Supervision Debt

This paper introduces the concept of Supervision Debt, analyzing how current autonomous systems impact cognitive work by: • Transferring verification burden to humans • Creating an imbalance between output generation and human validation • Leading to increased metacognitive demand • Fostering cognitive complacency It explores a critical challenge in incorporating advanced computing systems into knowledge work. While these systems promise enhanced productivity through scalable output generation, they often shift the responsibility for accuracy, contextual relevance, and operational reliability onto the human operator. This creates a measurable divergence where perceived efficiency gains are offset by the growing need for human oversight. The paper demonstrates that this imbalance results in "Supervision Debt," a cumulative supervisory effort that silently erodes the benefits of high-throughput systems. Drawing on empirical findings from various fields like legal practice and software development, it highlights how human capacity for validation becomes the central constraint, rather than the system's ability to generate content. This reframes the core challenge, emphasizing the need to align computational power with realistic human validation capabilities. This research is essential for developers, researchers, and organizations deploying advanced computing systems, providing insights to design more balanced and genuinely productive human-computer collaboration models.