Focusing on the core direction of China's high-quality development of intelligent teaching and learning, and taking the change driven by the vertical large model of education as a breakthrough, we deeply analyse the key constants and essential attributes in the process of digitalisation of education. It reveals the heterogeneous impact of shallow and deep learning characteristics on learning concepts and learning effectiveness, and systematically clarifies the complex interaction mechanisms of learning feedback, methods, concepts and contexts in enhancing learning effectiveness. These findings not only provide solid theoretical support for the transformation of China's digital education, but also provide an innovative practical path for building a high-quality and sustainable education development model, which further empowers the effective implementation of the national high-level education innovation system and talent development strategy.