Beginners Guide to Prompt Engineering

Let me start by saying that I’m not an AI expert…yet. But I do find AI fascinating and am determined to learn more about it and utilize it in any way possible. If I’m being honest, I’ve noticed a significant shift in my thought process regarding AI as of late. Before AI was sort of an afterthought, but now it’s at the forefront of my mind and it’s the first thing I navigate to when diving into a new task. How can AI help me do this better, faster, or more efficiently? 

Understandably though, it can be overwhelming to know where to start your AI journey. My advice there is, just pick a topic and dive in. Start googling, watching YouTube videos or reading blogs. The first topic I wanted to delve deeper into was prompting because I wanted to get more out of my interactions with AI. So, let’s dive into what AI prompting is all about.

Google describes it as “the process of instructing an AI model to generate a specific type of output, such as text, images, or code. Effective prompts provide clear instructions and context, guiding the AI towards the desired result.”

Prompt Engineering

A good prompt should provide explicit, clear instructions to the model. Below I’ve outlined some of the important elements to keep in mind when developing a good prompt.

Specificity - What do you want your prompt to do? Be precise with your instructions to ensure AI understands what you want. 

Components - Background and contextual information that will help achieve accurate and impactful results. I’ll provide more specifics on individual components later in this blog.

Iteration - You won’t get exactly what you want on the first try, so experiment and refine your prompt after each iteration to optimize your output

Output - This is an important one! You as the human need to review the output. Is the information accurate? If not, let AI know about it. Were your instructions clear or did AI misinterpret what you were trying to say? This goes back to the iterative nature of prompting. 

Prompt Components

It’s important to note that there is no standard for prompt engineering, but I’ve provided some common prompt components that I’ve used in my own prompts or that I’ve come across in my research. Additionally, there is no specific order to place these elements in a prompt nor does every element listed below need to be used in a single prompt. 

• Role - Define AI’s role. What is its expertise? 

• Objective - Have a clear goal of what you are using AI to achieve. All other components should be targeted towards this objective

• Context - Set the scene….provide relevant background information and details to help AI understand its task

• Action - Describe the primary action AI should perform

• Format - Describe your desired output format. Do you want a bulleted list, essay-style response, questions with feedback, or diagram? 

• Tone - The voice and style of AI’s response. Examples include casual, formal, informative or  humorous

Practical Examples

So now that you have a better idea of what prompt engineering is, do any real-world applications come to mind? To get the creative juices flowing here are a couple examples of how I utilize prompts that are both fun and helpful.

Strength coach

I created a prompt that assigned AI the role of an experienced strength and conditioning coach. I provided context about my fitness level, available equipment, and specific goals (like "improve deadlift form" or "design a 4-week program for improved back strength"). I specified the format which included the exercise, sets, reps, and progression strategies. This saves me from having to program my own workouts and gives me personalized guidance without the cost of hiring a real coach.

Travel agent

When planning trips, I prompt AI to act as a travel specialist for specific destinations. I include context like my budget, travel dates, personal interests, and whether I'm traveling solo or with family. I ask for detailed itineraries with logistical considerations like transportation between sites and opening hours. By being specific about what I want (hidden gems over tourist traps, for example), I've discovered experiences I would have missed using standard travel guides.

Blog idea generator 

To overcome writer's block, I’ll give AI context about my audience, my expertise areas, and recent topics I've covered. I'll often specify that I want ideas that connect to current trends. I might request the output in a table format with columns for the blog title, key points to cover, and potential keywords. This approach has helped me develop content calendars that remain fresh and engaging.

Project plan development

For work projects, I’ll prompt AI with detailed context about project goals, stakeholders, available resources, and deadlines. I specify that I need a comprehensive breakdown with tasks, time estimates, responsible team members, and dependencies clearly identified. I’ll also request that critical path tasks be highlighted so I can focus on potential bottlenecks. This saves me significant planning time and helps identify risks I might have overlooked.

Final Thoughts

Prompt engineering is both an art and a science. As you begin experimenting with your own prompts, remember that practice makes progress. Start simple by defining a clear objective and gradually incorporate more components as you become comfortable with the process.

What I've found most valuable in my AI journey is the mindset of continuous iteration. My first prompts were basic questions, but over time, I've learned to craft detailed instructions that yield increasingly useful results. The beauty of prompt engineering is that you'll improve with each interaction. So, take that first step. Choose one of the examples I've shared or think of a task you'd like assistance with, and craft your first intentional prompt. Review the output, refine your approach, and watch as your AI interactions transform from basic exchanges to powerful collaborations.

Remember, you don't need to be an AI expert to benefit from these tools. You just need to start the conversation with purpose and clarity. And remember to say thank you from time to time for all the hard work your AI pal is doing for you!