On the verge of a new relationship between man and artificial learning machines?

Abstract

We are now living an exciting time in which astonishing results coming from the A.I. field trigger the usual expectations and fears about intelligent man-machine relationship. Some believe that the basic ingredients for the artificial general intelligence are already here and, in their opinion, it is just a matter of putting it all together. It seems that we do not have to program computers anymore; they will program themselves. Should this statement frighten us? Our position is both of recognizing the big potentialities of new A.I. developments and, simultaneously, to warn against yet another overselling and possibly damaging stage in the A.I. field. We advocate that, intelligent systems, relying in evolving machine learning algorithms, fully autonomous software agents or robots, should always follow the "human in the loop" principle, ensuring that the responsibility for all future intelligent entity activities can be traced back to some recognized and accountable individuals or organizations.