Screen for “Principal Basis” Before You Rely on Rankings
Explore how to identify when AI-generated rankings serve as the principal basis for decision-making in DoD and federal workflows. Learn to distinguish between convenience outputs and gatekeeping rankings, assess human review authenticity, and document meaningful discretion. Understand the need for early screening, evidence-based oversight, and compliance with M-25-21 requirements to ensure accountability and appropriate risk management in AI use.
Two AI uses can look the same on paper because both end with a human signature. One use asks a model to draft a weekly internal summary of open issues, and a supervisor initials it before it is posted to a team channel. The other use asks a model to rank people or entities for who gets contacted, inspected, or helped first, and a supervisor initials that list too. The common mistake is treating those signatures as proof the AI is not driving anything important.
A human signature only tells a reader someone touched the artifact. What matters is what the workflow does next, because the ...