Centralizing and systematizing our approach to evaluating AI bias would force companies to disclose their target demographics and offer the public an independent trustworthy evaluation of how well products serve those demographics, with results that can be directly compared across the marketplace.