1 00:00:20,252 --> 00:00:26,452 S1: All right. Welcome to episode 496. All right, so my 2 00:00:26,452 --> 00:00:30,852 S1: video on building personal AI system using cloud code is out. 3 00:00:31,572 --> 00:00:34,332 S1: This thing has been a blocker for me in like 4 00:00:34,372 --> 00:00:38,092 S1: life and work and just it's taken so long to 5 00:00:38,092 --> 00:00:41,252 S1: get this thing out. Recording it took like 5 or 6 00:00:41,252 --> 00:00:45,732 S1: 7 takes or something. The blog post, the editing, everything 7 00:00:45,732 --> 00:00:48,532 S1: about it has been difficult. I'm so glad it is 8 00:00:48,531 --> 00:00:53,732 S1: out and recommend you go check it out. Just upgraded 9 00:00:53,732 --> 00:01:02,052 S1: to a Keychron Q11 keyboard. It is a canted and tilted. 10 00:01:03,372 --> 00:01:07,612 S1: I can't remember like which words associate with which angle, 11 00:01:07,612 --> 00:01:11,051 S1: but it's basically tilted inwards because that's the way your 12 00:01:11,052 --> 00:01:16,412 S1: hands are when you type. And then it's also tilted 13 00:01:16,412 --> 00:01:21,332 S1: upwards on the inside. So it's like an upside down V. 14 00:01:21,532 --> 00:01:25,131 S1: Kind of not as extreme though. So if you imagine 15 00:01:25,132 --> 00:01:27,372 S1: you just put your hands, your elbows by your side 16 00:01:27,372 --> 00:01:29,412 S1: and you put your hands out, they're going to be 17 00:01:29,411 --> 00:01:33,092 S1: like angled at like a, an upside down V shape, 18 00:01:34,612 --> 00:01:37,532 S1: you know, slightly like a wide open upside down v. 19 00:01:39,452 --> 00:01:43,172 S1: And they're also going to be moving. Well. If you 20 00:01:43,212 --> 00:01:46,612 S1: stick them straight out they will be straight out. But 21 00:01:46,612 --> 00:01:52,052 S1: if oftentimes you'll want to slightly cant them inwards or 22 00:01:52,052 --> 00:01:56,212 S1: you know uh tilt them inwards or not tilt but um, 23 00:01:56,732 --> 00:02:00,652 S1: angle them inwards. And if you do that then of 24 00:02:00,692 --> 00:02:06,012 S1: course you can um, you could reach the keys on 25 00:02:06,012 --> 00:02:08,852 S1: a regular keyboard that's lined up right now on a 26 00:02:08,852 --> 00:02:12,532 S1: regular keyboard that's lined up just the way a traditional 27 00:02:12,532 --> 00:02:15,692 S1: keyboard is. You actually have to tilt your your wrist 28 00:02:15,692 --> 00:02:20,302 S1: wrists inward, and it's just not it's not good for 29 00:02:20,302 --> 00:02:24,901 S1: you over time. I don't really have any pain that 30 00:02:24,902 --> 00:02:27,062 S1: I'm dealing with and trying to address here. I just 31 00:02:27,062 --> 00:02:28,702 S1: want it to be more natural. I want to be 32 00:02:28,742 --> 00:02:31,382 S1: able to type faster and I don't know. I also 33 00:02:31,382 --> 00:02:34,022 S1: think it looks cool. It feels cool as well. So 34 00:02:34,382 --> 00:02:39,542 S1: I like that. And then I'm experimenting with, um, the 35 00:02:39,582 --> 00:02:45,941 S1: keyboard because the keyboard, basically it has um, it's two 36 00:02:45,942 --> 00:02:49,501 S1: separate pieces. Okay. So the left is separated from the right. 37 00:02:49,502 --> 00:02:53,862 S1: That's like the most important feature. Um, and that allows 38 00:02:53,862 --> 00:02:57,542 S1: you to do this, this canting and everything completely separate 39 00:02:57,542 --> 00:03:01,462 S1: for left and right. Um, obviously you want them to match. 40 00:03:01,502 --> 00:03:05,462 S1: You want it to be symmetrical, basically. But the idea 41 00:03:05,462 --> 00:03:07,901 S1: is you can have them all the way across, like 42 00:03:07,942 --> 00:03:11,302 S1: the distance for from your elbows, uh, one elbow to 43 00:03:11,341 --> 00:03:17,022 S1: the next, which is what, you know, One, you know, 44 00:03:17,062 --> 00:03:21,702 S1: two feet, maybe something like two feet apart. You could 45 00:03:21,702 --> 00:03:24,782 S1: do that and then your hands would be just straight out. 46 00:03:25,181 --> 00:03:27,262 S1: And a lot of people who do like the custom keyboards, 47 00:03:27,262 --> 00:03:32,342 S1: they actually do that. I have mine probably at the 48 00:03:32,382 --> 00:03:41,102 S1: bottom of the keyboard. It's probably separated by, I would say, um. Actually, 49 00:03:41,102 --> 00:03:55,382 S1: I have a tape measure. Like five six inches is 50 00:03:55,382 --> 00:03:57,662 S1: how far the bottom is and the top is only 51 00:03:57,662 --> 00:04:01,262 S1: a couple of inches separated because again, I have it, 52 00:04:01,742 --> 00:04:06,622 S1: you know, canted downwards in that way. And it's pretty cool. 53 00:04:06,622 --> 00:04:09,302 S1: So the inside of the keyboard I have elevated. So 54 00:04:09,342 --> 00:04:12,942 S1: it's in that upside down V shape. And yeah, it 55 00:04:12,942 --> 00:04:16,902 S1: just feels really natural. Feels awesome. I've tried some other 56 00:04:16,902 --> 00:04:22,662 S1: more customizable keyboards that are just like 60. They're called 40% 57 00:04:22,662 --> 00:04:26,622 S1: keyboards and those you have to use layers to be 58 00:04:26,622 --> 00:04:29,421 S1: able to get to a lot of the keys. Um, 59 00:04:29,422 --> 00:04:34,342 S1: I didn't find that very fun. Um, the muscle memory 60 00:04:34,342 --> 00:04:38,302 S1: there was not easy to grow, and I feel like 61 00:04:38,541 --> 00:04:40,702 S1: there's not a reason to get rid of that many 62 00:04:40,742 --> 00:04:43,902 S1: keys and to force you to have to do layers. 63 00:04:45,262 --> 00:04:47,302 S1: I think this is a better approach. It's a more 64 00:04:47,302 --> 00:04:53,582 S1: traditional keyboard. So I have numbers above my fingers. I 65 00:04:53,582 --> 00:04:59,142 S1: still have my spacebar on both sides. I still have 66 00:04:59,142 --> 00:05:02,062 S1: number keys in the bottom right of my right hand. 67 00:05:02,782 --> 00:05:06,782 S1: So it's like it's more classic, more traditional, but at 68 00:05:06,782 --> 00:05:10,902 S1: the same time just oriented in a more ergonomic way. 69 00:05:11,062 --> 00:05:16,912 S1: So again, it's a Keychron Q11 and I do watch 70 00:05:16,912 --> 00:05:21,752 S1: too many keyboard videos on YouTube, so this might not 71 00:05:21,752 --> 00:05:29,111 S1: be my last keyboard. I have way too many keyboards, unfortunately. Okay, um, 72 00:05:29,791 --> 00:05:33,072 S1: found a major problem with doing member essays here in 73 00:05:33,072 --> 00:05:35,952 S1: the newsletter, so I'm not actually going to do them anymore. 74 00:05:35,952 --> 00:05:39,592 S1: And the problem is very simple. It's like I'm what 75 00:05:39,592 --> 00:05:42,432 S1: I'm trying to do is orient my work and my 76 00:05:42,432 --> 00:05:46,832 S1: life around the central theme of ideas. Right? I've been 77 00:05:46,832 --> 00:05:51,032 S1: trying to do this for years now, and the central 78 00:05:51,032 --> 00:05:54,751 S1: theme of ideas is you try to have ideas first 79 00:05:54,752 --> 00:05:57,112 S1: of all, and then you try to get them out 80 00:05:57,112 --> 00:06:00,832 S1: into the world. And what I would do is I 81 00:06:00,872 --> 00:06:02,712 S1: would be writing a newsletter and I would come up 82 00:06:02,712 --> 00:06:06,551 S1: with a cool idea, and I would do a member 83 00:06:06,552 --> 00:06:09,032 S1: essay on it, and I was going to do like 84 00:06:09,032 --> 00:06:10,912 S1: one a month, but I ended up doing it like 85 00:06:10,952 --> 00:06:14,752 S1: almost every month. But the problem is, they were so 86 00:06:15,272 --> 00:06:18,672 S1: decent I thought they were halfway good a lot of times. 87 00:06:19,752 --> 00:06:23,392 S1: At least several of them were halfway decent. And I'm like, well, 88 00:06:23,392 --> 00:06:25,912 S1: that just needs to be a regular blog post. Like, 89 00:06:25,912 --> 00:06:30,432 S1: I can't, you know, just it doesn't make any sense 90 00:06:30,432 --> 00:06:33,872 S1: to put the effort into it. Have people actually enjoy it, 91 00:06:33,872 --> 00:06:37,072 S1: but it only goes out to like 1500 people or 92 00:06:37,072 --> 00:06:41,632 S1: whatever the current member count is, right? So it's like 93 00:06:43,072 --> 00:06:47,312 S1: it just it was not in harmony with me trying 94 00:06:47,312 --> 00:06:51,152 S1: to get ideas out into the world as like a 95 00:06:51,192 --> 00:06:55,952 S1: primary mission. So. So I'm not going to do that anymore. Um, 96 00:06:55,952 --> 00:06:58,551 S1: basically I'm just going to turn those into blogs, right? 97 00:06:58,592 --> 00:07:00,152 S1: So you're still going to get them. They're still going 98 00:07:00,152 --> 00:07:02,752 S1: to be in the newsletter. It's just you're not going 99 00:07:02,791 --> 00:07:05,392 S1: to have to pay for them. And instead I'm going 100 00:07:05,472 --> 00:07:11,312 S1: to basically raise the member perk that has been most 101 00:07:11,352 --> 00:07:15,032 S1: asked for and just make that the primary perk. And 102 00:07:15,032 --> 00:07:20,952 S1: basically a clearly stated explicit thing of 25 to 50% 103 00:07:20,952 --> 00:07:25,992 S1: off all UL paid content. So when human 3.0 launches, 104 00:07:25,992 --> 00:07:31,432 S1: when augmented online, the online version of augmented launches, which 105 00:07:31,432 --> 00:07:37,072 S1: will be sometime this year, sometime soon, like you're just 106 00:07:37,072 --> 00:07:40,192 S1: going to get a massive discount like hundreds of dollars 107 00:07:40,192 --> 00:07:44,112 S1: or in some cases for courses or whatever. Or, you know, 108 00:07:44,152 --> 00:07:47,872 S1: conceivably if I charge enough for a thing, it'll be 109 00:07:47,912 --> 00:07:51,632 S1: thousands of dollars off for members, which is kind of 110 00:07:51,632 --> 00:07:56,552 S1: ridiculous given that the membership is only like $100. So 111 00:07:57,232 --> 00:08:00,272 S1: I don't know. I'll figure that out later. Uh, content 112 00:08:00,272 --> 00:08:05,552 S1: is more important in the meantime. So yeah, you can 113 00:08:05,552 --> 00:08:10,482 S1: subscribe at Daniel or subscribe. My friend Monica is running 114 00:08:10,482 --> 00:08:14,242 S1: her Resilient Career Accelerator course again soon. I'd recommend you 115 00:08:14,242 --> 00:08:16,482 S1: check it out. It's a week long course. How to 116 00:08:16,522 --> 00:08:20,322 S1: build a career in security leadership. Run a security program 117 00:08:20,322 --> 00:08:24,882 S1: and generally make yourself more resilient career wise in this 118 00:08:24,922 --> 00:08:27,762 S1: AI world. And she talks about that a lot. She's 119 00:08:27,762 --> 00:08:31,602 S1: very bright, very passionate about these topics. And I'm also 120 00:08:31,602 --> 00:08:34,442 S1: going to be a guest speaker at the course that 121 00:08:34,442 --> 00:08:40,881 S1: is in November. Uh, very much in line with my 122 00:08:40,882 --> 00:08:44,242 S1: recent Westerberg shares. You probably seen some of those videos 123 00:08:44,242 --> 00:08:48,122 S1: from her. I'm loving this guy. His name is Mainstone, 124 00:08:48,122 --> 00:08:50,882 S1: or at least that's his brand or whatever. And he's 125 00:08:50,881 --> 00:08:56,842 S1: just like this. Extremely honest, like British guy. And, uh, yeah, 126 00:08:56,842 --> 00:09:00,122 S1: he wears sunglasses and talks at the camera sometimes, but 127 00:09:00,122 --> 00:09:07,602 S1: he also is a videographer, photographer, videographer. Anyway, he takes 128 00:09:07,602 --> 00:09:12,122 S1: videos of like, terrain. And so his whole theme is 129 00:09:12,162 --> 00:09:14,242 S1: like going back to basics. It's like, what are we 130 00:09:14,242 --> 00:09:17,362 S1: actually doing here? It's a very existential, like, if I 131 00:09:17,402 --> 00:09:22,962 S1: had to categorize this YouTube channel, it's an existential YouTube channel. 132 00:09:23,362 --> 00:09:28,362 S1: And he's just he's smart. He his takes on AI, 133 00:09:28,402 --> 00:09:31,202 S1: I think are pretty clean. Like some of the cleanest 134 00:09:31,202 --> 00:09:34,642 S1: that I've seen, which I find to be hilarious. Like 135 00:09:34,642 --> 00:09:37,802 S1: this guy who is I don't know, I'm not even 136 00:09:37,802 --> 00:09:45,242 S1: sure how technical he is. Um, computer wise, um, video wise, 137 00:09:45,242 --> 00:09:49,761 S1: he's like, obviously extremely technical. The guy is. Yeah, he's 138 00:09:50,162 --> 00:09:53,362 S1: he could probably be a professional, professional video guy if 139 00:09:53,362 --> 00:09:56,442 S1: he isn't already, but I'm not sure if he's like 140 00:09:56,442 --> 00:10:00,482 S1: a computer guy. I don't really know that. Um, he 141 00:10:00,482 --> 00:10:04,122 S1: mentions being broke all the time and having like, $1,500 142 00:10:04,122 --> 00:10:06,881 S1: to his name and like, trying to find work. So 143 00:10:06,881 --> 00:10:11,122 S1: I'm guessing he doesn't have or hasn't had like a 144 00:10:11,122 --> 00:10:14,202 S1: tech career. That's I'm guessing he doesn't have that to 145 00:10:14,242 --> 00:10:18,161 S1: fall back on if he's talking about that. But in 146 00:10:18,442 --> 00:10:21,482 S1: the economy right now, who knows if that's actually the case? 147 00:10:21,522 --> 00:10:24,082 S1: Could be that he just can't find a job there. 148 00:10:25,202 --> 00:10:28,722 S1: But anyway, he goes his one of his favorite things 149 00:10:28,722 --> 00:10:30,882 S1: to do is just like to disappear and just like 150 00:10:30,922 --> 00:10:34,602 S1: go far, far away. He's also really good at Lan Nev, 151 00:10:34,642 --> 00:10:37,921 S1: by the way. Um, because I'm not that good at 152 00:10:37,962 --> 00:10:41,842 S1: Lan Nev. Or at least I'd have to train up. Um, 153 00:10:41,842 --> 00:10:45,682 S1: I think he does carry, like, a satellite device and, like, 154 00:10:45,682 --> 00:10:48,242 S1: batteries and stuff. So he could, like, actually get signal 155 00:10:48,242 --> 00:10:51,761 S1: if he needed to. I think he gets, like, weather reports, 156 00:10:51,761 --> 00:10:55,322 S1: but he's going places like very few people have been. 157 00:10:56,802 --> 00:11:01,002 S1: And he, like, disappears, like in, like, random parts of 158 00:11:01,002 --> 00:11:03,881 S1: Egypt or something, or like some random mountains in, like 159 00:11:03,881 --> 00:11:07,322 S1: Alaska or whatever. Just like disappears for like seven days. 160 00:11:07,722 --> 00:11:10,722 S1: And he's doing video the whole time. He's talking the 161 00:11:10,722 --> 00:11:14,802 S1: whole time. And it's really cool to see him actually 162 00:11:14,802 --> 00:11:18,722 S1: talk to the camera when he's feeling these extreme, like 163 00:11:18,761 --> 00:11:22,602 S1: emotions of solitude and like the wisdom he's getting from 164 00:11:22,602 --> 00:11:28,322 S1: it and everything. And his voiceovers, his narrations are great. Um, 165 00:11:28,362 --> 00:11:31,802 S1: his honesty. Like, he's just like, look, I just can't 166 00:11:31,802 --> 00:11:34,882 S1: figure this out. I don't know what I'm going to do. Like, 167 00:11:34,881 --> 00:11:39,562 S1: he's just very plain and open. It's extremely refreshing, very authentic. 168 00:11:39,802 --> 00:11:44,282 S1: Highly recommend it. Especially if you're kind of thinking, you know, 169 00:11:44,322 --> 00:11:46,322 S1: what is all this tech for? I want to go 170 00:11:46,322 --> 00:11:50,602 S1: back to basics, you know, analog. It's got that sort 171 00:11:50,602 --> 00:11:53,362 S1: of vibe to it. It's got the existential vibe to it, 172 00:11:54,082 --> 00:11:57,042 S1: gorgeous scenery as well. Like some of these shots he's taking. 173 00:11:57,042 --> 00:12:05,852 S1: Just absolutely amazing stuff. So anyway, highly recommend that. We 174 00:12:05,852 --> 00:12:13,172 S1: read Epictetus, the Enchiridion for book club. And it was wonderful. 175 00:12:13,412 --> 00:12:18,772 S1: Absolutely wonderful. Um, also reminding me so the Westerburg thing. 176 00:12:19,172 --> 00:12:23,572 S1: This guy mainstone, I don't know, all my stoic studies. 177 00:12:23,612 --> 00:12:26,812 S1: A million different things have, like been bringing me and 178 00:12:26,812 --> 00:12:29,572 S1: I don't know the attribution amounts, so I can't really 179 00:12:29,572 --> 00:12:34,412 S1: properly give credit. But I've also been reading a bunch of, like, 180 00:12:34,812 --> 00:12:38,531 S1: you know, dead tree books with no tech around me. 181 00:12:39,131 --> 00:12:41,692 S1: And I feel like whenever I get an insight from 182 00:12:41,732 --> 00:12:46,292 S1: a real book, an old book with text on a page, 183 00:12:46,492 --> 00:12:49,372 S1: I feel like the inside just hits me harder than, 184 00:12:51,252 --> 00:12:54,932 S1: you know, all the cool shit that I'm getting from YouTube, right? 185 00:12:54,972 --> 00:12:58,932 S1: And I'm getting great distilled, highly produced, great stuff from YouTube. 186 00:12:59,292 --> 00:13:02,372 S1: And somehow the book stuff just hits me harder. And 187 00:13:02,372 --> 00:13:07,492 S1: I'm trying to disentangle Like, is it because it's better content? 188 00:13:07,492 --> 00:13:10,572 S1: Is it become because it's coming from a different angle 189 00:13:10,572 --> 00:13:13,292 S1: where I'm not ready for it and I'm like surprised 190 00:13:13,292 --> 00:13:18,052 S1: by it. But one of my possible frames for this 191 00:13:18,052 --> 00:13:26,572 S1: is that it's simply like. One of the models I 192 00:13:26,572 --> 00:13:30,612 S1: was thinking of is like, I heard about on Reddit somewhere, 193 00:13:30,612 --> 00:13:35,292 S1: that some people use tea bags like 12 times and 194 00:13:35,292 --> 00:13:37,652 S1: like they keep track with like a little counter of 195 00:13:37,692 --> 00:13:40,692 S1: like how many times it's like, oh, 11th time. Well, 196 00:13:40,692 --> 00:13:43,092 S1: we haven't got enough out of it. So they keep 197 00:13:43,092 --> 00:13:45,972 S1: using it. And I'm like, I wonder if the content 198 00:13:45,972 --> 00:13:49,252 S1: that I'm getting from YouTube and from like, all these, 199 00:13:49,292 --> 00:13:53,612 S1: you know, modern sources or whatever, it's just like used 200 00:13:53,652 --> 00:13:57,651 S1: tea bags like 12 times used tea bags, it's just 201 00:13:57,692 --> 00:14:02,772 S1: it's good, but it's massively diluted. Just like, not the 202 00:14:02,892 --> 00:14:05,172 S1: good stuff. And I wonder if the stuff that I'm 203 00:14:05,212 --> 00:14:09,692 S1: getting from these books. I don't know, I can't tell 204 00:14:09,732 --> 00:14:12,052 S1: if it's just a better tea leaf, or if it's 205 00:14:12,052 --> 00:14:15,612 S1: just an undiluted tea leaf, or if it's just. I'm 206 00:14:15,652 --> 00:14:20,572 S1: not used to taking tea leaves by snorting them or 207 00:14:20,692 --> 00:14:22,732 S1: shoving them into my ears like, I don't know if 208 00:14:22,732 --> 00:14:25,572 S1: it's the medium. Like I'm still trying to figure that out. 209 00:14:25,572 --> 00:14:30,652 S1: So if anyone has any ideas there. But bottom line 210 00:14:30,652 --> 00:14:35,092 S1: is I am getting extremely excited about this concept of. 211 00:14:38,612 --> 00:14:47,612 S1: Old sources. Classics, but not even classics like literature. Literature. 212 00:14:48,172 --> 00:14:57,692 S1: I'm getting so much out of the biography of. Roosevelt. 213 00:14:58,012 --> 00:15:00,372 S1: Teddy Roosevelt. I was thinking of the other one. I 214 00:15:00,412 --> 00:15:03,302 S1: get them confused. But yeah. FDR is the later one. 215 00:15:03,302 --> 00:15:08,062 S1: He was the economy thing. FDR policies and I think 216 00:15:08,062 --> 00:15:12,982 S1: New Deal and all that. But Teddy Roosevelt was much younger. 217 00:15:12,982 --> 00:15:21,702 S1: He was doing things early 1900s or actually, no, 1800s. Yeah. Um, 218 00:15:21,702 --> 00:15:26,782 S1: 18 like 76 ish or whatever. Uh, roughly around there. 219 00:15:26,782 --> 00:15:28,902 S1: I think he was in his teens, like 100 years 220 00:15:28,902 --> 00:15:33,181 S1: after the country was born or something like that. Garfield 221 00:15:33,182 --> 00:15:37,382 S1: was assassinated, uh, during that time, something that just happened 222 00:15:37,382 --> 00:15:44,462 S1: in the book. But anyway, uh, he was. Yeah. This 223 00:15:44,462 --> 00:15:48,622 S1: biography was talking about his life, and he was talking about, like, um, 224 00:15:49,342 --> 00:15:51,382 S1: so he was going through his journals. That's what the 225 00:15:51,382 --> 00:15:54,542 S1: biography is largely source based on is his journals like 226 00:15:54,582 --> 00:16:00,382 S1: Roosevelt wrote tons. And the guy was just absolutely insane, like, lie. 227 00:16:00,382 --> 00:16:03,542 S1: Completely insane. He would just. He was like a boxer, 228 00:16:03,582 --> 00:16:08,262 S1: like a fisher, like dancing, but not very well. But 229 00:16:08,262 --> 00:16:14,462 S1: he just spoke multiple languages and literature, like he just 230 00:16:14,462 --> 00:16:18,982 S1: read everything. And recitations, evidently was a huge thing at Harvard, 231 00:16:19,862 --> 00:16:23,262 S1: going from place to place, doing recitations, like, I don't know, 232 00:16:23,262 --> 00:16:28,061 S1: is that people reading literature is that people reading memorized 233 00:16:28,062 --> 00:16:33,222 S1: literature without the book. I'm not sure exactly what that is, 234 00:16:33,222 --> 00:16:38,502 S1: but I need to go research that. But. I feel 235 00:16:38,502 --> 00:16:42,422 S1: like this content I don't again, I don't know if 236 00:16:42,422 --> 00:16:45,102 S1: it's the tea leaf itself or if it's the medium, 237 00:16:45,462 --> 00:16:49,422 S1: or if it's just different and it's just a different 238 00:16:49,422 --> 00:16:53,662 S1: style with different, you know, diction and and it spawns 239 00:16:53,702 --> 00:16:56,622 S1: thoughts in the brain differently. But I just feel like 240 00:16:56,622 --> 00:17:03,022 S1: people were so sharp back then? Um, I don't know. 241 00:17:03,062 --> 00:17:07,062 S1: I think they were extremely limited in other ways as well. 242 00:17:07,542 --> 00:17:11,582 S1: So it's hard to, uh, you know, measure it on balance. 243 00:17:11,582 --> 00:17:20,142 S1: But anyway, I'm getting extraordinary creative, uh, boosts from reading 244 00:17:20,142 --> 00:17:25,662 S1: these older books. I'm reading a book right now about, um, 245 00:17:26,742 --> 00:17:33,702 S1: rhetorical figures. It absolutely insane. It's like it's rhetoric, but 246 00:17:33,702 --> 00:17:38,101 S1: specifically about something called figures in rhetoric. Uh, it's a 247 00:17:38,102 --> 00:17:42,342 S1: book about writing, and I'm just loving every single page. 248 00:17:42,342 --> 00:17:45,302 S1: I also just love how the page is written. I 249 00:17:45,302 --> 00:17:48,902 S1: love the writing itself in this book. I think it's 250 00:17:48,902 --> 00:17:52,022 S1: using like it's tricks that it's teaching me. It's using 251 00:17:52,022 --> 00:17:54,742 S1: them on me while it's teaching me. Pretty sure it's 252 00:17:54,742 --> 00:17:59,112 S1: doing that. But what I find is it's very similar. 253 00:17:59,112 --> 00:18:02,952 S1: I got to post somewhere like my favorite writers. So Hitchens, 254 00:18:03,832 --> 00:18:09,032 S1: Scott Adams, the Dilbert guy before he went crazy, although 255 00:18:09,032 --> 00:18:13,912 S1: he probably still writes well now, um, not so much 256 00:18:13,912 --> 00:18:16,592 S1: Sam Harris. I like him for a different reason, but 257 00:18:17,232 --> 00:18:22,672 S1: the writing style that I like the most, and that 258 00:18:22,672 --> 00:18:26,432 S1: I use and probably emulate quite a bit is this 259 00:18:26,432 --> 00:18:33,672 S1: alternate alteration alternation between long and short sentences? It's very powerful. 260 00:18:33,992 --> 00:18:39,351 S1: I'm seeing this happen in this book here again. Um, 261 00:18:39,352 --> 00:18:43,392 S1: Hitchens did it. It's it's in several books that I've 262 00:18:43,392 --> 00:18:46,431 S1: read about good writing, so I'm not sure what the 263 00:18:46,432 --> 00:18:48,311 S1: actual source is. I'm not sure if it's one of 264 00:18:48,311 --> 00:18:55,032 S1: these figures, these rhetorical figures. But anyway, really, really powerful stuff. Um, 265 00:18:56,832 --> 00:19:00,552 S1: and we are 700 minutes into this podcast. Um, let's 266 00:19:00,552 --> 00:19:06,192 S1: get into it. That was just the intro. Cybersecurity. Anthropic 267 00:19:06,192 --> 00:19:09,912 S1: says attackers are using models to automate full hacking operations. 268 00:19:09,912 --> 00:19:12,472 S1: So this is the thing I talked about in my 269 00:19:12,472 --> 00:19:15,472 S1: Future of Hacking video, which you should definitely go look 270 00:19:15,472 --> 00:19:18,712 S1: at if you haven't seen that. Um, and that's just 271 00:19:18,752 --> 00:19:26,872 S1: a derivation of or derivative video, a hacking focused version 272 00:19:26,872 --> 00:19:30,831 S1: of my unified entity context, which is like this unified 273 00:19:30,832 --> 00:19:34,472 S1: context for the enterprise that everything runs off of, where 274 00:19:34,472 --> 00:19:37,272 S1: the AI talks to it. And it's like, you know, 275 00:19:37,311 --> 00:19:42,872 S1: the context is all in one place. Well. Anthropic just 276 00:19:42,872 --> 00:19:45,152 S1: came out and basically said, look, we've we were tracking 277 00:19:45,152 --> 00:19:51,032 S1: this actor, and the actor basically was acting like a 278 00:19:51,032 --> 00:19:54,911 S1: giant team, like, like a team of multiple people. And 279 00:19:54,952 --> 00:20:01,552 S1: this one person accomplished what normally takes large teams, you know, 280 00:20:01,592 --> 00:20:08,152 S1: months to do. So, uh, 17 organizations in weeks using Claude. 281 00:20:08,792 --> 00:20:10,752 S1: So somehow they were getting Claude to do all this, 282 00:20:10,752 --> 00:20:14,232 S1: by the way. So they were bypassing whatever restrictions Claude 283 00:20:14,272 --> 00:20:19,591 S1: had healthcare, governments, defense contractors, even a church. It, uh, 284 00:20:19,712 --> 00:20:24,192 S1: Claude analyzed stolen data, created customized extortion plans for each victim. 285 00:20:24,752 --> 00:20:30,312 S1: North Korean agents use Claude to fake IT worker identities 286 00:20:30,311 --> 00:20:33,912 S1: at companies. They asked Claude basic questions like what is 287 00:20:33,912 --> 00:20:37,311 S1: a muffin to maintain cover? So this is one of the, uh, 288 00:20:37,512 --> 00:20:43,391 S1: prompt injection bypass techniques, or at least, you know, to 289 00:20:43,432 --> 00:20:48,232 S1: bypass the policy of, like. Hey, you seem to be suspect. Uh, 290 00:20:48,232 --> 00:20:51,831 S1: fake employees can earn high salaries. Romance scammers use Claude 291 00:20:51,832 --> 00:20:59,121 S1: as their emotionally intelligent conversation engine. That's diabolical. Chinese hackers 292 00:20:59,122 --> 00:21:04,362 S1: use Claude to guide espionage against Vietnamese telecom companies. Criminals 293 00:21:04,362 --> 00:21:08,442 S1: are building ransomware as a service. Businesses powered by AI. 294 00:21:09,282 --> 00:21:13,002 S1: So all of this is exactly what I've been talking 295 00:21:13,002 --> 00:21:16,842 S1: about here, where it's like it's their AI system against 296 00:21:16,842 --> 00:21:21,962 S1: our AI system, right? It's them defining goals and the 297 00:21:21,962 --> 00:21:25,602 S1: AI basically using all its different tools, which is essentially 298 00:21:25,602 --> 00:21:28,202 S1: what I just talked about in my personal AI video, 299 00:21:28,922 --> 00:21:32,482 S1: where I'm talking about, like, you have a central system 300 00:21:33,082 --> 00:21:35,321 S1: and you give it all these tools and you say, look, 301 00:21:35,362 --> 00:21:37,282 S1: go do this, and it picks the tool to go 302 00:21:37,282 --> 00:21:41,562 S1: do it, and it does it. And it looks like 303 00:21:41,561 --> 00:21:45,722 S1: a system probably very similar to what I've built is 304 00:21:45,722 --> 00:21:50,962 S1: being used by this attacker. And I'm sure many thousands 305 00:21:50,962 --> 00:21:54,641 S1: of others, or at least hundreds of others soon to 306 00:21:54,682 --> 00:21:57,802 S1: be thousands. And maybe, I don't know, maybe more than 307 00:21:57,802 --> 00:22:02,121 S1: that later. I just think this is the way to go, right? 308 00:22:02,402 --> 00:22:05,282 S1: Everyone's going to have an automation stack. The question is 309 00:22:05,282 --> 00:22:08,962 S1: what are they using it for? And so what do 310 00:22:08,962 --> 00:22:10,841 S1: you have to do as a defender. You've got to 311 00:22:10,842 --> 00:22:13,842 S1: have the same exact automation stack. That was my takeaway 312 00:22:13,842 --> 00:22:18,081 S1: from this. You know future of hacking is context. Video 313 00:22:18,082 --> 00:22:20,282 S1: is basically what's the answer. The answer is to have 314 00:22:20,282 --> 00:22:24,442 S1: the same thing that the attackers have. And the framing 315 00:22:24,442 --> 00:22:27,122 S1: that I have for all of this is you have 316 00:22:27,122 --> 00:22:34,002 S1: to have a, a world model of your system and 317 00:22:34,002 --> 00:22:37,082 S1: your company and your employees and your projects and your 318 00:22:37,082 --> 00:22:39,601 S1: budget and everything. You have to have a world model 319 00:22:39,602 --> 00:22:42,482 S1: of that whole thing, and you have to be able 320 00:22:42,482 --> 00:22:46,242 S1: to audit that and understand its state at any given time. 321 00:22:47,282 --> 00:22:50,282 S1: And then at the bigger picture, the whole game is. 322 00:22:53,482 --> 00:23:02,441 S1: Excuse me. Take a look. At the bigger picture. The 323 00:23:02,442 --> 00:23:13,962 S1: ultimate game is going from current state to desired state. And? Anyway, 324 00:23:14,002 --> 00:23:17,682 S1: that's kind of a bigger topic. But the point is, 325 00:23:17,682 --> 00:23:23,321 S1: attackers that I made before is attackers will have this 326 00:23:23,922 --> 00:23:26,322 S1: world model of your company, and they will use that 327 00:23:26,321 --> 00:23:31,321 S1: to use their automation system to continuously launch attacks against you. 328 00:23:32,162 --> 00:23:34,561 S1: And basically what your logs are going to look like 329 00:23:34,762 --> 00:23:42,642 S1: is thousands of AI automation systems just hitting you constantly. 330 00:23:42,642 --> 00:23:48,841 S1: Just constantly. And if you fart or you burp or 331 00:23:48,842 --> 00:23:53,132 S1: you step out of line one second With like an 332 00:23:53,132 --> 00:23:58,332 S1: S3 bucket or a misconfiguration or, you know, a DNS 333 00:23:58,332 --> 00:24:03,012 S1: domain that you're not maintaining whatever it is that is 334 00:24:03,012 --> 00:24:06,052 S1: going to get jumped on by a whole bunch of 335 00:24:06,052 --> 00:24:12,972 S1: AI who does not sleep. Right. So your attacker is 336 00:24:12,972 --> 00:24:16,252 S1: absolutely going to jump on that, and they're going to 337 00:24:16,252 --> 00:24:19,292 S1: do it fast. And if not, one of their competitors 338 00:24:19,292 --> 00:24:21,611 S1: is going to do it. So we're not going to 339 00:24:21,612 --> 00:24:25,252 S1: have the time. So our only chance is we get 340 00:24:25,252 --> 00:24:29,811 S1: there first. Our automated AI system sees that thing pop 341 00:24:29,811 --> 00:24:34,772 S1: up first before them and goes and fixes it, or 342 00:24:34,772 --> 00:24:37,212 S1: blocks it, or does whatever. You know, it puts up 343 00:24:37,212 --> 00:24:43,252 S1: a firewall rule or whatever. So this is the game. 344 00:24:43,332 --> 00:24:48,292 S1: To me, this is the security story. This is the 345 00:24:48,292 --> 00:24:52,252 S1: number one security story going forward. Everyone talks about, you know, 346 00:24:52,972 --> 00:24:55,851 S1: AI and security. Hey, what are the future trends or whatever? 347 00:24:56,492 --> 00:25:00,612 S1: This is my number one answer. Number one answer. Unified 348 00:25:00,612 --> 00:25:04,972 S1: entity context to get everything unified inside of a company. 349 00:25:05,732 --> 00:25:11,932 S1: So the future of basically company management is the founders. 350 00:25:12,811 --> 00:25:16,252 S1: The owners sit at a giant table with a bunch 351 00:25:16,252 --> 00:25:21,972 S1: of other leaders who are also extremely business savvy, creative 352 00:25:22,732 --> 00:25:27,532 S1: and and technical as well. And then there is a 353 00:25:27,532 --> 00:25:33,331 S1: number of similar minded people that are basically like wizards, 354 00:25:35,052 --> 00:25:38,372 S1: and that essentially is their IT staff. And their IT 355 00:25:38,412 --> 00:25:45,372 S1: staff isn't really IT staff. It's more like business IT liaison. 356 00:25:45,492 --> 00:25:49,812 S1: So it's like they are experts at understanding exactly what 357 00:25:49,811 --> 00:25:52,252 S1: the leaders at that table just said, because they're at 358 00:25:52,252 --> 00:25:55,132 S1: the table as well. But they're kind of like the 359 00:25:55,132 --> 00:25:59,772 S1: execution arm. And then what they do is they they're like, okay, 360 00:25:59,811 --> 00:26:03,452 S1: I just understood that I am parsing that and bringing 361 00:26:03,452 --> 00:26:07,452 S1: it into the unified entity context that is now just updated. 362 00:26:07,452 --> 00:26:11,612 S1: Our goals. Our goals have now been updated. All our 363 00:26:11,612 --> 00:26:17,332 S1: plans have been rewritten. Um, these additional details have been 364 00:26:17,332 --> 00:26:20,972 S1: written into all future iterations of all of our products. 365 00:26:21,772 --> 00:26:24,411 S1: For all the products that you just mentioned building, we 366 00:26:24,452 --> 00:26:29,571 S1: have now started specking those out and they're basically wielding 367 00:26:29,571 --> 00:26:36,972 S1: these giant teams, these giant armies of. AI systems. And 368 00:26:36,972 --> 00:26:41,651 S1: those AI systems are the planners and the speakers and the, 369 00:26:42,172 --> 00:26:44,371 S1: you know, they're going to build marketing campaigns and they're 370 00:26:44,372 --> 00:26:47,332 S1: going to build all this different stuff around that. But 371 00:26:47,332 --> 00:26:49,462 S1: it has to start at that table with the people 372 00:26:49,462 --> 00:26:54,141 S1: deciding actually what to do. Right now, of course, there's 373 00:26:54,382 --> 00:26:56,702 S1: arguments that, okay, well, in the future you don't need 374 00:26:56,742 --> 00:26:59,821 S1: them because, you know, the AI will just be smarter 375 00:26:59,821 --> 00:27:04,142 S1: than all of them. Honestly, I believe that. Right? And 376 00:27:04,142 --> 00:27:08,982 S1: that's that's the ACI story. Very possible. But that's not 377 00:27:08,982 --> 00:27:12,702 S1: a security thing. I care about right now. It's just 378 00:27:12,702 --> 00:27:16,142 S1: not tangible to me yet. Maybe in five years, maybe 379 00:27:16,142 --> 00:27:19,982 S1: ten years, maybe 20 years. And who knows, it could 380 00:27:19,982 --> 00:27:24,462 S1: be two years. But for the foreseeable future, and I 381 00:27:24,462 --> 00:27:28,621 S1: think it'll be a good amount of time, even if 382 00:27:28,662 --> 00:27:31,182 S1: ACI hits, it's going to take a while. It's going 383 00:27:31,222 --> 00:27:35,302 S1: to be that leadership team in the middle, um, with 384 00:27:35,302 --> 00:27:39,581 S1: an enforcement arm of these extremely bright people wielding giant 385 00:27:39,622 --> 00:27:46,422 S1: teams of thousands or millions of tiny little AIS doing execution. 386 00:27:47,022 --> 00:27:52,061 S1: And in terms of security, really, it's about everything. It's 387 00:27:52,061 --> 00:27:56,862 S1: continuous monitoring and continuous context management and understanding of the business. 388 00:27:56,862 --> 00:28:01,342 S1: But especially for security, you have to continuously understand the 389 00:28:01,342 --> 00:28:05,702 S1: state of what you have, which will previously have been 390 00:28:05,702 --> 00:28:09,102 S1: called asset management. But really it's like state management of 391 00:28:09,102 --> 00:28:14,222 S1: the business overall. And then once again, it's your AI 392 00:28:14,262 --> 00:28:19,662 S1: systems against their AI systems. It's just that's just reality. 393 00:28:20,182 --> 00:28:24,262 S1: How fast are yours. How well do they understand the context? 394 00:28:24,662 --> 00:28:27,302 S1: How small is the gap between a change and an 395 00:28:27,302 --> 00:28:34,302 S1: understanding of the change that is the game. Okay. Um, 396 00:28:34,742 --> 00:28:40,142 S1: anthropic shows how they detect sleeper agents hiding malicious behavior. Yeah. 397 00:28:40,182 --> 00:28:43,302 S1: Catching AI agents that pretend to be helping while secretly 398 00:28:43,302 --> 00:28:49,142 S1: planning harmful actions when deployed. FBI confirmed Salt Typhoon breached 399 00:28:49,142 --> 00:28:53,022 S1: at least 200 American companies plus firms in 80 countries. 400 00:28:54,102 --> 00:29:00,222 S1: NSA and NCIS trace the Salt Typhoon espionage campaigns to 401 00:29:00,582 --> 00:29:07,462 S1: Sichuan chun shui Beijing. How you knew? Why would I? 402 00:29:07,502 --> 00:29:09,582 S1: Why would I choose to read these things? It's not 403 00:29:09,582 --> 00:29:13,342 S1: going to help me. I don't want to butcher people's names. Um, 404 00:29:13,342 --> 00:29:16,702 S1: but these are companies. Actually, not not people. So those 405 00:29:16,702 --> 00:29:22,822 S1: are different companies providing Chinese company, providing cyber tools to 406 00:29:22,862 --> 00:29:29,022 S1: China's intelligence services. Uh, Claude will start training on personal 407 00:29:29,022 --> 00:29:31,421 S1: account data unless you opt out. So this is an 408 00:29:31,422 --> 00:29:34,902 S1: important one. They you probably seen the emails if you're 409 00:29:34,902 --> 00:29:37,662 S1: a user. Uh, they emailed everyone and basically said you 410 00:29:37,662 --> 00:29:41,422 S1: have until September 28th to opt out. I wouldn't opt 411 00:29:41,422 --> 00:29:48,112 S1: it out immediately. Um, if you don't want your data 412 00:29:48,112 --> 00:29:50,112 S1: to be trained on and if you have like a 413 00:29:50,112 --> 00:29:53,872 S1: custom like enterprise account or certain types of, you know, 414 00:29:53,912 --> 00:29:57,832 S1: paid accounts, they won't do this by default. But if 415 00:29:57,832 --> 00:30:00,112 S1: you just have a regular account, including like a pro 416 00:30:00,112 --> 00:30:03,672 S1: account or a max account or whatever, um, they will 417 00:30:03,672 --> 00:30:09,632 S1: turn this on. They will start using your content for training. Uh, 418 00:30:09,632 --> 00:30:11,671 S1: if you do not opt out. So definitely go and 419 00:30:11,672 --> 00:30:15,072 S1: do that if you want to. I'm also curious about 420 00:30:15,072 --> 00:30:19,832 S1: why everyone's freaking out about this. Don't most other AI 421 00:30:19,872 --> 00:30:23,432 S1: companies already do this? Like, I feel like this is 422 00:30:23,432 --> 00:30:26,671 S1: just pretty standard at this point. Um, what I don't 423 00:30:26,672 --> 00:30:30,432 S1: understand and what I wouldn't like is if they're like, 424 00:30:30,472 --> 00:30:34,272 S1: we don't have the ability to opt out before September 28th. 425 00:30:34,272 --> 00:30:38,672 S1: Why would they say before September 28th? I guess that means. 426 00:30:41,072 --> 00:30:43,512 S1: I guess that just means that they will turn it on. 427 00:30:44,512 --> 00:30:50,912 S1: They will turn on the use after September 28th, but 428 00:30:50,952 --> 00:30:54,272 S1: hopefully it also means that you can still opt out 429 00:30:54,272 --> 00:30:57,472 S1: after that point. I'm not sure they made that clear 430 00:30:57,592 --> 00:30:59,432 S1: in the email. I didn't see it in the email, 431 00:31:00,232 --> 00:31:02,752 S1: but hopefully that would be the case. It would be 432 00:31:02,752 --> 00:31:05,752 S1: super shitty if they were like, nope, we sent you 433 00:31:05,752 --> 00:31:07,712 S1: the thing. I told you you had to do it 434 00:31:07,712 --> 00:31:11,712 S1: before September 28th and if you didn't? Too bad. That 435 00:31:11,712 --> 00:31:14,952 S1: would be weaksauce. And the more I think about it, 436 00:31:14,952 --> 00:31:17,112 S1: that's so ridiculous. I don't think that's going to be 437 00:31:17,112 --> 00:31:23,712 S1: the case. Researchers hide malicious prompts in images that only 438 00:31:23,712 --> 00:31:29,032 S1: appear when AI systems downscale them. This is from Trail 439 00:31:29,032 --> 00:31:34,352 S1: of Bits. Really cool attack. Uh, worth checking out. DoD 440 00:31:34,512 --> 00:31:39,872 S1: uses node JS utility maintained by a Russian Yandex developer. 441 00:31:41,272 --> 00:31:45,152 S1: Supply chain Transparency. This is why I can't wait till 442 00:31:45,152 --> 00:31:49,752 S1: there's like, you know, millions upon millions of AI eyes 443 00:31:49,872 --> 00:31:54,112 S1: watching everything. Oh, here's a cool idea that, um, I'm 444 00:31:54,112 --> 00:31:56,792 S1: going to be doing a panel with my buddy Sasha's 445 00:31:56,792 --> 00:32:02,232 S1: dealer pretty soon. And by the way, sorry for my voice. 446 00:32:02,232 --> 00:32:07,992 S1: I'm just realizing because I'm trying not to sniff while 447 00:32:08,312 --> 00:32:12,192 S1: on the microphone because it's a sniffing sound. But if 448 00:32:12,192 --> 00:32:16,712 S1: I'm trying to sniff, that means I have the sniffles, 449 00:32:16,712 --> 00:32:19,872 S1: which means I probably sound a little bit nasal, so 450 00:32:19,872 --> 00:32:24,352 S1: apologize for that. Not sick, but I'm like one quarter sick. 451 00:32:24,352 --> 00:32:28,792 S1: I just don't feel bad. Anyway, uh, the idea here is, 452 00:32:29,232 --> 00:32:33,592 S1: you know how we were told many eyes sunlight is 453 00:32:33,592 --> 00:32:36,032 S1: the best disinfectant and all this. And then we have 454 00:32:36,032 --> 00:32:39,432 S1: the SSL thing, which was open source, and it was 455 00:32:39,432 --> 00:32:44,992 S1: a massive, massive vulnerability across the whole internet. Well, what 456 00:32:44,992 --> 00:32:51,832 S1: if AI is basically a revisiting of many eyes, right? 457 00:32:51,872 --> 00:32:54,552 S1: Because Many Eyes turned out not to be true. Not 458 00:32:54,552 --> 00:33:00,872 S1: to be real. Because many eyes implied that because people 459 00:33:00,872 --> 00:33:03,832 S1: could look at the code that they would, that they 460 00:33:03,832 --> 00:33:06,592 S1: actually would. Right. And that turned out not to be 461 00:33:06,592 --> 00:33:10,592 S1: the case. Nobody was looking at all these open source projects. 462 00:33:10,912 --> 00:33:14,472 S1: Just because they're open source doesn't automatically make them secure, 463 00:33:14,472 --> 00:33:17,552 S1: because that implies a second step, which is lots of 464 00:33:17,552 --> 00:33:20,552 S1: people are actually taking the time to look. Why are 465 00:33:20,552 --> 00:33:24,472 S1: they doing that? What time are they doing that in? 466 00:33:24,992 --> 00:33:28,552 S1: How are they incentivized to do that? There's no answers 467 00:33:28,552 --> 00:33:31,752 S1: to those questions, which means which is the reason that 468 00:33:31,792 --> 00:33:35,032 S1: it wasn't being done. But what if we could just 469 00:33:35,032 --> 00:33:38,392 S1: unleash like this army? Like the government can unleash an 470 00:33:38,392 --> 00:33:44,762 S1: army of millions of agents, and their job is to 471 00:33:44,802 --> 00:33:48,082 S1: literally just go and clean up all this stuff. And 472 00:33:48,562 --> 00:33:54,362 S1: funny enough, that's exactly what the AI project was about. 473 00:33:54,402 --> 00:33:58,482 S1: Guess who? It was launched by DARPA. Yes, the creators 474 00:33:58,482 --> 00:34:04,002 S1: of the interwebs. So they literally created a project that 475 00:34:04,002 --> 00:34:11,482 S1: does this exact thing. It autonomously, autonomously, using agents can 476 00:34:11,482 --> 00:34:15,042 S1: go and find bugs and fix them end to end 477 00:34:15,202 --> 00:34:23,282 S1: without breaking the software. Isn't that insane? And this is what? Trilobites. 478 00:34:23,282 --> 00:34:26,242 S1: They just won second place. This is what Michael Brown 479 00:34:26,242 --> 00:34:29,122 S1: was talking about, which I haven't interviewed, which you should 480 00:34:29,122 --> 00:34:32,722 S1: go check out. It's on YouTube. And yeah, this is 481 00:34:32,722 --> 00:34:36,642 S1: exactly what the government set out to build. And this 482 00:34:36,642 --> 00:34:40,762 S1: is like, this is many eyes. That is literally what 483 00:34:40,762 --> 00:34:42,682 S1: many eyes was supposed to be. It was supposed to 484 00:34:42,682 --> 00:34:48,562 S1: be humans. Turns out that doesn't really scale. Um, because 485 00:34:48,602 --> 00:34:52,122 S1: open source turns out to be just one guy. I 486 00:34:52,162 --> 00:34:55,882 S1: saw somebody say that recently. Actually, it wasn't a person. 487 00:34:55,882 --> 00:34:58,082 S1: I just saw it as a headline. Open source is 488 00:34:58,082 --> 00:35:01,962 S1: just one guy. Maybe it's actually in the, uh, in 489 00:35:02,002 --> 00:35:05,882 S1: the discovery section. Open source is just one guy. That 490 00:35:05,882 --> 00:35:11,802 S1: is hilarious. Uh, anyway. FBI warns about three phase phantom 491 00:35:11,802 --> 00:35:17,242 S1: hacker scam targeting seniors. Scammers use AI to analyze social 492 00:35:17,242 --> 00:35:21,442 S1: media and personalize their attacks. Oh yeah? What do you know? First, 493 00:35:21,482 --> 00:35:24,082 S1: they pose as tech support and gain remote computer access. 494 00:35:24,082 --> 00:35:28,322 S1: Then they fake bank employees, convince victims their accounts are 495 00:35:28,322 --> 00:35:34,762 S1: compromised overseas. Just so nasty, so nasty. Just going after seniors. 496 00:35:35,802 --> 00:35:38,082 S1: And it's just going to be done at scale now. 497 00:35:39,202 --> 00:35:48,282 S1: National security. Pentagon security agency admits China keeps beating their defenses. Sad. 498 00:35:48,602 --> 00:35:54,722 S1: AI Cloudflare radar tracks global AI bot traffic across the internet. 499 00:35:55,762 --> 00:36:02,202 S1: This thing is really cool looking. Yeah, it shows you 500 00:36:02,202 --> 00:36:05,162 S1: all the stats for all the crawlers and everything. Super 501 00:36:05,162 --> 00:36:10,202 S1: super cool. AI is the natural step in making computers 502 00:36:10,202 --> 00:36:15,002 S1: understand humans. AI browser agents fail at real work. Except 503 00:36:15,002 --> 00:36:21,482 S1: for one unknown tool. Yeah, this was a really interesting one. 504 00:36:21,522 --> 00:36:24,962 S1: You should go check out that one in the newsletter. 505 00:36:26,042 --> 00:36:31,842 S1: AI adoption linked to 13% job decline for young US workers. Yeah, 506 00:36:31,842 --> 00:36:34,882 S1: I saw, uh, one of my smart buddies who was like, yeah, 507 00:36:34,922 --> 00:36:38,812 S1: you know, there's no evidence that AI is affecting the 508 00:36:38,812 --> 00:36:44,132 S1: job market. I'm like, what are you reading? Like, like 509 00:36:44,172 --> 00:36:47,332 S1: the CEOs, so many founders are like, yeah, I'm not 510 00:36:47,372 --> 00:36:49,692 S1: hiring anymore. I'm looking into how I could do more 511 00:36:49,692 --> 00:36:53,532 S1: of this with AI. And it's just like dozens of 512 00:36:53,532 --> 00:36:57,092 S1: these people coming out and saying this. The job stats 513 00:36:57,132 --> 00:36:59,332 S1: are going down at the same time. And it doesn't 514 00:36:59,332 --> 00:37:04,412 S1: mean like that's necessarily proof, right? You know, it's correlation, right. 515 00:37:04,612 --> 00:37:08,732 S1: But then you just see more and more people saying, yeah, 516 00:37:08,772 --> 00:37:12,812 S1: I'm automated this, you know, I have automated this. I 517 00:37:12,812 --> 00:37:15,892 S1: was told directly that the team was being laid off 518 00:37:15,892 --> 00:37:19,332 S1: because of AI reasons. Like, there's just so much direct 519 00:37:19,332 --> 00:37:22,812 S1: evidence that we have from people, you know, which is 520 00:37:22,812 --> 00:37:26,052 S1: like anecdata you add those up and they turn into 521 00:37:26,092 --> 00:37:31,692 S1: like almost data. So 13% I think is low. I 522 00:37:31,692 --> 00:37:35,972 S1: think it's just just now starting. And yeah. And actually 523 00:37:36,092 --> 00:37:41,332 S1: I have things queued for the next newsletter that, um, 524 00:37:41,692 --> 00:37:44,372 S1: that backs that up. There's actually more studies coming out. 525 00:37:44,372 --> 00:37:48,692 S1: It's probably going to be way more than 13%. And yeah, 526 00:37:48,732 --> 00:37:52,252 S1: only getting worse is my guess. Creators are automating their 527 00:37:52,252 --> 00:37:56,372 S1: entire content workflow with AI. So you spot trending topics 528 00:37:56,372 --> 00:37:59,692 S1: with AI agents, trained Claude on your style. Clone your 529 00:37:59,692 --> 00:38:04,092 S1: voice with 11 labs, create your digital avatar with hey Jen, 530 00:38:05,812 --> 00:38:12,892 S1: generate B-roll with Vo three and runway ML. Merge everything 531 00:38:12,892 --> 00:38:19,612 S1: in Canva. Automate distribution across platforms. Complete the workflow. So 532 00:38:20,532 --> 00:38:23,492 S1: that is basically the dream, right? This is like end 533 00:38:23,492 --> 00:38:25,892 S1: to end. They're just doing everything or at least saying 534 00:38:25,892 --> 00:38:30,372 S1: that it can be done. I love this. I think 535 00:38:31,532 --> 00:38:34,132 S1: all anyone should have to do, like as a creator, 536 00:38:34,132 --> 00:38:37,292 S1: is think about their thoughts and their ideas and like 537 00:38:37,332 --> 00:38:39,812 S1: how to get them out there and like the format 538 00:38:39,812 --> 00:38:41,852 S1: and the style that they want to get them out there. 539 00:38:41,852 --> 00:38:45,692 S1: But all the busy work that isn't fun, of course, 540 00:38:45,732 --> 00:38:48,732 S1: obviously still do the fun stuff. All that stuff should 541 00:38:48,732 --> 00:38:51,132 S1: just be automated away so you could focus on the 542 00:38:51,132 --> 00:38:58,292 S1: actual creativity part. So yeah, interesting. AI coding makes me 543 00:38:58,292 --> 00:39:01,692 S1: faster but kills my flow state. So this guy Praful 544 00:39:01,692 --> 00:39:07,972 S1: is talking about how. He used to be able to 545 00:39:08,012 --> 00:39:12,812 S1: just listen to music and code, and now it's like 546 00:39:12,852 --> 00:39:16,612 S1: you're dictating or you're typing or you're going back and 547 00:39:16,612 --> 00:39:20,572 S1: forth with this AI. Uh, for a lot of the 548 00:39:20,572 --> 00:39:23,932 S1: code generation. And it's just like, it's not the same 549 00:39:23,932 --> 00:39:27,892 S1: as just listening to music and coding. And I thought 550 00:39:27,892 --> 00:39:31,732 S1: that was really, really interesting. I hadn't realized it until 551 00:39:31,972 --> 00:39:36,262 S1: he said that. And it's definitely true for me. ChatGPT 552 00:39:36,342 --> 00:39:42,582 S1: usage drops hard when summer vacation starts. Yeah. This is 553 00:39:42,582 --> 00:39:44,862 S1: sad for me because it means they're not doing full 554 00:39:44,862 --> 00:39:48,862 S1: life integration with with the stuff they're mostly using it 555 00:39:48,862 --> 00:39:51,062 S1: for like homework and research and stuff like that, which 556 00:39:51,062 --> 00:39:54,382 S1: is fine, but I would like to see it more 557 00:39:54,382 --> 00:39:59,742 S1: widely adopted. Stop calling automation AI unless it actually learns something. 558 00:40:01,102 --> 00:40:05,302 S1: I included this just because I want to include counter 559 00:40:05,582 --> 00:40:12,902 S1: opinions to my own. Um, yeah. I think it's fine 560 00:40:12,902 --> 00:40:15,742 S1: that we want employees like it's obvious that we want 561 00:40:15,742 --> 00:40:19,462 S1: employees who can learn, but that's not the main thing 562 00:40:19,462 --> 00:40:22,422 S1: that we pay employees for to learn. We the main 563 00:40:22,422 --> 00:40:31,222 S1: thing we pay employees to do is to. Execute the tasks. Right. 564 00:40:31,262 --> 00:40:33,502 S1: So I think the learning and the executing of the 565 00:40:33,502 --> 00:40:36,262 S1: tasks are two separate things, because if we had AI 566 00:40:36,302 --> 00:40:40,542 S1: that could just execute the tasks, then the learning thing 567 00:40:40,542 --> 00:40:44,782 S1: would not be as important. Um, because what's more likely 568 00:40:44,782 --> 00:40:49,062 S1: to happen is that you would simply have, you know, 569 00:40:49,102 --> 00:40:53,342 S1: separate systems that are like, okay, we detected a sufficient 570 00:40:53,342 --> 00:40:57,742 S1: change in this thing. Therefore, you know, some retraining needs to, 571 00:40:57,902 --> 00:41:01,742 S1: you know, be added, whether that's a, you know, a 572 00:41:01,742 --> 00:41:07,182 S1: fine tuning of a model or whatever it is. And then, um. 573 00:41:09,542 --> 00:41:13,622 S1: Or just prompting or context or whatever it is, the 574 00:41:13,622 --> 00:41:16,182 S1: method you're going to use to do the improvement and 575 00:41:16,182 --> 00:41:18,422 S1: then you roll it back out. And the other thing is, 576 00:41:19,182 --> 00:41:23,862 S1: as the stuff gets better, I think. A lot of 577 00:41:23,902 --> 00:41:27,182 S1: that will generalize into being the same task and maybe 578 00:41:27,182 --> 00:41:33,782 S1: not even require too much learning. Again, just keep in 579 00:41:33,782 --> 00:41:39,982 S1: mind that just keep in mind how similar problems will 580 00:41:39,982 --> 00:41:44,542 S1: start to look when eyes are seeing all the problems 581 00:41:44,542 --> 00:41:49,942 S1: for all these different companies. How different really are the 582 00:41:49,942 --> 00:41:53,902 S1: sales use cases and the marketing use cases, and the 583 00:41:53,942 --> 00:42:01,302 S1: product management problems and the writing code problems. Right? It 584 00:42:01,342 --> 00:42:05,182 S1: seems to me like the hardest part is the creating of. 585 00:42:05,182 --> 00:42:09,942 S1: Net new things. It is the true creativity. It's the 586 00:42:09,942 --> 00:42:14,342 S1: true coming up with new ideas. And guess what? That's 587 00:42:14,342 --> 00:42:17,702 S1: not what we're paying most people for. We're paying them 588 00:42:17,702 --> 00:42:22,902 S1: to execute on these other tasks, which are the other 99.99% 589 00:42:23,142 --> 00:42:29,862 S1: of knowledge work. So I think this argument about, you know, 590 00:42:29,902 --> 00:42:34,032 S1: I'm paying you to learn. And A's can't learn on 591 00:42:34,032 --> 00:42:36,832 S1: the fly. Really good. Right now. Therefore, it's not a 592 00:42:36,832 --> 00:42:42,072 S1: threat to us. That is not a good argument. DSP 593 00:42:42,112 --> 00:42:48,392 S1: turns prompt engineering from art into systematic optimization. And this 594 00:42:48,392 --> 00:42:52,832 S1: is a towards data science article. Really cool. DSP is 595 00:42:52,832 --> 00:42:58,672 S1: a really powerful framework. It's basically automation of intelligence tasks 596 00:42:59,312 --> 00:43:03,912 S1: in pipeline type structure where it's repeatable and consistent and 597 00:43:03,912 --> 00:43:07,512 S1: you can depend on it and kind of like an 598 00:43:07,512 --> 00:43:13,952 S1: enterprisey kind of Nan type of vibe technology. Three mystery 599 00:43:13,952 --> 00:43:17,512 S1: customers account for over half of Nvidia's data center revenue. 600 00:43:18,232 --> 00:43:21,952 S1: That's spooky to me as like an investor in Nvidia, 601 00:43:21,952 --> 00:43:25,432 S1: which I currently don't hold any Nvidia, but if I 602 00:43:25,432 --> 00:43:30,392 S1: were holding some, that would be scary. Like, who are 603 00:43:30,392 --> 00:43:32,952 S1: those people and can they go away? And if so. 604 00:43:35,032 --> 00:43:38,592 S1: What would happen to their stock? It's more likely that 605 00:43:38,592 --> 00:43:41,592 S1: they'll gain five other unnamed ones and they'll go to 606 00:43:41,592 --> 00:43:44,911 S1: the moon. But I don't know. It just seems weird 607 00:43:44,912 --> 00:43:47,352 S1: to me if like so, such a high percentage of 608 00:43:47,392 --> 00:43:52,632 S1: like they're good numbers are from these unknown people. It's 609 00:43:52,632 --> 00:43:56,472 S1: like one of them Satan. Like, I don't know. I 610 00:43:56,512 --> 00:44:00,472 S1: just don't feel good about it. Venture capital in trouble. 611 00:44:00,512 --> 00:44:05,112 S1: Based on SEC filing patterns, TJ Jefferson discovered that formed 612 00:44:05,152 --> 00:44:11,552 S1: filings containing fund one peaked in Q3 2022 and have 613 00:44:11,552 --> 00:44:16,672 S1: been dropping hard ever since. Pretty cool analysis there. CrowdStrike 614 00:44:16,672 --> 00:44:22,232 S1: buys On'em for $290 million to build a genetic socks. 615 00:44:23,112 --> 00:44:28,112 S1: Computer science grads are struggling to find new jobs despite demand. 616 00:44:30,632 --> 00:44:34,512 S1: Entry level tech hiring drops 25%. Who knows where they're 617 00:44:34,512 --> 00:44:37,912 S1: getting that stat from? There's a million different of these studies. 618 00:44:38,592 --> 00:44:41,232 S1: I wouldn't be surprised if it was 25%. I wouldn't 619 00:44:41,232 --> 00:44:49,792 S1: be surprised if it was 65% drop. Yeah, like many 620 00:44:49,792 --> 00:44:55,272 S1: applying to places like Chipotle while software engineer postings remain down. Oh, 621 00:44:55,272 --> 00:45:02,352 S1: here we go. Software engineer postings remain down 70%. Cognitive 622 00:45:02,352 --> 00:45:08,911 S1: load is what actually matters in software complexity. Good piece there. Humans. 623 00:45:09,432 --> 00:45:13,552 S1: South Korea bans smartphones in schools starting 2026. Great job 624 00:45:13,872 --> 00:45:19,912 S1: and great job to probably Jonathan Haidt. Therapists are secretly 625 00:45:19,912 --> 00:45:24,792 S1: using ChatGPT and clients are upset about it. Middle aged 626 00:45:24,792 --> 00:45:27,802 S1: are no longer the most miserable. Turns out it's now 627 00:45:27,962 --> 00:45:32,002 S1: younger people. Younger people are now more miserable than middle aged. 628 00:45:32,042 --> 00:45:38,762 S1: It used to be a U-shape. With younger people happy 629 00:45:38,762 --> 00:45:44,202 S1: and older, older people happy, now younger people are massively unhappy. 630 00:45:45,882 --> 00:45:50,602 S1: US retail giants are raising prices based on Trump tariffs. 631 00:45:54,122 --> 00:45:58,602 S1: Shocking everyone. A brain abscess gave Eric Markowitz the clearest 632 00:45:58,602 --> 00:46:04,002 S1: moment of consciousness he's ever experienced. Describes how facing death 633 00:46:04,002 --> 00:46:07,842 S1: from a cerebral abscess made him truly conscious for the 634 00:46:07,842 --> 00:46:12,802 S1: first time. Trade school enrollment surges as Gen Z ditches 635 00:46:12,802 --> 00:46:19,202 S1: degrees for HVAC and welding. I got an idea here. 636 00:46:21,922 --> 00:46:25,162 S1: Why go to school right now? Why go to college? 637 00:46:25,562 --> 00:46:28,082 S1: I still think you should. So I'm of two minds 638 00:46:28,082 --> 00:46:33,122 S1: of this. I'm very strong. Like polar minds here. You 639 00:46:33,122 --> 00:46:36,042 S1: should still go because of connections. You should still go 640 00:46:36,042 --> 00:46:39,681 S1: because you do learn some decent fundamentals. If you weren't 641 00:46:39,682 --> 00:46:42,082 S1: to get. If you weren't going to get them some 642 00:46:42,082 --> 00:46:45,042 S1: other way. Right. It's a rigid structure. You have to 643 00:46:45,082 --> 00:46:47,681 S1: show up to class. You have to get the grades right. 644 00:46:47,682 --> 00:46:50,642 S1: Your parents are usually pushing you. So these are like 645 00:46:50,882 --> 00:46:56,362 S1: all good reasons to go. But holy crap, compared to 646 00:46:56,722 --> 00:47:01,442 S1: having your AI build you a custom curriculum of like 647 00:47:01,442 --> 00:47:05,282 S1: the best YouTube videos to teach you computer science. And 648 00:47:05,282 --> 00:47:09,322 S1: I'm going to be talking about this a whole bunch actually, 649 00:47:09,602 --> 00:47:12,002 S1: in the next few months. I've got this whole concept 650 00:47:12,002 --> 00:47:15,642 S1: of slack in the rope and actually what a real 651 00:47:15,642 --> 00:47:18,802 S1: curriculum could look like. I'm actually going to be building 652 00:47:18,802 --> 00:47:24,802 S1: curriculums into substrate curriculums for computer science and like all 653 00:47:24,922 --> 00:47:29,482 S1: these different, um, all these different degrees and stuff. And 654 00:47:29,482 --> 00:47:31,722 S1: I'm actually going to go find some experts on this 655 00:47:32,082 --> 00:47:36,082 S1: to help build the degrees, to help build the content 656 00:47:36,082 --> 00:47:38,162 S1: that needs to be in there. And I'm also going to, 657 00:47:38,202 --> 00:47:40,602 S1: of course, look at like all the classic ones, I'm 658 00:47:40,602 --> 00:47:42,522 S1: going to look at Oxford, I'm going to look at Stanford, 659 00:47:42,522 --> 00:47:46,082 S1: I'm going to look at like all these different schools, um, 660 00:47:46,082 --> 00:47:49,082 S1: including looking at what they used to teach, look at 661 00:47:49,082 --> 00:47:52,402 S1: what the Romans used to teach, like the trivium. Right. 662 00:47:52,682 --> 00:47:56,362 S1: And basically build like these custom bespoke things. But the 663 00:47:56,362 --> 00:48:00,002 S1: point is, I bet you it's pretty easy to put 664 00:48:00,002 --> 00:48:04,922 S1: together a set of YouTube videos, um, and maybe some 665 00:48:04,922 --> 00:48:08,482 S1: custom content that is like, just gets to the heart 666 00:48:08,482 --> 00:48:11,962 S1: of it, right? Imagine learning this stuff from like, Richard 667 00:48:11,962 --> 00:48:16,922 S1: Feynman and like the best teachers you could find. And 668 00:48:16,922 --> 00:48:20,042 S1: maybe certain teachers are better for certain types of learners, right? 669 00:48:20,082 --> 00:48:24,132 S1: And that could be stacked in there. But there's just 670 00:48:24,132 --> 00:48:29,532 S1: no possible way that learning computer science in college is 671 00:48:29,532 --> 00:48:39,412 S1: anywhere close, anywhere remotely close to like optimizing the cost function. 672 00:48:39,932 --> 00:48:44,052 S1: Like nowhere close to ideal. Uh, I know I use 673 00:48:44,412 --> 00:48:50,132 S1: I learned computer science in college in a traditional four 674 00:48:50,172 --> 00:48:54,932 S1: year college, and I've seen it taught in many, many others. 675 00:48:55,212 --> 00:48:57,812 S1: And I know what the curriculum looks like. Anyone could 676 00:48:57,812 --> 00:49:01,292 S1: go look it up. They're following the same books, like 677 00:49:01,292 --> 00:49:04,132 S1: most of them are reading the same books. Most of 678 00:49:04,132 --> 00:49:09,012 S1: the lecture styles are very similar. The homework is very similar. 679 00:49:09,012 --> 00:49:13,612 S1: Like the algorithm is just so ancient for doing this. And. 680 00:49:16,012 --> 00:49:18,652 S1: Right now, at this particular moment, we're like four years 681 00:49:18,652 --> 00:49:23,572 S1: from now is like 14.5 years from now In the 682 00:49:23,572 --> 00:49:25,892 S1: sense of like how much progress is going to be made. 683 00:49:26,012 --> 00:49:30,132 S1: What is your opportunity cost for sitting in a class 684 00:49:30,132 --> 00:49:34,372 S1: for a semester, trying to learn computer science, versus having 685 00:49:34,372 --> 00:49:38,132 S1: an AI build you a custom thing and actually have 686 00:49:38,132 --> 00:49:43,052 S1: conversations with you about it? And how about this? It's like, hey, 687 00:49:43,052 --> 00:49:45,732 S1: write me some code that does this, and you're writing 688 00:49:45,732 --> 00:49:47,812 S1: the code and it's like, yeah, that's not quite right. 689 00:49:47,812 --> 00:49:51,052 S1: It's more kind of like this. And it dynamically changes 690 00:49:51,052 --> 00:49:53,372 S1: your code for you and you're like, oh, okay, cool. 691 00:49:53,852 --> 00:49:55,692 S1: Or it's like, hey, I'm not going to help you. 692 00:49:55,692 --> 00:49:58,532 S1: You go ahead and do this and see if it compiles. Um, 693 00:49:58,572 --> 00:50:00,692 S1: go ahead and do this and, you know, test out 694 00:50:00,692 --> 00:50:03,492 S1: the website. Okay. Why is it not rendering? Why is 695 00:50:03,492 --> 00:50:07,052 S1: my thing off center? Why do my code not compile 696 00:50:07,052 --> 00:50:08,892 S1: or whatever? And it's like, I don't know, what do 697 00:50:08,892 --> 00:50:11,812 S1: you think it is like a Socratic sort of tutor 698 00:50:11,852 --> 00:50:15,092 S1: type thing, except for it never gets tired and you 699 00:50:15,092 --> 00:50:18,612 S1: don't pay extra money to ask the question again. Like, 700 00:50:18,612 --> 00:50:23,692 S1: how does that even compare To like this, this older 701 00:50:23,692 --> 00:50:28,452 S1: system that I just described, it's completely ridiculous. So there's 702 00:50:28,452 --> 00:50:30,692 S1: only a matter of time before this is going to 703 00:50:30,692 --> 00:50:33,412 S1: become open source. Like I'm going to make it open source. 704 00:50:33,412 --> 00:50:37,892 S1: I'm going to make these curriculums open source. And the 705 00:50:37,892 --> 00:50:40,212 S1: part that I don't have that I don't think I 706 00:50:40,212 --> 00:50:43,252 S1: can quickly build, and I'm sure other people will and 707 00:50:43,252 --> 00:50:47,132 S1: they'll do a great job is building the tutor that 708 00:50:47,132 --> 00:50:51,412 S1: can then use that curriculum. Right. Like I have Kai here, 709 00:50:51,412 --> 00:50:54,892 S1: but Kai's not specifically trained for being a tutor. Kai 710 00:50:54,932 --> 00:50:57,572 S1: is specifically trained right now because he lives inside of 711 00:50:57,572 --> 00:51:07,812 S1: cloud code to actually be a, um. Developer, right? Kai 712 00:51:07,812 --> 00:51:11,812 S1: is currently built because of his current body and brain structure, 713 00:51:12,972 --> 00:51:15,492 S1: to be an executor of tasks, to be a writer 714 00:51:15,492 --> 00:51:18,931 S1: of code, and that sort of thing. Um, but, you know, 715 00:51:18,972 --> 00:51:21,862 S1: I could reskin him as a tutor. In fact, I'm 716 00:51:21,902 --> 00:51:24,342 S1: getting excited about it. Just thinking about it. Like having 717 00:51:24,342 --> 00:51:27,662 S1: a tutor mode or a set of commands which are 718 00:51:27,662 --> 00:51:32,182 S1: tutor based. Okay, never mind, I'm excited. I'm actually writing 719 00:51:32,182 --> 00:51:38,982 S1: this down right now. I'm making some custom commands for tutoring. 720 00:51:39,182 --> 00:51:42,462 S1: I'm really curious about this right now, pairing that together 721 00:51:42,462 --> 00:51:51,262 S1: with substrate. Oh my goodness. Tutoring commands. And then open 722 00:51:51,262 --> 00:51:56,182 S1: sourcing the whole thing free on substrate and the tutoring 723 00:51:56,222 --> 00:51:59,102 S1: to go along with it. Question is how do you 724 00:51:59,102 --> 00:52:01,902 S1: make that packageable like as a mobile app? Because it 725 00:52:01,902 --> 00:52:03,902 S1: needs to go out to lots of people who need it. 726 00:52:04,302 --> 00:52:08,472 S1: Not like the 18 people or the even the 1800 727 00:52:08,472 --> 00:52:11,742 S1: or 18,000 people who see it on YouTube or whatever, 728 00:52:11,742 --> 00:52:15,382 S1: and they use it. That's barely making a dent. And 729 00:52:15,382 --> 00:52:17,902 S1: it's helping people who already maybe don't need the help 730 00:52:17,902 --> 00:52:24,142 S1: as much. Right. So anyway, I digress. Opportunity cost for college. 731 00:52:24,382 --> 00:52:28,822 S1: It's a tough question. I still think in general better 732 00:52:28,822 --> 00:52:33,982 S1: to have a degree. But right now, at this particular moment, 733 00:52:33,982 --> 00:52:36,502 S1: I am really worried about telling somebody to go and 734 00:52:36,502 --> 00:52:39,702 S1: get a four year degree and basically shutting down their 735 00:52:39,702 --> 00:52:43,222 S1: mind the whole time. What are you doing when you're 736 00:52:43,222 --> 00:52:47,422 S1: in four years of college? You're doing the coursework. You're 737 00:52:47,422 --> 00:52:50,942 S1: trying to please that teacher. You're trying to please that curriculum. 738 00:52:50,942 --> 00:52:55,022 S1: You're trying to please that old structure of college courses 739 00:52:55,022 --> 00:53:00,942 S1: from whenever this structure started. Right. And that is an ancient, 740 00:53:01,422 --> 00:53:06,142 S1: broken structure, extremely inefficient. I'm going to give it a 3% 741 00:53:06,142 --> 00:53:11,422 S1: efficiency score compared to AI tutoring you based on like 742 00:53:11,462 --> 00:53:16,142 S1: custom match content. Custom match YouTube content? Are you kidding me? 743 00:53:16,422 --> 00:53:20,262 S1: Like no comparison? Person. I think we could get like 744 00:53:20,302 --> 00:53:23,862 S1: a less ideal is 100. I think we can get 745 00:53:23,902 --> 00:53:31,022 S1: like a 60% efficiency up from like, okay, three is 746 00:53:31,022 --> 00:53:34,102 S1: too low, up from like a 10%. I think we 747 00:53:34,102 --> 00:53:41,422 S1: can get what is that, 5X5X better. And like how 748 00:53:41,422 --> 00:53:44,342 S1: does that turn into reality. I don't know. I'm talking 749 00:53:44,342 --> 00:53:48,262 S1: about full degrees full understanding. Here's what I'm talking about. 750 00:53:48,702 --> 00:53:52,462 S1: Having a better grasp of computer science a better grasp 751 00:53:52,462 --> 00:54:00,062 S1: of history a better grasp of psychology, evolutionary biology, political science. 752 00:54:00,382 --> 00:54:04,622 S1: Getting a grasp of those things in one week's time 753 00:54:06,422 --> 00:54:11,022 S1: that is better than 99.99% of all people on the 754 00:54:11,022 --> 00:54:17,502 S1: planet in one week's time. You can do that with 755 00:54:17,502 --> 00:54:23,182 S1: 20 really good YouTube videos combined with active tutoring by 756 00:54:23,182 --> 00:54:28,342 S1: your AI in a Socratic, challenging way to quiz you 757 00:54:28,342 --> 00:54:30,022 S1: to make sure you're able to move on to the 758 00:54:30,022 --> 00:54:35,022 S1: next video. Think about that. That's not even counting. We're 759 00:54:35,022 --> 00:54:39,342 S1: going to make you custom videos, right? Which is not 760 00:54:39,342 --> 00:54:43,302 S1: here yet. And still coming. Right. That that's next level. 761 00:54:43,662 --> 00:54:47,822 S1: But I think we can get massive multiples of gains 762 00:54:48,542 --> 00:54:53,102 S1: just from using the tech we already have. So like 763 00:54:53,102 --> 00:54:57,862 S1: I said violently of two minds on this whole college thing. Yes. 764 00:54:57,862 --> 00:55:01,142 S1: You should go. Holy crap. Don't turn off your brain. 765 00:55:01,142 --> 00:55:03,982 S1: You need to be doing all this AI stuff, all 766 00:55:03,982 --> 00:55:06,942 S1: this modern shit at the same time. If you're doing 767 00:55:06,942 --> 00:55:08,902 S1: it to get the degree, go do it. Get the 768 00:55:08,902 --> 00:55:15,502 S1: degree right. Especially if you're almost done. But I don't 769 00:55:15,502 --> 00:55:18,192 S1: think the people being laid off are like, nope, we 770 00:55:18,192 --> 00:55:20,672 S1: can't lay that person off. They have a four year degree. 771 00:55:20,672 --> 00:55:25,392 S1: I don't think anybody fucking cares right now, honestly. Um, 772 00:55:26,232 --> 00:55:28,712 S1: some people do, okay? Some people do. If you're trying 773 00:55:28,712 --> 00:55:31,232 S1: to go into some traditional field, I'm just telling you, 774 00:55:31,232 --> 00:55:36,392 S1: the world might be very different in four years. Like, 775 00:55:36,392 --> 00:55:42,952 S1: dramatically fucking different. So really, really think about opportunity cost. 776 00:55:42,992 --> 00:55:45,392 S1: Best of both worlds is do all the shit that 777 00:55:45,392 --> 00:55:48,792 S1: I'm talking about at the same time. Then you don't 778 00:55:48,792 --> 00:55:52,272 S1: really take a hit by getting the degree right. Plus, 779 00:55:52,512 --> 00:55:55,632 S1: ideally you're using AI to help you get your degree right. 780 00:55:55,712 --> 00:56:00,192 S1: So if it's easy, it's easy. Plus it's friendships. It's 781 00:56:00,232 --> 00:56:04,672 S1: maybe sports, it's extracurricular activities. It's social. You're making a 782 00:56:04,672 --> 00:56:07,912 S1: friend network. There's tons of benefits to college that aren't 783 00:56:07,912 --> 00:56:11,832 S1: related to the curriculum. I'm mostly criticizing the curriculum stuff. 784 00:56:13,832 --> 00:56:19,272 S1: Man can't get through these podcasts just talking. Everyone loves 785 00:56:19,272 --> 00:56:21,552 S1: it when I do this though, and go into more 786 00:56:21,552 --> 00:56:24,792 S1: detail and talk about things, but it does take a 787 00:56:24,792 --> 00:56:27,992 S1: lot of time. Um, I got to figure out how 788 00:56:27,992 --> 00:56:30,512 S1: to get it down into my old format of just like, 789 00:56:30,512 --> 00:56:33,872 S1: read the title and boom, go through, be done in 790 00:56:33,912 --> 00:56:36,272 S1: 15 minutes. Didn't I say I'm supposed to be done 791 00:56:36,272 --> 00:56:38,292 S1: in 15 minutes? Isn't it supposed to be a 15 792 00:56:38,292 --> 00:56:42,672 S1: minute podcast? Haven't had one of those in two years. 793 00:56:43,472 --> 00:56:48,512 S1: All right. Brain abscess already did that one. The feeling 794 00:56:48,512 --> 00:56:52,192 S1: your work creates matters more than the checkboxes. This is 795 00:56:52,192 --> 00:56:56,112 S1: by Mitchell Hashimoto, who also made the ghosty terminal, which 796 00:56:56,112 --> 00:56:59,912 S1: is my go to terminal. A really good article there. 797 00:57:00,352 --> 00:57:05,312 S1: Physical technical books create memory palaces that websites can't match. 798 00:57:05,352 --> 00:57:08,112 S1: I was just hanging out with my buddy Tim Leonard, 799 00:57:08,632 --> 00:57:13,912 S1: who is a I met through UL and, um, Hanging 800 00:57:13,912 --> 00:57:18,352 S1: out in Vegas, and he opens up his his logbook 801 00:57:18,352 --> 00:57:23,352 S1: that he keeps with him. And he's got like these custom, 802 00:57:23,392 --> 00:57:27,272 S1: like hand-drawn images. So, like, he takes a note or 803 00:57:27,272 --> 00:57:30,032 S1: whatever talks about a plant or a, or a thing 804 00:57:30,032 --> 00:57:31,912 S1: that he saw with his wife or whatever. He takes 805 00:57:31,912 --> 00:57:36,832 S1: these notes. He had like little pressed flowers, like he 806 00:57:36,832 --> 00:57:40,872 S1: pulled from a place. It was just brilliant, just absolutely 807 00:57:40,872 --> 00:57:45,712 S1: brilliant analog. He's got a pen, he's got paper memories. 808 00:57:47,152 --> 00:57:49,192 S1: I think you can have that with technical books as well. 809 00:57:49,192 --> 00:57:51,712 S1: And that's what this article is arguing about. It was 810 00:57:51,752 --> 00:57:59,312 S1: it was really cool. MIT and Scripps researchers develop one 811 00:57:59,312 --> 00:58:03,792 S1: shot vaccines for HIV and Covid. Really, really cool. All right. 812 00:58:03,792 --> 00:58:09,232 S1: Discovery I'm going to try not to talk for 74 813 00:58:09,232 --> 00:58:13,322 S1: minutes about each one of these. I'm going to blast through. 814 00:58:13,362 --> 00:58:18,882 S1: I'm going to try. AI continues computing's evolution towards understanding humans. 815 00:58:19,362 --> 00:58:23,362 S1: A beginner friendly jiu jitsu tutorial for non git users. 816 00:58:24,842 --> 00:58:27,722 S1: This is an alternative to git. It's called jiu jitsu. 817 00:58:27,762 --> 00:58:30,242 S1: It's supposed to be awesome, but I do not have 818 00:58:30,242 --> 00:58:34,362 S1: the time for the switching cost right now. The opportunity 819 00:58:34,362 --> 00:58:39,122 S1: cost of me trying to learn jiu jitsu right now. Um. 820 00:58:41,282 --> 00:58:44,962 S1: And migrate everything. Get off of GitHub. Does GitHub even 821 00:58:45,002 --> 00:58:49,202 S1: do jiu jitsu? I don't know. I'm guessing not. Git 822 00:58:49,242 --> 00:58:54,842 S1: is in the name. FFmpeg pages simplifies video processing with 823 00:58:54,842 --> 00:58:59,242 S1: visual commands. AI agents should use files and git instead 824 00:58:59,242 --> 00:59:06,282 S1: of context windows. Sounds pretty familiar to me with like 825 00:59:06,282 --> 00:59:10,682 S1: this whole UFC thing I just came up with. Um, 826 00:59:11,322 --> 00:59:14,602 S1: That's one I actually didn't read yet. I need to 827 00:59:14,602 --> 00:59:19,722 S1: open this tab. Because it's in the discovery section. So 828 00:59:19,722 --> 00:59:25,082 S1: I'm just kind of like sometimes kicking off cool links. Um, 829 00:59:25,482 --> 00:59:28,442 S1: that you should go check out. Single servers can handle 830 00:59:28,442 --> 00:59:32,362 S1: millions of HTTP requests per session. Dot files don't belong 831 00:59:32,362 --> 00:59:36,882 S1: in application support. Loved that one. That was so spot on. 832 00:59:38,682 --> 00:59:46,322 S1: Google lets you block AI overviews from your site. Nardin 833 00:59:46,362 --> 00:59:52,442 S1: automates bug bounty enumeration via discord, training yourself to like 834 00:59:52,442 --> 00:59:56,642 S1: things you hate. Klein runs completely offline with LM studio. 835 00:59:57,202 --> 01:00:00,122 S1: New vibe coded web app from Kathy Helps. I don't 836 01:00:00,122 --> 01:00:04,802 S1: see many of my women friends doing the AI coding entrepreneurship, 837 01:00:04,802 --> 01:00:07,002 S1: so I'm posting this here. This is a woman who 838 01:00:07,002 --> 01:00:09,802 S1: just built an app and is having super fun is 839 01:00:09,802 --> 01:00:13,722 S1: talking about it, and she's just really enjoying the whole 840 01:00:13,722 --> 01:00:16,322 S1: development process. And I just want to see a lot 841 01:00:16,322 --> 01:00:21,362 S1: more of my women friends, uh, doing this whole entrepreneurship, 842 01:00:21,402 --> 01:00:25,202 S1: like build your own business, you know, AI coding, AI 843 01:00:25,242 --> 01:00:27,682 S1: hacking thing. So I'm going to be sending this to 844 01:00:27,722 --> 01:00:31,362 S1: a bunch of my friends. Trace prompt provides tamper proof 845 01:00:31,362 --> 01:00:38,522 S1: audit trails for LLM applications. Oh, a duplicate got a 846 01:00:38,522 --> 01:00:44,122 S1: duplicate there. Claude gets a chrome extension. Physical textbooks become 847 01:00:44,162 --> 01:00:47,442 S1: unchanging memory places talked about that one. Wire prompts don't 848 01:00:47,442 --> 01:00:53,202 S1: belong in git. Uh so oops oops on that one 849 01:00:54,642 --> 01:01:00,442 S1: guys or shares creative bash prompt designs and K-pop demon 850 01:01:00,442 --> 01:01:05,322 S1: Hunters is Netflix's most watched video ever. And I still 851 01:01:05,322 --> 01:01:07,602 S1: have not watched this, but I do want to watch it. 852 01:01:08,042 --> 01:01:10,852 S1: Recommendation of the week. Start thinking about what you would 853 01:01:10,852 --> 01:01:15,932 S1: do if you had a personal assistant and what you 854 01:01:15,932 --> 01:01:20,372 S1: would give them. Scheduling, planning, collecting information, finding the best books. 855 01:01:20,412 --> 01:01:23,532 S1: Keeping people aware of what you're working on, keeping your 856 01:01:23,532 --> 01:01:29,332 S1: family synced up. Finding materials for your hobbies. Teaching yourself topics. 857 01:01:29,532 --> 01:01:32,892 S1: Start building that list because you will soon have this. 858 01:01:33,572 --> 01:01:36,732 S1: Whether it's through, like what I'm doing or a more 859 01:01:36,732 --> 01:01:40,572 S1: consumer based approach that's going to come out soon. Like, 860 01:01:40,612 --> 01:01:42,932 S1: you should start thinking about this. You want to be 861 01:01:42,932 --> 01:01:45,972 S1: able to magnify yourself in this way. Um, and I 862 01:01:45,972 --> 01:01:48,812 S1: recommend you go check out the the personal AI video 863 01:01:48,812 --> 01:01:52,932 S1: that I just put out talking about how I'm doing it. And, um, 864 01:01:54,652 --> 01:01:58,452 S1: the aphorism of the week argue for your limitations, and 865 01:01:58,452 --> 01:02:02,692 S1: sure enough, they are yours. Argue for your limitations, and 866 01:02:02,692 --> 01:02:06,052 S1: sure enough, they're yours. Richard Bach.