Sharing Accountability of Versatile AI Systems: The Role of Developers and Practitioners

Abstract

AI systems pose both opportunities and threats in various industries. To harness these opportunities and mitigate risks, accountability is crucial. Traditionally, developers bear the responsibility for auditing and modifying algorithms. How-ever, in the evolving landscape of versatile AI, developers may lack contextual understanding across diverse fields. This paper proposes a theoretical framework that distributes accountability to developers and practitioners according to their capabilities. This framework enhances systemic com-prehension of shared roles, empowering both groups to col-laboratively avert potential adverse impacts.