WEBVTT 00:00:00.000 --> 00:00:02.700 Imagine a world where the internet kind of cleans 00:00:02.700 --> 00:00:05.740 itself up, where scams are just too easy to spot, 00:00:05.919 --> 00:00:09.220 and where everyone's baseline intelligence, well, 00:00:09.400 --> 00:00:12.759 it gets a significant boost. What if AI isn't 00:00:12.759 --> 00:00:16.579 just another tool, but maybe a catalyst for a 00:00:16.579 --> 00:00:19.300 massive upgrade to human intelligence? Welcome 00:00:19.300 --> 00:00:21.550 to the Deep Dive. This is where we unpack the 00:00:21.550 --> 00:00:23.489 most intriguing ideas from the sources you've 00:00:23.489 --> 00:00:25.710 shared with us. Today, we're diving into a really 00:00:25.710 --> 00:00:27.829 fascinating collection of insights about AI's 00:00:27.829 --> 00:00:30.870 surprising impact, both on mass intelligence 00:00:30.870 --> 00:00:33.570 and the evolving landscape of work. Our mission, 00:00:33.590 --> 00:00:35.310 as always, is to give you that shortcut to being 00:00:35.310 --> 00:00:37.909 well -informed on these critical topics. So we'll 00:00:37.909 --> 00:00:40.310 explore this idea of unshittification, look at 00:00:40.310 --> 00:00:42.310 some of the latest AI innovations, and then, 00:00:42.350 --> 00:00:44.329 yeah, confront some pretty stark realities about 00:00:44.329 --> 00:00:46.770 how AI is reshaping the entry -level job market. 00:00:47.130 --> 00:00:49.229 Let's start with a truly surprising idea that 00:00:49.229 --> 00:00:51.530 seems to be gaining traction. AI might actually 00:00:51.530 --> 00:00:53.909 be helping us escape the slop of the digital 00:00:53.909 --> 00:00:56.649 age rather than creating more of it. Ethan Moller 00:00:56.649 --> 00:00:59.049 calls this mass intelligence. Yeah, it's a compelling 00:00:59.049 --> 00:01:00.750 thought, isn't it? Especially when you think 00:01:00.750 --> 00:01:02.369 about the context. I mean, we've all experienced 00:01:02.369 --> 00:01:04.750 what people are calling inshittification. You 00:01:04.750 --> 00:01:06.310 know, think about Google search results just 00:01:06.310 --> 00:01:09.870 drowned in SEO sludge or Airbnb with all those 00:01:09.870 --> 00:01:12.670 surprise cleaning fees and frankly, some shady 00:01:12.670 --> 00:01:15.390 hosts or social media just turning into outrage 00:01:15.390 --> 00:01:17.629 machines. It's that slow creep where platforms 00:01:17.629 --> 00:01:21.000 just turn to garbage. And this counter narrative, 00:01:21.120 --> 00:01:23.879 the un -shitification suggests AI could actually 00:01:23.879 --> 00:01:27.140 reverse that trend. Precisely. Yeah. The core 00:01:27.140 --> 00:01:29.799 idea is basically this. If everyone and we're 00:01:29.799 --> 00:01:31.920 talking over a billion people now using models 00:01:31.920 --> 00:01:35.840 like GPT or Claude has access to a near genius 00:01:35.840 --> 00:01:38.560 level AI agent. Well, the economics just flip. 00:01:38.810 --> 00:01:40.689 It suddenly becomes cheaper, genuinely cheaper 00:01:40.689 --> 00:01:42.930 to create real value than to try and game the 00:01:42.930 --> 00:01:46.090 system. Scams, they become like trivially easy 00:01:46.090 --> 00:01:49.150 for AI to detect. Those old SEO hacks just stop 00:01:49.150 --> 00:01:51.069 working. Algorithmic tricks, they just die off. 00:01:51.189 --> 00:01:53.569 It's about making deception expensive again, 00:01:53.689 --> 00:01:56.189 economically unviable. And this isn't just theory, 00:01:56.290 --> 00:01:59.370 right? We're seeing huge bets being placed. Meta, 00:01:59.370 --> 00:02:01.390 for instance, putting down, what, $72 billion 00:02:01.390 --> 00:02:04.659 on... personal superintelligence. And there are 00:02:04.659 --> 00:02:07.579 these really incredible real world examples like 00:02:07.579 --> 00:02:10.639 Alpha School, AI tutors helping kids learn whole 00:02:10.639 --> 00:02:13.560 grade levels in just 20 or 30 hours. No, absolutely. 00:02:13.840 --> 00:02:15.699 Those examples really make it concrete. Like 00:02:15.699 --> 00:02:17.919 there was one student who jumped 50 SAT points 00:02:17.919 --> 00:02:19.939 because the AI figured out they just needed a 00:02:19.939 --> 00:02:21.919 quick refresher on some third grade math. Can 00:02:21.919 --> 00:02:24.400 you imagine? Or another kid who loved learning 00:02:24.400 --> 00:02:26.439 with their AI tutors so much, they actually canceled 00:02:26.439 --> 00:02:29.039 summer break. What's so fascinating here, I think, 00:02:29.039 --> 00:02:31.580 is how every smart user, you know, boosted by 00:02:31.719 --> 00:02:35.060 AI essentially becomes a node in this huge distributed 00:02:35.060 --> 00:02:37.879 intelligence network. And that just makes every 00:02:37.879 --> 00:02:40.659 decision, every interaction much harder to manipulate. 00:02:40.759 --> 00:02:43.020 The systems that rely on manipulation start to 00:02:43.020 --> 00:02:45.620 lose their grip. That's a powerful vision. a 00:02:45.620 --> 00:02:48.460 world where AI makes deception unprofitable. 00:02:48.520 --> 00:02:50.620 But it sounds like a huge shakeup for businesses 00:02:50.620 --> 00:02:53.539 too. If trust becomes the main thing, what's 00:02:53.539 --> 00:02:55.800 the first kind of visible shift you'd expect 00:02:55.800 --> 00:02:58.400 to see, you know, in how businesses operate ethically 00:02:58.400 --> 00:03:00.620 or how we interact online? Well, I think the 00:03:00.620 --> 00:03:03.900 immediate pressure would be for platforms, for 00:03:03.900 --> 00:03:06.379 businesses to pivot really sharply towards genuine 00:03:06.379 --> 00:03:09.219 value creation. It's almost forced upon them. 00:03:09.400 --> 00:03:11.740 Imagine if every user could instantly verify 00:03:11.740 --> 00:03:15.250 a claim or spot hidden fees or... see -through 00:03:15.250 --> 00:03:17.810 manipulative marketing, trust would just skyrocket 00:03:17.810 --> 00:03:20.409 as the ultimate currency, right? It makes it 00:03:20.409 --> 00:03:22.330 incredibly difficult for bad actors to even get 00:03:22.330 --> 00:03:25.250 a foothold. So it's not just about being good, 00:03:25.289 --> 00:03:28.550 it's sheer economic necessity. It rewards authenticity 00:03:28.550 --> 00:03:31.349 and just makes manipulation way less profitable. 00:03:31.590 --> 00:03:34.629 We could maybe see a return to a more meritocratic 00:03:34.629 --> 00:03:37.539 internet. You know, driven by real utility and 00:03:37.539 --> 00:03:40.300 transparency. OK, that makes sense. So the enforced 00:03:40.300 --> 00:03:42.620 honesty through technology. Let's pivot then 00:03:42.620 --> 00:03:45.120 from the theoretical to what's actually happening. 00:03:46.060 --> 00:03:48.560 Right now in the world, it feels like a constant 00:03:48.560 --> 00:03:51.080 stream of news, doesn't it? New tools, massive 00:03:51.080 --> 00:03:53.900 investments. Oh, it's definitely a flurry. And 00:03:53.900 --> 00:03:57.060 stepping back, you can kind of see two big trends 00:03:57.060 --> 00:04:00.560 emerging almost daily, I think. First, AI is 00:04:00.560 --> 00:04:02.599 rapidly democratizing really advanced creative 00:04:02.599 --> 00:04:05.180 capabilities. We're seeing tools pop up that 00:04:05.180 --> 00:04:07.639 let basically anyone generate hyper realistic 00:04:07.639 --> 00:04:10.900 images from text or create these lifelike digital 00:04:10.900 --> 00:04:14.199 avatars or even translate and publish video content 00:04:14.199 --> 00:04:16.579 in any language. almost instantly. Tools like 00:04:16.579 --> 00:04:19.279 HeyGen lets you create a digital version of yourself 00:04:19.279 --> 00:04:22.060 in minutes. VoiceCheap translates video across 00:04:22.060 --> 00:04:24.300 languages. It's wild. It really points towards 00:04:24.300 --> 00:04:26.019 a future where, you know, sophisticated creative 00:04:26.019 --> 00:04:28.300 work isn't just for specialists anymore. Right. 00:04:28.360 --> 00:04:29.740 It's not just content creation, though. It's 00:04:29.740 --> 00:04:32.029 about empowering people more broadly. Absolutely. 00:04:32.129 --> 00:04:34.910 That's the second trend. The deep integration 00:04:34.910 --> 00:04:39.170 into enterprise applications. Big stuff. Microsoft 00:04:39.170 --> 00:04:41.610 just did that deal with the US GSA offering Microsoft 00:04:41.610 --> 00:04:44.750 365 co -pilot free to millions of users for up 00:04:44.750 --> 00:04:47.490 to a year. They project it'll save over $3 billion 00:04:47.490 --> 00:04:49.959 in the first year alone. And then you've got 00:04:49.959 --> 00:04:52.540 Apple, who quietly shipped models that are, get 00:04:52.540 --> 00:04:56.060 this, 85 times faster and can caption video in 00:04:56.060 --> 00:04:58.439 real time, all running locally on your device, 00:04:58.600 --> 00:05:01.160 which shows powerful AI doesn't always need the 00:05:01.160 --> 00:05:04.639 cloud. Wow. 85 times faster locally. And the 00:05:04.639 --> 00:05:06.139 investment community is definitely noticing. 00:05:06.339 --> 00:05:09.839 Anthropic just secured an incredible $13 billion 00:05:09.839 --> 00:05:12.779 in Series F funding. It tripled their valuation 00:05:12.779 --> 00:05:16.360 to like $183 billion. That's a huge vote of confidence. 00:05:16.560 --> 00:05:19.300 $13 billion. That's staggering. A massive vote 00:05:19.300 --> 00:05:20.980 of confidence, as you say. And it also shows 00:05:20.980 --> 00:05:23.620 up in all the new tools designed to just empower 00:05:23.620 --> 00:05:25.959 us, right? Like Virally, which helps get niche 00:05:25.959 --> 00:05:28.199 -specific posts out on X and LinkedIn automatically. 00:05:28.480 --> 00:05:31.100 Or BAVA for AI -powered diagrams. It's not just 00:05:31.100 --> 00:05:32.959 big tech anymore. These are tools for everyone. 00:05:33.180 --> 00:05:35.319 Oh, and remember that viral video, the black 00:05:35.319 --> 00:05:37.180 plastic bag at the White House people thought 00:05:37.180 --> 00:05:40.000 was AI? Oh, yeah, that caused a stir. Well, experts 00:05:40.000 --> 00:05:43.019 analyzed it. Turns out, nope. No signs of AI. 00:05:43.199 --> 00:05:46.160 Just a plastic bag being tossed. Sometimes things 00:05:46.160 --> 00:05:51.269 are just... Good reminder. So with all this happening, 00:05:51.389 --> 00:05:54.629 AI becoming so accessible, so powerful, what's 00:05:54.629 --> 00:05:57.339 the biggest takeaway for someone? you know, trying 00:05:57.339 --> 00:05:59.660 to keep up without feeling completely overwhelmed 00:05:59.660 --> 00:06:01.660 by it all. Yeah, it is overwhelming. I think 00:06:01.660 --> 00:06:05.300 the key takeaway really is to focus on understanding 00:06:05.300 --> 00:06:08.199 the core capabilities of AI. What does it actually 00:06:08.199 --> 00:06:10.079 enable you to do? Rather than trying to track 00:06:10.079 --> 00:06:11.759 every single new tool that comes out, because 00:06:11.759 --> 00:06:14.220 you just can't, recognize that AI is making complex 00:06:14.220 --> 00:06:16.920 tasks more intuitive, more automated. It shifts 00:06:16.920 --> 00:06:19.500 our focus maybe from pure execution to strategy 00:06:19.500 --> 00:06:22.660 and creativity. So it requires a proactive shift, 00:06:22.779 --> 00:06:24.920 I think, in how we approach both work and learning. 00:06:26.189 --> 00:06:28.250 towards leveraging these tools as like powerful 00:06:28.250 --> 00:06:30.829 co -pilots rather than just fearing they'll replace 00:06:30.829 --> 00:06:33.029 us entirely. OK, leveraging them as co -pilots. 00:06:33.029 --> 00:06:35.850 But now let's pivot to where that impact is becoming 00:06:35.850 --> 00:06:38.930 really clear and maybe a bit sobering for some. 00:06:39.029 --> 00:06:41.389 Yeah. The job market. What are your sources showing 00:06:41.389 --> 00:06:44.149 about how AI is reshaping careers, particularly 00:06:44.149 --> 00:06:47.709 those entry level roles? Yeah, this. This raises 00:06:47.709 --> 00:06:50.189 an important and maybe somewhat challenging question. 00:06:51.050 --> 00:06:53.250 The data suggests that while senior roles are 00:06:53.250 --> 00:06:54.889 still growing, those entry level opportunities 00:06:54.889 --> 00:06:57.750 are, oh, they're drawing up pretty significantly. 00:06:57.889 --> 00:07:00.430 And it doesn't look like just a temporary blip. 00:07:00.430 --> 00:07:02.730 It feels like a profound shift. It's especially 00:07:02.730 --> 00:07:04.850 noticeable in sectors like wholesale, retail 00:07:04.850 --> 00:07:07.670 and software. And it seems to be disproportionately 00:07:07.670 --> 00:07:10.990 affecting graduates from, say, mid -tier schools. 00:07:11.170 --> 00:07:14.089 The numbers. They really do show some significant 00:07:14.089 --> 00:07:16.569 shifts here, don't they? Research indicates firms 00:07:16.569 --> 00:07:19.209 that adopt generative AI, they reduce junior 00:07:19.209 --> 00:07:21.930 hiring by about 22 % in the first six quarters. 00:07:22.069 --> 00:07:24.750 That means junior roles decline, what, 7 % to 00:07:24.750 --> 00:07:27.670 12 % compared to firms not using AI. And what's 00:07:27.670 --> 00:07:29.769 really striking, as you said, it's not mass layoffs. 00:07:29.910 --> 00:07:32.110 Exactly. It's companies just stopping the hiring 00:07:32.110 --> 00:07:35.250 of new junior talent. Precisely. That's the key 00:07:35.250 --> 00:07:37.889 finding. The entry level economy in certain areas 00:07:37.889 --> 00:07:39.889 is kind of vanishing before our eyes. The hardest 00:07:39.889 --> 00:07:43.050 hit. Wholesale and retail. Entry level hiring 00:07:43.050 --> 00:07:45.350 down by around 40 percent per quarter in some 00:07:45.350 --> 00:07:48.589 studies. And AI exposed feels, you know, anything 00:07:48.589 --> 00:07:50.910 with repeatable tasks like coding support, data 00:07:50.910 --> 00:07:53.790 entry. They're also feeling the brunt. Harvard 00:07:53.790 --> 00:07:56.009 did that big study linking millions of LinkedIn 00:07:56.009 --> 00:07:58.750 job records. They tracked the changes right after 00:07:58.750 --> 00:08:01.310 companies started posting jobs like AI integrator 00:08:01.310 --> 00:08:04.379 or similar roles. The correlation. seems pretty 00:08:04.379 --> 00:08:06.680 clear. So it's a silent shift in demand, not 00:08:06.680 --> 00:08:09.240 a mass firing. Yeah. Fewer doors opening. But 00:08:09.240 --> 00:08:11.000 there's an interesting nuance in the data, isn't 00:08:11.000 --> 00:08:13.339 there? Juniors already inside firms often get 00:08:13.339 --> 00:08:16.560 promoted faster. But then why would wages for 00:08:16.560 --> 00:08:18.339 those who do get hired or get promoted internally, 00:08:18.579 --> 00:08:20.639 why would they largely stay flat despite all 00:08:20.639 --> 00:08:22.579 these shifts? That seems counterintuitive. Right. 00:08:22.699 --> 00:08:24.779 That's a really good question. And it's complex. 00:08:25.019 --> 00:08:27.819 The thinking is, well, while existing juniors 00:08:27.819 --> 00:08:30.300 might get faster promotions, their wages often 00:08:30.300 --> 00:08:33.250 stay flat for a couple of reasons. One, Senior 00:08:33.250 --> 00:08:35.429 employees still hold considerable leverage. They 00:08:35.429 --> 00:08:38.370 have experience, domain knowledge, stuff AI can't 00:08:38.370 --> 00:08:41.149 fully replicate, at least not yet. And two, the 00:08:41.149 --> 00:08:43.350 juniors who do get hired in this new environment, 00:08:43.529 --> 00:08:45.590 they're likely already stronger, more adaptable. 00:08:45.690 --> 00:08:47.470 They probably come in with a certain baseline 00:08:47.470 --> 00:08:50.450 of AI fluency. So it maintains a kind of competitive 00:08:50.450 --> 00:08:53.009 equilibrium at that entry wage level. Essentially, 00:08:53.190 --> 00:08:55.990 the bar for what entry level means has just been 00:08:55.990 --> 00:08:57.909 raised significantly. The bar's been raised. 00:08:58.090 --> 00:09:00.809 Okay. So for someone starting out now, or maybe 00:09:00.809 --> 00:09:03.350 even someone looking to research, skill mid -career. 00:09:03.789 --> 00:09:05.690 What's the most critical piece of advice we can 00:09:05.690 --> 00:09:07.809 pull from this data about that vanishing entry 00:09:07.809 --> 00:09:10.009 -level market? Look, I think the critical advice 00:09:10.009 --> 00:09:13.289 has to be focus relentlessly on developing unique 00:09:13.759 --> 00:09:16.600 non -repeatable skills. You know, those human 00:09:16.600 --> 00:09:19.379 centric abilities that AI just can't easily replicate. 00:09:19.600 --> 00:09:21.200 We're talking critical thinking, creativity, 00:09:21.539 --> 00:09:23.740 complex problem solving, emotional intelligence, 00:09:24.080 --> 00:09:26.940 interdisciplinary thinking, that sort of thing. 00:09:27.399 --> 00:09:30.559 Embrace adaptability like never before and actively 00:09:30.559 --> 00:09:33.039 look for roles that leverage AI to boost your 00:09:33.039 --> 00:09:34.860 productivity rather than roles that might just 00:09:34.860 --> 00:09:37.480 be automated by it. And maybe for those already 00:09:37.480 --> 00:09:39.919 employed, really prioritize internal growth. 00:09:40.120 --> 00:09:42.159 Upskilling within your current company seems 00:09:42.159 --> 00:09:45.419 to be a key. many firms are adopting now. Develop 00:09:45.419 --> 00:09:48.120 those uniquely human skills, be adaptable, leverage 00:09:48.120 --> 00:09:50.759 AI, look for internal growth. Okay, we've covered 00:09:50.759 --> 00:09:52.679 a lot of ground in this deep dive from that really 00:09:52.679 --> 00:09:55.679 intriguing idea of AI driving mass intelligence 00:09:55.679 --> 00:09:59.039 and maybe even unshitification online to the 00:09:59.039 --> 00:10:02.159 explosion of new tools and then this very real 00:10:02.159 --> 00:10:04.580 quiet transformation happening in the entry -level 00:10:04.580 --> 00:10:06.879 job market. Yeah. And if we connect all this 00:10:06.879 --> 00:10:09.559 to the bigger picture, it's pretty clear AI isn't 00:10:09.559 --> 00:10:11.559 just changing how we work. It's potentially changing 00:10:11.559 --> 00:10:14.320 how we think, how we learn, maybe even how we 00:10:14.320 --> 00:10:16.500 define intelligence itself. The implications 00:10:16.500 --> 00:10:19.539 are just profound, touching everything from education 00:10:19.539 --> 00:10:22.259 with those AI tutors to how we interact with 00:10:22.259 --> 00:10:24.200 information, especially with the resurgence of 00:10:24.200 --> 00:10:27.039 concepts like world models for AI, basically, 00:10:27.039 --> 00:10:29.639 how AI builds its own internal understanding, 00:10:29.899 --> 00:10:33.240 its simulation of reality, its mind bending stuff. 00:10:33.539 --> 00:10:36.269 So here's a final thought then. If AI helps us 00:10:36.269 --> 00:10:38.929 detect deception, makes the world less susceptible 00:10:38.929 --> 00:10:41.809 to manipulation, while at the same time it's 00:10:41.809 --> 00:10:44.429 reshaping that entry -level job market so dramatically, 00:10:44.669 --> 00:10:47.470 are we heading towards a future where human ingenuity 00:10:47.470 --> 00:10:50.429 is both amplified by AI but also forced to specialize 00:10:50.429 --> 00:10:53.009 much, much earlier in our careers? What will 00:10:53.009 --> 00:10:55.049 entry -level even mean when the actual entry 00:10:55.049 --> 00:10:57.230 point itself is shifting so drastically? Something 00:10:57.230 --> 00:10:59.019 to think about. We invite you to consider your 00:10:59.019 --> 00:11:01.080 own skills, your learning paths, and how you 00:11:01.080 --> 00:11:03.700 engage with information in this rapidly changing 00:11:03.700 --> 00:11:05.840 landscape. What really stands out to you from 00:11:05.840 --> 00:11:06.440 this deep dive?