# Aligning Large Language Model Behavior with Human Citation Preferences

arXiv: https://arxiv.org/abs/2602.05205 · HTML: https://mlsanity.com/paper/2602.05205/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 16 of its 24 references that are in the map.
- First posted on arXiv 5 Feb 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Human values and AI feedback alignment](https://mlsanity.com/alignment/human-values-and-ai-feedback-alignment/)

## Builds on (most cited first, 16 of 16)

- [Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection](https://mlsanity.com/paper/2310.11511/) (2023)
- [OpenAI GPT-5 System Card](https://mlsanity.com/paper/2601.03267/) (2025)
- [Enabling Large Language Models to Generate Text with Citations](https://mlsanity.com/paper/2305.14627/) (2023)
- [ZeRO-Offload: Democratizing Billion-Scale Model Training](https://mlsanity.com/paper/2101.06840/) (2021)
- [Evaluating Verifiability in Generative Search Engines](https://mlsanity.com/paper/2304.09848/) (2023)
- [PaperQA: Retrieval-Augmented Generative Agent for Scientific Research](https://mlsanity.com/paper/2312.07559/) (2023)
- [Structural Scaffolds for Citation Intent Classification in Scientific Publications](https://mlsanity.com/paper/1904.01608/) (2019)
- [LongCite: Enabling LLMs to Generate Fine-grained Citations in Long-context QA](https://mlsanity.com/paper/2409.02897/) (2024)
- [Citations and Trust in LLM Generated Responses](https://mlsanity.com/paper/2501.01303/) (2025)
- [CiteWorth: Cite-Worthiness Detection for Improved Scientific Document Understanding](https://mlsanity.com/paper/2105.10912/) (2021)
- [Citation Needed: A Taxonomy and Algorithmic Assessment of Wikipedia's Verifiability](https://mlsanity.com/paper/1902.11116/) (2019)
- [CitaLaw: Enhancing LLM with Citations in Legal Domain](https://mlsanity.com/paper/2412.14556/) (2024)
- [Anchored Answers: Unravelling Positional Bias in GPT-2's Multiple-Choice Questions](https://mlsanity.com/paper/2405.03205/) (2024)
- [Adaptive Question Answering: Enhancing Language Model Proficiency for Addressing Knowledge Conflicts with Source Citations](https://mlsanity.com/paper/2410.04241/) (2024)
- [Abstention vs. Hallucination: Benchmarking LLM Source Attribution for Scientific Citations](https://mlsanity.com/paper/2405.02228/) (2024)
- [WikiSQE: A Large-Scale Dataset for Sentence Quality Estimation in Wikipedia](https://mlsanity.com/paper/2305.05928/) (2023)

## Built on by (most cited first, 1 of 1)

- [On the Role of Citations in Preference Data](https://mlsanity.com/paper/2608.21376/) (2026)

## Abstract (arXiv)

Most services built on powerful large-scale language models (LLMs) add citations to their output to enhance credibility. Recent research has paid increasing attention to the question of what reference documents to link to outputs. However, how LLMs recognize cite-worthiness and how this process should be controlled remains underexplored. In this study, we focus on what kinds of content LLMs currently tend to cite and how well that behavior aligns with human preferences. We construct a dataset to characterize the relationship between human citation preferences and LLM behavior. Web-derived texts are categorized into eight citation-motivation types, and pairwise citation preferences are exhaustively evaluated across all type combinations to capture fine-grained contrasts. Our results show that humans most frequently seek citations for medical text, and stronger models display a similar tendency. We also find that current models are as much as 27% more likely than humans to add citations to text that is explicitly marked as needing citations on sources such as Wikipedia, and this overemphasis reduces alignment accuracy. Conversely, models systematically underselect numeric sentences (by -22.6% relative to humans) and sentences containing personal names (by -20.1%), categories for which humans typically demand citations. Furthermore, experiments with Direct Preference Optimization demonstrate that model behavior can be calibrated to better match human citation preferences. We expect this study to provide a foundation for more fine-grained investigations into LLM citation preferences.