Tort Law, Corrective Justice and the Problem of Autonomous-Machine-Caused Harm

Developments in artificial intelligence and robotics promise increased interaction between humans and autonomous machines, presenting novel risks of accidental harm to individuals and property.1 This essay situates the problem of autonomous-machine-caused harm within the doctrinal and theoretical framework of tort law, conceived of as a practice of corrective justice. The possibility of autonomous-machine-caused harm generates fresh doctrinal and theoretical issues for assigning tort liability. Due to machine-learning capabilities, harmful effects of autonomous machines may be untraceable to tortious actions of designers, manufacturers or users.2 As a result, traditional tort doctrine—framed by conditions of foreseeability and proximate causation—would not ground liability.3 Without recourse to compensation, faultless victims bear the accident costs of autonomous machines. This doctrinal outcome reflects possible incompatibility between tort’s theoretical structure of corrective justice and accidents involving autonomous machines. As a practice of corrective justice, tort liability draws a normative link between particular defendants and plaintiffs, as doers and sufferers of the same tortious harm, grounding defendants’ agent-specific obligations to repair the harm. Where accidents are caused by autonomous machines, the argument goes, the essential link between defendants and plaintiffs is severed; since resulting harm is not legally attributable to the human agency of designers, manufacturers or users, victims have no remedy in tort.