Something New Versus Tried and True: Ensuring 'Innovative' AI is 'Good' AI

摘要

What does it mean for AI to be innovative, and does new always mean better, particularly in terms of the ethical and societal implications of AI? This interview-driven study of 26 stakeholders of AI in the fields of technology research, law, and policy elucidates key tensions in producing innovative AI, as they are understood across sectors. As these stakeholders articulate a discourse on innovation, there is revealed a complex relationship between how innovative AI is conceived, both in terms of what is considered innovation, and how that innovation is seen to be restricted or supported through policy and regulatory action. Ultimately, this discourse operates on similar terms across these stakeholder groups, and presents a knotted, interlinked view of regulatory, design, and policy concerns across the ecology of AI, from its data to its use.