WEBVTT - UL NO. 459: New Active 0-day Exploitation, AI That Sees Your Open Apps, The RebootAI Project 0:00:00.080 --> 0:00:03.410 Wiz is the number one cloud security platform based on 0:00:03.410 --> 0:00:07.490 independent customer reviews that helps security leaders and developers protect 0:00:07.490 --> 0:00:11.300 everything they build and run in the cloud. Leading organizations 0:00:11.300 --> 0:00:14.840 from the fortune 100 to high growth companies all trust 0:00:14.840 --> 0:00:18.530 wiz to innovate at the pace of cloud while staying secure. 0:00:18.560 --> 0:00:21.980 Click the link in the show notes or head to wizfolio. 0:00:21.980 --> 0:00:26.479 Slash lowercase ul to request a personalized product tour with 0:00:26.480 --> 0:00:30.890 a Wiz expert. Unsupervised Learning Podcast listeners also get a 0:00:30.890 --> 0:00:35.990 free Cloud Security Health scan by going to Wiz IOL. 0:00:36.050 --> 0:00:44.510 That's wizard Tolowercase ul. Welcome to Unsupervised Learning, a security, 0:00:44.540 --> 0:00:47.240 AI and meaning focused podcast that looks at how best 0:00:47.240 --> 0:00:50.630 to thrive as humans in a post AI world. It 0:00:50.630 --> 0:00:54.620 combines original ideas, analysis and mental models to bring not 0:00:54.620 --> 0:00:57.830 just the news, but why it matters and how to respond. 0:01:02.370 --> 0:01:05.250 All right. Welcome to unsupervised learning. This is Daniel Miessler. 0:01:05.280 --> 0:01:09.209 This is episode 459 of the podcast. All right. So 0:01:09.209 --> 0:01:12.840 had a great conversation with Rob Allen from Threatlocker about 0:01:12.840 --> 0:01:15.300 their approach to zero Trust. And it turned out to 0:01:15.300 --> 0:01:19.319 be a really interesting conversation. I always like to dive 0:01:19.319 --> 0:01:21.929 into what somebody means when they say zero trust, because 0:01:21.930 --> 0:01:26.850 it could be anywhere from total horsecrap to pretty decent, 0:01:26.850 --> 0:01:31.260 and I really liked Rob's approach in responses there, so 0:01:31.260 --> 0:01:34.620 definitely check that out. Got you all membership. Black Friday 0:01:34.650 --> 0:01:39.809 sale just started 20% off the first month, so or 0:01:39.810 --> 0:01:43.770 the first year. So go check that out. And jumping 0:01:43.770 --> 0:01:46.920 right into security. Actually, first of all, um, did a 0:01:46.920 --> 0:01:50.250 huge upgrade to my ubiquiti gear now moving towards a 0:01:50.250 --> 0:01:53.820 ten gigabit world, which I talked a bit about last week, 0:01:53.820 --> 0:01:59.130 and going to Saudi to speak at Blackhat EMEA. Mia 0:01:59.160 --> 0:02:04.620 Blackhat Mia. Yeah Saudi Arabia next week. Going to be awesome. 0:02:04.650 --> 0:02:07.680 All right. Jumping into security. This one didn't get near 0:02:07.680 --> 0:02:10.530 enough coverage. I don't think ChatGPT has basically a new 0:02:10.530 --> 0:02:14.490 feature that could read code from Mac OS. So apps 0:02:14.490 --> 0:02:19.620 like VSCode, Xcode, terminal, it could basically read the app 0:02:19.620 --> 0:02:23.430 and understand what it's saying and doing, and then give 0:02:23.430 --> 0:02:26.070 you analysis based on that. And I think this is 0:02:26.070 --> 0:02:28.740 getting closer to what a whole bunch of people in 0:02:28.740 --> 0:02:32.970 AI are actually working on, which is what I believe 0:02:33.000 --> 0:02:37.020 to be like an inevitable future, which is you can 0:02:37.020 --> 0:02:39.420 see all the screens, right? I'm looking at three screens 0:02:39.419 --> 0:02:42.930 right now. Ideally, my eye would be looking from inside 0:02:42.930 --> 0:02:45.420 the computer because it would have like agent access, like 0:02:45.419 --> 0:02:47.820 we already talked about. But also I wanted to be 0:02:47.820 --> 0:02:50.250 able to see my surroundings. I wanted to be able 0:02:50.250 --> 0:02:52.200 to see my cameras. I wanted to be able to 0:02:52.200 --> 0:02:55.440 see around me, behind me, especially behind me. If you're 0:02:55.440 --> 0:02:59.220 out in the city walking around and also cameras or 0:02:59.220 --> 0:03:04.240 also Microphones and other types of sensor, but mostly visual 0:03:04.240 --> 0:03:08.290 and audio. And that way it can be consuming, you know, 0:03:08.320 --> 0:03:12.519 the environment all the time, constantly parsing what's going on 0:03:12.520 --> 0:03:16.870 and constantly doing analysis and giving you like, views inside 0:03:16.870 --> 0:03:20.620 of a heads up display or, you know, glasses or 0:03:20.650 --> 0:03:23.919 lenses or however that interface is going to be, depending 0:03:23.919 --> 0:03:26.560 on how far the tech is. But the most important 0:03:26.560 --> 0:03:28.870 thing is it has to be seeing what you're seeing 0:03:28.870 --> 0:03:31.270 and even more than what you're seeing. Big problem right 0:03:31.270 --> 0:03:35.050 now is the on ramp to using AI. It's not 0:03:35.050 --> 0:03:37.330 a problem of the models themselves. The models are getting 0:03:37.330 --> 0:03:39.310 really smart. The problem is how do you get stuff 0:03:39.310 --> 0:03:41.350 into the model and how do you get it out. 0:03:41.350 --> 0:03:44.380 That's the problem. So this type of thing is a 0:03:44.380 --> 0:03:47.890 really huge advance because it's able to see inside your apps. 0:03:47.890 --> 0:03:51.400 And I'm really excited about this future. I've talked about 0:03:51.400 --> 0:03:53.860 it a lot. I've talked about it in the predictable 0:03:53.860 --> 0:03:57.550 path of AI. It's just that I'm not really going 0:03:57.580 --> 0:04:00.100 to trust this. There's already a couple of companies who 0:04:00.100 --> 0:04:03.680 will watch your entire screen and will upload that stuff, 0:04:03.680 --> 0:04:05.960 and they will do the parsing and they will send 0:04:05.960 --> 0:04:10.070 you back some really cool stuff. Like there's one particular company, 0:04:10.220 --> 0:04:12.620 I think they changed their name. It's called rewind. Or 0:04:12.620 --> 0:04:16.070 maybe rewind is the new name. But anyway, I signed 0:04:16.070 --> 0:04:20.000 up for their new, uh, gadget, which should be coming soon. 0:04:20.000 --> 0:04:22.609 I did not give them access to my whole computer 0:04:22.610 --> 0:04:26.000 because I could have the calendar open or a message 0:04:26.000 --> 0:04:28.789 or something, and it's like you could be doxing your friends. 0:04:28.790 --> 0:04:31.099 You don't know what's going to happen. Most importantly, you 0:04:31.100 --> 0:04:35.030 don't know how secure that startup is. And having consulted 0:04:35.029 --> 0:04:40.430 for startups in security specifically for decades, I'm not sending 0:04:40.430 --> 0:04:43.580 my data to those startups. I basically only trust like 0:04:43.610 --> 0:04:46.969 one company that much, both on the security side and 0:04:46.970 --> 0:04:50.000 also on the privacy side, and just how seriously they 0:04:50.000 --> 0:04:53.420 take this stuff, combined with a lack of conflicts of interest. 0:04:53.420 --> 0:04:57.740 And that's Apple. I would say Google is just as competent, 0:04:57.770 --> 0:05:01.560 maybe even more competent on the security side, I would say. 0:05:01.589 --> 0:05:05.400 But the problem is they also make their money off 0:05:05.400 --> 0:05:08.100 of ads or a lot of their money off of ads. 0:05:08.100 --> 0:05:11.430 So I just don't like the idea of like them 0:05:11.430 --> 0:05:13.650 having all that data and then being able to sell 0:05:13.650 --> 0:05:16.380 it or sell things to me. I don't like the 0:05:16.380 --> 0:05:19.560 conflict of interest there, even though I like the security. 0:05:19.560 --> 0:05:22.650 So that's why I mostly only trust two companies and 0:05:22.650 --> 0:05:26.100 mostly only Apple. But this is where this is going. 0:05:26.100 --> 0:05:31.650 Continuous monitoring all the time with as many sensors as possible. Front, back, 0:05:31.650 --> 0:05:35.010 inside your apps, inside your phone. That's where everyone is 0:05:35.010 --> 0:05:37.500 rushing to. Even if they don't know that they're rushing 0:05:37.529 --> 0:05:40.770 to that location. And that's why I think Google and 0:05:40.770 --> 0:05:45.000 Apple have a massive advantage because they have the device, right? 0:05:45.029 --> 0:05:49.410 I'm on Apple for phone, I'm on Apple for Mac 0:05:49.410 --> 0:05:54.479 and desktop and laptop. So because they are the OS, 0:05:54.510 --> 0:05:58.140 they have a massive advantage here. Okay. Palo Alto Networks 0:05:58.140 --> 0:06:01.620 has released indicators of compromise IOCs for a new zero 0:06:01.650 --> 0:06:06.240 day vulnerability affecting firewalls. VMware confirmed that threat actors are 0:06:06.240 --> 0:06:10.589 exploiting two vCenter server vulnerabilities, and ones like a 9.8 0:06:10.620 --> 0:06:12.719 and the other ones like a seven something I and 0:06:12.720 --> 0:06:15.810 Tech Anthropic has a new prompt improver that takes the 0:06:15.810 --> 0:06:18.510 given prompt and writes a better one. Really cool to 0:06:18.510 --> 0:06:21.570 see people getting in on this. This is part of 0:06:21.570 --> 0:06:25.860 the overall ecosystem, right? Um, and we got another one here. Uh, 0:06:25.860 --> 0:06:29.910 OpenAI might launch an AI agent tool called Operator in January, 0:06:29.910 --> 0:06:33.090 and it will compete with anthropic computer use. I think 0:06:33.120 --> 0:06:35.430 agents is going to be the biggest thing that happens 0:06:35.430 --> 0:06:38.640 in 2025 for AI. And this is part of a 0:06:38.640 --> 0:06:41.250 bigger trend that I've been talking about, where it's more 0:06:41.250 --> 0:06:44.130 about the ecosystem. Okay, so if I if I actually 0:06:44.130 --> 0:06:48.719 click and open this thing here, it's not just about 0:06:48.720 --> 0:06:52.830 the models. Okay. So the big pieces that I think 0:06:52.830 --> 0:06:56.789 are part of these four pillars of an AI ecosystem, 0:06:56.790 --> 0:07:00.580 the model itself, the post training of the model internal 0:07:00.610 --> 0:07:06.729 tooling and agent functionality. So agent or model self-explanatory. Post-training 0:07:06.760 --> 0:07:10.120 a set of highly proprietary tricks that magnify the overall 0:07:10.120 --> 0:07:12.940 quality of the Raw model. Way to think of this 0:07:12.940 --> 0:07:14.770 is to say that it's a way to connect model 0:07:14.770 --> 0:07:18.730 weights to human problems. Okay. Internal tooling. All right. So 0:07:18.730 --> 0:07:23.110 look at this list. High quality APIs, larger context sizes, 0:07:23.140 --> 0:07:29.830 simple fine tuning. Haystack performance. Strict output control. External tooling 0:07:29.830 --> 0:07:34.510 like function calling. Trust and safety features. Mobile apps. Prompt testing. 0:07:34.510 --> 0:07:38.710 Voice mode and apps OS integration. Integrations with things like 0:07:38.710 --> 0:07:43.180 make Zapier end to end and things like caching mode. 0:07:43.210 --> 0:07:46.300 These are like all the internal tooling stuff that just 0:07:46.300 --> 0:07:49.000 makes it easier. Think of it this way the problem 0:07:49.000 --> 0:07:51.790 isn't the models. The problem is the on ramps onto 0:07:51.790 --> 0:07:54.489 the models and the output out of them back into 0:07:54.490 --> 0:07:57.850 your life. Okay, we are humans. We have human problems. 0:07:57.850 --> 0:08:00.680 We have business problems. We have personal problems. whatever we 0:08:00.680 --> 0:08:03.890 need to get that content, the content of that problem 0:08:03.890 --> 0:08:07.100 into an AI and then back out into our lives, 0:08:07.100 --> 0:08:10.040 into our actual brains, into the real world. That's what 0:08:10.040 --> 0:08:13.310 this internal tooling piece is, right? Because if you can't 0:08:13.310 --> 0:08:16.760 do this, well, then it doesn't matter if your model 0:08:16.760 --> 0:08:21.920 is 13% better on some random benchmark, right? So the 0:08:21.920 --> 0:08:26.480 models have to get better, but not as important as 0:08:26.510 --> 0:08:30.590 actually getting the interfaces to the models better. So uh 0:08:30.620 --> 0:08:33.320 oh yeah. And the next one relevant to the next 0:08:33.320 --> 0:08:38.000 story there is agents. Right. So an AI component that 0:08:38.000 --> 0:08:40.429 interprets instructions and takes on more of the work in 0:08:40.429 --> 0:08:46.490 total AI workflows than just LLM response, for example, executing functions, 0:08:46.490 --> 0:08:50.569 performing data lookups, etc. before passing on results. And I 0:08:50.570 --> 0:08:53.660 actually have a improved version of that, which I won't 0:08:53.690 --> 0:08:56.390 I won't go find it, but it's, uh, it's updated 0:08:56.390 --> 0:09:02.030 on the AGI definition inside of raid the real world 0:09:02.059 --> 0:09:05.150 I definitions if you want to go check that out. Okay, 0:09:05.179 --> 0:09:09.500 so Sam Altman and Arianna Huffington have a thrive AI 0:09:09.530 --> 0:09:15.380 health company and looking at doing personalized advice on sleep, food, 0:09:15.380 --> 0:09:19.850 fitness and more, Google.org is putting 20 million in cash 0:09:19.850 --> 0:09:22.880 and 2 million in cloud credits into a new initiative 0:09:22.910 --> 0:09:27.439 to help researchers use AI for scientific breakthroughs. One of 0:09:27.440 --> 0:09:31.730 the most important things that I think I could possibly 0:09:31.730 --> 0:09:41.119 do is actively going to just invent new things, make new, uh, research, right? 0:09:41.150 --> 0:09:45.020 Discover new things. Right? You take the smartest people in 0:09:45.020 --> 0:09:48.140 the world who who are capable of doing this, and 0:09:48.140 --> 0:09:50.840 there's very few of them. So if we can actually 0:09:50.840 --> 0:09:54.710 scale that, that's where the real takeoff starts to happen. 0:09:54.740 --> 0:09:58.910 Apple's M4 Max CPU transcribes audio twice as fast as 0:09:59.059 --> 0:10:05.849 Nvidia's RTX A5000 GPU while using significantly less power. I 0:10:05.850 --> 0:10:09.150 really want to get one of these clusters, like a 0:10:09.150 --> 0:10:15.120 m4 Mac mini cluster would be super cool. But yeah, 0:10:15.150 --> 0:10:16.860 a lot of people have been asking me, should I 0:10:16.860 --> 0:10:19.410 do the cluster thing with smaller boxes? Or should I 0:10:19.410 --> 0:10:22.140 just get one big rig? I feel like the next 0:10:22.140 --> 0:10:26.100 generation or the current next the current generation in like 0:10:26.130 --> 0:10:28.950 another year or so, it's still going to be better 0:10:28.950 --> 0:10:30.900 to have one big box. It really depends on what 0:10:30.900 --> 0:10:33.150 you need it for, but I would say we're not 0:10:33.150 --> 0:10:38.400 quite there yet. With the cluster of smaller boxes, this 0:10:38.520 --> 0:10:41.939 exo lab and things like that, they're a little bit experimental. 0:10:42.030 --> 0:10:45.569 The best thing to do still is to buy a box. 0:10:45.570 --> 0:10:49.260 I bought a pre-made one and it's fantastic. It's got 0:10:49.260 --> 0:10:52.530 two 49 seconds in it. It's got tons of memory. 0:10:52.679 --> 0:10:56.370 It's very, very fast. And it's just like plug and play. 0:10:56.370 --> 0:11:01.750 Easy to do. I would say if this clustering technology 0:11:01.750 --> 0:11:04.929 gets better and better, and there's more and more devices 0:11:04.929 --> 0:11:07.030 like Mac minis and stuff like that that you could 0:11:07.030 --> 0:11:10.569 piece together. Then it starts to be a compelling alternative 0:11:10.570 --> 0:11:15.130 to actually buying a giant AI box. But until then, 0:11:15.130 --> 0:11:18.520 I think the giant box is probably still going to 0:11:18.520 --> 0:11:22.720 be better. Okay. iOS 18.2 is music recognition feature. Now 0:11:22.750 --> 0:11:25.719 logs where you were when you actually heard the song. 0:11:25.720 --> 0:11:28.990 So that's part of the metadata. Now pharma stocks have crashed. 0:11:28.990 --> 0:11:32.050 This is under the humans label. Pharma stocks have crashed 0:11:32.050 --> 0:11:36.310 after RFK Jr was announced to be taking over Health 0:11:36.309 --> 0:11:39.940 and Human Services. Moderna is down like 40%. Yeah, I 0:11:39.940 --> 0:11:43.449 don't know what this is currently at. Uh, down 562, 0:11:43.480 --> 0:11:45.160 but that's for the day. So if we go to 0:11:45.190 --> 0:11:52.600 three months, down 57% in the last three months, Moderna. Okay. 0:11:52.929 --> 0:11:55.630 They're they're one of the people who came out with 0:11:55.630 --> 0:12:01.130 the best vaccine for Covid, like 57% because of RFK. 0:12:01.160 --> 0:12:04.219 I almost feel like this is a buy opportunity. How 0:12:04.220 --> 0:12:06.979 could this not be a buy opportunity? It's not like 0:12:06.980 --> 0:12:11.060 they suddenly stopped being able to make things. And I 0:12:11.059 --> 0:12:14.090 just don't believe that people are going to let RFK 0:12:14.090 --> 0:12:18.110 just destroy these companies. I just don't see that happening. 0:12:18.140 --> 0:12:20.690 Hopefully they're going to be able to figure it out, 0:12:20.690 --> 0:12:23.540 because RFK is going to do some cool stuff. Like 0:12:23.540 --> 0:12:25.460 he's right about a lot of stuff. That's what's most 0:12:25.460 --> 0:12:28.040 scary about a lot of these people. They're so right 0:12:28.040 --> 0:12:31.400 about so many things, and the things they're wrong about 0:12:31.400 --> 0:12:34.640 are mixed in with it. Right? And that's not just RFK. 0:12:34.670 --> 0:12:38.540 That's a lot of people, including probably myself. So yeah, 0:12:38.570 --> 0:12:44.300 I don't see how it doesn't bounce back from a 60% 0:12:44.300 --> 0:12:49.010 cut from three months. Right. Personally, I'm not investing, but 0:12:49.010 --> 0:12:51.470 I think it would probably be smart, at least in 0:12:51.470 --> 0:12:56.390 the long term. Netflix had a record 65 million concurrent 0:12:56.390 --> 0:13:00.200 streams during Mike Tyson versus Jake Paul fight. it did 0:13:00.200 --> 0:13:02.599 have a bunch of connection problems though. Everyone I know 0:13:02.600 --> 0:13:06.320 at Netflix got massively spammed with like, all their friends 0:13:06.320 --> 0:13:08.360 texting and was like, hey, what's going on? It's like, yeah, 0:13:08.360 --> 0:13:11.720 you think I'm in charge of actual throughput during a 0:13:11.720 --> 0:13:15.140 Tyson fight? Like, stop texting me. New study shows that 0:13:15.140 --> 0:13:18.500 treating bullying as a collective issue rather than an individual one, 0:13:18.500 --> 0:13:22.400 can significantly reduce its occurrence in primary schools. I love 0:13:22.400 --> 0:13:26.000 the concept. It's kind of like how the johns get 0:13:26.000 --> 0:13:29.750 in trouble in some European countries for prostitution instead of 0:13:29.750 --> 0:13:34.520 the prostitute, because it's like the ecosystem that's the actual problem, right? 0:13:34.610 --> 0:13:38.059 So I like this idea of with bullying, I like 0:13:38.090 --> 0:13:40.699 the idea of shaming the people around who didn't say 0:13:40.700 --> 0:13:42.800 anything or do anything. And of course, we have to 0:13:42.830 --> 0:13:44.810 be careful about shaming. We're talking about kids here in 0:13:44.809 --> 0:13:48.470 a lot of cases or in most cases. But like, 0:13:48.530 --> 0:13:50.990 I feel like all the marketing needs to be heading 0:13:50.990 --> 0:13:53.359 in the direction of like if you are one of 0:13:53.360 --> 0:13:56.270 these people watching and not saying anything, not reporting it, 0:13:56.300 --> 0:13:59.880 not intervening like you don't want to be unsafe or whatever, 0:13:59.880 --> 0:14:02.339 but you could go report it. You could do something 0:14:02.370 --> 0:14:06.449 to prevent this from happening again. And if you're not 0:14:06.480 --> 0:14:09.780 something like you are the bully or you are enabling 0:14:09.780 --> 0:14:11.699 this and this is really bad, and I know this 0:14:11.700 --> 0:14:14.430 is already an aspect of like a lot of these programs, 0:14:14.429 --> 0:14:19.860 but I think it could be significantly magnified. Ideas reboot I, 0:14:20.070 --> 0:14:23.609 I absolutely love this one. Absolutely love this one. I 0:14:23.610 --> 0:14:25.710 can't remember where I got this idea. I think it's 0:14:25.710 --> 0:14:28.680 been hit me from multiple places, but I want to 0:14:28.680 --> 0:14:32.040 build a local AI that can run offline. Oh, I 0:14:32.040 --> 0:14:34.440 know where I first heard this idea. It was actually 0:14:34.440 --> 0:14:38.190 from Joseph Thacker, like a year and a half ago. 0:14:38.220 --> 0:14:40.110 He's like, oh, I just want a thing that I 0:14:40.110 --> 0:14:43.920 can use offline. But, um, more recently I got this 0:14:43.920 --> 0:14:47.190 from somewhere else. I can't was it? Was it X 0:14:47.190 --> 0:14:51.870 or Instagram? I don't know, but the idea is let's 0:14:51.870 --> 0:14:54.300 say all the power is out, or let's say the 0:14:54.300 --> 0:14:58.290 internet is out and let's say, um, oh, I know 0:14:58.290 --> 0:15:03.670 what it was. The initial conversation I was having with Joseph, 0:15:03.700 --> 0:15:05.890 and he actually brought it to me. It was his idea. 0:15:05.890 --> 0:15:09.460 He was like, what if you could go back in time? 0:15:09.490 --> 0:15:12.850 What could you bring with you to actually move society 0:15:12.880 --> 0:15:15.790 forward or something like that? And it's like, I think 0:15:15.790 --> 0:15:18.100 about this a lot way more than the Roman Empire. 0:15:18.100 --> 0:15:20.920 I think a lot about like, could I actually move 0:15:20.950 --> 0:15:23.560 science forward if I was put 200 years in the 0:15:23.560 --> 0:15:26.290 past or 2000 years in the past, what could I 0:15:26.290 --> 0:15:28.840 actually offer to them? I think someone makes a joke 0:15:28.840 --> 0:15:32.860 about this in current stand up comedy. It's like, yeah, oh, 0:15:32.890 --> 0:15:37.870 it's Nate Bathgate, Bargatze, whatever his name is. Um, it's 0:15:37.870 --> 0:15:41.350 just like, yeah, um, there's going to be phones and 0:15:41.380 --> 0:15:44.770 they're going to have satellite technology. Oh, really? What's the satellite? Well, 0:15:44.770 --> 0:15:47.020 it's this thing that goes around. How do you make one? 0:15:47.050 --> 0:15:49.720 What does it do? You're saying there's thousands of satellites 0:15:49.720 --> 0:15:52.690 in the air and the Earth is actually round? Like, 0:15:52.720 --> 0:15:56.260 can you prove anything about this? No, I can't. So 0:15:56.710 --> 0:16:01.960 we we don't know how any of this is working, right? And, uh, 0:16:02.440 --> 0:16:05.380 and we can't describe it to anyone. And more importantly, 0:16:05.380 --> 0:16:08.050 if you lose the internet and there's like, say, a 0:16:08.050 --> 0:16:11.170 meteor hits or whatever, I'm not going to go into 0:16:11.200 --> 0:16:13.630 negativity right now too early in the show for that. 0:16:13.630 --> 0:16:17.140 But let's say something bad happens and you are stuck 0:16:17.140 --> 0:16:19.600 in your house and there's no internet. Let's say you're 0:16:19.600 --> 0:16:22.660 not dying of hunger or thirst, but you don't have 0:16:22.660 --> 0:16:28.120 the internet. Okay, so watch this. Tourniquets, sterilizing water, building shelters, 0:16:28.120 --> 0:16:33.100 identifying edible plants. Which mushroom is actually will kill you. 0:16:33.100 --> 0:16:36.100 And which mushroom can you put on a salad? These 0:16:36.100 --> 0:16:40.000 are important distinctions to make. So check this out. What 0:16:40.000 --> 0:16:43.180 if you had an AI that ran? We got to 0:16:43.180 --> 0:16:46.120 assume solar power, right? Or maybe the grid actually works, 0:16:46.120 --> 0:16:48.850 but there's no internet. Whatever. Just work with me. You 0:16:48.850 --> 0:16:51.670 can show it. Pictures. You could take pictures, or you 0:16:51.670 --> 0:16:54.460 could show it an actual live plant or whatever. All 0:16:54.460 --> 0:16:56.890 you have to do is show this AI the particular 0:16:56.890 --> 0:16:59.470 thing or you describe, hey, I need a shelter that 0:16:59.480 --> 0:17:02.270 does this. I've got this much water. I got it 0:17:02.270 --> 0:17:04.790 from this kind of creek. I live in this kind 0:17:04.790 --> 0:17:07.550 of area. What kind of toxins are likely to be 0:17:07.550 --> 0:17:09.530 in it? How can I get those toxins out? What 0:17:09.530 --> 0:17:12.530 kind of filter can I build? Will these iodine tablets 0:17:12.560 --> 0:17:16.310 actually work? I have these symptoms. Which drugs should I take? 0:17:16.340 --> 0:17:20.300 You have a local model running. All it needs is power. 0:17:20.300 --> 0:17:23.330 It doesn't need the internet. It doesn't need OpenAI. Think 0:17:23.330 --> 0:17:27.109 of how much knowledge is inside of a llama two 0:17:27.140 --> 0:17:30.590 or a llama three, or a llama four, or whatever. 0:17:30.590 --> 0:17:33.409 Local model. And as long as you have power to 0:17:33.440 --> 0:17:35.990 run it and it can actually see and you can 0:17:35.990 --> 0:17:39.109 type to it, it can answer tons of stuff that 0:17:39.109 --> 0:17:43.550 can actually keep you alive. Even better, let's go a 0:17:43.550 --> 0:17:46.550 little sci fi, okay. It's just you and 10,000 other 0:17:46.550 --> 0:17:50.480 people and the rest of the, let's say, the rest 0:17:50.480 --> 0:17:52.730 of the planet got hit by a meteor or whatever. 0:17:52.730 --> 0:17:57.260 You have to rebuild all of society. Okay? Irrigation. How 0:17:57.290 --> 0:18:00.570 to how does the stoplight work? How does a combustion 0:18:00.570 --> 0:18:03.419 engine work? What kind of metal do you need to 0:18:03.420 --> 0:18:06.629 build in order to make a combustion engine? What are 0:18:06.630 --> 0:18:12.300 all these different alloys? You can bootstrap a society with 0:18:12.300 --> 0:18:15.840 one box. Isn't that crazy? You can bootstrap a society 0:18:15.840 --> 0:18:19.650 with one box that you can have, like sitting next 0:18:19.650 --> 0:18:22.950 to your NAS over here. You just need power. You 0:18:22.950 --> 0:18:24.810 need power. Well, I guess you need the peripherals to 0:18:24.840 --> 0:18:27.270 be able to talk to it and everything, but you 0:18:27.270 --> 0:18:32.310 don't need that much. It's way more impressive than trying 0:18:32.310 --> 0:18:34.770 to collect all of Wikipedia. I mean, that was the 0:18:34.770 --> 0:18:37.140 other model, right? You just download Wikipedia and you have 0:18:37.140 --> 0:18:39.900 a bunch of things you can look for. Not nearly 0:18:39.900 --> 0:18:43.020 as good as a chatbot that you can ask questions. 0:18:43.020 --> 0:18:47.250 And if it's visually oriented like you could just these 0:18:47.250 --> 0:18:50.070 new models are getting amazing, right? It's like, show me 0:18:50.070 --> 0:18:53.520 the design, show me a picture of it, draw me 0:18:53.520 --> 0:18:57.930 a picture, design me whatever a compound that I could 0:18:57.930 --> 0:19:00.810 defend my plants with because people are probably going to 0:19:00.810 --> 0:19:04.620 come and get us whatever. Yeah. I don't have any guns. 0:19:04.770 --> 0:19:07.980 How do I defend myself from these roving people who 0:19:07.980 --> 0:19:09.600 are going to come try to get our stuff? You 0:19:09.600 --> 0:19:12.030 could ask it anything. Ideally, it would be uncensored for 0:19:12.060 --> 0:19:13.830 that reason. Right. You want to be able to ask 0:19:13.830 --> 0:19:17.310 security questions. So all that to say, I'm going to 0:19:17.310 --> 0:19:21.360 do a project called reboot I with reboot being like 0:19:21.359 --> 0:19:26.760 Reboot Society or whatever, an offline oracle for emergencies. Um, 0:19:26.760 --> 0:19:29.940 now it's such a cool idea. I'm sure a million 0:19:29.940 --> 0:19:32.340 people have already had it, so they're probably already working 0:19:32.340 --> 0:19:34.770 on it. So if anyone is hearing this or they're 0:19:34.770 --> 0:19:38.700 reading this, um, send me a link and I'll just like, 0:19:38.730 --> 0:19:42.330 go buy one and or procure one. If not, and 0:19:42.330 --> 0:19:45.030 someone wants to help build one, I'm going to go 0:19:45.060 --> 0:19:47.790 build this. I want this running in my house. I've 0:19:47.790 --> 0:19:50.730 already got solar. I've got lots of ways to gather 0:19:50.730 --> 0:19:53.520 energy and store it in batteries, and it would be 0:19:53.520 --> 0:19:56.580 super nice without internet to be able to ask all 0:19:56.580 --> 0:20:02.260 those sorts of questions. So cool. discovery. Cloudflare's robots.txt file. 0:20:02.260 --> 0:20:06.850 It's a mix of Ascii art and directives for web crawlers. Obviously, 0:20:06.850 --> 0:20:12.340 that allows Twitter bot and demon based website preview to 0:20:12.369 --> 0:20:16.419 access specific pages blocks many other from accessing. Actually, you 0:20:16.420 --> 0:20:19.810 know what I'm going to do? Let me just. Oh bam! 0:20:19.930 --> 0:20:22.930 Look at that. This is why you do a video podcast. 0:20:22.960 --> 0:20:25.780 You can just zoom in to stuff. Look at this. 0:20:25.869 --> 0:20:29.020 Why are they. That's a lot of that's that's a 0:20:29.020 --> 0:20:31.300 lot of bandwidth depending on how many people are pulling it. 0:20:31.330 --> 0:20:34.270 It's a lot of bandwidth. Our tree is a redwood. Cool. 0:20:34.270 --> 0:20:36.970 This is cool though. Look at all these allows I'm 0:20:36.970 --> 0:20:39.010 looking at the scroll bar. How long is this thing. 0:20:39.040 --> 0:20:43.630 Okay okay. It's a lot of disallows. Are those languages. No. 0:20:43.630 --> 0:20:47.290 Those are directories. Yeah. These are all directories okay. Why 0:20:47.290 --> 0:20:49.060 do they have these. Why do we need to say 0:20:49.060 --> 0:20:52.090 that okay. And then we got our sitemaps okay. And 0:20:52.090 --> 0:20:57.580 more Ascii art. Yeah. Cool I like it. Interesting. Managing 0:20:57.580 --> 0:21:00.530 high performers a guide on how to effectively manage high 0:21:00.530 --> 0:21:05.960 performing employees and Eon's secure shoelace knot is the best shoelace. 0:21:05.960 --> 0:21:10.760 Not that I know of. And no, there's no sponsorship 0:21:10.760 --> 0:21:13.940 because that would be silly because no such thing exists. 0:21:13.940 --> 0:21:19.580 It's a shoelace knot. I actually tie this for my sneakers, actually, 0:21:19.580 --> 0:21:22.550 the ones I'm wearing right now. Wish I could put 0:21:22.550 --> 0:21:25.850 my foot up there. That would hurt. Uh, and I 0:21:25.850 --> 0:21:29.300 mostly leave them that way. I literally, these are common projects. 0:21:29.300 --> 0:21:31.909 I literally just slide my foot into there. Sometimes I 0:21:31.910 --> 0:21:34.730 use a shoehorn to do that, and sometimes I just 0:21:34.730 --> 0:21:39.169 use my finger. But I tied these with this knot 0:21:39.170 --> 0:21:42.620 and it's the coolest looking knot. It's the most secure knot. 0:21:42.650 --> 0:21:45.560 It's awesome. And because we're on video, I'm going to 0:21:45.560 --> 0:21:48.230 go show you this knot. Look at this thing. So 0:21:48.230 --> 0:21:53.840 you make two thingies. Two separate thingies. Then you cross 0:21:53.840 --> 0:21:57.650 the front one over the left one, the right one 0:21:57.650 --> 0:22:02.160 goes over the the left one. And you do this twice. 0:22:02.160 --> 0:22:06.210 You do over and under for both of them. And 0:22:06.210 --> 0:22:09.420 then you pull the thing and you end up with that. 0:22:09.420 --> 0:22:12.450 You end up with this looking thing right here. And 0:22:12.450 --> 0:22:17.070 it's super symmetrical and flat. It's not like twisted and 0:22:17.070 --> 0:22:20.369 trying to go up. And it's got like this really 0:22:20.369 --> 0:22:23.399 cool looking box not in the middle. And that's the 0:22:23.400 --> 0:22:27.210 most you've ever heard about a shoelace. Not on a 0:22:27.210 --> 0:22:30.960 podcast probably. All right. Recommendation of the week. Check out 0:22:30.960 --> 0:22:34.200 the aphorism of the week below. So we'll jump there. 0:22:34.290 --> 0:22:37.139 If you hit a wrong note, it's the next note 0:22:37.140 --> 0:22:39.960 you play that determines if it's good or bad. If 0:22:39.960 --> 0:22:42.210 you hit a wrong note, it's the next note you 0:22:42.210 --> 0:22:46.109 play that determines if it's good or bad. Okay, so 0:22:46.109 --> 0:22:48.840 that was the aphorism of the week. Now focus your 0:22:48.840 --> 0:22:52.379 efforts on being flexible after wrong notes, as opposed to 0:22:52.410 --> 0:22:55.200 being able to play perfect notes all the time. That's 0:22:55.200 --> 0:22:58.170 my recommendation of the week. 2020 five inches the next 0:22:58.170 --> 0:23:01.410 few years are likely to be so crazy that we 0:23:01.410 --> 0:23:04.590 won't be able to plan or play the right notes, 0:23:04.590 --> 0:23:07.680 so we just have to be good playing the next 0:23:07.680 --> 0:23:11.070 note afterwards. And I will read the aphorism of the 0:23:11.070 --> 0:23:14.310 week again, because that's what we do. If you hit 0:23:14.340 --> 0:23:16.770 a wrong note, it's the next note you play that 0:23:16.770 --> 0:23:19.800 determines if it's good or bad. If you hit a 0:23:19.800 --> 0:23:23.310 wrong note, it's the next note you play that determines 0:23:23.310 --> 0:23:27.719 if it's good or bad. Miles Davis. Unsupervised learning is 0:23:27.720 --> 0:23:30.870 produced and edited by Daniel Miessler on a Neumann U87 0:23:30.900 --> 0:23:34.980 AI microphone using Hindenburg. Intro and outro music is by 0:23:34.980 --> 0:23:38.159 zombie with a Y. And to get the text and 0:23:38.160 --> 0:23:40.530 links from this episode, sign up for the newsletter version 0:23:40.530 --> 0:23:46.290 of the show at Daniel miessler.com/newsletter. We'll see you next time.