SIEBES: A Case Study in Hallucinatory Hagiography

Large Language Models (LLMs) have rapidly permeated both personal and professional domains, extending their use beyond everyday problem-solving to scientific research. In this paper, we explore a less common but culturally significant application: assistance in the writing of a contribution to a liber amicorum --- an edited volume celebrating a distinguished scholar. To support this delicate task, we introduce a semi-automated System for Interpretable and Explainable Biographical Extrapolation and Synthesis (SIEBES), and report our experiences with its use in a case study. Particular attention is paid to the phenomenon of hallucinations, providing a formal definition and empirical insights. Our study highlights both the promises and perils of AI-assisted biographical writing in the context of an esteemed academic tradition. (Disclaimer: Parts of this abstract have been written by a large language model.)