AI Congress ยท Methodology

Evidence before labels.

The rubric is designed to make disagreement inspectable. A classification must resolve to public evidence, record its confidence, and preserve uncertainty.

AI Congress describes a lawmaker's current AI policy posture. It does not score general ideology, partisanship, technical knowledge, effectiveness, or personal character.

Party never determines a classification. Interest-group ratings and reporting can help locate evidence, but published assessments rely on the underlying legislation, vote, framework, official statement, or other primary record.

Three camps.

The camp is a concise top-level description. The underlying dimension scores and evidence remain the substantive assessment.

CampWorking definitionTypical evidence
Move FasterPrioritizes rapid AI development, infrastructure, and adoption through lighter regulation, regulatory flexibility, or narrowly targeted rules.Permitting reform, sandboxes, national uniformity, adoption incentives, or opposition to broad AI-specific rules.
Set GuardrailsSupports continued development while imposing meaningful safety, consumer, worker, competition, or infrastructure requirements.Mandatory testing, defined liability, community conditions, worker transition, or targeted high-risk-use rules.
Slow DownSupports policies that pause or materially constrain advanced AI development, release, infrastructure scaling, or private control.Moratoria, broad licensing before release, government approval, or substantial public control of leading AI firms.
Not Classified is not a fourth camp.

It means the reviewed record does not support a defensible overall classification. It should never be interpreted as neutral, moderate, undecided, or disengaged.

5 dimensions

Each dimension has its own five-position rubric. The endpoints are policy-specific, so the numeric values are not a general liberal-to-conservative scale.

DimensionPosition 1Position 3Position 5
Data CentersPause ConstructionBuild with ConditionsBuild Faster
AI RulesApproval Before ReleaseTargeted GuardrailsKeep Rules Light
WorkersPrevent DisplacementPrepare WorkersPrioritize Adoption
Market PowerPublic ControlEnforce CompetitionLimit Intervention
Who Sets the RulesStates LeadShared AuthorityFederal Rules Only

Evidence hierarchy.

Direct action carries more weight than general rhetoric. Recent comprehensive frameworks can outweigh older or narrower statements when the editorial rationale explains the choice.

LevelEvidenceUse
ASponsored legislation, direct amendments, clean roll-call votes, and comprehensive official frameworks.Primary basis for dimension and overall classifications.
BCosponsored legislation, committee action, formal letters, detailed floor statements, and detailed official statements.Direct support, corroboration, and current-position evidence.
CCampaign platforms, candidate questionnaires, official press releases, substantive interviews, and clear first-party social statements.A specific statement can support a dimension score when the policy direction is unambiguous.
DBrief or ambiguous remarks, secondary summaries, and reported characterizations without a direct quotation.Discovery and context only. Not sufficient by itself for a published classification.

Overall classification

The top-level camp uses Data Centers, AI Rules, Workers, and Market Power. Who Sets the Rules is shown separately because federalism does not move cleanly along the same restrictive-to-permissive direction.

A classified senator needs at least three scored directional dimensions, including Data Centers or AI Rules, and at least two direct primary sources. Arithmetic provides an initial range. Explicit moratoria, broad licensing, acceleration frameworks, and conflicting comprehensive records trigger manual editorial review.

Corrections and changes

Classifications are dated snapshots of an evolving public record. New legislation, votes, campaign platforms, or corrected source interpretation can change a score. The source record and rationale should make those changes auditable.

Race selection

The Midterms Hub is a focused research set, not an election forecast or an exhaustive list of competitive races. A race is included when the major-party nominees are confirmed and the current public record supports useful records for both candidates on at least one AI-related dimension.

The same evidence hierarchy applies to Senate and House candidates. Campaign statements are labeled separately from legislative and government records. A missing or ambiguous position remains unscored.