The Privilege of First Draft: Technical and Ethical Aspects of Using AI to Draft Scripture
The Bible Translator · 2026-08 · 77 (2) · 243-290
DOI: 10.1177/20516770261458250 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Bible Translator, online erschienen am 16.07.2026. Verfasst von Jeff Heath und 2 weiteren Autor:innen. Die Arbeit nennt 62 Literaturangaben.
Abstract (Originalsprache)
Using artificial intelligence (AI) to facilitate the work of Bible translation currently excites a great deal of interest. In this article, we draw on our experience in translation and language technology to examine the technical, practical, and ethical implications of providing translators with machine-generated drafts of Scripture in their language. We also present a case study of AI drafts of 1 Samuel and Psalms in Chadian Arabic produced using Scripture Forge, showing that these AI drafts contain errors of both exegetical accuracy and naturalness. Even if such errors can be reduced, we argue that initial drafts of Scripture should not be produced by an automated, statistical process lacking linguistic and cultural depth. We further identify risks associated with post-editing AI-generated drafts, such as automation bias, cognitive anchoring, and skill decay. In light of these concerns, we contend that the privilege of first draft should remain with in-culture, first-language speakers.
Zitieren
APA
Heath, J., Heath, J., & Kuipers, E. (2026). The Privilege of First Draft: Technical and Ethical Aspects of Using AI to Draft Scripture. The Bible Translator, 77(2), 243-290. https://doi.org/10.1177/20516770261458250
Vancouver
Heath J, Heath J, Kuipers E. The Privilege of First Draft: Technical and Ethical Aspects of Using AI to Draft Scripture. The Bible Translator. 2026;77(2):243-290. doi:10.1177/20516770261458250
BibTeX
@article{Heath2026,
author = {Heath, Jeff and Heath, Judy and Kuipers, Emma},
title = {The Privilege of First Draft: Technical and Ethical Aspects of Using AI to Draft Scripture},
journal = {The Bible Translator},
year = {2026},
volume = {77},
number = {2},
pages = {243--290},
doi = {10.1177/20516770261458250}
}
EUVIMED