Artificial intelligence is often discussed in terms of benchmarks, accuracy metrics, or technological leaps. But these measures don’t always capture how AI relates to human cognition. To help frame this relationship, I propose a “The Human-AI Augmentation Spectrum” - a way of mapping AI systems according to the kinds of human expertise they replicate or exceed. This is not a definitive taxonomy, but an interpretive lens: a subjective model designed to spark discussion, test against examples, and evolve as our understanding of AI deepens. On this spectrum, AI moves from performing basic non-expert tasks, to augmenting expert-level judgment, to reaching domains of cognition beyond human capacity. Framing AI in this way highlights not only its capabilities, but also its limitations — especially the bottlenecks of actionability and interface that determine real-world utility.