Bringing AI into the classroom: experimenting with approach

No educator can be unaware of the potential for AI to reshape teaching, learning and assessment.  The rapid evolution of Large Language Models (LLMs) like ChatGPT present challenges and opportunities for higher education. Exactly how to deal with these challenges and opportunities is a critical – and urgent - question for business schools today.

LLMs can generate content in seconds. They can, using prompts, generate answers to questions, write essays, read and analyse information and automate a huge variety of tasks.  For students, they are also potential study aids, a guide to grammar and spelling, and an idea generation and assignment planning tool. This not only presents practical challenges for universities (especially in relation to assessment) but may fundamentally reshape the kinds of jobs today’s undergraduates will be undertaking in the future.

As appealing as they might be at first glance, LLMs are fraught with problems. They work on a predictive basis, using their training data to generate the most likely or suitable answer. This training data may not be accurate or up to date. Outputs can be biased, misleading or simply hallucinated (responses that are inaccurate, illogical or referring to other content that does not exist).

Problems notwithstanding, ignoring AI in Higher Education is not an option. Students will inevitably become aware of tools and therefore need guidance on how to use them properly and safely. This needs to encompass rules on academic conduct, but go beyond this too. Business schools need to help students to maximise these tools to support their learning, prepare them to use such tools in the workplace and help them to understand the risks, limitations and ethics of AI.

Here, we set out how we introduced AI to first year business undergraduate students, building ChatGPT into classroom and assessment activities.

Our introduction to LLMs began with an initial lecture providing an overview of the subject and setting the context for future student activity. This addressed the key question ‘what is AI?’ and included a full demonstration of tools. Students were invited to suggest prompts for ChatGPT, creating live content in the moment. The tools were asked to write blog posts, cover letters for jobs, book reviews, answers to essay questions and undertake research on a particular academic subject.  These examples resulted in a range of content from good to poor, demonstrating both accuracy and inaccuracy, as well as hallucinations.  Using the demonstration to explore the issues, the lecture also discussed risks of accuracy and bias. We further explored the style of writing that LLMs generate, and how they lack criticality, tend to sound automated and one-dimensional, and provide very general responses.

This lecture was used as a starting point for a seminar in which students could experiment further for themselves and apply their learning. To demonstrate critical analysis as part of their assignment, students were given three essay questions. They were asked to choose one and obtain a ChatGPT generated response. They were then required to critique the content, highlighting where any bias occurred, if there were any statements lacking supporting evidence, and to ascertain whether the academic literature sources were genuine. Students were also encouraged to critique the overall writing style of the tool and the extent to which the question was fully addressed by the generated response.  This generated rich classroom discussions about how ChatGPT can be used and when it should not be.

Following the seminar, students were asked a few questions about their thoughts on being taught ChatGPT in the classroom. Few expected to be taught tools of this nature. Almost all students either agreed that it is important for them to understand AI and its implications. In addition, the students all agreed or strongly agreed that the lecturers explained the core concepts of AI clearly. The teaching team had extensive personal experience in using ChatGPT, highlighting a key challenge of bringing AI into the classroom. Faculty need to be skilled users of relevant tools, and keep up to speed with the latest developments in the field in order to share them effectively with students.  Business schools therefore need to consider carefully how they support this need.

It could be argued, perhaps, that we have provided students with a tool to cut corners in their learning or even commit academic misconduct. We hope that instead, we have given students a realistic picture of AI tools, and information on how they can use them to maximise their learning and development during their time at university – and beyond.

Perhaps the final word should be left to ChatGPT itself, proving as it does some of our points about depth and criticality: The integration of LLM tools into the classroom should be approached with careful consideration. It can be beneficial when used thoughtfully to enhance the learning experience, but it requires attention to potential challenges and ethical considerations. Balancing technology with traditional teaching methods and fostering a supportive learning environment is crucial for successful implementation.