Can Fairness Be Prompted? Prompt-Based Debiasing Strategies in High-Stakes Recommendations
This addresses fairness issues in high-stakes recommendations for users, offering a lightweight and accessible debiasing method, but it is incremental as it builds on existing debiasing approaches.
The study tackled the problem of implicit biases in Large Language Model Recommenders (LLMRecs) by exploring prompt-based debiasing strategies, showing that instructing LLMs to be fair can improve fairness by up to 74% while maintaining comparable effectiveness, though it sometimes overpromotes specific demographic groups.
Large Language Models (LLMs) can infer sensitive attributes such as gender or age from indirect cues like names and pronouns, potentially biasing recommendations. While several debiasing methods exist, they require access to the LLMs' weights, are computationally costly, and cannot be used by lay users. To address this gap, we investigate implicit biases in LLM Recommenders (LLMRecs) and explore whether prompt-based strategies can serve as a lightweight and easy-to-use debiasing approach. We contribute three bias-aware prompting strategies for LLMRecs. To our knowledge, this is the first study on prompt-based debiasing approaches in LLMRecs that focuses on group fairness for users. Our experiments with 3 LLMs, 4 prompt templates, 9 sensitive attribute values, and 2 datasets show that our proposed debiasing approach, which instructs an LLM to be fair, can improve fairness by up to 74% while retaining comparable effectiveness, but might overpromote specific demographic groups in some cases.