Topic 2: Prompt Engineering—How to communicate with an AI
Prompt engineering is the practice of designing effective instructions for AI systems. It is how we humans tell a model what we need. In professional settings, that skill has become critical to get the best from AI tools.
Anatomy of a good prompt A prompt is an instruction or question aimed at an AI system to get a specific, useful response. It can be a question, a request, or a description. The quality of the output depends heavily on how we phrase that request.
Simple prompt examples:
- “Draft a professional email to request a meeting.”
- “Explain what digital marketing is to someone with no prior experience.”
- “Summarize this text in three key ideas for a sales team.”
In all these cases, the AI does the writing, but we set the direction, scope, and style.
Practice: one question, many answers Different AI tools can deliver different results from the same prompt. Choose a clear, simple prompt relevant to your academic or professional life—for example: “Plan a week of study to prepare for three exams.” Try it in two conversational AI tools. Compare their answers for tone, depth, clarity, and style. Which is more useful? Why? What would you change in the prompt to improve the responses?
Why prompt writing matters AI does not infer intent. It follows language patterns and needs precise, clear, contextual instructions to produce good results. A well-crafted prompt is the difference between a useful answer and an irrelevant one. Asking “give me a summary” may yield a shallow recap. Asking “summarize this text in four key points using plain language for first-year business students” produces something specific, adapted, and understandable.
Key elements of a strong prompt Think of it like ordering takeout. If you only say “food,” you might not get what you expect. If you specify “salad without tomatoes, dressing on the side, packed to go,” you likely will. Prompts work the same way.
To be useful, a prompt should include clarity, context, and role.
☰ Clarity: say exactly what you need
AI does not read between the lines. Spell out the action, format, length, and style:
- What to do: write, summarize, explain, list, compare, etc.
- Format: paragraphs, bullet points, table, with examples.
- Length: brief, detailed, numbered points.
- Style: formal, friendly, technical, accessible, and what to avoid (jargon, common mistakes).
Example
Option 1: “Write a sales report.”
Option 2: “Write a brief report (max 300 words) summarizing last quarter’s sales, highlighting top products and comparing with the previous quarter. Use simple language for a team without data-analysis training.”
The second prompt wins because specificity produces more useful output.
☰ Context: help AI understand the need
Context gives purpose to the task. Without it, AI may reply with something generic or off-base. Context can include your role, the target audience, the channel (email, report, social post, presentation), and the goal (inform, persuade, summarize, teach).
☰ Role: tell the AI “who to be”
Assigning a role gives the system a frame: how to think, what style to use, and what knowledge to prioritize.
- “Act as a customer support advisor with ten years of experience.”
- “Teach this like a university professor explaining to new students.”
- “Respond as a graphic designer pitching an idea to a client.”
The role changes the angle and depth of the answer. “Summarize this text” is not the same as “Summarize this text like an editor preparing a newsletter for parents.”
Practice: craft a better prompt Test both prompts below and compare the responses.
- “Explain what artificial intelligence is.”
- “Explain what artificial intelligence is in one paragraph for first-year educational psychology students, without technical jargon, using simple examples.”
Which answer feels more accurate?
Prompting pitfalls and how to avoid them The most common mistakes when interacting with AI—and how to prevent them.
☰ Mistake 1: Being too general
Vague prompts lead to generic answers. Add detail about what you want, for whom, in what format, and with what angle.
Weak prompt: “Tell me about leadership.”
Stronger prompt: “In five sentences, explain transformational leadership for first-year business students.”
☰ Mistake 2: Skipping context
AI needs a frame to hit the right style and depth. Say who you are, who the content is for, and why you need it.
No context: “Write an introduction for a paper.”
With context: “Write an introduction for a report on digital trends, aimed at marketing professionals, with a technical but accessible tone.”
☰ Mistake 3: Asking for many tasks in one go
Stacking multiple jobs in a single prompt can confuse the model. Split the work into steps.
Messy: “Summarize the text, give examples, and make a pros/cons table.”
Ordered sequence: “Summarize the text in three key ideas.” → “Give one practical example for each idea.” → “Put the ideas into a pros/cons table.”
☰ Mistake 4: Skipping review
Copy-pasting AI output without checking can introduce errors, outdated info, or the wrong tone. Always read, correct, adapt to the real context, and confirm it meets your professional goal.
☰ Mistake 5: Writing ultra-short prompts
One-line prompts rarely yield rich results. Be precise about format, audience, and emphasis.
Weak: “Write a cover letter.”
Stronger: “Write a formal cover letter for a technology internship. I am a second-year computer science student with IT support experience. Highlight my interest in customer service and process improvement.”
Avoiding common mistakes is less about perfection and more about guiding the AI effectively. Good outcomes depend on both the tool and the judgment of the person using it. Asking well, reviewing carefully, and editing with intent are core skills for the modern professional working with AI.