From Access to Realized Affordances: University Students' Generative AI Engagement across Linguistic and Sociotechnical Contexts

TEXT START: Generative artificial intelligence (GenAI) is increasingly embedded in university students' academic work, yet student engagement is often examined through adoption, frequency of use, or general perceptions, with less attention to how it is shaped by linguistic and sociotechnical conditions.

The Dissection

This is an interface study presented as a theory of educational possibility. It maps how 42 students in China, Japan, and Mongolia convert nominal AI access into usable academic output under language, platform, privacy, and institutional constraints.

Its strongest contribution is descriptive: access is not binary, and linguistic friction changes what the same model can practically do. But the study remains confined to the user layer. It does not measure learning, durable skill formation, labor-market value, wage effects, ownership of AI capital, or displacement of human work. “Realized affordances” makes automation legible as a pedagogical usability problem while leaving the ownership structure untouched.

The Core Fallacy

The paper mistakes successful use for preserved human economic relevance.

Under the Discontinuity Thesis, realized affordance is only the last-mile variable. P1 makes cognitive automation cheaper and more capable. P2 prevents institutions from preserving stable human-only domains at scale. P3 removes the majority from economically necessary labor. Whether a student can obtain a better answer in Mongolian, Japanese, or English changes access friction; it does not preserve the mass employment-to-wage-to-consumption circuit.

The paper also implicitly treats improved academic engagement as improved educational value. That link is unproven. Students may become more effective at producing acceptable academic artifacts while the artifacts, and eventually much of the credentialed labor behind them, become cheaper and less scarce. The machine is being optimized for use while the human role it replaces is quietly losing market necessity.

Hidden Assumptions

• Universities will remain the main mechanism converting academic activity into scarce economic value.
• GenAI use will translate into genuine learning rather than increasingly efficient output production.
• Human evaluation, critical thinking, and linguistic mediation will remain durable differentiators rather than tasks progressively absorbed by better models.
• Language and platform barriers will remain meaningful moats instead of temporary lag defenses compressed by translation, localization, and model improvement.
• Student judgments about privacy, dependence, and critical thinking can be treated as manageable design variables rather than symptoms of a deeper transfer of agency to AI systems.
• Three universities and 42 interviews can support a broad framework about higher education, despite providing contextual patterns rather than causal or population-level evidence.
• The central inequality is unequal access to AI. Under DT logic, the decisive inequality is control of AI capital and the ability to remain indispensable to those who control it.

Social Function

The paper is partial truth packaged as transition management and ideological anesthetic.

It accurately records the early unevenness of AI access and provides institutions with a vocabulary for improving adoption across linguistic contexts. That is useful operationally. Systemically, however, it redirects attention from ownership, substitution, and productive participation toward better mediation and more inclusive use. It helps universities manage the arrival of the machine without asking whether the university’s mass credential-and-employment function survives the machine.

The Verdict

The study maps friction around the automating system; it does not challenge the system’s trajectory. Its “realized affordances” are transitional adjustments, not a defense against P1-P3. Linguistic inequality may create temporary niches for translators, verifiers, localizers, and institutional intermediaries, but those are lag positions—not permanent protection.

This paper is a competent map of how future users learn to touch the machine. It is not evidence that human academic participation remains economically necessary. The university can equalize access to automation and still produce a population whose work the automation no longer needs.