Transparancy requirements for algorithms and Ai, wishful thinking?

Algorithms are the basic engine of AI and are increasingly supporting and channelling the outcome of deciscion making in all kind of different areas . Algorithms will drive many different applications, from industrial products to administrative and governmental decision making. This includes relevant legal decisions that will require information for data-subjects. The GDPR requires transparency of decision-making and explainability as well as information to the datas-subjects. In what respect and to what extent is this demand feasible? Must algorithms always be explainable or do we accept the black box to be black for different functionalities. Is a separate legal framework for algorithms necessary? It is not always necessary to know what is under the hood of the car to drive...