Artificial intelligence (AI) is not yet able to reliably predict earthquakes. This was stated by Andrey Strelnikov, a senior researcher at the Laboratory of Paleoseismology and Paleogeodynamics of the O. Yu. Schmidt Institute of Earth Physics, Candidate of Geological and Mineralogical Sciences.
According to him, to make accurate predictions, it is necessary to have a large amount of data for processing.
To study and understand, and in the future predict earthquakes with greater accuracy, including using machine learning, we need more data and greater temporal coverage.
He explained that instrumental observations of earthquakes in Russia have been conducted for only about a hundred years. This time is not enough to get a true picture of seismic activity.
Strelnikov added that major earthquakes in Russia can occur at intervals of hundreds or even thousands of years. This also makes it difficult to collect statistical data needed to train models.
The most seismically hazardous regions of Russia include areas such as Kamchatka, the Kuril Islands, Sakhalin, as well as the Baikal region, Altai, and the North Caucasus.
