[en] Artificial Intelligence (hereafter, ‘AI’) systems are widely adopted by public administrations. Competition law does not escape the rule. This is unsurprising, given AI systems promise to address well-documented flaws in human decision-making, e.g. arbitrariness or bias. What is more, AI systems carries the potential to strengthen ex officio investigations. Collusive behaviours are, currently, mainly exposed through leniency applications whose number is decreasing. It has been suggested AI-driven cartel screening constitutes a useful complement to leniency programmes. AI-driven cartel screening flags indicators of collusion that then trigger the need for further investigation. This increases the probability of detection that destabilises the collusive equilibrium and incentivises leniency applications. This paper does not discard the benefits of AI-driven cartel screening, but argues it is not a silver bullet against collusive behaviour. This algorithmic solution faces three major challenges. The first is a data challenge. AI-driven cartel screening is a data-dependent solution that is therefore impacted by problems in the availability and quality of the data it relies on. The second is an algorithmic challenge. The opacity of some AI systems infringes the principle of good administration as limited explicability – given unknown parameters weight – prevent public servants to properly state reasons of their decision. The third is a human challenge. Human cognitive biases similarly challenge the duty to state reasons as explaining an administrative decision also means explaining how the public servant weighted that algorithmic recommendation in the decision-making process. None of these drawbacks constitute a dead-end. Although not applicable to EU competition law, the EC’s Proposal for an AI Act proposes embryonic solutions. The data challenge may be technically solved with data quality requirements (art. 10 AI Act); the algorithmic and human challenges with transparency (art. 13 AI Act) and human oversight (art. 14 AI Act). Given the prevalence of human cognitive biases however, technical solution may be insufficient. This paper therefore proposes a ‘four-eyes principle’ that mandates decision adopted by a first officer to be approved by a second officer before entering into force. This solution, based on Belgian National Competition Authority’s bicephalic institutional structure that largely separates investigation and fining, ambitions to provide a reciprocal oversight improving EU competition law in a way that is fundamental rights-proof.
