The rapid advancement of Artificial Intelligence (AI) is profoundly changing the operational logic and organizational models of scientific research, fundamentally reshaping the theoretical foundation and methodological system of the Science of Science (SciSci).Based on the rise of AI-driven scientific research, this paper systematically analyzes the holistic transformation of SciSci research across its objects, core problems, and methodological tools under the empowerment of AI.The study indicates that AI acts simultaneously as a revolutionary force within the scientific system, generating new SciSci research objects and questions, and as a powerful analytical tool, breaking through the technical bottlenecks of traditional SciSci methodologies. This series of transformations will enable SciSci to achieve a fundamental expansion: from external representation to intrinsic mechanisms, from phenomenological description to mechanistic explanation, and from passive observation to active design. The new research paradigm exhibits core characteristics such as prospectiveness, systematicness, synergy, and designability.At the methodological level, AI promotes the objectification and multi-modality of SciSci's observation and measurement, the predictive and networked nature of its analysis and modeling, and the systemic and explainable nature of its explanation and prediction. This significantly enhances SciSci's practical capabilities in research evaluation, knowledge discovery, and scientific ecosystem design.Nevertheless, risks and challenges in technical, ethical, and institutional aspects urgently need to be addressed. This paper proposes that, while maintaining the unique identity of the SciSci discipline, we should drive paradigm evolution through methodological innovation, and construct an open, transparent, explainable, and prospectively guiding methodological system to achieve the continuous leap forward of SciSci empowered by AI.