It's interesting they chose those 4 domains, those are notoriously hard problems to begin with. And financial markets in particular self-correct themselves all

It's interesting they chose those 4 domains, those are notoriously hard problems to begin with. And financial markets in particular self-correct themselves all the time. I'm pretty confident that whatever ML models they tried to test on those datasets would degrade over time. But title feels buzzword-ish for sure.