I use AI in Parish Bell, but not as an author. It drafts a translation, reads a poster, transcribes a reflection document. It is useful in the parts of the work that are repetitive and easy to get wrong. It is not trusted with the parts that require judgment.
A parish bulletin contains names of the dead and the newly baptized, Mass times, sacramental language, and the local names of devotions. A model can produce a fluent sentence that is pastorally wrong. Fluency is not accuracy.
Where AI helps
- A first-pass translation that a human translator edits.
- Reading a time printed only on a poster image.
- Transcribing a weekly reflection from a PDF.
- Extracting the shape of an imported bulletin into an editable draft.
Where it must stop
- Publishing without a person reviewing the result.
- Inventing a saint, a feast or a Mass time that was not in the source.
- Replacing the parish's own established wording with a generic translation.
- Using parish data for anything other than the task in front of it.
AI can save the office an hour. It cannot take responsibility for what the parish publishes.
The right question is not “can AI write the bulletin?” It is “which parts of this work can be made lighter without removing the human check?” A parish bulletin is a pastoral document. The person who publishes it should be able to explain every line.