English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recently, Large Language Models (LLMs) have enhanced user interaction,enabling seamless information retrieval and recommendations. However, concernsemerge as these LLMs have shown tendencies to display discrimination related tousers' sensitive characteristics (such as gender), leading to explicit userunfairness. Furthermore, our analysis uncovers a more discreet variant of biasin LLMs, defined as implicit user unfairness, wherein these models demonstratediscriminatory recommendation behaviors based solely on non-sensitive userdetails, like usernames or email addresses. This subtle form of unfairness,while more pervasive, poses a significant threat to the ethical integrity andrights of minority user groups. To comprehensively explore implicit userunfairness, our analysis unfolds in three key steps: (1) We uncover the reasonsfor this implicit user unfairness: LLMs can infer users' sensitive attributesfrom non-sensitive attributes (e.g. user names) due to their extensive worldknowledge. (2) Our findings expose that the magnitude of implicit userunfairness within LLMs surpasses the level of explicit user unfairness observedin traditional recommender models, signifying a more alarming issue ofunfairness, i.e. some non-sensitive features of users like names may result inmore serious discrimination phenomena. (3) We analyze the long-term effect ofimplicit user unfairness, identifying that it will reinforce informationbubbles at an accelerated rate compared to traditional RS. We emphasize theneed to identify and mitigate implicit user unfairness, aiming to avert thepotential human-LLMs recommendation systems deterioration.

Do LLMs Implicitly Exhibit User Discrimination in Recommendation? An Empirical Study