Image generated using Flux AI (Dev)

Purpose

A key part of assessment is the learning that happens through feedback. Feedback needs to be understandable by students and help them know how and where they can improve. A number of educators struggle with feedback anxiety and can find it tricky to phrase feedback in the right way that is most supportive for student learning. AI may be able to help here, by turning your briefly-worded feedback into more supportive and specific comments for students.

At the University of Sydney, this is an allowed use of AI for marking and feedback by educators.

Building a custom AI agent to expand marker feedback

A Cogniti AI agent could be built to help. The agent can be provided the assignment instructions, marking rubric, samples of good feedback comments, and other context. Once built, the agent can be used by markers to support them in writing better feedback.

The screenshot below shows such an interaction with a custom AI agent.

Prompt to use in general-purpose AI tools

Note: The following prompts were tested in Microsoft Copilot Chat. They are most suitable for language-based generative AI.

Act as an expert educator in higher education.

I will give you a summary of notes for feedback to students about an assignment. I want you to expand these into a paragraph or two of feedback to students. The feedback needs to be encouraging and help students understand how they can move forward.

Base your feedback on the **assignment brief** as well as the **marking rubric**.

Acknowledge that you understand this and have received the assignment brief and rubric. Wait for me to provide the summary of notes. When I provide the summary, you should reply with the expanded feedback. Keep your feedback a maximum of 2 paragraphs.

## Assignment brief:

<paste the assignment instructions here>

## Marking rubric:

<paste the marking rubric here, i.e. what the descriptor is for each criterion and standard>

Explanation

This prompt gives the AI a role (expert educator), a task (to expand short-form feedback), requirements (be encouraging, what the brief and rubric are, etc.), and instructions (how to interact with you).

Examples

The screenshot shows Microsoft Copilot Chat expanding upon short-form feedback. The screenshot shows the end of the initial prompt that sets the context, followed by the AI indicating that it is ready to accept the short-form feedback, and then the educator providing the comments which the AI subsequent helps to elaborate upon.