Algorithm and Learning. Ethical and Theoretical Hypothesis on Empathic Robotics and Social Robotics

The article explores the relationship between algorithms, machine learning and artificial empathy. The pivotal problem is whether and how artificial learning of emotions by artificial agents, guided by algorithms, is possible. The first question consists in an in-depth analysis of the concept of learning (through some references to anthropotechnics, to Aristotle, to Merleau-Ponty and to Bateson), showing that – even in the human field – learning involves the use of forms of automatic and procedural repetition, such as the body schema or the behavioural algorithm. The hybrid aspect (between technique and nature) of learning processes in a broad sense is therefore underlined. The second issue is emotional learning by artificial agents. Through the use of an extensive specific bibliography, especially in the IT-engineering field, the possibilities and limits of this possibility are shown, discussing the major knot, namely that the machine or the algorithm has no living body (Leib), but mechanical (Körper). In the artificial field, the repetition and learning of patterns appear to be very far from that of human beings. However, the article tries to stress the usual categories and to formulate hypotheses, highlighting the possibilities of heterogeneous hybridization between the repetition learned in the machine and its ability to complexify its own behaviours, making them much less dissimilar from those of natural agents. The third question is that of the different ways in which the ‘affective’ dimension of the machine can be interpreted, through a relational systemic approach (social robotics) in which artificial empathy takes shape only within the relationship with the other. Different topics are explored, from those of emotional grafting (emotional chip) to that of a form of autopoiesis in the direction of a form of artificial autonomy, today in a larval but not absent form, through references especially to Dumouchel and Guattari. In the concluding part, some hypotheses are formulated on a conceptual frame for managing the ethical issue of autonomy, potentially also on an emotional level, of artificial agents guided by algorithms, suggesting an attitude capable of thinking of an ethics of hybridization but at the same time a hybridization of the usual ethical attitudes, that goes beyond the opposition between roboethics and machine ethics.