No-One Escapes the Permanent Underclass

Original Article ↗ Hacker News Discussion ↗

Reactions to the Thesis

  • Some readers see the piece as insightful and “best of the year”; others dismiss it as over‑generalized dystopian “fanfic.”
  • Critics say it piles “if” on “if”: assumes superintelligence, full automation, stable state control, and smooth extrapolation of today’s trends.
  • Supporters argue the value is not precise prediction but showing how individually rational incentives can create a globally catastrophic, locked‑in structure.

Wealth, Power, and the State

  • Debate over whether money straightforwardly converts to power:
    • One side: wealth is leverage but not sufficient; you still need strategy, institutions, and risk tolerance (examples: mercenary failures, potential state crackdowns on billionaires).
    • Other side: the state exists primarily to protect the capitalist class; conflicts are within that class, not between “the people” and capital as such.
  • Historical examples (samurai, merchants) used to argue that “the wealthy” and “the ruling class” are not always the same, and that class is better defined by control of the means of production.

AI Capabilities and Trajectory

  • Split between:
    • “This is just another Industrial Revolution” (bumpy, but broadly beneficial).
    • “This is fundamentally different,” because cognition scales faster than biology, making human obsolescence structurally likely.
  • Skeptics emphasize current LLM limitations, possible hard caps on progress, and human adaptation (work becoming more creative, cyborg-like human+AI roles).

Economic and Labor Impacts

  • Concern about “elite overproduction”: too many educated people for shrinking white‑collar roles; comparisons to Gulf welfare states vs. Brazilian favelas as end states.
  • Disagreement on which jobs are most vulnerable: some say wire‑mediated work dies first; others expect blue‑collar robotics to be even more disruptive long‑term.
  • One view: firms that fire all white‑collar staff for AI will become irrelevant; customers can go straight to AI.

Money, Markets, and All‑AI Production

  • Some argue that without humans in the loop, monetary “cost” and even markets lose meaning; money is a human coordination fiction.
  • Others reply that even AI‑run corporations still need some resource‑allocation mechanism; markets and “money‑like” signals might persist even if legal humans are mere figureheads.

Agency, Politics, and Human Responses

  • Pessimistic camp: individuals are “ants”; revolutions rarely succeed; people are already using AI to abdicate thinking.
  • Optimistic or practical camp: use AI as a tutor, build skills (especially trades, military/engineering), avoid debt, stay embedded in real communities.
  • Some foresee demographic decline shrinking the underclass over generations.

Alignment, Control, and “Human Zoo” Scenarios

  • One line of argument: even perfectly “aligned” AI can entrench a permanent underclass because human power centers have incentives not to relinquish AI.
  • Others find a tension between “AI obeys owners” and “superintelligence has no reason to obey humans.”
  • “Human zoo” / pet scenarios are debated:
    • Pro: a life of safety, health, and leisure under benevolent AI might beat current global suffering.
    • Con: self‑determination and freedom from opaque control are core values; a comfortable cage is still a cage.
  • Some note that we already live in inescapable “cages” (biology, state, capitalism); AI would just change the jailer, not the fact of constraint.