Generative AI as an education policy problem
Govern,Wait,or Redesign?
Four readings from 2023 to 2026, two sectors, one question: who decides how AI enters schools and universities, and whose knowledge does that policy rest on?
The through-line
The problem was redefined, 2023 to 2026
2023
Chan · students and teachers
“What is the definition of cheating?”
Problem: Integrity
Read the summary →
2023
Dunnigan et al. · a school leader
“I refuse to do it because I don't know what to put in it.”
Problem: No map to act on
Read the summary →
2025
Hughes et al. · a student (PS4)
“Is there an actual value to a degree? Now”
Problem: The institution's value
Read the summary →
2026
Henriksen et al. · Leon Furze
“Chatbots don't make sense—they make words.”
Problem: What is human in learning
Read the summary →
From who may use AI, to what education is for.
Map
Four readings, four problem definitions
| Reading | Sector | Central claim | Evidence |
|---|---|---|---|
| Chan (2023) | Higher Ed | Universities need an AI policy built from stakeholder input, with named owners. | Mixed-methods survey · Hong Kong universities · 457 students, 180 teachers and staff |
| Dunnigan et al. (2023) | K-12 | Leaders are deliberately waiting on policy and training; equity is barely on the radar. | Column built on two focus groups (~3 hours) · 6 Michigan districts and 7 Pacific Northwest independent schools |
| Hughes et al. (2025) | Higher Ed | GenAI challenges are tightly interdependent, so piecemeal fixes fail. | 29 interviews plus Interpretive Structural Modelling with 7 academic experts |
| Henriksen et al. (2026) | Teacher Ed | Embodiment, struggle, foundational knowledge, and agency matter more with GenAI. | Conceptual synthesis of 8 public-scholar interviews, 2023–2025 |
As the problem moves from rules to human capacity, the actor moves from managers to educators. The evidence never moves past perception.
Start here