Giovanni Pilato’s research lies in the area of Cognitive Artificial Intelligence, with a particular focus on the design of language-driven intelligent systems capable of integrating neural models, structured knowledge, and reasoning mechanisms.A central research line concerns the study of hybrid neural-symbolic architectures that combine Large Language Models, Information Retrieval techniques, knowledge representation, and computational models inspired by human cognition. The objective is to address some of the limitations of current AI systems, which show strong capabilities in generation and pattern recognition but still face significant challenges in structured reasoning, planning, explicit knowledge control, reliability, and explainability.
From this perspective, natural language and retrieval mechanisms are considered not only as tools for accessing information, but also as computational primitives for building cognitive systems able to support reasoning, planning, and decision-making processes. Particular attention is devoted to the integration of these components within coherent frameworks inspired by models such as dual-process theories, BDI architectures, and global workspace models.
Application areas of interest include decision support systems, intelligent interfaces, cognitive systems for data-intensive environments, and explainable and controllable AI solutions, with the broader goal of contributing to the development of intelligent systems that are more robust, interpretable, and aligned with users’ needs.
