How to Write and Refine ChatGPT Prompts to Get Best Results

Using ChatGPT centers on the act of writing prompts, which are directions you give ChatGPT to guide its responses. The higher the quality of prompts you feed ChatGPT, the better the work it can do for you.

Editors who want to use ChatGPT to speed up and simplify their work may begin by simply asking ChatGPT to “edit this.” This broad prompt will give you a result of edited text, but the result may have more and different changes than you seek. Instead, giving the AI more directions will allow you to fine-tune the response to be more closely aligned with your goal.

Keep in mind that even with a well-phrased prompt, you will likely need to give ChatGPT follow-up prompts to help it dial in on your specific needs. This will become a routine part of your process.

ChatGPT Prompt-Writing Basics

There are four core best practices for writing an effective ChatGPT prompt. It all comes down to being clear and specific.

1. Use Statements

Format instructions as statements instead of questions. This helps ChatGPT determine what you want it to do.

❌ "Can you write a paragraph explaining first-person point of view?"

🟢 "Write a paragraph explaining first-person point of view."

2. Be Specific

Make your requests detailed. Tell ChatGPT its exact next steps.

❌ "Edit this."

🟢 "Edit this corporate report to improve clarity and consistency and to correct errors of grammar, spelling, and punctuation."

3. Define Format

Specify format requirements in your prompt.

❌ "Write a blog post about the history of singular they."

🟢 "Write a 1,500–word blog post about the history of singular they. Use subheadings and bullet points. Use The Chicago Manual of Style."

4. Provide Context

Include relevant background and nuance. This helps ChatGPT narrow its response.

❌ "Is this a good title?"

🟢 "Considering this is a thriller novel set in Victorian England, is The Shadow's Grasp a fitting title?"

Refining Initial Responses

Even with a well-crafted prompt, the first output from ChatGPT often won’t meet your exact needs. The skillful user engages ChatGPT in a back-and-forth to tweak the initial response.

Here are examples of how to ask ChatGPT to refine its output:

1. Specify Changes

Clearly explain how you want the initial response to be changed.

• "Refine this response to use title case in headings instead of sentence case."

• "Change the examples to be relevant to podiatrists."

2. Identify Errors

Being a new technology, ChatGPT requires a bit of patience. It does not respond with accuracy every time. When using it to edit text, this can look like the program not catching every error. When using it to write, this can look like ChatGPT replying with information that is factually incorrect.

Rather than manually correcting errors ChatGPT misses (the whole point is to save you time and wear on your hands), explain the errors and instruct it to fix them.

• "Refine this response to ensure all numbers 1 through 9 are written as numerals."

• "Revise by placing the commas before closing quotation marks, not after."

3. Adjust Length

If you prompt ChatGPT to write text and the first response is too long or short, ask it to adjust the length.

• "Add a section about how about developmental editors can use ChatGPT."

• "Condense your response into one paragraph."

Try and Try Again

If your attempt to refine ChatGPT’s response still does not give you what you want, keep trying. Rephrase your prompt and try to make your directions more specific. Or open up a new chat and try the original prompt again. ChatGPT won’t answer in the exact same way twice, and sometimes starting with this clean slate is quicker than refining and refining a response that went off the path.

Want to learn more about ChatGPT?

Sign up for the ChatGPT for Editors course.

RSVP: ChatGPT for Editors

Erin Servais is an editor, educator, and community builder. She founded the Editors Tea Club, an online space where editors gather for education and support, and she offers editor coaching through her company, Dot and Dash.

In her fifteen years in publishing, she has helped to bring hundreds of titles to print and has presented about editing and entrepreneurship across the United States and Canada. Erin serves on the board of directors for ACES: The Society for Editing.

She always tells ChatGPT please and thank you, just in case.

Email Erin: Erin@aiforeditors.com

The Editor's AI Dictionary: Key Terms You Need to Know

Artificial Intelligence is transforming the editing profession. It's important to know tech jargon to stay relevant and understand the changes happening around you in the editing world.

That's why I've crafted The Editor's AI Dictionary—to demystify AI vocabulary and help you navigate the crossroads of technology and editing. In this guide, we'll break down the core AI terms into easy-to-understand definitions and include real-world examples of AI you likely already use in your everyday life.

Ready to crack the code? Let's dive in.

Artificial Intelligence

Computer systems that can perform tasks typically requiring human intelligence, such as understanding natural language, recognizing patterns, and adapting to new situations.

Example: smart thermostat that automatically adjusts temperature

Model

A single AI system

Example: Google’s Gemini

Machine Learning

A type of artificial intelligence that allows systems to learn and improve from experience without being programmed with specific instructions.

Example: facial recognition

Neural Network

A computing system inspired by the structure of the human brain. It consists of a collection of nodes, or "neurons," and the connections between them. These networks are key components of many machine learning models and can recognize patterns and relationships in data.

Example: handwriting recognition for check processing

RSVP: AI for Editors Course

Generative AI

A type of Artificial Intelligence that makes new content by drawing on what it has previously learned from existing content.

Example: Midjourney, which creates an image from a text prompt

Language Model / Large Language Model

A type of Artificial Intelligence model that can understand and generate text in a human-like manner. It learns patterns in language. Then, when given a piece of text, it predicts what comes next. A Large Language Model is trained on a bigger amount of text/information and has a better understanding of nuanced language and reasoning.

Example: predictive text on iPhone texting, ChatGPT

Generative Pretrained Transformer (GPT)

The type of artificial intelligence tool ChatGPT is. “Generative” means it can create data. "Pretrained" means it has already learned from a vast amount of text. "Transformer" means the type of neural network it uses, which is good at understanding relationships between sentences.

Example: ChatGPT

Chatbot

A computer program that lets people talk to a Large Language Model in a conversational manner via text or voice. This often looks like a person asking the chatbot a question, and the chatbot answering with the requested information.

Example: Amazon's Alexa

Prompt/Input

A piece of text given to a Large Language Model for it to respond to.

Example: “What are the arguments for and against having a rabbit as a pet?”

Response/Output

The text the Large Language Model generates in reply to the prompt.

Example: “Here are the arguments for and against having a rabbit as a pet. Pro: 1) They don’t require walks...”

Hallucination

When a chatbot answers with false information.

Example: an attorney who asked ChatGPT for legal cases, which he then used in court filings, not knowing ChatGPT invented the cases and they did not exist

Sign Up for the Course

Erin Servais teaches editors to upskill with artificial intelligence through her AI for Editors courses and corporate training. She is known worldwide as the leader in AI education for editors, writers, and other publishing professionals.

Email Erin: Erin@aiforeditors.com