In an era where 92% use AI, what should universities preserve and what should they change: rewriting classes, grades, and governance

Introduction: The "Design Change" AI Brings to Universities

How will generative AI change universities? The debate is heating up, but the future is not monolithic. A research team at Chalmers University of Technology has depicted a "near future" spanning two years in six scenarios using a method of evidence-based imagination called "informed educational fiction." This is not a prophecy but a mirror for making choices. Phys.org

Research Highlights: Capturing the Future Through Stories

This study, based on insights from student interviews, involved workshops with faculty, postdocs, and educational developers to narrate "possible futures." The stories are rooted in data and theory, functioning as "thought experiment devices" to encourage deliberation among decision-makers. Phys.org

Six Near-Future Scenarios (Summary and Opinions)

1. Conflicting Learning Goals
   Students "complete" assignments using AI, prompting educators to reconsider "what they want students to learn." The key lies in designing evaluations that assess not only the learning outcomes but also the process evidence (thinking traces, arguments, revision history). Phys.org

2. Students' "Excessive Autonomy"
   While AI acts as a companion, pitfalls of mislearning and overconfidence also exist. Balancing the granularity of support with the development of self-regulated learning is a point of contention. Phys.org

3. Unpredictability of GenAI
   The curriculum update cycle cannot keep up with technological evolution. **Meta-design of learning outcomes (adaptable SLOs)** and refactoring of assessments become necessary. Phys.org

4. Contradictory and Counterproductive Regulations
   Discrepancies in policies among departments, faculty, and university headquarters lead to student confusion. Hierarchical alignment of policies (alignment design across internal layers) and transparent consensus building are key. Phys.org

5. Transformation of Faculty Roles
   From "transmitter" to "designer of learning experiences." Skills in personalized design and formative assessment using AI become core competencies. Phys.org

6. Forging an AI-ready Campus
   Rather than leaving it to individuals, organizations should establish support, norms, and infrastructure. LMS integration, audit logs, prompt literacy training, and data governance are the pillars. Phys.org

Field Data: Rapid Expansion of Use and the Reality of "Grading AI"

A survey in the UK found that 92% of students use generative AI for learning. Summarizing key points, generating explanations, and drafting assignments have become established practices. However, concerns about misinformation and unfairness have also been raised, and stress testing of assessment methods is being demanded of universities. The Guardian

On the faculty side, analysis of 74,000 conversation logs shows a growing use of AI in grading and assessment, with "almost complete delegation" observed in about half of the cases. This raises issues of accountability and fairness alongside efficiency. Axios

Furthermore, reports that cross-examine the realities of students and faculty make visible the current situation where generative AI affects both learning design and avoidance. In other words, without good design, it becomes a "shortcut," but with good design, it becomes "scaffolding." Phys.org

Reactions on Social Media: Aspirations, Distrust, and Practical Solutions

In student communities, anger towards AI detection tools and stories of blatant cheating in classrooms are shared, reflecting fatigue on the "detection vs. learning support" seesaw. Reddit
In university subreddits, there are voices expressing discomfort with the use of AI in university public relations and confusion over the "AI saturation" of campuses. Reddit
Threads in AskAcademia continue open discussions on the impact on research and publication. Reddit

On LinkedIn, provocative posts like "Higher education is heading towards the AI iceberg" and concerns about the academic world being swallowed by corporate-led initiatives garner sympathy, while practitioners propose a forward-looking approach, suggesting that "authenticity should be ensured through design beyond detection dependence." LinkedIn

A "Calm Perspective" Including Opposing Views

Criticisms such as "AI will hollow out higher education" and "accelerate the credential business" are persistent. These criticisms point to "structural issues" that cannot be avoided unless the purpose of learning is redefined along with course design and evaluation. The Australian

On the other hand, there are also observations that "AI has not yet made a substantial impact," questioning the quality of implementation. Phys.org

The Implementation Triangle: Assessment, Roles, Norms

1. Assessment

   • From single-point evaluation of outputs to process evaluation: Evaluating prompt drafts, reasoning notes, and version control logs.

   • Increase the weight of oral examinations and live compositions, making self-reporting of AI assistance

     mandatory.

   • "AI-assumed" task design (e.g., assignments that critically examine AI outputs).

2. Roles

   • Faculty as "architects of learning experiences." The core involves designing formative feedback and developing AI literacy.

   • Students as "self-regulated learners." Metacognition, evaluating the reliability of information, and debating with AI are essential skills.

3. Governance & Infra

   • Establish hierarchical alignment of AI policies within the institution (course ⇄ department ⇄ university).

   • Integration of LMS×AI infrastructure / log auditing / principles of privacy and data sovereignty.

   • Institutionalize continuous training for faculty and students, operating an **"AI-ready campus."** Phys.org