The concept of AI: between science fiction and legal fiction

As a figment of the imagination, AI dates at least as far back as the eight century BCE, when, in the works of Homer, we find Hephaestus, the Greek god of smithing and craft, using golden handmaidens endowed with autonomous movement, perception, judgment, and speech, to execute simple, repetitive labour. These works of fiction circulated more than two thousand years before Karel Čapek first introduced the term “robot” to refer to artificial slaves. As an engineer’s blueprint, it is the third century BCE automata building treaties of Alexandrian mathematicians and engineers that first committed dreams of creating AI to paper. Buddhist legends from the same period in north east of India mention an army of automata built with know-how stolen from the Greeks guarding the Buddha’s relics. Similar stories are found in in Chinese, Sanskrit, Hindu, and Tibetan texts. 1 Human capacities to catch up with and “bring to life” ideas of AI relegated today to the literary genre of science fiction may have greatly improved in the meantime. What stayed the same is the all too human yearning of having one’s repetitive work and drudgery performed by some other entity, subordinate to our will. Since building such entities requires vast amounts of money and expertise, it follows that such advantages as respite from work will be enjoyed and profited from by those few possessing the means and who are already in positions of power. This is where the law should come in to ensure balance between the rights and interests of all parties concerned. The trouble is that the anthropomorphic characteristics of AI are disrupting the traditional legal and ethical categories that conceptualize and organize relationships between humans and things. Their autonomy, adaptivity and self-learning abilities entail a certain degree of unpredictability – arguably emancipatory, from inanimate things to becoming akin to humans. This prompted even the EU Parliament to entertain the idea of granting “at least the most sophisticated autonomous robots” the status of “electronic persons”, met with strong backlash, since it appears a very radical solution to so-called “liability gaps”. Radical because, on the one hand, it calls into question the relative immobility of the law as to the person-thing binary and, on the other hand, because it entails deciding whether an artificial entity should be endowed with (the same set of) rights and duties (as a legal subject). Debates thereof become the scene of a tension between two apparently irreconcilable positions, namely realism and legalism. The first questions the ontological specificity and substrate of legal subjects. The second stems from the practical function of law, to organize and pacify relationships between legal subjects, for which purpose it creates legal fictions. This contribution intends to re-examine the “ancestral” division that Western morality and law have posited between people and things, exploring its conceptual and pragmatic dimensions