LLM vs traditional search

A traditional search engine finds; an LLM composes. Type a query into a search box and you get a ranked list of links pointing to pages other people wrote — you still do the reading and synthesizing. Ask an LLM the same thing and it writes a single, direct answer in its own words, saving you the click-through but also hiding where the information came from.

The difference runs deep. Search retrieves existing documents and ranks them by relevance, so its answer lives somewhere you can check. An LLM generates new text from patterns in its training, so there is no source page behind a given sentence, and it can sound confident while being wrong. Many modern systems combine the two: the model searches first, then writes an answer grounded in what it found.