The EU Proposal for Regulating AI: Foreseeable Impact on Medical Robotics

摘要

In April 2021, in the wake of a number of preparatory documents, the European Commission published a proposed regulatory framework for Artificial Intelligence (AI). This comprehensive proposal is the first institutional effort that goes beyond first principles to lay down detailed requirements. The impact is expected to be worldwide, with an ensuing potential to cause widespread process changes. This paper examines key aspects of the Proposal and discusses the framework's foreseeable impact, in particular on research in medical robotics, classified among the high-risk applications and thus subject to a heavy regulatory burden. On the technical side, the focus rests in particular on the shift toward system-level requirements and an implied shift toward systems thinking. After discussing some open issues on the nature of decision making in AI-based systems, the paper explores issues connected with developing medical robots with progressively higher levels of autonomy. On the organizational side, attention is on the tension between two conflicting drives: the differentiation of independent regulatory ecosystems and the universal adoption of the most restrictive standard. Finally, regulatory burden differentials are examined in the light of a new division of research labor between academia and commercial spin-offs.