Agentic AI-Enabled Closed-Loop Drug Discovery With Hypothesis Generation Testing and Optimization

Artificial intelligence in drug discovery is transitioning from decision-support tools to agentic, closed-loop systems capable of autonomous hypothesis generation, experimentation, and optimisation. This chapter examines the emergence of agentic AI as a transformative yet contested paradigm reshaping scientific workflows in pharmacology and toxicology. Integrating insights from cognitive science, AI theory, philosophy of science, systems biology, and regulatory studies, the chapter conceptualises closed-loop drug discovery as a cybernetic, feedback-driven process rather than a linear pipeline. It analyses architectural foundations, empirical evidence, industry practices, and governance challenges, and proposes the Autonomous Scientific Discovery Stack (ASDS) as a novel integrative framework that links epistemic reasoning, experimental execution, and ethical–regulatory oversight. Through critical engagement with debates on autonomy, explainability, creativity, and accountability, the chapter highlights unresolved tensions, global inequities, and data limitations.