What about immortality? The elite upperclass might feel that it is fair that the underclass live out high quality, but finite lives, while the elites live forever. "After all," say the elites, "people were mortal before we invented superintelligence. What is wrong with keeping them mortal?"
Are ordinary humans supposed to give up on their dreams of a transhuman future simply because another group beat them to superintelligence? And, just so I can make this 100% clear: It is NOT fair to let everyone die, just because they are dying now. Humanity would have solved immortality, with or without the elite's superintelligence, so keeping them mortal is extremely unfair.
Nevertheless, the mentality that "letting people die is ethical because you're not directly killing them" seems shockingly common, and absolutely something a selfish elite upperclass might believe. I don't think you can create a reasonable taxonomy of permanent underclasses without even mentioning this issue once.
Are ordinary humans supposed to give up on their dreams of a transhuman future
I am pretty sure that a survey of "ordinary humans" would suggest that most of them have never considered a transhuman future at all. Eternal youth, hotness, financial comfort, and status, sure, those are all pretty popular, but I wouldn't characterize any of those as dreams of actual transhumanism.
Maybe we don't need to become demigods and worry about the future of the "galactic endowment." Those are pretty esoteric dreams, and pretty sure most people would be OK with ruling those dreams out in exchange for not taking a 20% risk of human extinction within their lifetimes.
Eternal youth, hotness, financial comfort, and status, sure, those are all pretty popular, but I wouldn't characterize any of those as dreams of actual transhumanism.
My comment focused on immortality, which is a superclass of "eternal youth". I think most people would choose to take a 20% risk of human extinction for everyone's eternal youth. For clarity, suppose there was a button which had two possible outcomes:
I think most people would press that button. Eternal youth is, in fact, the promise of many major religions, and a big reason why people are attracted to them. How many emperors died of mercury poisoning trying to find the elixir of life? How many alchemists sought the philosopher's stone? I don't think it's fair to just dismiss the topic as weird and unimportant.
Every person and his/her descendants get eternal youth
This has worse implications than you might expect.
If everyone can have unlimited descendants and reproduce frequently, the population could, for example, double every 20 years. In this scenario the human population would be over 10^150 in 10 thousand years, but the observable universe only has about 10^80 atoms.
Notice the only problem here is unlimited reproduction, but without breaking multiple physics law's (at least constant energy and the speed of light) we are not allowed to have children freely.
With unlimited reproduction, it could get to the point that dying in the first place was the best option. But no, death is worse than living forever usually. Sorry if I was unclear, my comment was just meant to be a reminder that even a post-scarcity world might have some restrictions (not having children could be one of them) and one badly chosen post-scarcity world might end up ugly. I would press the button BTW.
Depending on the size and distribution of these classes, it's not clear that there ARE enough controlling/upper-echelon humans to constitute a class. If they're fewer than a few thousand, and don't really interact with normal people, it may be better to just consider them fully part of the AI complex, and not exactly human.
Epistemic status: My own thoughts. Not predictions, just illustrations for the taxonomy. I think the taxonomy is still not perfect.
AI use: research, discussion of the taxonomy and examples, grammar and phrasing, fact-checking/editorial review, and footnote formatting.
I've read Capital Ownership Will Make You Go Crazy First. [1] At first, I felt like the post didn't fully specify the world in which the permanent underclass exists, but after a deeper look, I was able to see the argument. This reminded me that people use "permanent underclass" to describe lots of different outcomes, which makes discussing the topic very hard. Often, the outcome of such a conversation is realizing that we understand "permanent underclass" differently. This understanding is mostly based on the future we imagine, and the crux of the disagreement is the difference between these imagined futures. Thus, I'm curious to try categorizing these different uses of "permanent underclass" to make up my mind about it and use the term more easily in conversations.
To start, I use "upper class" to describe people who control AI directly or indirectly (including its infrastructure), and the people they care about and bring with them into the upper class. "Underclass" means everyone else: people excluded from wealth and power, with no way to change that.
Dimensions
I see the following relevant dimensions for describing a "permanent underclass" world. There are interconnections between the dimensions, so they are not perfectly orthogonal, which is why the taxonomy isn't perfect:
Living conditions - death → severe deprivation → basic subsistence → comfortable provision → luxurious provision.
Agency, with these subdimensions:
AI control - how many stakeholders control AI? No one? One person? Thousands? Are there competing providers? Does the underclass have access to AI? Is it a weaker version than what the upper class uses?
Concrete examples
Extermination - the upper class no longer needs the underclass, so all outsiders are actively exterminated. Basically an extreme case of the Intelligence Curse. [2] Nowadays, elites depend on ordinary people for labor, so countries are incentivized to keep ordinary people reasonably healthy, educated, and happy. In addition, ordinary people can hypothetically act as one entity and amass power that way, either politically (elections) or through violence (revolution). In this world, AI can do everything, and the elites, for some reason, don't want non-elites around (perhaps they just want Earth to be in a more natural state). I find this somewhat unlikely; abandonment seems more likely.
Abandonment - similar to extermination, but there is no active desire to kill the permanent underclass. Instead, they are just ignored. They don't have any power in the old society, but they have access to land and resources and have to fend for themselves in the wild.
Active oppression - a variation of extermination where the underclass is not killed directly, but is deliberately placed in harsh conditions that kill them slowly.
Modest UBI - everyone gets enough to survive, but remains on the verge of poverty. This happens either through the kindness of elites or through the masses preserving their political power.
Luxurious UBI - everyone gets access to AI abundance and lives a luxurious life. You are still clearly distinct from the elite; for example, their quality of life is better.
Elite UBI - everyone basically lives an elite lifestyle, and elite status is only nominal, based on ownership. It's hard to imagine how this could happen.
Conditional abundance - as long as a person accepts the rules set by the elite, they can live a comfortable life. Materially, they are doing well, but they have no independence.
Democratic abundance - the majority of people lose their jobs, but have enough rights and political power to access AI output and live a comfortable life. The upper class is still much richer and has a higher quality of life, but the distinction is not dramatic (basically comparable to today's society). In this example, it's not even clear that there is an underclass.
Independent enclave - the underclass has the option to leave the old society and join a new, completely independent non-AI society. It requires old-school work and offers a lower quality of life, but basically preserves the status quo from before AI.
Jan Kulveit, Capital Ownership Will Make You Go Crazy First. https://www.lesswrong.com/posts/Stzb33XDsr4uHNerX/capital-ownership-will-make-you-go-crazy-first ↩︎
Luke Drago and Rudolf Laine, The Intelligence Curse. https://intelligence-curse.ai/ ↩︎