99Science · AI & science

SCIENCE UNDER AUTOMATION

AI in Seismology: capabilities and evidence

Applied to data from 11,241 catalogued earthquakes

What has AI demonstrated?

Applied to data from 11,241 catalogued earthquakes

Test
Seismic arrival picking on distributed acoustic sensing arrays; 11,241 is dataset coverage, not detection accuracy.
Comparison
Catalogued events and seismic phase-picking comparisons; not a whole-field baseline.
System and version
PhaseNet-DAS · 2023
Evidence checked
2026-10-21

Arrival picking uses existing sensors and labelled training data; it does not establish earthquake prediction.

What does the field require?

Build a catalogue from existing sensor records

Infer and test a new fault model

Research tasks

Build a catalogue from existing sensor records

Arrival picking uses existing sensors and labelled training data; it does not establish earthquake prediction.

Infer and test a new fault model

This broader workflow has not been evaluated by the cited demonstration.

Can AI replace the whole workflow?

The results above measure specific tasks. The field assessment also requires completion of the connected workflow, including the requirements listed here. A field-wide replacement percentage has not been established.

Read the assessment method

Sources and related evidence

Compare related fields

Computational genomics · Theoretical physics · Experimental molecular biology