I have never typed a complete sentence into Google and am still very satisfied with the quality of Google search. But... your comment made me realize, a good u

I have never typed a complete sentence into Google and am still very satisfied with the quality of Google search. But... your comment made me realize, a good use of LLMs with search would be not to ask them directly, but to use them as a front: detect what the user wants and devise the perfect Google search to find it. LLMs should be very good at discriminating between a plain error message, a business name, or a general open question, building an excellent Google search for it, and parse and interpret the results. It seems that's what Perplexity is doing, mostly. Personally I find it slower and more cumbersome than using Google Search directly, but maybe that's the direction we're all going. Machines to help us use machines.