Ethical Considerations of Using AI in Tertiary Education

Posted on: August 29, 2023

This opinion piece was contributed by the authors of Data Science and Analytics Strategy: An Emergent Design Approach: Kailash Awati, data and sensemaking professional and Adjunct Fellow in Human-Centre Data Science at the UTS Connected Intelligence Centre and Alex Scriven, senior data scientist at Atlassian in Sydney, Australia. 

How do you envision technology, such as generative AI, being integrated into tertiary teaching? And what impact do you think this will have on educators?

AI will have a massive impact on educators as they will have to rethink how they teach and assess students. Many educators have already started exploring ways to introduce AI into their classrooms. An early pioneer is Ethan Mollick at Wharton Business School who uses it not only to generate course and lesson plans, but also mandates that his students use it in the courses that he teaches. Professor Simon Buckingham Shum from the UTS Connected Intelligence Centre (which we are associated with) recently gave an excellent overview of ChatGPT from an educator's perspective. 

From a student perspective, there are myriad ways in which generative technologies can enhance learning. The most obvious is the generation of whole or part responses to tasks. Students can brainstorm ideas, be provided a TOC or sections, and then iteratively generate content for each of the sections. The ability of these tools to adopt a persona also opens possibilities for tutoring interactions whereby the tool can assist the student to learn a concept in different ways (for example, “explain quantum entanglement to a high school student using non-technical language” or “explain quantum mechanics to a second year physics undergraduate”). However, one has to keep in mind that the tool does not automatically abide by good tutoring practices (e.g not simply giving answers but guiding, remembering student’s abilities and concerns etc). 

For educators, there is great opportunity but also great risk. Given the ability of generative AI to generate comprehensive responses to direct queries by regurgitating facts, educators will need to think carefully about how to use these tools. In our opinion, these tools, appropriately used, can enable students to pick up higher order skills such as critical thinking that have been traditionally hard to teach. This will require teachers to completely rethink the way they teach and think about homework tasks and assessments. ChatGPT renders many traditional student tasks and assessments such essays obsolete. However, as noted earlier, it opens up new opportunities to integrate generative AI into learning (see some examples from our university here).

Interestingly, a recent University of Melbourne study suggests that though students see the potential of AI to enhance their learning, a significant proportion are wary of using it. We think this may be due to lack of clear guidance (from their universities). It is imperative that universities publish clear guidance on how these technologies should be used. We address this in our response to the next question.

It is predicted that AI will disrupt the workforce in the next five years. What impact do you think this has on student learning now? And if technology continues to change how jobs are performed, what role do tertiary institutions play in preparing students for the future?

As noted in the previous response, it is critical that generative AI technologies such as ChatGPT be integrated into tertiary teaching. The challenge for academics is to integrate these technologies in classes in ways that reflect the manner in which they are (and will be used) in different professions. Anything less would be short-changing students who will be expected to use these technologies at their future workplaces. We think this will involve experimentation with different approaches to using generative AI in classes. Different disciplines will use it differently - for example, a coding class may use ChatGPT as a tutor to identify errors and guide students towards a solution to the programming challenge; whereas a class on creative writing may use it as an idea generator. 

The value that tertiary institutions can bring is in showing how to apply the technology to specific industry contexts and tasks, whilst guiding students on how to extract the most value out of the tools. As a simple example: what are the best prompts which will balance the risk of a model providing incorrect answers, whilst allowing it the freedom to provide interesting and useful responses? Such considerations fall in the domain of the nascent field of prompt engineering. The issue of academic integrity is a tricky one as it can be hard to detect the extent to which an assessment turned in by a student is AI-assisted. To this end, our university (UTS) has recently published 5 student-centred principles for effective and ethical use of generative AI. Different industries and contexts will require different approaches (language, risk, and ethical considerations), which the tool itself cannot provide by design. Good educators will be guides to using these tools, bringing academic rigour, industry knowledge and ethical lenses to the students' work. These are early days and we think there are exciting developments ahead.