AI writing is full of jargon that sounds harder than it is. These are the terms you will actually meet as a student, in plain English.
What you will learn
Working definitions of the AI terms that appear in university policies, news, and tool documentation — enough to read any of them without getting lost.
The guide
Large Language Model (LLM) — an AI trained on huge amounts of text to predict likely next words. The engine inside ChatGPT, Claude, and Gemini.
Prompt — whatever you type to the AI. The craft of writing better prompts is prompt engineering.
Hallucination — a confident but false AI output: an invented reference, a wrong date, a fake quotation. A built-in property of LLMs, not a rare bug.
Generative AI (GenAI) — the umbrella term for AI that produces new content (text, images, audio, code) rather than just classifying existing content. The word your university's policy most likely uses.
Chatbot / assistant — the product wrapped around an LLM, adding memory of your conversation, web search, file reading, and safety rules.
Context window — how much text the model can consider at once: your conversation plus anything you paste. Exceed it and the model "forgets" the earliest parts.
Training data / knowledge cutoff — the text the model learned from, and the date it stops. Ask about anything after the cutoff and the model either searches the web or guesses — always check which.
Token — the word-fragments models actually read and produce; pricing and limits are counted in tokens (roughly ¾ of a word each in English).
Fine-tuning — additional training that specialises a general model for a task or field.
RAG (Retrieval-Augmented Generation) — a setup where the AI first fetches real documents, then answers from them. Reduces (does not eliminate) hallucination; used by "chat with your PDF" tools.
AI detector — software claiming to identify AI-written text. Unreliable in both directions; treated by institutions as a signal, not proof.
Multimodal — a model that handles more than text: images, audio, video in or out.
Currency note (July 2026): agents (AI that takes multi-step actions, not just answers) are the current frontier term; expect them to appear in policies next.
Why this matters for your studies
University AI policies are written in exactly these terms. Reading your own institution's policy with this page open turns it from legal fog into a clear set of rules.
What next
Return to the fundamentals guide to see these ideas working together, and the integrity guide for how policy applies them.