AI Ethics as Applied Ethics

Abstract

The need to design and develop artificial intelligence (AI) in a sustainable manner has motivated researchers, institutions, and organizations to formulate suggestions for AI ethics. Although these suggestions cover various topics and address diverse audiences, they share the presupposition that AI ethics provides a generalizable basis for designers that is applicable to their work. We propose that one of the reasons the influence of current ethical codes has remained modest, may be the conception of the applied ethics that they represent. We discuss bioethics as a point of reference for weighing the metaethical and methodological approaches adopted in AI ethics, and propose that AI ethics could be made more methodologically solid and substantively more influential if the resources were enriched by adopting tools from fields of study created to improve the quality of human action and safeguard its desired outcomes. The approaches we consider to be useful for this purpose are the systems theory, safety research, impact assessment approach, and theory of change.