Algorithmic Collusion or Algorithmic Confusion? Rethinking the Risks of AI Pricing for Companies and Policymakers

Concerns about algorithmic collusion are often driven by the belief that increasingly sophisticated AI systems are more likely to autonomously learn to collude. This article reviews a growing body of academic literature suggesting a more nuanced insight.  Across studies of AI pricing algorithms, supracompetitive prices frequently arise not from advanced strategic intelligence, but from limitations in algorithm design.  More sophisticated algorithms, by contrast, often learn to compete more eff