The Quest for Beneficial AI

摘要

Artificial Intelligence has been around for several decades, but it is now becoming mature enough to make an indelible imprint on the society. After going through several short periods of time in which AI winters were declared, since excessive promises had proved to be difficult to fulfil, we are now facing a definitive shift that make even mega companies to preach an “Artificial Intelligence-first” kind of world. We here clearly state our belief in AI potentialities while warning against the current hyping of its near future, arguing in favor of a more cautious interpretation of the AI-based systems potential outreach. AI is nowadays experiencing a new age of social recognition as well as increasing the need for ethical responsibility. A number of relevant applications are indeed improving human life quality through helping and collaborating in solving complex problems, diagnosing and predicting, gaming and entertainment and Intelligent Robots development. Although Artificial Intelligent is fascinating and has a tremendous social impact, our main ethical concern should mostly be about autonomous oeIntelligent software developed in the AI field for many different purposes including to control autonomous Robots. Intelligent Robots, strategic games winning programs or the more modest AI assistants, are all creatures of artificial intelligence and machine learning, built with the best intentions. The danger is that they will behave in accordance with the information we feed them, which is often incomplete, unbalanced or biased. Moreover, they can evolve and learn guided by either the best or the worst ethical principles. Hyping AI may lead once again to suggest wonderful horizons that we are not at all sure they will happen. This does not prevent us from discussing the possibility of approaching some kind of Artificial General Intelligence that could make us think about artificial consciousness. However, it is mandatory that AI and Robotics main objectives must always be to maximize the realization of human values. We finally advocate a set of practices and principles that may prevent the development of AI- based programs and systems prone to be misused for the bad of humans and raise some ethical issues that are being discussed since now.