1 00:00:00,020 --> 00:00:02,600 S1: Are you prepared for whatever shitstorm may hit your desk 2 00:00:02,600 --> 00:00:05,360 S1: during the workday? Otto Max has your back. Check out 3 00:00:05,360 --> 00:00:08,690 S1: the brand new autonomous IT podcast. Listen in as various 4 00:00:08,690 --> 00:00:12,559 S1: IT experts discuss the latest Patch Tuesday releases, mitigation tips, 5 00:00:12,560 --> 00:00:16,130 S1: and custom automations to help with CVE remediations make new 6 00:00:16,130 --> 00:00:20,630 S1: work friends. Listen now to the autonomous IT podcast on Spotify, Apple, 7 00:00:20,630 --> 00:00:25,310 S1: or wherever you tune in to podcasts. Welcome to Unsupervised Learning, 8 00:00:25,310 --> 00:00:28,310 S1: a security, AI, and meaning focused podcast that looks at 9 00:00:28,310 --> 00:00:31,310 S1: how best to thrive as humans in a post AI world. 10 00:00:31,910 --> 00:00:35,900 S1: It combines original ideas, analysis, and mental models to bring 11 00:00:35,900 --> 00:00:38,450 S1: not just the news, but why it matters and how 12 00:00:38,450 --> 00:00:46,280 S1: to respond. All right, welcome to unsupervised learning. This is 13 00:00:46,280 --> 00:00:51,350 S1: Daniel Miessler. All right. Episode 441 441 of these. That's 14 00:00:51,350 --> 00:00:54,430 S1: pretty insane. Okay. Friend of UL Ray ulnar is looking 15 00:00:54,430 --> 00:00:57,400 S1: for a new position as a systems engineer. You should 16 00:00:57,400 --> 00:01:01,570 S1: reach out to Ray directly and got his LinkedIn there. 17 00:01:01,570 --> 00:01:06,459 S1: He's one of the best members of UL and, uh, super, 18 00:01:06,459 --> 00:01:09,460 S1: super smart, super kind and very competent as well as 19 00:01:09,459 --> 00:01:12,280 S1: you could see from his LinkedIn. And, uh, he just 20 00:01:12,280 --> 00:01:15,160 S1: came onto the market. So I would grab him before 21 00:01:15,160 --> 00:01:19,060 S1: he goes, added a couple of new levels to the 22 00:01:19,060 --> 00:01:22,690 S1: AGI definitions. So if we click through there and we 23 00:01:22,690 --> 00:01:26,559 S1: go down here, here are the definitions or the levels. 24 00:01:26,560 --> 00:01:30,640 S1: So AGI one is better but with significant drawbacks. Two 25 00:01:30,670 --> 00:01:36,070 S1: is competent but imperfect. Three is a full $80,000 a 26 00:01:36,069 --> 00:01:41,050 S1: year white collar worker replacement four okay, this is where 27 00:01:41,050 --> 00:01:44,620 S1: it gets cool. for is a world class employee such 28 00:01:44,620 --> 00:01:49,420 S1: as Andre Karpathy or Jeff Dean. So imagine 1% of 1% 29 00:01:49,420 --> 00:01:53,470 S1: and vision creativity programming, execution ability. This is like top 30 00:01:53,470 --> 00:01:57,160 S1: top tier. So imagine there's only like 100 or 1000 31 00:01:57,160 --> 00:02:00,040 S1: of these employees on the whole entire planet at any 32 00:02:00,040 --> 00:02:02,440 S1: given time. Wherever you want to set that bar, it's 33 00:02:02,440 --> 00:02:07,560 S1: something like that. Level five is a pinnacle human employee. Okay, 34 00:02:07,560 --> 00:02:11,190 S1: so this is somebody like the smartest person who's ever lived. 35 00:02:11,190 --> 00:02:14,160 S1: So a John von Neumann or an Isaac Newton or 36 00:02:14,160 --> 00:02:17,460 S1: a Richard Feynman or a Claude Shannon. So what this 37 00:02:17,460 --> 00:02:21,180 S1: offers over tier four is the ability to invent completely 38 00:02:21,180 --> 00:02:23,850 S1: new things when they don't exist, or see and explain 39 00:02:23,850 --> 00:02:26,910 S1: the world in a completely new way. So Newton didn't 40 00:02:26,910 --> 00:02:31,470 S1: have calculus, so he invented it. That's always like, astounded me. 41 00:02:31,500 --> 00:02:34,980 S1: Feynman was a supreme teacher, like a lot of people 42 00:02:34,980 --> 00:02:37,020 S1: consider him to be one of the best teachers ever. 43 00:02:37,050 --> 00:02:42,180 S1: Von Neumann innovated in mathematics, physics, engineering, game theory, like 44 00:02:42,180 --> 00:02:45,209 S1: tons of stuff, and Claude Shannon gave us information theory, 45 00:02:45,210 --> 00:02:49,290 S1: the foundations of cryptography, and a million other things. So basically, 46 00:02:49,290 --> 00:02:51,540 S1: at this level, you not only have the creativity and 47 00:02:51,540 --> 00:02:55,139 S1: execution of a top, like, whatever, only 100 or 1000 48 00:02:55,139 --> 00:02:57,690 S1: in the entire world at any given moment. But you 49 00:02:57,690 --> 00:03:01,770 S1: also have the once in a generation level innovation capabilities. 50 00:03:01,770 --> 00:03:05,370 S1: And once you're at Pinnacle Human, okay, the level after 51 00:03:05,370 --> 00:03:08,730 S1: that is where I cross over into ASI, which is 52 00:03:08,730 --> 00:03:12,450 S1: defined logically as a level of general AI that's smarter 53 00:03:12,450 --> 00:03:15,690 S1: and more capable than any human that's ever lived. And 54 00:03:15,690 --> 00:03:17,700 S1: that's the reason I added these, is so we have 55 00:03:17,700 --> 00:03:22,040 S1: a smoother transition from the end of AGI right into 56 00:03:22,040 --> 00:03:25,370 S1: the beginning of ASI one. Right. And that's where we 57 00:03:25,370 --> 00:03:31,160 S1: have right here. ASI one. So that was that. Uh, 58 00:03:31,160 --> 00:03:33,919 S1: I feel like Apple Notes is my actual operating system 59 00:03:33,919 --> 00:03:36,650 S1: and Mac OS is just the window manager, just, uh, 60 00:03:36,650 --> 00:03:41,510 S1: stupid jokes. Um, how many notes do I have in 61 00:03:41,510 --> 00:03:44,240 S1: Apple Notes? I'm going to look right now to try 62 00:03:44,240 --> 00:03:50,630 S1: to emphasize this point. 3867 Apple notes my operating system. 63 00:03:50,630 --> 00:03:53,510 S1: Not really, but you see what I'm saying? All right, 64 00:03:53,510 --> 00:03:58,970 S1: let's get into it. Okay. So substrate I've been thinking 65 00:03:58,970 --> 00:04:01,610 S1: about and working on this one for months now and 66 00:04:01,610 --> 00:04:04,730 S1: finally announced it. And actually there's a separate whole substrate 67 00:04:04,730 --> 00:04:08,360 S1: video coming out very soon you should check out. So yeah, 68 00:04:08,360 --> 00:04:10,040 S1: go look at that when it comes out. I'm not 69 00:04:10,040 --> 00:04:12,170 S1: going to go into detail here because I just did 70 00:04:12,170 --> 00:04:14,600 S1: in that other piece, got a new piece on dynamic 71 00:04:14,600 --> 00:04:16,850 S1: content summaries and how I think they're going to be 72 00:04:16,850 --> 00:04:19,310 S1: the way we view content in the future. Essentially, the 73 00:04:19,310 --> 00:04:23,690 S1: idea here is that rather than get raw content from 74 00:04:23,690 --> 00:04:27,679 S1: different places, I will produce content for us dynamic. So 75 00:04:27,680 --> 00:04:30,680 S1: imagine that we love to see Richard Feynman explain things 76 00:04:30,680 --> 00:04:34,120 S1: to us. Well, anytime anything complex happens in the world, 77 00:04:34,120 --> 00:04:36,669 S1: we're going to have a video. It's going to be 78 00:04:36,670 --> 00:04:39,820 S1: a YouTube video or something like a YouTube video, and 79 00:04:39,820 --> 00:04:43,930 S1: it's going to be Richard Feynman fully deepfaked, fully Deepfaked. 80 00:04:43,930 --> 00:04:46,989 S1: It looks exactly like Richard Feynman. And he's actually got 81 00:04:46,990 --> 00:04:50,620 S1: a piece of chalk, and he's actually in his Caltech 82 00:04:50,620 --> 00:04:55,930 S1: classroom with a giant whiteboard, and he's dynamically whiteboarding out 83 00:04:55,930 --> 00:04:58,690 S1: this concept to you and explaining with his voice and 84 00:04:58,690 --> 00:05:01,930 S1: his jokes and his humor or whatever. And he's explaining 85 00:05:01,930 --> 00:05:04,360 S1: everything to you. In fact, it might be like if 86 00:05:04,360 --> 00:05:09,280 S1: it's science related or complexity related or whatever, something Feynman ish, 87 00:05:09,279 --> 00:05:12,190 S1: then use Feynman. But maybe you've got this other like 88 00:05:12,190 --> 00:05:15,729 S1: cute influencer girl who you like to explain, like modern 89 00:05:15,730 --> 00:05:18,789 S1: cultural topics or whatever, and she is deepfaked to do 90 00:05:18,790 --> 00:05:20,680 S1: that for you there. And of course, there's going to 91 00:05:20,680 --> 00:05:23,020 S1: be all sorts of considerations around like, well, do they 92 00:05:23,020 --> 00:05:25,030 S1: have permission to do that? And a lot of people 93 00:05:25,029 --> 00:05:27,310 S1: will say, yeah, go ahead, do it. As long as 94 00:05:27,310 --> 00:05:30,940 S1: it falls within these criteria. Some people won't care. They'll 95 00:05:30,940 --> 00:05:33,940 S1: just deepfake anybody and do that. Um, other times it'll 96 00:05:33,940 --> 00:05:36,849 S1: be like the avatar of your actual Da. So your 97 00:05:36,850 --> 00:05:39,670 S1: Da is going to your digital assistant, your personal AI 98 00:05:39,670 --> 00:05:42,670 S1: is going to have an avatar as well. So maybe 99 00:05:42,670 --> 00:05:45,029 S1: you like that avatar to tell it to you, right? 100 00:05:45,029 --> 00:05:48,390 S1: Because maybe that one is of a coach, or honestly, 101 00:05:48,390 --> 00:05:50,250 S1: it's just going to become like one of your best friends. 102 00:05:50,250 --> 00:05:53,640 S1: It knows you the closest. It's very, you know, intimately 103 00:05:53,640 --> 00:05:56,910 S1: connected with you. So maybe it's the one teaching you. 104 00:05:56,910 --> 00:05:59,910 S1: Maybe it's you and your own voice teaching you. Okay. 105 00:05:59,910 --> 00:06:02,700 S1: It's going to be very easy to deepfake yourself, so 106 00:06:02,700 --> 00:06:06,420 S1: maybe you're explaining it to yourself like you created the content. 107 00:06:06,420 --> 00:06:09,279 S1: That would be insane. Point is, if there is a 108 00:06:09,279 --> 00:06:14,200 S1: four hour video explaining this very complex topic, okay, it's 109 00:06:14,200 --> 00:06:16,570 S1: a new type of machine learning. It's a new type 110 00:06:16,570 --> 00:06:19,210 S1: of AI. It's a new hacking technique, but it's an 111 00:06:19,210 --> 00:06:21,609 S1: hour long. It's three hours long. It's four hours long. 112 00:06:21,610 --> 00:06:25,900 S1: It's five hours long. It's a 20 video course. The 113 00:06:25,900 --> 00:06:28,840 S1: AI will go and watch it, and then it'll be like, hey, um, 114 00:06:28,839 --> 00:06:31,120 S1: I created a thing. I know you don't have much 115 00:06:31,120 --> 00:06:35,450 S1: time this week, but I created a 14 minute summary 116 00:06:35,450 --> 00:06:39,140 S1: of this 20 video series, and I'm going to show 117 00:06:39,140 --> 00:06:42,260 S1: it to you. It's actually done by Richard Feynman. So 118 00:06:42,260 --> 00:06:46,010 S1: now Richard Feynman explains that entire 20 videos to you 119 00:06:46,010 --> 00:06:49,310 S1: in 14 minutes, perfectly taught by Richard Feynman. So this 120 00:06:49,310 --> 00:06:52,370 S1: is the concept of dynamic content summaries that the raw 121 00:06:52,370 --> 00:06:55,430 S1: version of a thing is usually not going to be 122 00:06:55,430 --> 00:06:58,280 S1: ideal for a person. The ideal will be to have 123 00:06:58,279 --> 00:07:01,250 S1: it pitched in the way they like to receive things. Okay, 124 00:07:01,250 --> 00:07:03,409 S1: because it won't always be a video. Sometimes it will 125 00:07:03,410 --> 00:07:05,839 S1: be an audio book. It'll just be audio in your ear. 126 00:07:05,839 --> 00:07:08,000 S1: Other times it'll be like a visual story, and it 127 00:07:08,000 --> 00:07:10,580 S1: won't actually have an avatar or a person teaching it 128 00:07:10,580 --> 00:07:13,520 S1: on a whiteboard. It'll just be like animation, and maybe 129 00:07:13,520 --> 00:07:15,500 S1: people will learn better from that. Maybe people will learn 130 00:07:15,500 --> 00:07:18,230 S1: better from a story, whatever they prefer. That content will 131 00:07:18,230 --> 00:07:22,130 S1: be dynamically created by eye to best give you the 132 00:07:22,130 --> 00:07:25,220 S1: content from the raw version. And of course, I'm a 133 00:07:25,220 --> 00:07:28,130 S1: huge fan of slow, slow learning, and there's a whole 134 00:07:28,130 --> 00:07:31,160 S1: bunch of stuff in fabric already for this. Like sometimes 135 00:07:31,160 --> 00:07:34,190 S1: you don't want that summary, and sometimes you should stop 136 00:07:34,190 --> 00:07:37,520 S1: watching those summaries and say in the AI should say, 137 00:07:37,520 --> 00:07:40,190 S1: you know what? I created you a 14 minute summary 138 00:07:40,190 --> 00:07:42,710 S1: of this thing, but this you need to go watch 139 00:07:42,710 --> 00:07:45,410 S1: the full thing because oftentimes because I've been using these 140 00:07:45,430 --> 00:07:50,739 S1: summaries with fabric for almost two years now and I often, well, 141 00:07:50,740 --> 00:07:54,220 S1: pretty much always I always get something more from doing 142 00:07:54,220 --> 00:07:56,950 S1: the slow version, from watching the four hour version, from 143 00:07:56,950 --> 00:08:00,490 S1: watching their facial expressions. I get value from the summary itself, 144 00:08:00,490 --> 00:08:03,700 S1: but I get more value from watching it happen slowly 145 00:08:03,700 --> 00:08:06,670 S1: and watching it seep in slowly. So maybe you do both, 146 00:08:06,670 --> 00:08:08,890 S1: maybe you do one or the other, but don't give 147 00:08:08,890 --> 00:08:10,960 S1: up the slow just because you could do the fast. 148 00:08:10,960 --> 00:08:13,690 S1: Exploring the idea of personal and business brands. Yeah, that 149 00:08:13,690 --> 00:08:17,020 S1: was another short piece. Okay. Kaspersky is shut down US 150 00:08:17,050 --> 00:08:21,250 S1: operations because of banning. AT&T says nearly all cellular customers 151 00:08:21,250 --> 00:08:24,370 S1: and some landline users have had their data stolen. But 152 00:08:24,370 --> 00:08:30,340 S1: Kim Zetter at Wired said somebody got paid like $300,000 153 00:08:30,340 --> 00:08:34,990 S1: from the hacking team by AT&T to delete what was 154 00:08:34,990 --> 00:08:39,189 S1: supposedly a copy of all the data, and supposedly the 155 00:08:39,190 --> 00:08:42,100 S1: only copy of all the data. So we'll see how 156 00:08:42,100 --> 00:08:44,860 S1: that plays out. But good reporting from Kim Zetter. Russia 157 00:08:44,860 --> 00:08:49,479 S1: is using AI enhanced software called Meliorator to create fake 158 00:08:49,480 --> 00:08:54,370 S1: online personas for disinformation campaigns. Ooh, love this, because the 159 00:08:54,370 --> 00:08:56,229 S1: better this gets, the more it's going to look like 160 00:08:56,230 --> 00:08:59,520 S1: a real human. And then if the AI starts acting 161 00:08:59,520 --> 00:09:02,339 S1: like a human with like clicking on things, being able 162 00:09:02,340 --> 00:09:03,930 S1: to tell the difference is going to get a whole 163 00:09:03,929 --> 00:09:05,729 S1: lot more difficult. And this is actually one of the 164 00:09:05,730 --> 00:09:08,670 S1: things I'm most worried about from AI is disinformation bots, 165 00:09:08,670 --> 00:09:11,490 S1: not just the sheer number of them, but also their 166 00:09:11,490 --> 00:09:14,610 S1: how advanced they are, how sophisticated they are. Basically, the 167 00:09:14,610 --> 00:09:18,180 S1: better AI gets, and especially from agent frameworks, the more 168 00:09:18,179 --> 00:09:21,660 S1: it's going to be exactly the same, or almost exactly 169 00:09:21,660 --> 00:09:26,490 S1: the same, as if our enemies have millions of employees 170 00:09:26,490 --> 00:09:29,640 S1: in their intelligence agencies. Okay, let me say that again. 171 00:09:29,640 --> 00:09:31,920 S1: That was that was too convoluted. One of the things 172 00:09:31,920 --> 00:09:35,010 S1: I'm most worried about is that the better AI gets, 173 00:09:35,010 --> 00:09:42,360 S1: it'll be like our foreign intelligence agencies for our enemies China, Russia, Iran, 174 00:09:42,360 --> 00:09:46,170 S1: North Korea. Instead of having a few hundred elite people, 175 00:09:46,170 --> 00:09:48,810 S1: they're going to have a few hundred elite people and 176 00:09:48,809 --> 00:09:52,800 S1: 40,000 right below them, elite people. And then the better 177 00:09:52,800 --> 00:09:56,010 S1: the AI gets, those move up into the elite, and 178 00:09:56,010 --> 00:09:59,520 S1: then the elite humans become augmented with that AI and 179 00:09:59,520 --> 00:10:02,430 S1: they become like super elite. And the better the AI gets, 180 00:10:02,429 --> 00:10:05,130 S1: I mean, it just keeps moving up that chain. So 181 00:10:05,130 --> 00:10:09,060 S1: if you think it's bad to have ten, 50, 100 182 00:10:09,059 --> 00:10:13,520 S1: elite people in these different countries who are both Intel 183 00:10:13,520 --> 00:10:17,540 S1: trained and manipulation trained, but they're also like super hackers, 184 00:10:17,540 --> 00:10:20,600 S1: and they're good at all these different skills. But there's 185 00:10:20,600 --> 00:10:23,150 S1: I mean, there's only so many in the world. Well, 186 00:10:23,150 --> 00:10:28,370 S1: now imagine them with a thousand of them, 10,000 of them, 100,000, 187 00:10:28,370 --> 00:10:32,540 S1: 100 million. And they're working consistently with no sleep to 188 00:10:32,540 --> 00:10:35,840 S1: go after all your targets. That that's what I'm personally 189 00:10:35,840 --> 00:10:39,260 S1: worried about right now. And if they are producing content 190 00:10:39,260 --> 00:10:42,170 S1: and the content is all manipulative, the internet is going 191 00:10:42,170 --> 00:10:45,200 S1: to have to switch from block list to an allow list. 192 00:10:45,200 --> 00:10:47,569 S1: I mean, think about that. The internet currently is an 193 00:10:47,570 --> 00:10:52,580 S1: allow list. I'm sorry, it's a block list. Everything is available, 194 00:10:52,580 --> 00:10:55,010 S1: and you have to specifically go into a place and say, 195 00:10:55,010 --> 00:10:56,540 S1: I don't want to see this. I don't want to 196 00:10:56,540 --> 00:10:58,640 S1: see this. I don't want to see this because of 197 00:10:58,640 --> 00:11:01,110 S1: this problem right here that I'm talking about. We're going 198 00:11:01,110 --> 00:11:03,150 S1: to have to switch. We're going to have to switch 199 00:11:03,150 --> 00:11:06,990 S1: to the other method, which is to say my I, 200 00:11:07,020 --> 00:11:13,980 S1: my internet platform, my mobile OS, okay, Android, macOS, iOS, 201 00:11:13,980 --> 00:11:16,620 S1: they're not going to allow anything in unless you have 202 00:11:16,620 --> 00:11:18,809 S1: the green check mark. What is the green check mark? 203 00:11:18,809 --> 00:11:22,890 S1: Green check mark is an agreement between the the tech platform, 204 00:11:22,890 --> 00:11:27,130 S1: the mobile platform, the OS and trusted providers like a YouTube, 205 00:11:27,130 --> 00:11:32,860 S1: which has approved sources from approved individuals and approved organizations. 206 00:11:32,860 --> 00:11:34,870 S1: So your news is going to have to come that 207 00:11:34,870 --> 00:11:37,150 S1: way because it's going to have to be vetted. So 208 00:11:37,150 --> 00:11:40,540 S1: it gets the green checkmark. This might be something that's required. 209 00:11:40,540 --> 00:11:42,370 S1: It's going to happen anyway, but it might be something 210 00:11:42,370 --> 00:11:47,080 S1: that's required within settings and within operating systems because it's 211 00:11:47,080 --> 00:11:49,840 S1: simply too hard to drown out everything. Because this problem 212 00:11:49,840 --> 00:11:53,950 S1: right here, the amount of disinformation bots, pretty soon it's 213 00:11:53,950 --> 00:11:58,060 S1: going to be like, whatever, 50% of all content is 214 00:11:58,059 --> 00:12:00,760 S1: going to be manipulation, then it's going to be like 75%, 215 00:12:00,760 --> 00:12:03,130 S1: then it's going to be 95%. So if you are 216 00:12:03,130 --> 00:12:05,920 S1: naive and you just walk onto the internet, especially as 217 00:12:05,920 --> 00:12:08,860 S1: like a regular person or like a grandma or something, 218 00:12:08,860 --> 00:12:11,109 S1: you're going to be screwed. You're going to be just 219 00:12:11,110 --> 00:12:13,780 S1: like pulled around, Like all your money will disappear because 220 00:12:13,780 --> 00:12:16,449 S1: you had to save your kids from some horrible accident. 221 00:12:16,450 --> 00:12:18,640 S1: And all of a sudden there was an asteroid that 222 00:12:18,640 --> 00:12:21,099 S1: hit the thing and aliens came. And pretty soon you're 223 00:12:21,100 --> 00:12:23,410 S1: in a bunker and you have no money, and it's 224 00:12:23,410 --> 00:12:26,079 S1: like none of it was real. The amount of and 225 00:12:26,080 --> 00:12:30,250 S1: quality of manipulation. Think about that as well. The quality 226 00:12:30,250 --> 00:12:34,360 S1: of the pitches and the manipulation and the deepfakes and 227 00:12:34,360 --> 00:12:36,780 S1: the trickery is going to get so good. It's going 228 00:12:36,780 --> 00:12:39,300 S1: to trick us to it's going to trick experts, not 229 00:12:39,300 --> 00:12:42,720 S1: just regular people. There's a new XM vulnerability with a 230 00:12:42,720 --> 00:12:46,319 S1: Cvss of 9.1. Google is now offering Passkeys for high 231 00:12:46,320 --> 00:12:49,740 S1: risk users who join Advanced Protection. Previously, they had to 232 00:12:49,740 --> 00:12:56,309 S1: get a physical token foreign influence campaigns analysis from US intelligence. Basically, 233 00:12:56,309 --> 00:12:59,939 S1: they say Russia is backing Trump, probably mostly because of Ukraine. 234 00:12:59,940 --> 00:13:03,000 S1: Iran is acting as a chaos agent in its campaigns, 235 00:13:03,000 --> 00:13:06,809 S1: focusing on exploiting US political and social tensions rather than 236 00:13:06,809 --> 00:13:10,590 S1: backing one side or another. And China's mostly staying out 237 00:13:10,590 --> 00:13:15,000 S1: of the elections just trying to data collect for future 238 00:13:15,000 --> 00:13:20,130 S1: influence operations. Think OPM, uh, Marriott. Yeah, though don't get 239 00:13:20,130 --> 00:13:23,100 S1: me started on those. Thanks to auto Max for sponsoring. 240 00:13:23,100 --> 00:13:27,140 S1: GitLab has a critical flaw in its CI CD pipelines. 241 00:13:27,170 --> 00:13:32,360 S1: Thanks to Project Discovery for sponsoring and OpenAI's AGI levels, 242 00:13:32,360 --> 00:13:35,630 S1: OpenAI has published their five tier ladder for AI progress. 243 00:13:35,630 --> 00:13:38,449 S1: I'm honestly not a fan. Other than level five, I 244 00:13:38,450 --> 00:13:41,360 S1: find level five really interesting. But the problem is they're 245 00:13:41,360 --> 00:13:44,420 S1: going from like chat bots to human level reasoners and 246 00:13:44,420 --> 00:13:47,300 S1: then to AI agents, then to innovators that can aid 247 00:13:47,300 --> 00:13:50,329 S1: in invention. So level two and level four are way 248 00:13:50,330 --> 00:13:55,790 S1: too close. Okay. So human level reasoners to agents. It's 249 00:13:55,790 --> 00:13:59,719 S1: not a big enough jump. Uh, and both are already possible. 250 00:13:59,720 --> 00:14:02,090 S1: That's the other problem. Level two and level four. They're 251 00:14:02,090 --> 00:14:04,580 S1: kind of already possible. Then you have this really interesting 252 00:14:04,580 --> 00:14:06,950 S1: jump at level five to something that can do the 253 00:14:06,950 --> 00:14:11,180 S1: work of an organization. So for OpenAI level five, AGI 254 00:14:11,179 --> 00:14:14,450 S1: is something that could do the work of an entire organization, 255 00:14:14,450 --> 00:14:18,380 S1: but that's not super helpful. Okay, there are lots of 256 00:14:18,380 --> 00:14:22,400 S1: very dumb organizations, okay? An organization could be for people. 257 00:14:22,400 --> 00:14:26,360 S1: An organization can be Apple, an organization could be OpenAI 258 00:14:26,360 --> 00:14:30,740 S1: in the year 2027. So that is not super descriptive 259 00:14:30,740 --> 00:14:33,410 S1: in my opinion. And even worse, the problem is they're 260 00:14:33,410 --> 00:14:37,070 S1: mixing criteria here. Right? So one is I don't know 261 00:14:37,070 --> 00:14:41,320 S1: what what criteria number one is Reasoners is about thinking 262 00:14:41,320 --> 00:14:45,010 S1: quality agents is just an attribute, which means the thing 263 00:14:45,010 --> 00:14:46,960 S1: can take an action. It doesn't say anything about the 264 00:14:46,960 --> 00:14:50,890 S1: quality of the action. So that's not necessarily above one 265 00:14:50,890 --> 00:14:54,700 S1: or below four or above four. It's it doesn't put 266 00:14:54,700 --> 00:14:57,850 S1: it anywhere because it's apples to oranges. Innovators is just 267 00:14:57,850 --> 00:15:02,050 S1: a descriptive output. So it aids in invention. Here's the 268 00:15:02,050 --> 00:15:05,710 S1: problem with this chat bots already aid with invention and 269 00:15:05,710 --> 00:15:07,960 S1: that's what they have as their level. One is chat bots. 270 00:15:07,990 --> 00:15:11,440 S1: The chat bots already help inventors invent better things. Then 271 00:15:11,440 --> 00:15:15,010 S1: you have level five, which is actually about scale more 272 00:15:15,010 --> 00:15:18,340 S1: than quality. Um, yeah. And like I said, scale isn't 273 00:15:18,340 --> 00:15:21,370 S1: super important because you don't know if you're scaling like 274 00:15:21,370 --> 00:15:23,920 S1: an ant colony. They're just kind of doing the same thing. 275 00:15:23,920 --> 00:15:27,580 S1: Or if you're scaling and the scale itself leads to 276 00:15:27,580 --> 00:15:31,510 S1: smarter things. So just not overall happy with it. I'm 277 00:15:31,510 --> 00:15:33,910 S1: happy they put it out and I feel bad because 278 00:15:33,910 --> 00:15:36,040 S1: I'm sure someone put some effort into it. And I'm 279 00:15:36,040 --> 00:15:38,950 S1: sure they're human and I don't want you to feel bad. 280 00:15:38,950 --> 00:15:42,610 S1: Just hopefully take this feedback and maybe the next one. 281 00:15:42,610 --> 00:15:46,550 S1: Just think about these ideas, okay? AI startups raising $100 282 00:15:46,550 --> 00:15:49,870 S1: million in 2024. Got a full list there. Anthropic has 283 00:15:49,870 --> 00:15:54,600 S1: more new features, uh, to help automate prompt engineering. And 284 00:15:54,600 --> 00:15:57,240 S1: I see a lot of my friends switching to cloud 285 00:15:57,240 --> 00:16:01,860 S1: over ChatGPT right now, especially sonnet 35. I'm actually using 286 00:16:01,860 --> 00:16:05,370 S1: sonnet 35 as my default within fabric right now, and 287 00:16:05,370 --> 00:16:07,530 S1: it's a leapfrog game because soon we're about to have 288 00:16:07,530 --> 00:16:12,690 S1: opus 35. Cannot wait for that llama 3.0 300 billion 289 00:16:12,690 --> 00:16:17,290 S1: and eventually GPT five, I'm guessing fall to right at 290 00:16:17,290 --> 00:16:21,609 S1: the very beginning of January. So I'm going to say August. September. 291 00:16:21,610 --> 00:16:27,190 S1: I'm going to say between October and February, 90% chance 292 00:16:27,190 --> 00:16:30,850 S1: of GPT five or the next GPT to come out, 293 00:16:30,850 --> 00:16:32,440 S1: whether they call it 4 or 5 or they call 294 00:16:32,440 --> 00:16:36,070 S1: it five, whatever, or some new name. And yeah, 2025 295 00:16:36,070 --> 00:16:38,950 S1: is going to be nuts. So we're going gonna we're 296 00:16:38,950 --> 00:16:41,560 S1: gonna have a new president. Well, maybe a new president, 297 00:16:41,560 --> 00:16:43,630 S1: but either way, it's going to be a new whatever. 298 00:16:43,630 --> 00:16:45,310 S1: There's going to be an election and an outcome. So 299 00:16:45,310 --> 00:16:47,739 S1: that's going to be crazy by itself. Um, hopefully not 300 00:16:47,740 --> 00:16:50,890 S1: the bad crazy, but, uh, yeah, there's just going to 301 00:16:50,890 --> 00:16:54,280 S1: be a whole lot going on in 2025, not the 302 00:16:54,280 --> 00:16:56,890 S1: least of which is all these new models. New fiber 303 00:16:56,890 --> 00:17:03,960 S1: optic network transmits data at above 400TB per second. And importantly, 304 00:17:03,960 --> 00:17:07,590 S1: this is using existing fiber, not some, you know, adamantium 305 00:17:07,590 --> 00:17:10,410 S1: special stuff. YouTube music is testing. A new feature lets 306 00:17:10,410 --> 00:17:13,470 S1: you use AI to generate a playlist by just describing 307 00:17:13,470 --> 00:17:16,020 S1: what you want to hear. I would use that a lot. 308 00:17:16,020 --> 00:17:19,889 S1: There's a surge in delivery startups like Hail Ify, okay, 309 00:17:19,890 --> 00:17:24,840 S1: Hail Ify, and they're focusing on shine and TMU, which 310 00:17:24,840 --> 00:17:28,490 S1: are the Chinese people competing with Amazon? Why women are 311 00:17:28,490 --> 00:17:31,790 S1: disappearing from tech Houston is on a path to an 312 00:17:31,790 --> 00:17:35,720 S1: all out power crisis. Tour de France riders are inhaling 313 00:17:35,720 --> 00:17:38,870 S1: carbon monoxide, which is a poison gas. It's the one 314 00:17:38,869 --> 00:17:41,330 S1: that kills you, I think, in your car, in a garage, 315 00:17:41,330 --> 00:17:44,719 S1: if you don't have, like, open ventilation, how how is 316 00:17:44,720 --> 00:17:50,750 S1: a toxic gas improving performance that okay. Yeah. 130 million 317 00:17:50,750 --> 00:17:56,030 S1: US adults have low literacy skills, over half of Americans 318 00:17:56,030 --> 00:18:01,040 S1: aged 16 to 74. Read below. A sixth grade level. 319 00:18:01,040 --> 00:18:03,020 S1: I'm gonna say that again for the people in the back, 320 00:18:03,020 --> 00:18:08,000 S1: over half of Americans aged 16 to 74 read below. 321 00:18:08,000 --> 00:18:11,930 S1: A sixth grade level moving on Colorado poultry workers test 322 00:18:11,930 --> 00:18:14,659 S1: positive for bird flu. But we're not talking about that 323 00:18:14,660 --> 00:18:18,010 S1: because we we're tired of talking about viruses. Uh, we're 324 00:18:18,010 --> 00:18:19,959 S1: tired of talking about Covid. We don't want to talk 325 00:18:19,960 --> 00:18:23,470 S1: about bird flu. That's just another Covid. It's, uh. We're exhausted. 326 00:18:23,470 --> 00:18:25,960 S1: You know what? I give people a break. People are exhausted. 327 00:18:25,960 --> 00:18:28,870 S1: They don't want to talk about negative stuff. They literally 328 00:18:28,869 --> 00:18:31,900 S1: are just going to ignore this with blinders until it 329 00:18:31,900 --> 00:18:35,020 S1: literally comes to their front door and, like, makes their 330 00:18:35,020 --> 00:18:38,290 S1: friends sick. And other than that, any other coverage of this, 331 00:18:38,290 --> 00:18:40,120 S1: they're just going to be like, yeah, I'll know about 332 00:18:40,119 --> 00:18:43,510 S1: it if it gets too bad. Yeah. And meanwhile, all 333 00:18:43,510 --> 00:18:46,540 S1: the experts are like, guys, this is kind of bad. 334 00:18:46,540 --> 00:18:51,070 S1: It's transferring it humans. It's killed 6 million birds and 335 00:18:51,070 --> 00:18:55,480 S1: is now infecting dairy cattle across the state, which means 336 00:18:55,480 --> 00:19:01,270 S1: it's going into milk and dairy workers are getting it. Uh, no. 337 00:19:01,270 --> 00:19:04,030 S1: Poultry workers are now getting it. So it's like. I mean, 338 00:19:04,030 --> 00:19:07,179 S1: I kind of feel it along with everyone else. I'm like, uh, 339 00:19:07,180 --> 00:19:11,380 S1: I guess if my friends start getting it, I'll pay attention. 340 00:19:11,380 --> 00:19:15,250 S1: Very strange times we live in, we're so, so exhausted 341 00:19:15,250 --> 00:19:19,419 S1: by negativity in certain types of negativity. Just foreign. 10,000 342 00:19:19,420 --> 00:19:22,990 S1: galaxies may host intelligent aliens. This is a new study 343 00:19:22,990 --> 00:19:26,109 S1: basically trying to explain the Drake equation of like why 344 00:19:26,109 --> 00:19:30,389 S1: we're not seeing aliens, but four out of 10,000 galaxies. 345 00:19:30,390 --> 00:19:34,919 S1: I believe there's 200 billion galaxies. Unless I'm mixing up 346 00:19:34,920 --> 00:19:37,560 S1: my numbers. Is it 200 billion galaxies? I think it's 347 00:19:37,560 --> 00:19:42,510 S1: 200 billion. So 4 in 10,000. But there's actually. So 348 00:19:42,510 --> 00:19:45,990 S1: what is that ten K to 200 billion. That's still 349 00:19:45,990 --> 00:19:50,130 S1: a lot of galaxies. That's a lot of galaxies. And 350 00:19:50,130 --> 00:19:52,500 S1: that's just this is just one study okay okay. Okay. 351 00:19:52,500 --> 00:19:59,130 S1: So it's saying 0.003% to 0.2% of exoplanets. Yeah, but 352 00:19:59,130 --> 00:20:02,160 S1: there's a lot of exoplanets like this is not super 353 00:20:02,160 --> 00:20:04,770 S1: depressing to me. But I guess it's interesting if it 354 00:20:04,770 --> 00:20:09,990 S1: explains the Drake equation. Okay, ideas VC's are buying medical practices. 355 00:20:09,990 --> 00:20:12,810 S1: I got a buddy who's talking about how VCs are 356 00:20:12,810 --> 00:20:16,080 S1: just coming into these practices, and they're just moving in 357 00:20:16,080 --> 00:20:18,119 S1: and being like, you've got to sell all this super 358 00:20:18,119 --> 00:20:22,260 S1: gross stuff. Order tests that you don't need, like sell 359 00:20:22,260 --> 00:20:25,169 S1: them miracle drugs that don't actually work. You've got to 360 00:20:25,170 --> 00:20:28,890 S1: do this to hype up the user base. They're basically 361 00:20:28,890 --> 00:20:33,900 S1: turning it into like sexy sales and, um, getting them 362 00:20:33,900 --> 00:20:38,460 S1: to do things other than fundamental, like high quality, proactive 363 00:20:38,460 --> 00:20:41,720 S1: health like diet and exercise. It's kind of going the 364 00:20:41,720 --> 00:20:48,169 S1: opposite direction towards more tests, more assessments, pills and things 365 00:20:48,170 --> 00:20:50,810 S1: that don't actually work. Why? Because they make more money, 366 00:20:50,810 --> 00:20:54,170 S1: which is good for VC's. So really, really bad. And 367 00:20:54,170 --> 00:20:57,380 S1: thanks to my buddy for calling that out. Most conspiracies 368 00:20:57,380 --> 00:21:00,590 S1: come from not realizing how often things fail. The tweet 369 00:21:00,590 --> 00:21:04,430 S1: thread I put out recently on X therapy, rumination, and 370 00:21:04,430 --> 00:21:08,550 S1: Untying knots. This is about therapy. Recommend you click on it. 371 00:21:08,580 --> 00:21:10,500 S1: We're already going long so I'm not going to talk 372 00:21:10,500 --> 00:21:15,359 S1: about it here. Discovery fluff on Lambda. Really cool. Uh 373 00:21:15,359 --> 00:21:18,750 S1: really cool project by this guy here. Um, definitely want 374 00:21:18,750 --> 00:21:21,930 S1: to check it out. Bullfrog GitHub action secures your workflows 375 00:21:21,930 --> 00:21:25,649 S1: by controlling outbound network connections. Everything you see is a 376 00:21:25,650 --> 00:21:29,370 S1: computational process if you know how to look. Um. Love it. 377 00:21:29,369 --> 00:21:33,879 S1: Sean Purry emotional elicitor. This is a prompt by Moritz 378 00:21:33,910 --> 00:21:37,120 S1: Kremp who's a really cool, uh, doing some really cool 379 00:21:37,119 --> 00:21:40,720 S1: prompt stuff. Got to check him out. WTF happened to blogs? 380 00:21:40,720 --> 00:21:44,050 S1: As an employee you are disposable. You never control the 381 00:21:44,050 --> 00:21:47,620 S1: arc of your career. Smoking versus lung cancer deaths. I'm 382 00:21:47,619 --> 00:21:49,629 S1: going to show this one. This is insane. Look at 383 00:21:49,630 --> 00:21:52,480 S1: this thing. Hmm. Wonder if there's a correlation. Well, we 384 00:21:52,480 --> 00:21:56,590 S1: see a correlation. I wonder if there's causation. That's that's compelling. 385 00:21:56,590 --> 00:22:01,210 S1: Look at this. Taxes increasing bans taxes. Look at this. 386 00:22:01,210 --> 00:22:04,900 S1: These are deaths going down. Great depression, World War two 387 00:22:04,900 --> 00:22:08,890 S1: boom US banned cigarette ads on the radio. And surgeon 388 00:22:08,890 --> 00:22:12,820 S1: General report linked smoking to death from cancer. That's the top. 389 00:22:12,820 --> 00:22:14,740 S1: And it starts going down from there with all these 390 00:22:14,740 --> 00:22:18,969 S1: other measures. Really interesting learning multiple concepts from a SQL image, 391 00:22:18,970 --> 00:22:22,870 S1: change detection in satellite imagery. This is a hobby. I 392 00:22:22,869 --> 00:22:25,060 S1: would be doing a lot more. I'm actually waiting for 393 00:22:25,060 --> 00:22:27,010 S1: the AI to get really good at this. I want 394 00:22:27,010 --> 00:22:30,640 S1: to create a bot that basically watches the latest satellite 395 00:22:30,640 --> 00:22:35,680 S1: images from choice locations, uses AI in a pipeline, and 396 00:22:35,800 --> 00:22:40,119 S1: sends me a constant flow of updates of what changed. 397 00:22:40,119 --> 00:22:44,700 S1: Like it looks like they did something underground. Um, a 398 00:22:44,700 --> 00:22:46,890 S1: whole bunch of trucks moved in. They appear to be 399 00:22:46,890 --> 00:22:50,429 S1: this kind of truck. Here are the possible explanations for that. Again, 400 00:22:50,430 --> 00:22:52,979 S1: going back to the Intel thing that I'm building, I 401 00:22:52,980 --> 00:22:58,170 S1: want continuous flows of information flowing into me based on 402 00:22:58,170 --> 00:23:03,419 S1: capturing artifacts, in this case images, and doing AI analysis 403 00:23:03,420 --> 00:23:06,660 S1: on them. 89 things I Know About Git Commits I 404 00:23:06,660 --> 00:23:10,740 S1: got the most comments about this one and substrate this week, 405 00:23:10,740 --> 00:23:14,100 S1: so definitely check this one out by Jamie Tanner and 406 00:23:14,100 --> 00:23:16,920 S1: the recommendation of the week. Check on your friends if 407 00:23:16,920 --> 00:23:20,340 S1: you haven't heard from them in a while, send them 408 00:23:20,340 --> 00:23:23,370 S1: a text. It's free and they will appreciate being thought 409 00:23:23,369 --> 00:23:26,520 S1: of and be consistent about it. Like push it right? 410 00:23:26,520 --> 00:23:31,200 S1: They they will appreciate that in the age of infinite leverage, 411 00:23:31,200 --> 00:23:35,820 S1: judgment is the most important skill. Naval has been crushing 412 00:23:35,820 --> 00:23:38,460 S1: it lately. Absolutely crushing it. I'm going to say this 413 00:23:38,460 --> 00:23:42,330 S1: one again. In the age of infinite leverage, judgment is 414 00:23:42,330 --> 00:23:48,330 S1: the most important skill. Naval. Unsupervised learning is produced and 415 00:23:48,330 --> 00:23:51,870 S1: edited by Daniel Miessler on a Neumann U87 AI microphone 416 00:23:51,869 --> 00:23:55,850 S1: using Hindenburg. Intro and outro music is by Zomby with 417 00:23:55,850 --> 00:23:58,670 S1: the Y, and to get the text and links from 418 00:23:58,670 --> 00:24:00,830 S1: this episode, sign up for the newsletter version of the 419 00:24:00,830 --> 00:24:06,350 S1: show at Daniel miessler.com/newsletter. We'll see you next time.