Your principal just sent an email about “implementing AI literacy across the curriculum” and you’re wondering what half the words mean. You’re not alone: a 2025 EdWeek survey found that 68% of teachers feel unprepared to teach about AI. This glossary covers every term you’ll encounter, explained without the tech-speak and with examples that actually make sense for educators.

A

AI (Artificial Intelligence): Software that can perform tasks that normally require human thinking: writing, analyzing, creating, and answering questions. ChatGPT, Claude, and Gemini are all AI tools.

Algorithm: A set of rules a computer follows to solve a problem. When people say “the algorithm,” they usually mean the system deciding what content to show you.

API (Application Programming Interface): A way for software to talk to other software. When MagicSchool generates a quiz, it’s using an API to communicate with an AI model behind the scenes. You don’t need to understand this to use AI tools.

C

Chatbot: An AI you interact with by typing messages. ChatGPT is a chatbot. So is Claude. The “chat” format is just the interface: the AI behind it can do much more than chat.

Context Window: How much text an AI can “remember” in a single conversation. A larger context window means you can paste longer documents. Claude has one of the largest context windows.

Copilot: Microsoft’s AI assistant built into Word, Excel, PowerPoint, and Teams. If your school uses Microsoft 365, you may have access to Copilot.

D

Data Privacy: How your information is stored and used. When you paste student information into an AI tool, that data may be stored or used to train the model. Always check a tool’s privacy policy. See also: FERPA.

Differentiation (AI-assisted): Using AI to create multiple versions of the same content at different reading levels or complexity. One of the most practical uses of AI for teachers.

F

FERPA (Family Educational Rights and Privacy Act): US federal law protecting student education records. Pasting identifiable student information into consumer AI tools like ChatGPT may violate FERPA. Use anonymized data or FERPA-compliant tools like MagicSchool’s school plan.

Fine-tuning: Training an AI model on specific data to make it better at a particular task. MagicSchool is fine-tuned for education. You don’t fine-tune models yourself: tool companies do this.

G

Generative AI: AI that creates new content (text, images, code) rather than just analyzing existing content. ChatGPT generates text. DALL-E generates images. Both are generative AI.

Gemini: Google’s AI assistant. Available free and integrated into Google Workspace (Docs, Slides, Sheets). Useful for teachers in Google-based schools.

GPT (Generative Pre-trained Transformer): The technology behind ChatGPT. GPT-4 is the current advanced model. You don’t need to know how it works: just know that GPT-4 is more capable than GPT-3.5.

H

Hallucination: When AI generates information that sounds correct but is completely made up. AI can fabricate quotes, statistics, book titles, and even research studies. Always verify factual claims.

I

Iteration: Refining AI output by giving follow-up instructions. “Make it shorter,” “add an example,” “change the tone to more formal.” Iteration is how you get good results from AI.

L

Large Language Model (LLM): The type of AI behind ChatGPT, Claude, and Gemini. It’s trained on massive amounts of text and predicts what words should come next. It doesn’t “understand”: it generates statistically likely responses.

M

MagicSchool: An AI platform built specifically for teachers. Has pre-built tools for lesson planning, quiz generation, rubrics, and more. Offers FERPA-compliant school plans.

Model: The AI “brain” that generates responses. GPT-4, Claude Sonnet, and Gemini are all models. Different models have different strengths.

P

Prompt: The instruction you give to an AI tool. “Write a quiz on fractions for 4th grade” is a prompt. Better prompts = better output.

Prompt Engineering: The skill of writing effective prompts. Includes being specific, giving context, specifying format, and iterating. Not as complicated as it sounds.

R

RAG (Retrieval-Augmented Generation): A technique where AI searches a database before generating a response, reducing hallucinations. Some education tools use RAG to ensure answers are based on actual curriculum standards.

S

Scaffold: In AI context, providing structure in your prompt so the AI follows a specific format. “Write 3 paragraphs: first about strengths, then growth areas, then next steps” is scaffolding your prompt.

T

Temperature: A setting that controls how creative or predictable AI output is. High temperature = more creative, more random. Low temperature = more focused, more predictable. Most tools set this for you.

Token: The unit AI uses to process text. Roughly, 1 token ≈ ¾ of a word. Token limits determine how much text you can input and how long the response can be.

Training Data: The text an AI was trained on. ChatGPT was trained on internet text, books, and articles. It doesn’t have access to your school’s data unless you provide it.

Z

Zero-shot: Asking AI to do something without giving it examples. “Write a rubric” is zero-shot. “Write a rubric like this example: [example]” is few-shot. Few-shot usually produces better results.

---

Bookmark this page and come back whenever you encounter a term you don’t recognize. We update this glossary as new AI tools and concepts emerge.

Related reading: 7 Best AI Tools for Teachers · 10 ChatGPT Prompts for Teachers · AI for Rubric Creation

🛠️ Try it yourself: Lesson Plan Generator or Rubric Generator: free, no signup needed.

FAQ

Do I need any special tools to get started with this?

For most AI applications, you just need a ChatGPT ($20/month) or Claude ($20/month) subscription. Some tasks benefit from specialized tools, but you can start with a general AI assistant and add specific tools as your needs grow.

How much time will this actually save me?

Most teachers report saving 3-8 hours per week once they’ve established their AI workflows. The first week is slower as you learn, but by week 2-3, the time savings compound. Focus on the tasks you do repeatedly: that’s where AI saves the most time.

Is the output quality good enough to use directly?

Rarely use AI output without editing. Think of AI as producing a strong first draft that’s 70-80% ready. Your expertise adds the final 20-30%: context, nuance, and accuracy that AI can’t provide. Always review before sending to clients or publishing.

What are the biggest mistakes teachers make with AI?

The top three: (1) not providing enough context in prompts, (2) trusting output without verification, and (3) trying to automate everything at once instead of starting with one workflow. Start small, verify everything, and expand gradually.

Will AI replace teachers?

No. AI replaces tasks, not jobs. The teachers who use AI will outperform those who don’t: they’ll handle more clients, produce better work, and spend less time on repetitive tasks. The value shifts from execution to judgment and relationships.