Abstract:

Generative artificial intelligence (GAI) can produce large amounts of content, but high-quality knowledge outputs depend on more than individual AI capabilities. This study proposes a Sustainably Evolving AI Knowledge Ecosystem by integrating the Knowledge Value Chain (KVC), the SECI Model (SECI), and the DIKW Pyramid (DIKW) into a three-layer architecture: Infrastructure, Fusion, and Decision. This architecture brings together human-AI collaborative thinking models, complementary AI approaches, and multidimensional insight. The study traces a ten-stage research journey, showing how repeated testing and model development led to the proposed ecosystem. Four research-developed configurations were compared with eight mainstream AI systems across five interdisciplinary tasks. Nine AI evaluators assessed response quality in two rounds, including randomized-order re-evaluation. All four research-developed configurations held the top four overall rankings in both rounds. These findings support further investigation of the proposed approach, while the contributions of individual mechanisms and sustained ecosystem evolution remain untested.