Abstract
The chapter provides a quantitative model of assessing critical thinking in AI-enhanced STEM learning setting. The model being proposed converts the AI-generated data in interaction with the student into quantifiable cognitive aspects through feature representation, normalization, and aggregation by multiple criteria. A hybrid weighting framework based on Analytical Hierarchy Process, combined with entropy, will enhance reliability and objectivity. The framework involves optimization, validation, and implementation of algorithms to provide scalability and accuracy. As illustrated through experimental examination, it has proved to be useful in evaluating critical thinking among learners. The research adds a mathematically based, evidence-supported perspective that facilitates real-time assessment, individual learning, and evidence-based decision making in current AI-infused STEM education systems.