As intelligent science and technology continue to advance, AI-enabled subject teaching has emerged as a vital component of school teaching. This paper applies the research methods of hierarchical analysis and Delphi method to analyze the indicators of AI-enabled subject teaching effect. First, the Delphi method was utilized to identify 3 primary indicators and 18 secondary indicators, while the hierarchical analysis method was employed to calculate and determine the weights of the evaluation indicators. And then a systematic assessment framework for the instructional effectiveness of AI-enabled disciplines grounded in fuzzy neural network was developed, with a view to providing reference for assessing the instructional outcomes of AI-integrated disciplines. Representative teaching samples were gathered to validate the reliability of the fuzzy assessment framework, and the optimal training effect was achieved when the model was trained for 60 times, with an error value of only 0.002355. The root mean square deviation of the model was 1.9432, satisfying the practical assessment benchmark for the instructional outcomes of AI-empowered disciplines, and is able to accurately reflect the comprehensive effect of AI-empowered disciplines teaching. It provides a new methodology for promoting the precise and scientific development of AI education application.
