Prompt injection voor LLM's met lange context als alternatief voor RAG?

But isn’t this just a (really inefficient and) “static” version of RAG?

All RAG does differently from this approach is select and include relevant chunks of content instead of including all content.