My Empire of Dirt

A viral blog post arguing that advanced AI will render humans economically irrelevant and trigger great‑power war has reignited debates over both AI “doomerism” and techno‑utopianism. Commenters challenge the author’s geopolitical assumptions and lack of evidence, contrasting speculative AI catastrophe with concrete risks like climate change, while still acknowledging genuine concerns about autonomous weapons, job loss, and concentrated power in AI labs. The thread also probes Effective Altruism’s influence on AI leaders, whether an “economy of agents” without human consumers is plausible, and how much weight to give predictions from high‑profile technologists whose views shift rapidly.

AI, War, and Geopolitics

  • Some readers accept the post’s premise that an AI/electricity/compute race could push the US and China toward conflict, borrowing “RTS game” metaphors and first-strike logic.
  • Others strongly reject this, arguing:
    • War is driven more by ambitions, territory, resources, culture, and personalities than by abstract metrics like energy production or AI progress.
    • AI is an instrument in existing power struggles (e.g., “One China” vs Western imperial interests), not the primary cause.
  • Historical analogy is debated: some invoke the “Thucydides Trap”; others note the 1945–49 US nuclear monopoly did not lead to preventive war.

US–China Capabilities and Blockades

  • One side stresses China’s maritime dependence for oil and gas, suggesting vulnerability in a hot war.
  • Others counter with:
    • China’s expanding navy and shipyards, its ability to contest local seas, and long-term planning with Brazil/Africa as resource suppliers.
    • Skepticism that the US could or would enforce a fully global blockade without massive geopolitical blowback.
  • Nuclear “first strike,” missile defense, and mutual assured destruction are discussed, with broad agreement that full-scale nuclear war would be catastrophic but still heavily deterred.

AI Doom, Responsibility, and Marketing

  • Some see frontier labs as genuinely fearful of near-term “self-improving superintelligence” and racing irresponsibly; an ex-researcher’s resignation letter is cited as evidence.
  • Others consider the post’s argument under-justified (“claims without evidence”) and lump it in with doomer marketing and LessWrong-style cultish thinking.
  • There is pushback that AI doom distracts from real, present crises (climate, inequality, energy poverty).

Economy, Jobs, and “Ascended Economy”

  • One cluster anticipates an “economy without you”:
    • AI as both producer and consumer, agents transacting with agents while most humans are economically irrelevant.
    • Concern that current debates obsess over production (AI replacing work) but ignore “means of consumption” (AI replacing consumers).
  • Others argue:
    • Economies need human consumers; if people lose incomes, the system will shrink or be politically overturned.
    • History suggests large populations will not accept permanent exclusion; unrest, parallel economies, or violent correction are expected.

Effective Altruism and Ideology

  • Links describe an EA-adjacent worldview: build ASI, delay only for alignment, then colonize the cosmos with conscious posthumanity.
  • Critics say:
    • This easily collapses into utilitarian “utility monster” logic where almost anything is justified “for the greater good.”
    • EA can function as a moral cover for wealthy actors doing what they wanted anyway.

Assessment of the Blog and Its Author

  • Several commenters see a pattern of dramatic swings (e.g., recent “AI agents will never program” → now x-risk doomerism) and question epistemic reliability.
  • Others defend rapid updates as reasonable given fast AI progress, but suggest more humility and less confident proclamations.
  • Multiple replies frame the tone as exaggerated, “terminally online,” and in need of more real-world perspective.