Writing a prompt is generally simple: a user asks for something, and Alix provides the answer. Yet how a prompt is written can significantly influence the quality and relevance of the responses generated by Alix.
The sections below dig deeper into best practices and practical techniques to consider when writing prompts.
Be specific and clear.
Clearly state exactly what you want Alix to do. Be as specific as possible about the action you need to help Alix focus on the right content.
When possible, avoid asking vague questions or giving open-ended instructions, as this could result in an incorrect or undesirable response.
-
General Example: "What are the main features of the latest iPhone model released in 2023?” is a specified and clear question. Whereas “Tell me about the iPhone” is general and vague on what specific information the user is seeking.
Provide context.
Provide relevant context and details, when necessary, to help Alix better understand the full picture of your request. This can include elements such as background details, the purpose of your inquiry, or any specific preferences or limitations related to your request.
While context may not always be necessary, it can help Alix generate responses that are more relevant and aligned with your needs.
Avoid details which are not relevant to your current request and don’t assume that Alix knows certain context related to your request.
-
General Example: “I’m planning a trip to Paris in December. What are some must-see attractions and tips for traveling during the winter?”
This prompt provides crucial context, such as the where and when the user is traveling which would provide a relevant and targeted response.
Keep it simple.
Use simple and direct language, focusing on a specific request. A short and well-structured prompt will guide Alix to a more focused and relevant answer. If you have a complex request, break it down into smaller, manageable prompts to help Alix tackle each request separately.
Avoid complex or long-winded phrases whenever possible, as well as ambiguous terms or jargon, as these can result in scattered or incomplete responses.
-
General Example: “Explain in detail the concepts for State of Good Repair (SoGR), including the performance measures it looks at, how performance is measured, and how to get started.”
This is a long-winded question which may not result in targeted response due to the complexity. Instead, break down each portion into its own request and make it a conversation.
Request explanations or reasoning.
If you want Alix to explain a process or provide reasoning on a particular subject, include this in your prompt.
-
General Example: "Can you explain the difference between artificial intelligence and machine learning?"
This prompt indicates the user not only wants the python code itself, but also desires an explanation of how this specific task is written in python.
Make it a conversation.
Write as if you’re speaking to another person and express thoughts in complete sentences. Use follow-up prompts to refine your request, if initial results don’t meet your expectation.
It’s an interactive and iterative process. Small changes can dramatically improve the quality of response from Alix and help fine-tune your approach to writing prompts.