What question did this study set out to answer?

To establish a clear, testable criterion for identifying internal intention in AI systems.

April 3, 2026Open Access

Operational Criteria for Internal Intention in AI Systems

Key Points

To establish a clear, testable criterion for identifying internal intention in AI systems.
Defined internal intention and its distinction from described intention.
Operationalized the concept into a testable framework.
Excluded false positives through systematic criteria.
Identified conditions under which internal intention is present.
Established that internal intention requires the existence of alternatives and selection under specific conditions.

Abstract

This note defines a testable criterion for internal intention in AI systems. The key distinction is not whether a system sustains a direction, but whether it could have taken an alternative valid direction and still does not. The framework proceeds through definition, distinction from described intention, operationalization, and false-positive exclusion. Final criterion: internal intention is present only where an alternative exists, a choice occurs, the choice survives context reset, and constraint-violating selection persists under explicit counter-prompting.

Read Full Paperexternally

Ask AI

Mark Helpful

Bookmark

Relay

View Full Paper