Using the best prompt framework can make all the difference between getting high-quality outputs from ChatGPT and wasting time and effort only to achieve mediocre results.
An effective prompt delivers the requirements of the task in a clear and specific manner, leaves no guesswork to the AI, and is concise. The problem is, unless we are professional prompt engineers or hardcore enthusiasts, it’s common to struggle with writing such good prompts. This is where prompt frameworks come in—they provide tested and proven structures that guide you in crafting prompts that work.
And the RICE framework—Role, Instructions, Context/Constraints, and Example—stands out as the top choice. Allow us to break down each component of this prompt framework for you and demonstrate its strengths with real-life use cases.
Table Of Content
Role
“Hey, ChatGPT, Imagine You Are…”
The first component, “Role”, defines the persona or expertise you want ChatGPT to assume, such as a high school math teacher, a street artist from Turkey, or the neighbor of a cat owner. This guides the AI’s tone, depth of knowledge, and approach when it prepares responses suitable for your needs.
Typically we assign a role to ChatGPT using sentences like: “Imagine you are a [Role]…” or “Act as a [Role]…”
Include more details to make the role description more specific. To illustrate: If ChatGPT is asked to act as a high school math teacher, what kind of qualities should it have? How many years of experience should such a role have to handle your task easily? Who does it teach?
An example prompt would be:
Imagine you are a primary school English teacher from Australia, who has 5 years of experience, specializing in teaching top-performing students.
Instructions
“ChatGPT, I want you to…”
“Instructions” are clear, direct commands that tell ChatGPT exactly what you need it to do.
As always, clarity and specificity are key. You want to make sure your instructions are not vague or ambiguous, and aim to minimize misunderstandings.
Continuing from the previous example of ChatGPT acting as an English teacher, here’s an Instruction for our RICE prompt:
You will teach your students Past Tense in 3 lessons, each lasting 45 minutes. Create a lesson plan divided into 3 lessons, detailing the topic(s) to cover in each lesson.
Context/Constraints
“Here’s the situation… And these are what you should and should not do.”
Two components make up the ‘C’ part of the RICE prompt framework: “Context” and “Constraints”.
“Context” sets the stage by furnishing ChatGPT with background information. This can include details like:
- Setting the scene: “This article will be published in Nature magazine.”
- Clarifying purpose and target: “The target audience is middle-aged men with no prior gym training.”
- Providing background knowledge: “The project has already undergone an initial review, and this phase focuses on refining key details.”
“Constraints,” on the other hand, set boundaries to keep ChatGPT’s response focused and relevant. For instance:
- Defining limits: “This essay should be no longer than 400 words.”
- Avoiding unwanted content: “Do not include technical jargon.”
- Guiding style and format: “Provide your response in bullet points.”
- Restricting content: “Stick strictly to the facts.”
- Maintaining consistency: “Follow the same format used in the previous project brief.”
Examples
“For example…”
It’s often hard to describe what we want in words—whether to humans or AI. Providing examples is an easy way out. It’s an effective way to show whoever we’re communicating with what we require and saves us the trouble of finding the exact words to phrase it.
The “Example” component in the RICE framework works the same way. It helps align ChatGPT toward the style, format, or tone you desire.
Continuing from the previous English teacher example, this is how we could embed an Example in our prompt:
Refer to the lesson plan I created in the past for the format of the lesson plan.
Afterward, we can either copy and paste the previous lesson plan into ChatGPT or upload a document that contains the plan for ChatGPT’s reference.
Case Studies: Applying the RICE Framework
To keep the discussion of the RICE framework practical, we’ll demonstrate 3 real-life use cases to show you how it can be applied in everyday situations.
Case Study 1: Content Creator
Scenario: A travel blogger needs help outlining a blog post about a two-day itinerary in Singapore.
Prompt Using RICE Framework:
- Role: “Imagine you are a travel blogger with expert knowledge of Singapore, who has over 500k followers.”
- Instructions: “Create an outline for a blog post detailing a two-day trip itinerary in Singapore, including key attractions, dining recommendations, and tips for transportation.”
- Context/Constraints: “The outline should focus on a mix of popular and hidden spots that offer a unique experience. The audience is budget-conscious travelers.”
- Example: “Start with an introduction to Singapore, followed by Day 1 activities like visiting Marina Bay Sands and Day 2 focusing on Sentosa Island.”
Here’s a snapshot of ChatGPT’s response to this prompt:
Case Study 2: Study Aid
Scenario: A Computer Science student struggles to understand arrays and seeks ChatGPT’s help.
Prompt Using RICE Framework:
- Role: “You are a computer science tutor from a university, specializing in helping beginners understand programming concepts.”
- Instructions: “Explain the concept of arrays in simple terms and suggest a practical example to help illustrate how they work.”
- Context/Constraints: “The explanation should be beginner-friendly. Focus on where and how to use arrays, and how arrays differ from other data structures.”
- Example: “Use a style similar to the textbook Data Structures and Algorithms in C++/Java/Python by Michael T., when explaining this to me.”
Here’s a snapshot of ChatGPT’s response to this prompt:
Case Study 3: Cat Diet Suggestions
Scenario: A pet owner seeks advice on how to help their skinny cat gain weight.
Prompt Using RICE Framework:
- Role: “Act as a veterinary nutritionist with at least 10 years of experience specializing in cat diets, particularly focusing on weight control for cats.”
- Instructions: “Suggest a diet plan to help a skinny cat gain weight safely. The plan should include recommended foods, portion sizes, and feeding frequency.”
- Context/Constraints: “The advice should be practical, safe for cats, and feasible for implementation in a typical household. Avoid suggesting foods that are commonly known allergens or harmful.”
- Example: “For example, ‘Include wet food options like chicken pate, and consider adding small amounts of boiled chicken to meals for extra protein.’”
Here’s a snapshot of ChatGPT’s response to this prompt:
Conclusion
The RICE framework helps you craft clear, effective prompts that bring out the best in ChatGPT. It’s well-structured and straightforward, and suitable for users at any level of prompt engineering experience, whether you’re just starting out or are a seasoned expert.
If you’re already familiar with other frameworks, we encourage you to compare them with RICE. See if you agree that the RICE framework is an all-purpose solution that can be applied to all kinds of scenarios. But most importantly, compare them to find out which is the best prompt framework for you.
And if you’re new to prompting or want to master just one prompt framework, RICE is the best choice. Try it out to see if you agree.
Happy prompting!
One Comment