A Fuzzy Approach to Risk-Based Ethical Decision Making in Symbiotic Artificial Intelligence

The ontological and epistemic complexities inherent in the moral domain make it challenging to establish clear standards for evaluating the performance of a moral machine. In this paper, we present a novel approach to ethical decision making (EDM) based on the assessment of AI-related risks. In particular, we consider the case of Symbiotic Artificial Intelligence (SAI), where the close human–AI interaction may have a larger ethical impact. To address the inherent vagueness and uncertainty of moral reasoning, we rely on fuzzy logic to model ethical risk assessment and decision selection. The needed knowledge is encoded in a modular fuzzy rule base, and used to compute the ethical risk level for some scenario and suggest the most appropriate action to take. We formally define fuzzy EDM models and show how they can be transformed into fuzzy Petri nets, enabling formal verification and validation. A case study of SAI in the healthcare domain is considered to illustrate the proposed approach.