This is the second in a series of articles designed to provide a framework for how churches should respond to AI. If you missed our first post, "How Should Churches Approach AI," check it out here.
What is AI, anyways?
A lot of tools get lumped into the ‘AI’ banner. At Mere, we prefer to use the word ‘machine learning’ to encompass this broader category of tools. This includes useful utilities like automatic translators (think Google Translate, live, for video), speech-to-text converters (transcripts of your Sunday sermon available on your website, same-day), and pattern detectors (tools that doctors use to help diagnose and identify odd diseases).
But, when most people say ‘AI,’ what they have in mind is ‘large language models:’ chatbot products like ChatGPT, Claude, Google Gemini, and Grok. They’re also probably thinking of ‘AI’ modes bolted on to products they use every day, like an AI help assistant, or Microsoft’s Copilot.
A large language model is a program that has absorbed huge amounts of writing, and uses statistical patterns to predict, word by word, the most plausible thing to say next — producing fluent, convincing language without actually understanding any of it. In fact, the previous sentence was generated by a large language model (the only one in this article).
For the purposes of this article, we'll continue to conflate AI and ‘large language models,’ because that’s how most ordinary people use the label: as a shortcut for the term large language model.
And, these things are pretty convincing facsimiles of human speech! They’re convincing enough that Richard Dawkins, of all people, has concluded that AI is sentient and conscious (If you're interested, Gary Marcus has a great response here). Marcus writes:
“The fundamental problem here is that Dawkins doesn’t reflect on how these outputs have been generated. Claude’s outputs are the product of a form of mimicry, rather than as a report of genuine internal states. Consciousness is about internal states; the mimicry, no matter how rich, proves very little. Dawkins seems to imagine that since LLMs say things people do, they must be like people, and that simply does not follow.”
That’s a rabbit trail there’s no time to get into on a simple blog post, but suffice it to say, LLMs can do incredible things, some people (largely the founders of the companies who own LLMs) are convinced they’re the ‘future of work,’ and that they’ll substantially ‘replace human labor’ (citation needed from the pro-AI crowd).
If this is true, the church cannot ignore this technological innovation, much like it could not ignore the printing press or the internal combustion engine.
Some will argue that the church must respond to this new technology, be on the bleeding edge of new developments, and use it for all its worth. Others, like The Gospel Coalition, will advise caution and refrain from using AI tools altogether. So where does this leave us?
Machines can talk like people now. More and more, people have started to talk like machines (customer support chats, rote business emails, even personal communication has started to become veneered with AI). What should we do about that?
Next week, we’ll start to look at how to engage with AI, if it’s ethical to use AI tools, and build our framework for what faithful engagement looks like for pastors and church members alike.
