Intelligent and Non-Intelligent Supervisory Data-Driven Tools: A General Legal Framework for Proportionate Oversight and Reciprocal Control within the Theory of Data-Driven Tools, with Application to AI-Assisted Competition Law Enforcement

Abstract

The expanding use of intelligent and non-intelligent data-driven tools across technology-mediated environments has improved operational efficiency while increasing the legal significance of errors in data acquisition, processing, state detection, and automated execution. Building on the “Theory of Intelligent and Non-Intelligent Data-Driven Tools,” this study develops the “Theory of Intelligent and Non-Intelligent Supervisory Data-Driven Tools” as a cumulative and integrative legal framework for continuous oversight. The framework distinguishes executing from supervisory tools, differentiates intelligent and non-intelligent supervisors, and requires supervisory capacity to be proportionate to the nature and complexity of the monitored tool. It further incorporates supervisory independence, reciprocal control, dual recording of assessments, and preserved human intervention. Where a potential error, anomaly, or violation is detected, the supervisory tool records and reports it, while the executing tool reassesses the determination and both assessments are retained for human review. The framework will be developed as a general legal theory applicable to intelligent and non-intelligent data-driven tools and will be tested through selected AI-assisted competition-law enforcement scenarios to assess its implications for procedural fairness, evidentiary reliability, explainability, and the rights of defence while preserving computational efficiency. Keywords: Supervisory Data-Driven Tools; Intelligent Data-Driven Tools; Non-Intelligent Data-Driven Tools; Data-Driven Oversight; Supervisory Proportionality; Reciprocal Control; Human Oversight.