Automated Decision Making and the Swiss Principle of the Constitutional Judge
Caroline Géraldine SCHÖNHOLZER
University of Bern, Switzerland
It is said that the Fourth Industrial Revolution is affecting all areas of work in ways that are hard to predict, and that the legal profession will be significantly affected by this industrialisation of law. Legal AI technology is reshaping the demand for, and functionality of legal services in both private and public sectors. In the name of increasing efficiency, harmonising the application of the law, improving quality, saving costs and more, AI technology has already been introduced into the judiciary in some countries, where AI is assisting court administration as well as the judges themselves. And even in countries where no AI technology has been implemented by the state, the use of this technology cannot be stopped, as AI technologies such as ChatPGT or Claude Sonnet are open to everyone (and even free to some extent).
Can robots be judges? Can we automate judicial decision-making? While this question may seem tedious in today's discussions about the application of AI in the legal sphere, the issues it raises are many and highly relevant as the technology develops at a rapid pace in unknown directions. In Switzerland, there is no specific AI regulation (yet), and AI technology has also not (yet) been implemented in the judicial decision-making process. However, this is potentially only a matter of time and technological development, which makes the timing of this research project favourable.
My research focuses on one aspect of the original question. I want to understand how the application of AI technology in the judiciary, and in particular the automation of judicial decision-making, is compatible with the Swiss right to a constitutional judge. This right includes the right of access to a court constituted by law and the aspect of its institutional independence. Judges are democratically legitimised functionaries who are regarded as the guardians of an independent judiciary under the rule of law, which also includes the trust in the judicial process and, in general, the legitimacy of the courts in a democratic constitutional state.
In order to answer these questions, three aspects are relevant: firstly, a technical basis has to be established together with an analysis of the technical standards of some pioneering states. In a second step, the legal basis for the application of automated decision-making systems by the judiciary in Switzerland will be examined by analysing the Swiss and international regulation of AI as well as the principle of the constitutional judge. The third part is dedicated to an outline of the legal issues that arise in the application of automated decision-making systems by the judiciary, so that in the last part the main results can be summarised, and conclusions can be drawn.
The primary objective of this presentation is to provide insight into the research project, its intended structure, preliminary findings, and key open questions.
Generative AI and European Courts: Between the pacing problem and judicial guidelines
Petra GYÖNGYI
Lund University, Sweden
This contribution aims to connect socio-legal scholarship on the integration of generative AI systems in European courts and their administration, with the relevant provisions of the EU AI Act concerning General Purpose Artificial Intelligence (GPAI). Drawing on socio-legal studies on technological governance and notions of the pacing problem between law and new technologies, the article: i) offers a critical reading of the EU AI Act’s systemic risk GPAI obligations (such as GPT-4) and ii) explores how non-binding judicial guidelines on the use of AI could become vehicles for addressing the pacing problem to enhance judiciary performance and address citizens’ expectations regarding judicial performance in the digital age. The article aims to highlight the complimentary nature of different legal mechanisms and how layered governance mechanisms could aid transformations in judicial services. Overall, the paper broadens the theoretical understanding of technological governance and the practical toolkit from a court administration perspective navigating the integration of AI in judicial systems.
Predictive Justice and Administrative Trial: The Role of Judicial Precedent
Piera VIPIANA, Matteo TIMO, Giovanni BOTTO, Sara SCAZZOLA
Università di Genova, Italy
The issue concerning the use of artificial intelligence systems in the legal field is particularly interesting, as well as complex and fraught with consequences, in the context of justice. The interest of the topic is entirely evident also following the adoption of the European Regulation on Artificial Intelligence (the so-called "AI Act"), which sets the fundamental objective of regulating the use of artificial intelligence at the European level in a way that renders it trustworthy, and within which, as is known, the matter of "justice" is included in the list (as per Annex III of the Regulation) of high-risk actions, that is, those applications that, subject to rigorous rules, are considered "acceptable". This research aims to delve into a topic of great relevance: the role of judicial precedent, with particular reference, of course, to the context of administrative justice.
In particular, within this macro-theme of interest, the authors participating in the panel will concentrate their presentations on some specific topics.
A first and introductory presentation will concern, following a framing of judicial precedent in administrative proceedings, the theme of the relationship between the latter and the use of artificial intelligence systems from the perspective of so-called "legal calculability", that is, the possible impact of automation tools on the predictability of judicial decisions. The use of new technologies in the procedural field, in fact, touches deep chords of the legal system, going so far as to call into question the very relationship that exists between judicial decisions and legal norms, and, logically, the principle of judicial subjection to the law. It is well to clarify – and an attempt will be made to argue it – that the possibility of "predicting", with greater precision than in the past, the decisional outcome of the judge, does not automatically imply a proportional benefit on the side of calculability understood in a substantial sense.
A second presentation, instead, will concern the application of "predictive" systems for the purposes of adopting simplified judgments as per Article 74 of the Italian Code of Administrative Procedure. The particular angle of perspective chosen seemed worthy of analysis, by reason of the specific reference to "judicial precedent" made by the cited norm, which contributes to dissolving part of the argumentative tensions inherent in the procedural use of computer systems that, notoriously, base their operation on the repertoire of past judgments.
Finally, the last presentation will deal with the interesting problem of the relationship between artificial intelligence tools and interim relief in administrative proceedings. There are at least two reflections of great relevance on this topic: the assessment of the prerequisites for interim relief and the issue concerning single-judge protection and the impartiality of the panel that must express itself on it (a mechanism provided for by the Italian procedural code, as well as by other regulations at the European level).
|
