A TA constructs the following prompt to generate a rubric. Match each prompt component to its correct term from the “PARTS” framework in order to generate the highest quality output.
To get the best results from generative AI, you need a solid underlying framework. The PARTS method breaks down your instructions into five distinct elements, ensuring the system understands exactly what you want.
Here is the correct alignment for building your custom grading rubric:
Persona: “Act as an expert instructional designer…”
This assigns a specific role, setting the baseline expertise and professional tone for the entire output.
Action: “…generate a rubric…”
This explicitly defines the core task the system must execute.
Recipient: “…for a 300-level undergraduate research paper.”
This identifies the target audience. It calibrates the AI so the vocabulary and academic rigor match the exact needs of upper-level college students.
Theme: “The paper is a persuasive argument for or against carbon tax.”
This establishes the core subject matter, giving the AI the necessary context to populate the rubric with relevant criteria rather than generic filler.
Structure: “The paper should be scored on a 4-point-scale for the following categories: Integration of knowledge, depth of discussion, focus, spelling and grammar, sources, and citations.”
This dictates the exact format. It sets strict boundaries on how the final deliverable must be organized and presented.
By organizing your prompt this way, you eliminate guesswork. The AI knows exactly who it is acting as, what it needs to build, who it is building it for, what the topic is, and how the final product should look.