How Google Scrapes the Internet to Find Relevant Links

Google 1.0 and Google Classic, as I guess we'd call it, scrapes the internet to find relevant links and then sort of ranks them and builds them versus what these AI systems are doing. It takes a preexisting human-built database that they keep adding material to. But it doesn't go out and just hunt around to find links that exist. Just takes billions and billions of words, billions and billion of bits of data and builds a training model to link them together into something that seems like perceptible speech. And oddly enough, it works.