Abstract: A critical gap remains in understanding the valuable issue of how to achieve AI innovation and advance it further. Based on the TOE framework, this study employs fsQCA and PSM methods to analyze Chinese listed firms (2018—2023), examining how configurations of technical factors (technological breadth, technological depth), organizational factors (ICT capabilities, AI stock, AI age), and environmental factors (regional AI ecological environment, industry ICT innovation atmosphere) influence AI innovation and subsequently shape firms’ high-quality development. Our findings reveal that AI stock serves as a necessary condition, and identify three high-level innovation paths: dynamic learning, self striving, and collaboration. The incremental effect test results of the configuration show that compared with the net effect of antecedent conditions, the configuration results have a better explanatory strength for AI innovation. The multi-period QCA results indicate that the changes in the three paths respectively present “emergent trajectories”, “buffer-dominant trajectories”, and “dominant trajectories”. High-level AI innovation triggered by multiple configurations has different effects on the high-quality development. The paper expands the research on the antecedents of AI innovation from a configuration perspective, provides a new explanatory for the “modern productivity paradox” of AI, and provides inspiration for firms on how to leverage AI innovation to foster and develop new qualitative productivity.

Key words: TOE framework, artificial intelligence innovation, fuzzy set qualitative comparative analysis (fsQCA), high quality development