LLMs have, for me at least, mostly replaced search. Because one of the most common things I search for is stuff like 'what is the function in [x] that does [y] in [z]'. Search for that on Google (or any other search engine) and your results will be complete trash. Search for it on an LLM and it's one of the few things they're nearly perfect at.

Search engines, especially with the resources of Google, could have developed at least basically functional natural language search decades ago. Instead they chose to rest on their laurels (pagerank even still plays a significant role!), and secure market control by anti-competitive behavior.

Google is certainly well on their way to becoming another Yahoo, but there's no extrinsic reason for it in the same way there was no reason for Yahoo to become what it became. In 100 years people will be searching the corpus of human writings (or whatever medium is dominant then) for information that's relevant to them. So there's in intrinsic reason for any search related company to just naturally die. It simply needs to keep innovating, or at least evolutionarily improving.