Artificial Intelligence does not understand, at least not in the way humans mean the word understanding. This is not a criticism of AI’s capability, but a clarification of context. The confusion arises because we collapse multiple meanings of “understanding” into a single, unexamined concept. When AI performs well in language, reasoning, or problem-solving tasks, we intuitively project human comprehension onto it. But this projection ignores a critical distinction: understanding is not a monolith, but is stratified across contexts. Without contextual stratification, discussions about AI intelligence, alignment, and consciousness become incoherent. We argue past each other, using the same word while referring to fundamentally different phenomena.