WEBVTT 00:00:00.000 --> 00:00:03.720 Half the country is using AI daily, yet nearly 00:00:03.720 --> 00:00:05.700 everyone is convinced it's going to ruin society. 00:00:06.820 --> 00:00:10.419 Welcome to the paradox of 2026. It's absolutely 00:00:10.419 --> 00:00:13.300 wild. We're outsourcing our daily emails to models 00:00:13.300 --> 00:00:17.160 we secretly fear. Welcome to the deep dive. Today 00:00:17.160 --> 00:00:19.379 we're unpacking the G7 summit's shift toward 00:00:19.379 --> 00:00:22.320 government AI equity grabs. Right, and we'll 00:00:22.320 --> 00:00:24.120 run through a rapid -fire update of the newest 00:00:24.120 --> 00:00:26.519 models hitting the market. Plus, we'll explore 00:00:26.519 --> 00:00:29.339 that massive disconnect between AI adoption and 00:00:29.339 --> 00:00:32.539 public trust. Now, neutrally speaking, there 00:00:32.539 --> 00:00:34.520 are several different political proposals right 00:00:34.520 --> 00:00:38.640 now. The Treasury Secretary wants to use AI equity 00:00:38.640 --> 00:00:42.240 to fund Trump accounts. Right. But the Commerce 00:00:42.240 --> 00:00:45.340 Secretary prefers routing equity into a sovereign 00:00:45.340 --> 00:00:47.880 wealth fund. And Senator Bernie Sanders proposed 00:00:47.880 --> 00:00:50.600 a totally different law. He wants to give Americans 00:00:50.600 --> 00:00:54.359 direct ownership stakes in top firms. So the 00:00:54.359 --> 00:00:56.880 source facts show different paths to the exact 00:00:56.880 --> 00:00:59.600 same idea. The public sector wants financial 00:00:59.600 --> 00:01:02.679 value directly from private labs. Exactly. The 00:01:02.679 --> 00:01:05.280 specific political flavor just determines where 00:01:05.280 --> 00:01:07.780 the money actually goes. The timing of this is 00:01:07.780 --> 00:01:09.640 incredibly suspect to me, though. Oh, totally. 00:01:09.840 --> 00:01:12.040 These talks happened right before the fable and 00:01:12.040 --> 00:01:14.260 mythos restrictions hit. You know, that massive 00:01:14.260 --> 00:01:17.390 export control clash. with Anthropic. Yeah, Anthropic's 00:01:17.390 --> 00:01:20.290 Dario Amadei and Google DeepMind's Demis Hassabis 00:01:20.290 --> 00:01:23.549 were pushing hard. They want a robust U .S.-led 00:01:23.549 --> 00:01:26.450 coalition on AI rules and chips. Right. They're 00:01:26.450 --> 00:01:28.030 trying to shape the cage before they get locked 00:01:28.030 --> 00:01:31.030 inside. Exactly. The industry's best case scenario 00:01:31.030 --> 00:01:33.750 is getting Fable 5 back online. And they want 00:01:33.750 --> 00:01:35.989 to do it without aggressive power grabs. I mean, 00:01:36.010 --> 00:01:37.890 there's also talk of naturally slowing down the 00:01:37.890 --> 00:01:40.989 release cadence. This whole new model every month 00:01:40.989 --> 00:01:43.469 schedule is just fundamentally broken. It exhausts 00:01:43.469 --> 00:01:47.090 the developers entirely. And it severely confuses 00:01:47.090 --> 00:01:49.709 everyday users trying to keep up. I completely 00:01:49.709 --> 00:01:52.370 agree with that exhaustion. I mean, I still wrestle 00:01:52.370 --> 00:01:55.629 with prompt drift myself. Oh, yeah. We all feel 00:01:55.629 --> 00:01:57.849 that pain. You know the exact feeling. Your go 00:01:57.849 --> 00:02:00.489 -to instructions, just quietly stop working under 00:02:00.489 --> 00:02:02.349 the hood. Yeah, it breaks the user experience 00:02:02.349 --> 00:02:04.790 completely. But zooming out, we're talking about 00:02:04.790 --> 00:02:07.189 frontier models here. Let's define that real 00:02:07.189 --> 00:02:09.530 quick. Frontier models are the most advanced, 00:02:09.669 --> 00:02:12.090 cutting -edge AI systems currently available. 00:02:12.349 --> 00:02:14.310 Right. And the worst -case scenario from the 00:02:14.310 --> 00:02:17.430 sources is pretty dark. Imagine frontier models 00:02:17.430 --> 00:02:20.349 locked entire... behind massive government gates. 00:02:20.669 --> 00:02:23.330 Oh, wow. Yeah. Open source development gets banned 00:02:23.330 --> 00:02:25.770 completely for national security reasons. So 00:02:25.770 --> 00:02:28.310 the market just faces zero actual competition. 00:02:28.389 --> 00:02:30.949 Exactly. We end up paying whatever the big labs 00:02:30.949 --> 00:02:35.009 demand. Or we settle for vastly subpar intelligence 00:02:35.009 --> 00:02:37.889 in our daily tools. Are we heading toward a world 00:02:37.889 --> 00:02:41.610 where AI is treated as a public utility or just 00:02:41.610 --> 00:02:44.639 a government sanctioned monopoly? It looks like 00:02:44.639 --> 00:02:46.919 they want the broad structure of a public utility, 00:02:47.159 --> 00:02:50.460 but the control remains tightly centralized among 00:02:50.460 --> 00:02:53.680 a few key political players. Public utility vibes, 00:02:53.939 --> 00:02:57.379 but entirely controlled by government gatekeepers. 00:02:57.539 --> 00:03:01.060 Beat. Precisely. And that central control is 00:03:01.060 --> 00:03:03.389 what labs are fighting. The government clearly 00:03:03.389 --> 00:03:06.090 wants to slow things down right now. Right. But 00:03:06.090 --> 00:03:08.770 the tech labs are sprinting in the absolute opposite 00:03:08.770 --> 00:03:10.909 direction. They really are moving faster than 00:03:10.909 --> 00:03:14.069 ever before. GPT 5 .6 and the pro version might 00:03:14.069 --> 00:03:16.439 actually arrive next week. Yeah, and early testing 00:03:16.439 --> 00:03:18.639 shows their logical understanding has improved 00:03:18.639 --> 00:03:21.680 massively. But the coding limitations are still 00:03:21.680 --> 00:03:24.080 a stubborn bottleneck. Yeah, front -end and web 00:03:24.080 --> 00:03:25.759 dev are definitely not fully solved yet. No, 00:03:25.759 --> 00:03:27.800 they still struggle with complex architectural 00:03:27.800 --> 00:03:31.099 planning. But Anthropix's Chris Ciari is very 00:03:31.099 --> 00:03:33.780 confident about CloudFable 5. He believes the 00:03:33.780 --> 00:03:36.000 models will return to the market very soon. And 00:03:36.000 --> 00:03:38.759 the tool updates are just as aggressive. Perplexity 00:03:38.759 --> 00:03:41.099 Computer just received a massive brain upgrade 00:03:41.099 --> 00:03:44.039 for max users. It literally makes you want to 00:03:44.039 --> 00:03:45.759 provide more content. text every day. It learns 00:03:45.759 --> 00:03:48.300 your specific workflows perfectly. Google's pushing 00:03:48.300 --> 00:03:50.460 hard into the enterprise ad space too, right? 00:03:50.500 --> 00:03:53.860 Yeah. Google's new Ask Ad Manager agent can troubleshoot 00:03:53.860 --> 00:03:56.979 complex marketing campaigns. But we're also seeing 00:03:56.979 --> 00:03:59.000 some fascinating cultural shifts around this 00:03:59.000 --> 00:04:01.960 technology. Match Group recently released some 00:04:01.960 --> 00:04:06.219 wild data about Tinder. Oh, right. 47 % of U 00:04:06.219 --> 00:04:08.659 .S. singles feel negative about AI in dating. 00:04:09.159 --> 00:04:10.860 Which isn't surprising. You don't want a robot 00:04:10.860 --> 00:04:12.879 writing your romance. Right, but look at the 00:04:12.879 --> 00:04:15.219 massive contradiction in their behavior. Many 00:04:15.219 --> 00:04:18.500 of those same singles still want help with their 00:04:18.500 --> 00:04:21.279 profiles. They hate the concept, but they love 00:04:21.279 --> 00:04:23.980 the practical utility. Exactly. They still want 00:04:23.980 --> 00:04:27.360 AI to help navigate awkward intro chats. On the 00:04:27.360 --> 00:04:29.680 corporate side, companies are getting very creative 00:04:29.680 --> 00:04:32.459 with structure. Snap is spinning off its internal 00:04:32.459 --> 00:04:36.189 AI video team as Dotmo. Which just sounds like 00:04:36.189 --> 00:04:38.589 basic corporate cost cutting on the surface. 00:04:39.069 --> 00:04:41.629 But Snap is keeping a large equity bet in the 00:04:41.629 --> 00:04:44.290 spinoff. Right. They want the upside. Without 00:04:44.290 --> 00:04:46.569 the massive operational burn rate. Right. And 00:04:46.569 --> 00:04:49.509 that burn rate is driven entirely by the hardware 00:04:49.509 --> 00:04:51.769 bottleneck. Absolutely. Amazon is apparently 00:04:51.769 --> 00:04:55.069 considering selling AI chips to outside companies. 00:04:55.230 --> 00:04:57.250 A challenge NVIDIA. Yeah, they're trying to carve 00:04:57.250 --> 00:05:01.350 out their own massive turf. AWS CEO Andy Jassy 00:05:01.350 --> 00:05:05.910 estimates a $50 billion chip run rate. $50 billion? 00:05:06.089 --> 00:05:08.740 Beat. That's a staggering amount of hardware. 00:05:08.899 --> 00:05:10.939 It's almost impossible to comprehend that scale. 00:05:11.120 --> 00:05:14.519 Whoa, imagine scaling to a billion queries. Two 00:05:14.519 --> 00:05:17.339 secs silence. It requires a fundamental rewiring 00:05:17.339 --> 00:05:20.279 of our global energy grid. And that compute power 00:05:20.279 --> 00:05:23.399 fuels incredibly ambitious new startups today. 00:05:23.540 --> 00:05:25.879 Like Odyssey, right? They just raised $310 million. 00:05:26.240 --> 00:05:29.579 Yeah, at a $1 .45 billion valuation. And they're 00:05:29.579 --> 00:05:33.019 building world models. Basically, AI that simulates 00:05:33.019 --> 00:05:35.500 real world physics and complex human behavior. 00:05:35.579 --> 00:05:37.860 Exactly. It's rendering a universe, not just 00:05:37.860 --> 00:05:40.040 writing an essay. So with world models perfectly 00:05:40.040 --> 00:05:42.399 simulating physical reality, is the hardware 00:05:42.399 --> 00:05:45.139 demand ever going to plateau? The physical world 00:05:45.139 --> 00:05:47.800 has infinite resolution and infinite complexity. 00:05:48.579 --> 00:05:51.399 Simulating it perfectly requires exponentially 00:05:51.399 --> 00:05:54.800 growing processing power. It definitely won't 00:05:54.800 --> 00:05:57.660 stop anytime soon. Not until the physical and 00:05:57.660 --> 00:06:00.120 digital worlds completely merge. That's the ultimate 00:06:00.120 --> 00:06:03.000 endgame for these massive frontier labs. But 00:06:03.000 --> 00:06:04.800 let's bring this back down to the human scale. 00:06:05.040 --> 00:06:08.079 Yeah, let's do that. That massive hardware bottleneck 00:06:08.079 --> 00:06:10.800 is exactly why everyday users feel squeezed. 00:06:11.360 --> 00:06:14.120 Which brings us to the microtools patching specific 00:06:14.120 --> 00:06:17.600 daily gaps. These microtools are solving specific 00:06:17.600 --> 00:06:20.420 friction points for regular workers. Seaton's 00:06:20.420 --> 00:06:22.819 2 .0 Mini is a fantastic example of this trend. 00:06:22.959 --> 00:06:25.360 Yeah, it generates high -quality AI videos in 00:06:25.360 --> 00:06:28.160 under a single minute. Faster, cheaper, and they 00:06:28.160 --> 00:06:31.180 offer 30 free credits. It completely democratizes 00:06:31.180 --> 00:06:33.560 video generation for the average creator. And 00:06:33.560 --> 00:06:35.660 then there's Upstream, completely changing the 00:06:35.660 --> 00:06:38.240 daily administrative grind. It sorts your messages 00:06:38.240 --> 00:06:40.519 and drafts your replies automatically. It makes 00:06:40.519 --> 00:06:42.740 checking your email feel light, fast, and fun 00:06:42.740 --> 00:06:45.220 again. Honestly is another tool providing massive 00:06:45.220 --> 00:06:48.339 value to modern digital marketers. It pulls sentiment 00:06:48.339 --> 00:06:51.040 insights directly from Reddit. TikTok and X. 00:06:51.300 --> 00:06:53.939 It also scrapes data from YouTube, Instagram, 00:06:54.019 --> 00:06:57.079 and Facebook automatically. You get unfiltered 00:06:57.079 --> 00:06:59.459 feedback without spending hours reading angry 00:06:59.459 --> 00:07:02.000 comments manually. I want to highlight Tabstack 00:07:02.000 --> 00:07:04.019 specifically for the developers listening today. 00:07:04.279 --> 00:07:07.279 Tabstack is a highly specialized web data and 00:07:07.279 --> 00:07:10.500 automation API. You pass a URL and it spits back 00:07:10.500 --> 00:07:13.759 pristine JSON. Which is just a simple text format 00:07:13.759 --> 00:07:16.899 for storing and moving digital data. It delivers 00:07:16.899 --> 00:07:19.240 highly reliable structured output from a single 00:07:19.240 --> 00:07:22.230 URL. It completely eliminates the parsing errors 00:07:22.230 --> 00:07:24.990 that plague legacy web scrapers. Are these micro 00:07:24.990 --> 00:07:27.589 tools permanently replacing whole software suites 00:07:27.589 --> 00:07:30.829 or just bridging temporary gaps? Right now, they 00:07:30.829 --> 00:07:33.250 act as brilliant little patches. But as they 00:07:33.250 --> 00:07:35.990 interlock seamlessly, they will obsolete massive 00:07:35.990 --> 00:07:38.870 legacy platforms entirely. They're bridging small 00:07:38.870 --> 00:07:41.629 gaps today, replacing entire suites tomorrow. 00:07:41.910 --> 00:07:45.680 Beat. Yeah. So let's recenter on the actual human 00:07:45.680 --> 00:07:47.899 beings using these platforms. Yeah, that's arguably 00:07:47.899 --> 00:07:50.040 the most important question we face today. We've 00:07:50.040 --> 00:07:52.360 covered massive models and the incredible utility 00:07:52.360 --> 00:07:55.279 of microtools. But how do we actually feel about 00:07:55.279 --> 00:07:57.560 integrating this technology? Well, Pew Research 00:07:57.560 --> 00:08:01.639 just dropped its massive 2026 data report. They 00:08:01.639 --> 00:08:05.439 surveyed over 5000 U .S. adults. And the findings 00:08:05.439 --> 00:08:08.920 highlight a very stark and deeply troubling contrast. 00:08:09.300 --> 00:08:11.720 Yeah. Adoption rates are climbing rapidly everywhere. 00:08:12.060 --> 00:08:14.800 Roughly half of all U .S. adults are now using 00:08:14.800 --> 00:08:18.879 AI chat bots. And a solid 25 percent log in every 00:08:18.879 --> 00:08:22.480 single day. But public pessimism is heavily outpacing 00:08:22.480 --> 00:08:25.060 the optimism right now. Nearly 40 percent expect 00:08:25.060 --> 00:08:27.639 AI to make. our society fundamentally worse. 00:08:27.800 --> 00:08:29.759 Exactly. They believe it will degrade society 00:08:29.759 --> 00:08:32.679 over the next two decades. Only 16 % actually 00:08:32.679 --> 00:08:34.460 believe it will make things better. We're rapidly 00:08:34.460 --> 00:08:36.879 adopting a technology we fundamentally fear right 00:08:36.879 --> 00:08:39.500 now. And there's a fascinating demographic quirk 00:08:39.500 --> 00:08:42.240 hidden inside this survey data. We really need 00:08:42.240 --> 00:08:44.480 to point out the massive under 30 divide here. 00:08:44.639 --> 00:08:48.009 Oh, right. Younger users lean on AI the absolute 00:08:48.009 --> 00:08:50.330 hardest. But they're also the absolute biggest 00:08:50.330 --> 00:08:53.549 skeptics in the entire survey. Just 14 % of them 00:08:53.549 --> 00:08:56.029 see a positive long -term payoff. It feels so 00:08:56.029 --> 00:08:58.549 counterintuitive at first glance. They grew up 00:08:58.549 --> 00:09:01.370 entirely immersed in modern digital technology. 00:09:01.590 --> 00:09:05.029 They see the potential for severe societal disruption 00:09:05.029 --> 00:09:07.889 very clearly. Let's look at the actual market 00:09:07.889 --> 00:09:11.289 share reality check here. ChatGPT is still the 00:09:11.289 --> 00:09:14.610 undisputed heavyweight at 44 % usage. Yeah, and 00:09:14.610 --> 00:09:17.549 Google's Gemini currently holds about 24 % of 00:09:17.549 --> 00:09:20.269 the market. But Anthropic's Claude sits way down 00:09:20.269 --> 00:09:23.509 at just 6%. Despite being constantly discussed 00:09:23.509 --> 00:09:26.730 in tech circles, Claude barely registers. It 00:09:26.730 --> 00:09:29.470 absolutely pales in comparison to the massive 00:09:29.470 --> 00:09:32.389 footprint of chat GPT. Why do we keep using these 00:09:32.389 --> 00:09:35.149 tools daily if we genuinely believe they'll eventually 00:09:35.149 --> 00:09:38.679 ruin society? The immediate benefit to your personal 00:09:38.679 --> 00:09:41.299 productivity is highly visible right now. That 00:09:41.299 --> 00:09:43.899 tangible benefit completely overrides abstract 00:09:43.899 --> 00:09:46.620 fears about long -term decay. Because personal 00:09:46.620 --> 00:09:49.320 immediate convenience almost always overrides 00:09:49.320 --> 00:09:52.039 long -term societal anxiety. Exactly. And if 00:09:52.039 --> 00:09:53.519 we look at the core narrative thread of this 00:09:53.519 --> 00:09:56.360 deep dive, we're trapped in a massive three -way 00:09:56.360 --> 00:09:58.860 tug of war. Tech labs are moving at breakneck 00:09:58.860 --> 00:10:02.220 speed to simulate reality perfectly. Right. And 00:10:02.220 --> 00:10:04.539 the government is frantically scrambling to lock 00:10:04.539 --> 00:10:07.460 down control. They literally want the deed to 00:10:07.460 --> 00:10:09.679 the technological foundation. And the broader 00:10:09.679 --> 00:10:12.639 public is caught squarely in the middle. We're 00:10:12.639 --> 00:10:15.320 relying heavily on these tools to survive the 00:10:15.320 --> 00:10:18.019 modern workday. But we fundamentally distrust 00:10:18.019 --> 00:10:21.559 where this technological revolution is actually 00:10:21.559 --> 00:10:24.740 heading. I want to leave you with a final provocative 00:10:24.740 --> 00:10:27.720 thought today. Something for you to deeply mull 00:10:27.720 --> 00:10:30.519 over on your own time. Let's hear it. We discussed 00:10:30.519 --> 00:10:32.879 the government securing equity in these massive 00:10:32.879 --> 00:10:35.820 AI labs. Imagine they eventually distribute those 00:10:35.820 --> 00:10:38.539 financial dividends directly to regular citizens. 00:10:38.779 --> 00:10:42.080 Like a universal basic income. Exactly. If that 00:10:42.080 --> 00:10:44.340 happens, will that pay out finally by the public's 00:10:44.340 --> 00:10:46.759 trust? Or will it just deepen our skepticism 00:10:46.759 --> 00:10:48.919 of the machines running our lives? Thank you 00:10:48.919 --> 00:10:50.919 for taking this deep dive with us. Otero Music.