Abstract

Recent growth in the adoption of Agentic AI technology within organizational workflows and processes holds tremendous promise for productivity gains and the potential to drive disruptive innovation. Unlike traditional AI systems such as Large Language Models (LLMs) that are prompted by humans, Agentic AI systems automate the human decision-making and workflow processes to a varying degree using a collection of tools, LLMs, and datasets to perform a series of sequential actions and different tasks with varying levels of human engagement (Acharya et. al., 2025). Such automation, while improving efficiency, could have a detrimental effect on an individual’s own perception of their own worth and importance, a term referred to as ‘dignity’ in the social sciences. While much of the IS research has focused on Trust in AI systems, algorithm aversion, AI bias, transparency, and autonomy, scant research has investigated dignity within the context of AI. Dignity is a multidimensional construct that has received relatively less attention within the IS discipline (Joseph & Mate-Toth, 2026). In this study, we propose to study the impact that varying levels of automation resulting from Agentic AI use could have on different forms of dignity: Inherent dignity (moral standing), social dignity (interpersonal treatment), and subjective dignity (subjective dignity) at an individual level (Killmister, 2020). We expect this relationship to differ from traditional decision-support technology, as it is driven by AI that helps them adapt to different conditions and situations. While higher levels of automation are expected to negatively impact dignity, we hypothesize that lower levels of Agentic AI automation could empower humans by providing more information and decision support, thereby improving dignity outcomes. Further, these relationships are expected to be moderated by factors such as task competency and responsibility attribution. For instance, a worker with lower task competency may find Agentic AI use more empowering, while a person with higher task competency may feel that this infringes on their true worth and ability to perform tasks. A similar logic applies to responsibility attribution. Policy guidelines and work culture may dictate who is responsible for tasks, depending on the level of autonomy. The study is set up as a field experiment with three levels of agentic AI task design and random presentation to subjects with a particular context. Different levels of AI autonomy will be presented to study subjects. Findings have strong implications for researchers and practitioners. This research provides guidance to researchers on developing better insights into how Agentic AI systems could shape workplace perceptions. Future studies could focus on developing a better balance between automation and workplace or task satisfaction. Practitioners could benefit from deciding the level of automation to inject for different tasks or workflows.

Download

Share

COinS