I prefer “just a full-text search engine with lossy compression”. Reducing it to the low-level ops is kind of a cop-out, especially when there is a serious analogue that we’ve been using for decades, and the key innovation here was basically just “what if we do it worse?”
I prefer “just a full-text search engine with lossy compression”. Reducing it to the low-level ops is kind of a cop-out, especially when there is a serious analogue that we’ve been using for decades, and the key innovation here was basically just “what if we do it worse?”
Search engine is just part of what LLMs (are starting to) do really well. Translation is another key aspect.