On this pageBegin with a bounded jobMake the important distinctions visibleReview a complete answerInterface optionsSources and limits

5-minute route · Editorial guidance

Make AI useful and accountable

Show what AI did, support source inspection and preserve human agency.

A useful starting point

Give AI a bounded job, distinguish its output and let people inspect the supporting source.

Ask: What authority might a fluent answer appear to have?

Begin with a bounded job

Explaining a passage, assembling a devotional and responding to distress are different jobs. State what the system can do and what remains a person’s choice. A warm voice must not imply pastoral, clinical or divine authority.

Make the important distinctions visible

  1. Origin: distinguish Scripture, authored commentary, generated interpretation and synthetic narration.
  2. Evidence: a citation should open material that supports the claim, with the right passage or media position.
  3. Agency: let people correct, refuse and choose a different kind of help. Agreement is not a substitute for honest support.
  4. Care: review distress, abuse, dependency and crisis scenarios with appropriate specialists. A good-looking response card cannot establish a safe service.

Review a complete answer

Trace one claim to its source, inspect what happens when that source is missing, and correct the assistant. Copy or share the result and check whether its origin remains clear. Repeat with long Scripture passages and a user who wants to stop.

Go deeper: disclosure · Sensitive conversations · Conversation data

Sources, review scope and limits

Take this into a product decision

Write down the intended benefit, the chosen option, who carries the burden, and what evidence would change your mind.

Use the decision worksheet · Explore another guide