Abstract

Artificial intelligence systems increasingly shape how users consume information, interact with digital platforms, and make decisions online. Recommender systems used by platforms such as TikTok, Instagram, and YouTube continuously learn from user behavior to maximize engagement and economic value. As these systems determine what users see, how preferences evolve, and which information gains visibility, their underlying learning processes often remain opaque and difficult for users to understand or influence. Existing research has raised growing concerns regarding algorithmic opacity, user autonomy, and the ethical implications of AI-driven personalization, yet considerably less attention has been devoted to actionable design mechanisms that allow users to meaningfully understand, modify, or contest algorithmic behavior (Floridi et al., 2018; Recker et al., 2022). Current ethical AI discussions frequently emphasize high-level principles such as transparency, accountability, and fairness, but provide limited guidance regarding how these principles should materialize within user-facing AI systems. Drawing on Digital Responsibility and Authenticity–Control–Transparency theory, this research examines how algorithmic control mechanisms can support more transparent, accountable, and user-centered AI systems. Specifically, the study develops a design-science artifact that operationalizes responsible AI principles through interface-level control features embedded within a recommender system environment. The artifact enables users to view why recommendations occur, adjust personalization settings, limit algorithmic learning signals, and reset or modify recommendation trajectories. These features are designed to reduce the asymmetry of knowledge and control that currently characterizes many AI-enabled platforms. The study argues that responsible AI requires more than abstract governance frameworks or policy-level principles; it requires concrete design mechanisms that preserve user autonomy, transparency, and meaningful oversight during human–AI interaction. By giving users visible and actionable control over algorithmic processes, the proposed artifact seeks to restore user agency within increasingly opaque recommendation environments. The study further examines how varying levels of algorithmic control influence users’ perceptions of transparency, autonomy, trust, ethical responsibility, and satisfaction. This research contributes to Information Systems literature by translating responsible AI principles into concrete and testable interface-level design features for AI-enabled platforms. More broadly, the study highlights the growing need for AI systems that balance personalization and economic optimization with transparency, accountability, and user agency in increasingly algorithm-driven digital environments.

Download

Share

COinS