Three recent academic papers explore the complex relationship between generative AI and linguistic diversity, particularly concerning World Englishes. The first paper discusses how AI tools can both democratize academic writing and marginalize minority English varieties, calling for equity-informed policies and inclusive co-design. The second paper examines how large language models (LLMs) reproduce dominant language ideologies, privileging Inner Circle norms and potentially challenging Global South English users, while also noting a paradox where AI might homogenize English yet pluralize it through diverse corpora. The third paper evaluates AI performance on African languages like Yoruba, Kinyarwanda, and Amharic, finding that while models excel on curated news data, they struggle with code-switched conversational data from platforms like Reddit, highlighting the need for models that can process both clean and mixed-language text. AI
IMPACT These papers highlight critical issues in AI development regarding linguistic equity and the potential for AI to either reinforce or challenge existing language hierarchies, urging for more inclusive design.
RANK_REASON The cluster consists of three academic papers published on arXiv discussing AI's impact on language diversity.
- Amharic
- English
- generative artificial intelligence
- GlotLID
- Hugging Face
- Kinyarwanda
- large-language models
- linguistic diversity
- linguistic inclusivity
- Llama 3.3-70B
- World Englishes
- Yoruba
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →
