From Access to Outcomes: Generative AI, Digital Inequality, and Career Development in Vocational Higher Education

• In Event: Evaluating Equity, Access, and Outcomes in Workforce and Technical Education

Abstract

Generative AI is rapidly entering education and career preparation, but its benefits are unlikely to be distributed evenly. For students in vocational higher education, who often face disadvantages in socioeconomic resources, digital skills, and institutional support, AI may function both as a tool for capability development and as a mechanism that reproduces inequality. This study examines how generative AI use reshapes educational equity and school-to-work transitions among vocational college students, with particular attention to who uses AI, how it is used, and what developmental and labor-market outcomes follow.The analysis builds on the digital divide literature, which conceptualizes inequality in technology through differences in access, usage, and outcomes. Extending this framework to generative AI, the study argues that an emerging “intelligence divide” lies not simply in access to tools, but in the capacity to use them strategically for learning and career advancement. To capture this process, the paper develops a two-dimensional framework of AI use organized by task orientation and enablement mechanism. Task orientation distinguishes learning-related uses from career-related uses, while enablement mechanism distinguishes cost reduction from barrier reduction. This yields four analytically distinct use patterns: learning efficiency enhancement, learning innovation and knowledge transfer, career preparation and information acquisition, and career capability expansion and skill leapfrogging. These patterns provide a way to connect AI engagement to capability formation and employment outcomes.Empirically, the study uses a large-scale national graduate employment survey covering 35,686 vocational college graduates. The dataset includes measures of AI use frequency, duration, and usage scenarios, along with indicators of digital skills, self-reported capability development, employment outcomes, and socioeconomic background. Fixed-effects regression models are used to estimate the relationships between AI use patterns, digital skill development, and employability, while interaction terms test heterogeneity across student groups.The findings reveal a stratified pattern of generative AI use consistent with a three-level digital divide. At the access level, AI engagement varies systematically by both individual and structural characteristics. Students from urban backgrounds, stronger academic tracks, and more advantaged family settings show higher overall usage intensity, while female students tend to use AI more selectively but with greater depth. At the usage level, both the breadth and continuity of AI engagement are positively associated with digital skill development. However, not all forms of use produce the same returns. Diversified and task-aligned use is more strongly associated with employability than simple usage frequency alone. At the outcome level, generative AI reduces information costs and lowers barriers to complex tasks in both learning and career preparation, but unequal access translates into unequal usage patterns and, ultimately, unequal outcomes.The study contributes to current debates on AI and education by shifting attention from general adoption to stratified use and differentiated returns. Its policy implication is that equitable AI integration requires more than tool access. Vocational institutions need to embed AI literacy, guided use strategies, and career-oriented applications into curricula so that generative AI expands opportunity rather than deepening existing inequality in the transition from school to work.

Author

• yaxuan liu, Peking University
  Presenting Author