Nick Koudas

Nick Koudas is a professor of computer science at the University of Toronto. With Mahe Chen, Xiaoxuan Wang and Kaiwen Chen, he compared AI search systems with Google across verticals, languages and phrasings.

The study's practical punchline is about sources. AI engines in their analysis relied overwhelmingly on earned media over brand-owned and social content. They also favored large brands in ways that smaller companies have to work deliberately to overcome. Those findings are in the 2025 comparative paper.

Those two findings travel well into commercial GEO briefings because they are actionable. If earned media dominates, PR and third-party proof are not optional garnish. If large brands get a default boost, challengers need a plan that is more than a homepage rewrite.

The comparative design across languages and phrasings also undercuts the excuse that a single English prompt sample explains the whole channel. Teams that only test a handful of domestic queries are measuring a slice of the behavior Koudas and coauthors mapped more carefully.

For agencies and brand teams, the Koudas-led work is one of the strongest academic cases for treating third-party corroboration as central to generative engine optimization. If AI answers lean on journalism, reviews and other external sources, then authority work sits at the center of the program. That finding rhymes with how builders such as Jason Barnard talk about corroborating brand facts.

The paper also gives researchers a comparative frame: AI search is not simply Google with a chat UI. Source mix and brand bias differ enough that teams need separate measurement. That point alone has changed how some analytics teams design their AI visibility reports.

As an architect on the TGA 20, Koudas represents comparative evidence at academic rigor, which practitioners can cite when a stakeholder asks why earned media budgets belong next to content budgets.

Favoritism toward large brands in AI answers is an especially hard finding for mid-market clients to hear, which is why having it in an academic comparative study helps agencies set expectations early.

Start with "Generative Engine Optimization: How to Dominate AI Search" (2025). Read it for the source-mix findings, then decide what your own brand evidence would look like under the same lens.

Academic comparative work also travels into legal and policy conversations about AI search bias. When earned media dominates and big brands win by default, the GEO advice for smaller firms has to be more than hope.