Agentic AI and Autonomous Agents: Conceptual Continuity and Epistemological Reinterpretation

The rise of Large Language Models (LLMs) has fueled the popularization of the terms “Agentic AI” and “Multiagentic AI.” However, these concepts do not represent a new ontology of agents but a technological reconfiguration of ideas long established in the theory of autonomous agents and multi-agent systems (MAS). These new terms often mask concepts already consolidated in the classical literature on intelligent agents and MAS. This paper argues for the conceptual invariance of the autonomous agent paradigm in the face of technological evolution and examines the conceptual continuity between classical autonomous agents and new architectures based onLLMs, integrating epistemological and neuro-symbolic contributions. It contends that Agentic AI does not redefine agency but reinterprets it through a cognitive-linguistic lens, maintaining the functional structure of perception, decision, and action of the agent abstraction defined within the field of Autonomous Agents and Multi-Agent Systems. Finally, it discusses the risks of terminological inflation and the opportunities for a neuro-symbolic synthesis for the future of artificial intelligence.