Models and frameworks for crafting effective prompts, which are designed to guide AI systems like GPT in generating accurate and useful responses:
| Model | Components | Description | Example |
|---|---|---|---|
| Four-Point Model | Goal, Context, Instructions, Constraints | Focuses on defining the goal, providing context, giving instructions, and setting constraints for the AI. | “Explain the impact of inflation on small businesses in under 200 words.” |
| Prompt-Response Model | Prompt, Response | Emphasizes the direct link between the prompt and the desired outcome. | Prompt: “Write a reminder email for tomorrow’s meeting.” Response: Email text. |
| RTF (Role, Task, Format) | Role, Task, Format | Specify the AI’s role, define the task, and indicate the format for the response. | Role: “HR Manager.” Task: “Write a welcome email.” Format: “Formal and concise.” |
| CRISP Model | Context, Role, Instructions, Specifics, Politeness | Provides detailed context, role for AI, specific instructions, and tone of response. | “As a financial advisor, explain retirement planning for someone in their 30s.” |
| TAP Model | Task, Audience, Purpose | Focuses on the task, audience, and purpose of the prompt to clarify expectations. | Task: “Summarize an article.” Audience: “High school students.” Purpose: “Explain basic concepts.” |
| CARA Model | Context, Action, Response, Accuracy | Ensures accurate, clear, and concise outputs by specifying the context and desired action. | Context: “You are summarizing a research paper.” Action: “Write a brief summary.” Response: “Keep it under 100 words.” Accuracy: “Ensure it’s factual.” |
| PIERCE Model | Purpose, Instructions, Examples, Reference, Constraints, Evaluation | Focuses on goal clarity, specific instructions, and measurable outcomes with clear evaluation criteria. | “Write a 300-word blog post on cloud computing, referencing the Gartner 2023 report. Keep it informative and easy to understand.” |
| SIC Model | Situation, Input, Constraints | Provides a scenario and specific inputs, along with clear constraints like length or format. | Situation: “Student writing an essay on climate change.” Input: “Provide three arguments for renewable energy.” Constraints: “Each argument under 50 words.” |
| IDEA Model | Identify, Describe, Example, Action | Identifies the task, describes the context, and provides examples with clear actions for the AI to take. | “Explain photosynthesis in simple terms and include an example of how plants convert sunlight to energy.” |
| SAFE Model | Structure, Audience, Format, Examples | Focuses on organizing the structure, audience, format, and examples to guide the AI. | Structure: “Introduction, main points, conclusion.” Audience: “College students.” Format: “Three-paragraph essay.” |
Each of these models helps ensure that the prompts given to AI systems are clear, focused, and result in useful and relevant outputs.