What is the purpose of a system message in Azure OpenAI?

To set the AI's persona, behavior, constraints, and response format. The system message sets the AI's persona (who it acts as), tone (how it communicates), format (how to structure responses), and constraints (what it should not do). It is the most powerful tool for controlling model behavior in chat completions.

A developer provides three example input-output pairs before asking the model to classify a new text. What prompt engineering technique is this?

Few-shot prompting. Few-shot prompting provides examples of the desired input-output pattern before presenting the actual task. The examples help the model understand the expected format and behavior. Zero-shot uses no examples; chain-of-thought asks for step-by-step reasoning.

Which temperature setting would produce the MOST creative and varied responses?

Temperature 0.9. Higher temperature values (closer to 1.0) produce more creative, varied, and unpredictable responses. Temperature 0 produces the most deterministic and focused responses. For factual tasks, use low temperature; for creative tasks, use higher temperature.

What is the purpose of "grounding" in prompt engineering?

Providing specific factual data for the model to base its response on, reducing hallucinations. Grounding provides the model with specific, relevant data or documents to use as its source of truth. This constrains the model to the provided information, significantly reducing hallucinations and improving factual accuracy. The model answers from the grounding data, not from its pre-trained knowledge.

Adding the phrase "Think step by step" to a prompt is an example of which technique?

Chain-of-thought prompting. Chain-of-thought prompting asks the model to show its reasoning process step by step. Phrases like "think step by step," "explain your reasoning," or "show your work" activate this technique. It significantly improves accuracy on math, logic, and multi-step reasoning tasks.

Prompt Engineering

Quick Answer: Prompt engineering is designing effective instructions for generative AI models. Key techniques include system messages (set behavior), few-shot prompting (provide examples), chain-of-thought (step-by-step reasoning), and grounding (provide factual data). A well-crafted prompt can dramatically improve the quality, accuracy, and relevance of AI responses.

What Is Prompt Engineering?

Prompt engineering is the art and science of crafting effective prompts (instructions) to get desired outputs from generative AI models. Since you are not modifying the model itself, the prompt is your primary tool for controlling behavior.

Think of it this way: the generative AI model is a highly capable assistant, and your prompt is the instructions you give that assistant. Better instructions lead to better results.

Anatomy of a Prompt

Chat Completion Message Structure

Azure OpenAI uses a chat completion API with three message roles:

Role	Purpose	Example
System	Set the AI's persona, behavior, constraints, and format	"You are a helpful customer service agent for a bank. Always be polite and never provide financial advice."
User	The actual question or request from the end user	"What are your hours of operation?"
Assistant	Previous AI responses (for conversation history)	"Our branches are open Monday to Friday, 9 AM to 5 PM."

The System Message

The system message is the most powerful tool for controlling AI behavior. It sets:

Persona: Who the AI should act as
Tone: How the AI should communicate
Format: How responses should be structured
Constraints: What the AI should NOT do
Knowledge: What context the AI should use

Example system message: "You are an expert Azure certification tutor. Explain concepts clearly using simple language. Always provide examples. Format answers with bullet points when listing items. If you do not know the answer, say so honestly instead of guessing. Never provide actual exam questions."

Key Prompt Engineering Techniques

1. Zero-Shot Prompting (No Examples)

Give the model a task with NO examples — rely on its pre-trained knowledge.

Prompt: "Classify the following text as positive, negative, or neutral: 'The movie was absolutely fantastic!'"

Response: "Positive"

When to use: Simple tasks where the model understands the task from the instruction alone.

2. Few-Shot Prompting (With Examples)

Provide examples of the desired input-output pattern before the actual task.

Prompt: "Classify the sentiment of each review:

Review: 'Great product, fast shipping!' → Positive Review: 'Terrible quality, waste of money' → Negative Review: 'It works fine, nothing special' → Neutral Review: 'The customer service team was incredibly helpful and resolved my issue within minutes' →"

Response: "Positive"

When to use: When you want the model to follow a specific pattern or format, or when zero-shot results are inconsistent.

3. Chain-of-Thought Prompting

Ask the model to think step by step, showing its reasoning process.

Prompt: "A store has 47 apples. They sell 23 and receive a new shipment of 35. How many apples do they have now? Think through this step by step."

Response: "Step 1: Start with 47 apples. Step 2: Subtract 23 sold: 47 - 23 = 24 apples. Step 3: Add 35 from new shipment: 24 + 35 = 59 apples. Answer: The store has 59 apples."

When to use: Math problems, multi-step reasoning, complex analysis. Adding "think step by step" or "explain your reasoning" significantly improves accuracy on reasoning tasks.

4. Grounding with Context

Provide the model with specific data or documents to base its response on, rather than relying on its pre-trained knowledge.

Prompt: "Based on the following document, answer the question.

DOCUMENT: 'Azure AI-900 exam has 40-60 questions, costs $165, and requires a score of 700/1000 to pass. The exam duration is 45 minutes.'

QUESTION: How much does the AI-900 exam cost?"

Response: "The AI-900 exam costs $165."

When to use: When accuracy is critical and you want to reduce hallucinations. The model answers from the provided data, not from its potentially outdated or incorrect pre-trained knowledge.