The Ethics of Algorithms

Algorithms have become a core societal infrastructure that shapes our environment and experience, whether at an individual or at a group level. As volumes of data keep growing and computational techniques keep on being refined, algorithms become increasingly more valuable tools to parcel and analyse data to extract information, which contributes to define individual lives and the functioning of our societies. We all interact with recommender systems on a daily basis, be it to select a song, a movie, a product, or a “friend”. At the same time, schools and hospitals, courts, local governmental bodies, and national governments, all rely on algorithms to make their operations and decisions.

The potential of algorithms to improve individual and social welfare is coupled with significant ethical risks. This is because, algorithms are not ethically neutral. In this paper we build on the conceptual map of ethical problems related to algorithms proposed by Mittelstadt et al. (2016) to review the current debate on the ethics of algorithms and identify what ethical problems algorithms raise and what solutions have been offered in the relevant literature to address these problems. The map identifies six ethical concerns, which define the conceptual space of ethics of algorithms as a field of research. Three of the ethical concerns refer to epistemic factors that span from inconclusive and inscrutable evidence to misguided evidence; two are explicitly normative, i.e. unfair outcomes and transformative effects; while one – traceability – is relevant both for epistemic and normative purposes.