Artificial Intelligence and Moral intelligence

Abstract

We discuss the thesis that the implementation of a moral code in the behaviour of artificial intelligent systems needs a specific form of human and artificial intelligence, not just an abstract intelligence. We present intelligence as a system with an internal structure and the structural levels of the moral system, as well as certain characteristics of artificial intelligent agents which can/must be treated as 1- individual entities (with a complex, specialized, autonomous or selfdetermined, even unpredictable conduct), 2- entities endowed with diverse or even multiple intelligence forms, like moral intelligence, 3- open and, even, free-conduct performing systems (with specific, flexible and heuristic mechanisms and procedures of decision), 4 – systems which are open to education, not just to instruction, 5- entities with “lifegraphy”, not just “stategraphy”, 6- equipped not just with automatisms but with beliefs (cognitive and affective complexes), 7- capable even of reflection (“moral life” is a form of spiritual, not just of conscious activity), 8 – elements/members of some real (corporal or virtual) community, 9 – cultural beings: free conduct gives cultural value to the action of a ”natural” or artificial being. Implementation of such characteristics does not necessarily suppose efforts to design, construct and educate machines like human beings. The human moral code is irremediably imperfect: it is a morality of preference, of accountability (not of responsibility) and a morality of non-liberty, which cannot be remedied by the invention of ethical systems, by the circulation of ideal values and by ethical (even computing) education. But such an imperfect morality needs perfect instruments for its implementation: applications of special logic fields; efficient psychological (theoretical and technical) attainments to endow the machine not just with intelligence, but with conscience and even spirit; comprehensive technical means for supplementing the objective decision with a subjective one. Machine ethics can/will be of the highest quality because it will be derived from the sciences, modelled by techniques and accomplished by technologies. If our theoretical hypothesis about a specific moral intelligence, necessary for the implementation of an artificial moral conduct, is correct, then some theoretical and technical issues appear, but the following working hypotheses are possible: structural, functional and behavioural. The future of human and/or artificial morality is to be anticipated.