Navigating the challenges of Gen-AI in Chinese higher education: Balancing technological innovation with academic integrity and intellectual engagement

Synthesis: Bai and Costa (2026) use a qualitative design — semi-structured interviews with 33 staff and students in Jiangxi Province, China — to explore how Gen-AI is reshaping teaching, learning and thinking in Chinese Higher Education. Drawing on Bourdieu's field and habitus and Arendt's view of thinking as an inward, untransferable activity, they identify two principal tensions: the erosion of independent, critical and creative thinking, and the destabilisation of established understandings of Academic Integrity. While Gen-AI tools save time and offer emotional support, they risk fostering cognitive passivity that undermines the intellectual processes fundamental to higher education. The authors call for critical curricular reform and a redefinition of intellectual Learner Agency in an increasingly AI-mediated academic environment.

Key Findings

Three themes. Reflexive thematic analysis of 33 transcripts surfaces three themes: (1) concerns over independent, critical and creative thinking; (2) the impact of Gen-AI on educational and study practices (chiefly workload reduction); and (3) Academic Integrity and ethical concerns.

Theme 1 — cognitive inertia and dependency. Both staff and students worry that Gen-AI encourages cognitive inertia and reduces independent thinking, despite its efficiency. Participants described an "Over-Reliance on the technology" that "makes individuals think less actively; the ability to innovate decreases," and that "excessive reliance on Gen-AI will reduce our ability to analyze independently." There is an awareness that the convenience of Gen-AI can lead to superficial engagement with knowledge — echoing the knowledge base's Over-Reliance and Cognitive Offloading concerns.

Theme 2 — practical benefits vs. workload. Despite the dependency worries, participants acknowledge Gen-AI's practical value for performative, time-consuming tasks — creating test questions and test papers, guiding student thinking, analyzing learning situations, and generating work materials, summaries and reports. A reported need to reduce workload sits at the center of these uses, illustrating the ambivalent everyday negotiation of Gen-AI in Higher Education.

Theme 3 — academic integrity and ethical concerns. The ease with which Gen-AI generates content — including academic papers — raises questions about originality and research integrity. Participants also noted that Chinese-company-managed Gen-AI tools "do not cater for political debates or diverse views and stances on key issues," surfacing an ethical and geopolitical dimension to AI-mediated academic work.

Theoretical frame. Bourdieu's field and habitus locate Gen-AI use within shifting institutional structures and dispositions, while Arendt's conception of thinking as an inward, untransferable activity supplies the normative criterion against which cognitive passivity is judged; hooks and Freire further anchor the analysis of curriculum and education as engaged practice and of knowing as "earned, not given."

Method. A case-oriented, reflexive thematic approach (Braun & Clarke) was applied to 33 semi-structured interviews from Jiangxi Province, with within-case then cross-case synthesis of staff and student transcripts, reflexive memos, co-researcher coding, and an Ethics-approved protocol (#EDU-2024-2716-2733).

Reform imperative. The authors argue for critical curricular reform and reimagined practices that prioritize intellectual development, reframing the purpose of higher education in an AI-mediated environment. Safeguarding the humanistic aspect of higher education demands redefining intellectual agency and critical engagement with technological change, linking to AI Literacy and Learner Agency.

What this means for practice

• Instructors. Publish course-level rules that say how Gen-AI use and Academic Integrity coexist in each assignment, because the study found official policy endorsed both aims without explaining how they fit together, leaving staff and students to devise their own standards.
• Learners. Protect the work whose purpose is the thinking: keep analytical and essay-generation tasks unaided when the aim is independent reasoning, and reserve Gen-AI for the laborious, performative tasks—test questions, materials, summaries—where participants reported their workload relief came from.
• Instructors. Require students to show the reasoning behind AI-assisted work—source checking, critique of AI output, an account of what they changed—so that efficiency does not become the cognitive inertia participants described as thinking "less actively."
• Administrators. Fund curricular reform that names intellectual development as the purpose of AI-mediated study, and audit what the endorsed tools will not discuss (participants flagged Chinese-company-managed Gen-AI tools for avoiding political debate and diverse viewpoints).
• Faculty developers. Treat Gen-AI's emotional and relational role—students described a tool as a "loyal friend"—as a signal of unmet campus needs, and pair AI provision with community and advising rather than letting the tool absorb them.

Limitations

• The study draws on 33 semi-structured interviews—21 students and 12 academic staff—recruited through a conference group of the Jiangxi Provincial University Alliance, with no demographic criteria beyond prior Gen-AI experience, so the sample is self-selected.
• All participants came from one province (Jiangxi), chosen partly because it is a researcher's home region with convenient local contacts, and the interviews ran only from December 2024 to January 2025.
• The evidence is participants' accounts of their own thinking and conduct rather than any measured change in critical thinking or integrity, so the cognitive-inertia finding is perceptual rather than demonstrated.

Citation

Bai, L., & Costa, C. (2026). Navigating the challenges of Gen-AI in Chinese higher education: Balancing technological innovation with academic integrity and intellectual engagement. Computers and Education Open, 100397. https://doi.org/10.1016/j.caeo.2026.100397