The student
“Am I learning the subject — or just learning how to please the system?”
An interactive research story · 7 minutes
Who pays
for progress?
A university embraces AI. The dashboards glow. The headlines celebrate innovation. But behind the promises, someone is carrying the cost.
Inspired by Janine Arantes, Critical Studies in Education (2026). The scenarios are thought experiments, not documented case studies.
01 / The question behind the headlines
What if our solution is quietly creating the next problem?
AI is often introduced to save time, improve learning, and make institutions more agile. But implementing it can create new work, uncertainty, and unequal demands on the people meant to benefit.
Arantes calls this the Innovation–Enervation Paradox: the strain generated by innovation becomes an argument for still more innovation.
Imagine running faster every year — and calling the exhaustion proof that you need a faster machine.
A creative paraphrase of the paper’s central argument, not a direct quotation.
↗We promised a better future.
02 / A short human story
One announcement. Many realities.
SCENE 01 / 04
“Our university is leading the AI revolution.” The room applauds. The announcement is beautifully polished.
A silent, caption-led animated story. Read at your own pace or play the scenes automatically.
03 / You're in the meeting
You're shaping AI policy. What do you do?
The university wants a new AI prompting programme. Leadership sees a bright future. Educators are worried about uneven digital experience and mounting workload. You have a seat at the table.
Make your choice.
Each path carries a different trade-off. There is no magic button that makes governance simple.
These consequences are interpretive prompts based on Arantes’s heuristic scenario, not predictions or outcome data.
04 / Try the invisible workload
Just one more prompt.
A student is told that better prompting leads to better answers. Try improving the answer. Notice what happens to your attention.
Student workspace / draft response
“Education is important because it helps people learn new skills.”
VERSION 01 · STARTING DRAFT
Could this be better?
The feedback says: “Make it stronger.” What does that actually mean?
A deliberately designed simulation of iterative effort. Energy is illustrative, not a research measurement.
05 / The double bind
“Use AI.”
“Prove you didn't.”
Now imagine the same university encouraging AI-enabled learning while introducing AI detection in assessment. A student follows the advice — and then faces a review.
From the teaching team
“Experiment. Iterate. Use AI as a learning partner.”
Prompting is embedded into learning activities and workshops.
From the integrity office
“Your submission has been flagged for review.”
Staff must investigate, interpret, and communicate decisions amid shifting rules.
This is the paper’s second heuristic scenario, not evidence that a particular institution behaved this way.
06 / Whose voice counts?
Three people. One policy.
Different stakes.
The educator
“Who finds the hours to rewrite assessments and resolve every grey area?”
The decision-maker
“We need to move forward. But what don't we yet understand?”
These are original composite voices written to illuminate the paper’s governance tensions — not interviews or quotations from research participants.
07 / Take the conversation somewhere
Maybe the boldest innovation is asking who gets to say no.
Arantes does not argue for a simple rejection of AI. The challenge is to govern it with genuine participation, recognition of expertise, and room for refusal.
What would you change first?
Choose one to reveal your conversation starter.
Read beneath the surface
The research behind the story.
This experience is an interpretation of an open-access scholarly article. Source buttons throughout the story identify the relevant pages. The paper’s two scenarios are analytic thought experiments, not empirical case reports.
THE CENTRAL PARADOX · PP. 7–8
Innovation can generate strain that becomes a justification for more technological intervention.
CONFIDENCE & EXPERTISE · PP. 8–10
Partial, distributed knowledge can coexist with strong institutional certainty; Dunning–Kruger is used as a heuristic, not a diagnosis of individuals.
SCENARIO 1 · PP. 12–13
Prompt engineering can shift effort toward repeated output optimisation and privilege some forms of knowledge over others.
SCENARIO 2 · PP. 14–15
AI promotion and AI detection can impose competing expectations on students and generate interpretive labour for staff.
CONCLUSION · PP. 16–17
AI governance should examine the distribution of authority, labour, risk, participation, and the capacity to refuse.
ORIGINAL PUBLICATION
Janine Arantes (2026), “The Innovation–Enervation Paradox in higher education: AI and the Dunning–Kruger effect,” Critical Studies in Education.
Open original paper ↗ · View supplied PDF ↗