Institutional Responsible AI Edge Use Case • S-A-R v2.2 • W3C PROV-O
Arizona Western College (AWC) Pathway DOM GraphRAG Workbench
Focused, zero-PII Retrieval-Augmented Generation (RAG) cockpit scoped exclusively to a single 5-tier institutional pathway: Adjunct Faculty -> Exercise and Wellness Curriculum -> Allied Health Programs -> Entrepreneurial College -> Arizona Western College.
In Plain English • Responsible AI Edge Architecture
Before you read the details below, here is the short version: this workbench proves how a college can deploy Retrieval-Augmented Generation (RAG) at institutional scale without risking student privacy (FERPA), AI hallucinations, or recurring cloud fees. Every search runs right inside your browser across 1,005 verified public nodes connecting classroom Exercise and Wellness labs up through Allied Health Programs, the Entrepreneurial College, and Arizona Western College's HSI mission. Every result includes a cryptographic SHA-256 receipt and traces all five institutional layers.
En palabras sencillas: Es como un mapa interactivo blindado que conecta cada práctica del aula en Ejercicio y Bienestar con Salud Aliada, el Colegio Emprendedor y la misión estudiantil de Arizona Western College, sin subir datos de estudiantes a la nube ni inventar respuestas. Si los términos técnicos se ponen pesados, toca cualquiera de los 5 bloques de la ruta abajo para explorar cómo cada curso apoya el éxito estudiantil con costo $0 en libros.
5-Tier Institutional Pathway Graph Explorer
Select any node in the pathway chain to traverse connected curriculum, clinical, workforce, and institutional nodes:
Pathway-Scoped Local Edge Synthesis (127.0.0.1 • Public Pathway Nodes Only)
Synthesized locally via Apache 2.0 Ollama strictly from the public AWC pathway nodes retrieved below (zero access to non-pathway files; zero cloud egress).
Why Pathway-Scoped Edge GraphRAG Is an Institutional Responsible AI Reference Case
Aligned with U.S. Department of Labor ETA TEN No. 07-25 (Area 1: Understand AI Principles, Area 4: Evaluate Outputs, and Area 5: Use AI Responsibly) and NSF TechAccess: AI-Ready America, this workbench illustrates how community colleges can scale RAG responsibly across a focused academic-to-workforce pathway:
| Institutional Dimension | Unguarded Generic Cloud Chatbot | AWC Pathway-Scoped Edge DOM GraphRAG |
|---|---|---|
| Domain Containment | Draws from unverified open-web training weights; mixes unrelated institutions and outdated prerequisites. | Strictly bounded to verified public nodes along Adjunct Faculty → EXW → Allied Health → Entrepreneurial College → AWC. |
| Hallucination & Clinical Gate | Frequently invents citations or crosses licensed healthcare Scope-of-Practice boundaries. | Every node carries a W3C PROV-O SHA-256 receipt anchored in PAGA 2018, DGA 2025–2030, ACSM 12th Ed., NASM OPT, and statutory Scope-of-Practice gates. |
| FERPA & Student Privacy | Student prompts and pasted records transit external commercial servers. | 0 bytes of student PII; executes 100% client-side in browser memory with zero login and zero tracking cookies. |
| HSI Equity, Accessibility & Cost | Per-seat subscription tiers ($20+/user/month) and inconsistent screen-reader / contrast compliance. | $0 ZTC / $0 SaaS, bilingual English + Norteño-Spanish scaffolding, and 9:1 Double-Contrast Guard (ADA Title II / WCAG 2.2 AA). |