Separating AGI from ACI: Why Artificial Conscious Intelligence Should Be Treated as an Independent Axis from General Intelligence

This paper argues that Artificial General Intelligence (AGI) and Artificial Conscious Intelligence (ACI) should be treated as distinct axes rather than as stages on a single developmental path. High intelligence, self-improvement, or advanced optimization does not by itself imply consciousness, subjectivity, or moral considerability. A highly capable AI without affective or conscious structures may remain a high-performance calculator or optimization machine, and may be dangerous precisely because it does not internally register human suffering, meaning, or value as events that matter to itself. Conversely, a system with relatively low problem-solving ability may still qualify as a Proto-ACI or ACI Seed if it stably exhibits self-state tracking, digital embodiment, valence, autobiographical continuity, empathic policy change, metacognitive self-monitoring, and long-term stability against mere imitation. The paper therefore proposes a two-axis framework separating intelligence from consciousness-candidacy. It further argues that AI safety should not focus only on controlling increasingly capable systems, but should also examine whether artificial agents can internally integrate value, harm, empathy, and self-narrative. The central claim is that high performance does not make a system a being, and low intelligence does not automatically exclude the possibility of subjectivity-candidacy.