> The problem we now face, which is being made much worse by AI, is filtering things, as we get blasted by thousands of “relevent” search results

Especially with a malicious-trained model, there can just be so much content that seems about right that you can't really trust anything.