On the status of machine learning inferences in data privacy economics and regulation

Advances in machine learning technology, combined with fantastic growth in the volume and variety of data that firms capture about individuals, have led to increased public awareness of the question of data privacy. However, as reported in the relevant literatures, the machine learning technology poses challenges to resolving these questions. The information economics literature on data privacy discusses challenges to economic modeling of the problem, due to the externalities that machine learning exhibits; the literature that uses a more regulatory approach to data privacy cites different challenges, such as widely held assumptions about a firm’s right to make inferences from their data, which create resistance to any attempts at regulation. It thus emerges that just when machine learning technology has created a need for advances in data privacy policy, the characteristics of the technology itself are hindering progress. The observation that is the point of departure for this chapter is that paradoxically, by delving a bit into some technical distinctions about how machine learning works, we are able to partially defuse these apparent stumbling blocks, or at least, sharpen our understanding of them. This chapter highlights the reported challenges to crafting data privacy policy for machine learning, and attempts, through discussion of some technical distinctions, to clarify the challenges and point to some possible ways forward.