At its core, the race in AI for Science is a race for talent. To grasp the full scope of this transformation, we must look beyond hard inputs such as data, computing power, and algorithms, and recognize that a softer dimension, innovation culture, is being reshaped at a systemic level. The emerging paradigm of human, AI cognitive collaboration is fundamentally altering how scientific knowledge is produced and how innovation unfolds. In this shift, defining the right problems matters more, and the need for cross−disciplinary, composite talent has become acute. Open−source practices have turned into the primary mode of research collaboration, while the accelerating pace of technological iteration makes long term commitment and tolerance for failure more valuable than ever. Widely accessible AI tools are empowering the public to engage in innovation, and ethical competence is now an integral part of innovative capability itself. The United States' recent Genesis Mission and its Science: A New Golden Age report both underscore the importance of investing in people. An innovation culture aligned with the AI era will unlock the full potential of AI for Science. Moving forward, cultivating this new culture must serve as a strategic pillar for achieving greater self-reliance and strength in science and technology. This requires focused efforts to reshape talent development systems, build open and collaborative ecosystems, reform evaluation and incentive mechanisms, strengthen technology ethics governance, and systematically enhance AI competency across the whole of society.