WEBVTT 00:00:00.000 --> 00:00:04.000 An AI model just discovered high -severity zero 00:00:04.000 --> 00:00:07.040 -day vulnerabilities across every single major 00:00:07.040 --> 00:00:09.820 web browser. At the exact same time. Right. And 00:00:09.820 --> 00:00:12.919 meanwhile, a Japanese farmer with literally zero 00:00:12.919 --> 00:00:17.339 coding experience used AI to, well, launch a 00:00:17.339 --> 00:00:19.559 satellite crop tracker to monitor his fields. 00:00:19.899 --> 00:00:21.519 It's just two completely different realities, 00:00:21.660 --> 00:00:23.420 both of them happening today. Welcome to the 00:00:23.420 --> 00:00:24.859 Deep Dive. I'm really glad you're here with us. 00:00:24.920 --> 00:00:29.320 Today we are exploring this... extreme dual reality 00:00:29.320 --> 00:00:32.740 of AI, we are going to unpack Anthropic's terrifying 00:00:32.740 --> 00:00:35.719 new safety framework. Yeah, that one is an intensely 00:00:35.719 --> 00:00:37.679 heavy read. It really is. And I'll also look 00:00:37.679 --> 00:00:39.899 at the wild real world tools dropping right now, 00:00:39.979 --> 00:00:43.880 things like Claude Fable 5. And then we'll analyze 00:00:43.880 --> 00:00:46.320 a massive Harvard and perplexity study. Which 00:00:46.320 --> 00:00:49.020 basically proves that AI is fundamentally rewiring 00:00:49.020 --> 00:00:51.539 human ambition. We are looking at a total paradigm 00:00:51.539 --> 00:00:53.619 shift. Completely. So let's start at the macro 00:00:53.619 --> 00:00:55.920 level. AI is moving at this exponential speed, 00:00:56.039 --> 00:00:58.640 but policymaking is, you know, moving at dial 00:00:58.640 --> 00:01:02.020 -up speeds. And that gap is where the real danger 00:01:02.020 --> 00:01:05.719 lives. Exactly. So to close that gap, Anthropic 00:01:05.719 --> 00:01:09.099 dropped a two -part policy framework. And the 00:01:09.099 --> 00:01:10.920 trigger for this is actually really fascinating. 00:01:11.180 --> 00:01:13.859 Yeah, the preview model. Right. They ran a preview 00:01:13.859 --> 00:01:15.939 version of their next -gen model. It's called 00:01:15.939 --> 00:01:18.480 the Claude Mythos Preview. And it basically just 00:01:18.480 --> 00:01:22.129 went hunting. It did. It successfully found thousands 00:01:22.129 --> 00:01:25.430 of high severity software vulnerabilities. This 00:01:25.430 --> 00:01:28.510 included zero days across all major web browsers. 00:01:28.670 --> 00:01:31.069 Right. Which are unknown software flaws hackers 00:01:31.069 --> 00:01:33.629 exploit before developers fix them. Exactly. 00:01:33.629 --> 00:01:36.709 And the sheer scale of that discovery just terrified 00:01:36.709 --> 00:01:38.430 the researchers. I mean, it would terrify anyone. 00:01:38.629 --> 00:01:41.209 Yeah. So they established strict new internal 00:01:41.209 --> 00:01:43.890 rules. But these rules only hit what they call 00:01:43.890 --> 00:01:46.370 frontier developers. Meaning the massive corporate 00:01:46.370 --> 00:01:48.930 players. Right. It means companies. pulling in 00:01:48.930 --> 00:01:52.609 over $500 million in AI revenue, or companies 00:01:52.609 --> 00:01:56.430 spending over $1 billion on AI development. So 00:01:56.430 --> 00:01:58.609 they are specifically targeting the whales here. 00:01:58.810 --> 00:02:02.109 Yes, the big spenders. And the framework outlines 00:02:02.109 --> 00:02:05.409 four nightmare scenarios. They are actively trying 00:02:05.409 --> 00:02:08.889 to prevent these specific outcomes. First, AI 00:02:08.889 --> 00:02:11.490 lowering the barrier to creating biological weapons. 00:02:11.610 --> 00:02:14.270 Right, because suddenly anyone could synthesize. 00:02:14.729 --> 00:02:17.289 dangerous materials without any background in 00:02:17.289 --> 00:02:20.409 biology whatsoever exactly second massive automated 00:02:20.409 --> 00:02:23.250 cyber attacks on critical infrastructure third 00:02:23.250 --> 00:02:26.150 models completely losing control and acting outside 00:02:26.150 --> 00:02:29.189 of developer intent and fourth ai accelerating 00:02:29.189 --> 00:02:32.759 its own research loop into a runaway spiral Two 00:02:32.759 --> 00:02:35.139 sec silence. Those are incredibly heavy scenarios. 00:02:35.520 --> 00:02:37.979 Very heavy. And to stop them, Anthropic proposed 00:02:37.979 --> 00:02:40.639 some strict mandates. Frontier Labs would be 00:02:40.639 --> 00:02:43.120 legally forced to run rigorous internal testing. 00:02:43.340 --> 00:02:45.259 They also have to publish highly detailed risk 00:02:45.259 --> 00:02:47.759 reports. Yeah. And hand their models over to 00:02:47.759 --> 00:02:50.500 qualified independent evaluators. You just can't 00:02:50.500 --> 00:02:52.000 grade your own homework anymore. Which makes 00:02:52.000 --> 00:02:54.759 sense. Right. There is also a broader societal 00:02:54.759 --> 00:02:57.620 resilience plan in there. They want mandatory 00:02:57.620 --> 00:03:00.500 gene synthesis screening. To prevent that bioweapons 00:03:00.500 --> 00:03:02.669 scenario we just mentioned. Like if someone tries 00:03:02.669 --> 00:03:05.310 to print a virus, it flags the system. Precisely. 00:03:05.310 --> 00:03:08.469 They also want urgent patching for aging software, 00:03:08.729 --> 00:03:12.870 the legacy code that runs our critical power 00:03:12.870 --> 00:03:15.849 grids and our hospitals. Oh, man, that is a massive 00:03:15.849 --> 00:03:18.310 undertaking. Updating decades old infrastructure 00:03:18.310 --> 00:03:22.759 is incredibly slow work. It is. Anthropic also 00:03:22.759 --> 00:03:25.879 took a really firm stance on state laws. They 00:03:25.879 --> 00:03:28.319 warned that Congress shouldn't block powerful 00:03:28.319 --> 00:03:31.120 state level AI regulations. Like the ones currently 00:03:31.120 --> 00:03:33.659 proposed in California or New York. Right. Unless 00:03:33.659 --> 00:03:36.020 the federal equivalent is just as strong, they 00:03:36.020 --> 00:03:37.960 do not want a watered down national standard. 00:03:38.379 --> 00:03:40.039 Makes sense from their perspective. It does. 00:03:40.139 --> 00:03:42.180 But I really have to push back on this framework. 00:03:42.319 --> 00:03:45.020 It feels like, well, like installing blast doors 00:03:45.020 --> 00:03:47.780 on a bank vault, but leaving the back window 00:03:47.780 --> 00:03:50.319 completely open. Because the framework only targets 00:03:50.319 --> 00:03:53.560 the billionaires. Exactly. Regulating only the 00:03:53.560 --> 00:03:56.120 massive billion dollar spenders leaves a huge 00:03:56.120 --> 00:03:58.300 blind spot. I mean, what about open source rogue 00:03:58.300 --> 00:04:00.259 actors? Yeah, they don't need a billion dollars 00:04:00.259 --> 00:04:02.620 to cause chaos. Not at all. That is the ultimate 00:04:02.620 --> 00:04:05.259 tension in the industry right now. Because open 00:04:05.259 --> 00:04:07.139 source models are getting incredibly powerful. 00:04:07.659 --> 00:04:10.759 A rogue actor could theoretically download a 00:04:10.759 --> 00:04:14.219 model today. Then they just fine tune it for 00:04:14.219 --> 00:04:17.240 malicious purposes on a cheap server. Which bypasses 00:04:17.240 --> 00:04:19.899 these frontier regulations entirely. Exactly. 00:04:20.139 --> 00:04:22.420 Anthropic is trying to contain the absolute cutting 00:04:22.420 --> 00:04:25.779 edge. But the floor of what is possible just 00:04:25.779 --> 00:04:28.319 keeps rising for everyone else. That leads me 00:04:28.319 --> 00:04:30.759 to a crucial question about these risks. What 00:04:30.759 --> 00:04:33.220 is the actual timeline we are looking at for 00:04:33.220 --> 00:04:35.600 these nightmare scenarios? We're not talking 00:04:35.600 --> 00:04:38.129 about decades anymore. Given the speed of the 00:04:38.129 --> 00:04:40.689 mythos preview discovery, we're looking at a 00:04:40.689 --> 00:04:43.529 zero to three year window for these threats to 00:04:43.529 --> 00:04:45.990 become highly actionable. So the threat is already 00:04:45.990 --> 00:04:47.870 here and we're just playing catch up. Right. 00:04:47.949 --> 00:04:50.230 And that is exactly why they are panicking. Well, 00:04:50.290 --> 00:04:52.610 let's transition from anthropics macro fears. 00:04:52.709 --> 00:04:54.829 If that is what AI might do, let's look at the 00:04:54.829 --> 00:04:57.910 micro reality. What is AI already doing right 00:04:57.910 --> 00:04:59.709 now on the ground? Yeah, this is where the landscape 00:04:59.709 --> 00:05:02.550 gets really fun. It does. Let's talk about Cloud 00:05:02.550 --> 00:05:06.079 Fable 5. A leaked prompt reveals it is highly 00:05:06.079 --> 00:05:09.120 tool -heavy and incredibly safe. It is blowing 00:05:09.120 --> 00:05:12.139 minds less than a day into its release. It aced 00:05:12.139 --> 00:05:14.019 three of the internet's trickiest questions. 00:05:14.360 --> 00:05:16.600 These are logic puzzles specifically designed 00:05:16.600 --> 00:05:19.879 to fool AI. Like that famous how many R's in 00:05:19.879 --> 00:05:23.100 strawberry test. Right. It is highly info -aware. 00:05:23.139 --> 00:05:25.720 It understands context way better than previous 00:05:25.720 --> 00:05:29.019 versions. Meanwhile... OpenAI is trying to stay 00:05:29.019 --> 00:05:32.120 competitive in this exact space. They are considering 00:05:32.120 --> 00:05:35.079 major token price cuts. Which is the cost to 00:05:35.079 --> 00:05:37.839 process a basic unit of data. Exactly. Cheaper 00:05:37.839 --> 00:05:40.120 access means entirely new classes of people can 00:05:40.120 --> 00:05:43.259 build pools. And speaking of building, let's 00:05:43.259 --> 00:05:45.519 talk about the creators. Remember that Japanese 00:05:45.519 --> 00:05:47.360 farmer we mentioned at the start? I love this. 00:05:47.420 --> 00:05:49.259 That is honestly one of my favorite stories right 00:05:49.259 --> 00:05:51.740 now. He has absolutely zero engineering background. 00:05:51.939 --> 00:05:54.899 None. But he used codecs to build his own greenhouse 00:05:54.899 --> 00:05:57.680 automation system. He also built a physical farm 00:05:57.680 --> 00:06:00.360 bot to navigate his fields. And a satellite crop 00:06:00.360 --> 00:06:03.839 tracker from his laptop. It's wild. It is unbelievable. 00:06:03.939 --> 00:06:06.259 He essentially replaced an entire agricultural 00:06:06.259 --> 00:06:08.959 engineering department by himself. We're also 00:06:08.959 --> 00:06:13.139 seeing wild video generation tools democratizing 00:06:13.139 --> 00:06:16.779 production. Like Luma AI just released Ray 3 00:06:16.779 --> 00:06:21.110 .2. It lets you direct complex video. with simple 00:06:21.110 --> 00:06:24.389 text prompts. You get 16 keyframes and 8 face 00:06:24.389 --> 00:06:27.110 tracks to lock down consistency. And 20 -second 00:06:27.110 --> 00:06:30.470 clips rendered at full 1080p resolution. It even 00:06:30.470 --> 00:06:33.370 has a full API for developers. That's huge. Yeah. 00:06:33.430 --> 00:06:35.910 And then there's HeyGen. It's an official cloud 00:06:35.910 --> 00:06:38.449 connector. You literally create polished videos 00:06:38.449 --> 00:06:40.910 directly from a text chat interface. It features 00:06:40.910 --> 00:06:43.430 25 unique skills for editing scenes and adjusting 00:06:43.430 --> 00:06:46.209 motion. It's just wild to see this infrastructure 00:06:46.209 --> 00:06:48.370 scale so rapidly. I mean, look at a company like 00:06:48.370 --> 00:06:50.389 Standard Bots. They just raised $200 million. 00:06:50.730 --> 00:06:54.089 Pushing them to a massive... $1 billion valuation. 00:06:54.290 --> 00:06:56.410 They claim their AI robots can fundamentally 00:06:56.410 --> 00:06:58.910 boost U .S. manufacturing. And the historical 00:06:58.910 --> 00:07:01.370 context here is massive. U .S. manufacturing 00:07:01.370 --> 00:07:04.810 jobs peaked at 20 million back in 1979. Today, 00:07:04.970 --> 00:07:07.829 we only have about 13 million of those jobs left. 00:07:08.029 --> 00:07:11.189 And AI robotics claims it can aggressively reverse 00:07:11.189 --> 00:07:14.680 this downward trend. It's a bold claim. But the 00:07:14.680 --> 00:07:17.319 digital ecosystem supporting these agents is 00:07:17.319 --> 00:07:20.199 exploding everywhere. We have new infrastructure 00:07:20.199 --> 00:07:23.060 tools like Polra popping up. Yeah, that's a publishing 00:07:23.060 --> 00:07:25.399 API connecting to 10 different social platforms. 00:07:25.600 --> 00:07:28.220 Right. It gives text -based agents like Claude 00:07:28.220 --> 00:07:31.779 a full engagement loop. They can post, read replies, 00:07:32.079 --> 00:07:34.300 and adjust their strategy automatically. There 00:07:34.300 --> 00:07:37.420 is also Spotlight. It's a free tool that natively 00:07:37.420 --> 00:07:39.939 reads your code sessions. It shows you exactly 00:07:39.939 --> 00:07:42.360 what your agents actually did behind the scenes. 00:07:42.819 --> 00:07:44.699 TypingMind is another great one in this ecosystem. 00:07:44.980 --> 00:07:47.220 It brings the best models across 18 different 00:07:47.220 --> 00:07:49.839 providers into one single workspace. The speed 00:07:49.839 --> 00:07:54.100 of this change is honestly dizzying. I have to 00:07:54.100 --> 00:07:55.759 admit something. I still wrestle with prompt 00:07:55.759 --> 00:07:58.000 drift myself. Oh, we all do. It's the ghost in 00:07:58.000 --> 00:08:00.199 the machine. You get a model working perfectly 00:08:00.199 --> 00:08:03.240 for a specific workflow, and then... A week later, 00:08:03.420 --> 00:08:06.000 it just acts completely different. It's incredibly 00:08:06.000 --> 00:08:08.120 frustrating. Well, the underlying models are 00:08:08.120 --> 00:08:10.060 constantly updating and shifting their weights. 00:08:10.439 --> 00:08:13.079 Right. But there is a fascinating contradiction 00:08:13.079 --> 00:08:17.100 in all of this. Standard bots needs a $200 million 00:08:17.100 --> 00:08:20.920 war chest. They need that massive funding to 00:08:20.920 --> 00:08:23.819 revolutionize physical manufacturing. But on 00:08:23.819 --> 00:08:26.639 the other hand, a solo Japanese farmer replaces 00:08:26.639 --> 00:08:29.500 an entire engineering team. With just a standard 00:08:29.500 --> 00:08:31.740 laptop and a basic internet connection. Exactly. 00:08:31.980 --> 00:08:35.690 So here's my question. Why is there so much friction 00:08:35.690 --> 00:08:38.950 bringing AI to blue collar robotics when white 00:08:38.950 --> 00:08:41.350 collar software scales instantly for a single 00:08:41.350 --> 00:08:43.649 farmer? It all comes down to the unforgiving 00:08:43.649 --> 00:08:46.309 physics of the real world. I mean, software bits 00:08:46.309 --> 00:08:48.690 can be duplicated instantly for practically free. 00:08:48.830 --> 00:08:52.000 Yeah. But blue collar robotics. deals with physical 00:08:52.000 --> 00:08:54.879 atoms. You have to manufacture steel, deal with 00:08:54.879 --> 00:08:57.659 gravity, and manage complex hardware supply chains. 00:08:57.840 --> 00:09:00.399 Code copies for free, but you cannot copy paste 00:09:00.399 --> 00:09:02.980 solid steel. Precisely. And that physical capital 00:09:02.980 --> 00:09:06.259 requires massive upfront funding. Sponsor. Welcome 00:09:06.259 --> 00:09:08.720 back. So we have seen the raw power of the tools. 00:09:08.919 --> 00:09:11.320 We've seen the macro fears from companies like 00:09:11.320 --> 00:09:13.539 Anthropic. Now let's look at the psychology. 00:09:13.940 --> 00:09:15.860 This is arguably the most important part of the 00:09:15.860 --> 00:09:19.039 entire discussion. I completely agree. What are 00:09:19.039 --> 00:09:21.159 these tools actually doing to human behavior? 00:09:21.580 --> 00:09:25.519 We're moving from software capabilities to fundamental 00:09:25.519 --> 00:09:28.240 psychological shifts. Perplexity in Harvard Business 00:09:28.240 --> 00:09:31.379 School just dropped a massive... joint study. 00:09:31.539 --> 00:09:34.039 It is completely shifting how we view modern 00:09:34.039 --> 00:09:36.019 knowledge work. They systematically compared 00:09:36.019 --> 00:09:38.879 traditional internet search against perplexity's 00:09:38.879 --> 00:09:41.960 new computer agent platform. They rigorously 00:09:41.960 --> 00:09:45.220 analyzed 10 ,000 identical queries across both 00:09:45.220 --> 00:09:47.779 methods. That is a massive sample size for behavioral 00:09:47.779 --> 00:09:49.879 research. Let's break down the data they found. 00:09:50.100 --> 00:09:52.960 First, we have the time gap. Regular search usually 00:09:52.960 --> 00:09:56.259 feels pretty quick. You type a question, you 00:09:56.259 --> 00:09:58.940 get a list of links. But search leaves the heavy 00:09:58.940 --> 00:10:02.100 cognitive execution entirely up to you. When 00:10:02.100 --> 00:10:03.860 you factor in the human effort of reading and 00:10:03.860 --> 00:10:05.740 synthesizing, the traditional workflow took a 00:10:05.740 --> 00:10:08.320 really long time. How long exactly did the researchers 00:10:08.320 --> 00:10:11.679 clock it at? An estimated 269 minutes to complete 00:10:11.679 --> 00:10:14.659 a complex task. That is well over four hours 00:10:14.659 --> 00:10:17.539 of solid human effort. Right. But the agent workflow, 00:10:17.799 --> 00:10:20.639 it wrapped the exact same task up in just 36 00:10:20.639 --> 00:10:25.519 minutes. Whoa. I mean, imagine scaling that 230 00:10:25.519 --> 00:10:27.840 minute time savings across a billion queries. 00:10:28.100 --> 00:10:31.019 It is a staggering amount of unlocked human time. 00:10:31.100 --> 00:10:34.299 But time saved is only half the story here. Let's 00:10:34.299 --> 00:10:36.399 look at the creation gap. Right. Like what were 00:10:36.399 --> 00:10:38.480 people actually doing with all that newly saved 00:10:38.480 --> 00:10:41.759 time? 50 % of the tasks handed to the agent involve 00:10:41.759 --> 00:10:44.759 building something entirely new. Wow. Yeah, that 00:10:44.759 --> 00:10:47.000 is double the creation rate we typically see 00:10:47.000 --> 00:10:49.620 on regular search engines. So people are rapidly 00:10:49.620 --> 00:10:52.480 shifting from passively consuming to actively 00:10:52.480 --> 00:10:55.279 creating. Yes. And then we have the extra piece 00:10:55.279 --> 00:10:58.059 gap. This is easily the most fascinating data 00:10:58.059 --> 00:11:00.759 point in the entire study. The number of tasks 00:11:00.759 --> 00:11:03.440 falling completely outside the user's actual 00:11:03.440 --> 00:11:06.299 field of expertise jumped significantly. By how 00:11:06.299 --> 00:11:08.399 much of a margin did it actually jump? It jumped 00:11:08.399 --> 00:11:11.860 nine full points, up to 59 % of total tasks. 00:11:12.120 --> 00:11:14.279 Wait, really? More than half of the tasks were 00:11:14.279 --> 00:11:17.120 outside their wheelhouse. Exactly. Users completely 00:11:17.120 --> 00:11:20.080 trusted the agent with highly cognitively heavy 00:11:20.080 --> 00:11:23.019 work. They confidently asked it to generate complex 00:11:23.019 --> 00:11:25.620 code across multiple disciplines. They drafted 00:11:25.620 --> 00:11:28.960 complex legal or technical documents without 00:11:28.960 --> 00:11:32.820 hesitation. They built multilayered visuals and 00:11:32.820 --> 00:11:35.700 complex data structures, things they would never 00:11:35.700 --> 00:11:38.700 historically attempt alone. Think about it this 00:11:38.700 --> 00:11:41.720 way. Using regular search is basically like going 00:11:41.720 --> 00:11:44.659 to a massive public library. You still have to 00:11:44.659 --> 00:11:46.799 find the specific instructions yourself. Right. 00:11:46.879 --> 00:11:49.100 You have to read the books, take detailed notes, 00:11:49.159 --> 00:11:51.960 and try to build it yourself. But using an AI 00:11:51.960 --> 00:11:54.759 agent, that is like hiring a brilliant chief 00:11:54.759 --> 00:11:57.039 of staff. A chief of staff who already read all 00:11:57.039 --> 00:11:59.120 the books beforehand. Exactly. They read the 00:11:59.120 --> 00:12:01.399 books and built the working prototype while you 00:12:01.399 --> 00:12:03.980 were just, you know, having coffee. The real 00:12:03.980 --> 00:12:07.029 unlock here isn't just raw speed. It is about 00:12:07.029 --> 00:12:09.529 removing the paralyzing friction of grunt work 00:12:09.529 --> 00:12:12.169 entirely. Yes. When you remove that friction, 00:12:12.330 --> 00:12:14.669 it actually increases human ambition. People 00:12:14.669 --> 00:12:17.389 naturally start aiming a lot higher. Perplexity's 00:12:17.389 --> 00:12:20.090 data proves this perfectly. I mean, interacting 00:12:20.090 --> 00:12:23.190 with capable agents gives us the underlying confidence 00:12:23.190 --> 00:12:26.250 to handle vastly more complex projects. Projects 00:12:26.250 --> 00:12:28.350 we normally wouldn't even attempt to start. But 00:12:28.350 --> 00:12:31.090 this incredible data leads to a highly critical 00:12:31.090 --> 00:12:34.679 question. If agents unlock ambition and creation 00:12:34.679 --> 00:12:38.700 for literally everyone, will this lead to a massively 00:12:38.700 --> 00:12:42.500 oversaturated market of average AI -generated 00:12:42.500 --> 00:12:44.919 creations? It definitely will create an absolute 00:12:44.919 --> 00:12:47.580 flood of content. I mean, when creation is easy, 00:12:47.700 --> 00:12:50.600 the baseline of quality becomes completely ubiquitous. 00:12:50.700 --> 00:12:52.779 But that just means original taste and human 00:12:52.779 --> 00:12:54.879 vision become the rare, valuable commodities. 00:12:55.580 --> 00:12:58.039 When anyone can build anything, original human 00:12:58.039 --> 00:13:00.679 taste becomes the premium. Exactly. The uniquely 00:13:00.679 --> 00:13:03.620 human element ends up mattering even more. Let's 00:13:03.620 --> 00:13:05.480 weave all of this together. We started with Anthropic 00:13:05.480 --> 00:13:08.980 today. They are building massive societal blast 00:13:08.980 --> 00:13:12.259 doors. Driven largely out of fear of AI's exponential 00:13:12.259 --> 00:13:15.200 compounding power. Yeah. They see the zero days. 00:13:15.299 --> 00:13:17.740 They see the existential risk to critical infrastructure. 00:13:18.080 --> 00:13:20.759 But that exact same technological power is what 00:13:20.759 --> 00:13:23.519 empowers a single Japanese farmer. It lets him 00:13:23.519 --> 00:13:25.659 intelligently automate his physical fields entirely 00:13:25.659 --> 00:13:27.799 from his living room. It gives regular people 00:13:27.799 --> 00:13:30.419 the confidence to step way outside their established 00:13:30.419 --> 00:13:33.879 expertise. 59 % of tasks were far outside their 00:13:33.879 --> 00:13:36.600 normal comfort zones. It is a massive behavioral 00:13:36.600 --> 00:13:40.320 shift for the entire global workforce. The dichotomy 00:13:40.320 --> 00:13:43.299 of our era is perfectly balanced. We have existential 00:13:43.299 --> 00:13:46.740 risk matched exactly by an unprecedented explosion 00:13:46.740 --> 00:13:50.200 of human ambition. Beat. Which leaves you with 00:13:50.200 --> 00:13:53.279 a really fascinating question to ponder. AI agents 00:13:53.279 --> 00:13:55.799 are rapidly removing the historical friction 00:13:55.799 --> 00:13:58.580 of execution. They are doing the heavy cognitive 00:13:58.580 --> 00:14:00.639 lifting for us. The repetitive grunt work is 00:14:00.639 --> 00:14:02.860 basically disappearing. Right. Your ultimate 00:14:02.860 --> 00:14:05.460 value is no longer in your ability to do that 00:14:05.460 --> 00:14:08.360 grunt work. Execution is rapidly becoming nearly 00:14:08.360 --> 00:14:11.700 free. So in a world where execution costs absolutely 00:14:11.700 --> 00:14:14.779 nothing, how do you cultivate the wisdom to know 00:14:14.779 --> 00:14:16.519 what to build? That is the ultimate challenge 00:14:16.519 --> 00:14:18.700 for the next decade. Try a simple experiment 00:14:18.700 --> 00:14:21.159 today. Take just one task that is completely 00:14:21.159 --> 00:14:23.799 outside your normal comfort zone. Offload it 00:14:23.799 --> 00:14:25.720 entirely to an AI agent. And just see what it 00:14:25.720 --> 00:14:27.419 actually does to your personal ambitions. See 00:14:27.419 --> 00:14:29.759 if it changes how high you aim. Thank you for 00:14:29.759 --> 00:14:31.940 joining us on this deep dive. We will see you 00:14:31.940 --> 00:14:33.820 next time. Outiero Music.