Prompting 101

Prompting refers to how users communicate with and instruct Large Language Models (LLMs) to get desired outputs. It involves crafting specific text inputs - which can range from simple questions to complex instructions with examples and context - that guide the AI in generating appropriate responses. The art and science of creating effective prompts, often called "prompt engineering," can significantly impact the quality and relevance of the LLM's output.

Prompting Formula

Prompting Best Practices

1. Be Clear and Specific: Articulate your needs clearly. Include key details for specific information requests. Vague or broad prompts can lead to generalized responses.
2. Use Concise Language: Balance detail with conciseness. Avoid overly verbose prompts to prevent confusion.
3. Provide Context: Include relevant background information for context-dependent questions or requests.
4. Structure Your Prompt Logically: Organize prompts logically, especially when asking multiple questions.
5. Specify the Desired Format: Indicate if you need responses in a specific format (list, summary, detailed explanation, guide, etc.).
6. Use Correct and Precise Language: Employ correct grammar and terminology, particularly for technical or specialized topics.
7. Set the Tone: Indicate the desired tone (professional, casual, humorous, etc.) for the response.
8. Be Ethical and Responsible: Avoid prompts leading to harmful, biased, or unethical content.
9. Iterative Approach: Refine your prompt based on initial responses for more precise information.
10. Understand the Model’s Limitations: Acknowledge LLMs' capabilities and limitations for various tasks.
11. Include Examples When Necessary: Provide examples or describe the required structure for specific styles or formats.
12. Utilize Keywords Wisely: Include relevant keywords for quicker subject matter recognition, especially for niche topics.
13. Differentiating Text with Special Characters:
    ‍
     - Quotation Marks (""): Use quotation marks to highlight specific phrases, terms, or direct questions. This can signal the LLM to focus on the quoted text.
    
     - Parentheses (): Employ parentheses to add supplementary information or clarification without disrupting the main flow of the prompt.
    
     - Brackets []: Brackets are useful for inserting editorial notes or additional context that can aid in the LLM’s understanding of the prompt.
    
     - Hyphens and Dashes –, —: Use hyphens for compound words and dashes for adding emphasis or an explanatory note within a sentence.
    ‍

By incorporating these best practices, including the effective use of special characters, you can enhance the clarity and specificity of your prompts, leading to more accurate and relevant responses from LLMs.

For more information, visit OpenAI’s articles here:

• Prompt engineering

Open AI has been very helpful in producing supporting educational information to help users better use and understand their products, such as this image they released around prompting best practices

Advanced Prompting Techniques

Click to learn more about any of the following prompting techniques to level-up your competency when interacting with LLMs like ChatGPT and understand both how to get the most out of these amazing tools, and also what to be careful of when reviewing responses for accuracy.