Agentic AI Redefined: A New Paradigm in Artificial Intelligence
Yogesh Awasthi · Journal of Software Engineering and Simulation · 2025
The evolution of Artificial Intelligence (AI) from rule-based systems to deep learning has enabled significant technological advancements, but it has also raised complex questions about autonomy and agency. Agentic AI refers to AI systems capable of initiating goal-directed actions, making context-sensitive decisions, and adapting over time with minimal human oversight. This paper explores the conceptual boundaries of agentic AI and provides empirical analysis based on case studies from autonomous vehicles, intelligent tutoring systems, and AI-enabled robotics. By evaluating behavioural data and decision-making patterns, we demonstrate how these systems exhibit agentic properties that represent a paradigm shift in AI. The paper concludes with implications for AI design, ethics, and governance.