A COMPREHENSIVE STUDY OF AGENTIC AI SYSTEMS: EXPLORING THE EVOLUTION FROM PREDICTIVE MACHINE LEARNING MODELS TO AUTONOMOUS, GOAL-DIRECTED, AND DECISION-MAKING ARTIFICIAL AGENTS IN COMPLEX ENVIRONMENTS

This study investigates the transformative evolution of artificial intelligence from predictive machine learning models to agentic AI systems capable of autonomous, goal-directed decision-making in complex environments. Employing a mixed-methods approach, including systematic literature review, simulation-based experiments on realistic datasets, and performance benchmarking, we analyse key architectural shifts, empirical outcomes, and theoretical implications. Main findings reveal that agentic systems enhance task completion rates by up to 40% in dynamic settings compared to traditional models, driven by advancements in reinforcement learning and multi-agent collaboration. However, challenges such as ethical alignment and scalability persist. We conclude that agentic AI represents a paradigm shift toward proactive intelligence, with implications for industries like healthcare and robotics. Future directions emphasize hybrid human-AI frameworks to mitigate risks while maximizing societal benefits. This research bridges gaps in understanding long-term adaptability, offering a reproducible methodology for ongoing evaluation.