Prompting Key Terms

Essential AI Prompting Vocabulary

Understanding these terms will help you talk about AI tools more confidently and get better results.

---

Prompt

What you type to the chatbot to get a response. Your instruction, question, or request.

Write a professional email declining a meeting

---

Chat

A conversation with the chatbot. Most AI tools maintain context within a chat, remembering what you've discussed earlier.

Important: Start a new chat for unrelated topics to avoid confusion.

---

Conversation / Thread

Another term for a chat. The full back-and-forth exchange between you and the AI.

---

Model

The specific version of AI you're using. Different models have different capabilities, speed, and cost.

Examples:

• GPT-4o, GPT-5 (OpenAI / ChatGPT)
• Claude Opus, Claude Sonnet (Anthropic)
• Gemini Pro, Gemini Ultra (Google)

Generally: Higher numbers or more advanced model names = newer and more capable (but sometimes slower or more expensive).

---

Token

The unit AI models use to process text. Roughly, 1 token = 4 characters or 0.75 words.

Why it matters: Models have token limits for how much they can read (input) and write (output) in one go.

---

Context Window

The maximum amount of text (measured in tokens) that a model can "remember" at once. This includes both your prompts and the AI's responses.

Larger context windows = you can have longer conversations or work with bigger documents.

---

Temperature

A setting that controls how creative or predictable the AI's responses are.

• Low temperature (0-0.3): More focused, deterministic, consistent
• High temperature (0.7-1.0): More creative, varied, unpredictable

Most users don't need to adjust this - default settings work well.

---

System Prompt / Instructions

Background instructions given to the AI before you start chatting. These shape the AI's behaviour, tone, and approach.

Example: "You are a helpful assistant that explains technical concepts simply."

You usually don't see these, but they influence how the AI responds.

---

Fine-Tuning

Training an AI model on specific data to make it better at particular tasks. This is advanced and typically done by organisations, not individual users.

---

API

Application Programming Interface. A way to connect AI tools directly to your own apps or workflows, rather than using the web interface.

Useful for: Automating tasks, integrating AI into products, processing data at scale.

---

Hallucination

When an AI confidently states something that's incorrect or made up. AI tools don't "know" when they're wrong - they predict what sounds right.

How to avoid: Ask for sources, verify important facts, use AI for drafts not final answers on critical topics.

---

Zero-Shot / One-Shot / Few-Shot

Prompting techniques based on how many examples you provide:

• Zero-shot: No examples, just ask
• One-shot: Provide one example
• Few-shot: Provide several examples

---

RAG (Retrieval-Augmented Generation)

A technique where the AI retrieves relevant information from a database or documents before generating a response. Makes outputs more accurate and grounded in specific sources.

You don't need to do anything special - some tools do this automatically when they search the web or access your files.

---

Prompt Engineering

The practice of crafting effective prompts to get better AI outputs. Sounds fancy, but it's really just: being clear, providing context, and iterating.

You don't need to be a "prompt engineer" to use AI well. Just practice and learn what works.