File APB-2026 · Office of AI Policy
A policy simulation for EDU 280, Social Justice and Urban Education
AI Policy BuilderRun AI policy for a school district. Live with what it does.
Four rounds, three school years, twelve decisions, with real evidence behind every option.
There is no policy here that helps everyone. The game scores whether you saw who it cost.
No campaign on file. Nothing is saved until you begin.
4
Bands playable
K-5 to higher ed
232
Sources in the registry
public and auditable
12
Decisions
per campaign
45-60
Sitting time
minutes, autosaved
Two courses, one campaign
Social justice and foundations
Who bears the harm, who was never in the room, and what the aggregate conceals. Consultation costs capacity, so nobody can hear everyone, and one outcome number looks fine until it is disaggregated.
Education policy and leadership
What kind of instrument this is, whether the bodies that must approve it would, how much of it actually reached classrooms, and what the road not taken would have cost. Students leave with a policy memo waiting for their argument.
The instructor page names every framework a student practices, with its citation, and says plainly what the game does not teach. Read the instructor page.
How a campaign works
- Step01
Choose who to hear
Consultation costs staff capacity, so you can never bring everyone into the room. Whoever you leave out stays left out, and the game remembers.
- Step02
Decide and commit
Each decision offers real options with real precedents, openable evidence cards, and six advisors who disagree with each other. Commits are final.
- Step03
Watch it land
Time advances. Some policies do not survive the bodies that must approve them, and the ones that do arrive only partly. Consequences come in three tiers: seen, unseen, unforeseen. One aggregate number looks fine, until you disaggregate it.
A full campaign takes about 45 to 60 minutes and saves automatically in your browser. It can be split across sittings, and it can be shared as a link that carries only the seed and the choices.
One system, K through 20
All four bands are playable now. Each runs on the same engine with its own campaign, its own stakeholders, and its own verified evidence base, so the same mechanic lands differently at five, at thirteen, and at twenty.
- Playable
Grades K-5
Reading tutors, child data privacy, and the youngest students in the dataset.
- Playable
Grades 6-8
First phones, deepfakes, monitoring, and the ages the law splits at 13.
- Playable
Grades 9-12
Four rounds, Fall 2026 to Spring 2028. Detection, procurement, surveillance, devices, IEPs, and the graduating class of 2029.
- Playable
Higher Education
Proctoring, campus AI licenses, faculty governance, and academic freedom.
Cited, or cut
Every policy option traces to something a real district, state, or institution actually did. Every consequence is grounded in published research, government data, or verified reporting, and the simulation’s own numbers are labeled as modeled, with the evidence they extrapolate from. Retracted studies are excluded, and contested findings are labeled suggestive, not settled.
The whole list is public. You can audit it without playing, and every citation in the game links straight back into it.
Audit the evidence base- Peer reviewed or authoritative
- 85
- Government, court, district
- 50
- Journalism, for events only
- 89
- Suggestive, labeled as such
- 8
What comes out of it
At the end, the campaign produces four documents, generated in the browser, fully cited, available as a filed PDF or as editable Markdown. Three of them are records of what happened. The fourth is deliberately unfinished.
Exhibit A
Policy brief
The policy they built, decision by decision, with the evidence that was on the table at each point.
Exhibit B
Equity impact report
Disaggregated outcomes, harms seen, missed and caused, who was consulted, who never was, and when they first disaggregated.
Exhibit C
Decision trace
Every evidence card, marked opened or never opened. Students meet their own reading behaviour.
Exhibit Dscaffold, not a record
Policy memo
The professional deliverable of education policy, structured as problem definition, alternatives, criteria, projected outcomes, trade-offs, recommendation, implementation plan, and evaluation plan. Everything the campaign recorded is set as cited evidence, including the instruments the student actually wrote and how much of each policy reached classrooms.
Everything that is the writer’s own reasoning is left as a ruled blank, with a statement of what belongs there. A memo handed in with the blanks still blank is visibly unfinished, which is the point of building it this way.
Assigning this in a course?
The instructor page names every framework a student practices, with its citation, and sets out six assignments built on what the game actually produces. It also says plainly what the game does not teach. Free to assign, no accounts, no setup.