to do 102 00:06:15,227 --> 00:06:20,947 S1: that 99.99% better and faster and much cheaper than most humans. 103 00:06:21,067 --> 00:06:24,227 S1: So my whole thing with AI is that it matters 104 00:06:24,227 --> 00:06:29,267 S1: because of how it impacts humans, right? Getting rid of 105 00:06:29,267 --> 00:06:33,347 S1: millions of jobs affects humans. That's why I define AGI 106 00:06:33,387 --> 00:06:37,747 S1: as an AI that can replace an average knowledge worker. 107 00:06:38,027 --> 00:06:41,187 S1: Because I don't care about AI in a theoretical sense. 108 00:06:41,467 --> 00:06:43,307 S1: I don't care about it in a technical sense. I 109 00:06:43,347 --> 00:06:47,307 S1: care about it mostly in how it impacts humans and 110 00:06:47,307 --> 00:06:51,267 S1: human society and human meaning. So that's what our debate 111 00:06:51,267 --> 00:06:54,707 S1: is going to come around. And anyway, the way I 112 00:06:54,707 --> 00:06:56,987 S1: got on to this is that that's one view is 113 00:06:56,987 --> 00:07:00,947 S1: like this very anti-ai, very negative view of like, we 114 00:07:00,947 --> 00:07:03,427 S1: don't even have it. It's not real. It's all hype. 115 00:07:03,427 --> 00:07:05,547 S1: It's just like crypto. It's just a bunch of billionaires 116 00:07:05,547 --> 00:07:09,577 S1: trying to get rich. All these stats and benchmarks, everything 117 00:07:09,577 --> 00:07:14,497 S1: being said is basically marketing from executives for these lame, stupid, 118 00:07:14,537 --> 00:07:18,417 S1: temporary AI companies that are going to be gone immediately. 119 00:07:18,417 --> 00:07:21,417 S1: And basically it's not going to have much job impact 120 00:07:21,417 --> 00:07:23,937 S1: because it can't actually replace a human. So that's kind 121 00:07:23,937 --> 00:07:26,177 S1: of their argument. I'm not lumping everything I just said 122 00:07:26,177 --> 00:07:29,137 S1: into Marcuse's argument. I'll let him make his own arguments 123 00:07:29,297 --> 00:07:34,257 S1: when we have that discussion. But that's one group generally now. 124 00:07:34,257 --> 00:07:37,497 S1: The other group is like myself and Joseph Thacker and 125 00:07:37,817 --> 00:07:42,017 S1: my friend Jason Haddox and my friend Sasha Geller. Like, 126 00:07:42,057 --> 00:07:46,017 S1: we're all thinking, you know, we kind of saw where 127 00:07:46,017 --> 00:07:49,577 S1: this was going, especially myself and Joseph Thacker and Sasha. 128 00:07:50,257 --> 00:07:55,417 S1: It's like end of 22. We're like, oh my goodness. Right. 129 00:07:55,457 --> 00:07:57,737 S1: So all the stuff I've been talking about for the last, 130 00:07:58,097 --> 00:08:02,897 S1: you know, two and a half, three years is we 131 00:08:03,217 --> 00:08:06,527 S1: kind of saw the glimpses of that already happening at 132 00:08:06,527 --> 00:08:09,367 S1: the end of 2022. It was kind of inevitable. And 133 00:08:09,367 --> 00:08:11,087 S1: the more the more the AI started to happen, the 134 00:08:11,087 --> 00:08:13,207 S1: more obvious it gets or whatever. And of course, it's 135 00:08:13,207 --> 00:08:15,087 S1: part of the book. That was in 2016 as well. 136 00:08:15,087 --> 00:08:19,807 S1: But the point is, one group of people just still 137 00:08:19,847 --> 00:08:24,327 S1: today in mid 2025 doesn't accept it. Other another group 138 00:08:24,327 --> 00:08:26,727 S1: of people like myself and a bunch of other people 139 00:08:26,727 --> 00:08:30,287 S1: just like me or similar to me in mindset. They 140 00:08:30,287 --> 00:08:33,967 S1: got it instantly. Or if I'm right, they got it instantly. 141 00:08:33,967 --> 00:08:36,287 S1: If I'm wrong, then they were also instantly wrong, just 142 00:08:36,287 --> 00:08:39,647 S1: like me. And what's fascinating to me is these are 143 00:08:39,646 --> 00:08:44,607 S1: all security experts. These are all deeply technical, deeply experienced, 144 00:08:44,607 --> 00:08:49,686 S1: multiple decades of experience in cybersecurity, and we have drastically 145 00:08:49,687 --> 00:08:54,326 S1: different views on this. That's what's so fascinating to me, right? Why? 146 00:08:55,166 --> 00:08:57,206 S1: Why do they think that way and why do we 147 00:08:57,207 --> 00:09:00,526 S1: think this way? What makes the difference between being in 148 00:09:00,526 --> 00:09:02,446 S1: one group or the other? And I think the answer 149 00:09:02,447 --> 00:09:06,916 S1: comes down to just how the overall view of the world. Um, 150 00:09:06,916 --> 00:09:11,037 S1: Marcus in particular is very, I want to say socialist, 151 00:09:11,036 --> 00:09:13,516 S1: but not not socialist in a, in a bad way, 152 00:09:13,557 --> 00:09:18,797 S1: socialist in a human uplifting way, which I am as well. Right. 153 00:09:18,796 --> 00:09:23,597 S1: I just I'm also a capitalist. So it, uh, complex topic. 154 00:09:23,597 --> 00:09:25,597 S1: I'm not going to go into that, but the point 155 00:09:25,597 --> 00:09:29,157 S1: is he is very sort of anti billionaire anti-big business. 156 00:09:29,156 --> 00:09:32,036 S1: He sees all this as a giant scam, a big, uh, 157 00:09:32,396 --> 00:09:36,276 S1: you know, a big push to try to just, uh, 158 00:09:36,317 --> 00:09:39,396 S1: remove jobs or trick people into removing jobs when it's 159 00:09:39,396 --> 00:09:42,357 S1: not even real. He's just very cynical about the entire 160 00:09:42,357 --> 00:09:47,797 S1: thing based on, you know, kind of being anti-capitalist, anti, 161 00:09:47,997 --> 00:09:51,436 S1: you know, big business anti billionaire for sure. So I 162 00:09:51,437 --> 00:09:54,156 S1: think people who have that view or I think the 163 00:09:54,156 --> 00:09:57,037 S1: other view that that puts people in this category often 164 00:09:57,036 --> 00:10:00,477 S1: is just being anti-change. And a lot of people in 165 00:10:00,957 --> 00:10:05,016 S1: information security are actually anti-change. I didn't realize this until 166 00:10:05,016 --> 00:10:07,217 S1: a few years ago, but it's like there are many 167 00:10:07,217 --> 00:10:09,857 S1: people who are just like, oh, don't change that. You know, 168 00:10:09,896 --> 00:10:11,896 S1: when I was growing up in the 80s or 90s, 169 00:10:11,937 --> 00:10:14,136 S1: you know, the protocol worked like this and that's how 170 00:10:14,136 --> 00:10:17,177 S1: it should always work and can't stand these, these crazy 171 00:10:17,176 --> 00:10:19,657 S1: kids with all their JavaScript and all their dynamic web 172 00:10:19,697 --> 00:10:23,377 S1: apps and, you know, all the crazy APIs, like, I 173 00:10:23,416 --> 00:10:26,216 S1: just see a lot of get off my lawn type mentality. 174 00:10:26,416 --> 00:10:28,977 S1: And people who have that mentality tend to be very 175 00:10:29,256 --> 00:10:32,377 S1: anti AI in this way. And then my other group 176 00:10:32,377 --> 00:10:35,016 S1: of friends are like, yeah, tech changes all the time. 177 00:10:35,016 --> 00:10:38,097 S1: No big deal. And they kind of have this mentality 178 00:10:38,097 --> 00:10:40,937 S1: that I have which is shepherding. So I see myself 179 00:10:40,937 --> 00:10:43,657 S1: as a shepherd, you know, with like the stick and 180 00:10:43,656 --> 00:10:46,537 S1: the robes and, um, I don't want to use sheep 181 00:10:46,536 --> 00:10:49,617 S1: here because I don't consider all of humanity to be sheep, 182 00:10:49,617 --> 00:10:53,696 S1: but whatever, we'll use sheep. It's not actually pertinent in this. 183 00:10:53,937 --> 00:10:57,296 S1: The idea is you. It's your job to walk up 184 00:10:57,296 --> 00:11:00,446 S1: ahead and see if there's danger and to, like, help 185 00:11:00,447 --> 00:11:05,447 S1: steer people, because in this metaphor, it's not superiority of 186 00:11:05,447 --> 00:11:08,086 S1: the human to the inferiority of the sheep. It's more 187 00:11:08,087 --> 00:11:14,087 S1: like superiority of your knowledge of the risks versus inferiority, 188 00:11:14,087 --> 00:11:16,727 S1: of the knowledge of the risks. So it's my job 189 00:11:17,487 --> 00:11:19,487 S1: switching to a military metaphor. It's my job to go 190 00:11:19,487 --> 00:11:21,767 S1: out there and step on mines. It's my job to 191 00:11:21,766 --> 00:11:24,327 S1: go out there and detect all the mines, and then 192 00:11:24,327 --> 00:11:27,727 S1: spray paint a path on this minefield so people can 193 00:11:27,727 --> 00:11:31,247 S1: walk out, walk on it without getting hurt. Right. That's 194 00:11:31,247 --> 00:11:33,127 S1: the way I see it. And I see the the 195 00:11:33,447 --> 00:11:36,926 S1: this minefield is constantly changing because the tech is constantly changing. 196 00:11:36,926 --> 00:11:39,367 S1: And we should just accept that and everything's fine, right? 197 00:11:39,406 --> 00:11:41,646 S1: Change is constant. That's fine. Oh, we're going to lose 198 00:11:41,646 --> 00:11:45,086 S1: all these jobs. Cool. What's next? Um, AI is going 199 00:11:45,087 --> 00:11:47,087 S1: to replace a whole bunch of stuff. Cool. What's next? 200 00:11:47,087 --> 00:11:49,006 S1: Let's talk about the upsides. Oh, we're all going to 201 00:11:49,006 --> 00:11:51,607 S1: become way more creative. Okay, cool. Let's talk about that. 202 00:11:51,646 --> 00:11:54,286 S1: Let's figure out how to uplevel and upskill. So I 203 00:11:54,286 --> 00:11:58,207 S1: think that mentality versus the other anti-change mentality is the 204 00:11:58,317 --> 00:12:00,436 S1: primary thing that's guiding whether or not someone is in 205 00:12:00,437 --> 00:12:05,117 S1: one of these camps versus the other. So that's that's 206 00:12:05,117 --> 00:12:07,437 S1: essentially what that essay is about. And what I just 207 00:12:07,437 --> 00:12:09,957 S1: said is like 20 times longer than the actual essay, 208 00:12:10,237 --> 00:12:13,516 S1: which is like a one minute read. But that's the 209 00:12:13,516 --> 00:12:18,557 S1: advantage of doing a long form podcast. All right. Let's 210 00:12:18,557 --> 00:12:23,957 S1: see here. Gukesh Indian current world champion beat Magnus Carlsen 211 00:12:23,957 --> 00:12:26,717 S1: in the for the first time in classical chess and 212 00:12:26,717 --> 00:12:31,516 S1: Magnus like hammer fisted the table super super loud. Gukesh 213 00:12:31,516 --> 00:12:33,436 S1: didn't jump, so I gave him credit for that. But 214 00:12:33,957 --> 00:12:38,117 S1: that was cool. Got a video of that. All right. Cybersecurity. 215 00:12:38,156 --> 00:12:41,636 S1: Google patches new Chrome zero day bug in attacks Microsoft 216 00:12:41,636 --> 00:12:45,877 S1: and CrowdStrike create a shared threat actor dictionary. This one's 217 00:12:45,877 --> 00:12:48,517 S1: cool because some threat actors have like 15 different names 218 00:12:48,516 --> 00:12:52,836 S1: for them, so they unified that. That's, I think, progress. 219 00:12:53,276 --> 00:12:56,747 S1: OpenAI's O3 discovers Linux kernel zero day vulnerability I Vulnerability. 220 00:12:56,747 --> 00:12:59,506 S1: I talked about this last week, but, um, yeah, this 221 00:12:59,506 --> 00:13:02,386 S1: article is pretty good here. Meta plans to automate product 222 00:13:02,386 --> 00:13:05,467 S1: risk assessments with AI. This is really cool. 90% of 223 00:13:05,467 --> 00:13:08,906 S1: app updates they're going to do the assessments using AI. 224 00:13:09,107 --> 00:13:11,067 S1: This has always been one of my favorite things. I 225 00:13:11,107 --> 00:13:17,426 S1: did this very early, um, when I first started to pop, um, 226 00:13:17,627 --> 00:13:21,307 S1: a filtering system for a big problem with security teams 227 00:13:21,307 --> 00:13:23,707 S1: is they are the bottleneck between engineering being able to 228 00:13:23,707 --> 00:13:25,627 S1: ship and not being able to ship. So they get 229 00:13:25,627 --> 00:13:28,227 S1: really mad at security about that. I love the idea 230 00:13:28,227 --> 00:13:29,987 S1: of having all the context of the app that they're 231 00:13:29,987 --> 00:13:33,627 S1: trying to push through and sorting it into, like buckets. 232 00:13:33,627 --> 00:13:35,507 S1: You could actually use lots of different buckets. You could 233 00:13:35,506 --> 00:13:37,826 S1: use like 2 or 3, but you might be able 234 00:13:37,827 --> 00:13:41,347 S1: to sort it into like 15 different filtering buckets where 235 00:13:41,587 --> 00:13:44,906 S1: your pipeline of security checks is vastly different based on okay, 236 00:13:44,947 --> 00:13:48,667 S1: is it privacy focused? Is it, um, customer data security focused? 237 00:13:48,666 --> 00:13:52,947 S1: Is it proprietary? Data focused? Like what are what's the 238 00:13:52,947 --> 00:13:56,207 S1: threat model which could automatically do as well? I can 239 00:13:56,207 --> 00:13:58,687 S1: understand your risks of the company. It can understand worst 240 00:13:58,687 --> 00:14:01,447 S1: case scenarios for the company. And let's say it's sorting 241 00:14:01,447 --> 00:14:06,167 S1: into 16 different granular security check types. Hopefully most of 242 00:14:06,166 --> 00:14:08,847 S1: those being automated. But if it's a super high priority 243 00:14:08,886 --> 00:14:10,927 S1: thing it goes into the top one. It goes into 244 00:14:10,967 --> 00:14:13,727 S1: number 16. And that one is actually a manual red 245 00:14:13,766 --> 00:14:16,767 S1: team pen test, which is associated with automated scans that 246 00:14:16,766 --> 00:14:19,847 S1: already happened or whatever. But imagine being able to speed up, 247 00:14:19,887 --> 00:14:26,087 S1: you know, 85%, 95%, 98% of projects that engineering is 248 00:14:26,087 --> 00:14:30,007 S1: trying to push out for the company and have only 2% 249 00:14:30,007 --> 00:14:34,887 S1: actually involve manual testing, um, or whatever. I think that's really, 250 00:14:34,887 --> 00:14:37,687 S1: really powerful use case for this. And anyway, so meta 251 00:14:37,687 --> 00:14:40,647 S1: is doing that or something like that for 90% of 252 00:14:40,687 --> 00:14:46,527 S1: app updates using AI. Massive Asus router botnet uses persistent backdoors. 253 00:14:46,567 --> 00:14:50,607 S1: Over 8000 Asus routers affected by this, and it survives updates. 254 00:14:50,607 --> 00:14:52,927 S1: So if you reboot the router, the thing is still there. 255 00:14:53,477 --> 00:14:57,357 S1: Russian market becomes top destination for stolen credentials. So this 256 00:14:57,357 --> 00:15:01,477 S1: is the Russian market cybercrime platform. That's the name of it. 257 00:15:01,477 --> 00:15:04,997 S1: And it's filling the gap left by Genesis Markets takedown. 258 00:15:05,557 --> 00:15:08,637 S1: DOJ takes down four major services using that were used 259 00:15:08,637 --> 00:15:13,556 S1: by cyber criminals. So it's a basically like URL hiding 260 00:15:13,637 --> 00:15:18,397 S1: uh infrastructure. China linked hackers exploit SAP and SQL server 261 00:15:18,397 --> 00:15:21,357 S1: flaws in attacks across Asia and Brazil. And this actor 262 00:15:21,357 --> 00:15:27,037 S1: is called Earth Lamia. Lamia. And they're hitting multiple organizations 263 00:15:27,157 --> 00:15:33,597 S1: since 2023. National security Ukraine hides explosive drones in wooden 264 00:15:33,597 --> 00:15:37,237 S1: sheds to hit parked Russian bombers. This one is absolutely insane. 265 00:15:37,237 --> 00:15:40,157 S1: I'm sure you've seen stuff on it already. It's. It 266 00:15:40,157 --> 00:15:42,997 S1: reminds me of what actually happened with the Israeli attacks 267 00:15:42,997 --> 00:15:46,877 S1: on Hezbollah, with this ingenious, sort of like personalized little 268 00:15:46,877 --> 00:15:50,077 S1: mini bombs or whatever that went after leadership. So it 269 00:15:50,077 --> 00:15:53,186 S1: wasn't like giant drone strikes. It wasn't like giant bombing raids. 270 00:15:53,187 --> 00:15:57,387 S1: It was these personalized little things against leadership, which, uh, 271 00:15:57,787 --> 00:16:00,547 S1: a lot of people are saying, you know, tactical win here. 272 00:16:00,867 --> 00:16:03,427 S1: And it was pretty effective. And I think this is 273 00:16:03,427 --> 00:16:09,827 S1: similar in the sense that it's asymmetrical use of force. Um, 274 00:16:10,507 --> 00:16:14,947 S1: you have Ukraine, which is out outmanned, outgunned. And they 275 00:16:14,987 --> 00:16:20,707 S1: snuck these drones, uh, over a thousand miles inside of Russia, 276 00:16:20,707 --> 00:16:23,707 S1: I think maybe even close to 2000, 3000 miles in 277 00:16:23,707 --> 00:16:30,066 S1: some cases. Um, snuck them in and, um, basically, they're 278 00:16:30,067 --> 00:16:32,867 S1: inside of Russia. They they get all the way to 279 00:16:32,907 --> 00:16:37,427 S1: the airfield, they remotely open them. They're now sitting on Russian, uh, 280 00:16:37,427 --> 00:16:40,947 S1: cell networks, uh, which is what's used to activate them 281 00:16:40,947 --> 00:16:45,187 S1: and control them. And they, um, they hit these Russian bombers, 282 00:16:45,187 --> 00:16:48,347 S1: and these Russian bombers cannot be reproduced. Russia does not 283 00:16:48,347 --> 00:16:51,497 S1: have the ability to make these bombers again. And what's 284 00:16:51,497 --> 00:16:56,497 S1: interesting about this is they didn't diminish Russian bombing against Ukraine. 285 00:16:56,497 --> 00:17:03,257 S1: They diminished Russian bombing capabilities completely strategically as compared to 286 00:17:03,257 --> 00:17:06,976 S1: the United States as compared to China. So they significantly 287 00:17:06,977 --> 00:17:11,216 S1: damaged Russia's capabilities, right? Not to mention, you know, the 288 00:17:11,337 --> 00:17:15,257 S1: hundreds of thousands of actual troops they've lost. Um, but 289 00:17:15,297 --> 00:17:21,057 S1: this this war is seriously, seriously harming Russia's overall strategic 290 00:17:21,057 --> 00:17:25,257 S1: military capabilities. And, uh, the thing to call out here 291 00:17:25,257 --> 00:17:31,296 S1: is just how much Ukraine has changed warfare using drones. 292 00:17:31,777 --> 00:17:35,657 S1: And again, I implore you to go read Kill Decision 293 00:17:35,657 --> 00:17:40,297 S1: by Daniel Suarez, which talks about specifically autonomous drones, which 294 00:17:40,297 --> 00:17:43,936 S1: can't be taken out by GPS blockers or Em blockers 295 00:17:44,377 --> 00:17:48,137 S1: because they could do everything, uh, self-sufficient inside the drone itself. 296 00:17:50,077 --> 00:17:53,956 S1: China's deep network penetration signals war preparations. So former Trump 297 00:17:53,956 --> 00:17:57,516 S1: adviser H.R. McMaster told lawmakers that China's extensive hacking of 298 00:17:57,557 --> 00:18:02,477 S1: US infrastructure is actually them preparing for kinetic war. FBI 299 00:18:02,476 --> 00:18:06,397 S1: arrests defense intelligence IT worker for Pak drop espionage. So 300 00:18:06,557 --> 00:18:08,917 S1: he basically went to go do one of these, you know, 301 00:18:08,956 --> 00:18:13,436 S1: spy versus spy dead drops in a park in Virginia. And, uh, 302 00:18:13,837 --> 00:18:18,117 S1: the guy was actually part of DIA's internal threat division. 303 00:18:18,436 --> 00:18:20,757 S1: So so he's part of the team designed to help 304 00:18:20,757 --> 00:18:27,117 S1: people find or design to find spies inside of Dia. 305 00:18:27,517 --> 00:18:30,397 S1: And he was one of these spies, and he dropped 306 00:18:30,397 --> 00:18:32,917 S1: this material, and it was actually dropped to an FBI 307 00:18:32,956 --> 00:18:37,956 S1: agent and got all the write ups there. I Mary 308 00:18:37,997 --> 00:18:41,716 S1: Meeker returns with First Trends report since 2019. Definitely go 309 00:18:41,716 --> 00:18:44,917 S1: check this out. She is a beast of analysis and 310 00:18:44,917 --> 00:18:48,466 S1: she's been dormant for a while and now she's back. 311 00:18:48,466 --> 00:18:52,026 S1: And this analysis is like 300 something pages. Got a 312 00:18:52,027 --> 00:18:54,506 S1: decent summary here in the newsletter, but go check it out. 313 00:18:55,147 --> 00:18:58,586 S1: Dwarkesh Patel has longer AGI timelines than his podcast guests. 314 00:18:58,587 --> 00:19:01,667 S1: I think he's wrong about this, I think. So basically 315 00:19:01,667 --> 00:19:04,506 S1: what Dwarkesh is saying is that AI is not smart 316 00:19:04,507 --> 00:19:09,827 S1: enough to do continuous learning. And I think I'm not 317 00:19:09,827 --> 00:19:12,027 S1: sure about this, but I think he thinks that's an 318 00:19:12,027 --> 00:19:15,027 S1: IQ problem, whereas I see it as a scaffolding and 319 00:19:15,027 --> 00:19:18,667 S1: system problem. It's a memory problem. It's a context size problem. 320 00:19:18,867 --> 00:19:21,066 S1: It's a even more than a context size problem for 321 00:19:21,067 --> 00:19:25,946 S1: the model. It's more of a context knowledge base management problem. 322 00:19:25,946 --> 00:19:28,387 S1: So this is all scaffolding to me. And I think 323 00:19:28,387 --> 00:19:31,107 S1: this scaffolding is going to improve as fast or faster 324 00:19:31,107 --> 00:19:35,947 S1: than the IQ of the models themselves. So I think 325 00:19:35,946 --> 00:19:38,787 S1: this is not as big of a hurdle as Dwarkesh 326 00:19:38,827 --> 00:19:42,387 S1: is thinking it is. McKinsey says the future of work 327 00:19:42,387 --> 00:19:46,536 S1: is agentic. They released this big article Cool. Um, talking 328 00:19:46,537 --> 00:19:49,777 S1: about becoming digital workers that can think, decide and execute 329 00:19:49,817 --> 00:19:53,617 S1: tasks on their own. And it's worth a read. You 330 00:19:53,617 --> 00:19:55,057 S1: should go check it out. It's a it's a good 331 00:19:55,057 --> 00:19:58,777 S1: article by McKinsey. The truth about AI and job loss. 332 00:19:58,817 --> 00:20:03,056 S1: Neruda from meta dug into historical data to find out 333 00:20:03,057 --> 00:20:07,617 S1: which AI, which jobs AI will actually eliminate, and whether 334 00:20:07,617 --> 00:20:10,617 S1: there's still room for junior developers. That was a good read. 335 00:20:11,337 --> 00:20:15,617 S1: Google Gemini integration with Siri could fill Apple's personal context gap. 336 00:20:16,017 --> 00:20:18,497 S1: I think this analysis is wrong. I included it because 337 00:20:18,497 --> 00:20:22,457 S1: it's an alternative view to my own, but I think 338 00:20:22,456 --> 00:20:25,577 S1: this analysis is wrong. The problem is not Apple's access 339 00:20:25,577 --> 00:20:29,096 S1: to personal data. They have the best. They've been building 340 00:20:29,337 --> 00:20:31,617 S1: iOS all of this time. They have the best personal 341 00:20:31,617 --> 00:20:35,496 S1: contacts on people, I believe, on the planet. The problem 342 00:20:35,497 --> 00:20:38,776 S1: is having Siri prompt injected while it has access to 343 00:20:38,817 --> 00:20:41,657 S1: all that context. I don't think they know how to 344 00:20:41,696 --> 00:20:45,286 S1: give Siri or whatever the super agent is going to 345 00:20:45,287 --> 00:20:47,887 S1: be is probably still going to be Siri. They don't 346 00:20:47,887 --> 00:20:50,486 S1: know how to give Siri access to all that data 347 00:20:50,486 --> 00:20:54,407 S1: in a secure way, in a comprehensive, powerful way that 348 00:20:54,407 --> 00:20:56,567 S1: is secure. I think that is what is stopping them. 349 00:20:56,567 --> 00:20:58,887 S1: That's my guess. I don't know that for sure. Not 350 00:20:58,887 --> 00:21:01,127 S1: talking to people on the inside about it, but that 351 00:21:01,127 --> 00:21:04,006 S1: is my guess. Based on working in security at Apple 352 00:21:04,007 --> 00:21:08,767 S1: for like three years. Snowflake buys crunchy data for $250 million. 353 00:21:08,767 --> 00:21:11,767 S1: And this is snowflake, which is a giant data platform 354 00:21:12,127 --> 00:21:16,167 S1: making an AI agents play technology. McKinsey uses AI to 355 00:21:16,206 --> 00:21:21,687 S1: automate PowerPoint creation and proposal writing so their platform, Lily 356 00:21:21,726 --> 00:21:25,726 S1: now handles PowerPoint creation and proposal drafting over 75% of 357 00:21:25,726 --> 00:21:31,006 S1: people using it monthly. Internally, employees. Workday plans to rehire 358 00:21:31,007 --> 00:21:32,806 S1: the same number of people they laid off, but with 359 00:21:32,807 --> 00:21:35,367 S1: different skills. This is what I've been saying. They want 360 00:21:35,407 --> 00:21:41,647 S1: completely different people, completely different people. Nvidia develops new AI 361 00:21:41,686 --> 00:21:45,306 S1: chip for China that meets export controls. I don't think 362 00:21:45,307 --> 00:21:48,787 S1: this stuff matters that much. I think if China gets 363 00:21:48,787 --> 00:21:50,747 S1: less chips, they'll learn how to make more AI with 364 00:21:50,747 --> 00:21:53,466 S1: less chips. And in the meantime, they're making their own chips. 365 00:21:53,466 --> 00:21:56,307 S1: In fact, they just released one that performs really well. 366 00:21:57,027 --> 00:21:59,186 S1: I think ultimately China is going to get everything that 367 00:21:59,186 --> 00:22:02,587 S1: we have. I think ultimately, most AI advancements are happening 368 00:22:02,587 --> 00:22:05,027 S1: at the software layer via what I used to call 369 00:22:05,027 --> 00:22:08,187 S1: in 2023 tricks, a whole bunch of different tricks that 370 00:22:08,186 --> 00:22:11,787 S1: they're using, right? And those tricks, I think get leaked. 371 00:22:12,226 --> 00:22:16,787 S1: I think they get stolen. And so, for example, there's 372 00:22:16,787 --> 00:22:19,826 S1: an argument that something that OpenAI was doing with O1 373 00:22:19,867 --> 00:22:24,147 S1: got stolen by deep sea and deep sea copied it. 374 00:22:24,147 --> 00:22:27,227 S1: And guess what? Their R1 model was nearly as good 375 00:22:27,226 --> 00:22:31,306 S1: as O1. After OpenAI spent all that money in deep sea, 376 00:22:31,627 --> 00:22:35,107 S1: spent way less. That's an example of a trick or 377 00:22:35,147 --> 00:22:37,627 S1: a set of tricks getting stolen. I think that's going 378 00:22:37,667 --> 00:22:41,057 S1: to continue to happen. And it's why not only that, 379 00:22:41,057 --> 00:22:44,097 S1: the pinnacle model makers are going to stay close to 380 00:22:44,137 --> 00:22:46,657 S1: each other, but also open source is going to stay 381 00:22:46,657 --> 00:22:49,777 S1: close to the pinnacle model makers. So I think this 382 00:22:49,777 --> 00:22:54,097 S1: whole thing just kind of churns along, chugs along, advances, 383 00:22:54,097 --> 00:23:00,537 S1: slowly rolls forward with them, largely similar with them largely 384 00:23:01,897 --> 00:23:04,576 S1: staying neck and neck. Not always. There's going to be 385 00:23:04,577 --> 00:23:07,617 S1: these these big jumps, right? But I think that thing 386 00:23:07,617 --> 00:23:11,497 S1: will be copied. What I thought was would happen, and 387 00:23:11,497 --> 00:23:14,617 S1: what I think most people thought would happen is when 388 00:23:14,617 --> 00:23:17,817 S1: somebody made a major advancement, like OpenAI or anthropic or 389 00:23:17,817 --> 00:23:21,056 S1: someone in China, they would be a leader for six months, 390 00:23:21,097 --> 00:23:23,257 S1: a year or two years. They would just be like, oh, 391 00:23:24,017 --> 00:23:26,137 S1: they did something crazy, and no one can find out 392 00:23:26,137 --> 00:23:29,377 S1: what it is. And they're just way ahead. And everyone's 393 00:23:29,377 --> 00:23:32,937 S1: developing like independently in these silos. But it turns out 394 00:23:33,297 --> 00:23:35,497 S1: so many of these researchers know each other. So many 395 00:23:35,497 --> 00:23:40,446 S1: of these researchers are moving between the model maker companies. 396 00:23:40,686 --> 00:23:44,767 S1: So many of them talk and share their stuff in 397 00:23:44,767 --> 00:23:49,086 S1: coffee shops and in conferences and stuff. These secrets, these tricks, 398 00:23:49,767 --> 00:23:52,367 S1: these scaffolding things, because a lot of this is not 399 00:23:52,367 --> 00:23:54,607 S1: about the pre-training of the model. It's about what you 400 00:23:54,607 --> 00:23:57,607 S1: do after post training where a lot of these tricks 401 00:23:57,607 --> 00:24:00,887 S1: come into play and it's just being spread. It's just 402 00:24:00,887 --> 00:24:03,727 S1: being leaked. It's just being shared so much that I 403 00:24:03,726 --> 00:24:06,327 S1: think we're going to have a lot of parity with, 404 00:24:06,327 --> 00:24:10,086 S1: with some exceptions, uh, going forward. And I think the 405 00:24:10,087 --> 00:24:12,247 S1: only thing that changes this and gets really weird is 406 00:24:12,247 --> 00:24:17,007 S1: when you cross through AGI, where I think that still maintains, uh, 407 00:24:17,007 --> 00:24:20,807 S1: what I was saying, but when it hits ASI, that's 408 00:24:20,807 --> 00:24:22,887 S1: where you're going to have these giant jumps. And I 409 00:24:22,887 --> 00:24:25,767 S1: don't know necessarily. Well, no one knows what that's going 410 00:24:25,807 --> 00:24:28,806 S1: to look like because it's going to be ridiculous. Computer 411 00:24:28,807 --> 00:24:34,286 S1: science unemployment hits six 1% despite Major's popularity. So a 412 00:24:34,287 --> 00:24:37,006 S1: lot of questions of like, is computer science still a 413 00:24:37,007 --> 00:24:40,877 S1: good thing to go into. Um, I think dialectic, I 414 00:24:40,917 --> 00:24:48,477 S1: think studying, um, philosophy, um, rhetoric. Dialectic. Um, how to 415 00:24:48,517 --> 00:24:54,637 S1: think clearly, studying history, studying computer science, um, studying physics. 416 00:24:54,997 --> 00:24:57,357 S1: I think these are the core things that people should 417 00:24:57,397 --> 00:24:59,837 S1: actually be studying. And you should be getting your kids 418 00:24:59,837 --> 00:25:03,356 S1: and your younger young ones, or even people who are 419 00:25:03,357 --> 00:25:06,597 S1: advanced in their careers. This is how you retool. And 420 00:25:06,597 --> 00:25:08,397 S1: this is what I'm going to be putting into the 421 00:25:08,397 --> 00:25:12,997 S1: curriculum for H3 for human 3.0. Uh, this curriculum, I think, 422 00:25:12,997 --> 00:25:16,477 S1: is the core sort of first principles type education that 423 00:25:16,476 --> 00:25:18,557 S1: you're going to need. And of course, you can specialize 424 00:25:18,557 --> 00:25:20,837 S1: in all the stuff that you want like that you're 425 00:25:20,837 --> 00:25:23,837 S1: super interested in. But this core, I think, is what's 426 00:25:23,837 --> 00:25:29,036 S1: going to make you resilient to all these different changes. Humans. 60% 427 00:25:29,037 --> 00:25:32,317 S1: of Americans have retirement savings accounts, but it's very lumpy. 428 00:25:33,077 --> 00:25:36,536 S1: So 83% of people making over 100 K have retirement accounts. 429 00:25:36,577 --> 00:25:41,377 S1: Only 28% under 50 K have retirement accounts. There's a 430 00:25:41,377 --> 00:25:44,776 S1: massive racial gap. There's also a massive gap between college 431 00:25:44,777 --> 00:25:49,977 S1: graduates and, uh, people who don't have any college. So 81% 432 00:25:49,976 --> 00:25:54,976 S1: versus 39% and white versus people of color, it's 68% 433 00:25:54,976 --> 00:26:01,936 S1: versus 42%. Uh, US economy contracts, contracts more than expected 434 00:26:01,936 --> 00:26:05,056 S1: in Q1. It actually looks like it contracted by 0.2 435 00:26:05,057 --> 00:26:09,217 S1: percent in Q1. Younger generations less likely to develop dementia. 436 00:26:09,216 --> 00:26:13,096 S1: I think this is because previous generations, they would stop 437 00:26:13,097 --> 00:26:16,736 S1: working at 60 or 65, and they would just be like, oh, 438 00:26:16,777 --> 00:26:18,457 S1: I'm just going to play with grandkids and I don't 439 00:26:18,456 --> 00:26:20,576 S1: really need to do anything. And I think when you 440 00:26:20,577 --> 00:26:23,137 S1: tell your brain to shut off, it like literally starts 441 00:26:23,137 --> 00:26:27,417 S1: shutting down and you start getting dementia, that's that's me 442 00:26:27,417 --> 00:26:30,697 S1: being a fake doctor. Do not listen to me. I 443 00:26:30,696 --> 00:26:32,937 S1: think it's backed up by tons of data. But I 444 00:26:32,936 --> 00:26:36,527 S1: think if you treat your brain and your body like 445 00:26:36,527 --> 00:26:39,647 S1: you are a 22 year old in terms of like 446 00:26:39,686 --> 00:26:42,966 S1: you're lifting weights, you're working out, you're in the sun, 447 00:26:42,966 --> 00:26:46,647 S1: you're doing cardio, you're learning new things constantly. You are 448 00:26:46,767 --> 00:26:51,887 S1: executing on new cognitive tasks constantly, and you're constantly relearning 449 00:26:52,087 --> 00:26:56,086 S1: and in upgrading yourself, then you are basically sending a 450 00:26:56,087 --> 00:26:58,887 S1: signal to your brain and body, hey, we are 22. 451 00:26:59,167 --> 00:27:02,766 S1: You better stay in shape. And me, as old as 452 00:27:02,767 --> 00:27:06,327 S1: I am, that's exactly how I feel. I feel 22 453 00:27:06,927 --> 00:27:10,446 S1: virtually all the time. Unless I'm like off routine and 454 00:27:10,446 --> 00:27:14,287 S1: low energy or whatever. Then I feel like my age. But, um, 455 00:27:14,287 --> 00:27:18,407 S1: I would say 95% of the time I feel 22. 456 00:27:18,446 --> 00:27:20,807 S1: And this is why, and this is why I think 457 00:27:20,807 --> 00:27:25,127 S1: there's such a big difference between dimension numbers. Uh, in 458 00:27:26,047 --> 00:27:28,967 S1: I think the number is 25% versus 15%. So what 459 00:27:28,966 --> 00:27:34,397 S1: is that? That's a massive difference. Um, is just because 460 00:27:34,397 --> 00:27:36,957 S1: the old mindset of retirement versus the new mindset, which 461 00:27:36,956 --> 00:27:41,357 S1: is more active American versus European mindset on life. Really 462 00:27:41,357 --> 00:27:44,397 S1: good piece. If you're useful, it doesn't mean you're valued. 463 00:27:44,397 --> 00:27:47,196 S1: This was a good piece about economic value and how 464 00:27:47,196 --> 00:27:50,276 S1: companies view us. How much coffee is too much? Really 465 00:27:50,277 --> 00:27:54,237 S1: good news about 3 to 5 cups of coffee. Uh, discovery. 466 00:27:54,476 --> 00:27:58,317 S1: Run your own. I locally on your Mac Anthropic's interactive 467 00:27:58,317 --> 00:28:03,597 S1: prompt engineering tutorial indirect prompt injection overview. My AI skeptic 468 00:28:03,597 --> 00:28:07,237 S1: friends might are nuts. My AI skeptic friends are nuts. 469 00:28:07,236 --> 00:28:10,837 S1: This is by another security guy who's basically saying the 470 00:28:10,837 --> 00:28:12,917 S1: same thing that I was talking about in the beginning 471 00:28:12,917 --> 00:28:16,157 S1: from that essay. Um, and I've got a cool, cool 472 00:28:16,157 --> 00:28:19,716 S1: quote here from it. Um, quote. But the code is 473 00:28:19,716 --> 00:28:22,397 S1: shitty like that of a junior developer. And his response 474 00:28:22,397 --> 00:28:25,637 S1: to that. Does an intern cost $20 a month? Because 475 00:28:25,637 --> 00:28:29,277 S1: that's what cursor costs $20 a month. I think that's 476 00:28:29,277 --> 00:28:33,587 S1: a great point. Cloud code is my computer. You too. Inky. 477 00:28:33,627 --> 00:28:38,347 S1: Turns video into vocabulary. Flashcards gabenator filters. Hacker news job 478 00:28:38,347 --> 00:28:41,907 S1: posts with AI powered metadata. Tensor product. Attention is all 479 00:28:41,907 --> 00:28:44,867 S1: you need. The Metamorphosis of Prime Intellect. This is a 480 00:28:44,867 --> 00:28:47,627 S1: really cool book. It's an online text. You can just 481 00:28:47,627 --> 00:28:49,667 S1: click the link in the newsletter and you get the 482 00:28:49,667 --> 00:28:52,027 S1: full book. It's fantastic. One of my top ten sci 483 00:28:52,027 --> 00:28:55,707 S1: fi books ever. Andor season two shows how insider threats 484 00:28:55,747 --> 00:28:59,867 S1: actually work in real world organizations. I have not seen 485 00:28:59,867 --> 00:29:02,467 S1: all of Andor one, so I need to go watch that. 486 00:29:02,467 --> 00:29:05,267 S1: And two evidently it's the best Star Wars in a 487 00:29:05,267 --> 00:29:10,147 S1: long time. GitHub repository of end to end workflows. Someone 488 00:29:10,147 --> 00:29:13,427 S1: created a GitHub repo of tons of scraped any workflow 489 00:29:13,467 --> 00:29:16,267 S1: like hundreds of them, and you could just go get 490 00:29:16,267 --> 00:29:19,867 S1: them because they're in GitHub now, and because it's like 491 00:29:19,867 --> 00:29:23,827 S1: a text configuration, you could just like paste them into Mnd-n, which, 492 00:29:23,827 --> 00:29:26,827 S1: by the way, is becoming one of my favorite AI 493 00:29:26,827 --> 00:29:31,247 S1: automation frameworks. N the number eight A10. That and bedrock 494 00:29:31,247 --> 00:29:34,847 S1: are my favorites and landgraaf, but I'm using that last now, um, 495 00:29:35,247 --> 00:29:39,847 S1: preferring to use my five year experiment with UTC. So 496 00:29:39,847 --> 00:29:42,367 S1: someone is now using UTC for everything. I thought I 497 00:29:42,367 --> 00:29:45,367 S1: was crazy for going to metric this guys using UTC 498 00:29:46,167 --> 00:29:49,167 S1: book of Secret Knowledge GitHub repository. Jason Chan and Clint 499 00:29:49,167 --> 00:29:52,367 S1: Gibler have a brilliant conversation in the latest TLDR sick. 500 00:29:52,767 --> 00:29:56,167 S1: So my buddy Clint, who runs TLDR comm, which you 501 00:29:56,167 --> 00:29:59,647 S1: should go sign up for that newsletter. Um, he had 502 00:29:59,647 --> 00:30:02,327 S1: a guest post by Jason Chan, who's like a absolute 503 00:30:02,327 --> 00:30:09,607 S1: security guru, actually kind of pioneered, uh, security guardrails and, um, really, 504 00:30:10,767 --> 00:30:16,207 S1: really cool, innovative security guy who ran security at Netflix. 505 00:30:16,207 --> 00:30:18,367 S1: But anyway, he wrote a piece about building a security 506 00:30:18,367 --> 00:30:22,207 S1: program for TLDR, and it is extraordinary. You got to 507 00:30:22,207 --> 00:30:26,367 S1: go check it out. And the member essay, everyone is 508 00:30:26,367 --> 00:30:29,357 S1: multiple people. So Ellen just went to Washington and ruined 509 00:30:29,357 --> 00:30:32,877 S1: his reputation, damaged the value of his company's massively, and 510 00:30:32,877 --> 00:30:36,836 S1: then basically got ejected. Also, he could increase efficiency and 511 00:30:36,837 --> 00:30:39,397 S1: cut costs, which it turns out he was actually passionate about. 512 00:30:39,757 --> 00:30:42,197 S1: Now he's super pissed because the administration's new bill is 513 00:30:42,237 --> 00:30:44,837 S1: way more wasteful than anything. He cleaned up with Dodge 514 00:30:45,197 --> 00:30:49,117 S1: or Dodge, and it's basically a complete train wreck. So 515 00:30:49,117 --> 00:30:53,277 S1: lots of people, including Trump's former lawyer Cohen, are predicting 516 00:30:53,277 --> 00:30:56,277 S1: that Trump and the administration in general will come after Elon. 517 00:30:56,597 --> 00:30:59,757 S1: And after his money in various ways like stopping contracts, 518 00:30:59,797 --> 00:31:03,877 S1: launching investigations, auditing him, like with IRS or whatever. And 519 00:31:03,877 --> 00:31:06,717 S1: I think this whole weird arc with Elon and like 520 00:31:06,717 --> 00:31:11,077 S1: all this Trumpy stuff is interesting and it provides multiple lessons. 521 00:31:11,077 --> 00:31:14,037 S1: So first one is like someone could be insane committing 522 00:31:14,037 --> 00:31:17,197 S1: evil acts, and they could be principled about certain things 523 00:31:17,197 --> 00:31:20,397 S1: all at the same time. And everyone is multiple people, right? 524 00:31:20,437 --> 00:31:24,036 S1: And there are those people, those different people at different times. 525 00:31:24,517 --> 00:31:27,827 S1: So some people are just way more extreme than others, 526 00:31:27,827 --> 00:31:29,627 S1: and how many people they are, and how different the 527 00:31:29,627 --> 00:31:34,467 S1: various selves are from each other. Like lots of Elon's people, 528 00:31:34,707 --> 00:31:37,227 S1: that he is inside his brain or whatever. I like, 529 00:31:37,267 --> 00:31:39,307 S1: like I like the old Elon. I like the old 530 00:31:39,307 --> 00:31:42,907 S1: innovative like SpaceX, Elon or whatever. And now I can 531 00:31:42,907 --> 00:31:46,707 S1: barely tolerate hearing about him because he's too often these 532 00:31:46,707 --> 00:31:49,067 S1: versions of Elon that are mean and harmful and I 533 00:31:49,107 --> 00:31:53,587 S1: just can't abide it. Number two, Elon actually did believe 534 00:31:53,627 --> 00:31:56,667 S1: he was helping fix government and he was excited about that. 535 00:31:57,307 --> 00:32:00,347 S1: While the team was actually doing massive harm, by the way, 536 00:32:00,387 --> 00:32:02,467 S1: especially with US aid, if you look at that. Oh 537 00:32:02,507 --> 00:32:06,427 S1: my goodness. It's it's nasty. But we know for sure 538 00:32:06,427 --> 00:32:09,587 S1: that he was actually genuine about this cost saving stuff 539 00:32:09,827 --> 00:32:13,387 S1: because he's risking the wrath of losing hundreds of billions 540 00:32:13,387 --> 00:32:17,067 S1: of dollars, pushing back against Trump and this bill, pushing 541 00:32:17,067 --> 00:32:20,507 S1: back against, like, all these other people. And this isn't 542 00:32:20,507 --> 00:32:24,027 S1: really a political post. I mentioned Trump, so it's automatically 543 00:32:24,027 --> 00:32:26,177 S1: political because I mentioned it, but that's not what this 544 00:32:26,177 --> 00:32:29,377 S1: is about. What this is about is everyone is multiple people. 545 00:32:29,377 --> 00:32:33,977 S1: So I like Ellen. I can't stand Ellen, okay? Ellen 546 00:32:33,977 --> 00:32:37,537 S1: is principled about this cost saving stuff. He's principled about 547 00:32:37,537 --> 00:32:41,857 S1: certain things. He's not principled about a million other things. Uh, 548 00:32:41,857 --> 00:32:44,937 S1: because he's crazy. Because he's actually an asshole or because 549 00:32:44,937 --> 00:32:49,377 S1: whatever reasons, the point is, be very cautious of anyone 550 00:32:49,377 --> 00:32:53,297 S1: telling you, Ellen or anyone else for that matter, is 551 00:32:53,297 --> 00:32:59,177 S1: like any one thing. Like if they insist that he's 552 00:32:59,177 --> 00:33:02,177 S1: some brilliant, kind hearted genius who needs to be protected, 553 00:33:02,417 --> 00:33:06,697 S1: or that he's 100% evil tycoon who only cares about money, 554 00:33:07,177 --> 00:33:10,577 S1: they are both wrong. Wisdom is based in reality and 555 00:33:10,577 --> 00:33:16,097 S1: reality is complex. Recommendation of the week. Make sure you're 556 00:33:16,097 --> 00:33:18,737 S1: not trying to win the wrong game. This is a 557 00:33:18,737 --> 00:33:20,816 S1: text I sent to a mentee this week that I 558 00:33:20,817 --> 00:33:23,757 S1: think is worth sharing. Hey, I have a concept that 559 00:33:23,757 --> 00:33:25,677 S1: I've shared with you before that I want to keep 560 00:33:26,717 --> 00:33:29,597 S1: front of mind for you, especially going to all this 561 00:33:29,597 --> 00:33:32,277 S1: AI shit over the next few years. That is going 562 00:33:32,277 --> 00:33:34,677 S1: to change a lot of the ground under our feet. 563 00:33:34,997 --> 00:33:38,357 S1: So make sure you're not setting up to win the 564 00:33:38,357 --> 00:33:42,117 S1: wrong game. Nothing is worse than celebrating after years of work, 565 00:33:42,477 --> 00:33:44,596 S1: doing a project or something, and then realizing it was 566 00:33:44,597 --> 00:33:47,877 S1: the wrong target. Let's be super thoughtful about the goals 567 00:33:47,877 --> 00:33:49,797 S1: we pick out for the next 1 to 3 years. 568 00:33:50,637 --> 00:33:56,157 S1: So that's what I sent to her and I wanted 569 00:33:56,157 --> 00:33:58,757 S1: to share that. So I sent I screenshotted the text, 570 00:33:58,757 --> 00:34:00,677 S1: which is actually what's in the newsletter, but I also 571 00:34:00,677 --> 00:34:02,757 S1: sent that screenshot of text to some other people that 572 00:34:02,757 --> 00:34:06,677 S1: I'm helping out in the aphorism of the week live 573 00:34:06,717 --> 00:34:08,877 S1: as if you were living a second time and as 574 00:34:08,877 --> 00:34:12,076 S1: though you had acted wrongly the first time. Live as 575 00:34:12,077 --> 00:34:15,037 S1: if you were living a second time, and as though 576 00:34:15,077 --> 00:34:18,317 S1: you had acted wrongly the first time. Viktor Frankl.