Using Artificial Intelligence
Lee Gatiss considers the uses and abuses of artificial intelligence in ministry. 
The Prime Minister said recently that “Artificial Intelligence will drive incredible change in our country” and that he wants to “turbocharge” the industry and make Britain a world leader in this area. Growth in this arena has really taken off and caught the public imagination in the last few years, and naturally that has led many Christians to be curious and ask questions about it.
So I decided to dig into AI, and did a bit of training with IBM courses on AI and Generative AI and an AI Masterclass with Alison training. It was an enlightening experience and opened my eyes to this whole new world. I think it has a lot of potential applications for Christians and churches and ministers, but also some potential pitfalls that we need to look out for.
You’re already using it
Many of us already use artificial intelligence without perhaps being aware that that’s what we’re doing. Spell checkers on our word processing programs for example, or autocorrect on your phone which uses AI to predict your next word (not always successfully). Online booksellers and social media platforms use it to serve up content they think you might like, and your email program may well be using discriminative filters to weed out content they think you won’t (so-called SPAM). Many will have had a frustrating experience with low-grade AI in the form of automated Chatbots, which some companies use to respond to queries and complaints on their websites. They never seem very intelligent to me, but they are certainly artificial. There has been a more recent revolution in what’s called generative AI, that can actually generate new content not just guide you through a decision tree to the appropriate complaints department. Though generative AI has roots going back to the 1950s, greater computing power and the vast amounts of data on the internet have led to huge leaps forward. It will change how we work and is fast becoming a huge billion, even trillion dollar industry.
Church Society has used AI. The programs we’ve used for podcasts and videos can automatically clean up background noise from our content, and maybe even add captions. It can save time and improve the audience’s experience. I’ve also used it in some fruitful ways, such as uploading the text of the 39 Articles and the Homilies and getting an AI app to convert it into speech and read it out to me on car journeys (with a choice of different accents). AI helped me chase down a Latin quotation that John Owen misattributed (yes, I’m a nerd and I enjoyed it). And it suggested some interactive Bible study questions I could use on a passage of scripture. It can be employed in unfruitful ways too, such as transforming my profile picture so I look like I’m from the seventeenth or twenty-fifth centuries. There’s endless hours of fun in that sort of thing. AI is also fully integrated into the latest Apple iPhone and into Microsoft (Co-Pilot). It is among us.
It’s dangerous!
It can also be dangerous. That entertaining image generation technology can be turned to more illicit purposes, and political disinformation. Technology that can suggest alliterative or rhyming headings for my sermon, and ways to make it shorter or more interesting, can also be used to cheat in university. But isn’t that always the way with human inventions? We are “fearfully and wonderfully made” beings, created in the image of God with a huge capacity for creativity and potential for good. But we are also fallen and flawed, too often captive to “the devices and desires of our own hearts”. We are more than capable of perverting every aspect of God’s world so it serves our disordered and sinful ends.
In the nineteenth century, a group of English textile workers became worried that the automated machinery introduced by the industrial revolution was threatening their jobs. They were concerned about their pay, their conditions, and the quality of the work being produced. They organised raids to destroy the machines. They were called the Luddites. Should we become Luddites with respect to artificial intelligence today? In 2023, Goldman Sachs published a report showing that AI could replace the equivalent of 300 million full-time jobs. Should we be worried that the machines are taking over, and refuse to go along with this development, not least because AI has the potential to make us all stupid? Should we yearn for the old ways? Should we long to get back to pen and paper, papyrus Bible fragments, and candlelight….?
Why did I waste time on education?
This is a real question, and an urgent one because the temptations of AI are very real. The Bible study platform Logos have recently introduced an AI “sermon assistant” feature to help write your next barnstormer. ChatGPT, one of the the most popular AI platforms, can analyse and summarise text and videos, and generate mountains of words in seconds, from a simple prompt such as “Give me a 10 minute evangelical homily on Proverbs 31, with an illustration, a joke, and an application to every age group in my church”. So perhaps I wasted my time with all that theological and homiletical training? If ChatGPT can summarise any passage in 7 words (with 10 options to choose from) in a microsecond, and also compose a nice Anglican collect based on the passage, why did I bother?
The thing is, AI isn’t really all that intelligent. AI that is based on Large Language Models (LLM), such as ChatGPT, simulate and mimic human intelligence, drawing on huge amounts of data that they have been trained on. It learns patterns and linguistic structures from that data and tries to produce something new based on those observed frameworks. It doesn’t really know what it is doing, and it can be skewed by bias in the training data. So it often gives wildly incorrect answers. Technically, it’s called hallucinating (but you might call it lying!). For example, when I asked it to tell me 10 books that Lee Gatiss had written, it came up with an impressive list — but I hadn’t written any of them! They were the kind of things I might have written, but I hadn’t actually done so. An Amazon search would have done better. LLMs are not search engines. And they are not designed to give you the truth; just plausible fabrications that sound a bit like it. Does sound a little devilish, doesn't it? We know who the father of lies is.
Be the human in the room
But maybe we shouldn’t dismiss it too quickly. Let’s take a lesson from history. In 1547, Archbishop Cranmer published the First Book of Homilies. This was a set of sample sermons on key topics, designed to be read out in churches where the minister wasn’t able to preach their own sermons. It was supposed to be a temporary stopgap measure, while things were put in place to train up preachers in Reformation faith and practice. The number of ministers trained and able to preach their own sermons increased tremendously over the course of the sixteenth century. But some people were wedded to the set Homilies, and thought that those should be used more than freshly-composed sermons designed by an in-person pastor to address a specific congregation at a specific time. Archbishop Grindal (1519-1583) resigned over this issue when Queen Elizabeth I insisted there only needed to be maybe two preachers per county and everyone else should just use the set Homilies.
Grindal was right. And he has a lesson to teach us in the twenty-first century. A well-trained and pastorally-minded minister is far better at preaching than someone who only uses a set homily written by someone else. A human on the ground is better than one in a book. Maybe there’s a spectrum from using a set Homily, to plagiarising someone else’s sermon, to using ChatGPT. Sure, it’s better to have half a loaf of bread than none at all (as Grindal said). But appetites grow and people should and will want more than that as they mature. A ChatGPT sermon could contain edifying nuggets. But it can’t do the human thing and apply a passage right into the hearts of the people in your pews, with theological nuance, pastoral precision, and authentic compassion and conviction. As Chris Green said recently, “AI might be able to suggest 5 songs for a funeral but it’s the wise pastor who knows what the grieving widow might prefer.”
Ethical AI
Major companies such as IBM, Google, and Meta are keenly aware of the issues in AI governance and its responsible and ethical use. And we should be too. There are complex data privacy issues to be considered. When you take a photo of the contents of your fridge and ask AI to suggest a recipe for something you could make from those ingredients, that’s fun; but where does the image go? Who sees it? Who knows what you’re eating tonight and what kind of fridge you have, and will they monetise that information somehow? Are you using artificial intelligence or is it using you? When your phone enables you to artificially remove someone or something from a photo you just took, does it give you pause to realise that Stalin used the same trick with his political opponents? And who is watching the watchers?
Still, you may want to explore using DALL-E or Synthesia for image and video generation. Or experiment with tools that can produce slides from your sermon text, or schedule meetings and draft emails for you. Why not transcribe your sermons or meetings automatically instead of laboriously by hand? AI also has a myriad of applications in health diagnostics, traffic control, banking fraud detection, and in automating a lot of tasks that take up the majority of our time today. It will change the anatomy of work. It may or may not take your job; but being able to use AI effectively in the workplace is going to distinguish employees in the future, and indeed is already doing so.
It can be utilised well as a study assistant, as long as we remember that it is a bit stupid and needs an expert to handle its “creative inspiration” well and not just regurgitate its output. It should be seen as a supplement to, rather than a replacement for, the prayerful, reflective, and pastoral dimensions of ministry. It requires prayerful discernment, but it doesn’t teach you how to have prayerful discernment, and it won’t do prayerful discernment for you. So don’t allow it to make you lazy. “Go to the ant, O sluggard; consider her ways, and be wise” (Proverbs 6:6). Real human intelligence is greater than artificial intelligence, because real intelligence knows how to use (and not use) artificial intelligence.