Civil liability in accidents caused by artificial intelligence deep learning (case study: self-driving cars)

Artificial intelligence can have countless benefits for humanity, one of which is fully self-driving cars. With the entry of various branches of information technology, it is necessary to provide its non-technical infrastructure, such as explaining the legal aspects of the upcoming issues. Artificial intelligence can have countless social benefits, but at the same time, they may cause harm. The requirement for proper protection of consumers is that before the introduction of new technology along with the preparation of road infrastructure and legal infrastructure, it is also reviewed and modified. This research specifically and case-by-case deals with the civil responsibility of deep learning of artificial intelligence in a fully self-driving car. Who will be responsible when the AI in the self-driving car does not act according to the instructions given by the manufacturer or software developers, and the AI acts outside the framework of its instructions, according to deep (reinforcement) learning? How can artificial intelligence or those involved be held responsible by relying on the general rules of civil responsibility? After examining and pondering the laws, we find that several people, including the software designer, the self-driving car manufacturer, and the technical supervisor, may be identified as responsible, and the results of this research show that they are not up-to-date and blame-oriented. Iran's civil liability system is related to the civil liability of artificial intelligence operators.