UNESCO has issued a new warning about the potential risks of bias in AI-driven language learning systems. The report highlights that many AI models are trained predominantly on English-language data from specific cultural contexts, which may lead to biased outputs and limited representation of global English varieties. This has implications for learners developing intercultural communicative competence. The organization calls for more inclusive datasets and transparent AI systems that reflect linguistic diversity. In ELT, this is particularly critical, as English is used globally in diverse contexts. Teachers are encouraged to act as mediators, helping students critically evaluate AI-generated language and understand its limitations. The report reinforces the need for ethical guidelines and responsible AI integration in language education.
UNESCO Highlights Risks of AI Bias in Language Education
