Intelligenza artificiale generativa e fonti storico-educative: prospettive metodologiche

Abstract

The paper examines the methodological potential of generative artificial intelligence in historical-educational research, addressing a subject that has so far remained unexplored. After outlining the functioning of Large Language Models (LLM) and Retrieval Augmented Generation (RAG) systems, it analyses their most promising applications, including the automatic transcription of manuscripts, the translation of texts in ancient languages, and the processing of extensive documentary corpora. Through case studies, the research demonstrates how RAG architectures can effectively overcome the limitations of LLMs in analysing large collections of historical sources. Finally, a structured methodological framework is proposed to integrate these technologies into historical-educational research, establishing an operational protocol for documentary analysis.