WEBVTT 00:00:00.000 --> 00:00:02.319 Imagine an astronaut. You know, they're floating 00:00:02.319 --> 00:00:05.040 up there in zero gravity. It looks so peaceful, 00:00:05.179 --> 00:00:07.000 right? There's no resistance, no heavy lifting. 00:00:07.179 --> 00:00:09.320 But while they're floating there, just enjoying 00:00:09.320 --> 00:00:15.099 the view, something dangerous is quietly happening 00:00:15.099 --> 00:00:18.199 inside their body. Without gravity to push against 00:00:18.199 --> 00:00:21.859 the muscles, they start to atrophy. They can 00:00:21.859 --> 00:00:25.079 lose up to 20 % of their mass. Now, imagine that 00:00:25.079 --> 00:00:28.329 exact same process. But it's happening to your 00:00:28.329 --> 00:00:30.370 brain. That's such a vivid image, isn't it? But 00:00:30.370 --> 00:00:32.950 it's also a little deceptive. How so? Well, with 00:00:32.950 --> 00:00:35.070 zero gravity, you know you're floating. You know 00:00:35.070 --> 00:00:37.130 something is wrong. But with this kind of cognitive 00:00:37.130 --> 00:00:40.009 atrophy, people feel productive. They feel like 00:00:40.009 --> 00:00:41.890 they're sprinting. Right. And that's the really 00:00:41.890 --> 00:00:44.729 dangerous part. You treat AI as a shortcut. You 00:00:44.729 --> 00:00:46.350 get the result and you think, wow, I won the 00:00:46.350 --> 00:00:48.710 race. But really, you just skip the workout entirely. 00:00:49.350 --> 00:00:52.210 Welcome to the deep dive. It's Friday, January 00:00:52.210 --> 00:00:55.609 30th, 2026. We are looking at a guide that just 00:00:55.609 --> 00:00:58.810 came out this morning from AI Fire by Max Ann, 00:00:58.990 --> 00:01:03.149 and it tackles this exact idea, this training 00:01:03.149 --> 00:01:05.609 your replacement trap. And I have to say, reading 00:01:05.609 --> 00:01:10.510 this, it made me a little uncomfortable. It suggests 00:01:10.510 --> 00:01:13.430 that by outsourcing the friction of work, we 00:01:13.430 --> 00:01:16.170 aren't just becoming efficient, we are becoming... 00:01:17.239 --> 00:01:19.540 It's a harsh reality check for sure. We have 00:01:19.540 --> 00:01:21.560 this built -in assumption that easier is better. 00:01:22.079 --> 00:01:24.140 But the core argument Max is making here is that 00:01:24.140 --> 00:01:26.420 most professionals, I mean smart people, people 00:01:26.420 --> 00:01:28.920 listening right now, are using AI to outsource 00:01:28.920 --> 00:01:30.739 the actual thinking. Handing over the cognitive 00:01:30.739 --> 00:01:33.000 load. Exactly. So today we're going to walk through 00:01:33.000 --> 00:01:35.200 a framework to, well, to reverse this, we need 00:01:35.200 --> 00:01:37.260 to talk about intelligent laziness, which sounds 00:01:37.260 --> 00:01:39.280 like a total contradiction. It does. Then we 00:01:39.280 --> 00:01:40.500 need to break down something called the D -RAG 00:01:40.500 --> 00:01:42.540 framework for delegation. And then we have to 00:01:42.540 --> 00:01:44.519 talk about agents because that whole landscape 00:01:44.519 --> 00:01:46.700 has just changed completely in the last six months. 00:01:47.260 --> 00:01:50.519 And finally, my favorite part, this concept of 00:01:50.519 --> 00:01:53.260 the intelligent gym. Because if you're not using 00:01:53.260 --> 00:01:55.299 AI to make your brain sweat, you're probably 00:01:55.299 --> 00:01:58.060 using it wrong. OK, so let's start with the trap, 00:01:58.159 --> 00:02:01.019 the zero gravity effect. The source argues that, 00:02:01.019 --> 00:02:03.459 you know, by just copy pasting prompts and taking 00:02:03.459 --> 00:02:06.040 the first answer, we're weakening our analytical 00:02:06.040 --> 00:02:09.099 muscles. Yeah. But let me play devil's advocate 00:02:09.099 --> 00:02:12.139 here. Isn't the whole point of technology to 00:02:12.139 --> 00:02:14.560 remove friction? If I can get an answer in 10 00:02:14.560 --> 00:02:17.180 seconds, why should I struggle for an hour? That 00:02:17.180 --> 00:02:19.580 is the billion dollar question, right? And the 00:02:19.580 --> 00:02:22.909 answer, really. lies in what you're struggling 00:02:22.909 --> 00:02:26.009 with. The guide introduces this idea of two curves, 00:02:26.189 --> 00:02:28.710 curve one and curve two. If you remove the friction 00:02:28.710 --> 00:02:31.370 from everything, you just, you flatline. But 00:02:31.370 --> 00:02:33.689 if you remove friction from the low value stuff 00:02:33.689 --> 00:02:35.889 to free you up to obsess over the high value 00:02:35.889 --> 00:02:38.050 stuff, well, that's where you win. Okay, so break 00:02:38.050 --> 00:02:40.150 down curve one for me. Max calls this capped 00:02:40.150 --> 00:02:42.349 payoffs. Yeah, this is what he calls the lazy 00:02:42.349 --> 00:02:44.849 zone. Just think about your week. How many tasks 00:02:44.849 --> 00:02:47.669 do you do where, honestly, trying harder just 00:02:47.669 --> 00:02:50.360 doesn't matter? formatting a slide deck. Perfect. 00:02:50.759 --> 00:02:53.060 Rewriting a status update. Scheduling meetings. 00:02:53.360 --> 00:02:55.800 The source mentions a study from HBR, I think 00:02:55.800 --> 00:02:59.199 it was, that the average CEO spends 72 % of their 00:02:59.199 --> 00:03:02.919 time in meetings and most of them just... Don't 00:03:02.919 --> 00:03:04.979 move the needle. Right. And there's a psychological 00:03:04.979 --> 00:03:07.759 trap here. It's called completion bias. You spend 00:03:07.759 --> 00:03:09.960 two hours making the font perfect on a slide 00:03:09.960 --> 00:03:12.300 nobody's even going to read. Your brain gives 00:03:12.300 --> 00:03:14.199 you a little hit of dopamine because you finished 00:03:14.199 --> 00:03:17.000 something. But you added zero value. Zero. So 00:03:17.000 --> 00:03:19.379 Max suggests this concept from Herbert Simon, 00:03:19.580 --> 00:03:22.759 a Nobel laureate, called satisficing. Satisficing. 00:03:22.759 --> 00:03:26.240 It's a mashup of satisfy and suffice. It basically 00:03:26.240 --> 00:03:30.030 means. do just enough the test question you should 00:03:30.030 --> 00:03:32.870 ask yourself is what actually changes if this 00:03:32.870 --> 00:03:35.569 task is perfect instead of just good enough and 00:03:35.569 --> 00:03:37.550 if the answer is nothing then you're in curve 00:03:37.550 --> 00:03:40.849 one be lazy automate it use ai to get it done 00:03:40.849 --> 00:03:42.990 and move on don't waste your energy there which 00:03:42.990 --> 00:03:45.530 of course brings us to curve two the uncapped 00:03:45.530 --> 00:03:48.210 payoffs and this is the obsession zone this is 00:03:48.210 --> 00:03:51.610 strategy hiring Designing a new product. This 00:03:51.610 --> 00:03:53.449 is the kind of work where the effort to reward 00:03:53.449 --> 00:03:56.969 ratio is. It's exponential. The example they 00:03:56.969 --> 00:03:59.289 gave was skeeve jobs obsessing over the circuit 00:03:59.289 --> 00:04:02.990 boards inside the very first iPhone. You know, 00:04:02.990 --> 00:04:05.150 stuff nobody was ever going to see. Exactly. 00:04:05.270 --> 00:04:08.030 That wasn't efficient. It was, frankly, insane. 00:04:08.729 --> 00:04:11.270 But that obsession is what built the trust in 00:04:11.270 --> 00:04:14.439 the brand. If you use AI to automate the good 00:04:14.439 --> 00:04:17.500 enough stuff, you literally buy back the mental 00:04:17.500 --> 00:04:19.860 energy to be insane about the stuff that actually 00:04:19.860 --> 00:04:22.300 matters. So it's not about working less. It's 00:04:22.300 --> 00:04:24.959 about reallocating the suffering. I like that. 00:04:25.040 --> 00:04:27.199 Yes. Reallocate the suffering to where it pays 00:04:27.199 --> 00:04:30.019 dividends. OK, so let's get practical. If we're 00:04:30.019 --> 00:04:32.560 pushing all this curve one work to AI, how do 00:04:32.560 --> 00:04:35.079 we actually do it without losing quality? The 00:04:35.079 --> 00:04:39.120 source uses this acronym DRAG, D -R -A -G. And 00:04:39.120 --> 00:04:41.180 normally I hate acronyms. They feel so corporate. 00:04:41.300 --> 00:04:44.279 But this one, this one felt descriptive. It is. 00:04:44.300 --> 00:04:46.819 It frames the work that literally drags you down. 00:04:47.040 --> 00:04:49.139 Drafting, research analysis and grunt work. Max 00:04:49.139 --> 00:04:51.379 claims 70 to 80 percent of repetitive work fits 00:04:51.379 --> 00:04:53.279 in there. So let's strip this down. D is for 00:04:53.279 --> 00:04:55.339 drafting. You're saying this is the cure for 00:04:55.339 --> 00:04:57.779 writer's block. Right. And this is where so many 00:04:57.779 --> 00:04:59.660 people get it wrong. They ask the AI to write 00:04:59.660 --> 00:05:02.439 the final email. No, you ask AI to break the 00:05:02.439 --> 00:05:05.060 inertia. You say, act as a senior product marketer, 00:05:05.079 --> 00:05:07.959 draft a launch email based on these three bullet 00:05:07.959 --> 00:05:11.319 points. It's not going to be perfect. It might 00:05:11.319 --> 00:05:14.560 be a C plus. But editing a C plus is way faster 00:05:14.560 --> 00:05:17.279 than staring at a blank page. Infinitely faster. 00:05:17.660 --> 00:05:20.519 You have research and analysis. The R and the 00:05:20.519 --> 00:05:22.379 A. I like to group these two because they're 00:05:22.379 --> 00:05:24.680 really about data compression. We are all just 00:05:24.680 --> 00:05:27.220 drowning in information. We have like infobesity. 00:05:27.660 --> 00:05:30.259 Infobesity. Yeah. Too much intake, not enough 00:05:30.259 --> 00:05:33.620 processing. You can use an AI to crawl 20 websites, 00:05:33.959 --> 00:05:35.920 summarize all the themes, and then tell you where 00:05:35.920 --> 00:05:38.060 the gaps are. It can turn days of reading into 00:05:38.060 --> 00:05:40.339 minutes of reviewing. I want to pause on the 00:05:40.339 --> 00:05:43.519 G, though. Grunt work. And specifically how this 00:05:43.519 --> 00:05:46.040 connects to decision making. Because I see a 00:05:46.040 --> 00:05:49.240 real risk here. If I use AI to do my drafting 00:05:49.240 --> 00:05:52.699 and the research and all the formatting, at what 00:05:52.699 --> 00:05:54.939 point do I stop actually understanding the material 00:05:54.939 --> 00:05:57.660 myself? That is the danger zone. And the guide 00:05:57.660 --> 00:06:00.420 has what it calls a golden rule for drag. You 00:06:00.420 --> 00:06:03.100 only, and I mean only, apply it to curve one. 00:06:03.259 --> 00:06:06.360 You never, ever outsource judgment. Okay, give 00:06:06.360 --> 00:06:08.899 me a concrete example. Let's say hiring someone. 00:06:09.389 --> 00:06:12.850 Perfect example. Hiring has both curves. Scanning 00:06:12.850 --> 00:06:15.329 a thousand resumes to filter for specific technical 00:06:15.329 --> 00:06:18.230 skills. That's curve one. That's pure curve one. 00:06:18.269 --> 00:06:21.009 That is analysis and grunt work. Give it to the 00:06:21.009 --> 00:06:23.970 AI. Let it do the filtering. But the interview, 00:06:24.149 --> 00:06:26.850 looking a candidate in the eye, deciding if they 00:06:26.850 --> 00:06:30.089 have the grit to survive a crunch period. That's 00:06:30.089 --> 00:06:32.709 curve two. That's pure intuition. That's taste. 00:06:33.029 --> 00:06:35.790 If you outsource that, if you ask an AI, should 00:06:35.790 --> 00:06:38.189 I hire this person? You're not being efficient. 00:06:38.329 --> 00:06:40.089 You're abdicating your responsibility. You are 00:06:40.089 --> 00:06:42.410 letting the muscle atrophy. So the human in the 00:06:42.410 --> 00:06:44.910 loop isn't just a safety check. It's the whole 00:06:44.910 --> 00:06:46.910 point. It's the only thing that keeps you relevant. 00:06:47.250 --> 00:06:49.610 Okay, let's shift gears to the interaction itself. 00:06:50.310 --> 00:06:52.550 Because knowing what to delegate is one thing, 00:06:52.550 --> 00:06:54.689 but getting the model to actually do it well 00:06:54.689 --> 00:06:58.649 is another beast entirely. The source talks about 00:06:58.649 --> 00:07:01.329 the intelligent hill of prompting. And it starts 00:07:01.329 --> 00:07:04.009 with this comparison to quantum mechanics versus 00:07:04.009 --> 00:07:08.759 Newtonian physics. Yeah, this is so key. We treat 00:07:08.759 --> 00:07:11.399 AI like it's a Google search. It's Newtonian. 00:07:11.480 --> 00:07:14.220 Input A leads to output B. It's predictable. 00:07:14.420 --> 00:07:18.180 But these large language models... They're probabilistic. 00:07:18.220 --> 00:07:20.959 They're quantum. They're just a cloud of possibilities. 00:07:21.240 --> 00:07:23.720 Meaning you can ask the same exact question twice 00:07:23.720 --> 00:07:25.519 and get two different answers. And we totally 00:07:25.519 --> 00:07:27.300 ignore that. We just type a prompt and expect 00:07:27.300 --> 00:07:29.540 a fact. But really, what we're doing is spinning 00:07:29.540 --> 00:07:32.000 a roulette wheel. So the whole intelligent hill 00:07:32.000 --> 00:07:34.079 is about stacking the odds in your favor. Now, 00:07:34.100 --> 00:07:35.959 the first few camps on this hill are pretty standard. 00:07:36.180 --> 00:07:38.779 One shot means giving one example. Few shot means 00:07:38.779 --> 00:07:40.519 giving a few examples. I think most of our listeners 00:07:40.519 --> 00:07:42.579 get that. But I want to zoom in on the top of 00:07:42.579 --> 00:07:45.959 the hill. Camp four. Agents. This is the frontier. 00:07:46.410 --> 00:07:48.949 The source drops a massive number here. Salesforce 00:07:48.949 --> 00:07:52.670 reported that AI agents drove over $67 billion 00:07:52.670 --> 00:07:56.610 in sales during Cyber Week. Yeah. That is not 00:07:56.610 --> 00:07:59.170 a typo. But I feel like the word agent is getting 00:07:59.170 --> 00:08:01.509 thrown around like a buzzword. What does it actually 00:08:01.509 --> 00:08:03.689 look like for a normal user? Yeah. How is it 00:08:03.689 --> 00:08:05.449 different from a chatbot? It's the difference 00:08:05.449 --> 00:08:08.389 between a chat and a loop. Think of a standard 00:08:08.389 --> 00:08:11.410 prompt. You say, write me a travel itinerary 00:08:11.410 --> 00:08:14.430 for Tokyo. The AI just spits out a block of text. 00:08:14.670 --> 00:08:18.279 Done. It's linear. An agentic workflow is a loop. 00:08:18.379 --> 00:08:20.879 You give it a goal. Plan a trip to Tokyo that 00:08:20.879 --> 00:08:23.319 optimizes for food and minimizes travel time. 00:08:23.639 --> 00:08:26.660 The agent doesn't just write. It pauses. It thinks. 00:08:26.860 --> 00:08:30.519 It breaks the task down. Step one, search flight 00:08:30.519 --> 00:08:33.620 data. Step two, search restaurant reviews. Step 00:08:33.620 --> 00:08:36.669 three, cross -reference locations on a map. So 00:08:36.669 --> 00:08:38.990 it's basically talking to itself. Effectively, 00:08:38.990 --> 00:08:41.450 yes. And here's the magic part. It critiques 00:08:41.450 --> 00:08:43.250 its own work. It might find a great restaurant, 00:08:43.350 --> 00:08:45.009 realize it's closed on Tuesdays, and then it'll 00:08:45.009 --> 00:08:46.690 go back and find another one before it ever even 00:08:46.690 --> 00:08:48.710 shows you the answer. So instead of me doing 00:08:48.710 --> 00:08:50.690 that loop prompting, checking, correcting, prompting 00:08:50.690 --> 00:08:53.370 again, the software does that entire loop for 00:08:53.370 --> 00:08:56.309 me. Exactly. You move from being the writer to 00:08:56.309 --> 00:08:58.769 being the manager. You're managing a little digital 00:08:58.769 --> 00:09:01.629 employee that goes away, does the work, checks 00:09:01.629 --> 00:09:03.669 the work, and comes back with a polished result. 00:09:04.110 --> 00:09:06.809 It sounds great, but it also sounds complicated. 00:09:07.610 --> 00:09:09.950 How does a normal person actually set that up? 00:09:10.549 --> 00:09:13.350 Do I need to be a coder? You know, six months 00:09:13.350 --> 00:09:17.610 ago, yes. Today, no. Most of the major models, 00:09:17.730 --> 00:09:20.269 Claude, ChatGPT with its new reasoning models, 00:09:20.450 --> 00:09:23.289 they're starting to do this natively. You just 00:09:23.289 --> 00:09:25.490 have to prompt for it. You have to say, don't 00:09:25.490 --> 00:09:28.350 just answer. Create a plan. Execute the plan. 00:09:28.570 --> 00:09:30.889 Check your work for errors. Then present the 00:09:30.889 --> 00:09:32.850 final result. You literally have to tell it to 00:09:32.850 --> 00:09:34.710 be an agent. It's interesting. It feels like 00:09:34.710 --> 00:09:37.009 we're moving from search to service. That's a 00:09:37.009 --> 00:09:38.809 great way to put it. I just want to take a moment 00:09:38.809 --> 00:09:42.870 to digest that. Because if we have... These agents 00:09:42.870 --> 00:09:44.929 doing all the heavy lifting. And we have the 00:09:44.929 --> 00:09:48.769 DRAG framework handling the boring stuff. What 00:09:48.769 --> 00:09:50.950 happens to us? Are we just the button pushers? 00:09:51.250 --> 00:09:53.629 The source pivots here to something called the 00:09:53.629 --> 00:09:55.789 intelligent gym. We'll unpack that right after 00:09:55.789 --> 00:09:59.830 this. We are back. We've talked about efficiency. 00:09:59.929 --> 00:10:01.909 We've talked about agents doing all the loops 00:10:01.909 --> 00:10:04.370 for us. But the part of this guide that really, 00:10:04.429 --> 00:10:06.409 really stuck with me, and honestly the part that 00:10:06.409 --> 00:10:09.009 made me feel a bit guilty, was this concept of 00:10:09.009 --> 00:10:11.679 the intelligent gym. Yeah. This is the plot twist 00:10:11.679 --> 00:10:14.179 in the whole thing. Up until now, we've been 00:10:14.179 --> 00:10:16.340 selling you on removing friction. But the guide 00:10:16.340 --> 00:10:18.240 makes this really sharp distinction between a 00:10:18.240 --> 00:10:21.379 wheelchair and a gym. A wheelchair removes friction 00:10:21.379 --> 00:10:24.240 to help you move. A gym adds friction to help 00:10:24.240 --> 00:10:26.500 you grow. And I'll be honest, I think I've been 00:10:26.500 --> 00:10:28.679 using AI as a wheelchair. You know, summarize 00:10:28.679 --> 00:10:31.059 this article, explain this concept. It's just 00:10:31.059 --> 00:10:34.419 so easy. We all do it. It's seductive. But if 00:10:34.419 --> 00:10:36.879 you use AI to bypass the struggle of learning. 00:10:37.419 --> 00:10:40.100 Your brain gets soft. You're outsourcing the 00:10:40.100 --> 00:10:42.659 understanding. The Intelligent Gym is about flipping 00:10:42.659 --> 00:10:45.600 the script. You use AI to intentionally increase 00:10:45.600 --> 00:10:48.080 the difficulty. Progressive overload. Just like 00:10:48.080 --> 00:10:49.960 lifting weights. So instead of saying, summarize 00:10:49.960 --> 00:10:52.600 this book, you read the book. Then you go to 00:10:52.600 --> 00:10:55.220 the AI and you say, okay, I just read this. Here's 00:10:55.220 --> 00:10:58.419 my takeaway. Now quiz me on everything I missed. 00:10:58.820 --> 00:11:01.960 The guide has these levels of difficulty. Level 00:11:01.960 --> 00:11:04.149 one is... Quiz me like a high school student. 00:11:04.309 --> 00:11:06.769 Yeah, that's gentle mode. But level four. Level 00:11:06.769 --> 00:11:10.049 four is challenge me like an irate boss who thinks 00:11:10.049 --> 00:11:12.580 I'm unprepared. Think about the value of that. 00:11:12.980 --> 00:11:15.440 Most of us are terrified of that scenario in 00:11:15.440 --> 00:11:18.220 real life. But with an AI, you can simulate it. 00:11:18.259 --> 00:11:20.659 You can role play a high stakes negotiation or 00:11:20.659 --> 00:11:23.299 crisis meeting all in a safe environment. You're 00:11:23.299 --> 00:11:25.460 forcing your brain to defend its ideas against 00:11:25.460 --> 00:11:27.539 a machine that has basically read the entire 00:11:27.539 --> 00:11:29.879 Internet. So let's try this. If I wanted to do 00:11:29.879 --> 00:11:31.759 this right now, say I was preparing for this 00:11:31.759 --> 00:11:33.940 deep dive, instead of asking AI to write the 00:11:33.940 --> 00:11:36.240 script, what should I have asked it? You should 00:11:36.240 --> 00:11:39.330 have pasted your outline and said. Critique this. 00:11:39.570 --> 00:11:42.610 Tell me where my logic is weak. Tell me which 00:11:42.610 --> 00:11:45.190 arguments a skeptic would completely tear apart. 00:11:45.250 --> 00:11:48.330 Be ruthless. Be ruthless. You have to invite 00:11:48.330 --> 00:11:50.769 the friction. You have to ask for the resistance 00:11:50.769 --> 00:11:53.490 because that is the only way you actually sharpen 00:11:53.490 --> 00:11:55.970 your own thinking. And this connects to the final 00:11:55.970 --> 00:11:57.889 piece of the framework, doesn't it? The intelligent 00:11:57.889 --> 00:12:00.330 fool. This is my favorite part. It's all about 00:12:00.330 --> 00:12:03.710 the beginner's mind. We're all so afraid of looking 00:12:03.710 --> 00:12:06.669 stupid, especially in meetings. Right. You don't 00:12:06.669 --> 00:12:08.169 want to be the one person asking, wait, what 00:12:08.169 --> 00:12:10.570 does that acronym mean? There's a story in there 00:12:10.570 --> 00:12:13.269 about Satya Nadella at Microsoft. Oh, it's a 00:12:13.269 --> 00:12:17.379 legendary story. When he took over in 2014, Microsoft 00:12:17.379 --> 00:12:20.519 was a culture of know -it -alls. It was apparently 00:12:20.519 --> 00:12:22.879 toxic. If you didn't know the answer to something, 00:12:23.019 --> 00:12:26.299 you were dead. Nadella shifted the entire culture 00:12:26.299 --> 00:12:28.559 to be about learn -it -alls. And the market cap 00:12:28.559 --> 00:12:31.039 went from something like $300 billion to over 00:12:31.039 --> 00:12:33.460 $3 trillion. Because they stopped pretending. 00:12:33.919 --> 00:12:37.070 And biologically, this really matters. Neuroscience 00:12:37.070 --> 00:12:40.090 shows us that our brains only rewire. We only 00:12:40.090 --> 00:12:42.990 actually learn when we make errors, when we feel 00:12:42.990 --> 00:12:45.070 that little spike of frustration. So if you aren't 00:12:45.070 --> 00:12:47.830 feeling stupid, you aren't really learning. Precisely. 00:12:47.870 --> 00:12:50.690 And AI is the ultimate tool for being an intelligent 00:12:50.690 --> 00:12:52.590 fool because it doesn't judge you. You can go 00:12:52.590 --> 00:12:55.549 to it at 2 in the morning and say, explain quantum 00:12:55.549 --> 00:12:59.169 computing like I am 5 years old. And then? Explain 00:12:59.169 --> 00:13:01.850 it again, simpler, give me a different analogy. 00:13:02.389 --> 00:13:05.529 You can strip away all the jargon without taking 00:13:05.529 --> 00:13:07.750 that ego hit in front of your colleagues. You 00:13:07.750 --> 00:13:10.009 can practice failing. You can practice failing, 00:13:10.090 --> 00:13:12.649 and that is a superpower. While everyone else 00:13:12.649 --> 00:13:14.509 is nodding along in the meeting pretending they 00:13:14.509 --> 00:13:17.179 understand, you've actually done the rest. You 00:13:17.179 --> 00:13:20.279 know the material deep down. So we have a choice. 00:13:20.620 --> 00:13:22.960 The source describes it as a fork in the road. 00:13:23.100 --> 00:13:25.360 It is. You can take what they call the high floor, 00:13:25.379 --> 00:13:28.600 low ceiling path. That's using AI for automation, 00:13:29.039 --> 00:13:32.159 lazy drafting, zero shot prompting. You'll look 00:13:32.159 --> 00:13:34.539 productive. You'll clear your inbox, but you 00:13:34.539 --> 00:13:37.220 will hit a ceiling and your cognitive muscles 00:13:37.220 --> 00:13:41.039 will atrophy. Or you choose the cognitive gym. 00:13:41.309 --> 00:13:44.490 you use the dreg framework to ruthlessly clear 00:13:44.490 --> 00:13:48.500 out that low value curve one work and why So 00:13:48.500 --> 00:13:50.659 you have time to suffer. So you have time to 00:13:50.659 --> 00:13:53.200 climb the intelligent hill, build these agentic 00:13:53.200 --> 00:13:56.460 workflows, and use AI as a sparring partner to 00:13:56.460 --> 00:13:58.840 challenge your own worldview. One path makes 00:13:58.840 --> 00:14:00.639 you dependent on the tool. And the other makes 00:14:00.639 --> 00:14:03.320 you sharper because of the tool. It's the difference 00:14:03.320 --> 00:14:05.240 between looking smart because you have a smart 00:14:05.240 --> 00:14:07.759 assistant and actually becoming dangerously intelligent. 00:14:08.179 --> 00:14:10.220 The source leaves us with one final challenge. 00:14:10.419 --> 00:14:12.519 It's a really simple one. Open your prompt history. 00:14:13.250 --> 00:14:16.029 Right now, just look at the last five things 00:14:16.029 --> 00:14:18.470 you asked your AI. I am almost afraid to look 00:14:18.470 --> 00:14:21.210 at mine. It's a very sobering audit. Ask yourself, 00:14:21.389 --> 00:14:24.110 did I ask for an answer to save time? Or did 00:14:24.110 --> 00:14:26.409 I ask to be challenged so I could grow? Did I 00:14:26.409 --> 00:14:29.029 use the wheelchair? Or did I use the gym? Exactly. 00:14:29.070 --> 00:14:30.690 I think I have some reps to do this weekend. 00:14:30.850 --> 00:14:31.590 See you in the gym.