Why Data Protection and Transparency Are Not Enough When Facing Social Problems of Machine Learning in a Big Data Context

Abstract

Neither data protection nor transparency are effective answers to large part of the social challenges coming along with Machine Learning in a Big Data context (MLBD). Data protection is not enough, simply because input and output data of MLBD need not qualify as personal data according to the definition stipulated in relevant legislation such as the General Data Protection Regulation (GDPR), nor do they have to be about human beings at all, in order to affect humans in questionable ways. Transparency falls short for another reason. Although the opacity due to technical and contextual dimensions are basic problems in attempts to cope with ethical and legal problems concerning MLBD (Vedder, Naudts, 2017; Burrell, 2016; Kroll et al., 2017), transparency can only play a role at the very first start of the deliberations. For the actual observation, articulation and solution of possible problems, a broader normative framework (ethical or legal) is needed.

Keywords: machine learning, artificial intelligence, big data, profiling, ethics, law, privacy, data protection, non-discrimination, justice