Against the backdrop of the accelerated evolution of artificial intelligence (AI) technology, the integrated development of AI and industries has gradually become a trend and the core driving force for promoting industrial transformation and upgrading. As a result, the importance of intelligent assessment and structural analysis of industries has emerged and the demand is urgent. This study proposes an evaluation model (i.‍e., Patent-AIIIA) for the integrated development of AI and industry, which uses large language models to deeply explore the intrinsic connection between patent data and AI technology. Combined with the BGE (BAAI general embedding) model and the UMAP (unified manifold approximation and projection) algorithm, a problem space and a solution space of patents are formed. Based on the density peak clustering algorithm and the information entropy theory, an index quantification method for industrial span and technological width was constructed, and a comprehensive assessment of the entire industry and individual clusters was finally achieved. The Patent-AIIIA model was applied to conduct an empirical study on multiple types of industrial clusters in the advanced equipment manufacturing industry in a certain region. The indicators of industrial span and technological width, the distribution of AI technology industrial clusters, and enterprise technological benchmarking were analyzed, and the practicability of the relevant models was verified. Based on the assessment results, suggestions such as implementing targeted industrial incentive policies, optimizing the AI technology layout of industrial clusters, and building a collaborative innovation network with leading enterprises as the core were put forward. The Patent-AIIIA model can be regarded as a general analytical model with macro- and micro-characteristics. It has the application potential to expand into multiple fields and can provide data support and scientific basis for enterprise strategy planning and industrial policy formulation.