What is the role of ChatGPT and other large language model AI in Higher Education?

What is the role of ChatGPT and other large language model AI in Higher Education? This is a question that is exercising many as we grapple with the realisation that we have access to an unfamiliar and powerful new technology. For many who work in education, an early response was concern that this tool might be used by students in place of learning. For some students, initial interest might have been around how ChatGPT could make the labour of assessment less onerous. These reactions are predictable and familiar. Many technologies that aid thinking are demonised by the gatekeepers of learning as a risk to their power, and for the less powerful can raise hopes of freedom. From Plato's concern that writing would ruin the memory skills of the oral tradition of poets, to the more recent resistance among educators to internet search, Wikipedia and smartphones in the classroom, we who are interested in interactive learning environments are familiar with this cycle of resistance followed by adoption and eventually mass take up. Frameworks such as the technology adoption model and UTAUT give us access to quantifiable variables to measure. These can help us sample a specific population and weigh up the usefulness they perceive in a tool. As a technology matures, measuring what it is good for, can offer valuable insight, but may be less helpful for detailing a disruptive paradigm shift, as I asked in an earlier editorial on this topic, with ChatGPT it is knowing what questions to ask (Rospigliosi, 2023).

While ChatGPT may still seem new in some regards, it is less than two years since the release of the first version that was downloaded on a large scale and the pace of take-up has already passed beyond early adopters to verge on majority take up. Rogers’ modelling of technology adoption helps identify these stages (1962), as early majority encroaches on laggards, institutions and organisations recognise the need to accommodate use, and many stakeholders, from education regulators to teachers and students are realising the necessity of moving from the initial stage of unfamiliarity to the necessary questions about what we use it for, what are the benefits to us, and what risks does it pose? This journal can help, and we encourage authors to engage with this important new form of interactive learning. We welcome practical insights derived from experience of the uses of large language models to generate learning materials, assessment, and feedback, but encourage authors to reflect critically on the implications beyond their own set of results. We know that these new tools are effective, but need to also try to understand what their effect will be.

For many educators, the opportunity to use large language models in our teaching and learning will require the use of the existing range of systems such as ChatGPT, Bard and Ernie. These are the systems on which we are often dependent, yet they raise risks that we cannot ignore, even as we integrate them into practice. Student use of AI (LLMs) can enhance the capacity for student research: developing skills in asking the right questions and reviewing responses. We can hope for “prompt engineering” to emerge as a new vocationalism (Bourner et al., 2011). There are significant risks of the adoption of AI in education exacerbating the existing digital divide. Algorithmic injustice and the perpetuation of discriminatory stereotypes are endemic in the machine learning datasets used for the major LLM AI systems (Rospigliosi, 2021).

Generally, it is at the institutional level that choices of system adoption are made, yet it will be in the practice of teachers and learners, researchers and scholars that the consequences will be explored. This journal aspires to play a part in the debate in addressing the question: what is the role of ChatGPT and other large language model AI in Higher Education?