PURSUING EFFICIENCY WHILE UPHOLDING ACCOUNTABILITY: The Efficiency–Accountability Paradox in GenAI– Assisted Knowledge Work

Abstract

Generative artificial intelligence (GenAI) is increasingly used by employees in commercial and operational roles to support everyday work activities. While these tools can improve efficiency by accelerating the creation of content, analyses, summaries, and communications, employees must still determine whether outputs are accurate, relevant, and suitable for organizational use. Although research has explored GenAI’s capabilities and productivity implications, less attention has been given to how employees pursue efficiency gains while upholding accountability in everyday work practices. This thesis examines how employees use GenAI to pursue efficiency while remaining accountable for outputs generated with GenAI, and how they navigate this tension in everyday work practices. Drawing on 14 semi-structured interviews with chartering and operations employees, the study adopts an interpretive qualitative case study approach. Findings show that GenAI use emerged through individual experimentation and informal adaptation, often before formal organizational guidance. Participants consistently treated GenAI outputs as drafts rather than final products. Before organizational use, employees engaged in verification, contextualization, and professional judgment to assess suitability. As a result, efficiency gains were accompanied by ongoing accountability demands, shifting employee effort from draft production towards verification, contextualization, evaluation, and approval of GenAI-assisted outputs. This study conceptualizes an efficiency–accountability paradox in GenAIassisted work, where employees gain efficiency while remaining accountable for the outputs they ultimately choose to own and use in the workplace. Consequently, efficiency gains and accountability demands persist simultaneously, creating an ongoing tension that employees must continuously navigate in practice.