Sure, in brief. Emilio Calvano, Giacomo Calzolari, Vincenzo Denicolò and Sergio Pastorello cowl this subject within the newest problem of the American Financial Overview:
More and more, algorithms are supplanting human decision-makers in pricing items and companies. To research the potential penalties, we research experimentally the conduct of algorithms powered by Synthetic Intelligence (Q-learning) in a workhorse oligopoly mannequin of repeated value competitors. We discover that the algorithms persistently study to cost supracompetitive costs, with out speaking with each other. The excessive costs are sustained by collusive methods with a finite part of punishment adopted by a gradual return to cooperation. This discovering is strong to asymmetries in price or demand, adjustments within the variety of gamers, and numerous types of uncertainty.
Right here is the paper.