Beth Bryson:  AI and Translation Issues
(December 2025)

Beth Bryson (Dictionary/Lexicography Services and FLExTrans team) gave an overview of AI issues for the October 2025 Chiangmai, Thailand meetings of the Area Translation Coordinators and the International Translation Services staff. 

The Scripture Forge AI Drafting team reports that there are roughly 300 projects currently using AI Drafting.  In the past, of teams that tested it out, roughly 50% found it useful enough to continue.  Recently that number has risen to around 60%.

The Scripture Forge team has developed a set of “readiness criteria” to ensure that teams are more likely to have a positive experience.  Some of the criteria include:

• Readiness of the project: Translation brief in place, clarity about which dialect is being targeted, stable orthography

• Readiness of the translators: already trained in translation skills and with some experience, preferably having manually translated at least one book

• Readiness of the stakeholders (including the church, branch leadership, consultant, funders): Are all of these stakeholders on board with the idea of AI Drafting?  Have they been trained about what the technology can and can’t do, and how to work with it responsibly?

At the BT Conference, the paper “Ready or Not, Here AI Comes! Practical Considerations for Ethical AI Drafting in Bible Translation (Taeho Jang, et al)” explored these ideas. Email Taeho Jang for the full written paper. 

In discussions about AI Drafting on the Linguistics Discussion List, there was agreement that training for working with AI drafts would be similar to the training needed for a traditional approach.  It is still essential to develop awareness of target language features that are not used in the source language, especially features that force a choice.  This requires a good and conscious understanding of the morphosyntax of the target language, as well as the necessary exegetical skills to determine the contextual meanings of the source that underpin the needed choices.

In a recent presentation through EMDC online (https://emdc.academy/list/696), Eshinee Veith explored the kinds of mistakes AI Drafting tends to make, and thus we need to train those who are reviewing these drafts to watch out for these kinds of issues.

AI drafting was explored in a number of other papers at the BT Conference last week. 

For example, the research project from Prescience labs, A Study of Equivalence in Methods of Using AI-Generated Drafts in Bible Translation  found that using an AI draft as a reference text led to quicker drafting, better quality from a consultant's perspective and equal or better internalisation by the team than manual drafting.  

Meanwhile. Martha Wade's AI Drafting of Scripture: A Detailed Analysis of the Results in a Language in Papua New Guinea gives a fantastic breakdown of some of the specific issues from AI drafting with Scripture Forge, but also notes the positive life impact on the translator. 

In Steering from Harm: Applying an Ethical Research Lens to AI and Bible Translation, Eshinee Veith gives a clear and robust framework for thinking through the ethical issues around using AI in Bible Translation projects.