What if the biggest barrier to Bible translation isn't time or money, but data? Thousands of languages are missing from today's AI, not because no one speaks them, but because no one is capturing what's said.
In this episode, two pioneers share how they're tackling that challenge:
This episode has a slightly different format: you'll see the questions on screen, then hear each guest's answer.
AI is already speeding up translation drafting and research; the next challenge is scaling those gains beyond a few specialists.
Daniel sees AI cutting the "time to insight" across knowledge work, from a question to an actionable answer in moments. In Bible translation, drafting is accelerating as methods improve and systems become cheaper and easier to use. He expects the same acceleration in Scripture engagement and Christian media, in high-resource and low-resource languages alike.
Jacob expects the next two to three years to move us past early adoption and into the challenge of scale. Right now, highly specialized people are leading the way. The work ahead is bridging the training gap so pastors, missionaries and translation teams can use these tools productively every day, and turning discoveries into concrete tools that non-specialists can use.
The first question many translators ask about AI is personal: will it replace me? Jacob helps them see that their expertise is what makes the tool useful.
Many people Jacob works with have heard of AI but don't really know what it is. Once they understand it, a common first reaction is fear about whether it will make them less useful in ministry or work.
So SIL's approach starts with identity. Translators are skilled craftsmen, and AI is a new tool. The tool doesn't replace the craftsman; everything the craftsman has learned helps them make the best use of it. Once people find their identity in the world of AI, Jacob says, they become enthusiastic about experimenting and feel free to bring it into their work. He finds that "who am I in the world of AI?" question almost universal, in ministry and likely in many other fields.
No language truly lacks data; it simply hasn't been captured. So XRI focuses on getting the most value from every sentence it collects.
"It's almost a scandal that we think of any language as low resource," Daniel says. We pour out language data every day; it's just uncaptured and unstructured.
His team ran the numbers for Swahili, often labeled low-resource. Kenya has about 300,000 teachers who teach around six hours a day. If every one of them recorded what they said in Kiswahili, how long would it take to reach a trillion tokens, the scale some open-source language models are trained on? About two months.
Of course, privacy and other concerns mean no one would actually do that. But it shows that the data exists; it just isn't online for models to learn from.
With limited time and resources for collecting data from native speakers, XRI's guiding question is how to maximize "intelligence per sentence." What is the ideal dataset? Through extensive experimentation, the team is getting close to knowing exactly which sentences, and how many, will predictably produce a strong model.
The pipeline is already working. XRI collects the data and trains a model, then passes the data to partners at SIL, who train their own models and generate drafts that translation teams are using. Translation is accelerating as a result.
Be clear about your mission, then build the habit of reaching for AI in everyday knowledge work.
Jacob's advice is to start with focus. What are you trying to accomplish in your ministry? From there, network to find out which tools others are using well. Many builders are looking for early adopters to give feedback, and clarity about your goal will help you cut through the noise about what AI should or could be used for.
Daniel's advice is to build a new muscle memory. Use tools like Claude, ChatGPT and Meta's models until asking AI becomes your default, instead of scrolling through search results and sponsored links. XRI encourages its whole team, from operations and HR to finance and engineering, to work "AI first."
He shares a Saturday morning example. Before cleaning his garage, he asked OpenAI's then-new Operator agent to find organizations deploying humanitarian chatbots and log their names, websites, countries and languages in a Google Sheet. It worked for 45 minutes while he cleaned. Meanwhile, he set Deep Research on a question about AI chip import and export restrictions. By the time he finished, both tasks, which would have taken multiple people many hours, were done. The more you practice, he says, the more you'll see acceleration everywhere.
Go deeper with these related talks in the Missional AI Video Library:
Want to learn more or work with our guests?
Listen to the full episode on the Missional AI Podcast, and subscribe so you don't miss future conversations.
Mark your calendar for the Missional AI 2026 Global Summit, April 7–9, 2026, in Silicon Valley, the largest gathering of Christians working at the intersection of faith and AI. Sign up at missional.ai to be first to know when tickets are available.
This episode was made possible in partnership with Gloo, the technology platform connecting the faith ecosystem.