Whose English Does Generative AI Improve? Algorithmically Mediated Unequal Englishes in the Workplace

Generative artificial intelligence (GenAI) is frequently presented as a linguistic equalizer in workplaces where English mediates recruitment, collaboration, documentation, and customer interaction. Productivity research partly supports this claim, particularly for less-experienced workers. This conceptual piece argues that such gains do not constitute linguistic equality and that, for business communication scholarship, the key issue is how GenAI redistributes communicative authority, adaptation, and credibility within organizations. I call this process Algorithmically Mediated Unequal Englishes (AMUE). It operates through four connected mechanisms: normative compression into AI-polished English, unequal linguistic and affective labor, automated misrecognition and screening, and a credibility trap surrounding authorship and disclosure. These mechanisms offer resources while preserving institutional ‘standards’ and listener expectations. Linguistically equitable AI governance must therefore move beyond providing access to tools. It must redefine effective communication through task relevance, intelligibility, accuracy, reciprocal accommodation, worker control over voice, and institutional responsibility for technological risk.