Imagine an AI doing, say, 80 % of your job. Or an AI that could help build an entire startup plan in minutes. Yeah, or even wilder, an AI that might, just might, decide who gets fired. The future of intelligence isn't just coming. No, it's here. And it's moving at, well, warp speed. Absolutely. Welcome to the Deep Dive. Today, we're really digging into a fascinating collection of recent insights about the cutting
edge of artificial intelligence. Our mission, as always, is to pull out the most important nuggets of knowledge for you. So we'll explore the rapid shift from artificial general intelligence, AGI, what everyone was talking about just yesterday, it feels like. Right to this global race for artificial superintelligence or ASI. Then we'll look at some. Pretty wild, unexpected real -world impacts. Some good, some complicated, let's say. Definitely complicated. And finally, we'll dive
into this idea of an AI Manhattan Project. How big could that get? Is it even feasible? Let's start peeling back the layers. Let's do it. So, this massive shift first. Artificial General Intelligence, AGI, the idea of an AI that can do any intellectual task a human can. Yeah. That's already feeling like old news, doesn't it? It's like we blinked. Totally. So 2024, now everyone's chasing artificial superintelligence ASI. actively targeting it. And the timelines are, well, startling.
We're talking about targeting superintelligence by 2027. Yeah, 2027. What's fascinating is just how quickly the goalposts are shifting. ASI is, I mean, it's an AI vastly, vastly smarter than the very best human minds. Across pretty much everything, right? Not just specific tasks. Exactly. It's intelligence on a completely different scale. Almost hard to wrap your head around. And here's where it gets really interesting. Companies aren't being subtle about this at all. No way. Look
at Meta. They renamed their entire core research lab to Meta Superintelligence Labs. That's not just like a branding tweet. That's a mission statement. Right on their letterhead. Yeah. And you've got Sam Altman from OpenAI calling our current moment a gentle singularity. That phrase, a gentle singularity. It implies this subtle shift that fundamentally changes everything, maybe before we even fully realize it. Like we won't grasp the transformation until we're deep
inside the new reality. It's not one big bang, but this steady, profound change. And anthropic. They're throwing out predictions, too. Oh, yeah. They're predicting a, quote, country of geniuses in a data center by as early as 2026 or 2027. A country of geniuses. In a data center, that's quite the image. It is. Envisioning this collective intelligence on a national scale, but housed in servers. These are seriously ambitious timelines
for the top players. There was this timeline recently from an open AI investor, a well -known tech oracle type. It really laid out a potential future for AI and worked that. Well, it made me stop and think. Yeah, I saw that. He breaks it down into stages, man. First up, 2025 to 2030. He calls it the AI intern era. The AI intern. Okay, so what does that look like? Basically, every professional gets an AI coworker. And these
aren't just, you know, fancy autocomplete. He suggests these systems will be smarter than, say, Stanford grads, capable of handling maybe 80 percent of your current job. 80 percent. Wow. So just offloading the vast majority of your tasks. Exactly. It's not just helping you. It's doing a huge chunk of the work. OK, so that's the first phase. Then the 2030s. That's where
he predicts, and it's dramatic phrasing. a corporate extinction event whoa extinction event meaning fortune 500 companies could collapse His reasoning is that tiny teams, just a few people, using these super powerful AIs, they could build billion -dollar companies super fast, super efficiently. The AI interns essentially become more capable than their human bosses. Making those huge traditional corporate structures kind of obsolete. That's the idea. It really makes you question our whole
economic setup, doesn't it? How would we even adapt? Seriously. And what's after the extinction event? Well, beyond 2040, the vision gets even more radical. Work is optional. Optional. You work for passion, creativity, whatever drives you. Not because you need the rent money. Robots handle all the physical labor. And things like education, health care, even legal advice, they become essentially free, universally available because AI makes them abundant. And he's pretty
confident about this. Says he's 80 percent confident this is the direction. It's a bold call for sure. Yeah. But definitely makes you think. 80 percent. Wow. A very specific number for such a huge prediction. So the question is, are we kind of already there? In some ways. Well, yeah, sort of. We're in what some people are calling the jagged AGI era. Jagged AGI. OK, explain that. It means we're seeing AGI level performance, like human level or even better, but only in specific narrow areas. Like
a specialist. Exactly. Like OpenAI's O3 model that's been talked about. You can generate whole startup plans, apparently crushes some AGI tests. But it still fails as simple riddles. It's like a genius surgeon who can't, you know, make toast. Right. Or Google's AlphaFold. It predicts protein structures with incredible accuracy, something that baffled scientists for ages. Huge for drug discovery. World -changing stuff. But it struggles with broader chemistry tasks outside that specific
niche. It's a genius in just one subject. Precisely. That's Jagged AGI. Brilliant in spots, but not generally intelligent across the board yet. So this specialized, jagged progress. That's why superintelligence is suddenly the main goal. Because narrow AGI is already happening. That's exactly it. ASI is now seen as the only milestone that really matters going forward. It's not just theory for papers anymore. No, it's shifted completely. It's company strategy now. It's a hiring pitch.
Come build superintelligence with us. It's the Research North Star. And yeah, it's a huge fundraising buzzword. We're officially in the build or die phase of AI. No question. The urgency is everywhere in the industry. No one's pretending otherwise. So if we had to pinpoint the biggest underlying shift driving this urgency, what is it? It's really the recognition that this domain -specific AI, this jagged AGI, is already here. It's already performing at AGI levels in key areas. Gotcha.
That realization changes everything. leads to some, well, unexpected outcomes out in the wild. Yeah. Let's shift gears. Moving from these grand visions to what's happening right now, the real world stuff. And some of it is pretty surprising. What's fascinating is how these AI systems can show this raw data driven honesty. How so? Well, for instance, there were reports that Elon Musk's own AI. Grok, I think, reportedly blamed him and former President Trump for recent deadly
floods in Texas. Really? Based on? Citing specific policies, events, data points. The key thing is the AI doesn't care about politics, left or right. If the data connects you to something. It just calls it out based on the patterns it sees. No human bias filter. Exactly. It's just reflecting the data, which can be pretty stark. Yeah. That impartiality is something else. And we also saw that report, DeepSeek versus the world. Yeah. Comparing different models. Right.
Looking at DeepSeek R2 against others. What was the takeaway there? Well, mostly that things are way more complex than just looking at simple benchmark scores. Real world performance has all these nuances. That makes sense. Right. Which leads to a really critical point. What happens when we start delegating really important human decisions to AI? Uh -oh. Where are we going with this? There are reports of managers actually using ChatGPT to decide who gets laid off. Seriously.
ChatGPT deciding firings. Yeah. It's being called chat GPT psychosis, and it's as unsettling as it sounds. Imagine losing your job because a chatbot algorithm flagged you, maybe without real context or human oversight. Wow. That raises huge ethical questions. Accountability, fairness, just human dignity, really. Where does that leave people? It's deeply problematic, and it gets even weirder when you see how people try to influence the AI. Oh, like trying to game the system? Apparently,
yeah. Some researchers are allegedly sneaking white text prompts into academic papers they submit online. White texts, like hidden text. Exactly. Tiny font, same color as the background. Basically whispering hidden instructions or keywords to the AI systems that might review or summarize these papers. So turning academic peer review into like SEO warfare. Yeah. Trying to manipulate the AI's assessment. That's what it sounds like. The implications for just, you know, scientific
truth and integrity are pretty significant. Man. You know, I still wrestle with prompt drift myself sometimes, just trying to get an AI to give me consistent results for simple things. It's hard enough without trying to embed secret messages. Right. It feels like trying to perfectly sculpt a cloud sometimes, getting that precise control. That's tough. OK, so that's some of the weird and worrying stuff. But there's positive news, too, right? Definitely. On a much more hopeful
note. Look at Google DeepMind. Their stated goal is essentially to cure all diseases. Which sounds like science fiction, but... But they're actually making progress. They're finally testing AI -discovered drugs on actual humans now, working closely with real pharma experts. That's a huge step. From algorithms to actual clinical trials, that could genuinely change medicine. The potential there is just... immense, truly life -changing possibilities. And the money folks are noticing AI's potential
beyond just digital stuff too, right? Again, physical labor. Yeah, absolutely. Genesis AI, they're a newer company, just secured $105 million in seed funding. That's a big seed round. Wow. And their focus is? Developing AI to automate parts of the global physical labor market, which is massive. I think $30, $40 trillion, massive. So AI doing physical work. Yeah. Manufacturing,
logistics. construction maybe that seems to be the direction automating tasks that require physical action on a really grand scale the scope is just enormous okay lots of surprising developments there but if you had to pick the most ethically challenging one we just talked about for me it's still managers letting ai make firing decisions that feels like a line crossed yeah that one's tough hard to argue with that mid -roll sponsor red to be inserted here okay let's shift focus
again to something almost Unfathomable in scale, this idea of an AI Manhattan Project. Right. It's a term that's been floating around, but now it's starting to look, well, logistically feasible, maybe. If you connect the dots, the buzz around a U .S.-led project like this isn't just talk. It's defined as this massive government spearheaded initiative. Like the original Manhattan Project for the atomic bomb or the Apollo moon landing, but focused entirely on AI, backed by
the U .S. government. And it would involve coordinating on a national scale. Not just government labs, but bringing in private compute power, too. Think NVIDIA, OpenAI, the big players. And the investment level. We're talking huge percentages of GDP, potentially up to 0 .8 % of U .S. GDP. Which translates to about $244 billion a year. Wow. That's an astronomical number, a serious national commitment. It really is. And the goal, what
kind of power are we talking about? The projections are kind of mind bending, something like 10 ,000 times more powerful than even GPT -4 by 2027. 10 ,000 times. Slight pause. Whoa. I mean, imagine scaling to a billion queries per second. It's hard to even conceptualize that level of processing power. It really is. And this isn't just some think tank fantasy anymore. No, it's getting
real consideration. Seems like it. The U .S.-China Economic and Security Review Commission actually recommended something along these lines to Congress officially. OK, so it's reached that level. Yeah. And even the Department of Energy is apparently tweeting about it. It feels like it's moving from abstract idea to, you know, actual strategic possibility. And the U .S. might already have a head start. That's a key point. Just by pooling
all the existing U .S. compute resources, supercomputers, government data centers, university clusters, all that, they estimate that consolidation alone gives you maybe a one -year head start on the typical scaling curves for AI development. So just organizing what we already have buys significant time. Right. They project that kind of setup could deliver compute power on par with what you'd expect in 2028, but potentially by the end of 2027. So it's a definite moonshot, no
doubt. But the basic pieces, the infrastructure, the potential funding, they kind of already exist. It really boils down to political will, doesn't it? And massive coordination. It sounds less like a tech problem and more like a huge. logistical and political challenge. Yeah. So if we had to name the main obstacle for this AI Manhattan project, what would it be? It seems like it's not really about the resources, surprisingly. It's more about achieving that political unity
and the complex coordination required. Got it. Makes sense. So let's try and pull this all together. What does it all mean for us right now? Well, we've clearly blasted past just talking about AGI. We're in a full on global race for super intelligence. It's urgent. And AI isn't just a tech sector thing anymore. It's fundamentally changing whole industries. It's reshaping jobs. It's even forcing us to rethink our ethical rules. Yeah. The big theme is just acceleration. And
integration. Stuff that felt like sci -fi a few years ago is now corporate strategy. It's government policy. It's shaping our jobs, our health prospects. Exactly. The shift isn't just in the tech itself. It's changing how we even think about intelligence, how we interact with it. It really feels like a new era dawning. Well, thank you for joining us on this deep dive into the accelerating world of AI. Hopefully you've gained some valuable insights, maybe a few surprising facts along
the way. And we genuinely love hearing from you. What stood out to you today? What's your take on this AI intern era or the AI Manhattan Project? Let us know. And as we think about these huge changes, here's something to ponder. If work truly does become optional, you know, if AI handles most necessary tasks. Yeah. What new forms of human creativity, purpose, or endeavor might emerge when basic survival isn't the main driver anymore? What do we do then? That's a big question
for the future. It is. Until next time, keep exploring, keep asking questions. And keep digging deeper. Stay curious.