EGU Blog by Dr. Denise Degen

 

At the latest after the recent developments around ChatGPT, the discussion about the role, risks, and opportunities machine learning poses is at a new high. The possibilities for machine learning seem to be endless, but the risks and concerns as well. Especially the aspect of how trustworthy these models are is a main discussion point. So, the question arises: is machine learning a curse or a blessing? A question that is not straightforward to solve, since the answers vary drastically for different fields. In this blog entry, we address it from a geodynamical perspective, detailing the requirements for machine learning in this scientific field along with its opportunities as well as open challenges. Read the full blog on EGU website.

 

Physics-Based Machine Learning – Curse or Blessing?

Schematic representation of the various classes of surrogate model construction techniques (Figure from Degen et al., 2023 [3]).