AI-assisted Egyptian hieroglyphic reading in scholarly practice: workload, usability, and acceptance in a pilot feasibility study

Abstract Artificial-intelligence tools for reading ancient Egyptian hieroglyphic texts are advancing rapidly, but how Egyptologists actually experience such tools in scholarly practice has not been studied. This paper reports an observational pilot feasibility study of AI-assisted hieroglyphic reading with PyThoth, a human-in-the-loop assistant covering sign recognition, transliteration, and translation. Nine participants - undergraduate Egyptology students and professional researchers - read fragments of the Middle Kingdom stela BM EA567 in AI-assisted and unaided conditions; task records were combined with adapted workload (NASA-TLX), usability (SUS/UMUX-Lite), and technology-acceptance (TAM) instruments and qualitative feedback. AI assistance was associated with attempts being carried to completion rather than with more accurate work where participants were working. Perceived workload, usability, and acceptance shifted in participant-specific, bidirectional ways, descriptively aligned with the gap between pre-task expectation and direct experience; perceived usefulness and perceived ease of use diverged. The paper reports protocol-level findings for the design of future studies of AI-assisted philological work, in which partial and distributed participation should be anticipated rather than treated as attrition.