Summary

This post addresses existential risks (XRisks) through the lens of Effective Decision Theory (EDT), which focuses pragmatically on scenarios with realistically meaningful probabilities while disregarding exceedingly rare catastrophic events. Despite tangible XRisks like nuclear war, bioengineered pandemics, climate disruptions, and unaligned AI, humanity’s outlook remains positive when rationally calibrated. Historical evidence shows civilization repeatedly overcomes significant challenges—wars, pandemics, economic depressions, authoritarian regimes—emerging progressively stronger through scientific innovation, technological advances, international cooperation, and effective governance. Axio assigns ~75% credence that global life expectancy (proxy for civilizational health) will increase over the next 25 years. This is balanced assessment, not naive optimism: prudent reasoning supports cautious optimism based on humanity’s proven capacity for adaptation and growth.

Key Concepts

  • Existential risks (XRisks) – Nuclear war, bioengineered pandemics, severe climate disruptions, unaligned AI.
  • Effective Decision Theory (EDT) – Focus on scenarios with realistically meaningful probabilities; disregard exceedingly rare events.
  • Exaggerated fear – Catastrophic consequences provoke disproportionate fear relative to actual probability.
  • Historical resilience – Civilization repeatedly overcomes challenges, emerging stronger through innovation and cooperation.
  • Life expectancy as proxy – Global life expectancy at birth tracks civilizational health comprehensively.
  • Calibrated optimism – 75% credence for life expectancy increase over 25 years; balanced, not naive.

Evolution Notes

  • Connects Effective Decision Theory to civilizational trajectory assessment.
  • Uses life expectancy metric from companion post as concrete, measurable outcome.
  • Balances XRisk discourse (often pessimistic) with historical evidence and rational probability calibration.
  • Demonstrates how Axio applies Bayesian reasoning to macro-scale predictions.
  • Anticipates later work on progress, stagnation, and civilizational phases.

Tags

Cross-References

Open Questions

  • How should we update credence as AI capabilities advance (potential AI XRisk increase)?
  • What about tail risks with genuinely incalculable probabilities (unknown unknowns)?
  • Does focus on “realistic probabilities” systematically underweight high-impact low-probability events?
  • What’s the threshold below which EDT says ignore a risk entirely?