Daniel Schiff, Justin Biddle, Jason Borenstein, Kelly Laas
Georgia Institute of Technology, School of Public Policy
Illinois Institute of Technology, Center for Study of Ethics in the Professions
Since 2016, more than 80 AI ethics documents – including codes, principles, frameworks, and policy strategies – have been produced by corporations, governments, and NGOs. Our ongoing empirical study addresses the ethics and policy issues in these emerging documents through coding of approximately 25 ethics topics and 15 policy sectors, resulting in an original data set. We observe whether the documents were created in a participatory fashion and how deeply engaged they are with specific laws and regulations. This data set provides the opportunity to understand similarities and differences in ethics and policy concerns, by sector, by region, and over time. Going forward, we intend to engage in further descriptive analysis through techniques like clustering and natural language processing. We have also begun to evaluate issues like: 1) the relative homogeneity of the documents creators; 2) motivations behind creation of documents, such as social responsibility, competitive advantage, strategic planning, starting intervention, signaling social responsibility, and signaling leadership; 3) the varied impacts these documents may have on public discourse, stakeholder groups, regulation and law, corporate operations, and on other ethics documents; and 4) what factors are relevant to assessing whether a given document is likely to be successful in achieving its goals.
