What is prompt engineering?

The term ‘prompt engineering’ sounds like a complicated technical skill, doesn’t it? It conjures images of coding experts or people in lab coats tinkering with advanced AI systems. But the truth is, it’s a lot simpler than it sounds. At its core, prompt engineering is just about knowing how to ask questions or give instructions to AI tools like ChatGPT to get the results you need. Think of it like having a conversation with a super-smart assistant—it’s all about knowing how to phrase your request.

Prompt engineering is the art and science of crafting the input (your ‘prompt’) that you give to an AI tool. The quality of the AI's response often depends on how well you phrase your prompt. The good news? You don’t need a degree in computer science to do it well. It’s like learning to ask better questions in life: clear, specific, and to the point.

The term ‘engineering’ might make it seem intimidating, but don’t let that fool you. Prompt engineering is just about learning how to communicate with AI effectively. It’s not about mastering complicated systems; it’s about trial and error, curiosity, and clear communication.

Think of it like giving directions to someone unfamiliar with your city. If they get lost, you don’t need to rewrite the map—you just need to clarify your directions. With practice, you’ll get better at predicting how the AI will interpret your prompts and tailoring them to suit your needs.

The basics: Talking to AI

Imagine you walk into a bakery and say, “I want something sweet.” The baker might hand you a cupcake, a cookie, or even a slice of pie. If you say, “I want a chocolate cupcake with sprinkles,” you’re much more likely to get exactly what you’re craving. Prompt engineering works the same way—specificity is key.

Think of AI as a chef and yourself as someone ordering a meal.

• A vague prompt is like saying, “Make me something delicious.”
• A specific prompt is like saying, “Make me spaghetti with marinara sauce, extra garlic, and a sprinkle of Parmesan.”

The more details you provide, the closer the result will be to what you’re imagining. However, you don’t have to overwhelm the chef (or the AI). Start with the essentials, and if the result isn’t quite right, refine your instructions.

For example, here are examples of a vague prompt and a specific prompt:

• Vague prompt: “Explain AI to me.”
• Specific prompt: “Explain AI in simple terms, like you’re talking to a 10-year-old, using examples from everyday life.”

The second version makes it easier for the AI to give you a focused, useful answer.

The power of iteration

Sometimes, even with clear instructions, the AI’s first response may not be perfect. That’s okay! Prompt engineering is an iterative process. If the result isn’t quite what you need, tweak the input and try again.

For instance, if you want to create a fun story, your iterative attempts might be:

• First prompt: “Write me a funny story.”

If the result is too generic, refine it:

• Refined prompt: “Write a funny story about a cat who learns to play chess.”

Still not quite there? Add more context:

• Further refined prompt: “Write a funny story about a mischievous orange cat who learns to play chess to outsmart a rival dog. Make it suitable for kids aged 8-10.”

Every tweak sharpens the AI’s understanding, much like adjusting the ingredients in a recipe until the dish tastes just right.

Tips for great prompts:

Here are some practical tips to make your prompts work for you:

1. Be clear and specific:
   • Instead of: "Tell me about history."
   • Try: "Explain the causes of World War II in simple terms."
2. Set the tone or style:
   • Instead of: "Write an article on climate change."
   • Try: "Write a conversational article on climate change that's easy for high school students to understand."
3. Use examples or context:
   • Instead of: "Summarise this text."
   • Try: "Summarise this text in bullet points for a busy professional."
4. Experiment and refine:
   • If the response feels off, adjust your prompt. Think of it like tweaking a search query on Google to get better results.
5. Think step-by-step:
   • If you need a detailed explanation, ask the AI to break it down. For example: "Explain how photosynthesis works step by step."

Real-world applications

Prompt engineering can be applied in countless ways:

• Creative writing: Generate story ideas, poetry, or scripts.
• Workplace efficiency: Draft emails, reports, or presentations faster.
• Learning and research: Get explanations, summaries, or tutorials on new topics.
• Fun and exploration: Create jokes, plan trips, or brainstorm DIY projects.

For example, a teacher could use prompt engineering to design a lesson plan by asking the AI:

“Create a 30-minute lesson plan on the water cycle for 10-year-olds, including one hands-on activity.”

Or a marketer could refine prompts to generate catchy slogans:

“Generate 10 catchy taglines for a sustainable fashion brand.”

Final thought: AI as a collaborator

At its heart, prompt engineering is about teamwork—you provide the vision, and the AI helps bring it to life. Think of it as a collaborative partner rather than a machine to command.

And remember, you don’t have to get it perfect on the first try. With practice, you’ll find that prompt engineering is less like rocket science and more like a conversation with a friend who happens to know a little bit about everything. Once you get the hang of it, you’ll wonder how you ever managed without it.