Ethical assessments for a privacy-friendly artificial intelligence

Abstract

For quite some time now, there has been a tendency to claim that the law or different types of standards are not adequate or applicable to the different new technologies that are continuously appearing (cloud computing, big data, the Internet of Things, robots, Artificial Intelligence, etc.). This paper will attempt to turn this reasoning on its head and, focusing on Artificial Intelligence, propose suitable criteria for this technology and many others to be developed and applied appropriately to fundamental rights, in general, and privacy in particular. Within these criteria, an analysis will be made of what the explainability of the aforementioned technology should provide (a requirement introduced by different groups of experts) and some additional concepts, such as the possibility/necessity of controlling (to the point of being able to stop an Artificial Intelligence system at any time) and auditing Artificial Intelligence systems, will be explored in greater depth. In order to carry out the aforementioned in-depth study, we will use what are known as protection goals (availability, confidentiality, integrity, transparency, unlinkability, and ability to intervene) and the principles of Cyberethics, including a special reference to the sustainability and resilience of these types of systems.