by Jim Killam, Wycliffe Global Alliance
Imagine a Scripture engagement tool that could quickly answer any Bible question accurately, in many languages, and point to resources.
A GPT (generative pretrained transformer) works with giant databases of information to generate words, images, code and more in response to prompts entered by the user. General-knowledge GPTs can already answer questions about the Bible based on what they’ve been fed from the Internet. Bible-specific GPTs are under development right now, too (here is a prominent example).
All of which raises some red flags. What if the user asks a bad question? (What does the book of Leviticus say about AI?) Or what if it’s a good question but the GPT has been fed data that includes serious theological errors?
Adam Graber, digital theology consultant and cohost of the Device & Virtue podcast, wrestled with these questions during the recent Missional AI Summit. First, he paraphrased N.T. Wright in stating the purpose of Bible engagement: To connect with God and have his Word permeate our entire being. To be inwardly equipped for the church’s outward mission in the world.
Designed and used well, he said, Bible GPTs could play a significant role in that process—but not if used out of order.
“God speaks first. Not the Internet and not us,” Graber said. “Our work is to feed on God’s Word, and to use that food for energy and for mission.”
As an example, he said, a Bible GPT user could ask, “With John 5, what questions or issues does this passage ask me to wrestle with?”
“And it keeps me in this position of Bible engagement, not Bible replacement,” Graber said. “It keeps the GPT in the role of supplement, and not substitute.”
Designers will need to design for that sort of practice, he said.
“I think we need to align our systems to the purposes of Bible engagement. Bible GPTs won’t rise to the intentions of designers. They will fall to the habits of users. As designers we need to ask, does the system default to Bible engagement or GPT engagement? Does the system encourage connections and reflections with God himself? Does the system make it easier for users to love God and to feed on his Word? How do we stay aligned with that purpose of connecting with God, being transformed by that interaction and being equipped for mission?”
That’s about much more than simply the AI super-version of Googling for answers to Bible questions—a practice Graber believes can inhibit spiritual growth. Bible knowledge has value, but knowledge alone does not necessarily engage people with God.
“It’s failing to allow us to digest Scripture, to connect with God, to be transformed by that interaction and to be equipped for the mission,” he said.
What’s in? What’s out?
One key question facing designers of Bible GPTs is, which Bible-resource writings go into a database? Which are left out, and why? How will denominational differences be accounted for? How to make sure that diverse Christian voices from different cultures are included?
“Whoever has the most content in the database is going to have the most visibility,” Graber said. “The prolific Methodists are going to have more visibility than the quiet Mennonites. The loudest traditions actually pull the average in their direction. And it becomes a popularity contest, if it wasn’t already.”
Graber’s message was less about specific solutions and more about encouraging AI developers to think proactively about them—or the church worldwide could reap the consequences.
“But if we’re clear and conscientious about how we design Bible GPTs, I think the potential opportunities can be incredible,” he said. “And I believe that the best-designed Bible GPTs will be those that encourage and enable Christians to connect with God to transform them and equip them for mission.”
Reproduced with permission from Wycliffe Global Alliance