Tag
The author argues that discomfort with AI writing stems from shifts in knowledge-based value and identity, but emphasizes that judgment, taste, and direction remain valuable human traits.
This paper evaluates whether multilingual sentence embeddings can replace translation for linguistic-integrated reliability auditing across multiple languages in educational assessments, finding that native-language embeddings reproduce translation-based reliability estimates closely.