A11 Global Catastrophic Risks (GCR) - X-Risk, Synthetic and Nuclear Pandemic

Title

Global Catastrophic Risks (GCR) are events that could cause permanent and severe damage to humanity's potential on a global scale: they include risks of natural origin (asteroid impact, supervulcanes, pandemics of natural origin), technological risks (synthetic pandemic, nuclear war, severe climate change, risk of non-aligned AI), and systemic risks (collapse of civilization, extreme reduction in population). The field of study of global catastrophic risks developed mainly at the Center for the Study of Existential Risk (CSER) at the University of Cambridge, the Future of Humanity Institute (FHI) at Oxford, and the Open Philanthropy Project in the US seeks to identify, quantify and develop strategies to reduce these risks.

X-risks (existential risks) are the subset of GCRs that could cause the extinction of the human species or the permanent and drastic reduction in its ability to flourish: they include global nuclear war (especially if it produced a nuclear winter that destroyed global agriculture), synthetic pandemic (a pathogen engineered to maximize transmission and lethality that escaped from a biosafety laboratory or was deliberately used as a biological weapon), and artificial general intelligence risk. aligned (an AI system with human-level or higher capabilities that pursues goals incompatible with human well-being). The case of nuclear risk is the most historically documented (there are documented cases of nuclear "near misses" during the Cold War that could have resulted in a nuclear exchange) and the most susceptible to quantitative analysis with historical data.

Quantifying GCRs is methodologically very difficult: most of these risks have very low probabilities (< 1% per century) and potentially catastrophic consequences, making them very difficult to estimate accurately because there is little or no historical data on which to calibrate the models. The main methodological tools include: scenario analysis (construction of plausible scenarios of how the event could occur and estimation of the probability of each scenario), expert elicitation (structured surveys of experts to estimate the probability of the event), and event class reference models (using the historical distribution of similar events to extrapolate to the probabilities of more extreme events). The expected behavior of the GCR framework is that even a very small probability (1% per century) of an event with irreversible consequences for all of humanity justifies investments in risk reduction that would be irrational from the perspective of ordinary risks.

ID:4356