High-risk justice

This article examines the structural limits of decision-making automation in civil justice by taking enforcement proceedings as a paradigmatic test case. The enforcement phase is frequently portrayed as the segment most exposed to standardisation and, for this reason, as the privileged field for the application of artificial intelligence. Such a representation rests on the idea of forced enforcement as a formalised procedure strictly bound to the enforceable title and reducible to a sequence of technical operations devoid of evaluative content. The article challenges this assumption, showing that the alleged automatism of enforcement does not constitute a structural feature of the system, but rather the product of a historically and dogmatically determined construction, consolidated through modern codification and crystallised in the 1940 Code of Civil Procedure. Through a historical and systematic reconstruction, the contribution demonstrates that the enforcement paradigm incorporates a program of automatism which, in practical application, encounters an internal limit. Even within a procedure conceived as formally constrained, there emerge decisive junctures at which the enforcement judge is required to exercise powers of procedural governance, legality control, and assessment of the proportionality of coercive measures. These interventions take place within a sphere of endo-executive cognition which, although not amounting to a full determination of substantive rights, entails contextual, attributable, and reviewable decisions, necessarily requiring adversarial participation and reasoned justification. From this perspective, the analysis identifies the hypotheses of early termination of enforcement proceedings as an emblematic case. Precisely where the procedure appears most amenable to standardisation, decisions arise that cannot be reduced to an automated output without undermining judicial accountability and the controllability of jurisdictional activity. On this basis, the figure of the so-called “robot judge” is employed not as a futuristic provocation, but as a critical category through which to assess the compatibility between automation and procedural guarantees. The article concludes that artificial intelligence may legitimately serve a supportive function with respect to repetitive and organisational tasks, but encounters an intrinsic limit when it purports to replace the attributable decision-making core of the judicial function. Any technological innovation must therefore be confined within a paradigm of augmented intelligence, preserving human responsibility, motivation, and judicial accountability.