The forgotten human autonomy in Machine Learning

Abstract

There are many rights, freedoms, and principles that build our society and nourish it day after day. No right, freedom, or principle is absolute; they must always be balanced. Our starting point is the respect for internationally recognized human rights and we focus on the principle of autonomy, which is not being adequately treated and protected in the ever-changing algorithmic world. In this article we review some of the most influential bodies of knowledge and ethical recommendations in artificial intelligence, and analyze the extent to which they address the principle of autonomy. We ground the concept of autonomy in operational terms such as being well-informed and being able to make free decisions. Under these two different aspects, we analyze the technical and social risks and propose different ways in which artificial intelligence requires further exploration with the aim of preserving human autonomy.