“The discerning heart seeks knowledge”: bringing human intelligence into AI in the study of the Babylonian Talmud

Over the past few decades, the use of stochastic approaches for automatic text analysis has grown significantly and steadily. Today, so-called Large Language Models are the benchmark in the field of Artificial Intelligence (AI) for tasks involving text processing. However, there are some texts, such as religious texts, which, due to their intrinsic complexity (in terms of language, domain-specific knowledge, and the skills required for their correct interpretation, etc.), are not well-suited to be treated with purely statistical methods. In this paper, the authors present a method designed to study the text of the Babylonian Talmud through the technique of question answering whereby a scholar can interrogate the text by asking questions in natural language. The approach is based on the use of a Large Language Model and the application of the Retrieval Augmented Generation technique.