Resolution criteria

This market resolves to "Yes" if an artificial intelligence system publicly and accurately predicts a magnitude 7.0 or greater earthquake in the San Francisco Bay Area, with the prediction made at least 24 hours prior to the event. The prediction must specify the earthquake's magnitude (7.0 or higher, does not need to be exact), location (within the San Francisco Bay Area, does not need to be exact), and timing (within a 24-hour window). "There will be a 7+ magnitude earthquake in the SF Bay Area at 2pm June 7 - 2pm June 8, 2031" would be sufficiently specific if it happened then or 24 hours before or after. Verification will rely on reputable sources such as the United States Geological Survey (USGS) and peer-reviewed publications.

Background

Accurate short-term earthquake prediction remains a significant challenge in seismology. While AI and machine learning have been applied to earthquake forecasting, these efforts have primarily focused on probabilistic hazard assessments and early warning systems that provide seconds to minutes of notice. For instance, the ElarmS system successfully provided a brief warning before the 2007 Alum Rock earthquake in California. (en.wikipedia.org) However, no AI system has yet demonstrated the capability to predict large earthquakes with precise timing and location days in advance.

Considerations

The San Francisco Bay Area is seismically active, with a 72% probability of experiencing a magnitude 6.7 or greater earthquake by 2045. (usgs.gov) Despite this, the ability to predict the exact timing and location of such events remains elusive. AI systems have shown promise in analyzing seismic data and identifying patterns, but translating these capabilities into reliable, specific predictions for large earthquakes is an ongoing area of research.