You have three minutes to escape the perpetual underclass

> none of them is based on surveillance, ads or social media.

Literally all of them are.

Labour is not a single task, sure, but it's also a fallacy to say "labor is not a single task category that technology can delete".

Before I say why though, there's a lot of bad futures possible even with remote controlled humanoid robots whose only true AI are the basics, to keep themselves standing upright and to walk and run on command without falling over. All those people (even here on HN) who claim (to quote a recent exchange) "Every $1 taken by illegal workers is $1 stolen from Americans", how are they going to react to finding every minimum wage job in the US has been replaced with a Tesla Optimus whose "AI" is "actually Indians" thanks to a low-latency Starlink connection to Jodhpur? How do anti-authority protests (whether anti-ICE or anti-Iranian-government) work with any robots where the "AI" is smoke and mirrors of remote control? This is plenty enough for techno-feudalism, even without AI replacing all jobs.

But back to the original. As a proof of possibility that tech can, in fact, delete labour: Labour is made of atoms. We have always known, for as long as records show, how to make a bunch of atoms that can be forced to do labour for us, even though we have not understood the details of production until very recently. It's called "slavery", and the rules enforcing and maintaining slavery are encoded into the oldest laws we know of, and that text itself references older laws we've not yet been able to identify[0].

That just leaves a question: can technology also arrange atoms to solve arbitrary labour?

Yes, obviously we can physically do it, we're a species that rearranges them at every level, from disassembling mountains for ore, to re-engineering DNA and RNA for crops and medicine and making animations by moving around single atoms.

But that leads to a more precise question: can we learn what pattern we need to do this deliberately, and not with a 20 year delay from simply using our existing biology? There's a quote for that, too:

  If the human brain were so simple that we could understand it, we would be so simple that we couldn’t.

I consider intelligence to be the inverse of how many examples you need to learn something, by which standard all machine learning is very very stupid, and only makes up for it to an extend by being very very stupid very very quickly. Will we solve how to reduce that to only "very stupid", or even just "stupid"? Perhaps, perhaps not. Without improvements, machine learning systems will not take all our jobs, but may still make slaves of us for the reasons previously given, remote controlled robotics. Current ML approaches still have yet to solve for fully-general all-conditions self-driving even in the easy mode that is the USA, they need too many examples to take all our driving jobs despite millions of vehicles on the road gathering training data from the behaviour of all the vehicles they can observe.

When you say:

> This is the lump of labor fallacy- the belief that there’s a fixed amount of economically valuable work that technology and capital can eliminate through automation or capital accumulation instead of transforming it.

some sure fear that way, e.g. it sure was weird to see the exchange with Sen. Richard Blumenthal, and the response from Altman which is pretty close to what you're saying rather than "let me rephrase that, we're all going to die if we get this wrong":

  I think you have said in fact, and I'm gonna quote, development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity. End quote. You may have had in mind the effect on, on jobs, which is really my biggest nightmare in the long term.

- https://www.techpolicy.press/transcript-senate-judiciary-sub...

Not that it takes much intelligence to become an existential threat: ML may only improve at the per-example rate of evolution, but evolution created covid and ebola and the black death and HIV.

But yeah, techno feudalism? All that needs is for Optimus' on-board AI to not fall over when told to walk and run, to keep its balance while holding heavy objects or wielding a nightstick or firing a gun.

[0] https://en.wikipedia.org/wiki/Code_of_Ur-Nammu

> Mass market handicrafts have never been a viable means of subsistence anywhere, ever, if we benchmark against what I think a reasonable lifestyle looks like.

There were plenty of craftsmen who lived lifestyles that were acceptable for the time periods they were living in. From around 1400, to around the early 1900s, European settlements consisting of more than a few hundred people would have had blacksmiths, cobblers, tailors, furniture makers, and various other craftsmen. It wasn’t until the late 19th century that this arrangement started to be replaced by the economies of scale made possible by the factory system.

> A weaver in the 1700s would weep tears of joy at the opportunity to weave things by hand today at market rates.

The idea that a medieval peasant would be envious of our standard of living is doing nothing to further the point you are trying to make.

> By analogy, however, AI is orders of magnitude less power efficient than humans.

It is both more and less power efficient, depending on the task. When a coding task is easy enough that an LLM can actually do it, I've seen e.g. Claude do a week's worth of human labour in a few hours of wall-clock time for what amounted to 0.25 euro of subscription cost. When it can't, it will churn as many tokens as you've got and leave a big ball of mud behind, as seen with the recent attempt at a vibe-coded browser.

When Stable Diffusion is good enough, the energy cost per image output is comparable to the calorific consumption of a human living long enough to type the prompt. When Stable Diffusion isn't good enough, no amount of re-prompting gets you something it can't do.

In most countries living in the woods on your own isn't allowed. You're forced to be connected if you want a job, social life, not be seen as a crazy person. You pretend like we're all living on some island where everything is merely decided by what you as an individual do.

In my home country several old people had to close their shop as they were forced to move to a digital accounting system, they didn't have a choice. My bank only allows me to go to their office without an appointment 1 day a week (maybe not even). My grandpa who doesn't have a phone (he never even got a landline), doesn't have internet and barely even drives, he has to depend on others to call and make appointments. If you want to apply for a job, you need internet connection. Many won't even hire you without owning a car (even if you could perfectly commute with a bicycle or public transport).

If you think we're at the end of this 'evolution', we're just getting started. My grandpa could perfectly do everything on his own until 2010, by 2018 it was getting almost impossible, 2026 he feels like a burden for not being into technology.