Article Preview
TopAim And Scope Of This Paper
In the past decade, many policy documents have discussed ethical issues and potential future directions related to new emerging technologies such as artificial intelligence (AI) or autonomous systems. This paper presents the work of the AI4People-Automotive Committee1 established to advise more concretely on specific ethical issues that arise from autonomous vehicles (AVs). The committee consisted of industry experts and researchers from the fields of ethics, law, philosophy, engineering, technology and policy. The aim of this paper is to provide the automotive sector, including both companies and public entities such as regulators, with concrete and practical guidelines to comply with ethical principles within the AI systems of AVs. Therefore, this paper could serve as a checklist for policymakers and companies as well as a basis for developing a certification of ethics, an ‘ecosystem of trust’ (European Commission, 2020b) and ultimately a ‘Good AI Society’ (Floridi et al., 2018) in the automotive sector. These guidelines are intended to provide a clearer vision and moral compass on how to proceed and what to consider when developing AVs, rather than additional barriers to innovation. The automotive sector is defined here in the broadest terms possible to encompass a wide range of companies involved in the development of vehicles, including private cars, trucks, busses and shuttles. Sea, air and military-type applications have been excluded due to their functional and ethical specificity. This paper will focus on the ethics of the AI-based tools that are used in automotive technology, rather than on the ethics of vehicles in general.
This paper distinguishes between high-level guidelines for policymakers (‘policy recommendations’) and concrete actionable recommendations for companies (‘industry recommendations’). However, the line between the two cannot always be drawn clearly which also highlights the importance of co-regulation (i.e. the interaction of legal regulation and self-regulation by companies) (Pagallo et al., 2019). The policy recommendations are designed to focus attention on pressing policy issues and assist in setting acceptable standards. Thus, the policy recommendations ultimately influence the industry recommendations. Responsible targets for the execution of the policy recommendations are: policymakers, legislators, ethics standards boards and commissions such as the United Nations Economic Commission for Europe (UNECE). The industry recommendations formulate guidelines for companies across their entire value chain (especially during research & development, production & operations and service). Therefore, original equipment manufacturer (OEM) and car manufacturers are the primary responsible targets for those recommendations.
Before turning to the principles and guidelines, we note three key points of consensus among the authors: (1) a responsible balancing of risks or estimated harm should be permitted at any time for AVs; (2) a large-scale introduction of full-mode AVs (level 4 and higher) onto streets is unlikely in the short run, so we must consider a more incremental, step-by-step approach; and (3) policymakers face significant challenges now, and so there are significant pressures to quickly develop a clear regulatory framework.