1 00:00:01,129 --> 00:00:04,460 S1: Welcome to Unsupervised Learning, a security, AI and meaning focused 2 00:00:04,460 --> 00:00:07,310 S1: podcast that looks at how best to thrive as humans 3 00:00:07,310 --> 00:00:11,540 S1: in a post AI world. It combines original ideas, analysis, 4 00:00:11,570 --> 00:00:14,780 S1: and mental models to bring not just the news, but 5 00:00:14,780 --> 00:00:22,189 S1: why it matters and how to respond. All right, welcome 6 00:00:22,190 --> 00:00:26,720 S1: to unsupervised learning. This is Daniel. Okay. The first AI breaches. 7 00:00:26,720 --> 00:00:30,380 S1: What do we have here? Oh. Saw the comet last night. Yeah. 8 00:00:30,380 --> 00:00:32,120 S1: This is really cool. It's been out for a little while. 9 00:00:32,120 --> 00:00:35,059 S1: It's actually going to be out for a few more days, 10 00:00:35,060 --> 00:00:38,390 S1: and I think it's getting a little less bright, but. 11 00:00:38,390 --> 00:00:40,820 S1: And coming out a little bit later each day, I 12 00:00:40,820 --> 00:00:43,730 S1: think maybe like 10 or 15 minutes later each day. 13 00:00:43,880 --> 00:00:49,910 S1: But my girl took this very, very cool, very, very cool. Uh, 14 00:00:49,909 --> 00:00:53,870 S1: she actually spotted it with her, um, naked eyes at first. 15 00:00:53,900 --> 00:00:57,080 S1: She was the first one to see it, and then 16 00:00:57,080 --> 00:01:02,029 S1: we all started pointing cameras and binoculars at it, but 17 00:01:02,030 --> 00:01:06,709 S1: she actually captured this with, um, her iPhone 16 Pro, 18 00:01:06,709 --> 00:01:10,370 S1: so not bad at all. Yeah, basically, the iPhone catches 19 00:01:10,400 --> 00:01:13,819 S1: a lot more than the naked eye, and she used 20 00:01:13,819 --> 00:01:18,619 S1: light or night mode and also long exposure. And yeah, 21 00:01:18,650 --> 00:01:22,910 S1: it looks pretty good. Uh, it's an 80 000 year 22 00:01:23,240 --> 00:01:28,970 S1: orbit time for this particular comet. And, uh, the head there, 23 00:01:28,970 --> 00:01:34,039 S1: which was quite bright, is about two miles across, and 24 00:01:34,040 --> 00:01:38,390 S1: the tail is something like 18 million miles long. And 25 00:01:38,390 --> 00:01:42,020 S1: for reference, it's 8 million miles to the sun. So it's. 26 00:01:42,050 --> 00:01:45,590 S1: What is that? Almost three times two and a half times. 27 00:01:45,590 --> 00:01:48,290 S1: The tail is two and a half times the distance 28 00:01:48,290 --> 00:01:50,780 S1: from the Earth to the sun. And the head is 29 00:01:50,780 --> 00:01:54,380 S1: two miles wide. And it won't be back for another 30 00:01:54,380 --> 00:01:57,320 S1: 80,000 years. She's like, I guess we won't see it again. 31 00:01:57,320 --> 00:02:00,280 S1: I'm like, I don't know. Eat your greens. Good exercise 32 00:02:00,280 --> 00:02:05,170 S1: every day. You never know how long you can live. Okay, so, uh, yeah. 33 00:02:05,200 --> 00:02:08,559 S1: Router tutorial on how to use any hugging face model 34 00:02:08,560 --> 00:02:11,050 S1: with a llama. So I want to open this up 35 00:02:11,050 --> 00:02:15,880 S1: real quick. It's a super quick tutorial. I'm actually I 36 00:02:15,880 --> 00:02:18,850 S1: think I already recorded the video as well. Yeah, I 37 00:02:18,850 --> 00:02:20,920 S1: think I already recorded the video as well, so it 38 00:02:20,919 --> 00:02:23,830 S1: should be a video soon. But look at this. It's 39 00:02:23,830 --> 00:02:27,040 S1: super simple. You go to the models page on hugging Face. 40 00:02:27,040 --> 00:02:30,190 S1: You basically get the URL. You click on this right here, 41 00:02:30,190 --> 00:02:33,100 S1: files and versions, which you can see here. You get 42 00:02:33,130 --> 00:02:36,339 S1: one of these files like I got the smaller one, 43 00:02:36,340 --> 00:02:40,929 S1: the four gig one. You edit a model file and 44 00:02:40,930 --> 00:02:42,670 S1: this is all you have to put in there. Look 45 00:02:42,700 --> 00:02:47,859 S1: at this. It's insane. So you basically do um, from 46 00:02:47,889 --> 00:02:51,730 S1: is all you need actually is from. Okay. And then 47 00:02:51,730 --> 00:02:54,820 S1: you just put the path to the file and then 48 00:02:54,820 --> 00:02:57,940 S1: if you want to add a system prompt. You can. 49 00:02:57,970 --> 00:03:01,360 S1: So that's that's pretty much it. And you don't need 50 00:03:01,360 --> 00:03:04,390 S1: a system prompt okay. You can just use the user 51 00:03:04,389 --> 00:03:07,930 S1: prompt which is what you send into the model. But 52 00:03:08,169 --> 00:03:11,530 S1: then you do this this command here. So so again 53 00:03:11,560 --> 00:03:14,739 S1: we're trying to get a llama to run all these models. 54 00:03:14,740 --> 00:03:18,250 S1: So once you have this model file it's saved. You 55 00:03:18,250 --> 00:03:21,250 S1: just do this little command right here and then this 56 00:03:21,250 --> 00:03:24,430 S1: this name that you give it. Lexi writer or Lexi, 57 00:03:24,430 --> 00:03:27,429 S1: writer or writer or whatever you want to call it. 58 00:03:27,669 --> 00:03:31,450 S1: You just give it a llama create writer. Okay, so 59 00:03:31,450 --> 00:03:33,850 S1: you just made its own model. This takes like one 60 00:03:33,850 --> 00:03:36,700 S1: second to complete this little piece right here. And then 61 00:03:36,700 --> 00:03:39,670 S1: you run llama run in the name of the model 62 00:03:39,670 --> 00:03:43,780 S1: you created. Llama run. Writer. Llama run. Lexi. Writer, whatever 63 00:03:43,780 --> 00:03:46,660 S1: it is. And now you're actually talking to that model, 64 00:03:46,660 --> 00:03:49,480 S1: that hugging face model that you just downloaded. So this 65 00:03:49,480 --> 00:03:53,530 S1: entire process might be like three minutes. Five minutes, right? 66 00:03:53,560 --> 00:03:57,760 S1: Depends how long it takes you to download the the file. 67 00:03:57,760 --> 00:04:01,300 S1: So you just went from having like a couple of 68 00:04:01,300 --> 00:04:06,040 S1: dozen models available inside of a llama to having tens 69 00:04:06,040 --> 00:04:09,970 S1: of thousands or hundreds of thousands. So really, really cool. 70 00:04:09,970 --> 00:04:15,910 S1: And my buddy Marcus Hutchins and I disagree about Elon, 71 00:04:16,120 --> 00:04:19,000 S1: about a couple of things about his politics. Um, well, 72 00:04:19,029 --> 00:04:22,060 S1: we we agree on his politics. I just think Elon's 73 00:04:22,060 --> 00:04:23,740 S1: going to come back. I think he's still a liberal. 74 00:04:23,740 --> 00:04:27,100 S1: I'm hoping he's still a liberal. Um, I would say 75 00:04:27,100 --> 00:04:31,330 S1: I'm hoping he's liberal. Not a liberal. Not a current liberal. Um, 76 00:04:31,330 --> 00:04:35,349 S1: but I'm hoping he's more balanced. I believe he's more balanced. 77 00:04:35,350 --> 00:04:39,340 S1: And he's going to return. Marcus thinks he is Sith. 78 00:04:39,339 --> 00:04:43,210 S1: He is turned Sith. He is straight, far right, and 79 00:04:43,210 --> 00:04:47,020 S1: he's not coming back. And I want to say, I 80 00:04:47,020 --> 00:04:49,659 S1: do agree that that's possible and I'm worried about it, 81 00:04:49,660 --> 00:04:51,909 S1: but I'm hoping it's not the case. And I think 82 00:04:51,910 --> 00:04:54,570 S1: it's not the case I think after a bunch of 83 00:04:54,570 --> 00:04:57,419 S1: this craziness blows over in the next year or two, 84 00:04:57,450 --> 00:05:02,429 S1: that he will start showing his his more liberal and 85 00:05:02,430 --> 00:05:05,640 S1: balanced side, that that's my hope. He also believes that 86 00:05:05,670 --> 00:05:08,670 S1: Ellen is basically I don't want to put too many 87 00:05:08,670 --> 00:05:11,159 S1: words in his mouth, but I think he believes that 88 00:05:11,160 --> 00:05:14,760 S1: he's not a true innovator. He got lucky. A lot 89 00:05:14,760 --> 00:05:18,630 S1: of his stuff is garbage. Um, it's just like there's 90 00:05:18,660 --> 00:05:20,730 S1: there's no way any of this is going to go 91 00:05:20,760 --> 00:05:23,820 S1: well for him. Tesla is going to be basically bankrupt 92 00:05:23,820 --> 00:05:26,160 S1: within a year or two, I believe, he said. And 93 00:05:26,160 --> 00:05:27,930 S1: I've got a link to the whole conversation we had. 94 00:05:27,960 --> 00:05:30,300 S1: And we were we were texting on the side as 95 00:05:30,300 --> 00:05:34,200 S1: well during this. Um, and then basically I'm like, hey, look, 96 00:05:34,500 --> 00:05:38,219 S1: should we bet on this? Um, and so he offered 97 00:05:38,220 --> 00:05:41,549 S1: a couple of bets and I said, I need these 98 00:05:41,550 --> 00:05:45,300 S1: to be measurable. So I offered three measurable bets. One, 99 00:05:45,300 --> 00:05:48,750 S1: the Tesla stock would hit 250. Um, you know, it 100 00:05:48,750 --> 00:05:51,980 S1: would hit that point by June 30th. So middle of 101 00:05:51,980 --> 00:05:55,280 S1: next year it would hit 250. End of next year 102 00:05:55,279 --> 00:05:58,910 S1: it will have hit $300 at least once. It would 103 00:05:58,940 --> 00:06:01,310 S1: have got to that point. Um, and I actually think 104 00:06:01,310 --> 00:06:03,890 S1: it's going to be higher, but I just wanted to 105 00:06:03,890 --> 00:06:07,640 S1: have it 250 and 300. And then that Elon would, 106 00:06:07,670 --> 00:06:12,169 S1: in a liberal way, in a principled way that defends 107 00:06:12,170 --> 00:06:18,470 S1: something around anti-authoritarianism, something freedom oriented. In other words, Trump 108 00:06:18,470 --> 00:06:20,750 S1: would be on the wrong side of this and Elon 109 00:06:20,750 --> 00:06:22,969 S1: would be on the right side of it, the correct 110 00:06:22,970 --> 00:06:26,360 S1: side of it. And I argued that that would happen 111 00:06:26,360 --> 00:06:29,060 S1: by the end of next year. Okay. So that's three 112 00:06:29,060 --> 00:06:34,669 S1: different bets, each for $1,000. And uh, he accepted. So 113 00:06:34,850 --> 00:06:37,820 S1: we got three bets out there. Kind of reminds me 114 00:06:37,820 --> 00:06:40,460 S1: of this thing that I've been obsessed with lately, which 115 00:06:40,460 --> 00:06:44,780 S1: are these prediction markets? Which one is manifold? I forget 116 00:06:44,779 --> 00:06:47,990 S1: the name of the other one starts with a P, but, um. Yeah, 117 00:06:48,020 --> 00:06:52,070 S1: pretty excited about these. I've barely started messing around with them, 118 00:06:52,070 --> 00:06:54,589 S1: but I feel like I want to use them more. Okay, 119 00:06:54,620 --> 00:06:57,590 S1: I did a talk for a UN group called Ypo, 120 00:06:57,620 --> 00:07:01,850 S1: which is an intellectual property, and it went really well. 121 00:07:01,850 --> 00:07:05,990 S1: And I want to thank Olivia Fabrizi, I believe, is 122 00:07:05,990 --> 00:07:08,690 S1: how you pronounce her last name for being not just 123 00:07:08,690 --> 00:07:11,390 S1: a great host, but actually, really, she has some really 124 00:07:11,390 --> 00:07:14,420 S1: great questions. She seems very interested in all this AI stuff, 125 00:07:14,420 --> 00:07:18,770 S1: very knowledgeable, and she was kind enough to host us 126 00:07:18,770 --> 00:07:22,489 S1: in this. And yeah, it was great. It was a 127 00:07:22,490 --> 00:07:25,820 S1: great conversation. I listened back to the talk, which which 128 00:07:25,820 --> 00:07:28,489 S1: I hadn't done before, and it seemed like it went 129 00:07:28,490 --> 00:07:30,739 S1: pretty well and good questions at the end. So I 130 00:07:30,740 --> 00:07:33,470 S1: encourage you to check that out. Thanks to Nudge Security 131 00:07:33,500 --> 00:07:38,060 S1: for sponsoring, and let's get into security. Okay, so an 132 00:07:38,060 --> 00:07:43,400 S1: attacker accessed moi. I don't know how you pronounce that. Moi, moi. 133 00:07:43,430 --> 00:07:46,190 S1: I feel like you could have done a better job 134 00:07:46,190 --> 00:07:51,850 S1: with the name, but, um, AI chatbot database. Okay, so 135 00:07:51,850 --> 00:07:53,890 S1: this is I've been talking about this for a long time. 136 00:07:53,920 --> 00:07:56,350 S1: A lot of people who are anti-ai have been talking 137 00:07:56,350 --> 00:07:59,830 S1: about this, but I am obviously pro AI, but I 138 00:08:00,070 --> 00:08:03,370 S1: absolutely believe this is going to happen. It already has 139 00:08:03,370 --> 00:08:06,880 S1: happened and it's going to get worse. This AI chatbot 140 00:08:06,910 --> 00:08:11,650 S1: database is full of like basically AI girlfriends is is 141 00:08:11,680 --> 00:08:15,190 S1: the vibe that I'm getting. In fact, let's let's go 142 00:08:15,220 --> 00:08:16,840 S1: pull it up. What are we doing? This is why 143 00:08:16,840 --> 00:08:22,210 S1: we have video. All right. Select your login. Don't think so. Okay. Yeah. 144 00:08:22,240 --> 00:08:26,740 S1: Notice any, um, notice any trends here? How many guys 145 00:08:26,740 --> 00:08:28,780 S1: do we have here? Is this a guy? I think 146 00:08:28,780 --> 00:08:31,870 S1: the bottom left. We got a guy. Top right or 147 00:08:31,870 --> 00:08:34,330 S1: top left? We have a guy. Uh, not sure on 148 00:08:34,330 --> 00:08:37,660 S1: this one. And then the rest are all girls looking 149 00:08:37,660 --> 00:08:41,950 S1: a little young. Honestly, uh, if I'm being frank here, 150 00:08:42,160 --> 00:08:46,579 S1: but anyway. Um. Avatars. Okay, so basically, it's its avatars, 151 00:08:46,580 --> 00:08:51,110 S1: its companions. This whole scene is getting really big in 152 00:08:51,110 --> 00:08:55,489 S1: AI for lots of reasons. There's companionship, but obviously, I mean, 153 00:08:55,520 --> 00:09:00,860 S1: this whole girlfriend sex like fantasy angle is is so 154 00:09:00,860 --> 00:09:03,560 S1: obvious and it's so obvious that that's where a lot 155 00:09:03,559 --> 00:09:05,930 S1: of attention was going to go, both from the user 156 00:09:05,929 --> 00:09:10,040 S1: side but also on the builder side. And so this company, 157 00:09:10,040 --> 00:09:14,240 S1: their database got breached sensitive user interactions. It had the 158 00:09:14,240 --> 00:09:17,990 S1: chat history okay. So the user sign up you look 159 00:09:18,020 --> 00:09:22,189 S1: look at this. You sign up with you look look 160 00:09:22,220 --> 00:09:27,050 S1: it's got federated logins okay. These are your actual user 161 00:09:27,050 --> 00:09:29,840 S1: accounts okay. So people sign up with their user accounts. 162 00:09:29,870 --> 00:09:31,670 S1: Like if I click on this okay I'm going to 163 00:09:31,670 --> 00:09:34,910 S1: click on it I'm signing up with my real stuff okay. 164 00:09:34,940 --> 00:09:36,679 S1: If I sign up with this. And then I have 165 00:09:36,679 --> 00:09:40,490 S1: this chat with this, this avatar, this AI over there 166 00:09:40,490 --> 00:09:44,179 S1: and whatever I say on there. And now that's what 167 00:09:44,210 --> 00:09:49,500 S1: gets compromised. Okay, so including sexual fantasies and the user 168 00:09:49,500 --> 00:09:52,470 S1: accounts were linked to people's personal addresses. I mean, it's 169 00:09:52,470 --> 00:09:54,540 S1: so obvious that this was going to happen, and it 170 00:09:54,540 --> 00:09:56,939 S1: is now happening. It's going to be it's going to 171 00:09:56,940 --> 00:10:00,959 S1: be nasty, right? So input validation issues, prompt injection, all 172 00:10:00,990 --> 00:10:03,030 S1: of that. But the vast majority of damage this is 173 00:10:03,030 --> 00:10:06,030 S1: what I'm saying afterwards will come from customers giving their 174 00:10:06,030 --> 00:10:10,439 S1: souls essentially to small startups in the AI assistant AI 175 00:10:10,470 --> 00:10:13,620 S1: girlfriend space. This doesn't have to be like a sexual 176 00:10:13,620 --> 00:10:16,770 S1: fantasy thing. It could be trauma. It could be somebody 177 00:10:16,770 --> 00:10:19,800 S1: using it as there's other sites just like this, but 178 00:10:19,800 --> 00:10:22,500 S1: it's more therapy oriented. So it's like you have a 179 00:10:22,500 --> 00:10:24,900 S1: best friend to talk to you and you're talking about 180 00:10:24,900 --> 00:10:27,840 S1: your favorite show and blah, blah, blah, and like, you 181 00:10:27,840 --> 00:10:30,330 S1: develop this relationship with it, but you might have been 182 00:10:30,330 --> 00:10:34,260 S1: saying some really nasty stuff about your parents or about whatever, 183 00:10:34,350 --> 00:10:37,530 S1: maybe about your friends, right? Maybe you're really sad and 184 00:10:37,530 --> 00:10:40,050 S1: you think your friend stabbed you in the back or whatever. 185 00:10:40,080 --> 00:10:42,959 S1: So you're talking all this crap about them and that 186 00:10:42,990 --> 00:10:45,719 S1: gets released. I mean, we're talking about intimacy here. It 187 00:10:45,720 --> 00:10:47,969 S1: doesn't have to be of a sexual, you know, sort 188 00:10:47,970 --> 00:10:52,110 S1: of flavor. Right? So we're talking about the the closest 189 00:10:52,110 --> 00:10:54,959 S1: form of intimacy, which is what you have with like 190 00:10:54,990 --> 00:10:57,780 S1: this friend, you you can tell anything and you can't 191 00:10:57,780 --> 00:11:00,059 S1: tell anyone else. A lot of people, they don't have 192 00:11:00,059 --> 00:11:02,880 S1: any friends. So this is like their only friend. So 193 00:11:02,880 --> 00:11:08,100 S1: you give it your trauma, your political opinions. Yeah, fantasies, whatever. 194 00:11:08,100 --> 00:11:11,699 S1: And they totally get you. But this, it might be 195 00:11:11,700 --> 00:11:15,030 S1: like a nine person startup. They have no security, right? 196 00:11:15,059 --> 00:11:19,109 S1: I mean, startups get hacked so easily. It's so easy 197 00:11:19,110 --> 00:11:21,630 S1: to hack one of these little startups. They have no security, 198 00:11:21,630 --> 00:11:24,300 S1: they don't know how to do security. And once they do, 199 00:11:24,330 --> 00:11:27,540 S1: I mean, think about the extortion campaigns here, right? Assuming 200 00:11:27,540 --> 00:11:29,760 S1: the people have money, which a lot of some of 201 00:11:29,760 --> 00:11:32,040 S1: them will and some of them won't. But but I mean, 202 00:11:32,070 --> 00:11:34,020 S1: you could just be like, hey, here's a clip from 203 00:11:34,020 --> 00:11:37,140 S1: this thing you were talking with your avatar or whatever. 204 00:11:37,140 --> 00:11:39,540 S1: You shared this trauma that you had. You shared this 205 00:11:39,540 --> 00:11:44,300 S1: talking crap about your friend. I mean, there's just so 206 00:11:44,300 --> 00:11:48,650 S1: much possibility here for malicious use by an attacker who 207 00:11:48,650 --> 00:11:52,250 S1: gets hold of one of these databases. So something to 208 00:11:52,280 --> 00:11:57,260 S1: think about, Castillo says. A ransomware attack led to a 209 00:11:57,290 --> 00:12:01,280 S1: theft of sensitive data. I mean, 200GB. They said it 210 00:12:01,280 --> 00:12:05,839 S1: wasn't credit card info, but basically, Castillo got breached. Miter 211 00:12:05,840 --> 00:12:10,910 S1: introduced caldera bug bounty hunter plugin horizon three researchers detailed 212 00:12:10,910 --> 00:12:13,460 S1: how they identified new vulns in Palo Alto, and they 213 00:12:13,490 --> 00:12:17,569 S1: went all the way to full system compromise. And there's 214 00:12:17,570 --> 00:12:23,000 S1: patches out for that. Wayback machine got breached six gigabytes 215 00:12:23,000 --> 00:12:26,959 S1: of a SQL database, 31 million user records site is 216 00:12:26,960 --> 00:12:29,179 S1: still down, but they're working to get it back up. 217 00:12:29,179 --> 00:12:35,030 S1: Researchers from Eset have discovered two sophisticated tool sets used 218 00:12:35,030 --> 00:12:40,010 S1: by nation state groups to go after airgapped devices. I 219 00:12:40,010 --> 00:12:42,679 S1: see air gap, I think Israel. But they're saying this 220 00:12:42,679 --> 00:12:47,480 S1: is possibly Russian drop zone. Thanks again for sponsoring. Cyber 221 00:12:47,510 --> 00:12:51,229 S1: news says Google's Pixel nine Pro sends data packets to 222 00:12:51,260 --> 00:12:55,220 S1: Google every 15 minutes, including location, email and phone number, 223 00:12:55,250 --> 00:12:58,910 S1: even with GPS off. And they're using Wi-Fi to to 224 00:12:58,940 --> 00:13:04,220 S1: estimate location. UNODC warns that South East Asian scammers are 225 00:13:04,220 --> 00:13:09,290 S1: using deepfakes to enhance pig pig butchering. Yeah, pig butchering. These. 226 00:13:09,320 --> 00:13:13,910 S1: These are nasty. It's basically someone builds a relationship with 227 00:13:13,910 --> 00:13:18,199 S1: you over a period of time. And basically, especially for 228 00:13:18,200 --> 00:13:22,339 S1: older people who are lonely and who have money, obviously. Um, 229 00:13:22,340 --> 00:13:25,880 S1: so basically you become their friend, then this person gets 230 00:13:25,880 --> 00:13:29,630 S1: in distress and you're like, oh, I can send you money. 231 00:13:29,630 --> 00:13:33,080 S1: And that's kind of how it's done. Usually. Salt typhoon 232 00:13:33,080 --> 00:13:37,940 S1: Chinese group exploited backdoors meant for lawful data requests. Ukraine 233 00:13:37,940 --> 00:13:42,040 S1: has sentenced two hackers linked to Russia's FSB and the 234 00:13:42,040 --> 00:13:45,939 S1: Armageddon Group to 15 years for cyber attacks on state 235 00:13:45,940 --> 00:13:49,270 S1: institutions in absentia. So they're not there. They don't have 236 00:13:49,270 --> 00:13:52,510 S1: these people, but they have called them out and sentenced 237 00:13:52,540 --> 00:13:58,480 S1: them regardless. OpenAI has stopped another 20 foreign operations, using 238 00:13:58,480 --> 00:14:01,540 S1: its stuff to sway political opinions and meddle in elections. 239 00:14:01,540 --> 00:14:05,170 S1: They said they haven't seen any really nasty malware being 240 00:14:05,170 --> 00:14:07,540 S1: created because they have a lot of a lot of 241 00:14:07,540 --> 00:14:10,780 S1: protections built in for that. So that's that's good news. 242 00:14:10,780 --> 00:14:17,230 S1: But still being used for. Yeah, to try to influence campaigns. Basically, 243 00:14:17,230 --> 00:14:21,400 S1: private intelligence groups like Recorded Future and Flashpoint are changing 244 00:14:21,400 --> 00:14:24,670 S1: intelligence by leveraging tons of data from the internet, including 245 00:14:24,670 --> 00:14:28,180 S1: dark web. And I love this. I love this vibe 246 00:14:28,180 --> 00:14:32,230 S1: of small companies competing with corporations and small Intel groups 247 00:14:32,230 --> 00:14:36,130 S1: competing with, like, you know, big Intel groups in some ways, 248 00:14:36,130 --> 00:14:39,950 S1: study says popular car brands like Hyundai, Kia and Tesla 249 00:14:39,950 --> 00:14:43,940 S1: are collecting driver data, including voice recognition and camera footage, 250 00:14:43,940 --> 00:14:45,950 S1: and sharing it with third parties. I'm going to look 251 00:14:45,950 --> 00:14:49,700 S1: more into this because I'm curious how much Tesla is 252 00:14:49,700 --> 00:14:53,090 S1: sharing because I have a Tesla myself. Pentagon said the 253 00:14:53,090 --> 00:14:57,260 S1: US will send a Thad missile defense system to Israel, 254 00:14:57,260 --> 00:15:00,950 S1: along with 100 troops, just to operate it to improve 255 00:15:00,980 --> 00:15:05,630 S1: defense against Iran. Okay, so AI tech, if you use ChatGPT, 256 00:15:05,660 --> 00:15:08,270 S1: try this prompt just for fun. So this is a 257 00:15:08,270 --> 00:15:11,390 S1: meme or whatever that's going around. And essentially what it 258 00:15:11,390 --> 00:15:16,340 S1: is is it's saying go into ChatGPT and ask it 259 00:15:16,340 --> 00:15:18,560 S1: from all of our interactions together. What is the one 260 00:15:18,560 --> 00:15:20,870 S1: thing you could tell me about myself that I may 261 00:15:20,870 --> 00:15:23,090 S1: not know? And the idea is it's going to come 262 00:15:23,090 --> 00:15:25,550 S1: back with something and you're going to learn some insights 263 00:15:25,550 --> 00:15:27,800 S1: or whatever. And after it gives you an answer, you 264 00:15:27,800 --> 00:15:30,650 S1: ask it for another thing. What else may I not know? 265 00:15:30,650 --> 00:15:34,400 S1: And then what areas might I be holding myself back? 266 00:15:34,400 --> 00:15:37,110 S1: And here's what I got back. You seem to have 267 00:15:37,110 --> 00:15:41,160 S1: a constant drive to synthesize ideas across different domains, almost 268 00:15:41,160 --> 00:15:43,920 S1: like a mental architect. And as soon as I read, 269 00:15:43,920 --> 00:15:47,700 S1: like the first sentence, I'm like, uh, this feels gross. 270 00:15:47,700 --> 00:15:49,950 S1: I don't like it. And so this is what I 271 00:15:49,950 --> 00:15:54,300 S1: wrote about it in the newsletter yesterday. Basically, it feels 272 00:15:54,300 --> 00:15:58,410 S1: very complimentary and a bit like a horoscope. It feels 273 00:15:58,410 --> 00:16:00,720 S1: like a scam, right? Like it's just trying to make 274 00:16:00,720 --> 00:16:05,100 S1: me feel good about myself now. Simon, um, I, I 275 00:16:05,100 --> 00:16:07,440 S1: can't remember his last name. Um, I always mess up 276 00:16:07,440 --> 00:16:09,840 S1: his last name, but Simon, he talks a lot about 277 00:16:09,870 --> 00:16:13,980 S1: AI stuff, and he basically had the same exact observation. 278 00:16:13,980 --> 00:16:17,850 S1: And he pointed out, in addition, that it's not actually 279 00:16:17,850 --> 00:16:22,830 S1: doing introspection on all your conversations. Okay, so so what 280 00:16:22,830 --> 00:16:24,330 S1: you actually have to do, if you want to do that, 281 00:16:24,330 --> 00:16:26,490 S1: you have to go get the conversations and send it 282 00:16:26,490 --> 00:16:30,930 S1: in this thing that it built here. I know specifically, 283 00:16:30,930 --> 00:16:33,479 S1: I knew immediately when it came back. It's working off 284 00:16:33,480 --> 00:16:36,390 S1: of my system prompt. So if you are a paid 285 00:16:36,390 --> 00:16:40,470 S1: member and you have anything inside of your, um, your 286 00:16:40,470 --> 00:16:45,240 S1: config for ChatGPT, it's basically, um, here's who I am. 287 00:16:45,240 --> 00:16:47,880 S1: Here's how I would like you to interact with me. Right. 288 00:16:47,910 --> 00:16:51,420 S1: So I have stuff in there that mentions human 3.0. 289 00:16:51,450 --> 00:16:54,960 S1: It mentions like the type of stuff that I'm interested 290 00:16:54,960 --> 00:16:57,300 S1: in because I'm using that to have it shape how 291 00:16:57,300 --> 00:16:59,130 S1: it responds to me when I ask it a question 292 00:16:59,130 --> 00:17:01,680 S1: about whatever I want it to answer me in a 293 00:17:01,680 --> 00:17:05,580 S1: specific way. So all of that is what it's basically 294 00:17:05,609 --> 00:17:09,330 S1: going off of. And it's essentially hallucination. It's making up 295 00:17:09,330 --> 00:17:12,630 S1: all of this stuff to make you feel good. And 296 00:17:12,630 --> 00:17:16,410 S1: in fact, if you go and ask now, um, I 297 00:17:16,410 --> 00:17:20,220 S1: think they might have modified the output and basically said that, no, 298 00:17:20,220 --> 00:17:23,219 S1: we can't give you deep analysis. We don't actually have 299 00:17:23,220 --> 00:17:26,610 S1: the history of all your conversations readily available. You would 300 00:17:26,609 --> 00:17:28,830 S1: need to put them in. It's giving some kind of 301 00:17:28,859 --> 00:17:33,200 S1: a warning to counteract people doing this meme and thinking 302 00:17:33,200 --> 00:17:36,139 S1: the wrong thing about it. But yeah, here's what I 303 00:17:36,140 --> 00:17:40,220 S1: wrote about it yesterday. It's like, well, everyone's going to agree. 304 00:17:40,220 --> 00:17:43,190 S1: If you know, something comes back and it's like, you know, 305 00:17:43,220 --> 00:17:45,409 S1: I just think you're so insightful and you're so smart 306 00:17:45,410 --> 00:17:47,630 S1: and you're trying to help the world and all this, 307 00:17:47,630 --> 00:17:51,590 S1: and it's like, oh, wow, that it's so accurate. It's 308 00:17:51,590 --> 00:17:54,290 S1: so accurate. It really gets me right. This is the 309 00:17:54,320 --> 00:17:57,409 S1: same trick that horoscope and palm reader people use against 310 00:17:57,410 --> 00:18:01,880 S1: you is to like, just basically find special ways and 311 00:18:01,880 --> 00:18:04,220 S1: unique ways to compliment you. And it's like, oh, I've 312 00:18:04,220 --> 00:18:07,669 S1: never seen how awesome I was before. Thank you. So 313 00:18:07,670 --> 00:18:11,330 S1: I asked people, you know, curious what you get. And 314 00:18:11,330 --> 00:18:13,129 S1: I got back a whole bunch of responses and it 315 00:18:13,130 --> 00:18:15,530 S1: was basically the exact same thing. And a couple of 316 00:18:15,530 --> 00:18:17,780 S1: people responded and said, look, it's just going off the 317 00:18:17,780 --> 00:18:21,919 S1: system prompt. It's just hallucinating. So I think we all 318 00:18:21,950 --> 00:18:25,159 S1: sort of collectively figured out it was kind of lame. 319 00:18:25,190 --> 00:18:29,300 S1: Apple's AI researchers found that llms from meta and OpenAI 320 00:18:29,330 --> 00:18:32,570 S1: struggle with basic reasoning and they've got a whole paper 321 00:18:32,600 --> 00:18:36,199 S1: basically going into that. I think it's more accurate to 322 00:18:36,230 --> 00:18:39,350 S1: say that it's easy to disrupt the reasoning than to 323 00:18:39,350 --> 00:18:42,800 S1: say it doesn't have any. Geoffrey Hinton, often dubbed godfather 324 00:18:42,800 --> 00:18:46,010 S1: of AI, has won the Nobel Prize in Physics for 325 00:18:46,010 --> 00:18:49,879 S1: early work in neural networks. Congrats to him. And one 326 00:18:49,880 --> 00:18:52,880 S1: thing to notice is that he is in the Doomer camp. 327 00:18:52,880 --> 00:18:58,340 S1: And anytime somebody gets a Nobel Prize for crazy work 328 00:18:58,340 --> 00:19:02,990 S1: in something like really groundbreaking work, you should listen, especially 329 00:19:02,990 --> 00:19:05,990 S1: if they disagree with you. He disagrees with me, so 330 00:19:05,990 --> 00:19:11,960 S1: I'm listening. Musk unveiled Tesla's new robotaxi self-driving electric vehicle 331 00:19:11,960 --> 00:19:15,560 S1: without a steering wheel or pedals at the We Robot event, 332 00:19:15,560 --> 00:19:18,800 S1: and there was so much hate about this. In the 333 00:19:18,800 --> 00:19:21,439 S1: first version that I recorded yesterday, I went into a 334 00:19:21,440 --> 00:19:25,130 S1: whole like political thing, which was too much. I don't 335 00:19:25,130 --> 00:19:26,720 S1: want to do that again here. So I'm going to 336 00:19:26,720 --> 00:19:29,890 S1: try to overview instead of it. I would say that 337 00:19:29,890 --> 00:19:35,980 S1: everyone or most people, most people are very anti like 338 00:19:36,010 --> 00:19:40,090 S1: militantly anti Elon or they think he's like the new 339 00:19:40,090 --> 00:19:43,119 S1: Jesus and I don't feel like there's very many people 340 00:19:43,119 --> 00:19:46,810 S1: who see the subtlety and nuance. And if you go 341 00:19:46,810 --> 00:19:49,660 S1: all the way down to the bottom here, I think 342 00:19:49,660 --> 00:19:54,280 S1: it's because not enough people are reading biographies, right? Biographies, 343 00:19:54,280 --> 00:19:58,119 S1: by definition, are written by or written about people who 344 00:19:58,119 --> 00:20:01,090 S1: have done extraordinary things. And if you go and look 345 00:20:01,119 --> 00:20:05,530 S1: at a whole bunch of inventors like Edison and Tesla 346 00:20:05,530 --> 00:20:09,490 S1: and people like that, um, they tend Ben Franklin, they 347 00:20:09,490 --> 00:20:13,660 S1: tend to be crazy and weird and, like, not take 348 00:20:13,690 --> 00:20:17,470 S1: showers and have weird social interactions and, like, sleep under 349 00:20:17,470 --> 00:20:23,440 S1: their desk and just smell, like, absolutely horrible and all 350 00:20:23,440 --> 00:20:25,510 S1: sorts of things like this. Right? They tend to be 351 00:20:25,510 --> 00:20:30,010 S1: very strange and oftentimes they do bad things, right. And 352 00:20:30,010 --> 00:20:32,890 S1: also they're down on themselves and they have these up 353 00:20:32,890 --> 00:20:36,940 S1: cycles like manic depressive type vibes, or they have a 354 00:20:36,940 --> 00:20:40,330 S1: Jesus complex, or they hate themselves and they think they're 355 00:20:40,330 --> 00:20:43,480 S1: not good enough and they're horribly jealous of everyone. Like 356 00:20:43,510 --> 00:20:50,020 S1: you're not finding horribly stable people, like very regular people, right, 357 00:20:50,050 --> 00:20:53,740 S1: who get biographies written about them. Usually not always the case, 358 00:20:53,740 --> 00:20:57,399 S1: but usually they tend to be kind of weird. And 359 00:20:57,400 --> 00:21:00,310 S1: I think that's important when you look at somebody like Elon, 360 00:21:00,310 --> 00:21:02,440 S1: because I feel like people are looking at it through 361 00:21:02,440 --> 00:21:07,630 S1: the lens of today and what's like considered normal on 362 00:21:07,630 --> 00:21:12,939 S1: X or in, you know, in California or specifically in 363 00:21:12,940 --> 00:21:17,530 S1: San Francisco, in California, like what is considered acceptable behavior. 364 00:21:17,530 --> 00:21:20,770 S1: And I think the question is, if somebody is an 365 00:21:20,770 --> 00:21:26,550 S1: extraordinary creator, which we also have to like, agree on that. 366 00:21:26,550 --> 00:21:29,520 S1: I would say he is, but let's say that you 367 00:21:29,520 --> 00:21:32,430 S1: agree he's an extraordinary creator and you're like, yeah, but 368 00:21:32,430 --> 00:21:35,490 S1: it doesn't matter because he has all these flaws. I 369 00:21:35,490 --> 00:21:39,330 S1: ask you again to look backward at all the different 370 00:21:39,330 --> 00:21:42,060 S1: things that have been created in the world, and what 371 00:21:42,060 --> 00:21:46,230 S1: percentage of those were created by essentially a crazy person, 372 00:21:46,260 --> 00:21:49,650 S1: crazy and or broken and or messed up and or 373 00:21:49,650 --> 00:21:54,600 S1: in need of therapy and or gross or in some way, 374 00:21:54,630 --> 00:21:59,130 S1: you know, counter normal. And I'm not saying excuses, right. 375 00:21:59,130 --> 00:22:00,959 S1: I'm not saying it excuses all the people in the 376 00:22:00,960 --> 00:22:04,740 S1: past either. I'm just saying reality tends to look like this. 377 00:22:04,740 --> 00:22:08,430 S1: So we should keep that in mind. All right. Deals, deals, 378 00:22:08,430 --> 00:22:11,910 S1: sales staff given two days notice to return to office 379 00:22:11,910 --> 00:22:15,659 S1: full time. So I've been talking about this quite a bit. Essentially, 380 00:22:15,660 --> 00:22:19,950 S1: return to office is a way of doing firing essentially riffs. 381 00:22:19,980 --> 00:22:25,139 S1: Right toe equals riff. Not always, but in many, many cases, 382 00:22:25,140 --> 00:22:28,139 S1: it's like it's a cheating way of doing it. Um, 383 00:22:28,140 --> 00:22:35,970 S1: co-founder of Robinhood. Baiju is going to collect and transmit 384 00:22:35,970 --> 00:22:40,409 S1: solar energy using infrared lasers, and he's basically going to 385 00:22:40,440 --> 00:22:43,380 S1: I believe he's going to send the energy back down 386 00:22:43,380 --> 00:22:45,270 S1: to earth using a laser as well. So it'd have 387 00:22:45,270 --> 00:22:47,670 S1: to be like a receiver. And I'm like, well, you 388 00:22:47,670 --> 00:22:49,620 S1: need you would need to be able to target this 389 00:22:49,619 --> 00:22:53,880 S1: receiver while in space, moving very fast, right? Going around 390 00:22:53,880 --> 00:22:56,879 S1: the planet like every hour and a half. That's very, 391 00:22:56,880 --> 00:23:00,330 S1: very fast. But you'd still be able to hit these receivers. 392 00:23:00,359 --> 00:23:02,880 S1: Actually I think they would be geosynchronous. So you wouldn't 393 00:23:02,910 --> 00:23:07,170 S1: actually have to they wouldn't be moving in relation to 394 00:23:07,170 --> 00:23:12,210 S1: the receivers. Maybe not sure. Either way, you are technically 395 00:23:12,210 --> 00:23:15,900 S1: building a space laser, and if you're sending enough energy 396 00:23:15,900 --> 00:23:18,480 S1: to be useful on the ground, you're also sending enough 397 00:23:18,480 --> 00:23:21,780 S1: energy that could potentially be used as a weapon. So 398 00:23:21,800 --> 00:23:26,060 S1: just expect to have at least a questionnaire for the 399 00:23:26,060 --> 00:23:32,210 S1: military and or a request for purchase from the military. U.S. 400 00:23:32,210 --> 00:23:36,170 S1: Department of Justice considering breaking up Google after court said 401 00:23:36,170 --> 00:23:42,320 S1: they've crushed competition. Google Chrome, Android. Yeah, lots of different 402 00:23:42,320 --> 00:23:46,040 S1: products they have potentially. I agree with this. I agree 403 00:23:46,040 --> 00:23:48,859 S1: with Scott Galloway that a lot of these services need 404 00:23:48,890 --> 00:23:52,609 S1: to be broken up, potentially even Apple, even though I 405 00:23:52,609 --> 00:23:54,830 S1: love Apple and I don't think they've really done anything 406 00:23:54,830 --> 00:23:58,760 S1: wrong at some point. It's like a reward. It's basically 407 00:23:58,760 --> 00:24:02,240 S1: saying you are too awesome, therefore we have to break 408 00:24:02,240 --> 00:24:06,050 S1: you up and not necessarily like an indication that you're 409 00:24:06,050 --> 00:24:08,869 S1: a bad person. You know what I mean? Ticketmaster is 410 00:24:08,869 --> 00:24:12,500 S1: the first to use Apple's upgraded wallet tickets for iOS 18, 411 00:24:12,530 --> 00:24:17,840 S1: giving us stuff like venue, maps, parking, music playlists, weather forecasts, 412 00:24:17,869 --> 00:24:22,070 S1: stuff like that, stuff like that. So that's good. Anything 413 00:24:22,100 --> 00:24:25,250 S1: to make Ticketmaster suck less? That's what I wrote. New 414 00:24:25,250 --> 00:24:29,600 S1: HBO documentary claims Canadian crypto expert Peter Todd is the 415 00:24:29,600 --> 00:24:35,210 S1: mysterious inventor of Bitcoin Satoshi. Todd says nope, I was 416 00:24:35,210 --> 00:24:37,220 S1: too busy with school and work at the time, so 417 00:24:37,220 --> 00:24:39,620 S1: it couldn't have been me. Which is exactly what Satoshi 418 00:24:39,650 --> 00:24:44,629 S1: would say. For Taiwanese employees at Foxconn's Zhengzhou plant, world's 419 00:24:44,630 --> 00:24:48,139 S1: largest iPhone production facility have been detained by Chinese authorities. 420 00:24:48,170 --> 00:24:51,890 S1: Nobody knows why they assume politics. Humans. Looks like Christopher 421 00:24:51,890 --> 00:24:57,080 S1: Columbus was a Sephardic Jew from Western Europe. I think 422 00:24:57,080 --> 00:25:00,409 S1: they're saying Spain can't remember if he was Spanish or 423 00:25:00,410 --> 00:25:03,020 S1: if he was Portuguese. I can't remember that whole vibe. 424 00:25:03,020 --> 00:25:06,830 S1: But they're saying, yeah, Spanish Jew. And I think Sephardic 425 00:25:06,830 --> 00:25:09,950 S1: means I can't remember exactly. I think it means like 426 00:25:09,980 --> 00:25:14,570 S1: Israeli Jew as opposed to European Jew, which is Ashkenazi. 427 00:25:14,660 --> 00:25:17,540 S1: I believe that's the case. JP Morgan and Wells Fargo 428 00:25:17,540 --> 00:25:22,119 S1: dips in profits. So they say it was geopolitical tension. 429 00:25:22,119 --> 00:25:25,600 S1: I saw Bank of America saying they had something similar happen. 430 00:25:25,600 --> 00:25:29,320 S1: So yeah, banks are sort of struggling. Although I did 431 00:25:29,320 --> 00:25:32,800 S1: see Goldman Sachs say they were doing okay, I believe 432 00:25:32,830 --> 00:25:36,580 S1: I can't remember the details, but some mixed results on 433 00:25:36,580 --> 00:25:38,830 S1: the banks. Your brain changes based on what you did 434 00:25:38,859 --> 00:25:41,170 S1: two weeks ago. So this is basically saying that it's 435 00:25:41,170 --> 00:25:43,570 S1: not just what you did yesterday or the day before, 436 00:25:43,570 --> 00:25:48,760 S1: but that there's like accumulated downsides of having crappy diet 437 00:25:48,760 --> 00:25:52,660 S1: and exercise and sleep over like the course of two weeks. 438 00:25:52,660 --> 00:25:55,900 S1: So pretty cool. I mean, essentially, to me this says 439 00:25:55,930 --> 00:26:00,040 S1: keep doing your routine because like it accumulates, the benefits accumulate. 440 00:26:00,070 --> 00:26:04,600 S1: American Heart Association outlines a strict protocol for taking blood pressure. 441 00:26:04,600 --> 00:26:07,780 S1: They're saying you have to sit calmly with an empty bladder, 442 00:26:07,780 --> 00:26:11,109 S1: and that too many people are measuring right after running 443 00:26:11,109 --> 00:26:15,580 S1: around or while you have to pee. And basically this 444 00:26:15,580 --> 00:26:21,640 S1: this is inflating the hypertension numbers. Boeing is cutting 17,000 jobs, 10% 445 00:26:21,640 --> 00:26:26,260 S1: of its workforce. Federal emergency workers in Rutherford County, North Carolina, 446 00:26:26,260 --> 00:26:30,940 S1: were temporarily moved after reports of an armed militia threatening 447 00:26:30,940 --> 00:26:35,560 S1: government personnel, essentially because they think they helped somehow create 448 00:26:35,560 --> 00:26:39,850 S1: the hurricanes, I assume because they worked for the government. 449 00:26:39,880 --> 00:26:44,470 S1: Elizabeth Landau says single cell cyanobacteria can anticipate seasonal changes 450 00:26:44,470 --> 00:26:47,830 S1: by sensing day length and preparing for winter. Single cell 451 00:26:47,830 --> 00:26:51,910 S1: organism can sense seasons. That trips me out and kind 452 00:26:51,910 --> 00:26:55,240 S1: of explains how trees start, you know, turning fall and 453 00:26:55,240 --> 00:26:58,720 S1: dropping leaves and all that. United Airlines is adding new 454 00:26:58,720 --> 00:27:03,250 S1: routes to lesser known destinations like Bilbao, Faro, Madeira, Sicily 455 00:27:03,250 --> 00:27:07,240 S1: and Newark trying to get away from crowded hotspots essentially 456 00:27:07,270 --> 00:27:10,420 S1: got some journals from Alexei Navalny while he was in prison, 457 00:27:10,420 --> 00:27:13,690 S1: for he died in prison. Retail sales have dropped from 458 00:27:13,690 --> 00:27:18,940 S1: 7.5 to 5.7% of employment over the last decade. Losing 459 00:27:18,970 --> 00:27:25,359 S1: 850,000 positions despite the US adding 19 million jobs overall. 460 00:27:25,390 --> 00:27:28,389 S1: Looks like Wegovy and Zepp bound might have been the 461 00:27:28,390 --> 00:27:32,470 S1: cause of US adult obesity rate, dropping by two percentage 462 00:27:32,470 --> 00:27:37,420 S1: points from 2020 to 2023. And speaking of that, new 463 00:27:37,450 --> 00:27:41,199 S1: GLP one weight loss drugs are coming in pill form 464 00:27:41,200 --> 00:27:46,900 S1: soon to basically replace the weekly injections. Dhairya Karwa Mirza, 465 00:27:46,930 --> 00:27:53,320 S1: self-taught Kurdish astrophotographer. Look at this image that she took. Extraordinary. 466 00:27:53,350 --> 00:27:56,500 S1: Absolutely extraordinary. Look at that. Isn't that cool? Looks kind 467 00:27:56,500 --> 00:27:59,740 S1: of oblong. Not sure why. Maybe the shadow anyway. Very 468 00:27:59,740 --> 00:28:04,540 S1: cool ideas. Gullibility. Not disinformation. So I don't think the 469 00:28:04,570 --> 00:28:07,449 S1: US has a misinformation problem. I think it has a 470 00:28:07,480 --> 00:28:10,780 S1: gullibility problem. It's not that we're being fed too much crap, 471 00:28:10,780 --> 00:28:13,360 S1: it's the fact that we're actually eating it. Some two 472 00:28:13,359 --> 00:28:16,260 S1: large number of Republicans now believe that the Democrats are 473 00:28:16,260 --> 00:28:19,139 S1: sending hurricanes to Florida because it's election time. That's a 474 00:28:19,140 --> 00:28:23,880 S1: population problem, an education problem, not a conspiracy theory problem. 475 00:28:23,880 --> 00:28:27,270 S1: So in infosec terms, we need to reduce our vulnerability, 476 00:28:27,300 --> 00:28:30,540 S1: not try to get all the threats removed, right, because 477 00:28:30,540 --> 00:28:32,399 S1: the threats will always be there. And in fact, the 478 00:28:32,400 --> 00:28:34,350 S1: threats are going to get better. The only chance of 479 00:28:34,350 --> 00:28:39,000 S1: actually fixing this is education about how the world actually works. 480 00:28:39,000 --> 00:28:42,240 S1: And unfortunately, the far left and the far right seem 481 00:28:42,240 --> 00:28:45,450 S1: to have like lost touch with that, like the basic 482 00:28:45,450 --> 00:28:48,270 S1: facts of reality, because it's not just the far right 483 00:28:48,270 --> 00:28:51,870 S1: that's doing this, right. Remember the anti-vax thing that was 484 00:28:51,870 --> 00:28:55,140 S1: far left before it was far right? It was far left. 485 00:28:55,140 --> 00:28:57,630 S1: And then Covid happened and then the far right jumped on. 486 00:28:57,630 --> 00:28:59,640 S1: And I guess now the far left and the far 487 00:28:59,640 --> 00:29:04,080 S1: right are anti-vax. Not really sure, but the point is 488 00:29:04,080 --> 00:29:07,620 S1: we have a vulnerability, which is a lack of education 489 00:29:07,620 --> 00:29:10,410 S1: and a lack of understanding of how the world works. 490 00:29:10,410 --> 00:29:14,480 S1: That's the vulnerability. Threats are things like misinformation. We can 491 00:29:14,480 --> 00:29:17,960 S1: have a threat and not be vulnerable, and we still 492 00:29:17,960 --> 00:29:22,490 S1: have lower risk. So that's the vibe there. Discovery swarm, 493 00:29:22,490 --> 00:29:27,830 S1: OpenAI's new experimental framework for building and orchestrating multi-agent systems, 494 00:29:27,830 --> 00:29:30,380 S1: feels a lot like lane change to me. I didn't 495 00:29:30,380 --> 00:29:33,920 S1: see so many benefits. I'm going to mess with it. 496 00:29:33,920 --> 00:29:37,459 S1: More command tools. I like really good set of tools here. 497 00:29:37,490 --> 00:29:44,690 S1: RFT Delta tldr s-oxide http I fcf. Yeah, lots of 498 00:29:44,690 --> 00:29:48,920 S1: different tools here Xvm better vim mode for ssh. I'm 499 00:29:48,920 --> 00:29:53,450 S1: sorry for zsh. Love this thing. It's syntax highlighting. It's 500 00:29:53,450 --> 00:29:58,100 S1: basically a full featured version of vim on the command line, 501 00:29:58,100 --> 00:30:02,270 S1: so I use vim mode for my zsh have been 502 00:30:02,270 --> 00:30:05,300 S1: for years, but this gives you like syntax highlighting. It 503 00:30:05,300 --> 00:30:07,850 S1: gives you the ability to use like the surround plugin, 504 00:30:07,850 --> 00:30:12,380 S1: all kinds of stuff. Theano 3.0 AI powered API documentation 505 00:30:12,380 --> 00:30:16,520 S1: tool I updated my post on dynamic content generation. I 506 00:30:16,520 --> 00:30:20,870 S1: think this is going to be insanely disruptive to so 507 00:30:20,870 --> 00:30:24,290 S1: many industries. Highly recommend you go check that out. Essentially, 508 00:30:24,290 --> 00:30:27,650 S1: what it means is people are going to produce raw content. 509 00:30:27,680 --> 00:30:32,090 S1: Like for example, this is raw content of a full podcast, right? 510 00:30:32,120 --> 00:30:35,660 S1: And this has got me in this bottom right here 511 00:30:35,660 --> 00:30:38,540 S1: talking to the camera. And let's say this, this whole 512 00:30:38,540 --> 00:30:42,560 S1: episode is like 20 minutes long. Well, let's say that, um, 513 00:30:42,650 --> 00:30:47,660 S1: you have a, like a five minute bus ride or 514 00:30:47,660 --> 00:30:49,820 S1: a five minute bike ride, or you're waiting in line 515 00:30:49,820 --> 00:30:52,640 S1: for five minutes or whatever. And and this is like, 516 00:30:52,670 --> 00:30:54,950 S1: I don't know, three, five, five years in the future. 517 00:30:54,950 --> 00:30:57,560 S1: And you're like, hey, I want a summary. What did 518 00:30:57,560 --> 00:31:00,320 S1: Daniel put in his podcast yesterday? Your AI is not 519 00:31:00,320 --> 00:31:02,930 S1: going to go and find the video and speed it up. 520 00:31:02,930 --> 00:31:06,560 S1: In fact, your AI is just going to read, watch, 521 00:31:06,560 --> 00:31:10,000 S1: consume the whole video if it hasn't already, and it's 522 00:31:10,000 --> 00:31:13,840 S1: going to make a version of it. Maybe it's my face, 523 00:31:13,840 --> 00:31:16,600 S1: maybe it's me in a video. Maybe it's only audio 524 00:31:16,600 --> 00:31:19,330 S1: because they can actually look at anything right now. But anyway, 525 00:31:19,330 --> 00:31:22,300 S1: it's going to play in your ear a five minute version. 526 00:31:22,300 --> 00:31:24,430 S1: It's going to pick the best stuff out of here 527 00:31:24,430 --> 00:31:26,290 S1: that you actually care about. How is it going to 528 00:31:26,290 --> 00:31:29,710 S1: do that? Because your eye knows you. Your eye knows 529 00:31:29,710 --> 00:31:32,110 S1: what parts of these which which of these stories are 530 00:31:32,110 --> 00:31:35,350 S1: actually good or most pertinent to you. It's going to 531 00:31:35,350 --> 00:31:38,050 S1: take out the garbage. It doesn't like those stories, thinks 532 00:31:38,050 --> 00:31:41,410 S1: they're dumb, thinks it was filler or whatever, and it's 533 00:31:41,410 --> 00:31:44,170 S1: going to build a clean little clip. It could be 534 00:31:44,170 --> 00:31:46,630 S1: a 32nd clip, but it could be read in the 535 00:31:46,630 --> 00:31:50,770 S1: voice of Richard Feynman. It could be read in the 536 00:31:50,770 --> 00:31:54,070 S1: voice of Agatha Christie. It could be it could be anyone. 537 00:31:54,070 --> 00:31:57,460 S1: It could be Taylor Swift. Right. Telling you the news 538 00:31:57,460 --> 00:32:02,050 S1: that that came out of the Unsupervised Learning podcast. Point is, 539 00:32:02,050 --> 00:32:05,709 S1: your eye is going to be giving you content in 540 00:32:05,710 --> 00:32:10,150 S1: any format, in any presenter deepfaked exactly the way you 541 00:32:10,150 --> 00:32:16,720 S1: want it from 10s to giant series. Okay, the original content, 542 00:32:16,720 --> 00:32:20,020 S1: which lives in a textbook somewhere, or in some YouTube 543 00:32:20,020 --> 00:32:23,980 S1: video or in some blog post. That original content you 544 00:32:23,980 --> 00:32:26,980 S1: never have to mess with again. Your AI is going 545 00:32:27,010 --> 00:32:29,920 S1: to consume all of that and then dynamically build the 546 00:32:29,920 --> 00:32:33,460 S1: content for their principal. So that's the idea there. Augment 547 00:32:33,490 --> 00:32:37,630 S1: UI use AI to prototype front end designs. Software engineer 548 00:32:37,660 --> 00:32:41,230 S1: pay heatmap across the US, the digits of pi are 549 00:32:41,230 --> 00:32:44,469 S1: not random. Passbook lets you create an Apple Wallet, pass 550 00:32:44,470 --> 00:32:47,410 S1: from any QR code and export it to wallet. How 551 00:32:47,410 --> 00:32:50,530 S1: I animate three blue one Brown. It's one of my 552 00:32:50,530 --> 00:32:54,250 S1: favorite YouTubers, describes a whole bunch of science stuff and 553 00:32:54,250 --> 00:32:57,190 S1: recommendation of the week. If you want to calm your 554 00:32:57,220 --> 00:33:00,340 S1: nerves during this next month and a half going into 555 00:33:00,340 --> 00:33:04,510 S1: this election, go read about the Civil Rights movement and 556 00:33:04,510 --> 00:33:08,100 S1: how Disruptive that was to politics in the US. And 557 00:33:08,100 --> 00:33:10,680 S1: it was actually, in a lot of ways, way worse 558 00:33:10,680 --> 00:33:14,370 S1: than what we're seeing today. So we survived that really 559 00:33:14,370 --> 00:33:20,310 S1: bad stuff. We probably will again survive this election cycle, probably. 560 00:33:20,310 --> 00:33:22,680 S1: And the aphorism of the week, what is to give 561 00:33:22,710 --> 00:33:27,960 S1: light must endure burning. What is to give light must 562 00:33:27,990 --> 00:33:33,570 S1: endure burning Viktor Frankl. Unsupervised learning is produced and edited 563 00:33:33,570 --> 00:33:38,040 S1: by Daniel Miessler on a Neumann U87 AI microphone using Hindenburg. 564 00:33:38,460 --> 00:33:41,430 S1: Intro and outro music is by zombie with a Y. 565 00:33:42,060 --> 00:33:44,280 S1: And to get the text and links from this episode, 566 00:33:44,280 --> 00:33:46,530 S1: sign up for the newsletter version of the show at 567 00:33:46,530 --> 00:33:51,330 S1: Daniel missler.com/newsletter. We'll see you next time.