After an examination of the different models of artificial intelligence in the world of law (rule-based reasoning, case-based reasoning, generative AI) and an exposition of the fundamental principles that inspire international charters and draft laws on the subject, the Author dwells on the implications that a “replicating” algorithm, based on case law precedents, may have on criminal legality if forms of predictive artificial justice were to be admitted, aimed at anticipating, or supporting or even replacing the judicial decision. If this were to happen, the Author proposes a model of artificial intelligence operation based not on purely statistical criteria, but on favor rei, as a prerogative of the penal system, a model that also provides for a rebutative motivational burden on the human judge should he not wish to conform to the more favourable decision suggested by artificial intelligence.
Dopo la disamina dei diversi modelli di funzionamento dell’intelligenza artificiale nel mondo del diritto (rule-based reasoning, case based reasoning, IA generativa) e l’esposizione dei princìpi fondamentali che ispirano le carte internazionali e i progetti di legge sul tema, l’Autore si sofferma sulle implicazione che un algoritmo “replicante”, basato sui precedenti giurisprudenziali, può avere sulla legalità penale qualora si ammettessero forme di giustizia artificiale predittiva, volta ad anticipare, oppure a supportare o, addirittura, a surrogare la decisione giudiziale. L’Autore, qualora ciò avvenisse, propone un modello di funzionamento dell’intelligenza artificiale basato non su criteri puramente statistici, ma sul favor rei, come prerogativa del sistema penale, modello che prevede anche un onere motivazionale confutativo del giudice umano qualora lo stesso non si volesse conformare alla decisione più favorevole suggerita dall’intelligenza artificiale.
La giustizia artificiale replicante
Mario Caterini
2025-01-01
Abstract
After an examination of the different models of artificial intelligence in the world of law (rule-based reasoning, case-based reasoning, generative AI) and an exposition of the fundamental principles that inspire international charters and draft laws on the subject, the Author dwells on the implications that a “replicating” algorithm, based on case law precedents, may have on criminal legality if forms of predictive artificial justice were to be admitted, aimed at anticipating, or supporting or even replacing the judicial decision. If this were to happen, the Author proposes a model of artificial intelligence operation based not on purely statistical criteria, but on favor rei, as a prerogative of the penal system, a model that also provides for a rebutative motivational burden on the human judge should he not wish to conform to the more favourable decision suggested by artificial intelligence.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


