Forty-six percent of high school students now use AI tools like ChatGPT during their college search, up from 26% just months earlier — and 18% have already dropped a school from consideration based on what an AI told them, according to a 2026 EAB survey of over 5,000 students. Journalists and researchers show the same pattern: Perplexity alone processes roughly 50 million weekly queries, concentrated among exactly the people who'd otherwise call a university press office to ask "who studies this."
The uncomfortable part for university marketing teams: being visible on Google doesn't mean being visible to AI. Fewer than 10% of sources cited by ChatGPT, Gemini and Copilot rank in the Google top 10 for the same query — the overlap between what ranks and what gets cited has fallen from roughly 70% to under 20%. And 84% of AI citations trace back to earned, third-party coverage rather than a university's own pages, which makes one thing clear: a faculty member's expertise has to exist somewhere structured and crawlable before any AI system can cite it at all.
A PDF CV is invisible to a language model
Most university faculty pages are still built the way they were fifteen years ago: a directory listing, a linked CV in PDF, maybe a personal homepage nobody has updated since 2019. None of that is structured data. A crawler — or an AI model retrieving a page to answer "who at this university researches battery chemistry" — needs a clean, indexable page per professor: name, department, research areas, and an actual list of publications, not a PDF attachment.
That's the entire premise behind FanProfile, our public faculty portal. Every professor gets a real profile page — photo, biography, academic background, and a structured publication list — organized by department and searchable by research field, served through a REST API rather than a static PDF archive. It doesn't fix earned-media citations on its own, but it fixes the prerequisite: the university's own research expertise becomes something a crawler, and eventually an AI answer engine, can actually read and attribute correctly. Built and running for Lorestan University's faculty, and available to any institution facing the same gap.
If a prospective student, journalist, or another researcher asked an AI assistant right now who at your university works on their topic of interest, would there be a page it could point to?
