Date of Award

7-2024

Document Type

Dissertation

Degree Name

Doctor of Juridical Science (SJD)

Abstract

This dissertation examines Internet intermediary liability concerning algorithms, specifically Interactive Computer Services (ICS)’s utilization of machine learning-powered recommender systems. It posits that Section 230 of the Communications Decency Act (Section 230) and product liability are potential regulatory avenues to address the harms caused by this system. This research makes three substantial arguments. First, it identifies machine learning-powered recommender systems as pivotal within the formulation of informational capitalism, serving various economic functions such as enhancing consumer experiences, enabling behavioral advertising, and fostering social networks. The algorithmic harms associated with such systems encompass discrimination harm, informational harm, and mental harm. Second, through the interpretation-construction theory, this study analyzes the text of Section 230 and argues that, first, the linguistic meaning of “publisher” does not encompass ICSs' utilization of machine learning-powered recommender systems, and second, a two-factor analysis can be conducted on the meaning of “development” when platform uses algorithms. This research further proposes a legislative recommendation and explores whether such a proposal falls under the coverage of the First Amendment. Third, this research examines the intersection between product liability and AI. It proposes that the tangibility criterion for products should involve determining whether the item is a service and identifying the boundaries of the item when distributed commercially for use or consumption. In consideration of the difficulty in determining defects due to AI “autonomy,” it suggests employing the theory of “friction-in-design” to identify both design and warning defects of the product, as the user interface plays an instrumental role in causing algorithmic harms.

Available for download on Wednesday, July 25, 2029

Share

COinS