The first time you ask AI for help with a note to your volunteers, it comes back fine. A little polished, a little generic, but fine. So you send it.
A few weeks later you notice something. Everything you’ve written with AI help sounds about the same, and none of it sounds much like you.
Nothing broke. The tool did what it does. AI learned from millions of writers, not from you. It doesn’t know your church, your convictions, or your people unless you tell it. When a request leaves gaps, it fills them with the most common answer. And in a long conversation, your early instructions quietly fade.
I call that drift, and it shows up in six ways. Your voice starts sounding like a chatbot. Your theology gets flattened into a safe middle. Quotes and dates appear that nobody checked. The AI starts deciding what the message is instead of helping you say it. “Close enough” becomes the standard. And your instructions wear off the longer you talk.
The answer isn’t a better prompt. It’s writing down, once, the things that make you you: what you believe, how you sound, how you think through a decision, and what a finished piece of work has to look like before it goes out. Then every AI task starts from those anchors instead of from the average of everyone.
It’s less work than it sounds. A bivocational pastor can build the core four files in a weekend.
This week, try one thing. Take a piece of AI-assisted writing from the last month and ask the AI to check it for those six kinds of drift, quoting the lines that slipped. You’ll see the pattern fast.
If you want the full walk-through, it’s in the Pastor’s Workbench. The free lesson “Why AI Drifts” is a good place to start: https://www.skool.com/the-pastors-workbench-1893