Ethical AI frameworks: the missing governance piece

Abstract

In late 2019, after a period of public consultation, the Australian government released its Ethical Artificial Intelligence (AI) Framework which advanced eight entirely voluntary principles to 'encourage organisations using AI systems to strive for the best outcomes for Australia and Australians'. The summary of consultation inputs records the position that law and regulation have only a 'supplementary' role to play in governing socially responsible AI and that new regulation should only be considered where failure by the 'market' to give effect to the Framework's values showed up 'clear gaps' for intervention. This chapter critiques this regulatory meta-posture, understood as the unjustifiable abdication of a government's fundamental responsibility to attempt more deliberately and through law-based mechanisms to govern the social impact of the development and use by private actors of AI/ML (machine learning). This chapter uses the Australian example to offer a far broader critique of the prevailing trend globally of resorting to ethics-based frameworks as the principal modality for governing the responsible development and use of AI. It is critical of the relative paucity, in the context of this trend, of approaches with respect to governing private and corporate activity that are adequately and appropriately grounded in law and regulation.