New approach to directive driven AGI cognition