Evangelical Christians aren’t taking the challenge of AI seriously enough

What we should do about it

“A revolution cannot be mastered until it is understood… The temptation is always to seek to integrate it into a familiar doctrine: to deny a revolution is taking place.” Henry Kissinger, Nuclear Weapons and Foreign Policy (1957).

AI is overhyped, and hasn’t had much real world impact. That’s what Samuel James argued in a recent Substack post, and having taken a straw poll of some friends, I reckon that a lot of other Christians agree.

This stance is misguided and quite wrong, in my view. I’m going to set out why below (feel free to skip if you know lots about AI), then discuss the potential impact that AI might have on the specific type of evangelicalism which I’m part of. Relatively few people are giving this serious thought.

Don’t stick your head in the sand

Samuel James’s article was prompted by a viral article on Twitter which was hyping up AI. Let me be crystal clear: undoubtedly, some people overhype AI. But much of the overhype is based on overconfident predictions about timescales, not the underlying capabilities of the models.

The proof that these models are extremely powerful is stacking up. They are performing well on various benchmarks. They’re at PhD level of expertise in their fields. They’re increasingly good at writing code, which has prompted frenzied excitement on Twitter. They’re increasingly autonomous in the forms of agents who can solve longer and more complex tasks. And they’re even tackling novel technical problems successfully: last Friday, OpenAI announced that their newest model had helped solve a physics problem that no person had solved.

Importantly, current advances have been predictable given increased inputs. We’re still early in this process – the returns to scaling are still high, and there are lots of other potential ways to improve models that we haven’t yet fully tapped out. And implementation has barely started. Even if AI progress stopped right now, I think the current models will have a significant impact on our economy as we work out how to deploy them effectively. I think it was Dean Ball who suggested this framing: a technology like flight is a huge deal, but it takes the construction of airports and a whole host of other changes in order to fully utilise the technology and realise its potential.

To be clear, there are some things that AI still can’t do. There is also a lively debate over just how significant the employment impact will be. But we’re well past the point of being able to argue that this technology is pure vapourware, sustained by hype only.

And the past few weeks of improvements have really struck me. I’ve been following these trends closely for the past few years, but having seen things really happen, I’m starting to feel the AGI.

I don’t think this formulation is unique to me, but it’s a good description of my position: I’m more sceptical of AI’s impact than some of the most excitable engineers in San Francisco, but I’m less sceptical than most people I speak to outside of Silicon Valley.

But is this a Good Thing?

Like any technology, AI will be used for good and bad. As it’s deployed in scientific and medical research, for business and government efficiency, ultimately growing our economies and creating more material abundance for humans to enjoy, there will be lots to thank God for.

On the other hand, lots of ways that AI could be deployed are evil. Bioweapons and other national security threats are one category. Infinite, addictive video isn’t anything to get excited about either (although TikTok means we’re already facing that challenge). Excess passive consumption on screens is a vice – on that, Samuel James and I agree.

There is no real way to get around the fact that the underlying moral convictions of people matter for how they end up using technology. There is no system so perfect that people don’t need to be good.

Challenges (and opportunities)

So what will AI actually mean for churches and Christianity?

Here I am less boosterish than you may expect. It’s been over three years since ChatGPT was released, but I’ve only seen one use of AI in any church service in that time: simultaneous translation for non-English speakers, a kind of modern Pentecost every Sunday. It certainly hasn’t transformed Christianity the way some breathless commentators have predicted.

At the end of the day, Christian religion which highly prioritises the public teaching of the Bible by humans is somewhat insulated from trends like AI. As Ross Douthat pointed out in his interview with Dario Amodei, sometimes we’ll want to choose for humans to stay in charge. Church is a great example – there are strong biblical reasons to continue to do this. These ministers’ jobs are safe from AI. Where else can you go to hear someone teach the most important truths, apply them to a congregation whose deepest joys and struggles only he knows, and sing, talk, and care for each other?

More broadly, AI probably will affect some churches a bit. Some ministers are under pressure to serve their congregants’ needs, and it will be all too easy for AI to write a sermon for you. People will accumulate teaching to suit their itching ears, so AI models who claim to fill a spiritual hole in people’s hearts will probably be popular. (There may be a natural and limited role for AI when it comes to understanding a given topic, but it should not be allowed to govern a pastor’s interpretation of the Bible, and it certainly cannot apply a passage to his own heart or others’.)

But where a church has deep convictions about what it’s doing, or traditions – think Reformed churches like mine – not much is going to change, perhaps ever. “Man gets up and explains the Bible” is very Lindy, after all, having already survived two thousand years of historical and technological change.

So in a funny way, I think AI is going to be a huge deal, but I think we should keep doing what we’re doing. Preach the gospel in season and out of season, and entrust what we know to faithful men who will be able to teach others.

Despite this pleasant picture, there are two areas where I’m more leery of AI’s impact.

One is on what AI says about Christianity itself. It’s hard to know how big a problem it will be: how many people will go to a chatbot for answers to life’s most important questions? Some will, and more will be influenced somewhat more subtly as they ask AIs other questions.

The Gospel Coalition did some interesting research into this, effectively creating theological benchmarks for AI. I’m not sure if they have the right answer to all relevant questions, but we should recognise the danger of an intellectual monoculture accidentally imposed via AI when it comes to important religious questions. The same applies in other fields of inquiry too.

I’m also mildly concerned about economic disruption for white collar workers. As discussed above, it’s at least possible that some of these people will lose their jobs over time. This could have a significant impact on the financial profiles of a lot of affluent churches in America which fund a disproportionate amount of Christian missionary activity. The job of a minister might not be affected by AI, but they still need someone to pay them. On the other hand, AI is going to create a lot of wealth, so if the winners in the AI economy go to church, then maybe the net impact is zero in the end. We don’t know the answer to this yet.

Lastly, AI offers a big opportunity for Christians.

It’s possible that some jobs will disappear due to AI. When people’s careers don’t work out as planned, or they lose their jobs, the disappointment and dislocation – even devastation – are very real. What once seemed certain may be about to turn to vapour. Christians can encourage those who are struggling and help them find real foundations to build on.

Many applications of AI are inherently task-focused, helping us to get things done faster or better. But AI can only simulate relationships. LLMs are not conscious. They cannot be truly human, they cannot truly feel. And biblical Christianity is inherently people-centric: it’s surely significant that Jesus gives the church real people to do his work. If Christianity was either purely intellectual, or purely about correctly understanding the Bible, then AI might be useful. But it’s chiefly about a relationship with God, which overflows into love and service towards those around you.

Our struggle to help AI distinguish between right and wrong, to do what seems clear to us but is far from obvious to AI, can’t help but remind me of how humans struggle to understand and obey God’s rules fully. It should remind us of our own flaws and finitude, and the wonder of the incarnation: the Creator lived, died, and rose as one of the created.

Thanks to various friends for reading drafts of this piece.

Pieter Jansz. Saenredam - Interior of the Buurkerk, Utrecht (1645)