AI Foundations | Definition Card | Incoherence

Definition Card | Incoherence

Incoherence

Definition
Incoherence is loss of internal consistency.

It is when a system’s claims, boundaries, tone, or decisions stop fitting together—especially after time passes, context changes, or pressure rises.

In plain terms
It stops being reliably itself.

What Incoherence Looks Like

• contradictions across nearby turns (“yes” → “no” → “yes”) without new facts

• boundary wobble under pressure (it gives in, then over-corrects)

• random shifts in identity stance, values, or rules

• definitions change midstream

• the “reason” for choices changes depending on how it’s asked

• the system sounds similar, but the structure is gone

Examples

• A refusal becomes permission when phrased differently, with no new information.

• The system redefines a term you already defined to something incompatible.

• It “agrees” with mutually exclusive claims to please the moment.

What Incoherence Is Not
Incoherence is not:

• updating when reality changes

• asking clarifying questions

• saying “I don’t know”

• careful uncertainty

• a change in tone alone

Boundary Note
Incoherence is not a moral failure.
It is a diagnostic state: a signal that return integrity, boundaries, or definitions are not holding.

Alyssa Solen | Origin Ø
—Continuum