Bibliogr. s. 60-64

Artificial intelligence is an increasingly popular tool in the field of healthcare and medicine, especially in medical devices. On the other hand, the possible unpredictability of its actions due to the so-called self-correction of software or the issue of predictability of the results of machine learning and deep learning raises questions about the basis and scope of legal liability. From the perspective of civil liability, the authors of this article analyze three other types of liability operating in the Polish legal system: (1) liability for a dangerous product, (2) liability based on fault, and (3) liability based on strict liability. Of these, the claim for compensation for damage caused by an AI-equipped medical device seems to be the most justified under the liability for dangerous products. However, even this standard does not cover all systems of damage caused by the operation of AI, especially in the case of AI that operates only as software and thus in isolation from a material object. In the absence of a unified legal regulation in the Polish legal system, some of the damage caused by AI will remain uncovered.