The Efficiency-Accountability Tension in High-Accountability AI-Augmented Work: A Process Framework of Verification, Decision Density, and Anxiety

Generative AI is often framed as a time-saving technology, yet productivity gains may not reduce perceived time pressure in high-accountability work. This conceptual paper develops a bounded process framework explaining when AI-assisted efficiency may become subjective time poverty and anxiety. Integrating Social Acceleration Theory, the Job Demands-Resources model, and human-AI interaction research, we theorize an efficiency-accountability-control tension: AI accelerates upstream production while humans retain downstream verification, explanation, sign-off, and responsibility for consequences. We specify three mechanisms: liability-control mismatch, verification-intensity shift, and ambiguous micro-decision density. The framework applies to high-accountability AI-augmented knowledge work and is distinguished from low-stakes ideation and creative exploration. To support future empirical testing, we outline operationalization pathways and mechanism-aligned redesign levers. Ultimately, this framework reframes AI-related strain from a technological issue to a work-systems challenge, pointing to responsibility allocation, user control, system visibility, and work pacing as central design concerns for a psychologically sustainable AI transition.