Ten regulatory principles to scaffold the design, manufacture, and use of trustworthy autonomous systems, illustrated in a maritime context

Ten regulatory principles to scaffold the design, manufacture, and use of trustworthy autonomous systems, illustrated in a maritime context

연구 분야: Artificial Intelligence

학회: TAS '23: Proceedings of the First International Symposium on Trustworthy Autonomous Systems

Autonomous systems are increasingly prevalent around the world, with the benefits related to safety, efficiency, and sustainability attractive in addition to the opportunity to establish entirely new capabilities. In order to operationalise autonomous systems technology it is critical to have in place the legal, regulatory and ethical infrastructure necessary to enable safe and trusted operation. While there is a growing body of literature regarding ethical Artificial Intelligence (AI), there is a need for more academic exploration of legal and regulatory best practice for autonomous systems used in commercial and defence contexts. This paper addresses that literature gap by considering the role of regulation and its relationship with trust in a multi-disciplinary context, before proposing 10 principles to base regulatory development and implementation on. These principles, Trust-centred; Collaborative; Risk-based; Evidence-led; Facilitate experimentation; Systems-focussed; Usable; Consistent; Adaptable and Reviewable, collectively provide a domain and technology agnostic basis for a regulatory framework development and implementation approach that supports the design, manufacture and operation of safe and trusted autonomous systems. The paper concludes by recommending next steps towards the regulation of safe and trusted autonomous systems, including a focus on collaboration and experimentation.