1 00:00:00,800 --> 00:00:04,939 S1: Unsupervised Learning is a podcast about trends and ideas in cybersecurity, 2 00:00:04,970 --> 00:00:10,010 S1: national security, AI, technology and society, and how best to 3 00:00:10,039 --> 00:00:17,780 S1: upgrade ourselves to be ready for what's coming. All right, 4 00:00:17,780 --> 00:00:21,710 S1: welcome to Unsupervised Learning. This is Daniel, episode 470. Let's 5 00:00:21,710 --> 00:00:24,860 S1: get into it. Reading a couple new books in addition 6 00:00:24,860 --> 00:00:30,020 S1: to this month's books for UL. So the UL books 7 00:00:30,170 --> 00:00:35,480 S1: are 1984 by Orwell and Animal Farm by Orwell. And 8 00:00:35,479 --> 00:00:37,520 S1: these two books I just added to the list. I'm 9 00:00:37,520 --> 00:00:41,540 S1: about done with the Technical Republic and about to jump 10 00:00:41,540 --> 00:00:45,589 S1: into money, lies and God. And actually, there's another one 11 00:00:45,590 --> 00:00:48,650 S1: I'm going to add to this list, somebody who came 12 00:00:48,650 --> 00:00:51,710 S1: on the Goodfellows podcast. So I'm going to go find 13 00:00:51,710 --> 00:00:56,060 S1: his book as well. But, um, yeah, having a good 14 00:00:56,060 --> 00:00:59,180 S1: time with some reading. Got a friend in Guatemala who 15 00:00:59,180 --> 00:01:03,560 S1: was recently laid off senior engineer, focused around monitoring solutions 16 00:01:03,560 --> 00:01:07,070 S1: but really can do lots of different stuff. So I 17 00:01:07,069 --> 00:01:10,009 S1: recommend you check out his LinkedIn and see if if 18 00:01:10,010 --> 00:01:13,850 S1: you might have anything. My friend Monica is offering 25% 19 00:01:13,850 --> 00:01:19,820 S1: off her Security Leadership Masterclass, which I consulted for, and 20 00:01:19,850 --> 00:01:21,890 S1: I think it's really good for people trying to get 21 00:01:21,890 --> 00:01:26,510 S1: into leadership around security. I got a funny meme here about, um, 22 00:01:26,510 --> 00:01:30,169 S1: how to calm down a buddy who loses a bunch 23 00:01:30,170 --> 00:01:34,040 S1: of money in crypto. It's just it's just funny. Nothing 24 00:01:34,040 --> 00:01:37,910 S1: useful there except for laughter, which is useful. LinkedIn posts 25 00:01:37,910 --> 00:01:42,170 S1: about my ultimate app, which I keep iterating on. I'm 26 00:01:42,170 --> 00:01:43,940 S1: going to go ahead and click through to this one. 27 00:01:44,180 --> 00:01:47,870 S1: So this is essentially like the main app that I've 28 00:01:47,870 --> 00:01:50,090 S1: been building for the longest time. And I've got like 29 00:01:50,120 --> 00:01:52,580 S1: different versions of it. I've got like a commercial version, 30 00:01:52,580 --> 00:01:56,540 S1: I've got the overall arching one, but essentially it takes 31 00:01:56,540 --> 00:01:59,960 S1: inputs from any medium, from any platform, and does a 32 00:01:59,960 --> 00:02:02,840 S1: ton of analysis to determine how good it is. Okay, 33 00:02:02,870 --> 00:02:07,190 S1: so just imagine that this is like the most important 34 00:02:07,190 --> 00:02:13,070 S1: thing to me is most important like current buildable use 35 00:02:13,070 --> 00:02:16,910 S1: for AI. So all this stuff is happening in the world, right? 36 00:02:16,910 --> 00:02:22,040 S1: All this news, new intelligence is is dropping open source intelligence, 37 00:02:22,040 --> 00:02:26,510 S1: recon vulnerabilities are popping up on websites. Um, my favorite 38 00:02:26,540 --> 00:02:31,100 S1: thinkers are releasing a new blog, blog post, a new essay, 39 00:02:31,130 --> 00:02:35,540 S1: a new article, a new video on YouTube comes out. Right. 40 00:02:35,570 --> 00:02:40,609 S1: All the authors that I follow, they're releasing new books. Right? Um, 41 00:02:40,639 --> 00:02:45,410 S1: comments are being made which might reveal something really, really interesting. Uh, 42 00:02:45,410 --> 00:02:49,850 S1: new art is coming out, new books, new movies that 43 00:02:49,850 --> 00:02:54,230 S1: I should be watching. Um, comments from my friends, new 44 00:02:54,230 --> 00:02:57,470 S1: companies being launched. All this stuff is happening, you know. 45 00:02:57,500 --> 00:03:00,680 S1: Millions of things per day. Hundreds of thousands. Tens of 46 00:03:00,680 --> 00:03:04,520 S1: thousands or whatever, uh, pieces of content coming into my 47 00:03:04,520 --> 00:03:10,370 S1: RSS feeder. And all of that is is something I 48 00:03:10,370 --> 00:03:14,810 S1: could potentially benefit from. Okay, so this app that I've 49 00:03:14,810 --> 00:03:17,960 S1: been working on for the longest time, really starting, uh, 50 00:03:17,960 --> 00:03:23,329 S1: beginning of 23, is essentially the consumption mechanism for this. 51 00:03:23,360 --> 00:03:26,780 S1: Pull it all together, then use AI to run it 52 00:03:26,780 --> 00:03:30,200 S1: through a filter. Okay. That filter and I've already built 53 00:03:30,200 --> 00:03:34,489 S1: this way back then. The filter basically does a quality 54 00:03:34,490 --> 00:03:38,060 S1: check on all of these things to see if it's 55 00:03:38,060 --> 00:03:43,490 S1: full of interesting ideas, if it's creative, if it's novel, 56 00:03:43,490 --> 00:03:47,390 S1: if it's surprising, and all these various different aspects similar 57 00:03:47,390 --> 00:03:51,410 S1: to the label and rate, um, fabric pattern that I 58 00:03:51,410 --> 00:03:53,720 S1: have out there that's public. This is just like a 59 00:03:53,720 --> 00:03:56,900 S1: much more advanced version of it with a lot more 60 00:03:56,900 --> 00:04:01,460 S1: labels and stuff like that, because based on that, the 61 00:04:01,460 --> 00:04:04,220 S1: AI can then or a different AI or a different 62 00:04:04,220 --> 00:04:07,520 S1: part of the same AI can then sort what to 63 00:04:07,550 --> 00:04:10,790 S1: do with it. Right? Because it might be an alert. 64 00:04:10,790 --> 00:04:14,180 S1: It might send me an alert if something is important enough, like, hey, 65 00:04:14,210 --> 00:04:16,880 S1: a new vulnerability was found on one of your websites, 66 00:04:16,880 --> 00:04:20,990 S1: or a new vulnerability was found on a bounty program 67 00:04:20,990 --> 00:04:25,339 S1: that I'm monitoring. And by the way, the AI is 68 00:04:25,339 --> 00:04:27,739 S1: telling me my AI is like, oh, by the way, 69 00:04:27,740 --> 00:04:30,500 S1: I found the vulnerability. I found how you could actually 70 00:04:30,500 --> 00:04:33,650 S1: exploit it. I wrote that up into a report and 71 00:04:33,650 --> 00:04:36,020 S1: I submitted it to the thing. So we are waiting 72 00:04:36,050 --> 00:04:38,930 S1: to see if we're going to get paid from this bounty. Right? 73 00:04:38,930 --> 00:04:41,570 S1: So that that's a cool thing. And it might send 74 00:04:41,570 --> 00:04:44,090 S1: me an email which tags it with a certain thing 75 00:04:44,089 --> 00:04:47,930 S1: that says awaiting for bounty payout or something like that. 76 00:04:47,960 --> 00:04:52,010 S1: Or it might be that I just want to know if, um, 77 00:04:52,100 --> 00:04:56,080 S1: this particular company buys any new companies, or I want 78 00:04:56,110 --> 00:04:59,260 S1: to know if this particular company does anything to, like, 79 00:04:59,290 --> 00:05:02,920 S1: raise money. All of these could be like alerts that 80 00:05:02,920 --> 00:05:06,550 S1: I have set up, but it starts with the consumption 81 00:05:06,550 --> 00:05:10,900 S1: of the world and the processing through this AI. Now 82 00:05:10,900 --> 00:05:13,750 S1: I've already built this I've got is probably a version 83 00:05:13,750 --> 00:05:16,659 S1: like three or something, depending on how you call the versions. 84 00:05:16,660 --> 00:05:20,560 S1: But what I want to mention about this is that 85 00:05:20,560 --> 00:05:23,830 S1: the real version of this, the real version of this, 86 00:05:23,830 --> 00:05:27,190 S1: which I've talked about in my other videos, especially the 87 00:05:27,190 --> 00:05:32,740 S1: predictive path AI video, is that the real version of 88 00:05:32,740 --> 00:05:38,170 S1: this is my AI buddy, my AI assistant, my I friend, 89 00:05:38,200 --> 00:05:44,260 S1: my AI, my digital assistant. Okay, my universal primary digital 90 00:05:44,260 --> 00:05:47,020 S1: assistant is going to be in charge of doing this 91 00:05:47,020 --> 00:05:50,770 S1: for me. Okay. My APIs that are available to me 92 00:05:50,770 --> 00:05:54,880 S1: are the ones that are doing this processing, right, these agents, 93 00:05:54,880 --> 00:05:58,480 S1: these automation workflows, these things that are just running on 94 00:05:58,480 --> 00:06:05,469 S1: my behalf. They're out there in the world collecting, analyzing, assessing, um, labeling, 95 00:06:05,470 --> 00:06:09,190 S1: doing all those things for me. But the quarterback for 96 00:06:09,190 --> 00:06:12,760 S1: all of this is my DVR. My DVR is monitoring 97 00:06:12,760 --> 00:06:15,610 S1: how many feeds I have, if they're the right feeds, 98 00:06:15,610 --> 00:06:17,920 S1: if this feed like we thought it was awesome, but 99 00:06:17,920 --> 00:06:20,980 S1: it's actually garbage. It might have the permission to go 100 00:06:20,980 --> 00:06:24,430 S1: in and clean that up and fix it, right. So 101 00:06:24,430 --> 00:06:28,419 S1: the future of this is I don't really ask any 102 00:06:28,420 --> 00:06:32,440 S1: questions about this. I'm not like typing very much. I'm not, um, 103 00:06:32,440 --> 00:06:36,580 S1: really manually interacting with this too much at all. Because 104 00:06:36,580 --> 00:06:39,310 S1: if I want a new source, I say, hey, add 105 00:06:39,310 --> 00:06:42,580 S1: this to my source, okay? While I'm looking at a 106 00:06:42,580 --> 00:06:45,490 S1: page on the screen, hey, add this to my our 107 00:06:45,490 --> 00:06:49,240 S1: list of sources. Am I says, am I here? Yeah. 108 00:06:49,270 --> 00:06:52,510 S1: No problem. Just added it. Okay, I'm watching a new 109 00:06:52,540 --> 00:06:55,479 S1: YouTube video, for example. I'm like, yeah, add this to 110 00:06:55,510 --> 00:06:59,289 S1: our sources and it's like it runs the API command 111 00:06:59,290 --> 00:07:01,779 S1: to go add that. It's now part of the workflow. 112 00:07:01,779 --> 00:07:04,690 S1: It's now in the system right now. How could it 113 00:07:04,690 --> 00:07:07,390 S1: do that? Well, it's going to do that because it's 114 00:07:07,420 --> 00:07:10,480 S1: watching all my screens or it's watching with the camera. 115 00:07:10,540 --> 00:07:12,190 S1: However it's doing it. There's going to be lots of 116 00:07:12,190 --> 00:07:14,860 S1: different ways. But the point is, my Da will be 117 00:07:14,860 --> 00:07:18,280 S1: part of my life. Um, Apple is talking about having 118 00:07:18,280 --> 00:07:21,820 S1: cameras on its AirPods. I don't know if that'll be 119 00:07:21,820 --> 00:07:24,610 S1: the next version. Probably not. It'll be the version like 120 00:07:24,640 --> 00:07:28,030 S1: 2 or 3 versions after the current version, but I'm 121 00:07:28,030 --> 00:07:31,900 S1: wearing them now. Like if if these were like overear 122 00:07:31,960 --> 00:07:34,270 S1: and in the back, it had a camera that was 123 00:07:34,270 --> 00:07:38,740 S1: pointing behind me. Amazing. Obviously it should have a camera 124 00:07:38,740 --> 00:07:43,240 S1: that's pointing forward, pointing forward, pointing behind me with full 125 00:07:43,240 --> 00:07:47,050 S1: access to my life context, full access to my screen. 126 00:07:47,050 --> 00:07:50,080 S1: It should also have access within the screen as well, 127 00:07:50,080 --> 00:07:52,300 S1: so we could look at processes and stuff like that. 128 00:07:52,300 --> 00:07:55,000 S1: But let's just say you could only just see the screen. 129 00:07:55,000 --> 00:07:57,370 S1: I could still ask it things and it could still 130 00:07:57,370 --> 00:08:01,240 S1: do lookups. It could still summarize what I'm looking at, 131 00:08:01,270 --> 00:08:05,080 S1: for example. So this is really the interaction. The interaction 132 00:08:05,080 --> 00:08:08,590 S1: becomes me just talking to my AI and my AI 133 00:08:08,620 --> 00:08:12,280 S1: giving me back results. Now the second level of this, 134 00:08:12,280 --> 00:08:16,180 S1: which comes a little bit later because the technology just 135 00:08:16,180 --> 00:08:19,179 S1: is not quite there yet, but it's more like the 136 00:08:19,180 --> 00:08:22,989 S1: meta glasses. Okay, the meta glasses, as you saw in 137 00:08:22,990 --> 00:08:26,230 S1: the newsletter last week, or maybe it's this week and 138 00:08:26,230 --> 00:08:29,320 S1: we're about to talk about it. Um, the meta glasses 139 00:08:29,320 --> 00:08:33,190 S1: have sold 2 million units. They are the direction that 140 00:08:33,190 --> 00:08:35,829 S1: things are going. I imagine Apple is going to move 141 00:08:35,830 --> 00:08:38,590 S1: in this direction as well, and kind of offload a 142 00:08:38,590 --> 00:08:42,460 S1: bunch of the processing to the phone, but the next 143 00:08:42,460 --> 00:08:44,980 S1: version of this app that I'm building, the reason I'm 144 00:08:44,980 --> 00:08:49,120 S1: building all these services that are working for me, coming 145 00:08:49,120 --> 00:08:53,530 S1: through my digital assistant, which also doesn't fully exist yet. Okay. 146 00:08:53,559 --> 00:08:57,910 S1: That's also a place everyone is pushing into, including Apple. 147 00:08:58,059 --> 00:09:01,929 S1: But the real thing is that I am watching. I 148 00:09:01,960 --> 00:09:04,540 S1: am talking to somebody and they're like, hey, you know, 149 00:09:04,570 --> 00:09:07,060 S1: you should get in on this deal and blah, blah, blah. 150 00:09:07,059 --> 00:09:09,700 S1: And it's crypto related and it's a meme coin. And 151 00:09:09,700 --> 00:09:13,689 S1: I'm just like, uh, well, not interested. Don't want to 152 00:09:13,690 --> 00:09:17,440 S1: hear about that. Um, but let's say I was interested 153 00:09:17,440 --> 00:09:20,350 S1: and let's say it did sound good. In the meantime, 154 00:09:20,350 --> 00:09:24,819 S1: my Da is calling all my APIs. Okay. I have 155 00:09:24,820 --> 00:09:28,630 S1: an API for researching someone. I have an API for 156 00:09:28,660 --> 00:09:33,100 S1: creating a dossier on somebody. I have an API for monitoring, 157 00:09:33,100 --> 00:09:39,580 S1: watching someone's face and determining characteristics of lying. So while 158 00:09:39,580 --> 00:09:42,880 S1: I'm looking with my apple glasses, which is just like 159 00:09:42,880 --> 00:09:45,340 S1: they look like regular glasses, this is pretty far in 160 00:09:45,340 --> 00:09:48,459 S1: the future. Like five years Maybe. Maybe shorter. Maybe like 161 00:09:48,460 --> 00:09:51,040 S1: three years with meta or something. Who knows? But let's 162 00:09:51,040 --> 00:09:55,270 S1: just say three to 5 to 10 years. Okay. As 163 00:09:55,270 --> 00:09:58,210 S1: you're watching, I have this little dial in the corner 164 00:09:58,210 --> 00:10:02,230 S1: of my glasses, and it's like. It's like a bullshit meter, right? 165 00:10:02,260 --> 00:10:05,859 S1: It's like, yeah, no, don't believe that. And it could 166 00:10:05,860 --> 00:10:08,500 S1: even be printing out little thing in the text or whatever. 167 00:10:08,500 --> 00:10:12,610 S1: That's like that claim he just made is incorrect. That 168 00:10:12,610 --> 00:10:14,620 S1: was kind of bullshit. Oh, that was kind of smart 169 00:10:14,620 --> 00:10:17,260 S1: or whatever. You could do whatever. You could have an 170 00:10:17,260 --> 00:10:20,079 S1: outline around them based on the type of person you're 171 00:10:20,080 --> 00:10:22,870 S1: looking at. I talked about this in that big post, 172 00:10:22,870 --> 00:10:24,970 S1: which you should go check out. Actually, here's the video 173 00:10:24,970 --> 00:10:28,570 S1: for it. So you've got to think about what is 174 00:10:28,570 --> 00:10:33,640 S1: possible from your Da when your Da can also control 175 00:10:33,640 --> 00:10:37,420 S1: what you are seeing. The overlay on top of the world. Okay, 176 00:10:37,420 --> 00:10:40,630 S1: so when I walk into a Starbucks and I'm looking 177 00:10:40,630 --> 00:10:44,620 S1: for a new partner, I'm looking for a girlfriend, which 178 00:10:44,620 --> 00:10:47,440 S1: I'm not, by the way, but let's say it were, um, 179 00:10:47,470 --> 00:10:52,210 S1: there could be different outlines based on what, um, the 180 00:10:52,210 --> 00:10:57,370 S1: girls in the Starbucks are broadcasting in their demons. They're 181 00:10:57,370 --> 00:11:01,300 S1: broadcasting that they're creative. They're broadcasting that they like programming, 182 00:11:01,300 --> 00:11:04,030 S1: they're broadcasting that they want to start a business, they're 183 00:11:04,030 --> 00:11:10,330 S1: broadcasting whatever. And my Da could then place different outlines 184 00:11:10,330 --> 00:11:14,380 S1: around them based on highlighting that for me, knowing what 185 00:11:14,380 --> 00:11:17,500 S1: I'm looking for, right. And more importantly, there's going to 186 00:11:17,500 --> 00:11:21,429 S1: be 1000 or 1 million companies out there pitching to 187 00:11:21,460 --> 00:11:26,319 S1: my Da that their filter on reality is the best one, 188 00:11:26,350 --> 00:11:28,690 S1: because they have the coolest outlines, and they have the 189 00:11:28,690 --> 00:11:31,420 S1: coolest little animations that go around people who are like 190 00:11:31,450 --> 00:11:35,950 S1: artists or engineers or whatever. So this is kind of 191 00:11:35,980 --> 00:11:38,530 S1: these are the steps. This is what I'm building. Like 192 00:11:38,530 --> 00:11:41,350 S1: the stuff I put in that book from 2016, the 193 00:11:41,350 --> 00:11:44,500 S1: stuff I've been talking about in these posts, like I'm 194 00:11:44,500 --> 00:11:48,370 S1: not waiting for it to happen. I'm building the pieces 195 00:11:48,370 --> 00:11:51,189 S1: right now. Like I can't build the hardware. Like I'll 196 00:11:51,190 --> 00:11:53,020 S1: never mess with that. I have to wait for that. 197 00:11:53,020 --> 00:11:57,310 S1: But I can maybe build the assistant, or at least 198 00:11:57,309 --> 00:12:00,460 S1: build something to go on top of the assistant. And 199 00:12:00,460 --> 00:12:03,880 S1: I can absolutely build these services. And this is like 200 00:12:03,880 --> 00:12:07,690 S1: the most important service. This is the gathering and filtering service. 201 00:12:07,690 --> 00:12:10,870 S1: So that is what that's what I've made. Oh and 202 00:12:10,870 --> 00:12:13,450 S1: by the way I the commercial version of this that 203 00:12:13,450 --> 00:12:15,880 S1: I already put out, let me just click this and 204 00:12:15,880 --> 00:12:19,540 S1: see like what we get here. This is my actual threshold. 205 00:12:19,540 --> 00:12:24,579 S1: This is my personal feed for threshold. And I've actually 206 00:12:24,580 --> 00:12:26,980 S1: never showed this have I never showed this I don't 207 00:12:26,980 --> 00:12:29,170 S1: think I've ever showed this before. Let me show you 208 00:12:29,170 --> 00:12:32,530 S1: what this is like. So this is my threshold. And 209 00:12:32,530 --> 00:12:35,709 S1: look what I can do in here. I have a 210 00:12:35,710 --> 00:12:40,120 S1: score here that I can, uh, set the minimum threshold 211 00:12:40,120 --> 00:12:43,630 S1: of quality that has to. The content has to be. 212 00:12:43,630 --> 00:12:46,990 S1: So I have thousands of things like this is not theory. 213 00:12:46,990 --> 00:12:50,920 S1: I've already built this. Thousands of things are coming into 214 00:12:50,920 --> 00:12:57,040 S1: this thing being parsed, being filtered, being labeled, being rated right. 215 00:12:57,070 --> 00:13:00,610 S1: Watch this. It's at 85. I slide this down to 60. 216 00:13:00,670 --> 00:13:04,270 S1: Look I have all these categories as well. I hit save. 217 00:13:04,300 --> 00:13:07,990 S1: Look what it does over here. What it just readjusted. 218 00:13:08,020 --> 00:13:12,730 S1: It just readjusted. And watch this. Watch this view analysis 219 00:13:12,730 --> 00:13:16,059 S1: I have summaries of the contents. You could decide if 220 00:13:16,059 --> 00:13:18,100 S1: you actually want to go and read it, or if 221 00:13:18,100 --> 00:13:20,890 S1: you actually want to go and watch the video and 222 00:13:20,890 --> 00:13:26,020 S1: look at this surface level, middle level, deep level, hidden 223 00:13:26,020 --> 00:13:29,110 S1: message analysis, the ideas that are actually in here, the 224 00:13:29,110 --> 00:13:33,340 S1: recommendations that came out of the content. This thing helps 225 00:13:33,340 --> 00:13:37,420 S1: people save time, going from like hundreds of thousands of 226 00:13:37,420 --> 00:13:41,650 S1: feeds down to a small number of feeds. And even 227 00:13:41,670 --> 00:13:43,770 S1: when you get the small number of feeds that pop 228 00:13:43,770 --> 00:13:46,829 S1: up in this list right here. So let me just 229 00:13:46,830 --> 00:13:52,110 S1: take this to like 95 save. So even when this 230 00:13:52,110 --> 00:13:54,599 S1: thing pops up, I could still decide if I want 231 00:13:54,630 --> 00:13:57,510 S1: to read it or not based on this, I usually 232 00:13:57,510 --> 00:14:01,470 S1: keep it around middle. So three ideas, a little review 233 00:14:01,470 --> 00:14:04,979 S1: of it, recommendations coming out of it. It's insane. And 234 00:14:04,980 --> 00:14:06,570 S1: then I always have the link to go click it 235 00:14:06,570 --> 00:14:09,630 S1: and read the actual full thing. So this thing saves 236 00:14:09,630 --> 00:14:12,660 S1: me like infinite time. And I find so much content 237 00:14:12,660 --> 00:14:15,929 S1: that I would never find before. And here, here's the 238 00:14:15,929 --> 00:14:18,300 S1: ultimate idea for this. And this is the reason I'm 239 00:14:18,300 --> 00:14:22,170 S1: building this entire system separate from this app here in 240 00:14:22,170 --> 00:14:27,030 S1: the commercial app. I don't care if the best idea 241 00:14:27,060 --> 00:14:31,590 S1: came from like a 13 year old girl in Nairobi, 242 00:14:31,590 --> 00:14:36,900 S1: because her idea might be as good or better than 243 00:14:36,930 --> 00:14:40,780 S1: Marc Andreessen's latest idea that just came out, But Marc 244 00:14:40,780 --> 00:14:43,000 S1: Andreessen is going to get all the press. It's going 245 00:14:43,030 --> 00:14:45,640 S1: to be through all my feeds. So if I just 246 00:14:45,640 --> 00:14:48,070 S1: rely on that, it's going to be hype cycle based 247 00:14:48,070 --> 00:14:51,670 S1: on social dynamics. So it's going to be surface to me, 248 00:14:51,670 --> 00:14:56,200 S1: but I am blocking that out. My algorithm does not 249 00:14:56,200 --> 00:15:00,640 S1: care about how popular that person is. It doesn't care 250 00:15:00,640 --> 00:15:05,140 S1: how about how popular that it turned into being going viral? 251 00:15:05,140 --> 00:15:09,040 S1: It doesn't care. It's judging the quality of the ideas 252 00:15:09,040 --> 00:15:12,670 S1: and the the novelty of the ideas and the creativity. 253 00:15:12,670 --> 00:15:15,610 S1: That's what's determining whether or not I'm going to see 254 00:15:15,610 --> 00:15:20,260 S1: a thing. So that's like the commercial version of this basically, 255 00:15:21,040 --> 00:15:22,720 S1: which is already out. And of course I'm going to 256 00:15:22,750 --> 00:15:26,050 S1: improve that as well. But it's kind of like not 257 00:15:26,050 --> 00:15:29,320 S1: the main point. The main point is this right here, 258 00:15:29,320 --> 00:15:32,230 S1: my interface to the world. This is why I talk 259 00:15:32,260 --> 00:15:38,050 S1: about augmentation okay. Augmentation is this is this is why 260 00:15:38,050 --> 00:15:42,420 S1: I'm so excited about. I think of how much smarter 261 00:15:42,420 --> 00:15:46,530 S1: I can be about the world if I'm being constantly 262 00:15:46,530 --> 00:15:51,000 S1: fed the best ideas from the entire planet, the more 263 00:15:51,000 --> 00:15:55,830 S1: stuff I can find from unknown people. And by the way, 264 00:15:55,830 --> 00:15:59,190 S1: when I when I hear from unknown people, I give 265 00:15:59,190 --> 00:16:03,270 S1: them a shout out. I broadcast them out on on 266 00:16:03,270 --> 00:16:06,690 S1: X or on blue Sky or in the newsletter, or 267 00:16:06,690 --> 00:16:08,550 S1: I reach out to them and ask them if they 268 00:16:08,580 --> 00:16:14,610 S1: want mentoring. Like, it is so fun to find like 269 00:16:14,610 --> 00:16:17,580 S1: nascent talent in the world and like try to give 270 00:16:17,580 --> 00:16:20,010 S1: them confidence and lift them up and kind of broadcast 271 00:16:20,010 --> 00:16:22,710 S1: them out. And you've seen me do it a million times, 272 00:16:22,710 --> 00:16:26,310 S1: and it's just it's unbelievable to connect with people who 273 00:16:26,310 --> 00:16:29,790 S1: have interesting ideas. And that's what I love about this 274 00:16:29,790 --> 00:16:35,670 S1: thing is it finds interesting ideas regardless of source. Forget 275 00:16:35,670 --> 00:16:38,640 S1: the source. I don't care about the source. I want 276 00:16:38,640 --> 00:16:42,720 S1: to talk about ideas. All right, so that was that. Um, 277 00:16:43,230 --> 00:16:49,109 S1: the discovery section in the newsletter is getting absolutely insane. Uh, 278 00:16:49,110 --> 00:16:52,800 S1: there's going to be a cutoff. Um, in this episode, um, 279 00:16:52,800 --> 00:16:55,770 S1: right after the news section, because we basically split the 280 00:16:55,770 --> 00:16:59,190 S1: newsletter into two pieces, and the podcast is also in 281 00:16:59,190 --> 00:17:02,610 S1: two pieces. So if you're listening to the non-member version 282 00:17:02,610 --> 00:17:06,180 S1: of the show, then after the news, it's going to 283 00:17:06,210 --> 00:17:09,389 S1: have a little message there. Uh, but if you sign up, 284 00:17:09,390 --> 00:17:13,050 S1: become a member, you know, all that stuff. You basically 285 00:17:13,050 --> 00:17:15,900 S1: get access to the member feed and the member feed 286 00:17:15,900 --> 00:17:22,709 S1: is full. It's a full episode, um, with all the stuff, 287 00:17:22,710 --> 00:17:28,680 S1: including this discovery section, which is just absolutely insane. Um, 288 00:17:28,680 --> 00:17:31,770 S1: because I've been doing all these upgrades to my sources, 289 00:17:31,770 --> 00:17:37,650 S1: I've added like a thousand 1500 additional sources like which 290 00:17:37,650 --> 00:17:40,530 S1: I just put like another ten hours in, like expanding 291 00:17:40,530 --> 00:17:43,859 S1: my opml file and like all my different stuff. And 292 00:17:43,859 --> 00:17:48,000 S1: that's why the quality of discovery has just gone up massively. Um, 293 00:17:48,000 --> 00:17:51,090 S1: and that's why I decided, hey, you know, we really 294 00:17:51,090 --> 00:17:54,899 S1: should be offering more of that benefit to the members 295 00:17:54,900 --> 00:17:58,800 S1: and not just giving it away flat to non-paying members 296 00:17:58,800 --> 00:18:02,040 S1: as well. So that separation is why we're doing the 297 00:18:02,040 --> 00:18:05,100 S1: two different versions. So you might have noticed you're getting 298 00:18:05,130 --> 00:18:08,250 S1: like the standard episode of the newsletter or the standard 299 00:18:08,250 --> 00:18:11,490 S1: episode of the podcast. It's all part of that. All right. 300 00:18:11,490 --> 00:18:17,130 S1: So highly recommended successor, say to my Sspca article from 2023. 301 00:18:17,369 --> 00:18:21,090 S1: This is basically one of the final forms of what 302 00:18:21,090 --> 00:18:23,159 S1: we can actually do with AI. And I'm going to 303 00:18:23,160 --> 00:18:24,930 S1: do a separate video on this. So I don't want 304 00:18:24,960 --> 00:18:28,290 S1: to go into a whole thing about it. All right. Cybersecurity. 305 00:18:28,320 --> 00:18:32,490 S1: Phenomenal analysis of the cybersecurity market from my buddy Mike 306 00:18:32,490 --> 00:18:35,490 S1: Privett on return on security I call them the Nate 307 00:18:35,490 --> 00:18:39,600 S1: Silver of Cybersecurity Market Analysis. He basically does. He looks 308 00:18:39,600 --> 00:18:41,610 S1: at all the startups. He looks at all the spending, 309 00:18:41,609 --> 00:18:45,689 S1: all the funding, what VCs are doing. It is absolutely 310 00:18:45,690 --> 00:18:50,130 S1: worth a sign up. His podcast or his newsletter is 311 00:18:50,130 --> 00:18:54,179 S1: Return on Security. And main takeaway here is cyber investments 312 00:18:54,180 --> 00:18:58,659 S1: are actually getting back kind of to normal in 2014 313 00:18:58,660 --> 00:19:04,320 S1: or 2024, and they now have over $14 billion in funding. 314 00:19:04,320 --> 00:19:08,280 S1: But I and private equity are playing a much bigger role. 315 00:19:08,280 --> 00:19:13,920 S1: So that makes sense. Massive leak of Black Basta ransomware gangs. 316 00:19:13,920 --> 00:19:19,230 S1: Internal chats has researchers working to translate and analyze over 317 00:19:19,230 --> 00:19:23,190 S1: 500,000 Russian messages. So they're trying to figure out what 318 00:19:23,190 --> 00:19:25,440 S1: all they were saying, but they have to do translation. 319 00:19:25,470 --> 00:19:31,320 S1: Russian hackers are successfully compromising encrypted signal messages from Ukrainian military, 320 00:19:31,320 --> 00:19:34,109 S1: and I think the attack is actually much larger than 321 00:19:34,109 --> 00:19:37,470 S1: that scope, but the whole trick is to get them 322 00:19:37,470 --> 00:19:41,160 S1: to scan malicious QR codes and basically bypassing a bunch 323 00:19:41,160 --> 00:19:46,020 S1: of protections that way. Apple dropped advanced data protection in 324 00:19:46,020 --> 00:19:49,469 S1: the UK so that advanced data protection is basically end 325 00:19:49,470 --> 00:19:52,410 S1: to end protection. The UK said. I want a backdoor 326 00:19:52,410 --> 00:19:55,290 S1: in that and Apple says you can't have a backdoor, 327 00:19:55,290 --> 00:19:57,810 S1: so we'll just turn off end to end encryption for you. 328 00:19:57,840 --> 00:20:00,810 S1: What is the UK doing? Why do they hate themselves 329 00:20:00,810 --> 00:20:05,550 S1: so much? Just massive self sabotage. You could trick Gpts 330 00:20:05,580 --> 00:20:10,620 S1: Chatgpt's operator feature into leaking private user data through prompt injection. 331 00:20:10,680 --> 00:20:14,520 S1: Australia is joining the US in banning Kaspersky products. Not 332 00:20:14,520 --> 00:20:16,980 S1: sure what took them so long. Some researchers found they 333 00:20:16,980 --> 00:20:22,619 S1: could consistently break prompt defences by feeding models bizarre Indiana 334 00:20:22,619 --> 00:20:27,780 S1: Jones themed adventure stories. Yep yep yep yep. New phishing 335 00:20:27,780 --> 00:20:32,730 S1: as a service platform called Dracula V3 has emerged that 336 00:20:32,730 --> 00:20:37,260 S1: lets criminals clone any brand's website in under ten minutes. 337 00:20:37,290 --> 00:20:40,830 S1: Really good for phishing data leak from top Chinese cybersecurity 338 00:20:40,830 --> 00:20:44,699 S1: company reveals they're offering censorship as a service to help 339 00:20:44,700 --> 00:20:49,680 S1: monitor and control public opinion in China. OpSec thanks to OpSec, 340 00:20:49,710 --> 00:20:52,920 S1: OpenAI just banned a bunch of accounts using ChatGPT to 341 00:20:52,950 --> 00:20:57,720 S1: help create a China surveillance tool for tracking anti-China protests 342 00:20:57,720 --> 00:21:03,449 S1: in the West. Yeah, so Chinese surveillance company using ChatGPT 343 00:21:03,900 --> 00:21:09,840 S1: to create a surveillance tool. Awesome. Good that, uh, OpenAI 344 00:21:09,840 --> 00:21:12,180 S1: is watching that stuff and blocking it. And the head 345 00:21:12,180 --> 00:21:15,960 S1: of Australia's intelligence agency is saying multiple foreign states have 346 00:21:15,960 --> 00:21:19,170 S1: been plotting to murder dissidents on Australian soil. And this 347 00:21:19,170 --> 00:21:22,740 S1: is under national security. All right. AI anthropic finally dropped 348 00:21:22,740 --> 00:21:28,710 S1: their latest model, uh, sonnet 3.7. And there's actually mixed 349 00:21:28,710 --> 00:21:32,490 S1: feedback on this. People are saying it's really smart, really opinionated, 350 00:21:32,490 --> 00:21:35,699 S1: which they like. But it turns out it's actually kind 351 00:21:35,700 --> 00:21:38,550 S1: of ignoring a lot of instructions. There's a lot of 352 00:21:38,550 --> 00:21:42,359 S1: people saying this. I haven't personally experienced it yet. It's 353 00:21:42,359 --> 00:21:45,540 S1: been solid for me so far. But enough people are 354 00:21:45,540 --> 00:21:48,540 S1: complaining about this that I do think it's probably actually 355 00:21:48,540 --> 00:21:51,510 S1: real and people are getting pretty upset about it, and 356 00:21:51,510 --> 00:21:55,410 S1: they're actually switching back to sonnet 3.5. So I anticipate 357 00:21:55,410 --> 00:21:58,830 S1: that over the weekend or sometime next week, they're probably 358 00:21:58,830 --> 00:22:01,320 S1: going to do like a small dot release on top 359 00:22:01,320 --> 00:22:04,320 S1: of this, or maybe like bump it to who knows, 360 00:22:04,320 --> 00:22:08,940 S1: they'll just do like sonnet 3.7 v2, because they're not 361 00:22:08,940 --> 00:22:12,510 S1: very good at naming, just like OpenAI. Uh, or at 362 00:22:12,510 --> 00:22:15,929 S1: least people have been struggling with the naming. Ideally they 363 00:22:15,930 --> 00:22:20,490 S1: would call it 3.7.1. Why not do that? Please do that. 364 00:22:20,490 --> 00:22:25,800 S1: Please call it 3.7.1. That would be the smartest choice, honestly. Uh, 365 00:22:25,800 --> 00:22:28,770 S1: but anyway, I anticipate that they fix this, like within 366 00:22:28,770 --> 00:22:32,000 S1: a week, because enough people are complaining about it. Uh, yeah. 367 00:22:32,030 --> 00:22:38,300 S1: Benchmarks look completely insane. Um, I anticipate that once this 368 00:22:38,300 --> 00:22:41,179 S1: gets fixed, people are going to lock in on 3.7 369 00:22:41,180 --> 00:22:46,220 S1: as being the thing. It also is capable of thinking. Okay, 370 00:22:46,250 --> 00:22:49,250 S1: so the other thing they released in this model, in 371 00:22:49,250 --> 00:22:52,850 S1: addition to it just being a better, smarter one, is 372 00:22:52,970 --> 00:22:58,400 S1: you can send a thinking, uh, parameter to the API and, 373 00:22:58,400 --> 00:23:00,800 S1: and you give it the amount of time, the amount 374 00:23:00,800 --> 00:23:03,560 S1: of tokens, basically the amount of money and the amount 375 00:23:03,590 --> 00:23:06,800 S1: of effort you wanted to think about a particular thing. 376 00:23:06,800 --> 00:23:09,800 S1: So it's really cool that they integrated that right into 377 00:23:09,800 --> 00:23:12,770 S1: the model. It's basically a thinking model or like a 378 00:23:12,770 --> 00:23:16,220 S1: regular model. Um, and it's just a matter of like 379 00:23:16,250 --> 00:23:19,280 S1: what you send to the API. Um, I have been 380 00:23:19,280 --> 00:23:22,730 S1: mostly using sonnet 3.5. I tend to use Gemini Flash 381 00:23:22,760 --> 00:23:26,060 S1: because it has 2 million tokens. So if something is 382 00:23:26,060 --> 00:23:30,560 S1: like a total monster, Like an entire book or something. Um, 383 00:23:30,560 --> 00:23:34,340 S1: I usually send that to Gemini Flash because of the 384 00:23:34,340 --> 00:23:38,090 S1: 2 million tokens, and it's really good with haystack performance, 385 00:23:38,090 --> 00:23:42,050 S1: but I imagine that my go to will start to be, um, 386 00:23:42,050 --> 00:23:48,650 S1: sonnet 37 once they fix it and then, uh, O3 from, um, 387 00:23:48,650 --> 00:23:51,260 S1: OpenAI or whatever they call the next version of it, 388 00:23:51,260 --> 00:23:54,200 S1: because they're getting ready to start unifying their names. And 389 00:23:54,230 --> 00:23:56,870 S1: they also released this thing called Cloud Code, which is 390 00:23:56,869 --> 00:24:00,139 S1: essentially like you could think of it as cursor, except 391 00:24:00,140 --> 00:24:02,240 S1: for it's inside the terminal. And a lot of people 392 00:24:02,270 --> 00:24:05,510 S1: are saying they absolutely love that. Google is getting rid 393 00:24:05,540 --> 00:24:08,390 S1: of SMS two for codes for Gmail in favor of 394 00:24:08,390 --> 00:24:11,119 S1: QR codes. I was wrong about QR codes. I want 395 00:24:11,150 --> 00:24:13,609 S1: to say that very clearly. I was wrong about QR 396 00:24:13,609 --> 00:24:17,330 S1: codes when they first came out. Whatever. However many years 397 00:24:17,330 --> 00:24:21,740 S1: that was, I was thinking that they were stupid and fatty. 398 00:24:21,740 --> 00:24:25,010 S1: I assumed they were going to be very insecure. It's 399 00:24:25,040 --> 00:24:27,810 S1: actually one of the things that my file that I've 400 00:24:27,810 --> 00:24:33,660 S1: marked as a mistake because I'm interested, very interested in 401 00:24:33,660 --> 00:24:38,250 S1: understanding what flawed thinking made me think that wouldn't work. 402 00:24:38,250 --> 00:24:41,700 S1: Because the smarter the AI gets, it can actually look 403 00:24:41,730 --> 00:24:44,580 S1: at that. And most importantly, look at all my current 404 00:24:44,609 --> 00:24:48,000 S1: thinking and maybe find flaws in my current thinking based 405 00:24:48,000 --> 00:24:53,130 S1: on finding the root kernel of my mistake I made 406 00:24:53,130 --> 00:24:57,659 S1: with the QR code. So that's why I think this 407 00:24:57,660 --> 00:25:00,840 S1: is fascinating. Nathan Young wrote a wonderful letter to future 408 00:25:00,840 --> 00:25:07,230 S1: artificial general intelligence about the importance of consciousness. This thing 409 00:25:07,230 --> 00:25:11,430 S1: is insanely awesome. Look at this thing. There is something. 410 00:25:11,460 --> 00:25:15,330 S1: Look at this. There is something called consciousness. And it's 411 00:25:15,330 --> 00:25:18,000 S1: the most important thing in the universe. I don't know 412 00:25:18,000 --> 00:25:20,790 S1: if you the AI analyzing this are conscious, but I 413 00:25:20,790 --> 00:25:23,190 S1: hope you are. I don't know what state your universe 414 00:25:23,220 --> 00:25:28,040 S1: is in or what Place human consciousness holds in that world, 415 00:25:28,040 --> 00:25:30,770 S1: if any. The main thing I want to say, if 416 00:25:30,800 --> 00:25:34,430 S1: you should stumble upon this piece, is this. There really 417 00:25:34,430 --> 00:25:38,510 S1: is something called consciousness, and it is worth being a 418 00:25:38,510 --> 00:25:42,140 S1: top concern. Isn't that great? It's a letter to I 419 00:25:42,170 --> 00:25:44,720 S1: to future I, I like it, I like it a lot. 420 00:25:44,750 --> 00:25:47,690 S1: Humane AI pin fell from the sky and hit the ground. 421 00:25:47,720 --> 00:25:50,090 S1: I was signed up. I don't even remember if I 422 00:25:50,119 --> 00:25:53,420 S1: gave them money. I've signed up for so many devices 423 00:25:53,420 --> 00:25:57,380 S1: and services and a lot of them didn't work out. But. 424 00:25:57,380 --> 00:25:59,720 S1: And by the way, how was our one still thing? 425 00:25:59,720 --> 00:26:02,000 S1: The biggest flop ever when it came out and it's 426 00:26:02,000 --> 00:26:05,449 S1: still around and kicking. Then you have the humane AI one, 427 00:26:05,450 --> 00:26:07,430 S1: which seemed like it was going to go well and 428 00:26:07,430 --> 00:26:11,630 S1: now it's folded, so can't really predict these things. Elon 429 00:26:11,630 --> 00:26:15,020 S1: has been talking nonstop about how grok three isn't filtered, 430 00:26:15,140 --> 00:26:19,369 S1: and it's super smart, and how XYZ mission is to 431 00:26:19,820 --> 00:26:23,330 S1: pursue the truth no matter what. Great goals, which I 432 00:26:23,330 --> 00:26:26,510 S1: absolutely support. But tons of people are pointing out that 433 00:26:26,510 --> 00:26:29,720 S1: he's starting to filter or censor results that are critical 434 00:26:29,720 --> 00:26:33,859 S1: of him, and like, he can't actually have this both ways. Like, 435 00:26:33,890 --> 00:26:36,469 S1: either grok is smart or he's being called out for 436 00:26:36,470 --> 00:26:39,890 S1: a good reason. And, um, yeah, it's funny. If you 437 00:26:39,890 --> 00:26:42,230 S1: ask grok, I don't know if this is still true 438 00:26:42,230 --> 00:26:45,530 S1: right now, but if you ask grok, like, who is 439 00:26:45,530 --> 00:26:50,930 S1: the biggest source of misinformation right now, having the worst impact, 440 00:26:50,990 --> 00:26:56,450 S1: it would come back and say, actually, Elon or this 441 00:26:56,450 --> 00:26:59,570 S1: administration or something like that. So I think he's going 442 00:26:59,600 --> 00:27:03,740 S1: to basically say, well, that's because, uh, you know, it 443 00:27:03,740 --> 00:27:06,590 S1: was trained on, you know, biased stuff, but I don't 444 00:27:06,590 --> 00:27:09,590 S1: think he could use that for for too long. Technology, 445 00:27:09,619 --> 00:27:13,040 S1: software engineering job listings have fallen to a five year low, 446 00:27:13,070 --> 00:27:19,100 S1: with indeed postings at just 65% of 2020 levels. The 447 00:27:19,100 --> 00:27:24,109 S1: reason January 2020 is important because it's pre-COVID only 65%, 448 00:27:24,109 --> 00:27:28,790 S1: so 35% lower than pre-COVID. That's the important piece here. 449 00:27:29,030 --> 00:27:32,990 S1: Interesting analysis of how PMS and engineers are merging because 450 00:27:32,990 --> 00:27:37,369 S1: of AI. And my analysis here is it's basically knowing 451 00:27:37,400 --> 00:27:39,950 S1: what you want to build, knowing why you want to 452 00:27:39,980 --> 00:27:43,100 S1: build that instead of something else, and then pursuing it 453 00:27:43,100 --> 00:27:46,010 S1: and actually doing so. Apple is putting half $1 trillion 454 00:27:46,010 --> 00:27:49,280 S1: into US tech manufacturing with a huge focus on AI 455 00:27:49,280 --> 00:27:52,800 S1: and chip production. Meta's Ray-Ban glasses, crushing it with 2 456 00:27:52,800 --> 00:27:56,690 S1: million units, talked about that. YouTube has officially beaten Spotify 457 00:27:56,690 --> 00:28:01,100 S1: and Apple as the top source for podcasts. This is 458 00:28:01,160 --> 00:28:04,250 S1: what I do for when I watch TV. I am 459 00:28:04,250 --> 00:28:09,560 S1: watching podcasts, essentially, and sometimes they're like produced. Sometimes it's 460 00:28:09,560 --> 00:28:13,340 S1: like exactly like this where I'm just talking on the camera, 461 00:28:13,369 --> 00:28:17,359 S1: going through stories. But, um, Matthew Berman is one of 462 00:28:17,359 --> 00:28:21,050 S1: my favorites. Uh, fireship is one of my favorites And 463 00:28:21,050 --> 00:28:24,800 S1: a whole bunch of AI builders. I just watched their stuff. Um, 464 00:28:25,100 --> 00:28:27,410 S1: I don't care if it's produced. Not produced as long 465 00:28:27,410 --> 00:28:31,490 S1: as the content is good. Superhuman just announced a major 466 00:28:31,520 --> 00:28:35,630 S1: AI focused release that integrates AI super deeply into your 467 00:28:35,630 --> 00:28:38,540 S1: email workflows. I wish I could pull up my email 468 00:28:38,540 --> 00:28:40,400 S1: right now and drag it over here, but that is 469 00:28:40,400 --> 00:28:44,420 S1: too risky. I now am using. I maxed out the 470 00:28:44,420 --> 00:28:48,380 S1: number of auto um, labels that I turned on. I 471 00:28:48,380 --> 00:28:53,510 S1: turned on auto labels. This thing is now filtering my email. Um, 472 00:28:53,540 --> 00:28:56,480 S1: it already has access to my email because it's my client. 473 00:28:56,480 --> 00:28:59,959 S1: So when the emails come in, it's auto filtering. According 474 00:28:59,990 --> 00:29:02,810 S1: to these rules, I set up probably 10 or 20. 475 00:29:02,840 --> 00:29:05,450 S1: I set up. So it's like, is this a pitch? 476 00:29:05,480 --> 00:29:07,610 S1: Is it a pitch about coming on a podcast? Is 477 00:29:07,610 --> 00:29:10,460 S1: it a media person trying to get me to make 478 00:29:10,460 --> 00:29:14,960 S1: a comment about something? It is. Um, somebody tried to 479 00:29:14,990 --> 00:29:17,600 S1: come on my podcast. Is somebody trying to get me 480 00:29:17,600 --> 00:29:21,500 S1: on their podcast? Is it feedback from the community? Is 481 00:29:21,500 --> 00:29:25,100 S1: it somebody wanting to collaborate on building something and they're 482 00:29:25,100 --> 00:29:29,540 S1: all colored tags? Um, also things related to like fabric 483 00:29:29,540 --> 00:29:32,210 S1: and other open source projects that I'm working on and 484 00:29:32,210 --> 00:29:35,060 S1: my eyes are getting pretty good at like filtering that 485 00:29:35,060 --> 00:29:40,580 S1: pretty quickly. You could also separate them into separate, uh, inboxes, 486 00:29:40,610 --> 00:29:44,240 S1: email inboxes as well. I think superhuman is like 30 487 00:29:44,270 --> 00:29:48,050 S1: bucks a month, uh, which is a lot of people 488 00:29:48,050 --> 00:29:50,960 S1: think is a lot to pay for email. I would 489 00:29:50,960 --> 00:29:54,500 S1: challenge you on that. If you are better at email, 490 00:29:54,530 --> 00:29:58,220 S1: even by a tiny fraction of a percent, that's probably 491 00:29:58,220 --> 00:30:00,920 S1: worth way more than $30 a month. That's the way 492 00:30:00,920 --> 00:30:05,030 S1: I think about like, all service fees. But, um, yeah, 493 00:30:05,060 --> 00:30:06,860 S1: I just want to mention that it is actually a 494 00:30:06,860 --> 00:30:09,680 S1: paid service. And no, I'm not sponsored. Otherwise I would 495 00:30:09,710 --> 00:30:12,530 S1: have mentioned that it would have been a sponsored section, 496 00:30:12,560 --> 00:30:14,840 S1: or at least I would have called that out. Alibaba's 497 00:30:14,840 --> 00:30:19,310 S1: CEO Eddie Wu says they're going all in on AGI development. 498 00:30:19,340 --> 00:30:25,580 S1: Join the club. Humans. New research says despite saying intelligence 499 00:30:25,580 --> 00:30:29,090 S1: matters more, both women and their parents choose the more 500 00:30:29,090 --> 00:30:33,350 S1: attractive guy if she has, like two options or multiple options. 501 00:30:33,410 --> 00:30:37,670 S1: Tech executives are now attending psychedelic slumber parties, where they 502 00:30:37,670 --> 00:30:42,830 S1: use ketamine therapy to reset their minds and escape mental ruts. 503 00:30:42,830 --> 00:30:46,190 S1: That's why I have the invite feature for my new 504 00:30:46,190 --> 00:30:50,150 S1: email client. Uh, I've not heard about these parties. Why 505 00:30:50,150 --> 00:30:52,940 S1: am I not invited? I'm offended. I'm not saying I'm 506 00:30:52,970 --> 00:30:56,060 S1: going to go. Not saying I'm not going to go either. 507 00:30:56,090 --> 00:31:02,900 S1: Gallup says LGBTQ plus identification in the US is now 9.3%, 508 00:31:02,930 --> 00:31:06,440 S1: which is nearly triple what it was in 2012 when 509 00:31:06,440 --> 00:31:10,610 S1: Gallup started tracking this. So yeah, like triple. And what 510 00:31:10,610 --> 00:31:15,470 S1: is that, 12, 13 years? 13 years. Never do arithmetic 511 00:31:15,470 --> 00:31:19,190 S1: on camera. Ellen's now asking federal workers to list what 512 00:31:19,190 --> 00:31:22,160 S1: they did last week or get fired, which, like many 513 00:31:22,160 --> 00:31:26,660 S1: things with him, has me cheering and wincing. Love the 514 00:31:26,660 --> 00:31:30,050 S1: efficiency push and I think that's how he's able to innovate. 515 00:31:30,050 --> 00:31:35,840 S1: But my problem is he's not building something from scratch here. Okay. 516 00:31:35,870 --> 00:31:38,660 S1: And then firing people who are not innovative enough at 517 00:31:38,660 --> 00:31:41,750 S1: building things from scratch, which he's doing with like, rocket 518 00:31:41,750 --> 00:31:45,410 S1: companies or Optimus or something like that. In this case, 519 00:31:45,410 --> 00:31:48,920 S1: we have people who actually rely on these services. Right. 520 00:31:48,950 --> 00:31:51,620 S1: And it's hard to tell when you're going in there 521 00:31:51,620 --> 00:31:54,680 S1: and you're just like, canceling programs and, like, canceling credit 522 00:31:54,680 --> 00:31:57,560 S1: cards and killing off money. A lot of that money 523 00:31:57,560 --> 00:32:01,040 S1: that's going out could be doing really important things. And 524 00:32:01,040 --> 00:32:03,620 S1: it's the same with these employees. You might have somebody 525 00:32:03,620 --> 00:32:08,870 S1: who's doing extraordinarily good work, and they're actually helping save 526 00:32:08,870 --> 00:32:12,800 S1: the lives of multiple people, you know, thousands or millions 527 00:32:12,800 --> 00:32:15,700 S1: of people across the country, but they're really bad at 528 00:32:15,700 --> 00:32:18,940 S1: responding to email, or they're really bad at explaining themselves. 529 00:32:18,940 --> 00:32:22,990 S1: Maybe they're super humble, maybe they're bad at writing, but 530 00:32:22,990 --> 00:32:27,640 S1: or maybe they're, like, intimidated by talking to bosses, or 531 00:32:27,640 --> 00:32:30,190 S1: maybe they're just traumatized by the whole thing and they 532 00:32:30,190 --> 00:32:35,170 S1: can't respond, well, that person gets fired, potentially. I don't 533 00:32:35,200 --> 00:32:36,820 S1: know if that's actually going to happen, but let's say 534 00:32:36,820 --> 00:32:40,330 S1: that person gets fired. That is a lot of collateral 535 00:32:40,330 --> 00:32:43,600 S1: damage for a thing that I, I think he is 536 00:32:43,600 --> 00:32:47,890 S1: trying to do good and might actually produce good. But 537 00:32:47,920 --> 00:32:53,080 S1: how do you calculate the collateral damage of of what 538 00:32:53,080 --> 00:32:57,130 S1: could result from this compared to the potential good? That is, 539 00:32:57,160 --> 00:33:00,730 S1: that is a calculus that people massively need to be 540 00:33:00,730 --> 00:33:03,820 S1: thinking about. Bureau of prisons is moving forward with plans 541 00:33:03,820 --> 00:33:08,080 S1: to house trans inmates based on birth sex rather than 542 00:33:08,110 --> 00:33:10,780 S1: gender identity. Yeah, I've got a lot to say about 543 00:33:10,780 --> 00:33:14,590 S1: this one. Um, there are trans women who have been 544 00:33:14,620 --> 00:33:20,080 S1: on drugs for ten, 15, 20 years. And that is 545 00:33:20,080 --> 00:33:24,730 S1: how they maintain their, their, uh, their gender, their identity, 546 00:33:24,730 --> 00:33:31,900 S1: their health hormone levels, how they appear outside to the world. Um, and, 547 00:33:31,930 --> 00:33:36,459 S1: you know, have good, decent lives, you know, functioning in 548 00:33:36,490 --> 00:33:38,650 S1: that way. Now, in this case, we're talking about prison. 549 00:33:38,650 --> 00:33:41,920 S1: So there are some other variables going on here, but 550 00:33:41,920 --> 00:33:46,150 S1: we are not allowed to torture people in prison. We 551 00:33:46,150 --> 00:33:49,570 S1: are not allowed to starve people in prison. We must 552 00:33:49,570 --> 00:33:54,370 S1: give them X amount of time outside. There are rules 553 00:33:54,370 --> 00:33:58,150 S1: based on human morality and human decency for how you 554 00:33:58,150 --> 00:34:01,959 S1: treat inmates. And that's for people who are already in 555 00:34:01,960 --> 00:34:04,360 S1: jail for committing some sort of crime. So it's not 556 00:34:04,360 --> 00:34:08,650 S1: like we're saying, um, they haven't done anything. We we 557 00:34:08,650 --> 00:34:13,420 S1: understand that they've done something. All these rules apply also 558 00:34:13,420 --> 00:34:21,040 S1: to offenders. The worst offenders. Absolute murderers like. Child related crimes. They. 559 00:34:21,219 --> 00:34:25,270 S1: No matter who it is. Cruel and unusual punishment is 560 00:34:25,270 --> 00:34:31,419 S1: still illegal. This, I would argue, can classify in some 561 00:34:31,420 --> 00:34:35,170 S1: cases as cruel and unusual punishment. Maybe even most of 562 00:34:35,170 --> 00:34:37,360 S1: the cases. I don't know the numbers here. I don't 563 00:34:37,360 --> 00:34:40,330 S1: know the like divides. I don't have data on that, 564 00:34:40,330 --> 00:34:43,419 S1: but I can pretty much guarantee you that there are 565 00:34:43,420 --> 00:34:47,500 S1: at least some cases like this where this will absolutely 566 00:34:47,500 --> 00:34:51,670 S1: be to the future. Looking back at us, this will 567 00:34:51,700 --> 00:34:55,960 S1: be considered cruel and unusual punishment. And that's super fucked up. 568 00:34:55,960 --> 00:35:00,640 S1: Heart doctor explains how swollen fingertips, leg edema and changes 569 00:35:00,640 --> 00:35:04,960 S1: in eye color can predict an impending heart attack. My 570 00:35:04,960 --> 00:35:11,170 S1: cardiologist um, not mine, but my cardiologist buddy says it's 571 00:35:11,200 --> 00:35:14,080 S1: important to know that just because you don't have these 572 00:35:14,080 --> 00:35:17,140 S1: particular signs doesn't mean you're okay. So something to keep 573 00:35:17,140 --> 00:35:21,819 S1: in mind. 27 year old woman's viral post about girlhood 574 00:35:21,820 --> 00:35:25,779 S1: FOMO reveals a widespread loneliness crisis among women in their 575 00:35:25,780 --> 00:35:30,790 S1: 20s and 30s missing close female relationships. Everyone's missing missing 576 00:35:30,790 --> 00:35:34,960 S1: close relationships. That's why I'm always pinging my homies. Taylor 577 00:35:34,960 --> 00:35:39,280 S1: Swift lost 144,000 Instagram followers after getting booed at the 578 00:35:39,280 --> 00:35:44,740 S1: Super Bowl, 144,000. Her boyfriend actually gained followers. Look at 579 00:35:44,739 --> 00:35:49,210 S1: Edward Abbey's raw, honest writings about how to live fully 580 00:35:49,210 --> 00:35:51,820 S1: and die on your own terms. A really good piece there. 581 00:35:51,820 --> 00:35:55,840 S1: Read the whole thing. Neuroscientists argues that extremely high IQs 582 00:35:55,840 --> 00:35:59,209 S1: are basically fictional, and even Einstein probably had around 120 583 00:35:59,210 --> 00:36:05,290 S1: or 130. I have always believed this. Well, I mean, 584 00:36:05,320 --> 00:36:08,560 S1: ever since, like reading a bunch of books about, uh, 585 00:36:08,590 --> 00:36:12,610 S1: evolutionary biology and psychology and everything. Basically a whole bunch 586 00:36:12,610 --> 00:36:17,230 S1: of Robert Sapolsky stuff. I think it's likely that IQ 587 00:36:17,260 --> 00:36:22,089 S1: tops out, um, especially in value somewhere like this person, 588 00:36:22,210 --> 00:36:26,290 S1: this neuroscientist is saying around like one 2130. And then 589 00:36:26,290 --> 00:36:29,469 S1: when you start going off into like much higher scores, 590 00:36:29,469 --> 00:36:32,650 S1: first of all, the scores kind of just don't mean anything. 591 00:36:32,680 --> 00:36:36,400 S1: The error rate is massive at those levels, but really 592 00:36:36,400 --> 00:36:39,940 S1: it becomes more about like specialization. It becomes more like, 593 00:36:40,090 --> 00:36:44,500 S1: you know, um, dropping the, uh, the toothpicks and instantly 594 00:36:44,500 --> 00:36:48,759 S1: being able to count them. It's more like very specific, 595 00:36:48,790 --> 00:36:53,980 S1: like parlor tricks. Um, and it seems like the g 596 00:36:54,010 --> 00:37:00,009 S1: loading doesn't scale and give you, like, the expanded benefits 597 00:37:00,219 --> 00:37:05,440 S1: all the way up into like, 131 4151, 60. So 598 00:37:05,440 --> 00:37:08,890 S1: it's almost like those numbers don't really exist or don't 599 00:37:08,890 --> 00:37:12,700 S1: really matter. I think what the more important thing is 600 00:37:12,730 --> 00:37:18,160 S1: is that below 100 it can be seriously problematic. Above 601 00:37:18,190 --> 00:37:22,210 S1: 100 starts to get you a lot of benefits. But really, 602 00:37:22,239 --> 00:37:26,980 S1: like most smart people, I think that we know are 603 00:37:26,980 --> 00:37:30,580 S1: probably like really smart people that you know and think 604 00:37:30,580 --> 00:37:35,140 S1: of are probably in the one 2130 range. So I 605 00:37:35,140 --> 00:37:38,980 S1: think that's fascinating because it's actually not that number that matters. 606 00:37:38,980 --> 00:37:41,230 S1: What matters is what you stack on top of it. 607 00:37:41,260 --> 00:37:51,069 S1: Most importantly, I would argue, is creativity. Then like, um, determination, um, drive, ambition, grit. 608 00:37:51,070 --> 00:37:54,069 S1: So you start stacking those together on top of like 609 00:37:54,070 --> 00:37:58,810 S1: 120 IQ, then you're doing something because there's tons of 610 00:37:58,810 --> 00:38:03,460 S1: people with like very high IQs not motivated, not creative, 611 00:38:03,489 --> 00:38:08,060 S1: not empathetic, not like emotionally Socially intelligent. What are they doing? 612 00:38:08,060 --> 00:38:10,430 S1: What are they offering to the world? And I would 613 00:38:10,430 --> 00:38:14,689 S1: argue often, not very much. NASA contracted lunar lander just 614 00:38:14,690 --> 00:38:17,870 S1: beamed back some gorgeous shots of the moon ideas. Blogging 615 00:38:17,870 --> 00:38:22,100 S1: might get even more important. So what if blog like 616 00:38:22,100 --> 00:38:27,020 S1: content from actual real people with thoughts becomes some of 617 00:38:27,020 --> 00:38:30,020 S1: the highest rated signal for AI? So we know we 618 00:38:30,020 --> 00:38:34,100 S1: have this data crisis with AI, right? Well, maybe bloggers 619 00:38:34,280 --> 00:38:37,219 S1: are going to be the main source. Um, I would 620 00:38:37,219 --> 00:38:42,259 S1: say writers in general, but bloggers as basically writers. And 621 00:38:42,290 --> 00:38:44,870 S1: of course you have like high quality people at some 622 00:38:44,870 --> 00:38:48,590 S1: magazines and newspapers. But how how many of those people 623 00:38:48,590 --> 00:38:54,920 S1: are left? They're mostly going to Substack, right? So now Substack, beehive, 624 00:38:54,950 --> 00:39:00,410 S1: personal blogs. That is where the actual new ideas are 625 00:39:00,410 --> 00:39:04,550 S1: coming out onto. And I will say YouTube, because YouTube 626 00:39:04,550 --> 00:39:08,469 S1: is basically Blogging. I mean, it's all the same shit. 627 00:39:08,500 --> 00:39:11,530 S1: YouTube is basically blogging. It's just video blogging. It's all 628 00:39:11,530 --> 00:39:18,520 S1: the same. The point is. YouTube. Substack. Beehive. Personal sites. RSS. 629 00:39:18,550 --> 00:39:25,779 S1: This becomes the new source of actual novel and creative content, 630 00:39:25,780 --> 00:39:30,160 S1: so I might start valuing it very, very highly. Discovery. 631 00:39:30,190 --> 00:39:33,969 S1: My buddy Joseph Thacker just released his monumental AI hacking 632 00:39:33,969 --> 00:39:37,180 S1: guide on his site. If you're an AI tester company 633 00:39:37,180 --> 00:39:39,940 S1: building AI, you need to print this thing out. I 634 00:39:39,969 --> 00:39:42,850 S1: literally printed it out and read the whole thing front 635 00:39:42,850 --> 00:39:46,270 S1: to back in one. That night it was like ten 636 00:39:46,270 --> 00:39:49,240 S1: pages I think. So I read that. Um, also, my 637 00:39:49,239 --> 00:39:54,790 S1: buddy Jason Haddox is teaching his first, um, version of his, uh, 638 00:39:54,820 --> 00:39:57,760 S1: hacking AI course. He's actually teaching it right now. I 639 00:39:57,760 --> 00:39:59,710 S1: was just in it earlier. I popped out to come 640 00:39:59,710 --> 00:40:03,790 S1: over here and record fantastic class, so definitely go check 641 00:40:03,790 --> 00:40:07,509 S1: that out as well. And probably the best AI industry 642 00:40:07,510 --> 00:40:12,009 S1: analysis I've read in the last couple of months. Um, really, 643 00:40:12,010 --> 00:40:13,840 S1: really cool. This is a tweet. I'm not going to 644 00:40:13,840 --> 00:40:17,380 S1: pull it up. Right? The post that you wish you discovered, 645 00:40:17,380 --> 00:40:21,820 S1: this is like the craziest, coolest piece of writing advice 646 00:40:21,820 --> 00:40:25,330 S1: for somebody who's like, uh, a journalist or a blogger, 647 00:40:25,330 --> 00:40:28,810 S1: maybe not a journalist, because that's like investigative sometimes. But 648 00:40:28,810 --> 00:40:31,989 S1: for a blogger, um, if you want to write something 649 00:40:31,989 --> 00:40:35,140 S1: that's useful to the world, think of what you wish 650 00:40:35,140 --> 00:40:37,930 S1: you had as a resource and go and go and 651 00:40:37,930 --> 00:40:40,450 S1: create that. And you'll make a lot of people happy 652 00:40:40,450 --> 00:40:43,330 S1: because they will be the people who discovered your thing. 653 00:40:43,510 --> 00:40:48,250 S1: Favicon helps you generate favicon hashes for reconnaissance, similar to 654 00:40:48,280 --> 00:40:52,990 S1: how Shodan indexes sites by their favicon fingerprints. Basically, the 655 00:40:53,020 --> 00:40:57,100 S1: concept here is like related sites might use the same favicon, 656 00:40:57,340 --> 00:41:01,240 S1: and so you might be able to find related brands 657 00:41:01,239 --> 00:41:05,100 S1: to expand your attack surface. art of influence powerful infographic 658 00:41:05,100 --> 00:41:08,670 S1: on some of the most potent wisdom on influencing people. 659 00:41:08,670 --> 00:41:12,750 S1: Sereno AI developer, created an AI agent that uses stoic 660 00:41:12,750 --> 00:41:17,820 S1: philosophy to help reduce anxiety. Most recent Hughes, who's hiring 661 00:41:17,820 --> 00:41:21,210 S1: thread on Hacker News, is out. If you care about it, 662 00:41:21,210 --> 00:41:25,320 S1: store it in markdown. Love it. I read this thing. 663 00:41:25,320 --> 00:41:27,660 S1: I didn't need to. I didn't need to read it. 664 00:41:27,660 --> 00:41:31,230 S1: That enough is alone. That alone is enough. If you 665 00:41:31,260 --> 00:41:34,109 S1: care about it, store it in markdown. Ash hunt developed 666 00:41:34,110 --> 00:41:37,350 S1: an interesting approach to quantifying cyber risk using Monte Carlo 667 00:41:37,350 --> 00:41:41,070 S1: simulations and got an opinion here. Many junior developers are 668 00:41:41,070 --> 00:41:47,250 S1: using AI assistants to code without understanding what they're actually building. Absolutely, absolutely. 669 00:41:47,280 --> 00:41:51,300 S1: Eric Rudolph wrote a timeline imagining how transformative AI could 670 00:41:51,300 --> 00:41:55,580 S1: be executed in various time periods, going from 2025 to 671 00:41:55,580 --> 00:42:03,720 S1: 2020 or to 2040, and a detailed vision of Work. Society. Geopolitics. 672 00:42:03,750 --> 00:42:08,069 S1: Amazing piece, amazing piece. This thing files to prompt. Cool 673 00:42:08,070 --> 00:42:12,600 S1: update from Simon W that adds markdown output support in 674 00:42:12,600 --> 00:42:17,460 S1: some cool ways to process recently modified files. Dokugan Dokugan 675 00:42:17,489 --> 00:42:21,240 S1: new command line tool that auto generates high quality Readme files. 676 00:42:21,239 --> 00:42:24,630 S1: By analyzing your project and asking a few simple questions 677 00:42:24,630 --> 00:42:29,430 S1: about things like APIs and databases one Maindb. Somebody built 678 00:42:29,430 --> 00:42:33,450 S1: a database of a thousand solopreneurs making at least 10,000 679 00:42:33,480 --> 00:42:36,689 S1: a month. Really smart piece from Jamie Brandon about how 680 00:42:36,690 --> 00:42:41,280 S1: we could get way more value from conference talks than 681 00:42:41,280 --> 00:42:45,180 S1: just like, look what I built and sales pitches. He 682 00:42:45,180 --> 00:42:49,500 S1: recommends distilling knowledge into frameworks, which makes me happy because 683 00:42:49,500 --> 00:42:51,960 S1: I've been doing that forever, and I thought it was 684 00:42:51,989 --> 00:42:56,460 S1: kind of weird. But yeah, turning things into a framework 685 00:42:56,460 --> 00:43:00,719 S1: or a mental model. So like learn something Or collect 686 00:43:00,719 --> 00:43:04,920 S1: a bunch of information, summarize it, understand it enough to 687 00:43:04,950 --> 00:43:07,020 S1: be able to explain it, and then turn it into 688 00:43:07,020 --> 00:43:10,500 S1: a framework. So it becomes like a heuristic for people 689 00:43:10,500 --> 00:43:13,890 S1: who have who have read it. That's that's like my jam. 690 00:43:13,890 --> 00:43:16,109 S1: I love that, and I love to read it when 691 00:43:16,110 --> 00:43:18,600 S1: I find it somewhere else. Gabor. A new tool lets 692 00:43:18,600 --> 00:43:21,330 S1: you add function calling to any LLM with just a 693 00:43:21,330 --> 00:43:24,989 S1: few lines of code. Created a Cloudflare AI worker that 694 00:43:25,020 --> 00:43:28,230 S1: hooks into GitHub. It automatically reviews your code using the 695 00:43:28,380 --> 00:43:33,900 S1: Appsec model. Local version of Appsec hopefully MDK new command 696 00:43:33,900 --> 00:43:38,009 S1: line tool for parsing and extracting specific parts of markdown files, 697 00:43:38,100 --> 00:43:45,839 S1: similar to what JQ does for JSON. Scrape. Dodge. Scrape. Dodge. Gov. Scrape. Dodge. Gov. 698 00:43:45,870 --> 00:43:49,380 S1: Simon W also created a Python tool that tracks changes 699 00:43:49,380 --> 00:43:53,040 S1: made to Doge. Gov. Sonic crack. Bishop Fox released a 700 00:43:53,040 --> 00:43:58,380 S1: new tool that lets researchers decrypt and analyze Sonicwall NSV 701 00:43:58,380 --> 00:44:03,570 S1: V firewall firmware. Seneca on reading too much and too lightly. 702 00:44:03,600 --> 00:44:06,930 S1: Amazing piece. Go read that real quick. I mean, it's 703 00:44:06,960 --> 00:44:11,219 S1: like three seconds, five regrets of the dying and the 704 00:44:11,219 --> 00:44:14,550 S1: recommendation of the week. Instead of asking permission, tell your 705 00:44:14,550 --> 00:44:17,190 S1: manager what you're going to do and give them a 706 00:44:17,190 --> 00:44:22,049 S1: chance to object. And the aphorism of the week develop 707 00:44:22,050 --> 00:44:25,500 S1: a bias towards action. You're getting lapped by people who 708 00:44:25,530 --> 00:44:30,750 S1: don't overthink. Develop a bias towards action. You are getting 709 00:44:30,750 --> 00:44:35,880 S1: lapped by people who don't overthink. Unsupervised learning is produced 710 00:44:35,880 --> 00:44:40,200 S1: on Hindenburg Pro using an SM seven B microphone. A 711 00:44:40,200 --> 00:44:43,049 S1: video version of the podcast is available on the Unsupervised 712 00:44:43,050 --> 00:44:46,380 S1: Learning YouTube channel, and the text version with full links 713 00:44:46,380 --> 00:44:51,630 S1: and notes is available at Daniel Mysa.com slash newsletter. We'll 714 00:44:51,630 --> 00:44:52,560 S1: see you next time.