Intelligence Is Not a Scale: A Structural Taxonomy for Evaluating Artificial Minds

The debate about whether AI systems understand, feel, or deserve moral consideration has stalled. Not because the questions are unanswerable, but because everyone is arguing on the wrong axis. This paper argues that intelligence is not a single scale where AI sits somewhere below humans. It is a layered structure with three qualitatively distinct levels, each requiring conditions that the level below cannot produce through scaling alone. I call these the Computational Layer, the Viability Layer, and the Phenomenal Layer. Current AI genuinely occupies the first. It cannot reach the second through more parameters, more data, or better training, because the gap between them is a phase transition, not a gradient. To show why, I examine what every living system from a slime mould to a human being possesses that no language model does, drawing on biology, neuroscience, thermodynamics, and physics. The paper closes with three frameworks specifying what closing this gap would actually require, grounded in the convergent insights of biologists Maturana and Varela, neuroscientist Antonio Damasio, and physicist Erwin Schrödinger, among others.