This session interrogates a shared problematic across three papers: how theological knowledge is mediated, interpreted, and reshaped by the structures through which it passes. Moving between the controversies of Late Antiquity and the contemporary deployment of Artificial Intelligence is the humanities, the papers collectively examine what is preserved and what is changed in the transmission of meanings across time, geography, and medium in Christianity. The session draws on reception history, empirical evaluation methodology, and patristic theology to address questions of enduring relevance: the knowledge, tools, and medium that scholars in the History of Christianity use. The session will be of interest to scholars working in patristics, Late Antique Christianity, reception history, gender studies, pedagogy, and the emerging field of AI and religious studies.
This paper examines the intervention of the Scythian monks in the early sixth century as part of the ongoing reception of the Council of Chalcedon (451). In the decades following the council, the meaning of its Christological definition remained contested across the eastern Mediterranean. The Scythian monks, associated with Abbot Maxentius, entered these debates through their defense of the theopaschite formula, “one of the Trinity suffered in the flesh.” They argued that this language did not introduce a doctrinal innovation but clarified the implications of Chalcedonian Christology and protected the unity of Christ’s person. By examining the monks’ writings and appeals to ecclesiastical authorities, this paper situates their intervention within the broader process through which Chalcedon’s meaning continued to be interpreted and debated in the sixth century. The episode highlights how actors from frontier monastic communities participated in shaping the reception of conciliar doctrine in Late Antiquity.
Large Language Models (LLMs) are increasingly used for historical and theological inquiry, yet their reliability in specialized scholarly domains remains unexamined. This paper presents a systematic empirical evaluation of LLM accuracy in early Christian studies, using two fourth-century figures as case studies: Macrina the Younger (c. 327-379 CE) and Olympias of Constantinople (c. 368-408 CE). These figures were selected to probe LLM behavior across axes of scholarly versus popular reception, source type, and gender representation. Using a structured benchmarking methodology - testing biographical accuracy, chronological precision, theological positioning, and source-critical reasoning across multiple models - we aim to identify consistent failure patterns, including factual conflation, hallucination, and what we term association collapse: the systematic narration of women's significance through male contemporaries. We conclude with practical guidance for educators on integrating critical AI literacy into religious studies pedagogy and a replicable framework for evaluating LLMs in other historical and theological contexts.
In addition to the many and varied critiques of AI, history, and theological writing, one additional concern will be considered in this paper: AI is quite bad at theology and in particular historical theology. Driven by its own and secret means of building symbolic structures and exacerbated by a base training of freely available sources, the result is an idiosyncratic system with little regard for the meaning, relationships, and symbols of Christian history. In a fascinating way it both mimics the Cappadocian critiques of Eunomianism and, at times, even replicates its precise failings.