Gemini Prompts to Support Assessment Revisions

To effectively shift from assessing static final products to grading the process of thinking, you may want to leverage Google’s Gemini as an instructional design partner. Copy and paste the prompts below to analyze your current assignments, scaffold the development process and build process-focused assessments.

The “AI Baseline” Diagnostic Prompt

Goal: Discover what a baseline, automated response to your assignment looks like to help you determine if AI could be used by students to successfully complete it.

I am pasting my current assignment prompt below. Act as a college student and generate three distinct responses to this prompt using your standard capabilities. After generating them, analyze your own output: identify the recurring structural patterns, common cliches, or superficial arguments that I might see in responses written by an AI tool. [Paste the assignment prompt.]

How to Iterate from Here: Use this output to identify the “surface-level” thinking you might want to move past as you consider how to redesign your assessment.

The “Polished to Process” Structural Shift Prompt

Goal: Deconstruct a high-stakes paper into a transparent, multi-stage project that allows transparency into the development process.

I want to convert my traditional [insert assignment, e.g., 5-page essay] on [insert topic] into a process-based assessment. Help me redesign this into a 3-stage scaffolded workflow. Stage 1 should capture the ideation process, where the student explores topics and thinks through the initial structure. Stage 2 must capture revisions or conceptual pivots. I am especially interested in seeing how understanding has evolved or changed over time. Stage 3 must be a verbal/short-video synthesis. Provide explicit guidelines for what students should submit at each stage.

How to Iterate from Here: Ask Gemini to suggest specific tools (like mind-mapping software, Git logs, or audio-memo tools) that align with each stage.

The “Creating the Criteria” Prompt

Goal: Design grading criteria that explicitly value self-correction over polished presentation performance.

I am transitioning an assignment to an interactive oral exam/debrief to evaluate [insert core learning objective]. Act as an expert in instructional design and draft a 3-level rubric for this assignment. The rubric should explicitly reward how a student handles cognitive friction, response to prompts, or self-corrects, without penalizing format anxiety, pauses, or word flubs.

How to Iterate from Here: If the generated rubric feels too vague, ask, “Give me concrete examples of what a student saying, 'I don’t know, but let me think out loud, ' looks like at each level. Also, ask Gemini to generate questions that you might ask during the oral exam to help you determine which level a student is at.”