Markdown

Faith AI Authority Checklist

Use with Faith and Spiritual Reflection: Use Questions, Not Synthetic Authority.

Sort every request into one of four categories

  1. Quotation and comparison — retrieving a passage, comparing translations, finding a cross-reference. Safe.
  2. Interpretation aid — summarizing published commentary, clearly attributed to named sources. Safe with verification against the actual source.
  3. Reflection prompting — generating open-ended questions, without the model supplying “the” answer. Safe.
  4. Doctrinal or pastoral verdict — settling a theological dispute, claiming to speak for your tradition, acting as confessor, counseling a personal crisis. Never appropriate.

A reflection-prompting template

I am reflecting on [a specific passage or teaching from my own tradition].
Generate 5 open-ended questions that would help me think about how this
applies to my own life right now. Do not tell me what the passage means
or what conclusion I should reach — I want questions, not interpretation.

The accountability test

Before accepting an AI-generated answer on a faith question, ask: if this guidance were wrong or harmful, who would answer for it? A model answers for nothing. If the task requires someone to be accountable to a congregation, student, or person in crisis, it belongs with a person your tradition recognizes for that role.

Before you use AI for study or sermon preparation

What this cannot do

AI can retrieve, compare, and prompt reflection. It cannot hold faith, cannot be held accountable by a religious community, and cannot carry the lived weight of pastoral relationship. Bring interpretation, confession, and pastoral guidance to your own tradition and community.