Artificial intelligence and competition law. A computational analysis of the DMA and DSA

This Article investigates whether all stakeholder groups share the same understanding and useof the relevant terms and concepts of the DSA and DMA. Leveraging the power ofcomputational text analysis, we find significant differences in the employment ofterms like “gatekeepers,” “self-preferencing,” “collusion,” and others in the positionpapers of the consultation process that informed the drafting of the two latestCommission proposals. Added to that, sentiment analysis shows that in some casesthese differences also come with dissimilar attitudes. While this may not besurprising for new concepts such as gatekeepers or self-preferencing, the same isnot true for other terms, like “self-regulatory,” which not only is used differently bystakeholders but is also viewed more favorably by medium and big companies andorganizations than by small ones. We conclude by sketching out how differentcomputational text analysis tools, could be combined to provide many helpfulinsights for both rulemakers and legal scholars