Does Reasoning Make Search More Fair? Comparing Fairness in Reasoning and Non-Reasoning Rerankers

ArXi:2603.10332v1 Announce Type: cross While reasoning rerankers, such as Rank1, have nstrated strong abilities in improving ranking relevance, it is unclear how they perform on other retrieval qualities such as fairness. We conduct the first systematic comparison of fairness between reasoning and non-reasoning rerankers. Using the TREC 2022 Fair Ranking Track dataset, we evaluate six reranking models across multiple retrieval settings and graphic attributes. Our findings nstrate reasoning neither improve nor harm fairness compared to non-reasoning approaches.