Like our PhD project for example, we're doing machine learning on special forces selection in the Netherlands (for details, see [1]). The aim is basically just to reduce costs for the military and disappointment for the recruits. Furthermore, we hope to learn more about how very capable individuals can be detected early. This is a topic that is useful for many more situations than just the military.
[1]: https://osf.io/preprints/psyarxiv/s6j3r