Under the hood
What happens when someone asks for the best university for AI
Engines differ in the details, but the shape of the process is similar: retrieve, synthesize, cite.
Modern assistants combine two things: what the model absorbed during training, and what it retrieves live from the web when a question benefits from fresh information. For a question about universities, most engines run searches in the background, read a set of pages, and compose an answer from them.
Which pages get read is where rankings still matter. Ranking and comparison sites are heavily represented in the sources engines consult because they are structured, comparative, and regularly updated. But they now sit alongside sources league tables never had to compete with: forum threads, student discussions, subject-specific blogs, and recent news.
Citation behavior varies by engine. Perplexity and Google's AI Overviews attach visible source links to nearly everything they say, while ChatGPT and Gemini cite when they browse, and other engines lean on retrieval to different degrees depending on the question. The practical consequence is the same everywhere: a small set of third-party pages ends up speaking for your institution.