The words behind the tools you already use — so you can study smarter, spot the traps, and hold your own in any conversation about AI.

Academic integrity — Doing and crediting your own work honestly — including being upfront about any AI help, under your school or uni's rules.

Agent — AI that can carry out tasks — click, browse, run steps — not just reply in a chat.

AI detector — Software that tries to guess whether you used AI. It's unreliable and biased against non-native and plain writers — it should never be the only evidence.

Cognitive offloading — Letting a tool do thinking for you. Fine for some things — risky when it's the very skill you're meant to be learning.

Context window — How much the AI can “hold in mind” at once. Go past it and it forgets the start of the conversation.

Generative AI — AI that makes new things — essays, images, code, audio — rather than just sorting what already exists.

Hallucination — When AI confidently makes something up — a fake quote, a wrong number, a citation that doesn't exist. It sounds right, so always check.

LLM (large language model) — The tech behind chatbots like ChatGPT: trained on huge amounts of text to predict the next word, over and over.

Productive failure — The evidence that trying a hard problem before you get help makes the learning stick deeper.

Prompt — The instruction you give the AI. Clearer, more specific prompts get better answers.

Sycophancy — AI's habit of agreeing with you even when you're wrong. Don't mistake confidence for correctness.

The sandwich method — Do the work yourself first, use AI to critique or extend it, then revise — and note what you changed. You keep the learning; you gain the polish.

Token — The word-chunks an AI reads and writes. Length limits and pricing are counted in tokens, not words.

Verification — Checking what AI tells you against a reliable source before you trust it or hand it in.