Comment by amadeuspagel

> Although this might seem counterintuitive based on the bottom-line results, we think this experiment suggests that AI middle-managers are plausibly on the horizon.

An AI doing something badly (pricing and restocking inventory) that existing algorithms do perfectly is going to be a middle-manager?