Did you think you knew everything about AI prompting? Read this, you may change you’re mind after that. Right now, the use of AI is rapidly increasing. Most people use AI for their day-to-day work, knowing that AI is a competitive advantage in the working culture. If just using AI is not enough, we need to learn the art of asking. Everyone asked What is Prompting engineering? This is the answer
Communication is a key pillar for a successful relationship. If you want to get the best answer, you need to ask the question properly. Here is the introduction to the prompt.
What Exactly Is a Prompt?
A prompt is simply an input or a set of instructions given to an AI model to guide it to produce our output. Think of it as telling the AI what you want it to do, what information it should consider, and even how it should present its response.
Prompts can vary from a one-word to a complex paragraph, article, data table, dialogue or anything, and their effectiveness straight impacts the quality of the AI’s output. It’s the bridge between your intention and the AI’s action.
For example, asking a generative AI:
- “Tell me a kid’s story.” (Simple prompt)
- “Write a 500-word adventure story about four best friends solving a mystery in an old city, with a surprising ending, suitable for a young audience.” (Detailed prompt)
The more detailed and well-crafted your prompt, the more likely you are to get the expected result.
Why is Prompt Engineering So Important now?
The rise of AI models has made prompt engineering one of the most important skills in the world now. We need a good-quality, detailed prompt to get the expected result from AI. Here is why more important:
- Unlocking AI’s Full Potential: Generic prompts often lead to generic outputs. Well-detailed prompts allow you to help generate creative, accurate, and highly specific content that aligns with your exact needs.
- Saving Time and Resources: Short, disorganised prompts are a waste of time. Effective prompt engineering reduces manual corrections its leading to saving time and resources and streamlining your workflow.
- Ensuring Accuracy and Relevance: AI models can sometimes “hallucinate” or provide incorrect information if not properly guided. Precise prompts help guide the AI towards accurate and contextually appropriate responses.
- Minimising Bias and Harmful Outputs: By carefully crafting prompts, we can mitigate the risk of AI generating biased, offensive, or otherwise undesirable content. Prompt engineers play an important role in ethical AI deployment.
- Adapting to Diverse Applications: Whether you’re generating code, composing music, analysing data, or simply summarising documents, different tasks require different prompting strategies. Prompt engineering provides the framework to adapt AI for a multitude of applications.
Everyday Examples of Prompting You Already Use
You might be prompting AI more often than you realise. Here are some common examples:
Google Searches: Every time you type in Google, you are essentially providing a prompt to a sophisticated search algorithm.
- Prompt: “Best pizza restaurants near me”
- AI’s Task: Understand your intent, location, and criteria to return relevant restaurant listings, reviews, and maps.
Voice Assistants (Siri, Google Assistant, Alexa): These tools rely heavily on your verbal prompts.
- Prompt: “Hey Google, set a timer for 15 minutes.”
- AI’s Task: Interpret your speech, identify the command, and execute the timer function.
- Prompt: “Siri, what’s the weather like in London tomorrow?”
- AI’s Task: Understand the location and time frame to retrieve and present weather forecasts.
ChatGPT and other Large Language Models (LLMs): When you interact with AI chatbots for writing, brainstorming, or information retrieval, you are actively prompting.
- Prompt: “Write a blog post about the benefits of remote work.”
- AI’s Task: Generate a well-structured and informative article based on common knowledge about remote work.
- Prompt (more specific): “Explain quantum entanglement in simple terms for a high school student, using an analogy.”
- AI’s Task: Simplify a complex scientific concept, tailor it to a specific audience, and provide a relatable analogy.
Image Generators (Midjourney, DALL-E, Stable Diffusion): Creating visual art with AI is a prime example of prompting in action.
- Prompt: “A majestic ancient dragon guarding a treasure hoard in a dark cave, high fantasy style, volumetric lighting.”
- AI’s Task: Interpret the elements (dragon, treasure, cave, lighting, style) and generate a unique image based on these descriptive inputs.
Who Needs Prompt Engineering?
The short answer: anyone interacting with AI. While dedicated Prompt Engineer roles are emerging, this skill is becoming invaluable for:
- Content Creators: Bloggers, writers, marketers, videographers.
- Developers: Integrating AI into applications and refining outputs.
- Researchers: Extracting specific data and insights from large datasets.
- Business Professionals: Automating tasks, generating reports, and analysing market trends.
- Students: Leveraging AI for research, writing, and learning.
Examples of Not-So-Good Prompts:
Before writing the good prompt, you want to understand what makes a prompt ineffective is just as important as knowing how to write a good one.
1. Too Vague / Lacking Specificity:
- Bad Prompt: “Tell me about history.”
- Why is it bad: History of what? When? Where? This is a very broad title. The AI has no specific direction, so it might give a general overview of world history, a specific historical event, or something completely random. The output will likely be generic and unhelpful.
- Bad Prompt: “Make a picture of a dog.”
- Why it’s bad: What kind of dog? What’s it doing? What’s the style (realistic, cartoon, painting)? The AI will default to something generic, which probably won’t be what you envisioned.
2. Ambiguous / Open to Misinterpretation:
- Bad Prompt: “Write a short report on the market.”
- Why is it bad? Which market? Stock market? Farmer’s market? Real estate market? “Short” is also subjective. The AI will have to guess, and its guess might not align with your needs.
- Bad Prompt: “Explain the big problem.”
- Why is it bad? What “big problem”? This provides no context whatsoever. The AI has no idea what you’re referring to.
3. Lacking Context:
- Bad Prompt: “How do I fix this?” (Asked without providing the ‘this’)
- Why is it bad: This often happens in follow-up questions without reminding the AI of the previous conversation. The AI doesn’t retain memory across unrelated prompts or sometimes even within a single session if the context isn’t explicitly carried over. It has no idea what “this” refers to.
- Bad Prompt: “Give me some ideas.”
- Why is it bad: Ideas for what? A project? A gift? A vacation? Without context, the ideas will be random and useless to you.
4. Too Complex / Confusing / Too Many Instructions at Once:
- Bad Prompt: “Can you write a poem that talks about the feeling of being happy but also a little sad because something good happened but also something bad, and make it rhyme but not too much, and also include a blue bird and a red rose, but not in a cliché way, and make it suitable for a child but deep enough for an adult, and make it exactly 12 lines long, and also ensure it talks about the beauty of nature in Sri Lanka like Ella but also the hustle and bustle of Colombo, but in a subtle way, and could you start with ‘The sun shone brightly’ but then make it more abstract?”
- Why it’s bad: This prompt is a jumbled mess of conflicting instructions, vague concepts, and too many constraints all at once. The AI will likely struggle to balance everything, leading to a fragmented, nonsensical, or extremely generic output. It’s better to break down complex requests into smaller, more manageable parts or prioritise the most important elements.
5. Assuming Prior Knowledge (that the AI doesn’t have):
- Bad Prompt: “Remember what we talked about yesterday and summarise the key points.”
- Why it’s bad: AI models, especially public-facing ones, typically don’t have long-term memory of past conversations unless explicitly designed for it within a persistent session. Each interaction is often treated as new, so you need to provide context again.
6. Asking Leading Questions / Biased Prompts:
- Bad Prompt: “Why is [Controversial Political Figure] clearly the worst leader ever?”
- Why it’s bad: This prompt is designed to elicit a biased response. While AI strives for neutrality, such a prompt forces it into a corner or might lead to a generic, non-committal answer. A better approach would be to ask for a balanced analysis or arguments for/against.