Tracing the Moving Target: Locating Value Propositions on AI Ethics in the European Union (2018-2025)

Abstract

Anu Bradford (2020) argues that the EU creates a ‘Brussels Effect’ on the digital economy through its data regulations. In this context, understanding the evolution and lifespan of the value propositions of AI Ethics in EU regulations becomes crucial, as it tends to have global implications. The study employs computational tools to systematically analyze the emergence and disappearance of AI-related concepts in the EU legal-policy discourse. Preliminary findings indicate that AI-related policy discourse within the EU is highly dynamic, and many concepts such, as Green AI, Safe AI, Trustworthy AI, etc., are emerging and disappearing from discussions over time. It also indicates that several AI Ethics propositions, such as Fair AI or Explainable AI, have a short-lived presence. At the same time, the presence of Green AI and Sustainable AI has increased in the documents in recent years, indicating a shift in political priorities.The study also provides comparative insights on the weightage of these propositions in legal acts vs preparatory documents within the EU regulatory landscape. Furthermore, through a case study of Green AI the study examines the institutional authorship of these propositions over time to understand how different EU institutions shape regulatory priorities. In this context the paper aims to offer insights on how various concepts on AI Ethics are introduced, discussed, and ultimately fade from the discourse within the EU’s official preparatory documents and regulatory framework.