How we use AI
AI helps us handle breadth and structure. It does not replace evidence, permission checks, editorial judgment, or responsibility for what we publish.
Where AI helps
- Understanding a traveler's intent and matching it to relevant places or guides.
- Extracting structured facts and proposing links between evidence and fields.
- Organizing research, comparing drafts, and flagging stale or conflicting information.
- Synthesizing readable place and guide drafts from approved inputs.
Language is not a mechanical translation
Editorial synthesis is generated independently for each language and audience. The English layer is written for international visitors; it is not a sentence-by-sentence copy of the Chinese layer.
The human boundary
AI never publishes by itself. Automated systems can create drafts and suggestions, but the publication action is restricted to authenticated administrator accounts. Separate reviewer enforcement is still being built. High-risk content also depends on authoritative sources and real verification dates.
What AI is not allowed to do
- Invent a source, verification date, first-hand experience, or permission.
- Turn material restricted to internal use into redistributable public content.
- Resolve a meaningful source conflict by hiding the uncertainty.
- Publish or materially overwrite reviewed content without human approval.
If an AI-assisted passage is wrong
The correction standard is the same regardless of how a draft was produced. Please use our corrections channel; we verify the claim and update or withdraw it when warranted.