Starting Point
I studied Catholic theology at the University of Tübingen and graduated with a diploma: Latin, Hebrew, Greek, historical-critical method. Source analysis, textual criticism, literary criticism — philological work that transfers directly to NLP and text analysis. I am an atheist — and this profile, as far as I know, is not very common.
Research Project: AI-assisted Biblical Exegesis
Historical-critical exegesis breaks biblical texts into their components: source separation, genre classification, redaction history. These methods are considered hermeneutic acts. In reality, many of them are formal operations on text: variant comparison, vocabulary statistics, pattern matching, classification.
Dershowitz et al. (2015) showed that unsupervised clustering separates the Priestly source from the rest of the Pentateuch material with over 90% accuracy, without theological prior knowledge, without a hermeneutic circle. The algorithm reproduces in seconds what exegesis has worked out since Wellhausen over two centuries.
My concept is based on an escalation model: from concordance (level 0) through TF-IDF and clustering to LLM prompting (level 5) and autonomous agents (level 6).
The central question: if an algorithm understands the text better than a human, what does that mean for the concept of revelation? Is understanding scripture bound to human cognition, or is it a property of the text itself?
Why the Field is Growing
Philological labour was the foundation of exegesis for two centuries: building concordances, counting word fields, comparing variants, separating sources. That labour is losing its value. An agent does the same work in hours and misses nothing.
One suspects that this is exactly what scares exegetes in the field. My inquiry about a talk on AI and exegesis at a university was initially met with scepticism — whether "AI" could even do that — but then also with interest; after submitting the concept, no reply came. Christoph Heilig (PD Basel, LMU Munich) has argued publicly in interviews that exegesis in its current institutional form will implode — journals, grant-funded projects, dissertation series. The thesis was widely discussed. An academic response is still pending.
Retreating into specialisation offers no protection. On the contrary: AI can work through the entire secondary literature in any niche, identify gaps in argumentation, and provide counter-evidence. Those who retreat into a speciality are therefore not specialists but vulnerable — because the machine surveys the entire field.
Meanwhile, three academic communities work on partial aspects without reading each other: theologians discuss AI ethics but don't touch algorithms. Computational linguists build tools for biblical texts but don't ask hermeneutic questions. And the digital theology debate stays at "the Holy Spirit is missing" without even formulating the epistemological question.
Exegesis is the starting point. History of religion, church history, liturgical studies, pastoral practice — the same questions arise everywhere.