This study evaluates ChatGPT-4’s potential to support validation of Q-matrices and analysis of complex skill–item interactions. By comparing its outputs to expert benchmarks, we assess accuracy, consistency, and limitations, offering insights into how large language models can augment expert judgment in diagnostic assessment and cognitive skill mapping.