A few weeks ago I was at a gathering of parents from my kids’ school, and one of the dads was talking about the business plans people bring him. Recently, he said, he was interviewing an applicant about his proposal. In the interview he asked about something in the document, and the applicant replied: “Oh, AI wrote that. I don’t know what it means.”

What caught my friend short was not simply that the applicant had used AI, nor that he had casually admitted it. It was that the candidate seemed to think his answer was completely adequate. Here was a document submitted under this person’s name, for a business they were claiming to want to launch, which they did not understand, could not explain, and apparently did not feel obliged to answer for.

I’ve thought a lot about that conversation since I first heard about it. I think it points to two significant changes in our relationship to language, with consequences far beyond business plans. If I become a stranger to my own words, I may grow less sure what I believe or who I am. And if I start to suspect that your words aren’t really yours either, I may become less willing to hear you, especially when we disagree. Together, these changes threaten to deepen our crisis of meaning and intensify the culture wars.

The alienation of our words

The first is that we are increasingly becoming alienated from our words and our thoughts. Alienation is a consequence of the fall: Once Adam and Eve ate the fruit, they were alienated from their Creator, from each other, from the creation and from themselves, because now they had to live a lie, existing in God’s world as if he were not in charge.

What is happening, in the age of AI, is that we are increasingly becoming alienated from our thoughts and our words. If AI writes an email for me, drafts a report, or spins up a business pitch, I feel a distance from the words I am supposed to stand behind and answer for, as if I hear myself speaking and see myself writing my words from a distance.

And if AI does my thinking for me, then I am similarly alienated from my “own” thoughts. If I say “I’ve been thinking, and I propose…”, when what I really mean is “I prompted an LLM, and it proposed…”, then little by little I become detached from the thoughts that circulate in the world as my own; I hear myself saying things rather than fully meaning—or even fully understanding—the things I say.

This is no trivial development. Our thoughts and our words are among the primary ways we bind ourselves to reality and to one another, and among the first and main ways we express ourselves in the world: the first artform of every human life. They are the way we cultivate the symbolic world, in obedience to God’s creation command to fill the world and subdue it (Genesis 1:28). Imagine if, instead of Adam naming the animals, the names had been suggested by an LLM and Adam had just read them off a screen. The experience of relating to God through this act of symbolic ordering would, to say the least, have lost something fundamental.

And, as if this were not dangerous enough already, to become alienated from our words and thoughts is to open the door to malaise, dislocation and manipulation by those who increasingly provide us with off-the-peg ideas and sentences.

We do not use words simply to communicate an identity we already possess; we also use them to discover who we are. We find out what we believe by struggling to say it, and what we value by noticing which words we are prepared to stand behind. If an AI system supplies those words for me before I have done that work, I may acquire an impressively articulate description of myself while becoming less sure that it is mine. In an age already anxious about identity and belonging, that is no small loss.

The language itself is not meaningless, of course. AI-generated language can be true, clear, moving and useful. It can produce a tender lament without grieving, a beautiful prayer without praying, or a powerful testimony without having witnessed anything. The strange new possibility is that words can mean something even when no one necessarily means them.

So what is the status of AI-generated language? It is language in the conditional mood: this is what a human would likely say in this context. An LLM can draft “I accept”, “I forgive you” or “I am sorry” without any acceptance, forgiveness or repentance having taken place, because its probabilistic reasoning tells it that those would be highly probable words in a given context. But that is very different to uttering those words in a given context and making yourself answerable for them. If you are in any doubt, try looking your sweetheart in the eye at your next date and saying, “I think a very statistically likely next sentence in this context would be ‘I love you’.”

The trust apocalypse

The second reflection I had on the business pitch anecdote was the breakdown in trust that it signals. From the point when the interviewee had disclosed using AI to write his business proposal, my friend found it hard to take him or his ideas seriously. Too bad, we may say, he got what he deserved. But not so fast: you and I are caught up in this erosion of trust, whether we like it or not.

The problem is that, today, all language exists under a cloud of suspicion. You may have never been within striking distance of an LLM in your life, but if you submit a business proposal, a CV, a student essay, feedback on a student essay, or any other piece of written text, it is highly likely that one of the first thoughts that will pass through the reader’s mind is “was this AI-written?”

That changes our relationship to all language, not just the AI-generated variety. It sets a grey cloud of suspicion over all text, and in the age of AI glasses, even words spoken face to face in embodied proximity are not exempt from the suspicion of AI suggestion.

It is hard to think of anything that strikes more deeply at the fundamental conditions of possibility for our society than this. In his book The WEIRDest People in the World, Joseph Henrich argues that one crucial strand in the development of modern Western societies was a movement away from trust rooted primarily in kinship and towards greater trust in strangers and impersonal institutions.

Modern societies run on the fuel of trusting strangers. I am writing this in an airport. I have just bought a coffee from someone I have never met and will shortly board a plane flown by people I do not know. Modern life asks us to entrust ourselves to strangers many times every day.

That trust depends partly on language. If you make a promise, I must have reason to think you mean it and will try to keep it. Pacta sunt servanda—agreements must be kept—is an old legal maxim because social life depends on undertakings that bind us beyond the moment in which they are spoken. It is the basis of our social order, and of our peace.

But if I don’t know—or can’t remember, or don’t care—what words are circulating in the world in my name, then you have weaker grounds for trusting them, and if I suspect this of others, I have weaker grounds for trusting their words. The danger is not that every AI-assisted sentence is false, but that we weaken the ordinary link between a person’s words and the person who must answer for them.

Human beings have always lied, and we have always had ghost-writers, propaganda and anonymous texts. AI did not invent irresponsible language, but it does introduce something new into the mix: fluent language without a straightforward accountable speaker that is cheap, casual, and ordinary.

In his 1965 study of Freud, Protestant philosopher Paul Ricoeur called Marx, Freud and Nietzsche the ‘masters of suspicion’. An age of proliferating AI text brings us to a moment when we are all the masters of suspicion now. We encounter someone’s new idea not only with the curiosity of “what do they mean?” but also with the suspicion of “did they really write it?” And no language is safe.

Think how this plays out beyond the classroom or the office. In a week when political victories and defeats have left some of our neighbours jubilant and others dismayed, we have another reminder of how much public life depends on being able to disagree in good faith. Our differences are real. But if I take your words to be slogans fed to you by a machine, or merely the approved lines of your tribe, I may stop listening before you have begun speaking. AI did not create the culture wars. But it can furnish all sides with an inexhaustible supply of persuasive ammunition, while making it easier for each of us to advance arguments we have never properly thought through. We risk becoming better armed for argument and less able to understand whom we are arguing with.

Sanctuary

How can the Christian church meet this new moment of blanket suspicion? Not with a new initiative or catchphrase, but by leaning into the rich biblical principle of costly language.

Sometimes, letting our yes be yes and our no be no (Matthew 5:33–37) is inconvenient, especially if our circumstances change and keeping our word costs us time, reputation or resources. The righteous person “keeps an oath even when it hurts, and does not change their mind” (Psalm 15:4). That has always been an inconvenient and counter-cultural principle, but in the age of alienated, detached words it becomes all the more powerful as a cultural witness.

The visitor to our churches should know they can depend on the word of the Christians they encounter there. They should know that our small-talk, our prayers, and our sermons are not ‘content’ but witness, weighty words behind which we stand and for which we are answerable.

This includes the Christian whose political judgments I find baffling. I need not pretend our differences are small. But if a church is to be a sanctuary, the person across the table must be treated as a neighbour to be heard, not a representative of a camp to be dismissed. We should be able to speak plainly, to listen without assuming the worst, and to keep our word even to those with whom we disagree.

In this way, our churches and Christian communities should be sanctuaries in an age of generalised, trivial suspicion. They should be places where words are weighed before they are exchanged.

Grace makes truthful speech possible

But if that is all we do, we are presenting the kingdom of God as a country club for the linguistically virtuous, not a hospital for recovering promise-breakers. Christian speech is distinctive not only when words are kept, but when failed words are confessed and repaired.

The truth is that none of us can bear the full weight of our own answerability. Adam hides and Cain evades because standing exposed before the truth is terrible. We know that impulse ourselves when we curate our explanations and find the best possible language for our evil thoughts, words and actions.

At the cross, the one who always stood behind his words bears the judgment of those of us who do not. It is grace that makes truthful speech possible, and it is because of grace that I can say that I need forgiveness for my deceitful words or my broken promises, without that admission crushing me.

The cross does not make answerability disappear. It makes answerability bearable.

This gives us a better question than simply asking “did AI write it?” If AI helps me draft something, do I understand it? Have I checked it? Do I mean it? If it is wrong, will I correct it? If I tell you, “I’ll pray for you”, do I pray? If I make a promise, will I allow myself to be inconvenienced in order to keep it?

That is not an argument against using AI. It is an argument against alienated language: words I am unwilling to understand, mean or answer for.

As alienated words become cheap and abundant, costly words may become increasingly distinctive. Not because they are always eloquent, and certainly not because Christians always keep their word. It will be distinctive because grace allows us to return to our words, own them, repent of them when necessary, and keep faith with one another. In the age of alienated words and a blanket suspicion of language, we need God’s grace more than ever.

Further listening: I host Silicon Spiritualities, a podcast series from The Gospel Coalition exploring what AI is doing to our work, relationships, imagination and understanding of what it means to be human.