Causation in AI Tort Litigation: Legal Dilemmas of Algorithmic Black Boxes and Burden of Proof Allocation

Abstract

Through a comparative legal analysis of different jurisdictions, this study explores the difficulties algorithmic black boxes bring to traditional causation and the allocation of the burden of proof in artificial intelligence tort claims. The study employs a cross-jurisdictional doctrinal analysis and a methodical case study methodology to examine important AI tort cases related to algorithmic decision-making platforms, medical AI systems, and autonomous vehicles. Because courts cannot determine causal relationships using traditional evidentiary frameworks, the results show that traditional “but-for” tests of causation are intrinsically limited when applied to black box machine learning systems. Systemic differences in burden allocation mechanisms are revealed by cross-jurisdictional canvassing, with different jurisdictions implementing a range of strategies, from mandates for algorithmic audits and presumptive liability frameworks to stricter requirements for expert testimony.  According to the research, there are significant informational gaps between plaintiffs and AI system controllers, which calls for creative legal solutions like updated collective liability and causation presumptions. The results show that, while maintaining core tort law principles, legal frameworks must be modified to allow probabilistic algorithmic decision-making. Advocating for the shift to technologically adaptive liability regimes that strike a balance between victim protection and innovation incentives is also necessary. It is suggested that the aforementioned be put into effect through increased judicial technical proficiency and standardized transparency requirements.