Abstract

Artificial intelligence (AI) is reshaping scientific discovery, evolving fromspecialized computational tools into autonomous research partners. We positionAgentic Science as a pivotal stage within the broader AI for Science paradigm,where AI systems progress from partial assistance to full scientific agency.Enabled by large language models (LLMs), multimodal systems, and integratedresearch platforms, agentic AI shows capabilities in hypothesis generation,experimental design, execution, analysis, and iterative refinement -- behaviorsonce regarded as uniquely human. This survey provides a domain-oriented reviewof autonomous scientific discovery across life sciences, chemistry, materialsscience, and physics. We unify three previously fragmented perspectives --process-oriented, autonomy-oriented, and mechanism-oriented -- through acomprehensive framework that connects foundational capabilities, coreprocesses, and domain-specific realizations. Building on this framework, we (i)trace the evolution of AI for Science, (ii) identify five core capabilitiesunderpinning scientific agency, (iii) model discovery as a dynamic four-stageworkflow, (iv) review applications across the above domains, and (v) synthesizekey challenges and future opportunities. This work establishes adomain-oriented synthesis of autonomous scientific discovery and positionsAgentic Science as a structured paradigm for advancing AI-driven research.