Compliance is Necessary But Not Sufficient in AI

Many artificial intelligence (AI) developers, managers, and other professionals in positions related to bringing AI applications and autonomous systems to market express the desire to have laws and regulations governing AI established quickly so that they can comply with these and not have to do the labor and emotional work of ethics, shouldering “the burden of AI ethics.” However, the vast majority of firms surveyed do not take a compliance-only approach to ethics. This paper examines some of the reasons why firms have adopted values-based ethics as integral parts of their compliance with laws and regulations rather than having compliance policies alone. It will be argued that these reasons become more, not less, compelling in the context of the swift evolution of the ethical landscape around AI. In particular, it is argued that organizations that provide AI products and services that embrace integrity-based or values-based ethics are better able to navigate the rapidly changing AI landscape, deal with catastrophic failures of autonomous systems, and develop in some of their people the sort of practical, integrated, technical, business, and ethical wisdom needed to bring the benefits of AI to humanity while minimizing its harms. In part, this is due to the mutual reinforcement that values-based ethics has with Values-Based Engineering of AI.