Long-term Cost-effectiveness of Interventions for Loss of Electricity/Industry Compared to Artificial General Intelligence Safety
- D. C. Denkenberger, A. Sandberg, R. J. Tieman, J. M. Pearce
Summary
Loss of electricity/industry and artificial general intelligence (AGI) are both considered to be major risks to humanity. Denkenberger et al. (2021) calculated a 50%–88% probability that spending an average of $40 million USD on planning for loss of industry is more cost-effective than AGI research. The cost-effectiveness is even higher for money spent now than for the 40 millionth dollar, highlighting the need for urgent work in this area.
Abstract
Extreme solar storms, high-altitude electromagnetic pulses, and coordinated cyber attacks could disrupt regional/global electricity. Since electricity basically drives industry, industrial civilization could collapse without it. This could cause anthropological civilization (cities) to collapse, from which humanity might not recover, having long-term consequences. Previous work analyzed technical solutions to save nearly everyone despite industrial loss globally, including transition to animals powering farming and transportation. The present work estimates cost-effectiveness for the long-term future with a Monte Carlo (probabilistic) model. Model 1, partly based on a poll of Effective Altruism conference participants, finds a confidence that industrial loss preparation is more cost-effective than artificial general intelligence safety of ~ 88% and ~ 99+% for the 30 millionth dollar spent on industrial loss interventions and the margin now, respectively. Model 2 populated by one of the authors produces ~ 50% and ~ 99% confidence, respectively. These confidences are likely to be reduced by model and theory uncertainty, but the conclusion of industrial loss interventions being more cost-effective was robust to changing the most important 4–7 variables simultaneously to their pessimistic ends. Both cause areas save expected lives cheaply in the present generation and funding to preparation for industrial loss is particularly urgent.