Educator demo scripts
Run-it-live demo scripts for showing BetaNYC’s grounded AI + NYC/NYS open data MCPs to faculty, curriculum designers, and instructional staff.
These differ from the other demo folders in this repo by the question they answer. ../council-members/ and ../community-boards/ answer “what is happening in my district?” — the audience wants their own data back. An educator audience is asking something else: “what does this teach, and can I assign it?”
So the arc is built around transferable concepts rather than local findings:
- provenance and resolvable identifiers,
- API contracts and schema reading,
- verification, including the discipline of stating a bounded negative,
- privacy constraints as they actually appear in public data (suppressed cells, not hypotheticals).
Available demos
| Audience | File |
|---|---|
| CUNY computer science curriculum | cuny-cs-curriculum.md |
| Liberal arts and general education — writing, critical thinking, political science | cuny-liberal-arts.md · deck: cuny-liberal-arts-deck.html |
Adapting these
The CS script leans on education-sector datasets (CTE reports, graduation outcomes) because that is what a computer science curriculum audience recognizes. For a different discipline, swap the dataset in Act 3 and keep the structure:
- Act 1 — demonstrate the failure mode rather than asserting it.
- Act 2 — catalog search as an API contract.
- Act 3 — one dataset the audience already cares about, in depth.
- Act 4 — a gap, a limitation, or a data-quality problem. This act is where the honesty discipline gets taught, and it is the one worth keeping in every version.
- Act 5 — setup cost and assignment shapes.
Worked swaps, from the two scripts in this folder:
| Discipline | Act 3 | Act 4 (the honesty act) |
|---|---|---|
| Computer science | CTE programs report (gyjk-fbss) with schema and NYSED-approval columns |
Local Law 174: the law says annual, the record stops at 2019–2020 |
| First-year writing | One agency document plus the record behind its central claim | What the document asserts that the data does not support |
| Critical thinking | The suppressed-cell CTE enrollment file (8vqd-3345) |
The silent try/except that yields a wrong average |
| Political science | Council discretionary funding for the student’s own district | The same Local Law 174 gap, read as an oversight failure |
Still unwritten: for a journalism school, Act 3 might be 311 or procurement. For a statistics course, the suppression discussion in Act 3 can carry the whole session on its own.
Name the question the audience came in with before you pick anything. That is the real adaptation step. The council and community-board folders answer “what is happening in my district?” An educator audience asks “what does this teach, and can I assign it?” — and within that, a CS audience wants the systems property while a humanities audience wants the reading-and-writing property. Same connectors, different arc.
A standing rule for this folder
Dry-run every prompt against the live MCPs before a demo, and record what happened in a “Presenter notes (verified YYYY-MM-DD)” block. Prompts that read well can fail live — a search that silently returns [], a response large enough to stall the room. The presenter notes in cuny-cs-curriculum.md document several real ones found this way. A script without verified notes is a draft, not a demo.
See also
../user-journeys.md— the full, audience-neutral prompt catalog.../grounding-ai-agents.md— the rationale these demos dramatize.../training-resources.md— BetaNYC’s own classes and programs.