I wish Fable really was only a minor upgrade so that it wouldn't be missed, but this feels like the difference between having a post-doctoral colleague and educating a student that I have to constantly guide and correct. It's so profound for me that some of the reactions in this thread feel like they come from another reality entirely. Or maybe they just got instantly diverted to Opus, who knows.

Living standards should not be worse for younger generations than older generations. But more importantly, younger generations spend money on ridiculous "luxuries" because the situation for things people do need is actually much worse. The luxuries are the last straw, not the first thing to go.

> the elites would need to field their own robot army to challenge governments

Nationally, not through military power, through economic power and law. Since Citizens United, the elite own the government in the US, very clearly. They don't need to challenge anything, they don't need their own armies. They just need to maintain the status quo.

> the power of the sword is the only real power

Doesn't this sound familiar lately? Isn't this the post-Trump era headline?

Yes. Why do you need to ask this? This is the K-shaped economy. Demand is dropping (especially in luxury handbags) as the middle class gets hollowed out. Maybe you're well-off enough that you can still pretend neoclassical economics is still holding up. Must be nice.

> they can't afford to buy a family home in a lot of markets but this has to do with generational wealth and a housing shortage in many parts of the USA.

Yes it has to do with generational wealth, that's my point. A shortage is true in some cases, but not all. That's mostly fueled from massive demand from the wealthy. Buying the family home is the most obvious asset that is becoming out of reach. This past year, many other assets have gone the same way. I think that will continue.

> nobody buys the mountain of goods the robots and AI are producing

I don't get why you're appealing to modern economic theory when the whole point of this scenario is that the standard economic relationship entirely breaks down. You can dismiss it as sci-fi, but people are thinking about this scenario. The wealthy no longer need a mass consumer market in this scenario to stay wealthy. They could simply trade proprietary algorithms, real estate, raw resources, and automated services exclusively among themselves. The global economy shrinks down to a private, self-contained club. Everyone else is locked out of the market. This should sound familiar if you're paying attention to the current markets. There would only need to be a few to maintain the status quo, and they remain inside the market in exchange.

The exit from this bleak future is societal unrest, which needs to occur sooner rather than later in order to succeed. That's the source of instability. Later on, not so much.

> Even during the industrial revolution, jobs moved from the farm to the factories.

Yes, but living standards seriously deteriorated for a long time. That's not too far from the exterminism in the above scenario, just not as dire since factories still need far more human labor to run them. If the labor itself becomes redundant, that's very bad.

I'm not saying this will happen. But dismissing it as sci-fi doesn't seem wise when we're seeing the signs of that future already.

This is a forum filled with experts. Putting marketing aside, in a forum like this, it is most useful to assess models according to the toughest problems in the domain they were specifically refined on. For DeepSeek, that's math. For Claude, that's programming. Gemini and ChatGPT are generalist. Yes, you can use every model for anything you like. But Fable is a bit special, it's very expensive, and very clearly designed for particular types of tasks.

> Fable is just as prone to moronic mistakes as Opus was.

"Just as" is up for debate, but yes, all models are capable of moronic mistakes. That's not helpful information though.

> Codex is still a better model

You're comparing agentic workflows, which relies on a lot more than just the underlying model. It sounds like you're using it like a precision instrument, which is great! It's very different compared to my use cases though, and the ones that Fable seems to excel at. I'm using it for scientific computing, and you really, really want it to one shot a solution. It's either the right algorithm for the task, or the wrong one. So for the hardest problems, it needs to successfully implement a solution in effectively one shot. I use Codex too, but it's often too careless for the delicate tasks. If it gets it wrong, it is really hard to steer it back. You have to start from scratch.

> Bad engineers think Claude is better because it writes more lines of code and is more "proactive".

Think you missed the mark on this one. Not really an engineer, have as much experience as you do in my job. A solution to my problems comprises few lines of code. Fable actually gets it right, first time, every time (so far), but this is with a very long prompt and a bunch of attachments. No other model has done this for me. Not shilling for Anthropic, just impressed. This isn't particularly subjective for me; it is quantitatively measurable.

Don't assume everyone using AI is going to have the same experience you have, or the same types of use cases. And please don't assume that because others have different experiences that it makes them "bad".

Also, Claude has always been mediocre at creative tasks. For your line of work, I would have already recommended Codex hands down.

Fable was designed for _really_ hard software engineering problems. Possibly large, but especially hard. For those tasks, you feel the difference immediately.

I don't see facts in any of these arguments, that's really the point. Doomerism isn't particularly productive, but I'm tired of the complacency and the suggestions that everything will sort itself out. There is a chance it won't.

> we’re talking about state of the art stuff that can reliably be used to do serious work right

The enormous data centers are needed to train new models and deliver to hundreds of thousands of people. For us plebs, yes, the biggest models are out of reach. But a medium-sized business can readily buy the hardware needed to run the state-of-the-art locally if they have the weights. Inference is not so bad. That will be even more true in the future. So it's crazy to me to suggest that AI would just go away everywhere because it is too expensive. The problem is that the current arms race is wasteful and cares not for profitability.

> elite need the working class to enjoy their luxuries

I think this is an extremely critical misconception, and it's sending the world into an increasingly bad place. They really don't, and the assumptions that underpin this statement breaks down when the elite own all the critical assets. If you need proof of this, look at how the working class is being increasingly priced out of almost all luxuries right now. That's the norm. Almost all of human history has been that way. The formula could get a lot worse if there is even the remotest chance that robots or AI can take the place of the workers that might desperately be fighting for the scraps of the wealthy.

> It would be extremely fragile

Actually I think the current state of affairs is fragile. Could you explain this?

> grounded in sci fi instead of reality

More history than sci fi, but a fair criticism. Still, I don't believe there are any "factual" refutations of my concerns, and that should be worrying.

I'm still very skeptical of arguing for KANs as an eventual replacement, like I've seen some papers on the subject argue. The reduced depth may not be an advantage. For example, higher depth for standard neural networks doesn't just add to expressivity, it actually induces spectral sparsity bias. KANs have a bias of their own, but it is different, and is sometimes better, sometimes worse, depending on the task. If increasing depth turns out to be important, KANs might remain less efficient overall.

Precision in the activation function is targetting a part of neural networks that you don't want. There are many other methods that work with high precision. You use neural networks because of their implicit bias toward regular solutions. That means there is a sweet spot at low precision that you're targetting.

Regarding AI companies having capital to expand infrastructure; this is largely irrelevant. The cat is out of the bag, and you can already make serious gains by finetuning to local problems on a desktop machine. There is enough hardware out there to run these things en masse; it's more a question of power. Regardless, this stuff will always keep progressing, regardless of who is doing it.

Regarding the economy, it may be largely irrelevant if we, the people, don't do something very soon. The wheel keeps spinning as long as there are productive workers; it's just that those workers are being replaced by machines. The last year has increasingly demonstrated that you don't need normal people to buy your stuff to remain afloat. You can just keep selling amongst your rich friends while the masses starve, as long as _something_ is still producing what the wealthy want, and enough systems are in place to protect them.

I also remember the "video games are art" debate and the fury from one side of the aisle. I agree that a better understanding of the opposite side should have been part of the debate. But I don't believe that debate was existential. A better comparison to me is the climate change debate. I'm fine with having that debate in an environment where there is little at stake. But it's too late to be doing it with policymakers; we need to be talking about what to do.

cognitive: as in reasonable; of, relating to, or involving conscious mental activities (such as thinking, *understanding*, learning, and remembering)

Look, this isn't necessarily directed at you, but I've been a researcher into the theory of deep learning for many years now. I've seen all the phases, heard all the criticism, had to constantly argue against this. Gary Marcus was one of the loudest voices of this argument, but every would-be philosopher came out of the woodwork to explain why LLMs are no more than stochastic parrots because of their design. Geoffrey Hinton famously had to debunk these arguments multiple times.

And now that LLMs start to clearly exhibit intelligent behavior and can be somewhat reliable, now "nobody ever thought that LLMs could not possibly be good models of some cognitive capacity because of next-token predictions, or linear algebra, etc."? No, that's not okay.

By the way, I know it's a parody of another story that makes this exact refutation. But I think this only serves to highlight the point.