Evaluating Multi-level Risk in AI Implementation

AI affects commerce and security for governments, organizations and individuals. The impact of this transformation, however, lies not merely in the technology itself, but in who — individuals, organizations or governments — is primarily driving it. At its worst, AI transformation led by government may lead to scenarios such as social credit schemes and the elimination of privacy for individuals and organizations alike. Transformation left to organizations likewise carries risks of leveraging network effects to gain disproportionate advantage. AI implementation driven primarily by businesses or other organizations poses a threat both to individuals and to the ability of governments to effectively regulate power built upon far stronger network effects than the world has encountered. This paper develops a risk model for evaluating multiple forms of AI implementation and their implications for individuals and society.