Wikipedia in the Era of LLMs: Evolutions and Risks

ACL ARR 2025 February Submission1444 Authors

Abstract: Wikipedia helps both people and machines seeking knowledge about the world. In this paper, we present a thorough analysis of the influence of Large Language Models (LLMs) on Wikipedia, examining both human and machine perspectives. We begin by analyzing page views and article content to study Wikipedia’s recent evolutions and assess the impact of LLMs. Subsequently, we examine how LLMs affect various Natural Language Processing (NLP) tasks related to Wikipedia, including machine translation and retrieval-augmented generation. Our findings and simulation results reveal that while LLMs have not yet fully permeated Wikipedia’s language and knowledge structures, their current influence is significant enough to warrant careful consideration of potential future risks.

Paper Type: Long

Research Area: Computational Social Science and Cultural Analytics

Research Area Keywords: Wikipedia, Large Language Model, Natural Language Generation, Chronological Analysis, RAG, Document Readability, Wold Frequency

Contribution Types: NLP engineering experiment, Data analysis

Languages Studied: English

Submission Number: 1444