Promises and perils of algorithmic cartel detection

• Main Author: De Cooman, Jérôme
• Format: Lecture
• Language: English
• Published: 2023
• Subjects: artificial intelligence; cartel screening; competition law; dawn raid; human oversight; Law; criminology & political science; European & international law; Droit; criminologie & sciences politiques; Droit européen & international; White whale
• Online Access: https://orbi.uliege.be/handle/2268/304267; https://orbi.uliege.be/bitstream/2268/304267/1/De_Cooman_EUI.pdf

Abstract: Since Ariel Ezrachi and Maurice Stucke's book Virtual Competition (2016), Artificial Intelligence (AI) has become the white whale of competition law. Algorithmic tacit collusion and price discrimination seem to be public ennemy number one. Yet, AI could also be used by (national) competition authorities to dynamically detect anticompetitive behaviours by algorithmically screening available data (i.e., AI-driven cartel screening). Numerous studies have demonstrated this works. It is, however, not a panacea. AI-driven cartel screening has at least three drawbacks: a data-, an algorithmic-, and a human-challenge. This paper investigates all three pitfalls and proposes technical and non-technical solutions based on the bicephalic institutional model of French and Belgian competition law authorities.