It’s interesting to think about an information-deduplication solution for internet. Maybe now we can use LLMs and embeddings to create an archive of all (factual) information, then build a search on top of that?