It is AI generated. Or was written by someone a bit far from the technical advances IMHO. The Johnson-Lindenstrauss Lemma is a very specific and powerful concept, when in the article the QLJ explanation is vacuous. A knowledgeable human would not have left the reader wanting for how that relates to the Lemma.
Honestly, the bigger miss is people treating JL as some silver bullet for "extreme" compression, as if preserving pairwise distances for a fixed point set somehow means you still keep the task-relevant structure once you're dealing with modern models.
Try projecting embeddings this way and watch your recall crater the moment you need downstream task performance instead of nearest-neighbor retreival demos. If you're optimizing for blog post vibes instead of anything measurable sure, call it a breakthrough.