AI for scientific discovery is a social problem.

AI is being increasingly applied to scientific research, but its benefits remain unevenly distributed across different communities and disciplines. While technical challenges such as limited data, fragmented standards, and unequal access to computational resources are already well known, social and institutional factors are often the primary constraints. Narratives emphasizing autonomous "AI scientists," the underrecognition of data and infrastructure work, misaligned incentives, and gaps between domain experts and machine-learning researchers all limit the impact of AI on scientific discovery. Four interconnected challenges are highlighted in this paper: community coordination, the misalignment of research priorities with upstream needs, data fragmentation, and infrastructure inequities. We argue that addressing these challenges requires not only technical innovations but also intentional community-building efforts, cross-disciplinary education, shared benchmarks, and accessible infrastructure. We call for reframing AI for science as a collective social project, where sustainable collaboration and equitable participation are treated as prerequisites for achieving technical progress.