Botsitting: The Unpaid Labour Behind Every AI Productivity Claim (smarterarticles.co.uk)

A recent survey conducted by Glean's Work AI Institute reveals a critical insight into the hidden costs associated with AI tools in workplaces across the U.S., U.K., and Australia. While employees report gaining approximately 11 hours of extra productivity per week from AI, they spend an average of 6.4 hours on "botsitting"—the often unacknowledged work of overseeing, correcting, and validating AI outputs. This phenomenon highlights a significant disconnect between perceived productivity gains and actual human effort required to make AI systems functional and trustworthy. The findings suggest that organizations are underestimating the labor and resources necessary for effective AI integration, ultimately affecting performance metrics. The term "botsitting" encapsulates the challenge of managing AI systems, emphasizing a crucial point: efficiency is not measured merely by the time AI takes to generate results but also by the extensive validation labor that follows. With 69% of surveyed workers admitting to risking the quality of their outputs by submitting unverified AI-generated content, the integrity of work is at stake, leading to increased job dissatisfaction. This emerging discourse on botsitting calls for a reevaluation of productivity metrics and how AI's impact is recorded in the workplace, underscoring the need for greater recognition and compensation for the human labor intertwined with AI operations.