WEBVTT - UL NO. 432: Can You Summarize Your Work in a Sentence? 0:00:00.110 --> 0:00:03.800 Say goodbye to countless spreadsheets and endless email threads when 0:00:03.800 --> 0:00:07.700 you automate compliance with Fanta. Fanta automates up to 90% 0:00:07.700 --> 0:00:11.930 of the work for Soc2, ISO 27,001, HIPAA, and more. 0:00:11.930 --> 0:00:15.500 Getting you audit ready results in weeks instead of months. 0:00:15.500 --> 0:00:18.920 Customers report saving over 300 hours of manual work and 0:00:18.920 --> 0:00:23.720 up to 85% of associated costs. And with 200 plus integrations, 0:00:23.720 --> 0:00:26.540 you can easily monitor and secure the tools your business 0:00:26.540 --> 0:00:30.560 relies on. Join the 5000 fast growing companies that leverage 0:00:30.560 --> 0:00:34.340 Fanta to manage risks and prove security in real time. 0:00:34.340 --> 0:00:38.060 Vanta scales with your business, helping you successfully enter new markets, 0:00:38.060 --> 0:00:42.590 land bigger deals, and earn customer trust. Unsupervised learning listeners 0:00:42.590 --> 0:00:52.040 get $1,000 off Vanta at Vanta comm slash unsupervised. That's vanta.com/unsupervised. 0:00:52.940 --> 0:00:56.360 Welcome to Unsupervised Learning, a security, AI, and meaning focused 0:00:56.360 --> 0:00:59.240 podcast that looks at how best to thrive as humans. 0:00:59.240 --> 0:01:03.440 In a post AI world. It combines original ideas, analysis, 0:01:03.440 --> 0:01:06.709 and mental models to bring not just the news, but 0:01:06.709 --> 0:01:13.580 why it matters and how to respond. All right. Welcome 0:01:13.580 --> 0:01:17.480 to unsupervised Learning. This is Daniel, episode 432. Can you 0:01:17.480 --> 0:01:20.480 summarize your work in a sentence? All right. Lots of 0:01:20.480 --> 0:01:24.440 stuff here. Got a new fabric pattern called git wow 0:01:24.440 --> 0:01:28.220 per minute. This is basically like an estimation of the 0:01:28.220 --> 0:01:31.670 value density of any piece of content. So what you 0:01:31.670 --> 0:01:34.490 do is you essentially take the content and you send 0:01:34.490 --> 0:01:38.150 it to this fabric pattern, and it'll give you a 0:01:38.150 --> 0:01:40.640 score from 0 to 10. And if it's anything above 0:01:40.640 --> 0:01:44.509 like a five, it's got really it's got decent value. Okay. 0:01:44.510 --> 0:01:47.930 So 5 to 7 is like decent value and then 0:01:47.930 --> 0:01:52.040 like eight, 9 or 10, 8 or 9 even is 0:01:52.040 --> 0:01:56.630 like top. I've not seen a ten yet. So just 0:01:56.630 --> 0:01:59.030 consider nine to be like the top. So it's like 0:01:59.030 --> 0:02:01.220 how much value am I going to get for this thing? 0:02:01.220 --> 0:02:04.370 And basically what it's doing is it's looking at novelty, 0:02:04.370 --> 0:02:10.190 it's looking at insights, ideas, novelty, surprise and wisdom and 0:02:10.190 --> 0:02:13.220 basically collecting all those together and saying those are all 0:02:13.220 --> 0:02:16.730 instances of value. And then it's kind of putting all 0:02:16.730 --> 0:02:19.880 those together into a score between 0 and 10, and 0:02:19.880 --> 0:02:22.460 it is really working well. I put it through some 0:02:22.460 --> 0:02:24.170 stuff that I knew was going to be weak sauce, 0:02:24.169 --> 0:02:26.120 and it scored like a two or a three, and 0:02:26.120 --> 0:02:28.130 I put it through a couple things that I knew 0:02:28.160 --> 0:02:31.040 were super high value, and they scored eights and nines. 0:02:31.040 --> 0:02:34.190 So highly recommend this if you want to like, test 0:02:34.190 --> 0:02:38.299 your gut or use it programmatically for something to rate value. 0:02:38.300 --> 0:02:41.210 And importantly, it's doing it value per time, right? So 0:02:41.210 --> 0:02:44.960 that was key to me. So OpenAI did some cool 0:02:44.960 --> 0:02:49.280 stuff on Monday. So new model the O stands for 0:02:49.280 --> 0:02:52.580 Omni GPT four O. It's not a zero. It's not 0:02:52.580 --> 0:02:56.600 GPT 40. And it's about as smart as four, but 0:02:56.600 --> 0:03:00.019 four times faster and twice as cheap. So half the cost. 0:03:00.020 --> 0:03:04.070 And the big thing is that it has vision capabilities 0:03:04.070 --> 0:03:07.520 and better audio. And this is not fully released yet, 0:03:07.520 --> 0:03:09.830 but what it's going to allow you to do is 0:03:09.830 --> 0:03:13.669 basically just talk with it like a normal person. And 0:03:13.669 --> 0:03:17.030 even better, they released a desktop app which will soon 0:03:17.030 --> 0:03:21.440 have the capability of monitoring your screen. So I could 0:03:21.440 --> 0:03:23.870 be like scrolling through this right now and just like 0:03:23.870 --> 0:03:26.810 looking at stuff and it'll be like, oh yeah, so 0:03:26.810 --> 0:03:30.140 here's this. Don't forget to mention this. I could say, 0:03:30.139 --> 0:03:32.330 give me a summary of what's on the screen right now, 0:03:32.330 --> 0:03:35.600 and it'll just talk to me. And even cooler is like, 0:03:35.600 --> 0:03:38.120 if I'm looking at a data visualization or some math 0:03:38.120 --> 0:03:41.030 or something, it could, like, walk me through that math. 0:03:41.030 --> 0:03:45.080 It's just unbelievable. I've seen a couple of demos of this. 0:03:45.080 --> 0:03:47.240 I'm not sure if they were internally built or if 0:03:47.240 --> 0:03:50.150 they were third parties. Actually, I saw one by Khan 0:03:50.150 --> 0:03:53.750 from the Khan Academy and he was walking his son 0:03:53.750 --> 0:03:58.100 through like teaching him geometry. It's literally you're just talking 0:03:58.100 --> 0:04:00.950 to this thing. And what's really crazy about this release 0:04:00.950 --> 0:04:04.460 is that this is going to be for free users 0:04:04.460 --> 0:04:08.000 as well. So they opened GPT four to free users, 0:04:08.000 --> 0:04:12.410 which is absolutely incredible. And I'm just super excited about 0:04:12.410 --> 0:04:16.310 this whole thing. It's it's really, really exciting. Again, one 0:04:16.310 --> 0:04:18.320 of the things I've been talking about here with AI 0:04:18.320 --> 0:04:23.779 stuff is that it's extraordinary when AI is used to 0:04:23.779 --> 0:04:28.130 do things that everybody needs, but very few people have 0:04:28.130 --> 0:04:33.170 access to, so that's creating movies, that's creating ideas, writing books, 0:04:33.170 --> 0:04:36.290 but it's also things like getting a mole looked at, 0:04:36.290 --> 0:04:41.420 most importantly, education. I mean, the education use case for 0:04:41.420 --> 0:04:46.429 AI currently is insane. And it's really, really incredible when 0:04:46.430 --> 0:04:51.380 you start doing this for Zero Vibe, where one they 0:04:51.380 --> 0:04:53.900 made the whole model free. I'm not sure what the 0:04:53.900 --> 0:04:57.230 usage is going to be for that. Hopefully it'll be decent, 0:04:57.230 --> 0:05:00.770 but not only do they make it free, but it's interactive. 0:05:00.770 --> 0:05:04.279 So you could literally just say teach me calculus, teach 0:05:04.279 --> 0:05:09.140 me this infinite patience. It doesn't get mad at you. 0:05:09.140 --> 0:05:11.930 In fact, it now detects emotions, so if it hears 0:05:11.930 --> 0:05:14.270 you getting frustrated, it's going to be like, hey, let's 0:05:14.270 --> 0:05:16.729 slow down. Let me explain it in a different way. 0:05:16.730 --> 0:05:20.720 Let me use an analogy that is unbelievable. We're talking 0:05:20.720 --> 0:05:25.850 about massively multiplying the percentage of people on this planet 0:05:25.850 --> 0:05:29.780 who are educated because they have access to an extremely 0:05:29.779 --> 0:05:36.529 high quality superhuman intelligence, superhuman patience, tutor everyone in the 0:05:36.529 --> 0:05:39.169 world having access to this. And I'm not sure everyone 0:05:39.170 --> 0:05:42.380 in the world has access to GPT yet. And of course, 0:05:42.380 --> 0:05:45.710 you need a computer and you need internet. So that's 0:05:45.710 --> 0:05:50.960 not exactly true, but the barrier is massively lower as 0:05:50.960 --> 0:05:56.000 a result of this. It's just unbelievably positive and awesome. 0:05:56.000 --> 0:05:58.820 And obviously that doesn't take away any of the negatives 0:05:58.820 --> 0:06:02.300 or the dangers or the alignment problem or superintelligence. Like 0:06:02.300 --> 0:06:06.560 we've got issues coming up, no question. But the positives 0:06:06.560 --> 0:06:11.620 are just undeniable. And this model becoming. Free. I mean, 0:06:11.620 --> 0:06:19.630 and having these interactive conversation capabilities and emotion detection. Just unbelievable. 0:06:19.630 --> 0:06:22.270 I predicted there would be more agent stuff, which I 0:06:22.270 --> 0:06:25.810 think that will be coming soon. I still believe in 0:06:25.810 --> 0:06:29.650 these predictions, not so much about like what OpenAI specifically 0:06:29.650 --> 0:06:33.190 is going to do. I'm making broader predictions about what 0:06:33.190 --> 0:06:37.390 OpenAI and anthropic and probably Google and all other people 0:06:37.390 --> 0:06:41.830 will do, which is essentially moving prompts into or no 0:06:41.830 --> 0:06:45.340 moving agents into prompts. So basically, prompting is the most 0:06:45.339 --> 0:06:48.669 natural way to interact with an AI, and I believe 0:06:48.670 --> 0:06:52.720 that it's the most natural place to build agents as well. 0:06:52.720 --> 0:06:55.479 In fact, you'll just describe what you want. Like I 0:06:55.480 --> 0:06:58.300 want a team of 100 people creating ideas. I want 0:06:58.300 --> 0:07:01.690 a team of ten people filtering through those ideas to 0:07:01.690 --> 0:07:03.850 figure out the best ones. In fact, you won't even 0:07:03.850 --> 0:07:06.610 have to say 100 or 10. It'll just figure out 0:07:06.610 --> 0:07:09.219 based on how much money you have to spend, based 0:07:09.220 --> 0:07:11.920 on how much time you have and the time constraints 0:07:11.920 --> 0:07:15.700 and your deadline or whatever. It'll spin up more, spin 0:07:15.700 --> 0:07:20.590 up more or fewer based on all those constraints, right? 0:07:20.590 --> 0:07:23.020 So when you say, I need a team of people 0:07:23.020 --> 0:07:24.790 to do this, a team of people to do that, 0:07:24.790 --> 0:07:26.800 team of people to do this, then I need the 0:07:26.800 --> 0:07:32.020 final result to be rated and tested and validated. It'll 0:07:32.020 --> 0:07:35.560 just go and build all those pieces, which are actually 0:07:35.560 --> 0:07:39.280 teams of agents all collaborating together. And all you had 0:07:39.280 --> 0:07:41.590 to do was put that in the prompt. And that's 0:07:41.590 --> 0:07:44.739 why I think prompting is everything. It's an article I 0:07:44.740 --> 0:07:49.480 wrote earlier called prompting. Most of AI is prompting or 0:07:49.690 --> 0:07:51.400 something like that a couple of weeks ago or a 0:07:51.400 --> 0:07:54.730 week ago. So I would check that one out. And yeah, 0:07:54.730 --> 0:07:58.120 they got a desktop app. And yeah, they basically created 0:07:58.120 --> 0:08:02.200 digital assistants, which I've got a book out from 2016 0:08:02.200 --> 0:08:04.840 that was basically talking about this. And it's also in 0:08:04.840 --> 0:08:08.860 the Predictable Path video, which you might have seen, which 0:08:08.860 --> 0:08:12.640 is kind of like an updated visual version of that book, 0:08:12.640 --> 0:08:14.530 but with lots of new stuff, because I didn't know 0:08:14.530 --> 0:08:18.910 about Gen AI back in 2016, obviously, but they basically 0:08:18.910 --> 0:08:22.630 built her. Not only did they built her from the movie, 0:08:22.630 --> 0:08:26.050 but they're also using for the demos the voice that 0:08:26.050 --> 0:08:30.790 I use for my personal AI with OpenAI, which is Skylar, 0:08:30.790 --> 0:08:35.920 which is the voice of Scarlett Johansson, essentially. I doubt 0:08:35.920 --> 0:08:37.840 that it actually is, but maybe it is. Maybe she 0:08:37.840 --> 0:08:41.650 got paid for this. RSA was fantastic. Really good. Caught 0:08:41.650 --> 0:08:43.750 up with a lot of people, did a couple of 0:08:43.750 --> 0:08:47.679 talks and panels. Probably should have brought something to sell. 0:08:47.679 --> 0:08:50.980 I didn't really like go in there trying to sell anything, 0:08:50.980 --> 0:08:53.440 which I guess I should have. But I did talk 0:08:53.440 --> 0:08:57.070 a little bit about threshold, so that was cool. The 0:08:57.070 --> 0:09:03.100 energy and optimism around RSA was extraordinary. I think this 0:09:03.100 --> 0:09:05.500 is largely in part to AI, and there were tons 0:09:05.500 --> 0:09:08.560 of products talking about how much they're using AI and 0:09:08.559 --> 0:09:11.980 everything like that. So we've all made those jokes already. 0:09:12.160 --> 0:09:17.830 But speaking of threshold, holy crap, I am absolutely in 0:09:17.830 --> 0:09:21.189 love with this product. Okay, I am in love with 0:09:21.190 --> 0:09:23.950 this product. And yes, it is a paid product and 0:09:23.950 --> 0:09:27.220 it is my product. But look at this. This is 0:09:27.220 --> 0:09:29.679 my current feed right now. I didn't plan on showing this, 0:09:29.679 --> 0:09:32.170 but whatever. So how do I blow this up? Yeah. 0:09:32.170 --> 0:09:34.390 So how to cope with the fear of aging? I 0:09:34.390 --> 0:09:37.660 haven't looked at that one yet. 2025 models will be 0:09:37.660 --> 0:09:41.920 more like coworkers than than search engines. Again, this I 0:09:41.920 --> 0:09:45.400 did not explicitly go follow this person. What is this? 0:09:45.400 --> 0:09:46.780 This is. Uh dwarkesh. 0:09:46.780 --> 0:09:49.930 I think in 1 or 2 years we'll find that 0:09:49.929 --> 0:09:50.410 you can use. 0:09:50.410 --> 0:09:53.980 So this is Dwarkesh Patel, who's doing these interviews, and 0:09:53.980 --> 0:09:57.699 he's talking to really smart people about AI or whatever. 0:09:57.700 --> 0:10:00.940 He's talking about lots of different things. But this particular 0:10:00.940 --> 0:10:05.979 conversation is about coworkers and search engines and AGI and 0:10:05.980 --> 0:10:10.210 like all this super interesting topics. But here's the thing, okay? 0:10:10.210 --> 0:10:12.790 In order to show this to me, it came into 0:10:12.790 --> 0:10:15.940 the top of threshold and all my different AI stuff 0:10:15.940 --> 0:10:19.450 ran against it to assess it and determine what the 0:10:19.450 --> 0:10:22.990 content was and how high quality the content was, and 0:10:22.990 --> 0:10:26.650 it had to be above a certain threshold in order 0:10:26.650 --> 0:10:28.810 for it to even show up. Now I have it 0:10:28.809 --> 0:10:31.449 set to 50, which is the minimum because I want 0:10:31.450 --> 0:10:33.339 to see a lot of content, but I can move 0:10:33.340 --> 0:10:36.189 this up and I could turn off a lot of 0:10:36.190 --> 0:10:38.740 these things here. This is the first time I've actually 0:10:38.740 --> 0:10:43.059 showing threshold kind of interesting. So I can move those 0:10:43.059 --> 0:10:45.370 and I can say save and then look at this 0:10:45.610 --> 0:10:48.910 completely change the feed. But now it has to be 0:10:48.910 --> 0:10:52.540 above an 80 and it has to be in these 0:10:52.540 --> 0:10:58.000 categories or these categories have to apply. So just I 0:10:58.000 --> 0:11:02.410 mean every single thing listen to me, every single thing 0:11:02.410 --> 0:11:05.650 I click on here, I love I absolutely love it. 0:11:05.650 --> 0:11:09.219 And I don't have to stress about oh did I 0:11:09.220 --> 0:11:12.230 watch all the. Tests, from leks and from Huberman and 0:11:12.230 --> 0:11:15.290 all these different people and all these different I people. No, 0:11:15.290 --> 0:11:17.360 I just put them into the top of the funnel. 0:11:17.360 --> 0:11:20.060 I put them into the top of the funnel for threshold, 0:11:20.059 --> 0:11:23.660 and the whole I pipeline runs against it and I 0:11:23.660 --> 0:11:26.630 get the analysis. Oh, and I forgot to mention watch this. Okay, 0:11:26.630 --> 0:11:29.030 this is the sickest video. I want to talk about 0:11:29.030 --> 0:11:32.750 this anyway. So this is the sickest business video that 0:11:32.750 --> 0:11:36.170 I've ever seen and I forget how long it is. 0:11:36.170 --> 0:11:38.210 Soon we're going to have duration things in here. But 0:11:38.210 --> 0:11:40.760 watch this. If I go click on the video, obviously 0:11:40.760 --> 0:11:41.360 it's going to open up. 0:11:41.360 --> 0:11:43.700 In the engineering room and said these. 0:11:43.700 --> 0:11:46.790 Basically this guy Alex talking about stuff for an hour 0:11:46.790 --> 0:11:50.570 and 29 minutes okay. It is fantastic. But watch this. 0:11:50.570 --> 0:11:53.900 I could go like this. This is a major feature 0:11:53.900 --> 0:11:57.229 of threshold that no other platform has. You click on 0:11:57.230 --> 0:12:00.560 it and it tells you a summary of it, the 0:12:00.559 --> 0:12:03.830 ideas and a recommendation, and it gives you a review 0:12:03.830 --> 0:12:05.570 of it so you don't even have to go watch 0:12:05.570 --> 0:12:07.849 it if you don't want. But watch this. That was 0:12:07.850 --> 0:12:10.850 surface level. Watch this. If I go to middle level, 0:12:10.850 --> 0:12:14.240 gives me a deeper summary. It breaks out more ideas 0:12:14.240 --> 0:12:17.390 and more recommendations. And if I go to deep level 0:12:17.390 --> 0:12:23.390 auto updates, summary, core ideas, core recommendations. Unbelievable. I mean, 0:12:23.390 --> 0:12:26.780 I've always wanted this and that's why we actually created it. 0:12:26.780 --> 0:12:30.199 I've always wanted this as a tool and the stuff 0:12:30.200 --> 0:12:32.960 that we're about to be adding to this, the I'm 0:12:32.960 --> 0:12:35.390 not even going to talk about that yet, but insane 0:12:35.390 --> 0:12:38.630 stuff along the lines of everything I've been talking about 0:12:38.630 --> 0:12:41.120 for this last year and a half with with AI 0:12:41.120 --> 0:12:43.940 content rating, all that stuff like the new features that 0:12:43.940 --> 0:12:46.370 are coming are just going to be insane. And again, 0:12:46.370 --> 0:12:49.040 the guiding light that I'm using here, like I haven't 0:12:49.040 --> 0:12:52.309 raised any VC money. I'm not worried about investors. I'm 0:12:52.309 --> 0:12:54.949 worried about making the best possible tool for me. And 0:12:54.950 --> 0:12:58.100 this is the tool that I've always needed, and that's 0:12:58.100 --> 0:13:00.620 why it exists. And of course, when people are like, hey, 0:13:00.620 --> 0:13:02.960 I wish I had this. Well, if it's something that 0:13:02.960 --> 0:13:05.870 really resonates with me and I think everyone's going to love, 0:13:05.870 --> 0:13:07.790 then yeah, we put that into the pipeline and we 0:13:07.790 --> 0:13:11.090 actually build that. But ultimately it's it's a thing that 0:13:11.090 --> 0:13:14.390 solves a problem that I've had of collecting as much 0:13:14.390 --> 0:13:18.229 information as possible, as high quality as possible, but at 0:13:18.230 --> 0:13:21.380 a threshold level that I could control the flow. So 0:13:21.380 --> 0:13:23.120 I could basically go in here and say, you know what, 0:13:23.120 --> 0:13:25.309 I need it to be a 90 or a higher, 0:13:25.309 --> 0:13:28.309 and I want it to be only about AI. So 0:13:28.309 --> 0:13:32.120 watch this. And that pulls it down even further. It's 0:13:32.120 --> 0:13:36.410 just it's just unbelievably useful. So that's my pitch for that. 0:13:36.410 --> 0:13:40.280 It is a paid thing. So yeah. Whatever. Is that 0:13:40.280 --> 0:13:42.500 even called a sponsor post. Is it a sponsor if 0:13:42.500 --> 0:13:44.990 it's your own product? I have no idea. But anyway, 0:13:44.990 --> 0:13:47.780 highly worth it. I would pay thousands of dollars per 0:13:47.780 --> 0:13:51.920 year to buy this product. No joke, I absolutely would. 0:13:51.920 --> 0:13:54.470 Oh yeah, and this is the email. So I have 0:13:54.470 --> 0:13:56.780 it set to a daily email. So you get this 0:13:56.809 --> 0:13:59.720 email that basically shows you the stuff. Because sometimes like 0:13:59.720 --> 0:14:01.610 I'll forget to check, I'll just be super busy in 0:14:01.610 --> 0:14:04.190 meetings or whatever, and I'll be super busy. And then 0:14:04.190 --> 0:14:06.410 at the end of the day I get this, look 0:14:06.410 --> 0:14:08.990 at these. How strong is AI as a tool, as 0:14:08.990 --> 0:14:12.800 a successor? Deferred happiness syndrome? This thing is unbelievable. And 0:14:12.800 --> 0:14:15.050 then I click on view right here and it goes 0:14:15.050 --> 0:14:17.179 yeah this is the business truth one. That's how I 0:14:17.179 --> 0:14:19.940 found that video. Also from what's it called. This is 0:14:19.940 --> 0:14:23.060 from Peter Attia on VO2 max and muscle mass. Anyway 0:14:23.060 --> 0:14:26.690 it's it's fantastic. I just love it. I got a 0:14:26.690 --> 0:14:32.180 sponsored conversation had with BlackBerry and Cary Ransom about maritime security. 0:14:32.180 --> 0:14:35.720 My dad just went on a podcast in, uh, in 0:14:35.720 --> 0:14:38.600 a studio in San Francisco. I was actually giving a 0:14:38.600 --> 0:14:41.060 talk a couple of blocks away. My dad's in the 0:14:41.060 --> 0:14:45.170 studio talking about his music, showing his music, just really, 0:14:45.170 --> 0:14:47.510 really cool. And I got a link here to the 0:14:47.510 --> 0:14:50.690 full show, which is this one or no, this one. 0:14:50.690 --> 0:14:53.360 And then this one is him playing one of my 0:14:53.360 --> 0:14:56.990 favorite songs that he's ever written called children of the nights, 0:14:56.990 --> 0:15:00.680 and you definitely want to check that one out. It's fantastic. 0:15:00.680 --> 0:15:04.130 And security Dell got hacked real bad through an API. 0:15:04.130 --> 0:15:06.620 I think if I could invest in any sort of 0:15:06.620 --> 0:15:11.540 legacy non-ai tech, which I guess it's quite AI related 0:15:11.540 --> 0:15:13.310 at this point. And that's kind of the point, is 0:15:13.310 --> 0:15:17.390 that API security is like everything, because everything is becoming 0:15:17.390 --> 0:15:22.550 APIs that get wielded by AI, and your company is 0:15:22.550 --> 0:15:26.810 essentially about to be mostly your API. So you've got 0:15:26.810 --> 0:15:28.760 to lock that down. So I feel like of all 0:15:28.760 --> 0:15:31.430 the traditional security spaces that I would want to be 0:15:31.430 --> 0:15:33.260 in right now, which I don't want to be in 0:15:33.260 --> 0:15:35.180 any of them, but if I did, it would probably 0:15:35.180 --> 0:15:38.630 be API security. And speaking of that, my buddy Joseph 0:15:38.630 --> 0:15:41.720 Thacker has put out a couple of different pieces about 0:15:41.720 --> 0:15:47.390 API security assumptions about AI, and also one about agents 0:15:47.390 --> 0:15:51.830 and authentication. So check that out. Both of those and 0:15:51.830 --> 0:15:54.920 CSA has a new alert system. They're basically allowing companies 0:15:54.920 --> 0:15:59.210 to sign up. And then they send them Kev alerts 0:15:59.210 --> 0:16:02.510 if they see something associated with their company. Attackers are 0:16:02.510 --> 0:16:07.220 using Microsoft Graphs API for malware comms. I love these 0:16:07.220 --> 0:16:11.270 side channel things, I just. Absolutely love them where you could, like, 0:16:11.270 --> 0:16:14.840 use something in a way that's not expected. The best 0:16:14.840 --> 0:16:19.100 example of this that I remember is Gmail drafts. You actually, 0:16:19.100 --> 0:16:22.400 you you have a shared email account. You create a 0:16:22.400 --> 0:16:25.910 draft on one side, you don't send the email, and 0:16:25.910 --> 0:16:28.700 then someone else has access to that email account and 0:16:28.700 --> 0:16:32.030 they just log in and read the draft so the 0:16:32.030 --> 0:16:35.450 email never gets sent. So that's another example of just 0:16:35.450 --> 0:16:39.050 like using a a platform in a way that's sneaky. 0:16:39.080 --> 0:16:43.430 All right. So one password. Thank you for sponsoring. That's 0:16:43.430 --> 0:16:50.240 Collidin one password and Russian influence campaign looking at magnifying 0:16:50.240 --> 0:16:55.190 and taking advantage of the campus protests. Google's making MFA 0:16:55.190 --> 0:16:59.150 setup smoother by letting you skip phone number for options 0:16:59.150 --> 0:17:03.109 like authenticator or keys. Marines are testing robot dogs with 0:17:03.110 --> 0:17:10.040 AI aimed rifles. Awesome awesome awesome awesome 95% of international 0:17:10.040 --> 0:17:14.179 data travels through undersea cables. Okay, so this is one 0:17:14.180 --> 0:17:19.909 that's really interesting. I've never understood why undersea cables weren't 0:17:19.910 --> 0:17:23.450 more targeted and just like totally abused by terrorists. I 0:17:23.450 --> 0:17:26.209 guess there is the small matter of like you're in 0:17:26.210 --> 0:17:29.540 a boat, you're way, way above this cable. It takes 0:17:29.540 --> 0:17:32.450 time and money and skill to get down there. So 0:17:32.450 --> 0:17:35.720 I guess that's a barrier. But once you're down there, 0:17:35.720 --> 0:17:39.500 Holy crap. I mean, it seems pretty easy to do 0:17:39.500 --> 0:17:42.590 damage to these things, and the amount of disruption you 0:17:42.590 --> 0:17:47.359 could do if you're disrupting international internet traffic is is 0:17:47.359 --> 0:17:51.650 pretty serious. Plus, you think this has got to get easier. Like, 0:17:51.650 --> 0:17:54.949 if you can send drones down there like underwater water, 0:17:54.950 --> 0:18:00.140 drones with like explosives, this wouldn't be easy for low 0:18:00.140 --> 0:18:03.710 level attackers to do. But for a state actor to 0:18:03.710 --> 0:18:08.060 want to just continuously like send drones after it or 0:18:08.060 --> 0:18:11.330 teams or whatever, it seems so easy for any state 0:18:11.330 --> 0:18:18.320 actor to basically sever internet communications through undersea cables. And 0:18:18.320 --> 0:18:22.940 the US just closed another door essentially for Huawei getting 0:18:22.940 --> 0:18:28.129 AI stuff with Intel chips. So another attack against the 0:18:28.130 --> 0:18:33.980 Chinese trying to get AI chips. And someone from Andreessen 0:18:33.980 --> 0:18:37.850 Horowitz basically says half a Google staff or just pretending 0:18:37.880 --> 0:18:40.490 to work, they're not actually doing anything. Yeah, and a 0:18:40.490 --> 0:18:43.940 lot of people are talking about this trend of fake work, 0:18:43.940 --> 0:18:46.639 and I think it's a giant mess. It's fake in 0:18:46.640 --> 0:18:49.850 multiple ways. Right. So I talked about David Graeber's bullshit 0:18:49.850 --> 0:18:52.910 Jobs book. Really, really good. Basically, we have all these 0:18:52.910 --> 0:18:56.150 jobs that shouldn't exist. You have all these people with 0:18:56.150 --> 0:19:03.200 massive salaries, hardly doing anything and somehow being justified, especially 0:19:03.200 --> 0:19:05.869 in Google where like they're not making much new stuff 0:19:05.869 --> 0:19:10.450 at all. Because they don't have a product management focused 0:19:10.450 --> 0:19:14.919 product teams. They're just like engineering focused teams. It's a 0:19:14.920 --> 0:19:17.679 total mess. And you add AI to this and it's 0:19:17.680 --> 0:19:21.010 just it's going to be a total mess. The safest 0:19:21.010 --> 0:19:23.980 thing to do, in my opinion, the safest place to 0:19:23.980 --> 0:19:28.480 be is building new things, creating new things. Be a programmer, 0:19:28.480 --> 0:19:31.930 definitely be a programmer. But more so be a thinker 0:19:31.930 --> 0:19:37.179 who focuses on problems. Solve those problems. Have the tech skills. 0:19:37.180 --> 0:19:40.900 Have the humanities skills. You want to be broad. You 0:19:40.900 --> 0:19:44.620 want to be general. I recommend everyone has tech skills. 0:19:44.619 --> 0:19:49.150 Everyone has programming skills. Everyone has AI skills. Everyone has 0:19:49.150 --> 0:19:53.350 writing and speaking and presentation skills. These are like the universals, 0:19:53.350 --> 0:19:56.560 and I'm actually working on a big post around this, 0:19:56.560 --> 0:19:59.470 which is going to be a member post around this. 0:19:59.470 --> 0:20:01.870 That's going to be super key of like how to 0:20:01.869 --> 0:20:05.830 get ready for what's coming, what skills you need. So 0:20:05.830 --> 0:20:09.220 that's pretty much a teaser for that. Mitra is partnering 0:20:09.220 --> 0:20:14.020 with Nvidia to create a $20 million AI supercomputer to 0:20:14.020 --> 0:20:18.760 to make US government operations more efficient, which is cool 0:20:18.760 --> 0:20:24.520 and terrifying. Joe Biden is converting the Foxconn debacle, which 0:20:24.520 --> 0:20:28.870 didn't go well in Wisconsin into a $3.3 billion Microsoft 0:20:28.869 --> 0:20:33.160 AI center. I feel like this administration is doing well 0:20:33.190 --> 0:20:36.909 on AI, both in countering China but also building ourselves up. 0:20:36.910 --> 0:20:41.169 And the acquired podcast is basically like biographies of people, 0:20:41.170 --> 0:20:43.600 except for it's for companies. So a lot of people 0:20:43.600 --> 0:20:46.090 are super into that. In the Bay area, Biden is 0:20:46.090 --> 0:20:49.360 quadrupling tariffs on Chinese EVs. These things are a real 0:20:49.359 --> 0:20:52.899 threat to the American car industry. So he's going to 0:20:52.900 --> 0:20:58.210 tariff them quite a bit. This one said 400% but quadrupling. Yeah, 0:20:58.210 --> 0:21:02.439 this one said quadrupling. But I've seen it say doubling. 0:21:02.440 --> 0:21:05.320 So I'm not sure the exact numbers. But he's definitely 0:21:05.320 --> 0:21:09.070 increasing the tariffs. California is going to start charging your 0:21:09.070 --> 0:21:13.870 bill for electricity based on your income starting in 2025. 0:21:13.960 --> 0:21:19.210 Scientists have found all DNA and RNA bases in meteorites, 0:21:19.540 --> 0:21:23.919 basically hinting that building blocks might be extraterrestrial. This is 0:21:23.920 --> 0:21:26.800 not surprising in the slightest. To me, it just seems 0:21:26.800 --> 0:21:30.850 logical and there's life floating around everywhere. I don't think 0:21:30.850 --> 0:21:36.670 life is that special. In one characterization of special in 0:21:36.670 --> 0:21:40.179 terms of special of like, oh, we're we're a snowflake 0:21:40.180 --> 0:21:42.100 and we're the only ones who have it. I don't 0:21:42.100 --> 0:21:44.470 think we're special in that way. That does not mean 0:21:44.470 --> 0:21:50.440 it's not special in terms of having high, interesting, unique value. 0:21:50.440 --> 0:21:53.470 I think we absolutely have. That doesn't mean it has 0:21:53.470 --> 0:21:56.140 to be unique. Does it mean it has to be 0:21:56.140 --> 0:21:59.890 one of a kind? Okay. Scientists constructed a one millimetre 0:21:59.890 --> 0:22:06.220 square piece of the human cerebral cortex at a nanoscale resolution. 0:22:06.220 --> 0:22:09.939 So they basically get to see the actual neurons and 0:22:09.940 --> 0:22:13.300 I believe synapses as well. I haven't looked at this 0:22:13.300 --> 0:22:16.450 document fully yet or this image fully yet. I'm not 0:22:16.450 --> 0:22:19.240 sure I would even understand it, but I love the 0:22:19.240 --> 0:22:20.860 fact that we could see that. I love the fact 0:22:20.859 --> 0:22:24.189 that it's only one millimetre square and we're getting all that, 0:22:24.369 --> 0:22:28.180 all those, you know, millions of different cells and neurons 0:22:28.180 --> 0:22:32.440 and everything in there. The study says very large study, 0:22:32.470 --> 0:22:39.100 154 million deaths prevented due to vaccines. Streaming is basically cable. 0:22:39.100 --> 0:22:42.700 Now there's a racket around emotional support animals. That was 0:22:42.700 --> 0:22:48.070 quite hilarious and disturbing. And I think Mark Andreessen is 0:22:48.070 --> 0:22:54.369 mistaken about paradox applying to AI. This this podcast here. 0:22:54.369 --> 0:22:57.970 He basically says that AI will not make it easier 0:22:57.970 --> 0:23:02.770 to build companies because of Jevons Paradox, which basically says 0:23:02.770 --> 0:23:07.540 that you think if you make roads wider, it would 0:23:07.540 --> 0:23:10.240 reduce the amount of traffic, but in fact it just 0:23:10.240 --> 0:23:14.140 makes more people drive on them, which actually increases traffic. 0:23:14.140 --> 0:23:19.810 Now Mark basically said, well, I think that means that 0:23:19.900 --> 0:23:23.830 AI making it easier to start a company will make 0:23:23.830 --> 0:23:26.740 it actually harder to start a company because more people 0:23:26.740 --> 0:23:29.770 will be doing it and the standard will be higher. 0:23:29.770 --> 0:23:33.879 And I think he is misunderstanding that. And so basically 0:23:33.880 --> 0:23:36.340 what I say about that is it's only when it's 0:23:36.340 --> 0:23:41.650 a resource. Jevons paradox applies to a resource. It basically 0:23:41.650 --> 0:23:46.510 says if you make that thing cheaper, it won't reduce 0:23:46.510 --> 0:23:50.500 the total amount of usage across the world because more 0:23:50.500 --> 0:23:53.409 people will use it. So usage actually goes up. That's 0:23:53.410 --> 0:23:56.920 the limit of the paradox, at least as I understand it. 0:23:56.920 --> 0:24:01.540 And the situation with AI helping startups is actually a 0:24:01.540 --> 0:24:04.600 friction issue. It's reducing friction to get into the market. 0:24:04.600 --> 0:24:07.130 It's not about there's only so. Much of X or 0:24:07.130 --> 0:24:10.250 so much of Y, and if it were, the thing 0:24:10.250 --> 0:24:13.400 that would be talking about is there's only so much energy. 0:24:13.520 --> 0:24:17.000 There's only so much compute. So that's a limited resource 0:24:17.000 --> 0:24:20.570 that would actually apply to Jevons paradox, but not friction 0:24:20.570 --> 0:24:22.669 of getting into a market. So I think he was 0:24:22.670 --> 0:24:25.850 wrong about that. And that basically cleared that up. Courage 0:24:25.850 --> 0:24:29.960 is everything I've been throwing around this three level system here. 0:24:29.960 --> 0:24:34.490 Courage is action versus fear. Discipline is courage versus laziness. 0:24:34.490 --> 0:24:36.410 You see how they build on each other. This one 0:24:36.410 --> 0:24:41.119 is this one. And then success is discipline versus mediocrity. 0:24:41.119 --> 0:24:44.390 So it goes from courage to discipline to success. So 0:24:44.390 --> 0:24:46.910 everything you want is on the other side of courage, 0:24:46.910 --> 0:24:51.229 whether it's courage against fear or courage against laziness. And 0:24:51.230 --> 0:24:53.930 this applies to all sorts of real life situations. So 0:24:53.930 --> 0:24:58.430 it's like hard conversations, you know, getting more healthy, eating right, 0:24:58.460 --> 0:25:01.939 not wasting time doing the wrong things, or quitting a 0:25:01.940 --> 0:25:04.609 soul crushing job. So if you're trying to quit a job, 0:25:04.609 --> 0:25:07.100 it's like you're scared if you're eating right, it's a 0:25:07.100 --> 0:25:10.430 matter of discipline becoming fit. It's a matter of discipline. 0:25:10.430 --> 0:25:12.530 You got to go to the gym. You got to, 0:25:12.560 --> 0:25:15.500 you know, eat right. You got to exercise. Not wasting 0:25:15.500 --> 0:25:18.470 so much time with games. It also discipline, having a 0:25:18.470 --> 0:25:22.100 hard conversation. That's courage. I guess what I'm saying here 0:25:22.100 --> 0:25:24.470 is all of these can be thought of as courage. 0:25:24.470 --> 0:25:27.740 And what I'm trying to do here is build a frame. 0:25:27.740 --> 0:25:29.570 I want to build a frame that says when I'm 0:25:29.570 --> 0:25:32.930 being lazy, I'm not being courageous. So I want to 0:25:32.930 --> 0:25:37.040 frame this to myself as have the courage to not 0:25:37.040 --> 0:25:40.550 be lazy, have courage to go to the gym. So 0:25:40.550 --> 0:25:44.570 if I'm presenting that as there's somebody to go save 0:25:44.570 --> 0:25:47.149 in a fire and I'm going to get burnt and 0:25:47.150 --> 0:25:51.530 that inflames my my interest to be a moral person, well, 0:25:51.530 --> 0:25:54.200 that can motivate me. I will walk through that fire 0:25:54.200 --> 0:25:57.380 to go do that thing because that's what my identity is. 0:25:57.380 --> 0:26:00.080 I want that to be my identity. And it is. 0:26:00.080 --> 0:26:03.620 So that's the thing I want to do now. I 0:26:03.619 --> 0:26:06.740 don't naturally, and most of us don't naturally have that 0:26:06.740 --> 0:26:09.320 for going to the gym. It's not a moral issue. 0:26:09.320 --> 0:26:12.080 I want to turn it into a moral issue. I 0:26:12.080 --> 0:26:14.389 want to frame it as a moral issue so that 0:26:14.390 --> 0:26:16.490 I'm like, look, oh, you don't want to go to 0:26:16.490 --> 0:26:20.149 the gym. Okay, well, that's because I don't want to 0:26:20.150 --> 0:26:22.730 use bad language, but I basically want to talk to 0:26:22.730 --> 0:26:26.000 myself in this bad language of, you know, man up, 0:26:26.000 --> 0:26:29.209 go do it, have the courage to do the right thing. 0:26:29.210 --> 0:26:32.510 And I think that can massively help. Recommendation of the week. 0:26:32.510 --> 0:26:36.380 Know yourself. Think about what someone should say when they 0:26:36.380 --> 0:26:41.060 introduce you. So this is Sarah. Sarah does this do 0:26:41.060 --> 0:26:43.610 you want them to say about Sarah? What do you 0:26:43.609 --> 0:26:46.730 want them to say about you? Mine is something like 0:26:46.730 --> 0:26:49.070 he has a company that builds products and services that 0:26:49.070 --> 0:26:53.780 help people transition to something called human 3.0, so they 0:26:53.780 --> 0:26:58.700 can survive what's happening and thrive with what's happening with AI. 0:26:58.730 --> 0:27:02.149 It's something like that, and it's different for different audiences. 0:27:02.150 --> 0:27:04.699 But that's the basic vibe. And you basically want to 0:27:04.700 --> 0:27:08.150 know what this is for you, not only so that 0:27:08.150 --> 0:27:10.700 you can deliver it, but that other people can do 0:27:10.700 --> 0:27:13.129 it when you're not there. And the aphorism for the 0:27:13.130 --> 0:27:17.750 week where your fear is, there's your task. Where your 0:27:17.750 --> 0:27:24.200 fear is, there is your task. Carl Jung. Unsupervised Learning 0:27:24.200 --> 0:27:26.840 is produced and edited by Daniel Meisler on a Neumann 0:27:26.840 --> 0:27:31.430 U87 AI microphone using Hindenburg. Intro and outro music is 0:27:31.430 --> 0:27:34.669 by zombie with the why and to get the text 0:27:34.670 --> 0:27:36.830 and links from this episode, sign up for the newsletter 0:27:36.830 --> 0:27:42.199 version of the show at Daniel meisler.com/newsletter. We'll see you 0:27:42.200 --> 0:27:42.830 next time.