The paper aims to analyze the role of the concept of vulnerability in European Artificial Intelligence regulation (AI Act). The article will be developed in two parts. In the first we will conduct a conceptual analysis of vulnerability in its multiple dimensions, distinguishing between a universal and a particular sense, and between a categorizing and a situational approach. To do so, we will discuss the main lines of philosophical, legal and bioethical debate on the topic. In the second we will investigate the role of vulnerability in the AI Act, commenting on the most significant recurrences and major variations. We will adopt a synoptic look at the evolution of the document, from its first version to the final one, passing through the amendments proposed by the European Parliament. We will argue that in this document vulnerability is understood primarily in a particular sense, while the universal sense remains almost absent or not directly expressed. This absence, as we will show, seems only partially compensated for by the risk-based approach.
Il contributo si propone di analizzare il ruolo del concetto di vulnerabilità nella regolamentazione europea sull’Intelligenza Artificiale (AI Act). L’argomentazione sarà sviluppata in due parti. Nella prima ci dedicheremo all’analisi concettuale della vulnerabilità nelle sue molteplici dimensioni, distinguendo tra un senso universale e uno particolare, e tra un approccio categorizzante e uno situazionale. Per fare ciò, discuteremo le linee principali del dibattito filosofico, giuridico e bioetico sul tema. Nella seconda passeremo ad analizzare l’impiego del termine nell’AI Act, commentando le ricorrenze più significative e le principali variazioni. Adotteremo uno sguardo sinottico all’evoluzione del documento, dalla sua prima versione a quella definitiva, passando per gli emendamenti proposti dal Parlamento Europeo. Sosterremo che nel documento la vulnerabilità è intesa principalmente in senso particolare, mentre il senso universale rimane pressoché assente o implicito. Tale assenza, come mostreremo, sembra poter essere solo parzialmente compensata dall’approccio basato sul rischio.
Vulnerabilità. Note sul concetto nell'AI Act
Silvia Dada
2024-01-01
Abstract
The paper aims to analyze the role of the concept of vulnerability in European Artificial Intelligence regulation (AI Act). The article will be developed in two parts. In the first we will conduct a conceptual analysis of vulnerability in its multiple dimensions, distinguishing between a universal and a particular sense, and between a categorizing and a situational approach. To do so, we will discuss the main lines of philosophical, legal and bioethical debate on the topic. In the second we will investigate the role of vulnerability in the AI Act, commenting on the most significant recurrences and major variations. We will adopt a synoptic look at the evolution of the document, from its first version to the final one, passing through the amendments proposed by the European Parliament. We will argue that in this document vulnerability is understood primarily in a particular sense, while the universal sense remains almost absent or not directly expressed. This absence, as we will show, seems only partially compensated for by the risk-based approach.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
