This article is a kind contribution to the blog by Alessandro Burgognoni.
Ethics has always accompanied human life, ever since the necessity of social living led us, or at times forced us, to establish social rules of behavior, so that we could live in a way that offered a greater chance of survival within a social group, or in order to guarantee “from above” a hidden or latent form of control.
From the law of the strongest to respect for others, whatever their strength or social position, there are many shades in between, and sharing an ethic is not enough for it to be applied, because the ethic itself is a mediation between libertarian ideas and the will to power that have always accompanied human existence.
To be an ethic, it must be recognized by the individual and by the group they belong to, so that if certain acts are considered socially unacceptable by that group, the action will be sanctioned in the ways and according to the timing of that social group.
Our behaviors, the product of a thousand family, religious and social rules, often unwritten but confirmed by everyday acceptance, are shaped by circumstances, by our mental evolution and experience in life, and by the emotional moments that make it up.
Life runs along defined, often constraining tracks, which bring our behaviors back to the level of responsibility for our actions towards the group we belong to, to its behavioral conventions and to the acceptability of the actions that can be carried out without being sanctioned.
Today, after years of theoretical research on human ethics, we find it still shaky in everyday life, just as human nature is, just as the desire to project ourselves onto the world, the desire for social recognition at all costs and the desire of some to dominate are in conflict.
The ethic suggested or imposed as a social model is continually contradicted by gestures, speculation, fraud, abuse of power, the taking of lives and every form of failure to recognize the value of others in society.
Nothing new, however: in different forms it has always manifested itself in history, the desire of some to go against social rules, not only to spite common thinking and assert themselves or as a pure act of abuse of power but, rightly, to express their own independence and identity while still belonging to a “group”, or to make the group evolve towards a different ethic, often for reasons justified by the intolerable conditions of their present.
The inner drive that comes to us from our aggressiveness, which in modern times takes the form of a propulsive drive towards a conscious evolution of humankind, in others still manifests itself in the roots of our primordial aggressiveness, which seems not to have lost its strength despite a technologically advanced world; it is as if primitive man had put on a cloak of modernity without losing his essence: that of someone who defended himself or killed with blows of a club.
Our responsibility towards others has become more complex since our actions have been mediated by technology. Over the last fifty years, science and its application, technology, have progressively taken over the simple actions that people perform every day, replacing the manual gesture with the control of instruments, first analog and then digital.
In the short span of these last few years, the theoretical lines of artificial intelligence, born of the studies of the Second World War and their developments in the sixties and seventies of the last century, have found in the new technological means of computing and electronics the tools for their concrete realization, and we have progressively arrived at the construction of “artificial entities”, agents capable of acting in the real world and in the virtual world of the internet.
The extraction of tacit knowledge from data, and projection and prediction based on the knowledge acquired, are an asset applied more and more in everyday life. We are a society dominated by algorithms, by artificial neural networks which, by simulating the brain model in simplified form, operate, select, classify and create from the known, or from the synthesis of disparate data, something “new”, at times unexpected, as in algorithmic artistic creations or elaborations.
The Italian philosopher Luciano Floridi, who has long worked on the subject, devoting himself to machine ethics, has consolidated the idea that “…we have separated the capacity to act from intelligence”. In essence, the behavior of intelligent systems emulates apparently human behavior as if there were an intelligence “behind” it, but in reality there is only a highly complex rational, mathematical, logical, statistical and probabilistic process that achieves the goal without there being “something” behind it that is aware of the action or analysis carried out and of its consequences.
For us humans, awareness derives from existence, from the presence of a “Self”, of an identity aware of existing which, moment by moment, perceives living and itself, and is able to turn its attention not only to the outside world but also to the inner one; an intelligent system, however sophisticated, does not possess it. So is it wise to entrust our existence to intelligent systems?
This is where the discipline of “Machine Ethics” comes from, to which research groups all over the world are devoting themselves with the aim of predicting, steering and limiting in advance the behavior of intelligent machines, to prevent their actions in the real world from producing undesirable effects. The desire to move forward in technological development places us in a situation of progressive dependence on our discoveries, and in the case of artificial intelligence, the availability of new means and new solutions requires us to seek “correctness” and “fairness” in the solutions proposed by algorithms.
We obtain extraordinary results in the scientific field, in every field where algorithms are applied, but in the field of evaluation and decision-making – as in personnel selection – there have already been cases of results that gave rise to discriminatory or “biased” responses because of the data used as the evaluation pool for classification. The reasons are well known: that data had previously been selected with “biased” criteria by whoever had made it available to the system.
Artificial neural networks analyze according to the characteristics of the data made available to them. If the characteristics of the data tend to favor one type of “profile” or “attribute” over another, the result cannot fail to fall short of the fairness that would be required. Some scientists are searching for a “problem solver”, a “universal solver”, a “master algorithm” capable of tackling a problem with the same completeness of approach as a human being faced with the search for a solution.
Achieving this goal means bringing together all the schools of thought that have dealt with problem-solving methods and merging them into a single algorithmic model that is complete in itself. Research is far from a breakthrough, but after all, even without reaching these levels of complexity, our world has long been in the hands of computers – as the researcher Pedro Domingos, one of the proponents of the master algorithm, argues – “People worry that computers will get too smart and take over the world, but the real problem is that they’re too stupid and they’ve already taken over the world”.
Ultimately, machines endowed with the capacity for judgment and consequent action, for data analysis and for extracting knowledge from enormous volumes of data are all around us, and in the end they reflect what we are and what we desire; but because of their complexity they could go beyond our expectations and our desires and act or make choices that are humanly unacceptable, based solely on algorithmic logic and on the data made available.
Already today we have processes based on deep learning – complex artificial neural networks – that provide coherent solutions whose solving path we do not know, because it is enclosed in the depths of the network itself. This should show us the caution we need in order to obtain what we want, and push us to build effective systems for monitoring problem-solving networks so as to fully understand their underlying logic.
Finally, in the current situation, lacking stable reference values, we should try to understand whether our ancestral, often hidden, need to find an absolute point of reference is not leading us towards the mirage of building a new algorithm with hypothetical “omniscient” abilities, to give meaning back to existence through a “digital absolute”, someone to whom we can delegate responsibility, power and decisions, with the risk that those who have the means will take over this mechanism and use it for questionable or oppressive ends.




