Abstract

The article defines the AI system contained in the EU regulation on artificial intelligence. The purpose of the analysis is to evaluate whether the definition is accurate enough in order to differentiate between AI systems and other information systems, including traditional computer software.
The author analyses the way various interpretations of the definition may influence the classification of systems as AI systems, which has direct consequences for subjects developing and using such technologies.
The results of the analysis indicate that the definition of the AI system in the AI Act is unclear and may lead to contradictory interpretations. The lack of accurate definitions of key concepts, such as autonomy and the ability to adapt, restricts unequivocal classification of AI systems, furthermore, the definition contains elective components, which additionally cloud the picture. The author suggests that, in practice, interpretation of the definition of the AI system will be achieved based on the guidelines of the European Commission, which will help to accurately set boundaries between AI systems and other information systems and will act as official interpretation.