Artificial intelligence can be successfully applied to most aspects of social and professional life, including the work of lawyers and the justice system. The possibilities for its use in legal practice and court proceedings are very broad, ranging from streamlining lawyers’ work—for example, by drafting contracts and verifying documents—to AI issuing judgments. Can AI judges be more impartial than human judges? Research shows that human judges are never free from their own biases, which is an inherent aspect of being human.
Pursuant to Article 3 of the Artificial Intelligence Act, as adopted by the European Parliament on June 14, 2023, “artificial intelligence system” means a machine system that is designed to operate with varying degrees of autonomy and that is capable—for explicit or implicit purposes—of generating outputs such as predictions, recommendations, or decisions that affect physical or virtual environments.
When it comes to the technical aspects of AI that should be taken into account when considering the potential applications of AI in the work of lawyers and in the justice system more broadly, these include, first and foremost, machine learning, natural language processing (NLP), and the creation of artificial neural networks that would be trained in legal reasoning, inference, and legal logic. To properly harness the potential of AI for issuing rulings, it is essential, on the one hand, to develop an appropriate algorithm and enable the AI system to “learn” certain patterns; on the other hand, it must be emphasized that creating a functional AI-based tool is impossible without an adequate database containing a very large amount (thousands, tens of thousands) of relevant data points necessary for the AI system to learn. When developing algorithms that use AI to issue rulings in court proceedings, it is necessary to create a database and information repository that will enable the algorithm to learn the legal provisions, analyze the facts and evidence, and then apply the relevant legal provisions.
The use of AI systems in the justice system is classified as high-risk.
This means that entities using high-risk AI systems would have a number of obligations, such as data and risk management, detailed technical documentation for accountability purposes, extensive disclosure requirements toward users, as well as system security controls covering its entire lifecycle; failure to comply with these obligations would result in financial penalties.
