Public aggregate insights dashboard for Make AI Visible.
This repository owns the public-facing view of privacy-protected research findings. The MVP should make patterns visible without exposing individual contributors, conversations, schools, locations, or rare cohorts.
- Display aggregate counts and behavioral score distributions.
- Support cohort-level filters only when privacy thresholds are satisfied.
- Apply suppression and differential privacy noise before publication.
- Show methodology notes and data freshness metadata.
- Avoid individual-level records, quotes, or small-cell drilldowns.
The dashboard must only consume aggregate, publication-approved data. No raw conversations, anonymized full transcripts, or reviewer-only records should be shipped to the client.
- React or Next.js with TypeScript.
- Static aggregate JSON or a read-only API.
- Charting library with accessible labels.
- Automated checks for minimum cohort sizes.
Build a static dashboard using synthetic aggregate data, including suppression rules for cohorts below the publication threshold.
This repo includes a static synthetic aggregate dashboard:
python3 -m http.server 8000Then open http://127.0.0.1:8000.
The dashboard reads data/synthetic_aggregate.json, displays aggregate dimension counts,
and suppresses cohorts below min_public_cohort. It must not ship raw conversations,
quotes, individual records, schools, locations, or reviewer-only data.