Hastings Science and Technology Law Journal


Jake Feiler


Not only is Artificial Intelligence (AI) present everywhere in people’s lives, but the technology is also now capable of making unpredictable decisions in novel situations. AI poses issues for the United States’ traditional criminal law system because this system emphasizes mens rea’s importance in determining criminal liability. When AI makes unpredictable decisions that lead to crimes, it will be impractical to determine what mens rea to ascribe to the human agents associated with the technology, such as AI’s creators, owners, and users. To solve this issue, the United States’ legal system must hold AI’s creators, owners, and users strictly liable for their AI’s actions and also create standards that can provide these agents immunity from strict liability. Although other legal scholars have proposed solutions that fit within the United States’ traditional criminal law system, these proposals fail to strike the right balance between encouraging AI’s development and holding someone criminally liable when AI causes harm.

This Note illuminates this issue by exploring an artificially intelligent trolley problem. In this problem, an AI-powered self-driving car must decide between running over and killing five pedestrians or swerving out of the way and killing its one passenger; ultimately, the AI decides to kill the five pedestrians. This Note explains why the United States’ traditional criminal law system would struggle to hold the self-driving car’s owner, programmers, and creator liable for the AI’s decision, because of the numerous human agents this problem brings into the criminal liability equation, the impracticality of determining these agents’ mens rea, and the difficulty in satisfying the purposes of criminal punishment. Looking past the artificially intelligent trolley problem, these issues can be extended to most criminal laws that require a mens rea element. Criminal law serves as a powerful method of regulating new technologies, and it is essential that the United States’ criminal law system adapts to solve the issues that AI poses.