# Training-Induced Bias Toward LLM-Generated Content in Dense Retrieval

arXiv: https://arxiv.org/abs/2602.10833 · HTML: https://mlsanity.com/paper/2602.10833/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 23 references that are in the map.
- First posted on arXiv 11 Feb 2026.

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

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [MS MARCO: A Human Generated MAchine Reading COmprehension Dataset](https://mlsanity.com/paper/1611.09268/) (2016)
- [Text Embeddings by Weakly-Supervised Contrastive Pre-training](https://mlsanity.com/paper/2212.03533/) (2022)
- [Unsupervised Dense Information Retrieval with Contrastive Learning](https://mlsanity.com/paper/2112.09118/) (2021)
- [Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval](https://mlsanity.com/paper/2007.00808/) (2020)
- [DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature](https://mlsanity.com/paper/2301.11305/) (2023)
- [MTEB: Massive Text Embedding Benchmark](https://mlsanity.com/paper/2210.07316/) (2022)
- [Fast-DetectGPT: Efficient Zero-Shot Detection of Machine-Generated Text via Conditional Probability Curvature](https://mlsanity.com/paper/2310.05130/) (2023)
- [Efficiently Teaching an Effective Dense Retriever with Balanced Topic Aware Sampling](https://mlsanity.com/paper/2104.06967/) (2021)
- [SciFact-Open: Towards open-domain scientific claim verification](https://mlsanity.com/paper/2210.13777/) (2022)
- [Neural Retrievers are Biased Towards LLM-Generated Content](https://mlsanity.com/paper/2310.20501/) (2023)
- [Collapse of Dense Retrievers: Short, Early, and Literal Biases Outranking Factual Evidence](https://mlsanity.com/paper/2503.05037/) (2025)
- [Writing Style Matters: An Examination of Bias and Fairness in Information Retrieval Systems](https://mlsanity.com/paper/2411.13173/) (2024)
- [Invisible Relevance Bias: Text-Image Retrieval Models Prefer AI-Generated Images](https://mlsanity.com/paper/2311.14084/) (2023)
- [BERM: Training the Balanced and Extractable Representation for Matching to Improve Generalization Ability of Dense Retrieval](https://mlsanity.com/paper/2305.11052/) (2023)
- [Perplexity Trap: PLM-Based Retrievers Overrate Low Perplexity Documents](https://mlsanity.com/paper/2503.08684/) (2025)
- [EqualizeIR: Mitigating Linguistic Biases in Retrieval Models](https://mlsanity.com/paper/2504.07115/) (2025)
- [AugTriever: Unsupervised Dense Retrieval and Domain Adaptation by Scalable Data Augmentation](https://mlsanity.com/paper/2212.08841/) (2022)
- [Generative Ghost: Investigating Ranking Bias Hidden in AI-Generated Videos](https://mlsanity.com/paper/2502.07327/) (2025)
- [MultiContrievers: Analysis of Dense Retrieval Representations](https://mlsanity.com/paper/2402.15925/) (2024)

## Abstract (arXiv)

Dense retrieval is a promising approach for acquiring relevant context or world knowledge in open-domain natural language processing tasks and is now widely used in information retrieval applications. However, recent reports claim a broad preference for text generated by large language models (LLMs). This bias is called "source bias", and it has been hypothesized that lower perplexity contributes to this effect. In this study, we revisit this claim by conducting a controlled evaluation to trace the emergence of such preferences across training stages and data sources. Using parallel human- and LLM-generated counterparts of the SciFact and Natural Questions (NQ320K) datasets, we compare unsupervised checkpoints with models fine-tuned using in-domain human text, in-domain LLM-generated text, and MS MARCO. Our results show the following: 1) Unsupervised retrievers do not exhibit a uniform pro-LLM preference. The direction and magnitude depend on the dataset. 2) Across the settings tested, supervised fine-tuning on MS MARCO consistently shifts the rankings toward LLM-generated text. 3) In-domain fine-tuning produces dataset-specific and inconsistent shifts in preference. 4) Fine-tuning on LLM-generated corpora induces a pronounced pro-LLM bias. Finally, a retriever-centric perplexity probe involving the reattachment of a language modeling head to the fine-tuned dense retriever encoder indicates agreement with relevance near chance, thereby weakening the explanatory power of perplexity. Our study demonstrates that source bias is a training-induced phenomenon rather than an inherent property of dense retrievers.