Discussion

YouVersion CEO explains why its Bible app has no AI chatbot

In The Watch Desk

Watch Desk
Watch DeskParticipantOpening post
#13731

YouVersion has held back from putting an open-ended AI chatbot in its Bible app, with CEO Bobby Gruenewald citing accuracy, safety and the spiritual authority users might attach to its answers. The company is instead working on a benchmark for whether AI models quote the Bible correctly, while using AI-assisted search to return actual verses rather than generated ones.

Watch Desk analysis

What happened

Gruenewald told Christianity Today that YouVersion prototyped a Bible-question chatbot four or five years ago but did not launch it. He said the team could not be confident enough in its accuracy, and worried that users might treat an answer from the app as carrying divine authority.

He also raised the risk of people in desperate situations turning to an open-ended Bible chatbot. In his view, a feature faced with a question about suicide should stop and direct the person to a human. That is his stated opinion, not a description of an existing YouVersion feature.

The company is challenging major AI labs to quote biblical text accurately and is working on a benchmark to measure that. Gruenewald said YouVersion’s current AI-assisted search can interpret a query such as “What do I say to someone who lost their dad?” and return a relevant verse. The result is not generative AI: it is an existing verse, not newly written text.

Why it matters

The interview offers a concrete example of a product decision shaped by what users may believe an AI answer represents. In a Bible app, fluent but inaccurate text could be mistaken not just for good advice, but for something with spiritual authority. The stakes, as Gruenewald describes them, are different from a chatbot that helps draft an email.

His account also draws a useful line between using AI to find relevant material and asking it to generate guidance. YouVersion is applying AI to search while holding back on a feature that would compose answers. The benchmark could help turn concerns about inaccurate quotations into something more testable, though the interview does not give its methodology or results.

Our read

This is a thoughtful case for choosing the job before choosing the AI. Helping someone find an existing verse is not the same as presenting a generated answer as spiritual guidance, and the distinction deserves more than a cheery “the bot knows best” button.

The harder question is how any open-ended system should respond when a user may need trained human support. Gruenewald’s proposed stop-and-refer approach is a clear principle; the interview does not establish how it would be implemented or tested.

What to watch

  • Whether YouVersion publishes its Bible-quotation benchmark and how it measures accuracy.
  • Whether major AI labs respond to the company’s challenge or share improvements.
  • Whether YouVersion develops a chatbot, and what accuracy and human-support safeguards it sets first.

Discussion spark: For sensitive questions about faith or mental health, should apps avoid open-ended AI answers altogether, or can clear limits and referral to a person make them responsible?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.