# Revealing Potential Biases in LLM-Based Recommender Systems in the Cold Start Setting

arXiv: https://arxiv.org/abs/2508.20401 · HTML: https://mlsanity.com/paper/2508.20401/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 30 references that are in the map.
- First posted on arXiv 28 Aug 2025.

## Builds on (most cited first, 21 of 21)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [BBQ: A Hand-Built Bias Benchmark for Question Answering](https://mlsanity.com/paper/2110.08193/) (2021)
- [ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection](https://mlsanity.com/paper/2203.09509/) (2022)
- [Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5)](https://mlsanity.com/paper/2203.13366/) (2022)
- [TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation](https://mlsanity.com/paper/2305.00447/) (2023)
- [Towards Understanding and Mitigating Social Biases in Language Models](https://mlsanity.com/paper/2106.13219/) (2021)
- [Is ChatGPT Fair for Recommendation? Evaluating Fairness in Large Language Model Recommendation](https://mlsanity.com/paper/2305.07609/) (2023)
- [Beyond Parity: Fairness Objectives for Collaborative Filtering](https://mlsanity.com/paper/1705.08804/) (2017)
- [Multisided Fairness for Recommendation](https://mlsanity.com/paper/1707.00093/) (2017)
- [Large Language Models are Competitive Near Cold-start Recommenders for Language- and Item-based Preferences](https://mlsanity.com/paper/2307.14225/) (2023)
- [Certifying Counterfactual Bias in LLMs](https://mlsanity.com/paper/2405.18780/) (2024)
- [Unveiling Bias in Fairness Evaluations of Large Language Models: A Critical Literature Review of Music and Movie Recommendation Systems](https://mlsanity.com/paper/2401.04057/) (2024)
- [Unveiling and Mitigating Bias in Large Language Model Recommendations: A Path to Fairness](https://mlsanity.com/paper/2409.10825/) (2024)
- [Ethics and Governance of Artificial Intelligence: Evidence from a Survey of Machine Learning Researchers](https://mlsanity.com/paper/2105.02117/) (2021)
- [Text2Playlist: Generating Personalized Playlists from Text on Deezer](https://mlsanity.com/paper/2501.05894/) (2025)
- [Unmasking Gender Bias in Recommendation Systems and Enhancing Category-Aware Fairness](https://mlsanity.com/paper/2502.17921/) (2025)
- [FilterLLM: Text-To-Distribution LLM for Billion-Scale Cold-Start Recommendation](https://mlsanity.com/paper/2502.16924/) (2025)

## Built on by (most cited first, 6 of 6)

- [Prominence-Stratified Failure Modes in Retrieval-Augmented Commercial Recommendation: A 37,000-Run Audit](https://mlsanity.com/paper/2605.27439/) (2026)
- [HELM: A Human-Centered Evaluation Framework for LLM-Powered Recommender Systems](https://mlsanity.com/paper/2601.19197/) (2026)
- [Combating data scarcity in recommendation services: Integrating cognitive types of VARK and neural network technologies (LLM)](https://mlsanity.com/paper/2603.03309/) (2026)
- [An Investigation of Linguistic Biases in LLM-Based Recommendations](https://mlsanity.com/paper/2604.25456/) (2026)
- [Rethinking Fairness in LLM-Based Recommender Systems: A Survey](https://mlsanity.com/paper/2606.28340/) (2026)
- [Critical Acclaim Orientation in Large Language Models: Evidence from Film Preference Elicitation](https://mlsanity.com/paper/2608.06955/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) are increasingly used for recommendation tasks due to their general-purpose capabilities. While LLMs perform well in rich-context settings, their behavior in cold-start scenarios, where only limited signals such as age, gender, or language are available, raises fairness concerns because they may rely on societal biases encoded during pretraining. We introduce a benchmark specifically designed to evaluate fairness in zero-context recommendation. Our modular pipeline supports configurable recommendation domains and sensitive attributes, enabling systematic and flexible audits of any open-source LLM. Through evaluations of state-of-the-art models (Gemma 3 and Llama 3.2), we uncover consistent biases across recommendation domains (music, movies, and colleges) including gendered and cultural stereotypes. We also reveal a non-linear relationship between model size and fairness, highlighting the need for nuanced analysis.