Austrian economists have always discounted the use of math in economics because it is too complex and self referential, ie the things you are modelling can read your theories and adjust their behavior. Sounds like machine learning may finally bridge that gap, and take us out of the era of Voodoo Economics and into one where human action in aggregate is accurately modeled beyond the most basic "this policy will increase/decrease capital investment" used by the Austrians. Ever since I learned about AI and the potential upcoming technological Singularity, I have suspected that this will be the most important application of an ASI, because once you can accurately model human behavior on a massive scale, you can devise non-intrusive interventions a la the Butterfly Effect to maximize freedom and prosperity for everyone. Sort of like when something bad happens to you, like your family dies in a fiery car crash, or when a baby gets cancer, and some idiot tells you that "God has a plan". Well, now there is actually a real plan, and one that's actually GOOD for everyone involved.
Even if it doesn't run that far, hopefully this will finally bury the idiot Keynesian school. A school, I might add, that couldn't even COMPREHEND THE EXISTENCE of stagflation, nevermind model/predict it and hence should have been thrown into the dustbin of history in the 70's when it first appeared. Too bad it is so useful for justifying the government's eternal urge to spend more of the people's money buying votes.