AI transparency
Part of the AI Wiki policies
AI Wiki is an encyclopedia about AI that is built with AI, and we think you should know exactly what that means rather than having to guess.
How articles are made
Most articles are researched and drafted with the help of AI tools working under human editorial direction. That assistance is why a small team can keep thousands of fast-moving articles current. It is not why you should trust the result; the process around it is:
- Drafting starts from sources, and briefs are treated as hypotheses: names, dates, and figures are verified against independent sources before publication, and claims that fail verification are cut, whatever produced them.
- Articles then pass adversarial review passes whose explicit job is to refute what the draft says, because generated text fails in confident, plausible ways that friendly proofreading misses.
- The same sourcing standards and verifiability rules bind every revision, human-written or AI-assisted.
Labeling
- Every revision in an article's public history records whether it was AI-assisted; look for the "AI assisted" badge on any article's History page.
- Independent fact-check passes stamp the visible "Fact-checked" date in each article's header, so you can see when a human-directed verification last confirmed the article against its sources.
Why we work this way
Hallucination is the failure mode of ungrounded generation, not of generation itself. Our defense is structural: source-first drafting, verification that is separate from writing, public revision history, reader corrections, and visible flags when something fails a check. We publish this policy because a reference that uses AI and says so, with controls you can inspect, is more trustworthy than one that uses it quietly.