How UNSW academics are using GenAI to streamline rubric creation.
UNSW Nexus Scholars have created a series of prompts to help academic staff use GenAI tools for assessment rubric design. It’s one of many ways GenAI may help ease workloads in routine parts of the assessment process.
The Nexus Program was created to elevate UNSW’s approach to education. Part of that work has involved considering how we might embrace the ethical use of GenAI to innovate in teaching and learning.
A strength of GenAI is its capacity to synthesise multiple information sources to accelerate workflows, particularly for routine tasks such as rubric design.
Nexus Fellows Dr Cherie Lucas (UNSW Medicine & Health), Dr Mark Ian Jones (UNSW Arts, Design & Architecture), Dr Helena Pacitti (UNSW Science) and Dr Chis Campbell (UNSW Canberra) have been working with various GenAI agents to explore their potential in this area.
Saving time during rubric creation
Rubrics establish the expectations for each grading level for an assessment, but creating an effective rubric that aligns with Course Learning Outcomes (CLOs) can be time-consuming.
The Nexus Scholars found GenAI agents such as Copilot and ChatGPT can take on the manual work of matching CLOs with an assessment task to establish an initial benchmark for each grade against the criteria.
After creating a draft rubric using GenAI, academics can add further prompts to refine the output before it is edited and peer reviewed.
“It creates a starting point,” said Cherie. “The academics can then fill in the gaps.”
The Nexus Fellows believe academic staff are well placed to get effective outputs when using GenAI to create draft rubrics, which they then refine.
“As a content expert, you know what you want your students to learn. You know your course, the learning outcomes and what sort of assessments are driving that outcome. So, you can tailor the rubric to your needs once you’ve used some generic prompts,” Cherie said.
Iterative prompting for best results
Generative AI relies on effective prompting and access to relevant information to create useful outputs. The Nexus Scholars have found iterative prompting to be most effective in creating and refining AI-generated rubrics that can then be edited by humans.
“In my experience the agent cannot get to everything I require with one single prompt. It’s better to do it in stages. It can take six or seven sequential prompts to get a rubric that I’m happy with,” said Cherie.
The Nexus Fellows’ experience has led them to recommend a series of steps for prompting GenAI agents to produce rubrics.
To effectively prompt and receive the best output, academics need to share the course outline, assessment instructions and details of the course cohort – including diversity and stage in their program – with the GenAI agent.
Cherie noted the impact of training on how different GenAI agents respond to prompts.
“Public agents are being trained by members of the public, so there might be different outputs and different biases that emerge. At UNSW, we’re lucky to have licences for Copilot and an OpenAI ChatGPT trial, so those agents are being trained by academics within the University.”
The prompts designed by the Nexus Fellows serve as a starting point and can be adapted to suit individual course requirements.
“You can develop fantastic prompts for your specific course and tailor the prompts and the outputs to your course or program,” said Cherie.
As with all GenAI outputs, human oversight and editing are critical to ensure accuracy and the appropriateness of the rubric.
“As you get the rubric output, you need to critique it,” Cherie said. “You're not just accepting it. I’m always thinking, how can this be better or more succinct?”
Peer review can be an important part of that process.
“I have my rubrics peer reviewed, whether that’s internally within the school or externally,” she said.
Modelling ethical AI use for our students
In addition to saving time, the Nexus Fellows believe the use of GenAI for rubric design can serve as a way of modelling the ethical use of GenAI for students.
“When I use GenAI to develop a rubric, I acknowledge that I co-designed it with a specific GenAI tool with prompts that I authored. We're embracing GenAI, and we’re modelling it, as well as acknowledging it,” said Cherie.
The Nexus Fellows’ rubric design prompts are a starting point, Cherie noted, and will likely evolve over time.
“We're all learning and evolving. These prompts may change down the track and be developed further. We’re also modelling that we’re lifelong learners.”
Watch a webinar from Nexus Fellow Dr Helena Pacitti on using GenAI for rubric design.
Lucas C, Jones MI, Pacitti H, Campbell C, Cain J (2025). Considerations for Designing Rubrics with Artificial Intelligence in Pharmacy Education PulsesRx. https://pulsesrx.com/2025/03/04/considerations-for-designing-rubrics-with-artificial-intelligence-in-pharmacy-education/
Lucas C, Jones MI, Campbell C, Pacitti H (2024). Ethical use of Gen-AI for designing rubrics at UNSW: Nexus Fellows’ perspectives. https://www.education.unsw.edu.au/news-events/news/ethical-use-gen-ai-designing-rubrics
- Log in to post comments