Large language models show bias in judging ad relevance across identities
A Systematic Investigation of Bias in Large Language Models for Advertising Relevance
Artificial Intelligence
Summary
Large language models (LLMs) are used to decide how well ads match search queries, but their fairness is unclear. The authors show that changing the advertiser's identity or the language of the query can change the model's relevance score. They also found that the models sometimes reflect common stereotypes, especially involving gender and jobs. Their work suggests ways to spot and reduce these fairness issues in ad systems.
What this means in practice
- •For advertising technology teams: Improve fairness auditing and adjustments in AI systems that match ads to queries, by considering advertiser identity and language effects on relevance judgment.
- •For natural language processing engineers: Develop and test bias mitigation strategies during inference and training of language models to reduce stereotype-driven relevance errors in ad matching.
Authors
Weiwei Wang, Yinchuan Xu, Jialu Gao, Youkow Homma, Jian Jiao
Abstract
Large language models (LLMs) are increasingly used to judge how well an advertisement matches a query, but the fairness of these judgments has received limited attention. We conduct a systematic study of fairness in relevance judgments made by LLMs for queries and advertisements. Our counterfactual framework examines the effects of advertiser identity and possible popularity, input language, and demographic wording. We study GPT-4o as a categorical relevance judge and a Qwen-7B model trained specifically for relevance prediction. The advertiser and language experiments use query and advertisement pairs sampled from real advertising logs. Controlled synthetic queries are used to study demographic associations in employment, housing, and credit. For both models, changing the advertiser identity or input language can alter the relevance assessment. Selected demographic comparisons also show patterns consistent with common stereotypes, particularly those involving gender and occupation. We further study mitigation during model inference and training. The results indicate that its effectiveness depends on whether advertiser information is relevant to the query and how advertiser labels are distributed in the training data. These findings can help advertising practitioners identify fairness risks and develop suitable mitigation methods for LLM relevance systems.