Artificial intelligence, particularly large language models (LLMs), holds significant potential for advancing biblical studies. As these tools become more prevalent, scholars must address two questions: What are their most beneficial applications, and what best practices ensure reliable and meaningful results? This paper explores one specific application—using LLMs to detect intertextual connections in biblical texts—and evaluates their effectiveness in identifying quotations, allusions, and echoes in both English and Koine Greek.
This study employs Claude Opus to analyze a series of confirmed, potential, and spurious intertextual test cases. The results indicate that the model can generate insightful intertextual observations by leveraging lexical, semantic, and contextual data. In some instances, the LLM even identifies novel connections that could provide fresh avenues for scholarly inquiry. However, its susceptibility to false positives and occasional misinterpretations highlights the necessity of an expert-in-the-loop methodology. Rather than replacing traditional exegetical methods, LLMs can serve as powerful tools for augmenting intertextual research when guided by informed human oversight.
This paper systematically documents both the successes and limitations of LLM-generated intertextual analysis, outlining specific query strategies and conditions that yield the most reliable results. By integrating AI-assisted research with established hermeneutical practices, this study proposes a scalable framework for exploring the complex web of intertextuality within and beyond the biblical corpus.