In Christian Today, Duncan Williams writes of the use of AI for making films based on parts of the Bible. He sees that the technology has the potential to enable Christians to do this so much more cheaply and easily; but, there are dangers. Though AI is, in his view, just a tool which can be used for good or ill, he sees that thoughtless use of it could undermine Scripture. What warnings does he give, then? Firstly, that AI systems lack a deep understanding of the Bible due to their algorithmic nature and so may render a film which looks good on the surface but falls short in regard to nuance or meaning. Secondly, that AI systems are not indwelt by the Holy Spirit so their output will be mechanical rather than spiritual, and thus appealing to human delight in spectacle rather than truly engaging the soul. Nonetheless, if due care is taken, he sees that AI may be the next useful technology to help Christians tell the story of the Bible.
Duncan’s article is just a short piece for a Christian news site and so lacks the space to dive into all the issues one might cover. However, I found it also limited due to his understanding both of the technology and Scripture. Let me pick some examples.
Firstly, he opens by naming Jesus’s parables alongside Bible movies as ways of bringing “Scripture to life” because they both involve storytelling. But the first listeners to Jesus’s parables found some of them unclear, causing his disciples to wonder why he used the technique. Jesus replied by saying that the parables were to divide people and test their hearts, before then explaining privately to his friends what the story was about (Matthew 13:10-23). So though parables may be image-rich storytelling, Jesus doesn’t use them simply to make things clearer for those who like pictures cast in their minds rather than words on a page. Indeed, most of Scripture was actually written to be read out loud to people (e.g. Nehemiah 8:1-8; Revelation 1:3) in the context of trained teachers who can then verbally clarify and expand upon its meaning for the hearers. Duncan doesn’t discuss what altering that process to a visually focussed one might do to how the message is communicated.
Secondly, the article merges various AI tools into a single system. So, at the start, Duncan writes about AI producing film-quality images. But then he moves to discussing how AI would generate the storyline itself, not just the images. Thus, the AI becomes responsible for the whole filmmaking process. But, actually, a filmmaker could use a video generation system to create the visuals for a script that is completely written by humans. Treating AI as a single system, rather than a collection of tools, is confusing. Furthermore, he writes that “AI models generate content based on pre-existing data, which means they are shaped by algorithms rather than by a deep understanding of Scripture.” Well, generative AI – whether visual or verbal – produces its content based on the statistical relationships found in its training data. So in regard to language models, if the training includes vast amounts of a deep understanding of Scripture, then the model will regurgitate that content when prompted, though possibly with some novel elements thanks to the breadth of its training and randomisation. Writing, therefore, that it’s shaped by algorithms as opposed to understanding isn’t really the best way to express what’s happening inside the model.
I’m not saying that Duncan is completely wrong or unhelpful. For me, his article just highlights how we need to work harder at examining the multiple issues around this topic.
Photo by Marius Gire on Unsplash
All posts tagged under technology notebook
Introduction to this series of posts
Cover photo by Denley Photography on Unsplash
Scripture quotations are from the ESV® Bible (The Holy Bible, English Standard Version®), © 2001 by Crossway, a publishing ministry of Good News Publishers. Used by permission. All rights reserved. The ESV text may not be quoted in any publication made available to the public by a Creative Commons license. The ESV may not be translated in whole or in part into any other language.