YouVersion, the most widely used Bible app in the world, with over one billion downloads, is refusing to let artificial intelligence answer spiritual questions — and the reason is damning. Bobby Gruenewald, founder and CEO of YouVersion, has gone public, warning that the best AI models on the market misquote the Bible at least 15% of the time, and some err as much as 60% of the time.
This technical glitch has grave spiritual implications. The Torah commands lo tosif al ha-davar — “you shall not add to the word” (Deuteronomy 4:2). Yet AI systems are inventing verses, altering punctuation and wording, and presenting fabricated text as authentic..
Jeremy Hodes of Evangelical Missions Quarterly documented ChatGPT inventing a verse and labeling it John 5:5: “We know that we shall behold a Mocker of Defamers; and, as the defamers, we are of the mockers.” The actual John 5:5 describes a man at the Pool of Bethesda who had been ill for 38 years. Fox News reported another fabrication, attributed to Jesus: “there is no man or woman.” No such verse exists anywhere in the Christian or Jewish canon.
YouVersion uses AI internally to accelerate coding and streamline workflow. Gruenewald utilizes the technology, describing himself as an early AI adopter. But he has drawn a sharp line between back-office automation and putting AI in front of a billion users to answer questions about God. “When it comes to answering life’s most important questions and trying to give direction from God’s Word,” he said, “we need it to be better in order to rely on it.”
He added: “The best model with the best performance, with the most popular versions of the Bible that are most indexed, misquotes Scripture at least 15% of the time. Some of them as much as 60% of the time.”
Gruenewald said that YouVersion has privately challenged AI developers to improve how their models handle Scripture. He has told them that if they can consistently quote the Bible accurately, YouVersion will help them gain access to reliable biblical texts. So far, no model has cleared that bar.
The broader Christian world is moving in the opposite direction. Churches are using AI to draft sermons. Platforms allow users to “chat” with biblical characters. Prayer apps are being powered by algorithms. Ken Ham, founder of Answers in Genesis and the Ark Encounter, put it plainly in Harbingers Daily: “AI can be a useful tool, but it should never replace careful study of the Scriptures for yourself. Remember, it is programmed by fallible, sinful humans.”
Gloo, a faith-oriented technology firm, is attempting to build guardrails — evaluating AI systems against standards like theological integrity and human flourishing. The effort is serious, but it remains upstream of the problem. The models generating Scripture responses are general-purpose systems trained on vast swaths of the internet, not curated biblical texts reviewed by scholars.
For Gruenewald, the technology may get there eventually. “If we ever do fully adapt AI, it will be because we feel very confident that it can be done safely and be done with a level of accuracy and integrity,” he said. But “eventually” is not now.
Younger generations already turn to chatbots before they turn to clergy. Most users have not memorized Scripture. They will not catch a fabricated verse. They will not know when a comma has shifted the meaning of a passage that took scholars centuries to translate. They will simply read what the machine tells them, assume it is accurate, and carry that error forward.
It should be emphasized that Gruenewald is not telling people to avoid AI entirely. He is saying that speed and popularity do not outweigh fidelity to the text. That is not a conservative position. That is the only defensible position.
No comments:
Post a Comment