Bertrand Rouet-Leduc and his team at Los Alamos National Laboratory in New Mexico have created an A.I.-based algorithm that can predict laboratory earthquakes and could one day be useful in foretelling real temblors. The discovery, being hailed as a “tectonic shift” in the world of geology, comes after years of futile experimentation and study of several phenomena to forecast a quake.
The researchers created an artificial earthquake system and then recorded the acoustic emissions i.e. the tell-tale signs of sounds that usually precede the onset of an imminent quake. This recording was fed into the machine-learning algorithm, which to the surprise of geologists and experts, deciphered a pattern that researchers have missed for years, notes MIT Technology Review.
A similar approach to predict earthquakes was tested in the 1980s but with no results, however, the technological advancement such as improved machine-learning algorithms, supercomputers and the ability to process huge amount of data, have yielded positive outcome this time, writes Scientific American.