The use of generative AI tools in coursework

We have prepared some information and example instructions to students relating to the use of ChatGPT and other generative AI tools that teachers can adapt for their course.

From the Education Office’s bulletin, read the University’s response to the emergence of generative artificial intelligence (AI) tools in education. AI is moving at a fast pace so please consider that these suggestions are a moment in time.

What you should know

AI is here to stay, Prof Michael Witbrock (NAOInstitute) says we are still in early days of Artificial Intelligence, the output of LLM like ChatGPT gives no reflection or reconsideration to output but many Labs are working on far more complex reasoning, including scientific and legal reasoning. AI systems are going to become more competent quite rapidly.

There is no right answer or model for the use of generative AI tools. Each discipline and course is different and will need to decide how and what the best use of AI is.

Jason Lodge et al., suggests six options for redesigning assessment with two options, embrace and rethink being viable for the medium to long term.

Students will need to understand how to use AI for future employment, it has been suggested that AI may not replace a role but a person with AI skills will. Consider how AI is being used in the workplace and bring those skills into the course for students to learn.

AI will adapt and improve the more we use it. The language that it produces currently lacks human emotion or rationality. It can appear to reflect human thinking and therefore it is important to have an open discussion about the affordances and limitations of AI tools such as ChatGPT. Remind students not to rely on any information given by the tool.

What you should do now

Consider how AI is being used in the workplace within you discipline or industry and how you will incorporate and teach this in your course/programme.
Clearly communicate to students the decision and expectation of how AI should be used during your course. Use the templates found on the academic honesty declaration page to create your own instructions to students on the use of ChatGPT and generative AI.
Read the University’s guidelines for permitted use of software in assessment activities.
Consider the purpose of assessment, the ‘why’. Effective assessment design in the age of GenAI still focuses on principles of good assessment design.
Consider programmatic assessment, an arrangement of different assessment methods deliberately designed across the entire curriculum.
Decide what matters most in your discipline and what learning and skills students should be graduating with. Do the learning outcomes align with these skills, and therefore which assessment should be secure and which could be teaching collaboration and human skills.
Consider assessments that value and mark the learning journey rather than the end goal. Encourage feedback literacy in students. Boud and Carless (2018) Feedback literacy is based on social constructivist theory and focuses on students learning and sense making.
Danny Liu and Adam Bridgeman (UoS) make assessment suggestions that will have longevity even as AI advances. This article suggests to consider the humanness of teaching and learning. A full guide is available for download.

Examples on how teaching staff are incorporating AI into coursework and assessments

Dr Ruth Dimes & Prof Charl de Villiers

ACCTG 780: Sustainability Accounting and Integrated Reporting

Fostering critical evaluation in sustainable accounting assessment by integrating ChatGPT

In response to the launch of ChatGPT in November 2022, Ruth Dimes and Charl de Villiers redesigned the assessments to allow students to engage with generative AI and develop their critical evaluation skills in sustainability accounting and integrated reporting.

In this course students practise submitting prompts to Chat GPT then revising their prompts, to gradually achieve a satisfactory outcome in the form of a CEO report. The student generated reports are then combined with real reports and other AI generated ones, which are then critiqued and subject to peer review.

Each assessment has relevance to the finance and corporate industry and incorporating AI into the assessment prepares students for real world tasks that they will be expected to complete. It engages students critical evaluation and analysis skills. The assessments have a mixture of low and high stakes grading enabling students to receive feedback and build upon their learning as they progress through the course.

Dr Mohsen Mohammadzadeh

Urban Planning Programme

Assessments adaptation and innovation in the age of disruptive AI technologies

As a student, engineer, and educator, Mohsen Mohammadzadeh noticed how tech changes transformed education, work, and daily life. He’s curious about both the good and bad effects of new technologies, like Gen-AI. This curiosity is important for his Urban Planning courses, where he teaches Policy-making, Smart Cities, and Planning Theory. Dr. Mohsen uses his academic and professional network to stay updated and improve his teaching and pedagogies using new technology advancements. He adjusted how he teaches to match and dedicates lectures to technology’s philosophy and ethics. Assessments have also been redesigned to:

Using formative assessments instead of summative assessments.
Diversifying assessments, including quizzes, class tasks, group work presentations and video recordings, and individual practice-based projects.
Focusing on crucial skills such as problem solving, critical thinking, innovation, and creativity. Carefully write Learning Outcomes to align with the skills being assessed.
Designing the assignments and assessments as a process, considering the steps of the project, rather than solely focusing on the final work submission.

Dr Elham Bahmanteymouri

Urban Land Economics, Urban Economic Development and Urban Economics

Innovative pedagogies: adapting to change and embracing technology

Elham Bahmanteymouri, an economist, believes innovation and technology are vital amid economic crises. She adapts her teaching to technology advancements and stays updated on teaching methods, industry changes, and ethics. Her course strategies are:

Prioritising ongoing improvement and continuous feedback through formative assessments.
Diverse assessments such as quizzes, tasks, group and individual projects to allow students to engage with different aspects of the subject matter and develop diverse skills.
Using discussions, presentations and teamwork for AI-integrated assessments to encourage critical thinking.
Highlighting AI as a tool, not a replacement for deep thinking.
Students share how AI influenced their assignments.
Focusing on problem-solving, critical thinking, innovation, and creativity skills.
Design assignments as step-by-step processes to create final work submission which nurture continuous learning.
Using university technology tools for digital literacy and Gen-AI readiness.
Encouraging ChatGPT use for research, with a reminder to verify information from reputable sources to ensure accuracy and reliability.

Assoc Prof Ethan Cochrane

ANTHRO 306: Pacific Archaeology

Adapting Anthropology assessments with Gen-AI in mind

Ethan Cochrane has designed ANTHRO 306 using Bloom’s knowledge taxonomy. The first third of the course builds a basic foundation of the facts of Pacific archaeology. The second third is about comparisons with these data, and the final third focuses on evaluating knowledge claims.

The final assessment is an essay on best practice for evaluating competing knowledge claims using archaeological debates and literature discussed in class. For this essay students must also read and annotate assigned texts prior to class discussion. Their reading and engagement with the texts is automatically marked through the Perusall app within Canvas. Ethan’s “real world” goals for this essay, and other assessments, are plainly stated in the assessment briefs. For example, considering the final assessment, “You may not write many long-form essays outside of university, but in the professional world you will often have to aggregate different kinds of information – verbal, written, audio-visual – into a cohesive picture that supports a particular position. Ultimately, this is what you are learning to do here.”

Ethan last taught this course and used this assessment in 2020, before the proliferation of Gen-AI writing, and Perusall software for group-reading and automatic marking of student annotations. This year he has made a more specific requirement for the final essay to lessen dependence on Gen-AI: students must reference in-class discussions. He does not provide lecture recording of these discussions, requiring students to engage in face-to-face conversation, in real-time, and with the physicality and emotional appeals this can involve. He encourages students to consider that this too will benefit them in the world of work (and life in general).

Finally, Ethan uses ChatGPT in class to produce a final essay for this assessment to teach about the overly general responses made by Gen-AI text relative to the assigned readings and class discussions.

Case studies in teaching

How to make ChatGPT work as a teaching assistant: a case study in law

CASE STUDY – Dr Benjamin Liu provides a guide to working with AI writing tool GPT-4, to train it as a useful teaching assistant for answering students’ questions.

View all case studies

Stay informed