Artificial Intelligence Guidance Glossary

AI & K-12 Glossary (Jargon + Acronyms)

Quick definitions for terms used across DPI’s AI Guidance. Use your browser’s Find (Ctrl/Cmd+F) to jump to a term, or click a letter below.

▸A

Academic Integrity — Upholding honesty and original work when using AI (proper attribution, citations, allowed uses). Why it matters: protects learning and fairness.

Adaptive Learning — Software that adjusts content or pacing based on learner data. Why it matters: supports personalization and UDL.

AI (Artificial Intelligence) — Computer systems that perform tasks typically requiring human intelligence (recognizing patterns, generating text, etc.).

AI Literacy — Knowledge, skills, and dispositions to use, question, and create with AI ethically and effectively. Why it matters: foundation of DPI’s guidance and the AI Literacy Framework.

Anonymization / De-identification — Removing direct identifiers from data; true anonymization is hard. Why it matters: protects PII when using AI tools.

Attribution — Crediting AI tools when they contribute to a product (and citing sources used by the model when appropriate). Why it matters: transparency & integrity.

▸B

BadgerLink — Wisconsin’s statewide portal to licensed online resources (articles, reference sources). Why it matters: trusted sources to verify AI outputs.

Bias (Algorithmic) — Systematic, unfair outcomes in model training or outputs (e.g., stereotypes, exclusion). Why it matters: equity & legal considerations.

▸C

CESA — Cooperative Educational Service Agency (WI regional support organizations). Why it matters: training & implementation partners.

Chatbot — Conversational AI interface (text or voice) that answers questions or completes tasks.

Context Window — The amount of text (tokens) a model can consider at once. Why it matters: limits document length for analysis.

COPPA / PPRA — U.S. student privacy laws: Children’s Online Privacy Protection Act and Protection of Pupil Rights Amendment. Why it matters: parent consent & student data protections.

▸D

Differential Privacy — Adding statistical noise so individual records can’t be identified. Why it matters: safer analytics on student data.

Digital Durability — The idea that data and AI outputs persist and can be reused, resurfaced, or misused later. Why it matters: teach students long-term consequences.

▸E

Ethical Use — Using AI with fairness, transparency, privacy, safety, and accountability.

Embedding — Vector representation of text/images that capture meaning for search or retrieval. Why it matters: powers RAG and semantic search.

▸F

FERPA — Family Educational Rights and Privacy Act. Why it matters: governs disclosure of student education records.

Fine-Tuning — Training an existing model further on your examples to specialize behavior.

▸G

Generative AI — Models that create new content (text, images, audio, code) from prompts.

G&T (Gifted & Talented) — Programs for advanced learners; in this guidance, also “Academic Support Faculty.”

▸H

Hallucination — Confident but incorrect or fabricated AI output. Why it matters: teach verification habits.

H → AI → H (Human-in-the-Loop) — Human designs the task → AI drafts → Human reviews, revises, and owns final product.

HIPAA — Health Insurance Portability and Accountability Act (health data privacy; see School Nursing notes).

▸I

IEP — Individualized Education Program for students with disabilities.

Inference — A model generating outputs from inputs after training (the “use” phase).

ITL Standards — Wisconsin’s Information & Technology Literacy Standards (2025 revision).

▸J

Jailbreak — Techniques used to make a model ignore its safety rules; not appropriate for K-12 use.

▸K

K-12 GenAI Maturity Tool — Self-assessment rubric for district readiness and responsible AI implementation.

▸L

LEA / SEA — Local Education Agency (district) / State Education Agency (DPI).

LLM (Large Language Model) — A model trained on massive text to generate and reason over language.

LMS / SIS — Learning Management System (e.g., Canvas) / Student Information System (e.g., Infinite Campus).

▸M

Model Card — A transparency document describing how a model was built, tested, and its limitations. Why it matters: helps vet tools.

MTSS — Multi-Tiered System of Supports (tiered academic/behavioral supports).

▸N

NIST AI RMF — U.S. National Institute of Standards and Technology’s AI Risk Management Framework.

NLP — Natural Language Processing (AI working with human language).

▸O

Opt-Out / Consent — District procedures letting families control student data sharing with vendors and tools.

▸P

PII (Personally Identifiable Information) — Data that can identify a student (name, ID, image, voice, etc.). Why it matters: never paste PII into public AI tools.

Prompt / Prompt Engineering — The instructions you give an AI model (and techniques to improve results).

▸Q

Quality Review — The human evaluation step for accuracy, bias, privacy, and age-appropriateness (see DPI AI Output Review Checklist).

▸R

RAG (Retrieval-Augmented Generation) — Technique where the model looks up documents first, then writes using those sources. Why it matters: reduces hallucinations; enables local content.

Red Teaming — Testing a system to find safety, bias, or privacy failures before public use.

▸S

SEL — Social-Emotional Learning (skills for self-awareness, self-management, relationships, decision-making).

System Prompt — Hidden instructions that set a model’s role/behavior in a tool (“You are a student writing coach…”).

▸T

Temperature — Controls randomness/creativity in text generation (lower = more predictable).

Token — A chunk of text a model reads/writes (≈ 3–4 characters or 0.75 word on average). Why it matters: affects limits & cost.

▸U

UDL (Universal Design for Learning) — Designing learning that is accessible and flexible from the start.

▸V

Vector Database — Specialized database for embeddings, used to power semantic search and RAG.

Watermarking (AI) — Hidden or visible markers to signal AI-generated content.

▸W

WISELearn — Wisconsin’s OER hub for educator-curated resources (collections include AI PD and classroom tools).

WDLC — Wisconsin Digital Learning Collaborative (statewide online/virtual learning partnership).

Wisconsin DPI — Department of Public Instruction (SEA for WI; this guidance site).

← Back to AI Guidance Homepage