Toward an AI-Native Internet: Rethinking the Web Architecture for Semantic Retrieval
Muhammad Bilal · Zafar Qazi · Marco Canini
King Abdullah University of Science and Technology (KAUST)
Serving semantically chunked units instead of full HTML documents delivers statistically equivalent answer quality at 13–19% of the data — a 74–87% bandwidth reduction — pointing toward a web where meaning, not documents, is the fundamental unit of retrieval.
cs.NI
cs.IR
cs.AI
RAG Infrastructure
Semantic Retrieval