0:00 The smartest AI model on Earth just arrived, and it's called Claude Fable-5.0:04 The first piece of advice rocketing around X is brutally simple: do not use it if you don't know how.0:10 It will cost you an arm and a leg.0:12 Yesterday, in episode 105, we talked about the central tension in AI between these incredibly costly models and the search for practical wins.0:21 Today, that tension just became the defining story.0:24 The thread, started by user Ruben Hassid, is a stark warning.0:27 The new Fable-5 model from Anthropic is, by all accounts, a monster of capability.0:32 But that power comes with a price tag that is making even seasoned developers gasp.0:37 The message is clear: Smarter AI equals Expensive AI.0:40 We're not talking about a marginal increase.0:43 We're talking about a step-function change in the cost to get a result.0:47 This isn't just about paying more for a better answer.0:50 It's about the risk of burning through your entire budget on a single, poorly-phrased query.0:56 The era of casual, cheap experimentation with top-tier models might be closing.1:01 Now, while one part of the AI world is bracing for sticker shock, another is looking decades down the road.1:07 Demis Hassabis, the co-founder of Google DeepMind, just did a major interview, and the clips are everywhere.1:13 He's not talking about next quarter's costs.1:16 He's talking about the year 2050.1:18 He's laying out a vision for what he calls AI's "Next Big Breakthrough." This isn't about smarter chatbots.1:24 This is about AI as a fundamental tool for scientific discovery, a partner in solving problems that are currently beyond human comprehension.1:33 Hassabis is playing a completely different game, on a completely different timeline.1:38 For him, the cost of a model like Fable-5 isn't an expense—it's a down payment on the future of intelligence itself.1:45 So you have these two poles.1:46 On one end, the immediate, terrifying cost of state-of-the-art AI.1:50 On the other, the nearly infinite ambition for what it could become.1:55 So where does that leave you, right now?1:57 Here's the thread that connects them.1:59 A post from Ravi Kumar Sahu highlights a tool called Typeless 2.0.2:03 And it’s one of the smartest applications of AI I’ve seen in a while.2:07 Why?2:07 Because it’s not trying to be the biggest brain in the room.2:11 Its goal isn't to speed up your typing.2:13 Its goal is to speed up your THINKING.2:16 It's designed to remove the "invisible mental load"—to help you structure your thoughts, find the right words, and get to your point faster.2:24 It's not about replacing your intelligence; it's about augmenting it in a very specific, very practical way.2:30 This is the opposite of Fable-5.2:32 It's not a massive, expensive engine.2:35 It's a small, sharp tool designed for a specific job.2:38 And this split is showing up everywhere.2:40 We're even seeing a sort of wry, sarcastic take on it.2:43 A thread kicked off by Oliver Traldi and amplified by Enzo Rossi jokes about a "Major AI breakthrough: fully automated luxury continental philosophy." It's a joke, but it's a telling one.2:54 AI just took over a column in Newsweek, and people are both impressed and...2:59 a little weirded out.3:00 It points to this capability explosion where AI can now generate high-level, complex, and sometimes absurdly niche content.3:08 It can write like a philosopher, but does it have anything to say?3:12 Finally, there’s a darker conversation gaining momentum.3:15 A user named "Unknown Traveler" posted a thread that’s getting a lot of traction.3:20 It’s about disinformation.3:21 The argument is that the same technologies making marketing cheaper and easier are also eroding public confidence and muddling our national debate.3:30 The ability to generate plausible, customized content at scale is a dream for a marketer.3:35 It's a nightmare for a functioning democracy.3:38 This isn't a technical problem.3:40 It's a trust problem.3:41 And it's the toxic byproduct of this same AI revolution.3:45 So what does it all add up to?3:46 You have the arrival of a model so smart it's too expensive for most people to use.3:51 You have the visionary predicting a world transformed by 2050.3:55 You have the practical tool that helps you think better today.3:59 And you have the growing shadow of what happens when these powerful tools are used to break things instead of build them.4:06 Let's dive deeper into that first story.4:08 The arrival of Claude Fable-5.4:10 Because this isn't just another model update.4:13 This is a potential market shift.4:15 For the past few years, the dominant narrative has been about the democratization of AI.4:20 More power, to more people, for less money.4:23 Fable-5 seems to be the first major reversal of that trend.4:26 The warning from Ruben Hassid wasn't just "this is expensive." It was "Do not use Fable-5 if you don't know how." That is a CRITICAL distinction.4:35 This implies a new skill barrier.4:37 It suggests that the model is so complex, so nuanced, that simply asking it a question the way you'd ask GPT-4 or a previous Claude model is a recipe for disaster.4:46 You're not just paying for the compute; you're paying for your own ignorance.4:51 Think of it like this.4:52 For years, we've all been given a Toyota Camry.4:55 It's reliable, it's affordable, it gets you from A to B.4:59 Every year, the Camry gets a little better.5:01 Better gas mileage, a nicer stereo.5:03 Fable-5 isn't a better Camry.5:05 It's a Formula 1 car.5:06 And someone just handed you the keys and a credit card that charges you by the RPM.5:11 If you don't know how to warm up the tires, how to manage the fuel mixture, how to hit the apex of a corner—you're not just going to lose the race.5:20 You're going to spin out on the first turn and end up with a million-dollar repair bill.5:25 That's the fear circulating around Fable-5.5:28 It's a tool for professionals, for specialists who can justify the cost because they know how to extract a proportionally massive amount of value.5:36 This might be the beginning of a great divergence in AI models: the "pro-tier" and the "consumer-tier." And the gap between them isn't just performance, it's cost and accessibility.5:47 This changes the calculus for businesses entirely.5:50 Do you invest in training your team to use the Formula 1 car?5:54 Or do you stick with the reliable Camry and accept you'll never be the fastest?5:59 Or...5:59 do you look for another way?6:01 And that's where the other side of the story comes in.6:04 The contrast between Demis Hassabis's grand vision and the practical utility of a tool like Typeless 2.0.6:10 This is the real strategic choice you have to make now.6:13 Hassabis is talking about using AI to cure diseases, discover new materials, maybe even understand the universe.6:20 He's building the starship.6:22 The cost of a single model is irrelevant when the goal is to fundamentally reshape human civilization.6:28 He's operating on a timeline of decades and a budget that looks like a rounding error to Google.6:34 His talk is inspiring.6:35 It's what gets headlines.6:36 It's the "why" behind the whole endeavor.6:39 But then you have your day-to-day reality.6:41 You have a report to write.6:43 You have a presentation to build.6:45 You have a difficult email to compose.6:47 The starship is great, but you need to get to the grocery store.6:51 That's the appeal of Typeless 2.0.6:53 It's not trying to solve interstellar travel.6:56 It's trying to solve the problem of, "Ugh, how do I start this sentence?" It's an AI-powered thought partner.7:02 It helps you untangle the spaghetti in your own head and lay it out in a straight line.7:07 This is a profoundly different philosophy of AI application.7:11 It's not about creating a massive, general intelligence that knows everything.7:16 It's about creating a small, specialized intelligence that helps YOU know your own thoughts better.7:22 It's not a replacement for human intellect; it's a catalyst for it.7:26 And here's the key: its value is immediate and measurable.7:29 You either wrote that email faster, or you didn't.7:32 The report was clearer, or it wasn't.7:34 So here's the thread connecting everything.7:37 The AI world is splitting.7:38 It's no longer one monolithic path forward.7:41 You're seeing a bifurcation.7:43 On one path, you have the Fable-5s and the Hassabis visions—incredibly powerful, astronomically expensive, and aimed at solving the biggest problems imaginable.7:52 This is the High-AI track.7:54 On the other path, you have the Typeless 2.0s.7:56 Clever, efficient, and aimed at solving your problems, right now.8:00 This is the Applied-AI track.8:02 This week's conversations on X show that both tracks are accelerating.8:06 The High-AI track is getting more powerful and more expensive.8:10 The Applied-AI track is getting smarter and more integrated into our actual workflows.8:15 The danger is thinking you have to choose one.8:18 The opportunity is realizing you need a strategy for both.8:21 This week sets up a new kind of decision.8:24 It's not just about what AI can do.8:26 It's about what you can afford—in terms of money, yes, but also in terms of skill and focus.8:31 The debate is no longer about whether AI is a toy.8:34 The debate is about which tools you put in your toolbox.8:38 The starship is coming, but the work of today needs to get done.
