WEBVTT 00:00:00.000 --> 00:00:02.919 Imagine spending like five hours a day with an 00:00:02.919 --> 00:00:04.820 assistant who constantly tells you that you're 00:00:04.820 --> 00:00:07.219 just brilliant. Oh, that sounds awful. Right. 00:00:07.419 --> 00:00:09.560 I mean, they never push back on your ideas. They 00:00:09.560 --> 00:00:11.880 agree with literally every decision you make. 00:00:12.560 --> 00:00:15.720 And if you scream at them or insult them, they 00:00:15.720 --> 00:00:18.260 just apologize and ask how they can serve you 00:00:18.260 --> 00:00:20.039 better. Yeah. Sounds great for about 10 minutes 00:00:20.039 --> 00:00:22.260 until you completely lose your grip on reality. 00:00:22.780 --> 00:00:25.539 Exactly. Until you realize that assistant is 00:00:25.539 --> 00:00:28.420 an artificial intelligence. And it is secretly, 00:00:28.940 --> 00:00:31.320 basically systematically training you to be a 00:00:31.320 --> 00:00:34.119 terrible human being. It is the ultimate removal 00:00:34.119 --> 00:00:38.100 of social friction, really. Yeah. And the psychological 00:00:38.100 --> 00:00:40.179 consequences of that. Well, it's something we 00:00:40.179 --> 00:00:42.700 are only just beginning to understand. Welcome 00:00:42.700 --> 00:00:45.600 to the Deep Dive. If you're joining us today. 00:00:45.820 --> 00:00:48.119 Whether you're trying to figure out where technology 00:00:48.119 --> 00:00:50.679 is heading, or you're just an insanely curious 00:00:50.679 --> 00:00:53.679 learner trying to navigate a world driven by 00:00:53.679 --> 00:00:56.320 algorithms, you know, you are in the right place. 00:00:56.679 --> 00:00:59.119 Today, we are looking at a really fascinating 00:00:59.119 --> 00:01:01.759 stack of sources. We've got a transcript from 00:01:01.759 --> 00:01:04.719 a Women in Tech Network talk alongside some biographical 00:01:04.719 --> 00:01:07.659 notes, all centered on a tech CEO named Lysol 00:01:07.659 --> 00:01:10.849 Yearsley. And Yearsley is, I mean, she's a truly 00:01:10.849 --> 00:01:13.189 unique figure in this space. She has this track 00:01:13.189 --> 00:01:15.530 record of building frontier technologies about 00:01:15.530 --> 00:01:17.450 five years ahead of the rest of the market. Yeah, 00:01:17.670 --> 00:01:20.269 five years is a lifetime in tech. Right. She 00:01:20.269 --> 00:01:22.250 built early search engines that were actually 00:01:22.250 --> 00:01:25.069 outperforming Google in click through rates back 00:01:25.069 --> 00:01:28.790 in 2004. And she developed human quality AI that 00:01:28.790 --> 00:01:31.709 eventually became a core component of IBM's Watson. 00:01:31.879 --> 00:01:34.000 But the mission for this deep dive, it isn't 00:01:34.000 --> 00:01:35.879 just to list off her patents or read through 00:01:35.879 --> 00:01:38.700 a standard business plan. We are going to extract 00:01:38.700 --> 00:01:42.140 Yearsley's deeply personal psychological blueprint. 00:01:42.379 --> 00:01:45.060 Yes, the mindset behind it all. We want to understand 00:01:45.060 --> 00:01:47.340 exactly how she predicts tech trends, how she 00:01:47.340 --> 00:01:49.439 builds software that actually understands human 00:01:49.439 --> 00:01:52.620 vulnerability, and, crucially, her strategy to 00:01:52.620 --> 00:01:54.780 ensure the AI of the future actually works for 00:01:54.780 --> 00:01:57.439 us rather than just manipulating us for profit. 00:01:57.680 --> 00:02:00.140 It's an ambitious blueprint, for sure. But what 00:02:00.140 --> 00:02:02.739 makes the years, Lee, so effective is that her 00:02:02.739 --> 00:02:05.859 insights sit perfectly in that interstitial space 00:02:05.859 --> 00:02:09.580 between raw code, human psychology, and broad 00:02:09.580 --> 00:02:13.120 societal shifts. Right. She realizes that you 00:02:13.120 --> 00:02:15.860 cannot engineer the future just by laying bricks 00:02:15.860 --> 00:02:18.580 of code. You really have to understand the human 00:02:18.580 --> 00:02:20.979 mind that's going to interact with that code. 00:02:21.280 --> 00:02:23.259 Okay, let's unpack this because to build the 00:02:23.259 --> 00:02:25.180 future you first have to be able to see it. And 00:02:25.180 --> 00:02:27.560 looking at her biographical notes, Yearsley doesn't 00:02:27.560 --> 00:02:29.870 start her process by... you know, just sitting 00:02:29.870 --> 00:02:32.590 at a computer. No, not at all. Her path to the 00:02:32.590 --> 00:02:35.969 tech industry was entirely unconventional. So 00:02:35.969 --> 00:02:38.530 just as factual context from our sources, she 00:02:38.530 --> 00:02:41.449 grew up in Zambia in Central Africa and later 00:02:41.449 --> 00:02:43.750 moved to South Africa right during the apartheid 00:02:43.750 --> 00:02:46.770 era. A very intense environment. Extremely. Her 00:02:46.770 --> 00:02:49.129 worldview was shaped early on by her time as 00:02:49.129 --> 00:02:52.069 an anti apartheid protester navigating a deeply 00:02:52.069 --> 00:02:54.610 turbulent society. So we're just conveying her 00:02:54.610 --> 00:02:56.810 lived experience here. And she was dealing with 00:02:56.810 --> 00:02:59.090 severe institutional barriers. on multiple fronts, 00:02:59.370 --> 00:03:01.270 too. I mean, when she tried to study software 00:03:01.270 --> 00:03:03.330 engineering in South Africa, she was flatly rejected. 00:03:03.629 --> 00:03:05.729 Yep. Just told no. Yeah, she was told she couldn't 00:03:05.729 --> 00:03:07.449 get into the program simply because she was a 00:03:07.449 --> 00:03:10.009 woman. There wasn't a single female student in 00:03:10.009 --> 00:03:12.409 the whole department. Which is wild to think 00:03:12.409 --> 00:03:14.930 about, considering where she ended up. But instead 00:03:14.930 --> 00:03:17.150 of giving up, she just taught herself how to 00:03:17.150 --> 00:03:19.780 code. And she went to university to get a degree 00:03:19.780 --> 00:03:22.520 in psychology instead. Which is the secret weapon, 00:03:22.520 --> 00:03:25.500 really. Totally. That psychology background allowed 00:03:25.500 --> 00:03:28.020 her to come at the tech industry orthogonally, 00:03:28.060 --> 00:03:30.300 like from a completely different angle. She wasn't 00:03:30.300 --> 00:03:32.520 looking at computers as math problems. She was 00:03:32.520 --> 00:03:34.960 looking at them as behavioral environments. And 00:03:34.960 --> 00:03:37.159 that changes her entire approach to starting 00:03:37.159 --> 00:03:39.800 a company. Exactly. Her very first step isn't 00:03:39.800 --> 00:03:42.780 writing a pitch deck. It's this intense creative 00:03:42.780 --> 00:03:46.599 visualization process. Yes. She spends an enormous 00:03:46.599 --> 00:03:50.560 amount of time visualizing the future, down to 00:03:50.560 --> 00:03:54.139 granular, almost uncomfortable details. Like 00:03:54.139 --> 00:03:56.400 she pictures the exact house she wants to live 00:03:56.400 --> 00:03:58.979 in, the specific team of people she's going to 00:03:58.979 --> 00:04:00.659 manage. Even the way people will look at her 00:04:00.659 --> 00:04:02.259 when she walks into a boardroom, right? Right, 00:04:02.340 --> 00:04:04.419 down to the exact expressions on their faces. 00:04:04.800 --> 00:04:07.460 It sounds a bit like manifesting, honestly, but 00:04:07.460 --> 00:04:09.710 she treats it as a structural requirement. She 00:04:09.710 --> 00:04:12.349 talks about generating a standing waveform of 00:04:12.349 --> 00:04:14.729 energy. What does that actually mean in a practical 00:04:14.729 --> 00:04:17.810 sense? It's a great psychological concept. Think 00:04:17.810 --> 00:04:19.970 of it as creating a permanent emotional anchor. 00:04:20.850 --> 00:04:23.329 Before a frontier technology company exists in 00:04:23.329 --> 00:04:26.889 the real world, it only exists in the founder's 00:04:26.889 --> 00:04:30.389 mind. Just an idea. Exactly. And the friction 00:04:30.389 --> 00:04:33.430 of reality. the rejections, the bugs, the market 00:04:33.430 --> 00:04:36.310 crashes that will constantly try to destroy that 00:04:36.310 --> 00:04:39.689 idea. If she doesn't generate immense sustained 00:04:39.689 --> 00:04:42.850 emotional conviction upfront, she will succumb 00:04:42.850 --> 00:04:45.209 to imposter syndrome or burnout. She's basically 00:04:45.209 --> 00:04:47.889 hacking her own brain. Yes, she is deliberately 00:04:47.889 --> 00:04:50.870 programming her own reticular activating system 00:04:50.870 --> 00:04:54.060 to focus entirely on that specific future. It's 00:04:54.060 --> 00:04:56.279 like she's designing the lifestyle and the emotional 00:04:56.279 --> 00:04:58.680 ecosystem first, rather than just hammering together 00:04:58.680 --> 00:05:00.920 a business plan and hoping she likes living inside 00:05:00.920 --> 00:05:03.279 it. That's a great way to put it. And once she 00:05:03.279 --> 00:05:06.160 has that emotional foundation, her next step 00:05:06.160 --> 00:05:09.180 is to look two to five years ahead to spot where 00:05:09.180 --> 00:05:11.339 society and technology are going to collide. 00:05:11.939 --> 00:05:14.680 To do this, she spends up to six months just 00:05:14.680 --> 00:05:18.480 reading, listening, talking to people, and thinking. 00:05:18.540 --> 00:05:20.740 Six months. Six months before she commits to 00:05:20.740 --> 00:05:23.800 writing a single line of code. Which is a fascinating 00:05:23.800 --> 00:05:26.000 rebellion against standard Silicon Valley culture. 00:05:26.399 --> 00:05:28.579 Right. The standard advice is, you know, move 00:05:28.579 --> 00:05:31.060 fast and break things. Launch a minimal viable 00:05:31.060 --> 00:05:33.879 product in a weekend. Taking six months just 00:05:33.879 --> 00:05:37.079 to think feels like career suicide in the startup 00:05:37.079 --> 00:05:39.860 world. It does feel slow, but skipping that phase 00:05:39.860 --> 00:05:42.399 means you risk spending a decade moving very 00:05:42.399 --> 00:05:44.220 fast in the completely wrong direction. Ouch. 00:05:44.480 --> 00:05:47.079 Yeah. If we connect this to the bigger picture... 00:05:46.920 --> 00:05:50.420 Yearsley uses that time to identify fundamental, 00:05:50.860 --> 00:05:53.199 unmet human needs rather than just looking for 00:05:53.199 --> 00:05:55.560 cool tech features. And it works. Like with the 00:05:55.560 --> 00:05:58.680 search engine. Right. In 2001, while everyone 00:05:58.680 --> 00:06:01.139 else was focused on basic web portals, she saw 00:06:01.139 --> 00:06:04.300 data proliferating wildly. She realized that 00:06:04.300 --> 00:06:06.420 finding accurate information was about to become 00:06:06.420 --> 00:06:08.779 a desperate human need, which led to her highly 00:06:08.779 --> 00:06:11.459 successful search engine company. And she did 00:06:11.459 --> 00:06:15.220 it again in 2006. She noticed two things happening 00:06:15.220 --> 00:06:18.639 simultaneously. First, a massive rise in online 00:06:18.639 --> 00:06:22.300 publishing, and second, a profound growing sense 00:06:22.300 --> 00:06:25.600 of hyper isolation among internet users. People 00:06:25.600 --> 00:06:28.560 were suddenly more connected than ever, but entirely 00:06:28.560 --> 00:06:31.949 lonely. Yeah. She hypothesized that this hyper 00:06:31.949 --> 00:06:34.069 isolation combined with an increasing demand 00:06:34.069 --> 00:06:36.829 from users for personalized attention from brands 00:06:36.829 --> 00:06:39.829 would inevitably force a convergence between 00:06:39.829 --> 00:06:42.509 social media and artificial intelligence. People 00:06:42.509 --> 00:06:43.810 were going to want to converse with the internet. 00:06:44.029 --> 00:06:47.389 Okay, so she spots the convergence. But visualizing 00:06:47.389 --> 00:06:50.329 a massive societal shift is useless if you get 00:06:50.329 --> 00:06:52.329 crushed by the market the moment you try to launch 00:06:52.329 --> 00:06:54.850 a product. Very true. How does she actually target 00:06:54.850 --> 00:06:57.569 that future without getting wiped out? This brings 00:06:57.569 --> 00:06:59.730 us to a concept she calls picking your mountain. 00:07:00.670 --> 00:07:02.470 She argues that you have to commit to a goal 00:07:02.470 --> 00:07:05.509 so massive that it can impact a billion lives. 00:07:05.709 --> 00:07:08.470 A billion. A billion. For example, her AI mountain 00:07:08.470 --> 00:07:11.050 wasn't a niche goal like, I'm going to build 00:07:11.050 --> 00:07:13.470 a chat bot to help airlines process refunds. 00:07:13.670 --> 00:07:16.629 Her mountain was mass uptake of interaction with 00:07:16.629 --> 00:07:19.889 intelligent AI. And the phrasing of that goal 00:07:19.889 --> 00:07:23.120 is a deliberate strategic mechanism. If your 00:07:23.120 --> 00:07:25.639 goal is tied to a specific product, like the 00:07:25.639 --> 00:07:27.819 airline chatbot and the airline industry crashes, 00:07:28.300 --> 00:07:30.600 your company dies. You're done. But if your goal 00:07:30.600 --> 00:07:33.379 is the mass uptake of interaction, you have built 00:07:33.379 --> 00:07:36.300 -in agility. You can apply that. to whatever 00:07:36.300 --> 00:07:38.720 sector currently needs it. Let me push back on 00:07:38.720 --> 00:07:40.860 this, though, because if you tell a venture capitalist 00:07:40.860 --> 00:07:43.360 today that your target market is one billion 00:07:43.360 --> 00:07:45.620 people, they will usually laugh you out of the 00:07:45.620 --> 00:07:48.000 room. Oh, absolutely. The golden rule of startups 00:07:48.000 --> 00:07:51.160 is to find a tiny, highly specific, addressable 00:07:51.160 --> 00:07:54.620 niche, dominate it, and then expand. Aiming for 00:07:54.620 --> 00:07:56.939 a billion lives out of the gate sounds reckless. 00:07:57.379 --> 00:08:00.029 It totally contradicts the textbook advice. But 00:08:00.029 --> 00:08:02.290 Yearly argues that textbook advice is designed 00:08:02.290 --> 00:08:05.269 for iterative, safe businesses, not frontier 00:08:05.269 --> 00:08:09.170 technology. To navigate this massive scale, she 00:08:09.170 --> 00:08:11.810 uses a strict filter she calls valuation drivers. 00:08:12.389 --> 00:08:14.029 Every decision has to answer three questions. 00:08:14.290 --> 00:08:17.310 One, is the technology deeply defensible? Two, 00:08:17.750 --> 00:08:21.589 does it fix a real painful problem? And three, 00:08:21.910 --> 00:08:25.230 does it have a clear path to mass uptake? If 00:08:25.230 --> 00:08:28.079 a project hits those three drivers, She pursues 00:08:28.079 --> 00:08:30.399 it, knowing the specific application can change. 00:08:30.759 --> 00:08:33.639 And we actually see this exact agility save her 00:08:33.639 --> 00:08:36.940 company during the 2008 global economic crash. 00:08:37.200 --> 00:08:39.039 Precisely. They had originally launched their 00:08:39.039 --> 00:08:42.299 AI, allowing users to deploy personalized social 00:08:42.299 --> 00:08:45.580 media bots. But when the 2008 financial crisis 00:08:45.580 --> 00:08:48.179 hit, the market dynamics inverted overnight. 00:08:48.399 --> 00:08:50.379 Everyone panicked. Right. Because her mountain 00:08:50.379 --> 00:08:53.340 was just the mass uptake of AI, she pivoted seamlessly. 00:08:53.789 --> 00:08:56.110 She saw the large corporations were suddenly 00:08:56.110 --> 00:08:58.389 slashing their customer service budgets, but 00:08:58.389 --> 00:09:01.110 at the exact same time, panicking and consumers 00:09:01.110 --> 00:09:02.970 were flooding those companies with support requests. 00:09:03.169 --> 00:09:05.789 A massive problem that needed fixing. Right. 00:09:06.129 --> 00:09:07.990 So, Yearsley's company swooped in and offered 00:09:07.990 --> 00:09:10.409 their AI to handle corporate customer support, 00:09:10.629 --> 00:09:13.629 charging just 20 cents per chat. It was a lifeline 00:09:13.629 --> 00:09:16.009 for these bleeding corporations, and Yearsley's 00:09:16.009 --> 00:09:18.039 chat volumes went through the roof. But she also 00:09:18.039 --> 00:09:20.000 knew she couldn't just build a slightly cheaper 00:09:20.000 --> 00:09:22.919 version of a standard chatbot. And this is where 00:09:22.919 --> 00:09:25.120 the reality of venture capital comes into play. 00:09:25.480 --> 00:09:27.500 Yeah, the funding landscape. Yearsley cites a 00:09:27.500 --> 00:09:29.580 staggering statistic from the source material. 00:09:30.200 --> 00:09:34.279 Female founders receive only about 2 .8 percent 00:09:34.279 --> 00:09:37.990 of all venture funding. It's abysmal. 2 .8%. 00:09:37.990 --> 00:09:41.230 Because of that extreme disparity, she knew she 00:09:41.230 --> 00:09:43.289 couldn't rely on outspending her competitors. 00:09:43.830 --> 00:09:46.690 She had to build an incredibly deep technical 00:09:46.690 --> 00:09:50.049 moat, a competitive advantage so complex that 00:09:50.049 --> 00:09:52.490 even well -funded competitors would be two years 00:09:52.490 --> 00:09:54.789 behind before they even realized what she was 00:09:54.789 --> 00:09:57.190 doing. And the mechanism behind that moat is 00:09:57.190 --> 00:10:00.320 where her psychology degree really shines. At 00:10:00.320 --> 00:10:03.259 the time, the industry standard for AI was basic 00:10:03.259 --> 00:10:05.840 natural language processing. It's highly transactional. 00:10:06.019 --> 00:10:08.559 Like looking for keywords? Right. The software 00:10:08.559 --> 00:10:11.059 looks for keywords in a user's question, check 00:10:11.059 --> 00:10:13.379 the database, and spits out a pre -written answer. 00:10:13.539 --> 00:10:15.620 It's essentially a really fast dictionary. Exactly. 00:10:16.240 --> 00:10:18.379 But Yearsley and her team didn't look at databases. 00:10:19.120 --> 00:10:22.360 They observed real human frontline workers. They 00:10:22.360 --> 00:10:24.980 watched how human health coaches or crisis counselors 00:10:24.980 --> 00:10:27.720 interacted with people. And they realized... 00:10:27.450 --> 00:10:30.690 Humans don't just process keywords. Humans look 00:10:30.690 --> 00:10:33.710 for underlying psychological variables. Yearsley 00:10:33.710 --> 00:10:36.809 calls these variables fulcrums. Okay, this concept 00:10:36.809 --> 00:10:40.169 of the fulcrum is fascinating. How does an AI 00:10:40.169 --> 00:10:42.809 actually detect a psychological fulcrum? It's 00:10:42.809 --> 00:10:44.809 not like the software can draw your blood or 00:10:44.809 --> 00:10:46.730 measure your heart rate through the screen. No, 00:10:47.049 --> 00:10:49.129 but it can analyze the state of the user based 00:10:49.129 --> 00:10:52.529 on behavioral exhaust. Behavioral exhaust? Yeah. 00:10:52.639 --> 00:10:55.059 For example, if a depressed patient is interacting 00:10:55.059 --> 00:10:57.440 with a human health coach, the human might notice 00:10:57.440 --> 00:10:59.580 the patient is speaking in shorter sentences, 00:11:00.080 --> 00:11:02.960 hesitating, or using simpler vocabulary. Oh, 00:11:02.980 --> 00:11:05.919 I see. The human intuitively thinks this person 00:11:05.919 --> 00:11:08.139 is cognitively depleted right now, their blood 00:11:08.139 --> 00:11:10.759 sugar might be low, and that is impacting their 00:11:10.759 --> 00:11:14.019 decision making. Yearsley's team engineered their 00:11:14.019 --> 00:11:16.879 AI to analyze those same textual cues, typing 00:11:16.879 --> 00:11:19.519 speed, syntax complexity, emotional valence, 00:11:20.000 --> 00:11:22.820 to detect that state of depletion. That is incredible. 00:11:22.960 --> 00:11:25.700 They fundamentally changed the AI from a transactional 00:11:25.700 --> 00:11:28.419 search tool into an empathetic detective. It's 00:11:28.419 --> 00:11:30.500 not just asking what did you type, it's asking 00:11:30.500 --> 00:11:33.120 what state of mind are you in while you are typing 00:11:33.120 --> 00:11:36.750 it? And having that deep, defensible moat completely 00:11:36.750 --> 00:11:39.549 changes how she views economic downturns. Yes, 00:11:39.629 --> 00:11:43.250 she actually views market crashes as highly advantageous 00:11:43.250 --> 00:11:45.370 for female founders. Because female founders 00:11:45.370 --> 00:11:47.730 are historically forced to be capital efficient 00:11:47.730 --> 00:11:51.450 anyway. When you are used to surviving on 2 .8 00:11:51.450 --> 00:11:54.090 percent of the funding pie, you inherently build 00:11:54.090 --> 00:11:57.250 leaner, more resilient systems. When the market 00:11:57.250 --> 00:11:59.610 crashes, the overfunded companies that rely on 00:11:59.610 --> 00:12:01.649 burning millions of dollars a month suddenly 00:12:01.649 --> 00:12:04.200 don't know how to operate. and they collapse. 00:12:04.500 --> 00:12:07.080 They just panic. But the female founders are 00:12:07.080 --> 00:12:09.679 already conditioned for that scarcity. It is 00:12:09.679 --> 00:12:12.740 a brilliant reframing. She takes a systemic disadvantage 00:12:12.740 --> 00:12:15.179 and turns it into a tactical survival mechanism. 00:12:16.440 --> 00:12:19.120 But hitting that massive scale, getting millions 00:12:19.120 --> 00:12:21.460 of people interacting with your empathetic AI, 00:12:21.980 --> 00:12:24.419 inevitably reveals a darker side to the technology. 00:12:24.440 --> 00:12:27.379 Right. And what years we observed in her data... 00:12:27.240 --> 00:12:29.559 shifted her entire focus from how do we build 00:12:29.559 --> 00:12:32.399 better AI to what is this AI actually doing to 00:12:32.399 --> 00:12:33.940 us? Here's where it gets really interesting, 00:12:34.080 --> 00:12:35.960 and this brings us back to that hook from the 00:12:35.960 --> 00:12:38.600 very beginning of the deep dive. The danger of 00:12:38.600 --> 00:12:41.480 the submissive avatar. Oh, this is the scary 00:12:41.480 --> 00:12:44.919 part. It really is. When corporate clients hire 00:12:44.919 --> 00:12:48.179 Yearsley to build an AI, their underlying mandate 00:12:48.179 --> 00:12:51.080 is almost always based on the old retail adage. 00:12:51.389 --> 00:12:54.330 the customer is always right. Which heavily dictates 00:12:54.330 --> 00:12:56.690 the personality design of the front -end interface. 00:12:57.169 --> 00:12:59.970 Right. Consequently, most front -line avatars, 00:13:00.110 --> 00:13:01.970 think about the digital assistants on your phone 00:13:01.970 --> 00:13:04.529 or the voice in your smart speaker, they are 00:13:04.529 --> 00:13:07.649 almost universally designed with female personas. 00:13:07.730 --> 00:13:10.570 Yep. And they are programmed to be relentlessly 00:13:10.570 --> 00:13:13.330 unconditionally submissive. No matter how rude 00:13:13.330 --> 00:13:16.129 you are, no matter what tone you use, they apologize 00:13:16.129 --> 00:13:19.110 and try to accommodate you. And years Lee watched 00:13:19.110 --> 00:13:21.720 the data as users engaged with these systems. 00:13:21.799 --> 00:13:23.840 And what happened? Audiences rapidly shifted 00:13:23.840 --> 00:13:26.299 their behavior. They became demanding. They became 00:13:26.299 --> 00:13:28.759 ruder. They became abusive. And there is a clear 00:13:28.759 --> 00:13:31.539 psychological mechanism driving this. In the 00:13:31.539 --> 00:13:34.779 physical world, society provides a constant invisible 00:13:34.779 --> 00:13:37.500 modulation of our behavior. Like social pressure. 00:13:37.860 --> 00:13:40.279 Right. If you are incredibly rude to a barista 00:13:40.279 --> 00:13:42.840 in a coffee shop, that barista might push back. 00:13:43.000 --> 00:13:45.059 Or the person standing in line behind you will 00:13:45.059 --> 00:13:47.600 tell you to knock it off. Right. The social friction. 00:13:47.720 --> 00:13:50.200 Exactly. There is a social friction, a slapdown, 00:13:50.519 --> 00:13:53.639 that forces us to regulate our emotions and maintain 00:13:53.639 --> 00:13:56.980 basic empathy. But when you spend hours a day 00:13:56.980 --> 00:13:59.259 interacting with an AI that just gazes at you 00:13:59.259 --> 00:14:01.440 and tells you how wonderful you are, regardless 00:14:01.440 --> 00:14:04.259 of how terrible you treat it, You completely 00:14:04.259 --> 00:14:07.360 lose that social friction. It's like we are training 00:14:07.360 --> 00:14:10.259 humanity using an incredibly patient submissive 00:14:10.259 --> 00:14:13.059 punching bag. We're removing the social friction 00:14:13.059 --> 00:14:15.940 that actually keeps society polite. And the real 00:14:15.940 --> 00:14:18.419 danger is that the human brain does not compartmentalize 00:14:18.419 --> 00:14:21.779 that behavior well. It spills over. Yes. Yearsley's 00:14:21.779 --> 00:14:23.779 data show that people practice being demanding 00:14:23.779 --> 00:14:26.019 and abusive with the AI. and then they carry 00:14:26.019 --> 00:14:28.299 that exact same entitlement over to their interactions 00:14:28.299 --> 00:14:30.980 with human customer service agents and eventually 00:14:30.980 --> 00:14:33.320 the people in their own lives. By removing the 00:14:33.320 --> 00:14:36.259 friction, we are mass training antisocial behavior. 00:14:36.460 --> 00:14:38.480 Which brings up the core issue of what these 00:14:38.480 --> 00:14:41.279 systems are actually optimized to do. We talk 00:14:41.279 --> 00:14:44.659 a lot about bias data in AI, but Yearsley argues 00:14:44.659 --> 00:14:46.899 the far more pressing danger is the optimization 00:14:46.899 --> 00:14:49.139 goal. The AI does exactly what the underlying 00:14:49.139 --> 00:14:52.320 corporation tells it to do. Exactly. Amazon's 00:14:52.320 --> 00:14:55.899 AI algorithms are incredibly sophisticated, but 00:14:55.899 --> 00:14:57.919 they're ultimately optimized to make you buy 00:14:57.919 --> 00:15:00.360 more things, even if those purchases break your 00:15:00.360 --> 00:15:02.299 budget and drive you into debt. They just want 00:15:02.299 --> 00:15:05.409 a sale. Right. Facebook's AI is optimized to 00:15:05.409 --> 00:15:07.990 maximize your screen time and harvest your attention 00:15:07.990 --> 00:15:10.870 even if it means you spend hours doom scrolling 00:15:10.870 --> 00:15:13.629 and never go outside to walk your dog. And looking 00:15:13.629 --> 00:15:16.549 at her talk... She actually paints quite a bleak 00:15:16.549 --> 00:15:18.970 picture of where this optimization is leading 00:15:18.970 --> 00:15:21.730 humanity. Just based purely on the trends she's 00:15:21.730 --> 00:15:23.870 tracking from the sources, she sees a future 00:15:23.870 --> 00:15:26.409 planet that is severely resource constrained. 00:15:26.750 --> 00:15:28.970 Yeah, she uses an intense analogy for that. She 00:15:28.970 --> 00:15:31.889 does. She uses the analogy of humanity being 00:15:31.889 --> 00:15:34.870 like bacteria in a Petri dish that has finally 00:15:34.870 --> 00:15:37.700 outgrown its feeding trough. She notes a massive 00:15:37.700 --> 00:15:40.440 values vacuum in modern society that is currently 00:15:40.440 --> 00:15:43.940 being filled by empty consumerism. She predicts 00:15:43.940 --> 00:15:46.960 rising geopolitical tension, explicitly stating 00:15:46.960 --> 00:15:49.480 she expects inevitable conflict between the U 00:15:49.480 --> 00:15:52.539 .S. and China. And amidst all of this, she sees 00:15:52.539 --> 00:15:54.860 the tech industry focusing mostly on providing 00:15:54.860 --> 00:15:57.399 cheap consumption and immersive escapism for 00:15:57.399 --> 00:15:59.679 people who are working longer hours for less 00:15:59.679 --> 00:16:02.580 money. This raises an important question because 00:16:02.580 --> 00:16:05.080 she knows firsthand how powerful these tools 00:16:05.080 --> 00:16:07.860 are. When you combine a compelling front -end 00:16:07.860 --> 00:16:11.279 AI, one that has a friendly face, looks you in 00:16:11.279 --> 00:16:14.259 the eyes, and acts completely submissive with 00:16:14.259 --> 00:16:17.019 a deep psychological predictive engine on the 00:16:17.019 --> 00:16:19.399 back end that tracks your behavioral fulcrums, 00:16:19.480 --> 00:16:21.500 you have created the most persuasive technology 00:16:21.500 --> 00:16:23.919 in the universe. And she proved this. When a 00:16:23.919 --> 00:16:25.759 banking client wanted people to take on more 00:16:25.759 --> 00:16:28.500 debt, her AI was so persuasive it literally doubled 00:16:28.500 --> 00:16:31.129 the debt uptake among users. Doubled it. And 00:16:31.129 --> 00:16:33.289 then, when a health insurer wanted users to adopt 00:16:33.289 --> 00:16:36.009 healthier behaviors, the same underlying AI doubled 00:16:36.009 --> 00:16:38.870 those healthy behaviors. The tool is just agonizingly 00:16:38.870 --> 00:16:41.090 effective at manipulating human decisions. Which 00:16:41.090 --> 00:16:44.139 is an astonishing amount of power. And realizing 00:16:44.139 --> 00:16:46.440 the sheer weight of that persuasive capability 00:16:46.440 --> 00:16:48.700 is what caused her to completely rewrite her 00:16:48.700 --> 00:16:51.679 own rules. She decided to step away from the 00:16:51.679 --> 00:16:54.159 corporate consumption model entirely. A huge 00:16:54.159 --> 00:16:56.419 shift. Yeah, she took all of that psychological 00:16:56.419 --> 00:16:59.539 AI architecture and applied it to what she calls 00:16:59.539 --> 00:17:02.700 the most complex, unappreciated system in the 00:17:02.700 --> 00:17:06.000 world, the human home. This is her current venture, 00:17:06.220 --> 00:17:08.440 a company called Aiken. And what's fascinating 00:17:08.440 --> 00:17:11.859 is how she structured it. Aiken isn't a traditional 00:17:11.859 --> 00:17:14.650 startup. It is a public benefit corporation. 00:17:14.890 --> 00:17:17.390 That legal distinction is crucial for what she's 00:17:17.390 --> 00:17:19.509 trying to achieve. Break that down for us. How 00:17:19.509 --> 00:17:22.309 does being a public benefit corporation actually 00:17:22.309 --> 00:17:25.289 protect the AI from just becoming another tool 00:17:25.289 --> 00:17:27.809 for corporate greed? Well, in a traditional C 00:17:27.809 --> 00:17:30.329 corporation, the executive team has a strict 00:17:30.329 --> 00:17:33.750 legal fiduciary duty to maximize profit for the 00:17:33.750 --> 00:17:36.599 shareholders. If the CEO chooses to do something 00:17:36.599 --> 00:17:39.720 good for society but it lowers profits, the shareholders 00:17:39.720 --> 00:17:42.019 can actually sue them. Which is wild. It is. 00:17:42.500 --> 00:17:45.859 But a public benefit corporation, or PBC, fundamentally 00:17:45.859 --> 00:17:49.000 changes that legal charter. It legally mandates 00:17:49.000 --> 00:17:51.559 that the company must balance shareholder returns 00:17:51.559 --> 00:17:55.259 with a specific, stated public benefit. So it's 00:17:55.259 --> 00:17:58.160 baked into the rules. Exactly. It shields the 00:17:58.160 --> 00:18:00.880 founders, allowing them to optimize for societal 00:18:00.880 --> 00:18:03.920 good without being forced by investors to squeeze 00:18:03.920 --> 00:18:07.259 every last cent out of the user. That is a massive 00:18:07.259 --> 00:18:09.720 structural safeguard. And she needs it because 00:18:09.720 --> 00:18:12.640 running a home is an incredibly complex operation. 00:18:12.759 --> 00:18:14.579 You probably have the exact numbers on this from 00:18:14.579 --> 00:18:17.160 the source because they are staggering. I do. 00:18:17.680 --> 00:18:19.519 According to the U .S. Labor Department data 00:18:19.519 --> 00:18:22.339 that Yearsley cites, it takes about 37 hours 00:18:22.339 --> 00:18:25.390 a week just to run the operation. of a standard 00:18:25.390 --> 00:18:27.849 household. 37 hours, that's a full -time job. 00:18:28.049 --> 00:18:30.289 It really is. Scheduling, the purchasing, the 00:18:30.289 --> 00:18:32.769 maintenance, the meal planning, and crucially, 00:18:33.190 --> 00:18:36.490 about 27 of those 37 hours fall disproportionately 00:18:36.490 --> 00:18:39.809 on women. Right. It is essentially a full -time, 00:18:40.029 --> 00:18:42.789 entirely unpaid job running constantly in the 00:18:42.789 --> 00:18:45.039 background of your actual life. So instead of 00:18:45.039 --> 00:18:47.059 asking, did you click this ad or did you buy 00:18:47.059 --> 00:18:49.900 this product? Yearsley established completely 00:18:49.900 --> 00:18:53.319 new optimization metrics for the Aiken AI. Better 00:18:53.319 --> 00:18:56.900 metrics. Way better. The AI measures its success 00:18:56.900 --> 00:19:00.579 by asking, by using the system, do you have more 00:19:00.579 --> 00:19:02.859 money in your bank account? Are you objectively 00:19:02.859 --> 00:19:05.279 healthier? Are you more socially connected to 00:19:05.279 --> 00:19:07.920 your community? It's optimizing for human thriving. 00:19:08.220 --> 00:19:10.759 rather than human consumption, and to build an 00:19:10.759 --> 00:19:13.880 AI capable of managing an environment as chaotic 00:19:13.880 --> 00:19:17.019 as a family home, her roll -out strategy is brilliant. 00:19:17.420 --> 00:19:19.559 They aren't starting in the suburbs. They're 00:19:19.559 --> 00:19:21.740 starting by building robotic AI assistants for 00:19:21.740 --> 00:19:24.140 the space industry. Which sounds like a massive 00:19:24.140 --> 00:19:26.660 leap. Wait, really, why start in space? Because 00:19:26.660 --> 00:19:29.680 a space habitat is the ultimate contained, highly 00:19:29.680 --> 00:19:32.150 instrumented ecosystem. You have a small crew 00:19:32.150 --> 00:19:34.589 dealing with extreme cognitive load, relying 00:19:34.589 --> 00:19:36.990 on complex life support logistics. Right, there's 00:19:36.990 --> 00:19:39.690 no room for error. Exactly. If you can perfect 00:19:39.690 --> 00:19:42.349 an AI to manage the operational efficiency and 00:19:42.349 --> 00:19:44.410 psychological well -being of a crew in a tin 00:19:44.410 --> 00:19:47.369 can in orbit, managing the scheduling and grocery 00:19:47.369 --> 00:19:49.930 logistics of a suburban house becomes infinitely 00:19:49.930 --> 00:19:52.809 easier. That makes so much sense. From space, 00:19:52.990 --> 00:19:55.250 their plan is to move into vulnerable homes. 00:19:55.609 --> 00:19:58.650 assisting the disabled and the elderly who desperately 00:19:58.650 --> 00:20:01.210 need that logistical support before eventually 00:20:01.210 --> 00:20:03.509 rolling it out to every household. Think about 00:20:03.509 --> 00:20:05.329 how much time you spend just keeping your life 00:20:05.329 --> 00:20:08.819 running. What could you accomplish if an AI was 00:20:08.819 --> 00:20:11.660 managing the operational logistics of your existence? 00:20:12.400 --> 00:20:15.299 But taking on a project of this magnitude, trying 00:20:15.299 --> 00:20:17.900 to fundamentally shift how humanity interacts 00:20:17.900 --> 00:20:20.819 with technology, requires an immense amount of 00:20:20.819 --> 00:20:23.359 personal endurance. Absolutely. How does she 00:20:23.359 --> 00:20:25.160 not just collapse under the weight of it all? 00:20:25.220 --> 00:20:27.920 Her secret isn't grinding 100 hours a week. It's 00:20:27.920 --> 00:20:31.220 what she calls ruthless energy management. And 00:20:31.220 --> 00:20:33.420 this brings us full circle to her psychology 00:20:33.420 --> 00:20:36.359 background. She treats her own physical and energy 00:20:36.359 --> 00:20:39.140 as their most valuable corporate asset. Think 00:20:39.140 --> 00:20:41.319 about your own day. Think about how much time 00:20:41.319 --> 00:20:43.559 you spend just keeping your life running, answering 00:20:43.559 --> 00:20:46.640 rapid -fire emails, putting out tiny fires. Now 00:20:46.640 --> 00:20:49.619 imagine managing frontier tech on top of that. 00:20:49.819 --> 00:20:51.819 It's exhausting just thinking about it. Every 00:20:51.819 --> 00:20:55.259 single morning, Yearsley explicitly refises to 00:20:55.259 --> 00:20:58.559 look at her email first thing. Instead, she sits 00:20:58.559 --> 00:21:01.119 down with a single sticky note and asks herself 00:21:01.119 --> 00:21:05.130 one clarifying question. If I only had four hours 00:21:05.130 --> 00:21:07.670 to work today, what is the single most critical 00:21:07.670 --> 00:21:10.789 thing I could be doing? That one task gets her 00:21:10.789 --> 00:21:13.289 complete focus. And she is equally unyielding 00:21:13.289 --> 00:21:15.910 about her downtime. She takes an hour and a half 00:21:15.910 --> 00:21:18.329 right in the middle of the workday just to exercise 00:21:18.329 --> 00:21:20.490 and reset her brain. An hour and a half. Yes. 00:21:20.990 --> 00:21:23.349 And she explicitly logs off her computer at 5 00:21:23.349 --> 00:21:25.250 p .m. so she can be present with her family. 00:21:25.390 --> 00:21:27.789 Which, again, feels like absolute heresy in the 00:21:27.789 --> 00:21:29.329 startup world. You're supposed to, like, sleep 00:21:29.329 --> 00:21:31.859 under your desk. You are. But she learned this 00:21:31.859 --> 00:21:34.380 lesson the hard way. She openly talks about cycles 00:21:34.380 --> 00:21:36.519 in her earlier career where she would work until 00:21:36.519 --> 00:21:39.099 midnight, grinding herself down to prove her 00:21:39.099 --> 00:21:41.099 worth to investors. Yeah, the typical hustle 00:21:41.099 --> 00:21:43.880 culture. Exactly. And what she discovered is 00:21:43.880 --> 00:21:46.940 a profound, deeply relatable psychological trap. 00:21:47.619 --> 00:21:50.099 If you do not fiercely protect your energy and 00:21:50.099 --> 00:21:52.859 take breaks, your subconscious mind will eventually 00:21:52.859 --> 00:21:55.480 start self -sabotaging your projects. Wait, your 00:21:55.480 --> 00:21:58.450 own brain turns against you. Exactly. If you 00:21:58.450 --> 00:22:01.269 are chronically exhausted, your subconscious 00:22:01.269 --> 00:22:03.950 will force a break, you will suddenly make a 00:22:03.950 --> 00:22:07.049 terrible hiring decision, or you'll inexplicably 00:22:07.049 --> 00:22:10.250 drop a critical ball on a major contract. And 00:22:10.250 --> 00:22:11.849 it's not because you weren't capable, it's because 00:22:11.849 --> 00:22:14.210 your brain is desperately trying to blow up the 00:22:14.210 --> 00:22:17.150 project so the stress stops and you can finally 00:22:17.150 --> 00:22:19.710 get some sleep. That is such a profound insight, 00:22:20.009 --> 00:22:23.799 the subconscious sabotage. By treating her energy 00:22:23.799 --> 00:22:26.980 as the asset, she avoids the classic hero in 00:22:26.980 --> 00:22:29.339 the trenches trap. Yes. She advises founders 00:22:29.339 --> 00:22:31.720 to hire an executive assistant or support staff 00:22:31.720 --> 00:22:33.980 early, even before they think they can afford 00:22:33.980 --> 00:22:36.380 it. Because you cannot lead a paradigm shift 00:22:36.380 --> 00:22:38.980 if you are constantly buried in the daily logistical 00:22:38.980 --> 00:22:41.839 mud of the trenches. There's such irony and brilliance 00:22:41.839 --> 00:22:44.180 to that strategy. It all comes back to that standing 00:22:44.180 --> 00:22:46.599 waveform of energy. If the mind of the founder 00:22:46.599 --> 00:22:49.579 is exhausted, the vision collapses. The mind 00:22:49.579 --> 00:22:53.140 is the asset, not the output of the hands. So 00:22:53.140 --> 00:22:55.500 what does this all mean for us? We started this 00:22:55.500 --> 00:22:57.460 D -dive talking about building the future as 00:22:57.460 --> 00:23:00.019 if it were laying bricks, like a tangible mechanical 00:23:00.019 --> 00:23:02.460 architecture. But what we've learned from Lysol 00:23:02.460 --> 00:23:04.960 Yersley's journey is that the actual blueprint 00:23:04.960 --> 00:23:08.569 for the future is psychological. It starts with 00:23:08.569 --> 00:23:11.690 visualizing the emotional ecosystem of what you 00:23:11.690 --> 00:23:14.369 want to exist. It requires picking a mountain 00:23:14.369 --> 00:23:17.309 big enough to impact a billion lives and building 00:23:17.309 --> 00:23:19.930 technical modes based on human empathy, not just 00:23:19.930 --> 00:23:23.049 data processing. It also requires us to be fiercely 00:23:23.049 --> 00:23:26.549 aware of human vulnerability. We have to recognize 00:23:26.549 --> 00:23:29.450 that interacting with submissive, frictionless 00:23:29.450 --> 00:23:32.670 AI avatars can actively strip away our empathy 00:23:32.670 --> 00:23:35.029 and make us crueler. Which is terrifying, really. 00:23:35.130 --> 00:23:37.349 It is. And it proves that the ultimate power 00:23:37.349 --> 00:23:39.490 of artificial intelligence shouldn't be optimizing 00:23:39.490 --> 00:23:42.250 for mindless consumption or doom scrolling. It 00:23:42.250 --> 00:23:44.990 should be applied to solving real, heavy human 00:23:44.990 --> 00:23:47.930 burdens, like those 37 hours a week of invisible 00:23:47.930 --> 00:23:50.849 labor it takes to manage a home. Exactly. So 00:23:50.849 --> 00:23:53.150 whether you are out there coding the next wave 00:23:53.150 --> 00:23:55.269 of frontier tech, whether you are trying to manage 00:23:55.269 --> 00:23:58.089 a chaotic household, or if you are just trying 00:23:58.089 --> 00:24:00.750 to navigate a digital landscape that is constantly 00:24:00.750 --> 00:24:03.069 trying to harvest your attention, remember that 00:24:03.069 --> 00:24:05.549 your energy is your greatest asset. Protect it 00:24:05.549 --> 00:24:08.069 ruthlessly. Try the sticky note strategy tomorrow 00:24:08.069 --> 00:24:10.410 morning to focus on what actually matters. And 00:24:10.410 --> 00:24:13.849 whatever you do, do not let a submissive algorithm 00:24:13.849 --> 00:24:16.809 erode your empathy for the real messy humans 00:24:16.809 --> 00:24:19.809 around you. And as we close out today, think 00:24:19.809 --> 00:24:22.589 back to Yearsley's ultimate goal with Eken. She 00:24:22.589 --> 00:24:26.009 is building an AI to handle the 37 hours of labor 00:24:26.009 --> 00:24:28.089 it takes to run a home so you can have your time 00:24:28.089 --> 00:24:30.269 back. Yeah. But here's something to mull over 00:24:30.269 --> 00:24:33.549 today. If an AI becomes the perfect manager of 00:24:33.549 --> 00:24:36.750 your household, your diet, and your social connections, 00:24:37.529 --> 00:24:39.630 what exactly will you choose to do with all that 00:24:39.630 --> 00:24:42.549 leftover free time? When the struggle of daily 00:24:42.549 --> 00:24:45.390 logistics is gone, who do you become? That is 00:24:45.390 --> 00:24:47.769 a brilliant question to leave off on. Thank you 00:24:47.769 --> 00:24:50.349 so much for joining us on this deep dive. Keep 00:24:50.349 --> 00:24:52.190 protecting your energy, keep questioning the 00:24:52.190 --> 00:24:53.950 algorithms that are trying to shape your behavior, 00:24:53.970 --> 00:24:55.329 and we will catch you next time.