Dear friends,

This month we are announcing a new paper, “A Model for the Probability of Nuclear War”, co-authored by Robert de Neufville, Tony Barrett, and myself. The paper presents the most detailed accounting of the probability of nuclear war yet available.

The core of the paper is a model covering 14 scenarios for how nuclear war could occur. In 6 scenarios, a state intentionally starts nuclear war. In the other 8, a state mistakenly believes it is under nuclear attack by another state and starts nuclear war in what it believes is a retaliation. The model is supplemented with a dataset of 60 historical incidents that may have threatened nuclear war, from World War II (the only actual nuclear war yet to occur) to the Hawaii false alert earlier this year. We believe this to be the largest dataset of nuclear war incidents currently available. Finally, the paper also includes extensive background on probability theory and nuclear war probability analysis, making it a good read for anyone trying to get up to speed on the topic.

You can download the paper here. Stay tuned for next month when we plan to release our paper on the impacts of nuclear war.

Sincerely,
Seth Baum, Executive Director

Artificial Intelligence

GCRI Associate Roman Yampolskiy gave a talk on superintelligence risk titled “What We Need to Know about AI” at The World Government Summit in Dubai in February.

General Risk

GCRI Junior Associate Steven Umbrello reviewed Phil Torres’ book Morality, Foresight, and Human Flourishing: An Introduction to Existential Risks in Futures.