Neither Fear Nor Worship
A pastor I know sent me a news clip and asked me to address it. A researcher who spent three years at Anthropic, one of the companies building the most capable AI systems in the world, had resigned in public. His message was blunt. The labs, he wrote, are racing toward self-improving systems and "gambling with our lives."
She asked me because I build AI for a living. I build it for churches, which makes the question sharper, not softer. So here is the answer I gave her, expanded for every pastor and every executive who has watched a clip like that and wondered whether to be afraid.
Part of the warning is true
Start with what he got right, because it matters. The race is real. The companies building these systems compete on capability, and capability is easy to demonstrate. Safety is not. It is harder to measure, slower to show progress, and it does not win a sales meeting. Anyone who has run an organization knows what happens to the work that is hard to measure when the quarter is short.
And nobody fully understands these systems, including the people who train them. That is not a secret someone is hiding. It is the plain state of the field. So I will not tell you the risk is overblown. I do not know that, and neither does anyone else. That uncertainty is the fact worth sitting with, and it deserves better than a shrug.
The framing is wrong in a way that matters
"AI could kill us all" pictures the machine as a creature, something with a will and an appetite that one day decides we are in the way. That is not what these systems are. They have no survival drive. They have no goals they were not given. They do not want anything.
The danger I actually see is less cinematic and far more likely. It is people handing their judgment to a tool that is confident, fast, and sometimes wrong, and not noticing until it matters. That is a human failure with a technical accelerant. It is already happening in small ways: the report nobody checked, the summary everyone quoted, the recommendation that became a decision because questioning it felt slow. Let the same habit run up into power grids, weapons, and courtrooms, and the small failures become large ones.
I see the small version in my own work. A tool I rely on will do exactly what I asked, in exactly the wrong place, the moment I stop watching. Nothing about that is malicious. It is simply unwatched.
The doom story and the salvation story are the same mistake
Two stories are being told about AI right now. One says it will destroy us. The other says it will save us: cure disease, end poverty, write the sermon, run the company. They sound like opposites. They make the same mistake. Both give the technology a status it does not have. One treats it as a devil, the other as a god, and either way we stop treating it as a tool that answers to someone.
Fear and worship end in the same behavior. The frightened leader freezes and leaves the decisions to whoever is not afraid. The dazzled leader hands the decisions to the machine. In both cases the human steps back from the place where judgment belongs.
Scripture has a better word for the posture we need: "For God didn't give us a spirit of fear, but of power, love, and self-control" (2 Timothy 1:7, WEB). Self-control is not a small virtue here. It is the discipline of keeping your hand on the decision when something faster offers to take it.
What I would want a Christian to hear
I wrote Image Bearers because I think the deepest question AI raises is not what the machine is. It is what we are. The church's answer has always been that a person's worth is given by God, not earned by output. That answer turns out to be the steadiest ground there is for this moment.
If your worth is given, a machine that writes better than you cannot make you unnecessary, and a machine that fails cannot take that worth away. You are free to use the tool without being threatened by it, and free to refuse it without being left behind. People who know who they are do not need to fear the thing, or bow to it.
Sober stewardship, in practice
For pastors and executives alike, it comes down to a few habits.
Keep a person answerable for every decision that touches a person. A hiring call, a pastoral response, a loan, a diagnosis. The tool can draft. A human decides, and owns the decision.
Check before you trust, and keep checking. The cost of producing work has collapsed. The cost of verifying it has not. Budget for the checking, because that is where the risk now lives.
Never let the tool speak with a voice it does not have. A machine can help a pastor prepare. It should not pretend to be the pastor. The same goes for a letter from the CEO or a note of condolence.
Watch what it does to your people, not just your numbers. The question is not only whether the work got done. It is whether the people doing it grew or shrank.
Where that leaves me
This is why I build what I build the way I build it. Tools for the church, with a human in charge, and nothing that pretends to be a pastor. I do not think the people warning about AI are fools, and I do not think the people building it are villains. Both are right that this is serious, and both are tempted to turn it into something it is not.
The researcher who resigned did his part. He said out loud what worried him. Ours is quieter and harder: to stay awake, keep our judgment, and refuse both the panic and the praise. Neither fear nor worship. Stewardship.