What are some general best practices for creating prompts?

Best Practice	Description
Start simple and iterate	Prompt engineering is an iterative process that starts with a simple prompt, evaluates the AI's response, and then adds more details to get better results.
Experiment with the elements of prompt structures	Responses are highly sensitive to small changes in prompt structure. To improve your results, experiment with these simple structural elements to see how they affect the model's response: Word order: Changing the order of words can generate different responses. Line spacing: Adding or removing line breaks can affect the output. Placement: Information placed later in the prompt might hold more weight with some models.
Use detailed commands	Prompts are effective when you use commands to instruct the model for the exact task you're asking the model to achieve, such as Write, Classify, Summarize, Translate, Order, and so on.
Be specific	Specify the instruction and task you want the model to perform. Descriptive and detailed prompts usually generate better results. Being specific is crucial when aiming for a particular result or desired style of generating responses.
Be mindful of length	Aim for a balance between simplicity and detail. Ensure all information is relevant to the task and avoid unnecessary details.
Specify the output	Specify the exact output format needed for Oracle Fusion Cloud Applications.
Avoid using personally identifiable information (PII)	PII shouldn't be used for privacy and security reasons.
Test your prompts	Experiment with various phrasings or versions of the same prompt. Include both happy and sad case data (inputs designed to challenge the model) to ensure it stays on topic.