Social class,generative AI,and the negotiation of unequal digital literacies

Abstract

Focusing on two Chinese female university students from contrasting social class positions, Chen and Natalia, this article explores how engaging with generative artificial intelligence (AI) tools built on large language models constitutes social practice that can reproduce inequalities in language learning. Drawing on Darvin and Norton’s model of investment, we examine how these learners invest in AI literacy practices (as a specific type of digital literacies) shaped by differing dispositions toward technology and unequal access to material and symbolic resources. Data were collected through interviews, questionnaires, and digital artifacts, and analyzed using NVivo 12 following an inductive, case-based approach. Findings reveal that Chen, a middle-class student with extensive access to cultural, material, and social capital, was able to bypass barriers to entry and critically engage with ChatGPT. She agentively crafted targeted prompts and personalized chatbots, investing in the identity of a competent second language (L2) learner and AI user. In contrast, Natalia, a student from a small town in rural China, expressed a strong interest in AI but could only access 灵心AI (Lingxin AI), a third-party WeChat application powered by the ChatGLM model. Negotiating the material constraints of platforms, screen dimensions, memory capacity, and processing power, the two learners invested in contrasting digital literacies that ultimately offered unequal opportunities for English language learning. By shedding light on how social class creates unequal conditions for AI literacy practices, this study contributes to ongoing conversations about digital L2 learning, equity, and inclusion in an era of rapid technological change.