1 00:00:00,050 --> 00:00:03,200 S1: If you're building B2B SaaS, you probably want to sell 2 00:00:03,230 --> 00:00:06,320 S1: to enterprises at some point. The challenge is that selling 3 00:00:06,350 --> 00:00:09,590 S1: to enterprises requires you to support complex features like Saml, 4 00:00:09,590 --> 00:00:13,550 S1: single sign on, skim provisioning, audit logs, and fine grained 5 00:00:13,580 --> 00:00:18,500 S1: authorization that require months of development time and ongoing maintenance work. 6 00:00:18,530 --> 00:00:21,500 S1: OS solves this with APIs that are flexible and easy 7 00:00:21,500 --> 00:00:24,919 S1: to use. It makes adding enterprise features to your app 8 00:00:24,950 --> 00:00:28,700 S1: a quick and painless process. Work OS also has user 9 00:00:28,700 --> 00:00:31,700 S1: management solution called auth Kit that is free up to 10 00:00:31,730 --> 00:00:36,260 S1: 1 million monthly active users. It comes standard with MFA, Rbac, 11 00:00:36,260 --> 00:00:39,920 S1: bot protection, user impersonation and more. Auth kit is built 12 00:00:39,920 --> 00:00:44,690 S1: with radix components, which means zero compromises in design. You 13 00:00:44,690 --> 00:00:50,030 S1: get limitless customizations and modular templates designed for quick integrations. 14 00:00:50,060 --> 00:00:54,050 S1: Work OS is already used by hundreds of fast growing 15 00:00:54,080 --> 00:00:58,880 S1: companies like cursor, Vercel, Perplexity and Sierra. Check it out 16 00:00:58,880 --> 00:01:04,040 S1: at work OS Com that is working os. Com. Welcome 17 00:01:04,040 --> 00:01:07,610 S1: to Unsupervised Learning, a security, AI and meaning focused podcast 18 00:01:07,610 --> 00:01:10,039 S1: that looks at how best to thrive as humans in 19 00:01:10,040 --> 00:01:14,180 S1: a post AI world. It combines original ideas, analysis, and 20 00:01:14,180 --> 00:01:17,390 S1: mental models to bring not just the news, but why 21 00:01:17,390 --> 00:01:25,310 S1: it matters and how to respond. All right, welcome to 22 00:01:25,310 --> 00:01:28,640 S1: unsupervised learning. This is Daniel. All right. So settling in 23 00:01:28,670 --> 00:01:32,300 S1: for a couple of light holiday weeks trying to chill 24 00:01:32,300 --> 00:01:35,180 S1: out as much as possible, reading a bunch of lit RPG. 25 00:01:35,209 --> 00:01:37,640 S1: I don't think I'm going to play too much path 26 00:01:37,640 --> 00:01:40,940 S1: of Exile. It's like a religion. I've already got too 27 00:01:40,940 --> 00:01:44,449 S1: many religions. I've got reading as a religion, I've got 28 00:01:44,480 --> 00:01:47,600 S1: AI as a religion, got security work. I'm building a 29 00:01:47,600 --> 00:01:50,840 S1: bunch of tech. I just I don't have time to 30 00:01:50,870 --> 00:01:55,610 S1: invest in getting good at Poe. It's just too intimidating. Plus, 31 00:01:55,610 --> 00:01:58,040 S1: the little bit that I did play was mostly like 32 00:01:58,040 --> 00:02:01,900 S1: hitting the space bar and dodging way too much dodging. 33 00:02:01,900 --> 00:02:03,940 S1: So don't think I'm going to be playing that any 34 00:02:03,970 --> 00:02:06,670 S1: time soon. But I do watch other people play it, 35 00:02:06,670 --> 00:02:09,220 S1: so that's kind of fun. Please go subscribe to the 36 00:02:09,220 --> 00:02:11,680 S1: YouTube channel if you're watching this on YouTube. That's probably 37 00:02:11,680 --> 00:02:14,710 S1: kind of silly because you're probably already on the channel 38 00:02:14,710 --> 00:02:18,280 S1: and hopefully you're subscribed, but always forget to promote that. 39 00:02:18,280 --> 00:02:21,970 S1: So trying to remember wrote a piece for days. Been 40 00:02:21,970 --> 00:02:25,060 S1: wanting to get out for a while. It's basically about 41 00:02:25,240 --> 00:02:28,900 S1: how updated context is going to be like the centerpiece 42 00:02:28,900 --> 00:02:31,870 S1: of vulnerability management, because the problem isn't finding villains, it's 43 00:02:31,900 --> 00:02:36,130 S1: actually fixing them. So go check that out and let's 44 00:02:36,130 --> 00:02:41,140 S1: get into security. So full face masks for facial recognition evasion. 45 00:02:41,139 --> 00:02:45,010 S1: So realistic masks that you could put on I'm going 46 00:02:45,040 --> 00:02:47,200 S1: to open this up. This is why we do it live. 47 00:02:47,230 --> 00:02:50,230 S1: Look at this thing. So it's like a stocking or whatever. 48 00:02:50,230 --> 00:02:52,510 S1: But you put it over your face and then it 49 00:02:52,540 --> 00:02:55,240 S1: kind of pulls to contort to your face and look 50 00:02:55,270 --> 00:02:58,000 S1: at that from a distance. That is not so bad. 51 00:02:58,020 --> 00:02:59,910 S1: not so bad at all. And look at all these 52 00:02:59,910 --> 00:03:02,910 S1: different faces you got so pretty cool. Yeah. Bruce Schneier 53 00:03:02,940 --> 00:03:06,359 S1: is the one I think I saw that from. But yeah, 54 00:03:06,389 --> 00:03:09,240 S1: I'm curious. I'd love to see some testing done from 55 00:03:09,270 --> 00:03:11,460 S1: the security community to buy a bunch of these. And 56 00:03:11,460 --> 00:03:13,440 S1: then maybe if we have access to like a facial 57 00:03:13,440 --> 00:03:16,290 S1: recognition system to see if we could get it to 58 00:03:16,320 --> 00:03:19,860 S1: match someone else. I think the big thing is like, 59 00:03:19,889 --> 00:03:24,419 S1: is it only visual or is there like an active component? 60 00:03:24,419 --> 00:03:26,970 S1: I think, and I don't think most do have an 61 00:03:26,970 --> 00:03:30,300 S1: active component. I know face ID does, but I think 62 00:03:30,300 --> 00:03:34,470 S1: most facial recognition does not have an active component. And 63 00:03:34,470 --> 00:03:37,200 S1: what that would mean is like is something actually bouncing 64 00:03:37,200 --> 00:03:40,560 S1: off your face, which would go through the stocking. So 65 00:03:40,560 --> 00:03:42,660 S1: it'd be able to see like where your cheekbones are, 66 00:03:42,660 --> 00:03:46,710 S1: where your chin is, and, you know, eyes, foreheads, stuff 67 00:03:46,710 --> 00:03:50,790 S1: like that. That would seem to break the system or 68 00:03:50,790 --> 00:03:54,870 S1: break the masks. But yeah, I don't I don't think 69 00:03:54,870 --> 00:03:57,869 S1: most are doing that, but I'm not sure. Visual Studio 70 00:03:57,870 --> 00:04:02,070 S1: Code remote tunnel hacks. So some Chinese attackers are caught 71 00:04:02,070 --> 00:04:05,700 S1: stealing V's codes, remote tunnel feature, or using the feature 72 00:04:05,700 --> 00:04:08,970 S1: to hack it. Service providers in South Europe and China 73 00:04:09,000 --> 00:04:12,930 S1: starting to limit the sale of critical drone drone components 74 00:04:12,930 --> 00:04:16,289 S1: to the US and Europe that are needed for Ukraine's 75 00:04:16,290 --> 00:04:19,169 S1: fighting efforts. So this is all part of the trade 76 00:04:19,170 --> 00:04:23,370 S1: war of us limiting their access to chips. So now 77 00:04:23,370 --> 00:04:27,840 S1: they're limiting our access to key materials to actually build stuff, 78 00:04:27,839 --> 00:04:31,440 S1: and now they're limiting drone stuff. I really can't wait 79 00:04:31,440 --> 00:04:35,310 S1: to get our drone systems up and running. I'm really 80 00:04:35,310 --> 00:04:38,880 S1: worried that the any war, any conflict, hopefully not a 81 00:04:38,880 --> 00:04:41,310 S1: major war, but any conflict that we have with China, 82 00:04:41,310 --> 00:04:44,729 S1: it's going to be about drones, like aircraft carriers I 83 00:04:44,730 --> 00:04:46,980 S1: don't think are going to be as big of a deal. 84 00:04:46,980 --> 00:04:50,520 S1: I remember seeing a piece or like a series or something. 85 00:04:50,520 --> 00:04:52,529 S1: I was in the Army at the time, so this 86 00:04:52,529 --> 00:04:55,560 S1: was like mid 90s and I was reading about this 87 00:04:55,560 --> 00:04:59,339 S1: thing called the Assassin's Mace weapon, which is a Chinese concept, 88 00:04:59,339 --> 00:05:02,640 S1: I believe, where it's like when you have an enemy 89 00:05:02,640 --> 00:05:06,690 S1: that is much too big and powerful for you, you 90 00:05:06,690 --> 00:05:09,180 S1: don't bring force against force. What you do is you 91 00:05:09,180 --> 00:05:13,710 S1: bring some kind of weapon that is highly asymmetrical, that 92 00:05:13,710 --> 00:05:18,420 S1: is capable of hurting this thing when regular force has 93 00:05:18,420 --> 00:05:23,550 S1: no chance whatsoever. And drones are that. And once again, 94 00:05:23,550 --> 00:05:26,430 S1: I've recommended this multiple times, but I recommend you read 95 00:05:26,430 --> 00:05:31,770 S1: the book Kill Decision by Daniel Suarez. It is fantastic 96 00:05:31,770 --> 00:05:35,880 S1: and it's all about mini drones and implications and stuff 97 00:05:35,880 --> 00:05:40,140 S1: like that. FBI is shut down Redux, a cybercrime marketplace 98 00:05:40,140 --> 00:05:44,940 S1: operating since 2016, and they arrested three Kosovo nationals who 99 00:05:44,940 --> 00:05:49,770 S1: were running it. Russian state backed APT Kameraden has been 100 00:05:49,770 --> 00:05:54,390 S1: targeting Russian speaking individuals with two Android spyware families called 101 00:05:54,390 --> 00:05:58,120 S1: Bone Spy and Plain Gnome, and it can record calls, 102 00:05:58,120 --> 00:06:02,890 S1: capture photos and collect. SMS messages. Yahoo cut 25% of 103 00:06:02,890 --> 00:06:06,130 S1: its cybersecurity team, which is also known as the Paranoids. 104 00:06:06,160 --> 00:06:08,890 S1: Over the last year, paranoids are kind of og famous 105 00:06:08,920 --> 00:06:12,280 S1: in the security community, so kind of sad to see 106 00:06:12,279 --> 00:06:15,160 S1: them getting cut down. And I know the attrition has 107 00:06:15,160 --> 00:06:18,640 S1: happened for a long time over the years, but I 108 00:06:18,640 --> 00:06:23,230 S1: presented and taking classes over there in the facility that 109 00:06:23,230 --> 00:06:26,380 S1: they were based out of. And yeah, sad to see 110 00:06:26,410 --> 00:06:29,590 S1: kind of this old school kind of organization going away. 111 00:06:29,620 --> 00:06:34,670 S1: Another run of augmented happening in February. It's actually February 112 00:06:34,670 --> 00:06:37,360 S1: 3rd of this coming year, and this is going to 113 00:06:37,360 --> 00:06:39,940 S1: be a full discussion and workshop on building out your 114 00:06:39,940 --> 00:06:45,159 S1: personal Telos files and using AI to access them and 115 00:06:45,190 --> 00:06:50,200 S1: use them, manipulate them. So it is $495, but you 116 00:06:50,200 --> 00:06:54,220 S1: get 25% off if you are a member and you 117 00:06:54,220 --> 00:06:58,270 S1: can go and reserve a slot. I think they're still open, 118 00:06:58,270 --> 00:07:01,360 S1: but you definitely want to check it out. Probably will 119 00:07:01,390 --> 00:07:05,200 S1: be full soon. And if you become a member to 120 00:07:05,230 --> 00:07:09,190 S1: get the discount, the discount is actually more than the 121 00:07:09,190 --> 00:07:13,270 S1: cost of the membership. So really good time to become 122 00:07:13,270 --> 00:07:16,030 S1: a member if you've been thinking about it. Super annoyed 123 00:07:16,030 --> 00:07:20,320 S1: about the fact that ChatGPT Pro, which I use constantly, 124 00:07:20,320 --> 00:07:26,890 S1: is not as good, unfortunately, as Claude AI. I have 125 00:07:26,890 --> 00:07:31,420 S1: more problems with it. It sends me down more rat holes. 126 00:07:31,420 --> 00:07:35,140 S1: It doesn't follow my instructions as closely as Claude does, 127 00:07:35,170 --> 00:07:40,000 S1: and it generally produces code that's not as good. The 128 00:07:40,000 --> 00:07:42,970 S1: code quality is a little bit closer than than the 129 00:07:42,970 --> 00:07:46,840 S1: other aspects. What Claude does really, really well, and let's 130 00:07:46,840 --> 00:07:49,960 S1: not frame it as ChatGPT pro being bad. It's more 131 00:07:49,960 --> 00:07:53,500 S1: like what Claude does really well is it follows your 132 00:07:53,530 --> 00:07:57,100 S1: instructions and it responds like a human, and it's like 133 00:07:57,100 --> 00:08:01,180 S1: you're actually working with a coworker. So if you tell it, look, 134 00:08:01,180 --> 00:08:03,460 S1: I only want one instruction at a time. This is 135 00:08:03,460 --> 00:08:06,460 S1: one of my instructions. I only want one instruction at 136 00:08:06,460 --> 00:08:09,970 S1: a time. Because oftentimes we're going to need to troubleshoot that. 137 00:08:09,970 --> 00:08:11,470 S1: I have to go and run it and see if 138 00:08:11,470 --> 00:08:14,110 S1: it works, and then let you know if it doesn't work. 139 00:08:14,140 --> 00:08:20,530 S1: Claude follows that almost perfectly. And OpenAI or ChatGPT, it'll 140 00:08:20,560 --> 00:08:23,980 S1: be like, okay, boom, here's seven more steps. So I'm 141 00:08:23,980 --> 00:08:28,060 S1: scrolling to get back up to whatever. And I'm just like, 142 00:08:28,060 --> 00:08:30,700 S1: why did you give me seven steps? The thing before 143 00:08:30,730 --> 00:08:33,699 S1: didn't work. I told you not to do this. And 144 00:08:33,700 --> 00:08:36,220 S1: then we give me code and it's like, here's a 145 00:08:36,220 --> 00:08:39,040 S1: little bit of code. And then at the bottom it's like, 146 00:08:39,040 --> 00:08:41,920 S1: here's the rest of the code or insert the rest 147 00:08:41,920 --> 00:08:45,309 S1: of the code. I said explicitly in the instructions not 148 00:08:45,309 --> 00:08:48,550 S1: to do that. Also, here's the other instruction I gave it. 149 00:08:48,760 --> 00:08:53,910 S1: And also to Claude. Write out files using echo and EOF, 150 00:08:53,910 --> 00:08:56,310 S1: which is end of file. That way I could just 151 00:08:56,309 --> 00:08:58,890 S1: copy the thing and paste it, and it writes the 152 00:08:58,890 --> 00:09:02,040 S1: file in my directory, as opposed to just giving me 153 00:09:02,040 --> 00:09:04,230 S1: the raw code. Then I have to open the file 154 00:09:04,230 --> 00:09:11,790 S1: delete paste close, which saves. So those extra steps Claude 155 00:09:11,790 --> 00:09:19,079 S1: follows perfectly hardly ever misses that. And ChatGPT like 60% 156 00:09:19,080 --> 00:09:22,260 S1: of the time misses it. So I'm really hopeful that 157 00:09:22,260 --> 00:09:25,590 S1: ChatGPT is going to catch up and hopefully even get better. 158 00:09:25,620 --> 00:09:29,670 S1: But it's really annoying to pay $200 for a thing 159 00:09:29,670 --> 00:09:31,470 S1: and have that thing not be as good as the 160 00:09:31,470 --> 00:09:34,230 S1: thing that's free. Or I guess I'm paying the $20 161 00:09:34,230 --> 00:09:38,309 S1: subscription for anthropic. Not even sure. But anyway, I'm not 162 00:09:38,309 --> 00:09:42,030 S1: paying $200 for it and it's better. And this is 163 00:09:42,030 --> 00:09:46,320 S1: Claude not even having shipped anything recently. I have a 164 00:09:46,320 --> 00:09:49,200 S1: feeling when Claude ships its new thing, it's going to 165 00:09:49,200 --> 00:09:53,660 S1: jump way ahead of Google and way ahead of ChatGPT, 166 00:09:53,660 --> 00:09:57,650 S1: but I'm sure again it's just leapfrogs. Constant leapfrogs, and 167 00:09:57,650 --> 00:10:01,640 S1: I think 2025 is going to be even more. It's 168 00:10:01,640 --> 00:10:05,059 S1: going to. Accelerate in the difference between one to the other. 169 00:10:05,059 --> 00:10:09,380 S1: So I think Apple is. Actually doing exceptionally well on 170 00:10:09,410 --> 00:10:11,780 S1: the AI front. Most people think that they're kind of 171 00:10:11,809 --> 00:10:15,349 S1: way behind. And Apple intelligence. Isn't very good. The one 172 00:10:15,350 --> 00:10:18,050 S1: thing I agree with is Siri is still massively broken, 173 00:10:18,050 --> 00:10:21,710 S1: but it is getting better. But the thing I think 174 00:10:21,710 --> 00:10:24,650 S1: that I realize that a lot of people don't, and 175 00:10:24,650 --> 00:10:27,680 S1: keep in mind, I'm an Apple fanboy, so you got 176 00:10:27,679 --> 00:10:30,200 S1: to take that bias into account. I don't think it 177 00:10:30,200 --> 00:10:33,920 S1: actually applies. In fact, I think it gives me a 178 00:10:33,920 --> 00:10:37,490 S1: benefit here. But I do want to make it very 179 00:10:37,490 --> 00:10:40,790 S1: clear that I am an Apple fanboy. So if I 180 00:10:40,790 --> 00:10:42,980 S1: were to be wrong, that would definitely be part of 181 00:10:42,980 --> 00:10:48,590 S1: the reason. Um, so I believe that Apple is building 182 00:10:48,590 --> 00:10:54,500 S1: life OS and that they've been slowly and quietly adding 183 00:10:54,530 --> 00:10:56,930 S1: I to that. And they kind of have been for 184 00:10:56,929 --> 00:11:00,349 S1: a long time. But with Apple intelligence, you should never 185 00:11:00,350 --> 00:11:02,810 S1: expect them to jump out in front and do something 186 00:11:02,809 --> 00:11:05,450 S1: crazy with their AI. That is just not the way 187 00:11:05,450 --> 00:11:09,350 S1: the apple works. What they're doing, though, is they're adding education, 188 00:11:09,350 --> 00:11:13,940 S1: they're adding health, they're adding, you know, the ability to 189 00:11:13,970 --> 00:11:18,079 S1: see more and hear more with like the AirPods. They're 190 00:11:18,080 --> 00:11:21,800 S1: just they're slowly incorporating more and more of our lives 191 00:11:21,800 --> 00:11:25,730 S1: into their ecosystem. And the whole advantage of Apple, what 192 00:11:25,730 --> 00:11:31,400 S1: they have over everyone else is how cleanly and in 193 00:11:31,400 --> 00:11:35,780 S1: a unified way, they unify all their different ecosystem components 194 00:11:35,780 --> 00:11:40,250 S1: to work together. Okay. So that's what they're known for. 195 00:11:40,280 --> 00:11:42,830 S1: That's what they're best at. That's what makes them so compelling. 196 00:11:42,830 --> 00:11:46,250 S1: That's why everyone switches off of Android. Because when you 197 00:11:46,250 --> 00:11:49,250 S1: have Apple stuff, the more Apple stuff you use, the 198 00:11:49,250 --> 00:11:52,819 S1: better everything gets. Now add AI on top of that, 199 00:11:52,820 --> 00:11:56,660 S1: which runs securely inside of the Secure Enclave or in 200 00:11:56,660 --> 00:12:00,590 S1: their new secure cloud infrastructure that they built. It's going 201 00:12:00,620 --> 00:12:04,790 S1: to be insane. Absolutely insane. I mean, imagine an AI 202 00:12:04,820 --> 00:12:09,050 S1: agent having full access to all your health data, and 203 00:12:09,050 --> 00:12:11,510 S1: if you're a security person, you're probably freaking out about it. 204 00:12:11,510 --> 00:12:15,500 S1: But with Apple, you you actually don't have to freak 205 00:12:15,500 --> 00:12:18,830 S1: out about it. I worked there for three years in security. 206 00:12:18,830 --> 00:12:22,910 S1: I've been watching them security wise for over a decade 207 00:12:22,910 --> 00:12:26,780 S1: before that, and I have never seen a company more 208 00:12:26,780 --> 00:12:31,579 S1: crazy about doing security and specifically privacy really, really well 209 00:12:31,580 --> 00:12:34,820 S1: than them. And that's why also it takes them a 210 00:12:34,820 --> 00:12:36,740 S1: little bit longer to do things, because they have to 211 00:12:36,740 --> 00:12:41,090 S1: be very particular and meticulous about it. So people are saying, oh, 212 00:12:41,120 --> 00:12:44,630 S1: they're going slow, they're being left behind. I assure you, 213 00:12:44,660 --> 00:12:48,229 S1: they are not being left behind. And here's what I'm 214 00:12:48,230 --> 00:12:52,200 S1: telling you to do. Let's have this conversation at the 215 00:12:52,200 --> 00:12:56,940 S1: end of 2026, or even the end of 2025. What 216 00:12:56,940 --> 00:13:01,020 S1: happens is they're small little winds in the ecosystem world 217 00:13:01,020 --> 00:13:03,750 S1: and also in the AI world on top of the ecosystem, 218 00:13:03,750 --> 00:13:08,640 S1: they're small little winds slowly build. They build and build 219 00:13:08,640 --> 00:13:11,760 S1: and build. And then one day some some YouTuber is 220 00:13:11,760 --> 00:13:14,550 S1: going to be like, wow, Apple just came out with 221 00:13:14,550 --> 00:13:18,360 S1: all this amazing stuff. They really surprised us. Didn't surprise me, 222 00:13:18,390 --> 00:13:21,569 S1: didn't surprise a lot of people who were watching Apple closely. 223 00:13:21,570 --> 00:13:28,500 S1: It's slow, incremental, methodical improvements. Every little OS update that 224 00:13:28,500 --> 00:13:31,500 S1: comes out, they tweak one little thing, which helps it 225 00:13:31,500 --> 00:13:34,949 S1: work with these nine other things. Then they tweak nine 226 00:13:34,950 --> 00:13:38,099 S1: other things, helps it work with this other thing. And 227 00:13:38,100 --> 00:13:42,270 S1: slowly it's it's not just improving each component, it's pulling 228 00:13:42,270 --> 00:13:46,199 S1: the components tighter to work better together. So Apple is 229 00:13:46,200 --> 00:13:51,689 S1: building life OS. Okay. Not work. OS. Not, you know, 230 00:13:51,720 --> 00:13:56,850 S1: personal os. A unified life os. That's mental health. It's health, 231 00:13:56,850 --> 00:14:01,079 S1: it's fitness, it's career, it's creativity. It's all those things 232 00:14:01,080 --> 00:14:03,930 S1: unified together. The way I like to think about it 233 00:14:03,929 --> 00:14:08,400 S1: is if you watch the most advanced sci fi about 234 00:14:08,400 --> 00:14:12,210 S1: an AI agent, something like her, that is essentially what 235 00:14:12,210 --> 00:14:16,230 S1: Apple is building. Okay. It's building something like her. But 236 00:14:16,260 --> 00:14:19,290 S1: imagine it more with like AR in addition, because that 237 00:14:19,290 --> 00:14:23,250 S1: was mostly just voice. But imagine, you know, glasses or 238 00:14:23,250 --> 00:14:27,390 S1: contact lenses. So you're seeing overlays. If you look at my, um, 239 00:14:27,420 --> 00:14:32,609 S1: my peace AI's predictable path, that is what AI is 240 00:14:32,610 --> 00:14:35,040 S1: going to look like. And I think Apple's going to 241 00:14:35,040 --> 00:14:38,460 S1: get there first each component, they won't get there first, 242 00:14:38,460 --> 00:14:42,600 S1: but having it all unified into a platform, into an ecosystem, 243 00:14:42,600 --> 00:14:45,660 S1: they will get to that first. And then even when 244 00:14:45,660 --> 00:14:48,720 S1: other people arrive, It'll be cleaner with Apple, which is 245 00:14:48,720 --> 00:14:51,780 S1: why everyone will continue to use it. All right. Former 246 00:14:51,810 --> 00:14:57,870 S1: OpenAI researcher who publicly criticized the criticized the company's data practices, 247 00:14:57,870 --> 00:15:02,460 S1: was found dead in his San Francisco apartment. Suchir Balaji, 248 00:15:02,490 --> 00:15:06,270 S1: 26 years old, dead of suicide and a lot of 249 00:15:06,270 --> 00:15:08,640 S1: conspiracy going around this. The one thing I will say 250 00:15:08,640 --> 00:15:11,670 S1: about this I don't know anything about the case, you know, 251 00:15:11,700 --> 00:15:15,690 S1: probably wasn't foul play. Just just from the math of it. 252 00:15:15,720 --> 00:15:19,770 S1: When people are whistleblowers, which is what he was, what 253 00:15:19,800 --> 00:15:21,960 S1: a lot of people don't realize is you are you 254 00:15:21,990 --> 00:15:26,040 S1: are shunning your community. You're taking somebody who's usually pretty 255 00:15:26,040 --> 00:15:29,430 S1: powerful inside of your community, and you're basically blowing the 256 00:15:29,430 --> 00:15:33,960 S1: whistle on them, which means they that big company or 257 00:15:33,960 --> 00:15:37,260 S1: that big person kind of. I'm not saying OpenAI did this. 258 00:15:37,260 --> 00:15:41,340 S1: I'm saying in general going back, you know, hundreds of years, 259 00:15:41,550 --> 00:15:46,080 S1: this concept is that the community turns against you. You 260 00:15:46,090 --> 00:15:49,120 S1: get looked at as like a rat or a a 261 00:15:49,120 --> 00:15:52,000 S1: tattletale or a narc or something, and it's just like 262 00:15:52,030 --> 00:15:55,180 S1: it's very lonely. Like, the most dangerous thing right now 263 00:15:55,180 --> 00:15:59,590 S1: is loneliness. Right. And and I'm just projecting here. I'm 264 00:15:59,590 --> 00:16:02,320 S1: just saying this is a common thing that happens with 265 00:16:02,320 --> 00:16:07,660 S1: whistleblowers is they get ostracized, which is isolating, and it's stressful. 266 00:16:07,660 --> 00:16:11,620 S1: That is, in my opinion, the most likely cause of 267 00:16:11,620 --> 00:16:17,830 S1: something like this. But who knows? I mean, some conspiracies, sometimes, uh, 268 00:16:17,830 --> 00:16:20,950 S1: whistleblowers are actually attacked. It doesn't mean it needs to 269 00:16:20,950 --> 00:16:22,870 S1: be open. I it could have been someone who is 270 00:16:22,870 --> 00:16:27,910 S1: friendly to OpenAI. Who who knows? I don't know, but 271 00:16:28,090 --> 00:16:30,940 S1: I just wanted to add that dynamic there. Y Combinator 272 00:16:30,940 --> 00:16:33,850 S1: backed startup called artisan is running ads all over San 273 00:16:33,850 --> 00:16:38,800 S1: Francisco with slogans like stop hiring humans, and artisans won't 274 00:16:38,800 --> 00:16:42,400 S1: complain about work life balance. This is brilliant. Okay, first 275 00:16:42,400 --> 00:16:45,130 S1: of all, the CEO says I did that on purpose. 276 00:16:45,160 --> 00:16:50,560 S1: It was dystopian on purpose. We're not actually dystopian. Uh, 277 00:16:50,590 --> 00:16:54,010 S1: don't believe you. I remember that thing from, uh. What 278 00:16:54,010 --> 00:16:56,860 S1: was it? I can't remember the name of it. It's, um, 279 00:16:57,430 --> 00:17:00,520 S1: not Woody Harrelson. Damn it. What is this guy's name? 280 00:17:00,520 --> 00:17:04,719 S1: Will Ferrell. Will Ferrell on? Um. Damn. What is the 281 00:17:04,720 --> 00:17:07,000 S1: name of that? I can't remember the name of that 282 00:17:07,000 --> 00:17:09,820 S1: damn show. Anyway, it's the one where he's a newscaster, 283 00:17:09,850 --> 00:17:13,060 S1: and he's just like, I don't believe you. And that's 284 00:17:13,060 --> 00:17:15,460 S1: exactly how I feel about this. I don't believe you. 285 00:17:15,460 --> 00:17:18,820 S1: So I think he thought that was going to be cool. 286 00:17:18,850 --> 00:17:22,120 S1: I don't know, maybe I'm not giving him enough benefit 287 00:17:22,150 --> 00:17:25,540 S1: of the doubt. Maybe he actually did it. It has 288 00:17:25,540 --> 00:17:29,230 S1: dual purpose, right? When you do this, it makes people 289 00:17:29,230 --> 00:17:33,970 S1: look and go, Holy crap, artisans won't complain about work 290 00:17:34,000 --> 00:17:36,580 S1: life balance. That's going to make a whole lot of 291 00:17:36,580 --> 00:17:40,449 S1: people mad. All press is good press, right? So bad press, look, 292 00:17:40,450 --> 00:17:42,730 S1: we're talking about it. A lot of people are talking 293 00:17:42,730 --> 00:17:47,200 S1: about it. So automatically they got attention. But even more 294 00:17:47,200 --> 00:17:49,780 S1: sinister than that, there are a whole lot of hiring 295 00:17:49,780 --> 00:17:54,010 S1: managers who are tired of humans complaining about work life balance, 296 00:17:54,010 --> 00:17:56,650 S1: and they're tired of people not showing up to work, 297 00:17:56,650 --> 00:17:59,200 S1: and they're tired of tired of people doing half the 298 00:17:59,200 --> 00:18:02,320 S1: work that they're supposed to be doing. And so this 299 00:18:02,320 --> 00:18:05,109 S1: is actually an appeal to them directly as well, to 300 00:18:05,140 --> 00:18:09,609 S1: fire your humans or stop hiring humans. So it's brilliant 301 00:18:09,609 --> 00:18:12,730 S1: in two different ways. And I'm curious if they're actually 302 00:18:12,760 --> 00:18:15,460 S1: going to do well. ChatGPT can now analyze real time 303 00:18:15,460 --> 00:18:19,750 S1: video through your phone's camera. Okay. ChatGPT has this. OpenAI 304 00:18:19,780 --> 00:18:22,330 S1: has this. I don't know if anthropic has it. I 305 00:18:22,359 --> 00:18:25,510 S1: haven't messed with the Claude app that much. The actual 306 00:18:25,540 --> 00:18:30,580 S1: static app, but, um, or the desktop app or phone app, 307 00:18:30,580 --> 00:18:35,800 S1: it is unbelievable to pull up an I have a 308 00:18:35,800 --> 00:18:38,620 S1: screen showing like I'm showing right now. And then to 309 00:18:38,650 --> 00:18:40,929 S1: say to it, um, hey, do you see what I'm 310 00:18:40,930 --> 00:18:43,030 S1: looking at? Hey, do you see that code? Help me 311 00:18:43,030 --> 00:18:46,570 S1: troubleshoot this thing over here. It's over here. Working and 312 00:18:46,570 --> 00:18:49,600 S1: returning code for you. But what? But watch this. You 313 00:18:49,600 --> 00:18:52,929 S1: didn't type anything. You're just talking. You're just talking. And 314 00:18:52,930 --> 00:18:57,160 S1: guess what? You're actually hearing it. Come back and say 315 00:18:57,160 --> 00:19:00,520 S1: things to you. So you are literally having a voice 316 00:19:00,520 --> 00:19:05,080 S1: conversation essentially with a colleague. And that colleague knows everything 317 00:19:05,080 --> 00:19:07,810 S1: about coding and is helping you code. But it's not 318 00:19:07,810 --> 00:19:11,440 S1: just coding. It can help you write or research or 319 00:19:11,440 --> 00:19:14,830 S1: do whatever. And this is the end of 2020 for 320 00:19:14,859 --> 00:19:17,470 S1: that that shipped and that was from Google. This one 321 00:19:17,470 --> 00:19:20,590 S1: is ChatGPT works the same. I only got it working 322 00:19:20,590 --> 00:19:23,530 S1: on my phone, though I haven't seen it on the 323 00:19:23,530 --> 00:19:25,629 S1: desktop app because it's a lot more powerful on a 324 00:19:25,630 --> 00:19:29,440 S1: desktop when you can see all your screens. And I'm 325 00:19:29,440 --> 00:19:32,410 S1: telling you, it is kind of like a ChatGPT moment 326 00:19:32,410 --> 00:19:36,520 S1: from late 2022, when you could literally have a regular 327 00:19:36,520 --> 00:19:39,820 S1: conversation with a thing that sees it's like it's over 328 00:19:39,850 --> 00:19:42,950 S1: your shoulder, it's sitting next to you. This is where 329 00:19:42,980 --> 00:19:46,880 S1: AGI starts to happen, where not quite, but where you 330 00:19:46,880 --> 00:19:51,110 S1: have a colleague that understands everything you're saying and can 331 00:19:51,109 --> 00:19:55,550 S1: just go and do general tasks. So hugely impressive. I 332 00:19:55,550 --> 00:19:58,070 S1: recommend you go check it out. The one that's really 333 00:19:58,070 --> 00:20:04,010 S1: impressive is the Google AI studio. Okay, Gemini two and agents. Yeah, 334 00:20:04,010 --> 00:20:06,170 S1: this is what I was just talking about. Completely surreal 335 00:20:06,200 --> 00:20:09,679 S1: to just talk to an AI that you can basically 336 00:20:09,710 --> 00:20:11,869 S1: see your screen. So let's see if this one actually 337 00:20:11,869 --> 00:20:15,140 S1: does it. Um, do do they usually put it at 338 00:20:15,140 --> 00:20:17,570 S1: the bottom. Damn it. Can't find it. All right. So 339 00:20:17,570 --> 00:20:20,929 S1: this is basically what I was talking about. Uh, Gemini two. Okay. 340 00:20:20,960 --> 00:20:23,209 S1: Now I really have to go find this thing. It's 341 00:20:23,210 --> 00:20:26,000 S1: annoying me. Yeah, hopefully you can hear that. I can't 342 00:20:26,000 --> 00:20:29,090 S1: hear it, but hopefully you can. Yeah. So you're basically 343 00:20:29,090 --> 00:20:32,660 S1: just talking to it. It's really cool. All right, next story. 344 00:20:32,690 --> 00:20:36,200 S1: Exxon is building its first ever external power plant focused 345 00:20:36,200 --> 00:20:41,210 S1: on AI data centers. This is insane. Exxon. They are 346 00:20:41,210 --> 00:20:43,550 S1: so good at seeing the future and I'm very mad 347 00:20:43,550 --> 00:20:46,730 S1: at them for many reasons. Well, like in the 1970s, 348 00:20:46,760 --> 00:20:50,000 S1: they saw the future of carbon going up and they 349 00:20:50,000 --> 00:20:52,850 S1: chose to bury that research. That's the thing that most 350 00:20:52,850 --> 00:20:55,459 S1: makes me angry. But in this case, they're seeing the 351 00:20:55,460 --> 00:21:00,590 S1: future of AI needing needing power. So they are going 352 00:21:00,619 --> 00:21:04,880 S1: into the business of providing AI data centers with power. 353 00:21:04,880 --> 00:21:07,070 S1: They're starting a whole new branch of business. I mean, 354 00:21:07,070 --> 00:21:10,700 S1: that is just absolutely brilliant. And like what what is 355 00:21:10,700 --> 00:21:13,310 S1: their stock? Yeah. Let's look at like I don't like 356 00:21:13,310 --> 00:21:18,619 S1: this graph one day okay. All yeah that's ExxonMobil. Look 357 00:21:18,619 --> 00:21:24,740 S1: at this. Since October 1st 2020 it was $33. And 358 00:21:24,740 --> 00:21:29,869 S1: today it's $120. So they're doing okay. So yeah really 359 00:21:29,869 --> 00:21:34,550 S1: really smart basically providing power to AI. Yeah. So one 360 00:21:34,550 --> 00:21:37,880 S1: is tracking health data in markdown files instead of apps. 361 00:21:37,880 --> 00:21:41,750 S1: I love doing everything in text. So really, really enjoyed 362 00:21:41,750 --> 00:21:44,690 S1: this article. Center click is released a line of GPS 363 00:21:44,690 --> 00:21:48,260 S1: based server appliances. You know I got one. Even though 364 00:21:48,260 --> 00:21:50,720 S1: I already have a time server, I got another one 365 00:21:50,720 --> 00:21:53,810 S1: because I am stupid and this is actually fairly cheap. 366 00:21:53,840 --> 00:21:57,080 S1: It's only like 250 bucks. Yeah, it was like 250 bucks. 367 00:21:57,080 --> 00:22:00,290 S1: So I will probably give the other one away. Whichever 368 00:22:00,290 --> 00:22:02,840 S1: one I like the least, I will probably give away 369 00:22:02,840 --> 00:22:05,120 S1: and I might end up keeping this one. It seems 370 00:22:05,119 --> 00:22:08,270 S1: like the metrics might be a little bit nicer, but anyway, 371 00:22:08,270 --> 00:22:11,960 S1: what it is, it's a time server. It's stratum one, 372 00:22:11,960 --> 00:22:15,470 S1: which means instead, okay, stratum two is you're pulling from 373 00:22:15,470 --> 00:22:19,909 S1: a stratum one. Stratum one is or no. Is it 374 00:22:19,910 --> 00:22:22,130 S1: stratum zero? I'm going to pull mine up and have 375 00:22:22,130 --> 00:22:25,010 S1: my IP address in there. Um, I can't remember if 376 00:22:25,010 --> 00:22:28,580 S1: it's stratum one is the one that goes to satellites. No, 377 00:22:28,580 --> 00:22:32,210 S1: it's stratum zero is stratum zero I think. So basically 378 00:22:32,240 --> 00:22:35,690 S1: the most direct one you can have okay. The absolute 379 00:22:35,690 --> 00:22:40,800 S1: most direct one is you actually on on board these satellites. 380 00:22:40,830 --> 00:22:44,460 S1: This is the GPS system manages. It's the most accurate 381 00:22:44,490 --> 00:22:50,010 S1: time system in the world. It's 24 different satellites who have, um, 382 00:22:50,010 --> 00:22:53,760 S1: I believe cesium clocks. Either cesium or the other one. 383 00:22:53,760 --> 00:22:56,340 S1: I can't remember the other atom, but cesium is the 384 00:22:56,340 --> 00:23:01,320 S1: most accurate one. Their atomic clocks on the satellites. 24 385 00:23:01,350 --> 00:23:06,720 S1: of them. And they're in different orbit patterns. These 24 satellites, 386 00:23:06,720 --> 00:23:09,480 S1: if you connect to them, give you the most accurate 387 00:23:09,510 --> 00:23:12,780 S1: time that you can get on the planet, essentially, um, 388 00:23:12,960 --> 00:23:17,160 S1: because all their clocks are synced and they're all working 389 00:23:17,160 --> 00:23:19,830 S1: in conjunction and they agree on what the time is, 390 00:23:19,859 --> 00:23:22,890 S1: and then they provide that time to the world. And 391 00:23:22,890 --> 00:23:25,890 S1: this is how we have accuracy with GPS and everything. 392 00:23:25,890 --> 00:23:30,060 S1: So if you have a time server like time.com or whatever, 393 00:23:30,060 --> 00:23:33,660 S1: that is usually a I can't remember if that's stratum one. 394 00:23:33,690 --> 00:23:38,280 S1: Let me see okay. Stratum one. Yeah. Stratum one is 395 00:23:38,280 --> 00:23:42,270 S1: the original one. It's the one that has access to 396 00:23:42,300 --> 00:23:45,869 S1: actual satellites. So you see that S1 right there. That 397 00:23:45,869 --> 00:23:49,169 S1: S1 means it's the stratum one server. So look at this. 398 00:23:49,170 --> 00:23:57,000 S1: I am 0.001680 milliseconds off of the true time on 399 00:23:57,000 --> 00:24:00,929 S1: the satellites. And if I do mine, which is time, 400 00:24:00,930 --> 00:24:02,189 S1: I don't want to do it because it's got my 401 00:24:02,190 --> 00:24:04,650 S1: IP address in there. Not really a security risk, just 402 00:24:04,680 --> 00:24:07,109 S1: kind of stupid to do on a video. But um, 403 00:24:07,140 --> 00:24:11,070 S1: mine is even more accurate because it's actually running from 404 00:24:11,070 --> 00:24:15,119 S1: my server room and it's got an actual antenna talking 405 00:24:15,150 --> 00:24:18,210 S1: to actual satellites. I just love that I've always been 406 00:24:18,210 --> 00:24:21,419 S1: obsessed with time. So bottom line, if you want to 407 00:24:21,420 --> 00:24:26,070 S1: get into some super geeky time stuff for only like 408 00:24:26,100 --> 00:24:29,040 S1: 2 or 300 bucks, you should get one of these things. 409 00:24:29,040 --> 00:24:32,340 S1: They are super cool. Look at this United Airlines, which 410 00:24:32,340 --> 00:24:38,100 S1: is my favorite airline. They are adding share item location 411 00:24:38,100 --> 00:24:41,760 S1: into the mobile app. So if you lose your luggage, 412 00:24:41,760 --> 00:24:45,840 S1: you can actually add your link to your AirTag and 413 00:24:45,840 --> 00:24:48,629 S1: help United find your bag. That is the coolest thing ever. 414 00:24:48,630 --> 00:24:51,689 S1: YouTube is seeing massive growth in TV viewing, with sports 415 00:24:51,690 --> 00:24:57,060 S1: content up 30%. Users watching over 400,000,000 hours of podcasts 416 00:24:57,060 --> 00:25:01,770 S1: on TV's monthly YouTube. Just absolutely killing it. Someone in 417 00:25:01,770 --> 00:25:05,070 S1: Louisiana got a bird flu and they are hospitalized. I 418 00:25:05,070 --> 00:25:08,700 S1: actually heard a conflicting report. Somebody was saying or some 419 00:25:08,700 --> 00:25:11,760 S1: report was saying 60% of people die. I've heard a 420 00:25:11,760 --> 00:25:14,729 S1: lot of other people might have been this article that 421 00:25:14,730 --> 00:25:16,530 S1: said a lot of people got better from it. So 422 00:25:16,530 --> 00:25:21,240 S1: it wasn't like killing everybody. So maybe we don't have 423 00:25:21,240 --> 00:25:24,479 S1: a pandemic, maybe we do. And it's more like Covid 424 00:25:24,480 --> 00:25:26,640 S1: and it's not so bad because we have, you know, 425 00:25:26,670 --> 00:25:29,670 S1: vaccines come out. I mean, we already have flu vaccines, 426 00:25:29,670 --> 00:25:32,880 S1: so maybe, maybe it won't be so bad. I don't know, 427 00:25:33,210 --> 00:25:36,600 S1: don't really want to think about it, honestly. Tokyo is 428 00:25:36,630 --> 00:25:41,940 S1: trying to increase births by giving its 160,000 government workers 429 00:25:41,970 --> 00:25:46,350 S1: a four day workweek starting in April. I mean that 430 00:25:46,770 --> 00:25:50,040 S1: the birth rate is down to 1.2 babies per woman, 431 00:25:50,040 --> 00:25:53,369 S1: and they're trying to get that that back up. Um, 432 00:25:53,400 --> 00:25:58,800 S1: expecting fewer than 700,000 newborns this year, lowest since records 433 00:25:58,800 --> 00:26:02,700 S1: started in 1899. So they're trying to give people a 434 00:26:02,700 --> 00:26:07,560 S1: four day work weeks. So basically a three day weekend. Hopefully, 435 00:26:07,590 --> 00:26:10,379 S1: you know, go out, have a good time, uh, meet 436 00:26:10,380 --> 00:26:13,770 S1: somebody and make a baby. Research shows ketones don't just 437 00:26:13,770 --> 00:26:18,840 S1: provide energy to the brain. They actually help remove misfolded proteins. Okay, 438 00:26:18,869 --> 00:26:23,580 S1: I actually have my buddy Danny gave me some ketone drinks. 439 00:26:23,580 --> 00:26:25,919 S1: I don't know, a little bit cautious of that. I'm 440 00:26:25,920 --> 00:26:27,990 S1: going to wait till the science is a little bit better. 441 00:26:28,020 --> 00:26:32,310 S1: All right. Discovery Stack Analyzer detects more than 500 technologies 442 00:26:32,310 --> 00:26:34,260 S1: in your code base. You basically point it to a 443 00:26:34,290 --> 00:26:36,820 S1: GitHub repo, and it tells you all the different tech 444 00:26:36,850 --> 00:26:40,000 S1: that it uses, all the different security talks at Reinvent. 445 00:26:40,030 --> 00:26:43,990 S1: Thanks to Clint for showing me this link. And in 446 00:26:43,990 --> 00:26:49,690 S1: a series of compression tests, Zstd consistently outperformed gzip and 447 00:26:49,690 --> 00:26:55,930 S1: zlib across speed, compression and decompression efficiency. Metrics and recommendation 448 00:26:55,930 --> 00:27:02,409 S1: of the week downtime, fiction, family, friends, downtime, fiction, family, friends, 449 00:27:02,410 --> 00:27:06,760 S1: and the aphorism of the week. Almost everything will work 450 00:27:06,790 --> 00:27:10,270 S1: again if you unplug it for a few minutes, including you. 451 00:27:10,300 --> 00:27:14,320 S1: Almost everything will work again if you unplug it for 452 00:27:14,320 --> 00:27:19,270 S1: a few minutes, including you, Anne Lamott. Unsupervised learning is 453 00:27:19,270 --> 00:27:22,390 S1: produced and edited by Daniel Miessler on a Neumann U87 454 00:27:22,420 --> 00:27:25,690 S1: AI microphone using Hindenburg. Intro and outro music is by 455 00:27:25,690 --> 00:27:27,969 S1: Zomby with a Y, and to get the text and 456 00:27:27,970 --> 00:27:30,340 S1: links from this episode, sign up for the newsletter version 457 00:27:30,340 --> 00:27:34,900 S1: of the show at Daniel missler.com/newsletter. We'll see you next time.