KOSUNI KOREA
Korean News in English
Technology

3 Common Mistakes to Avoid When Writing ChatGPT Prompts

Published September 30, 2026 · 2 views
3 Common Mistakes to Avoid When Writing ChatGPT Prompts

Avoiding Vague Instructions: The Importance of Specificity and Repetition in Prompt Writing

When writing ChatGPT prompts, one common mistake is using vague language and overly long explanations. OpenAI's 'correct formula' emphasizes that the repetition structure matters more than the prompt itself. Users have found that providing instructions as if explaining to a junior colleague yields better results.

ChatGPT 프롬프트 작성 시 주의해야 할 3가지 오류 설명 이미지

OpenAI's Correct Formula: Repetition Structure Over Prompt Content

In April 2023, OpenAI revealed its 'correct formula' for prompt writing, stressing that the repetition structure is more important than the prompt itself. This means it's not about solving everything in one prompt but requires multiple rounds of feedback and adjustments to achieve the desired outcome.

Users applying the principle of 'don't write long explanations; define completion criteria first' have experienced time savings of over 30%. For example, when requesting a news summary, specifying 'summarize this in three sentences, highlighting only key points' yields more accurate results.

On March 23, 2023, ChatGPT experienced an error where other users' conversation histories were leaked. This issue highlighted the importance of user data protection, leading OpenAI to implement enhanced security measures afterward.

On January 8, 2024, AI bias was reported. ChatGPT generated results showing gender discrimination in some cases, suggesting that gender-related terms should be avoided when writing prompts. For instance, avoid 'female office worker' and instead use neutral terms like 'competent professional'.

Practical Use Cases: News Summaries, Pharmacy POPs, Development Support

The AI Korea community newsletter shared techniques for using ChatGPT to summarize news articles. An example prompt was provided: 'Summarize this article in three sentences, highlighting key figures and events.'

Pharmacy Forum shared how to generate pharmacy shelf price tags with ChatGPT. The sample prompt included specific information and design instructions: 'The price of this medication is 15,000 won, with main ingredient acetaminophen. Create a concise, visually appealing POP.'

A developer on Brunch shared how they use ChatGPT for work efficiency. For example, requesting 'Fix the bug in this code and suggest an optimized alternative' significantly improves development productivity.

Core Principles of Prompt Engineering: Specific Instructions, Reasoning Strength, Token Optimization

Three key principles should be followed when writing ChatGPT prompts. First, specific and clear instructions. Instead of 'Summarize this document,' use 'Summarize in three sentences, highlighting only key content.'

Second, adjust reasoning strength. In GPT-5.6, adjusting the default reasoning strength can reduce costs while improving result accuracy. For example, specifying 'Increase reasoning strength' leads to more reliable responses.

Third, optimize token usage. ChatGPT and other AI models like Claude have varying costs depending on the question, explained by 'token output disparity.' Thus, keeping prompts concise saves costs.

Response and Safety Information: Addressing Prompt Injection Attacks

OpenAI announced on December 22, 2025, continuous enhancements to ChatGPT Atlas for addressing prompt injection attacks. This security feature was developed to prevent malicious prompts from affecting the system.

On June 23, 2025, Brunch introduced three effective prompt engineering methods: 'Don't write long explanations; define completion criteria first,' 'Write as if explaining to a junior colleague,' and 'Use repetition structure to improve results.'

Important Considerations: Medical Work, Counseling, Senior Users

Caution is needed when using ChatGPT for medical work. A DailyMedi article on April 28, 2025, warned that medical information should only be used after consulting professionals.

Before sharing personal concerns with ChatGPT, four precautions are recommended. A Hankyoreh article (August 19, 2025) warned about 'risks of personal data leakage,' 'AI bias,' 'over-reliance,' and 'inaccurate advice.'

Guides for senior users also exist. An OMATE Senior article on August 11, 2025, provided tips including 'simple prompt writing,' 'managing conversation history,' and 'caution with medical information usage.'

Frequently Asked Questions

What is the most important factor when writing ChatGPT prompts?

Specific and clear instructions. Instead of 'Summarize this document,' specify 'Summarize in three sentences, highlighting only key content.'

What should you do if ChatGPT generates gender-discriminatory results?

Avoid gender-related terms in prompts and use neutral alternatives like 'competent professional.' If issues occur, revise and resubmit the prompt.

How can you reduce costs when writing prompts?

Optimize token usage by following the principle of 'don't write long explanations; define completion criteria first.' Concise and specific prompts save cost.

Frequently Asked Questions

Writing ChatGPT prompts goes beyond simple Q&A—it requires strategic thinking to achieve efficient results. OpenAI's 'correct formula' emphasizes the importance of repetition structure over prompt content, with users reporting time savings exceeding 30% by applying this approach.

Practical applications include news summaries, pharmacy POPs, and development support across various fields. However, caution is needed when using medical information, and avoiding gender-related terms helps prevent AI bias issues.

Prompt writing is an iterative process. If initial results aren't satisfactory, use specific instructions and repeated adjustments to achieve the best outcomes. This isn't just a technical skill but a core strategy for effectively collaborating with ChatGPT.

This article was written based on publicly available information. Note that all data referenced here is current as of 2026.

View More Related Articles

You can find similar articles in one place.

View Related Articles

FAQ

What is the most important factor when writing ChatGPT prompts?

Specific and clear instructions. Instead of 'Summarize this document,' specify 'Summarize in three sentences, highlighting only key content.'

What should you do if ChatGPT generates gender-discriminatory results?

Avoid gender-related terms in prompts and use neutral alternatives like 'competent professional.' If issues occur, revise and resubmit the prompt.

How can you reduce costs when writing prompts?

Optimize token usage by following the principle of 'don't write long explanations; define completion criteria first.' Concise and specific prompts save cost.

Korean Source

This article is an English localization of a Korean-language report from BlogScore. Original: ChatGPT 프롬프트 작성 시 주의해야 할 3가지 오류