Master Prompt Engineering: Essential Skills for the AI Era
Dr. Hernani Costa
AI Innovator & Fractional CAIO – Helping SMBs Implement Transformative AI with Zero Regrets, while keeping Humanity in and on the loop
When I first started using LLMs like ChatGPT, I felt like I was speaking a new language—sometimes getting brilliant responses and other times hitting frustrating dead ends. Does this sound familiar?
After hundreds of hours of experimentation and countless fine-tuning, I discovered that the secret wasn't finding the perfect AI tool but mastering how I communicated with it. This skill—prompt engineering—has transformed the way I utilize AI in my personal and professional life.
What is Prompt Engineering?
Prompt engineering is the process of designing and refining the language we use to communicate with AI models. As Diane from OpenAI Academy explains, "Prompts provide ChatGPT with the relevant context, information, instructions, and examples to help it formulate the most relevant response."
In the introduction to prompt engineering, OpenAI Academy highlights that a good prompt requires three key elements: context, role, and expectations. Context grounds the model in the information needed, role allows it to adopt an appropriate persona, and expectations help it understand how to structure its output.
The Foundation: Context and Expectations
At its core, effective prompting requires just two key elements according to OpenAI's framework:
Let's look at some examples directly from the OpenAI Academy videos:
Vague prompt: "Help me study for my biology test."
Improved prompt: "I have a biology test on human anatomy in two days. Can you quiz me with 10 multiple choice questions on the circulatory and respiratory systems?"
The second prompt provides specific context (a biology test on human anatomy in two days) and clear expectations (10 multiple-choice questions on specific systems).
Here's another example:
Vague prompt: "Help me with my history research on the gold rush."
Improved prompt: "I'm writing a research paper on the impact of the gold rush on California's economy. Can you help me find three key historical events, summarize primary sources related to the topic, and suggest credible sources for further research."
The Power of Personas
Without adopting a persona, if we ask ChatGPT to "summarize perfume trends in India for the past three years", it will give us a generic high-level summary.
However, when we tell it "you are a marketing manager at a beauty company exploring new territories" and then ask it to summarize perfume trends in India, it provides a much more targeted summary with details about sustainability, ingredients, digital trends, and key industry players—all tailored to a marketing perspective.
Advanced Techniques for Better Results
Once you've mastered the basics, these advanced tactics from OpenAI's advanced prompt engineering course can help you achieve more precise and nuanced responses.
1. Use Delimiters
Create logical breaks in your prompt to highlight specific text for processing. In the OpenAI Academy example, they use quotations to highlight input text:
Translate the text delimited by triple quotes to French:
"""
[Text to translate]
"""
Delimiters (like triple quotes) help the model distinguish between instructions and content to be processed.
2. Provide Step-by-Step Instructions
Break down complex tasks into concrete steps. The OpenAI video gives this example:
You will be given text delimited by triple quotes.
1. Read the text
2. Provide feedback on grammar and structure
3. Rewrite the text with recommended edits
4. Translate the text into French and Spanish
This structured approach guides the AI through a clear process, improving results for multi-stage tasks.
3. Use Few-Shot Prompting
Few-shot prompting provides an example within your prompt for ChatGPT to emulate. From the advanced video, their example combines multiple techniques:
This gives ChatGPT a clear pattern to follow for your request.
4. Leverage ChatGPT as a Prompt Engineer
As shown in the advanced techniques video, you can ask ChatGPT itself to help write prompts. For example, you could ask it to "write a prompt that processes company 10K documentation," and it will provide a detailed, well-structured prompt you can then use or modify.
5. Improve Prompts with Additional Hacks
The OpenAI Academy videos share these additional tips:
The Iterative Process
As emphasized in the advanced prompt engineering video, prompt engineering is always a process of iteration. Their best practices summary recommends:
Looking Forward
As AI tools continue to evolve, prompt engineering skills will become increasingly valuable across virtually every profession. I've already seen how mastering these techniques has given me an edge in my work, allowing me to accomplish in minutes what used to take hours.
Remember the advice from OpenAI Academy: "If you're unsure, treat it like you're giving instructions to a human. Ultimately, you are in control, directing AI to achieve the results that you need."
What prompt engineering challenges have you encountered? What would you like to learn next about working effectively with AI? I'd love to hear your thoughts and questions in the comments below.
Related Reading
If you found this article valuable, you may also enjoy this article:
Did you find this content valuable? Let me know what AI topics you'd like me to explore in future articles!
Created by Dr. Hernani Costa, with insights from copilot Claude.