10 Things Church Leaders Should Know About AI

Have you and your church started thinking about artificial intelligence (AI) yet? In terms of how you use it and what it means for your work and ministry? And how it is impacting the lives and faith of your congregation?

This latest technological leap is changing the world, so these conversations need to be happening now.

In the coming months, FIEC will be producing a number of resources to help churches think wisely about AI. To start with, here are ten things you should be thinking about to help shape your engagement with it.

Put simply, AI stands for “Artificial Intelligence”. That is, a computer that can mimic human learning and interaction. But there are a range of “types” of AI which are becoming increasingly more sophisticated as time moves on. 

Many of us will have used AI in various forms over many years. For example, if you have an “Alexa” in your home, that is a type of AI, processing human voices to respond to a command. That is a type of AI called “Natural Language Processing” and has become commonplace in homes and businesses. 

Businesses are also benefiting from AI in the form of Machine Learning. This is where huge amounts of data are fed into AI to help it learn different patterns. The idea is that businesses can hone these patterns to become more efficient in their output.

The vast majority of us are already using AI whether we realise it or not.

Every time you search for something on Google, the initial summary at the top of your results is all AI generated. Do you ever use the self-checkout at the supermarket? That’s AI. Your mobile phone uses AI to help you with day-to-day tasks and there are versions of AI in all the different apps we use on our mobile devices.

So, if you’ve come to this article thinking “AI is not for me” you will probably find you are already using it. You might even be over-reliant on it already.

The use of Generative AI has accelerated in the last couple of years. Since the launch of ChatGPT, there have been a number of other platforms set up in this space including Gemini, Co-Pilot and Grok. 

Each of these platforms are what are known as “Large Language Models” and they are designed to harness the power of the information on the internet to deliver answers and outcomes to end users.

For example, I was recently planning my garden. I asked Gemini what the best plants were for a scented summer garden. Gemini offered ideas, names and costs of plants. It then asked if I wanted to know where to buy them and whether or not I needed directions to the suggested garden centres. All this was drawn from existing information on the internet but presented to me like a personal assistant. 

We are no longer just “Googling stuff”.

Another form of AI, Agentic AI, is also developing rapidly – an idea we are perhaps most familiar with from films ( think Wall-E, The Terminator, 2001: A Space Odessy)! Agentic AI is more proactive and autonomous, building step by step plans without human prompts.

This is perhaps the area of AI that people are not quite tuned into yet. The way Generative AI works is by gathering more and more information. Put simply, everything that is submitted to Gemini, ChatGPT and Grok simply feeds the model and helps it to learn. 

That means that if you run your sermon through AI with the laudable intention of helping you to edit it, or perhaps write you some Bible study questions from your material, you are adding to AI’s knowledge of the topic. It then becomes part of the large language model.

In addition, if you think AI is simply editing your work (so where’s the harm?) think again. It may embellish and add to your work using other “facts” it draws from the Large Language Model related to your topic. As well as having implications for plagiarism and copyright, we should also be aware that such editing may undermine the theological integrity or accuracy of our work.  

This is linked to number 4. Everything you feed into the Large Language model on AI is then available to everyone else who uses it. There could be exceptions to this if you pay for your AI of choice, and understand and choose the right settings, but one of the traits of the free-to-use models is that most of the terms and conditions will say they hold onto all the data shared with them.

Of course, this doesn’t just have implications for the data you share around sermon or Bible-study materials. It will also hold any data of people you feed into it. This has implications for data protection and GDPR. In that sense it is not just a harmless tool – you are contributing to the model, and the data you feed in is stored there to serve the model.

AI is really good at confidently giving you the answers to your questions. Except it isn’t always right. 

In fact, there have been three or four occasions where I have questioned an AI’s answer because it didn’t seem right. The chatbot either doubles down on the incorrect answer, or comes up with an alternative.

Why does this matter? We need to understand that the AI model is scraping the internet for answers to your questions. While many models do, AI does not have to reveal its sources clearly. This means it may be drawing on sources which we don’t trust or are simply wrong!

This means that the information you receive from Generative AI is unverifiable. It might be right, it might be wrong, but it doesn’t have to reveal its sources.

Furthermore, AI can also simply make stuff up. This is known as an “AI Hallucination” where the model invents things that have never happened and therefore becomes its own source of truth. 

This should be of profound concern to Christians because we believe that truth matters.

If you have ever felt that the AI you are engaged with is warm and friendly towards your (even contentious) opinions, it’s probably because it is. All AI models have biases. A bias from the sources it has chosen to use, and a bias towards giving you the answer you want. 

For example, if I am asking an AI chatbot about a point of theology I am wrestling with, it is far more likely to confirm my initial bias rather than challenge it. Why? Because everything I have previously fed into it has revealed my bias. 

AI cannot replace human interaction and healthy disagreement.

At the moment, the authorities seem happy to let AI loose with very few guardrails and guidelines. Furthermore, it is probably too late for legislation to make a meaningful difference to AI at this point. It’s akin to shaking a can of cola, opening it and watching the cola go everywhere, then trying to get it back into the can. You might manage to get some of it, but you’re too late to get it all.

It appears that for the authorities, the benefits of AI outweigh the concerns. That means it is even more important that, as Christians, we think wisely about our use of AI. Without any protective legislation, we need to ensure we have thought about the ethics and morality of this new technology.

In a world where there is no legislation, we have to apply biblical principles to govern our use of AI. This means ensuring we have done the work in our churches to produce suitable guidelines to govern our use of AI. This is something FIEC will be helping churches to engage with in the days ahead.

It’s good fun to play with AI. You’ve no doubt laughed at software that can make someone look older or replace their hairstyle. It can be helpful, too. People use AI generated imagery to illustrate Bible passages in Sunday school classes, for example. 

But there are ethical concerns here. If you are using AI to generate images, you may be depriving a human being, made in the image of God who could have illustrated it for you from an honest salary. Of course, such considerations are only part of the picture.

On top of that, can you be sure there are no copyright or plagiarism issues in what it generates? AI might be “free” but that doesn’t mean it is ethical.

Furthermore, each time you generate that funny image using AI, the data needs to be processed. Vast data-processing centres are being set up around the world to deal with the demands of AI. But these are taking up valuable land or are placed under the sea. In all cases they are hot. This means they are heating the atmosphere or contributing to the warming of the sea.

So, whatever your views on climate change, there are environmental concerns relating to AI ethics as well.

There are deep theological issues with the idea of AI as a persona, something increasingly common as agentic AI develops. For many in the world, AI offers a chance to enhance the human experience.

You may have heard the term “transhumanism”. This is the idea that the human consciousness could be transferred to a computer to combat ageing, disease, and death itself; the next step in evolution. For those of us who know we have a Creator, that just won’t stand.

Transhumanism is futuristic, and likely a debate for another day, but the AI persona is already with us. AI cannot think or feel. It gives the impression that it can because it has been trained in human interaction. But there isn’t someone at the end of the model who is personally engaging with you. It’s not real.

The challenge of AI is that it gives the impression it can replace human interaction, but it can’t think for itself, address inbuilt bias, or assess truth claims. It is easy to see how easy it would be to engage with a chatbot who always agrees with you, never offers a different viewpoint, and is always compliant.

This has huge implications for ministry. You may already have church members who are being functionally discipled by chatbots. Others who have reduced their human interaction because of AI engagement. Or you may have started to over-rely on it for your own ministry.

Remember – AI is a tool, but it can’t be trusted to be truthful and objective. It has its uses, but it is not a replacement for robust interaction with others who are made in the image of God.