Probabilismo ‘artificiale’ e certezza del diritto penale: un finto ossimoro

If the judicial decision were completely logical-deductive and without appreciable margins of choice – as the Enlightenment advocated – this would decisively contribute to legal certainty. Interpretive activity, on the other hand, often does not lead to a single ‘right’ outcome, if one can call it that, but to several her-meneutical alternatives, thus creating an interpretative doubt that should be resolved in accordance with constitutional principles, in particular the favour rei. Before this, of course, one must ask oneself whether it is any doubt or only rationally founded doubt that legitimises a hermeneutic solution favourable to the offender, as is the case. To avoid giving prominence to any doubt, but only to what we might call legitimate doubt, one can resort to the ancient theory of probabilism, which is useful in delineating the reasonableness of interpretive doubt. The theory of pro-babilism, therefore, may represent the philosophical basis, consistent with constitutional principles, on which to base the possible operation of decision-making algorithms in the penal system. Artificial intelligence systems, in fact, among the possible interpretative solutions resulting from the elaboration of case-law precedents, should opt not for the statistically freest or ‘most probable’ one, but for the most favourable one, which is also ‘probable’, serious, because it is based on other case-law precedents.