Muñoz-Avila’s current research focuses on agentic AI, systems that can plan, reason, and act autonomously, including the problems of AI grounding and AI alignment. AI grounding addresses how systems connect abstract concepts to the outputs they generate, while AI alignment ensures those outputs remain consistent with societal and safety norms. These challenges have become increasingly important with the rise of generative AI, which is powerful but often unreliable. Muñoz-Avila’s work explores how to combine semantically robust paradigms, such as hierarchical representations, with large language and reasoning models to build more reliable and trustworthy AI systems.

Earlier in his career, Muñoz-Avila’s research focused on integrated AI, combining statistical learning, symbolic planning, and reasoning methods to achieve higher-level cognitive capabilities. This work laid the foundation for his current research on agentic AI by emphasizing the importance of systems that can reason, learn, and adapt in complex environments.

A common theme throughout his work is the integration of multiple AI paradigms, rather than focusing on isolated techniques. He has also collaborated extensively with researchers both within AI and across disciplines, including digital games, civil engineering, electrical engineering, cognitive psychology, education, and manufacturing.