My take away of a long post. << It’s a general feature of machine learning—and AI—techniques that they can be very useful if an approximate (“80%”) answer is g

My take away of a long post. << It’s a general feature of machine learning—and AI—techniques that they can be very useful if an approximate (“80%”) answer is good enough. But they tend to fail when one needs something more “precise” and “perfect”.