Uncovering inequalities in new knowledge learning by large language models across different languages
Proceedings of the National Academy of Sciences · 23.12.2025 · 122 (51)
DOI: 10.1073/pnas.2514626122 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 18.12.2025. Verfasst von Chenglong Wang und 11 weiteren Autor:innen. Beteiligte Einrichtung: School of Urban Planning & Design, Peking University Shenzhen Graduate School, Shenzhen 518055, China. Hauptthemen laut MEDLINE: Language, Learning, Knowledge, Models, Theoretical. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Natural Science Foundation of China, Shenzhen Science and Technology Innovation Program, Guangdong Basic and Applied Basic Research Foundation.
Abstract (Originalsprache)
As large language models (LLMs) gradually demonstrate their potential to boost productivity and become integral tools for problem-solving in daily life worldwide, understanding the linguistic inequalities they introduce is becoming increasingly important. Prior research has primarily focused on static analyses of disparities in existing knowledge and capabilities of LLMs across languages. However, LLMs are continuously evolving, acquiring new knowledge to provide current, relevant responses and deliver precise, expert-level answers in specific domains. Investigating linguistic inequalities within this dynamic learning process is, therefore, also essential. In this paper, we explore inequalities in new knowledge learning by LLMs across different languages and four key dimensions: effectiveness, transferability, prioritization, and robustness. Through extensive experiments in both in-context learning and fine-tuning settings, with proprietary and open-source models, we reveal four key findings: 1) LLMs face greater challenges in efficiently and accurately learning new knowledge in lower-resource languages; 2) knowledge learned by LLMs tends to be more easily transferred to higher-resource languages than to lower-resource ones; 3) new knowledge in higher-resource languages is more likely to be retained and prioritized; and 4) LLMs are more robust against incorrect or misleading information in higher-resource languages. We further analyze the underlying causes of these inequalities from linguistic perspectives, pretraining characteristics, and tokenizer design, and propose a preliminary mitigation strategy through the lens of linguistic neurons. This work highlights the urgent need to recognize and address emerging linguistic inequalities in the development of LLMs.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Urban Planning & Design, Peking University Shenzhen Graduate School, Shenzhen 518055, China.
- Key Laboratory of Earth Surface System and Human-Earth Relations of Ministry of Natural Resources of China, Peking University Shenzhen Graduate School, Shenzhen 518055, China.
- Gaoling School of Artificial Intelligence, Renmin University of China, Beijing 100872, China.
- Department of Computer Science, University of Illinois Urbana-Champaign, Champaign, IL 61820.
- Department of Computer Science & Engineering, Hong Kong University of Science and Technology, Hong Kong 999077, China.
- Paul and Marcia Center on Contemporary China, Princeton University, Princeton, NJ 08544.
ORCID-Profile: Chenglong Wang, Yueqi Xie, Junming Huang, Yu Xie, Pengjun Zhao, Xing Xie, Fangzhao Wu
Förderung
- National Natural Science Foundation of China
- Shenzhen Science and Technology Innovation Program
- Guangdong Basic and Applied Basic Research Foundation
Zitieren
APA
Wang, C., Tang, H., Yang, X., Xie, Y., Suh, J., Sitaram, S., Huang, J., Xie, Y., Zhao, P., Gong, Z., Xie, X., & Wu, F. (2025). Uncovering inequalities in new knowledge learning by large language models across different languages. Proceedings of the National Academy of Sciences, 122(51). https://doi.org/10.1073/pnas.2514626122
Vancouver
Wang C, Tang H, Yang X, Xie Y, Suh J, Sitaram S, et al. Uncovering inequalities in new knowledge learning by large language models across different languages. Proceedings of the National Academy of Sciences. 2025;122(51). doi:10.1073/pnas.2514626122
BibTeX
@article{Wang2025,
author = {Wang, Chenglong and Tang, Haoyu and Yang, Xiyuan and Xie, Yueqi and Suh, Jina and Sitaram, Sunayana and Huang, Junming and Xie, Yu and Zhao, Pengjun and Gong, Zhaoya and Xie, Xing and Wu, Fangzhao},
title = {Uncovering inequalities in new knowledge learning by large language models across different languages},
journal = {Proceedings of the National Academy of Sciences},
year = {2025},
volume = {122},
number = {51},
doi = {10.1073/pnas.2514626122}
}
EUVIMED