Abstract

Is the activity of software a neutral one? The set of algorithms and big data which help the framework of an artificial intelligence in recognizing, classifying or elaborating data, could be considered truly objective? As a matter of fact, data are never objective: statistical models represent reality as they modify it. Furthermore, we are living in the era of learning machines: robots are everyday more intelligent, useful and independent from human control. As they are implied in a long list of activities, we need to know which rules they have to respect. Most of all, we need to establish who, and in what terms, could be responsible for their faults. That is why the European Union is elaborating a set of principles and rules on Robots and Artificial Intelligence: new technologies have a great potential for the benefit of humanity, provided that ethics and law do not give up on ruling them.