KRIVIČNA PRAVDA U ERI VEŠTAČKE INTELIGENCIJE

Abstract

Artificial intelligence (AI), as an extension of the digital transformation of society, is becoming an integral part of criminal justice. The paper analyzes the use of AI in criminal law, as well as the risks that its use poses for criminal law. AI is not only a simple tool of criminal justice but also a path to the transformation of criminal law. The first part of the paper analyze the concept of AI and its influence in the world of the Internet of Things (IoT) in the way of detection of criminal acts, prevention and assessment of the degree of risk of recidivism. As AI is prone to prejudice (AI bias), algorithms of “predictive justice” software can have a discriminatory effect, which needs to be identified. It is essential that algorithms are subject to independent and thorough review of whether there is bias in the data generated by the criminal justice system. In the judicial systems of technologically developed countries, AI also acquires the function of helping the judge to perform the judicial work faster and more efficiently. The application of “machine learning” technology to automatically generate judgments for judges leads to numerous risks, especially highlighted in criminal cases. Questions are raised about the reshaping of the role of the judge, and the right to be informed about the logic of decision-making and to challenge the scientific validity of the algorithm, which is an element of the principle of equality of arms. AI has the character of autonomy, so the question inevitably arises whether there is criminal responsibility for artificial intelligence, to what extent and in what way, and whether it is necessary to create new forms of responsibility and new subjects of responsibility in criminal law? In the sphere of substantive criminal law, theoretical considerations point to the potential regulation of the responsibility of producers, developers and users for actions taken by the AI.