This review introduces Generative Metascience, a comprehensive framework for understanding how artificial intelligence (AI) transforms scientific discovery. We synthesize historical milestones and case studies across genomics, astronomy, materials science, and social sciences to illustrate AI's evolution from a research instrument into an autonomous co-investigator. Our analysis is structured around two core themes: AI-enabled data collection and analysis, and AI-driven hypothesis generation and testing. This dual focus highlights the iterative interplay between data-driven analytics and hypothesis-driven inquiry, showing how AI tools can simultaneously generate and evaluate scientific hypotheses. Key insights reveal AI-accelerated breakthroughs, such as automated protein-folding and materials design, and the rise of self-driving laboratories, which signal a shift from traditional inquiry toward an algorithmic discovery paradigm. By synthesizing data-driven pattern recognition with AI-facilitated hypothesis generation across disciplines, this review addresses a critical research gap, showing that AI has begun to automate hypothesis formation and serves as a meta-technology that is redefining scientific epistemology. We highlight urgent implications for future research, including the development of hybrid AI-human workflows and robust metrics for machine-generated insights. For policy, we articulate the need for greater transparency, open data standards, and interdisciplinary funding initiatives. For practice, we advocate for retraining researchers and updating curricula for AI-integrated labs. By articulating these developments and their contributions, this paper charts a roadmap for responsibly harnessing AI’s potential and guiding the scientific community as it navigates AI’s evolving role in a new era of discovery.
Shawn Ray (Mon,) studied this question.