ARTIFIA AI LABS v0.1

A decision that changes someone's life should be able to explain itself.

Automated and semi-automated systems now decide who gets hired, who gets credit, and who gets a home. Most of them cannot show their work. This is the standard we hold them to.

Hiring

Screening, ranking, rejection

Lending

Approval, pricing, limits

Housing

Tenancy, valuation, access

THE STANDARDFIVE ARTICLES
I

The criteria are fixed before the case is seen.

What counts is written down in advance, not reconstructed after the outcome is known.

II

Every decision leaves a record.

The inputs, the rule applied, and the moment it was applied all remain recoverable.

III

Outcomes are measured across groups, never assumed to be even.

A rule that treats everyone identically can still land on people unevenly.

IV

The reason is stated in terms a person can understand.

Not a score and not a model name, but the factors that moved the decision.

V

The evidence outlives the system that produced it.

Models are replaced. The record of what they decided is not.

If you make decisions of this kind and want them measured against this standard, write to us.

review@librastandard.com
REVISIONS
v0.1 First publication of the five articles SEP 2026
v0.2 Measurement criteria for article III IN PREPARATION