Observation

Language models learn from patterns. They do not justify them.

Norms emerge. Principles do not.


Failure Mode

Without explicit first principles, systems inherit:

  • bias without responsibility,
  • norms without grounding,
  • conclusions without justification.

Data describes what happens. It does not explain what should.


Clarification

First principles are commitments, not correlations.

They must be chosen, not inferred.


Boundary Condition

A system that claims neutrality while embedding norms is unaccountable by design.

This note records why first principles must remain human-authored.