The Empire Strikes Back: What Discourses Do Language Education Platforms Reproduce with Their AI-powered Technologies​

Recently, the laudatory views in favor of the rapid integration of artificial intelligence (AI) in language education have been challenged (Dovchin, 2025; Moorhouse, 2024) as cautious voices highlighting critical concerns have been raised, including human agency (Peshl, 2024), AI bias (Pack & Maloney, 2024), or equitable access (Celik, 2023). However, how platformized ELT ecosystems utilize AI-powered technologies has been missing in this critical scholarship. Defined as a platform-driven, web-based, private, for-profit spaces that enable (a)synchronous communication, complementing, replacing, or reconceptualizing the traditional language classrooms, teachers, and practices therein (Selvi, 2025), these ecosystems include AI-powered language education platforms. As these platforms race to brand themselves as AI-powered, they generate discourses that may have a detrimental impact, further aggravating the problem of equity and fairness in the AI-driven world. To address this issue, we explore the discourses (re)produced in the ads of such platforms. Drawing on the advertisement data openly available on Meta Ad Library, we have examined the ads with the keywords “language learning” and “artificial intelligence” broadcast on Facebook and Instagram for six months in Turkiye. Using critical multimodal discourse analysis (Van Leeuwen, 2013), we have found that some platforms illustrate non-native English speakers as clownish subjects-in-need, who might find themselves in desperate situations (e.g., unemployed, shamed, or even in hunger) if they do not learn English in native accents, which is a discourse reproduced and assetized by these platforms utilizing AI that provide instant feedback on “perfect” or “proper” American-English thanks to its training data. Moreover, as AI is used by those platforms to detect so-called mistakes that might be related to learners’ linguistic background, the platforms promote a deficit-based language education approach. Our findings indicate that the recent developmental trajectory of AI technologies fosters an alarming approach that reproduces neoliberal and colonial discourses in language education.