ChatGPT raises uncomfortable questions about teaching and classroom learning

In a now viral article, the Guardian reported on Dec 4 that OpenAI’s ChatGPT, the state-of-the-art development in text-generating artificial intelligence (AI), is able to produce answers to essay questions that would result in top marks in examinations.

The news prompted an English teacher to warn in a commentary in The Atlantic five days later that ChatGPT will “end high-school English”.

ChatGPT has taken the Internet by storm. In just one week, more than a million people around the world have tested it for themselves, having it generate responses to a range of questions, from critical expositions on sustainability priorities and what good governance is to creative expressions on sonnets and haiku on true love and the meaning of life. Many attested to the noteworthy texts which ChatGPT produced for them in seconds, with posts on social media waxing lyrical on the remarkable prowess of the system.

The capabilities of AI for text generation are not new. An earlier version of ChatGPT two years ago was able to generate opinion pieces for newspapers. The current version of ChatGPT, though, represents a significant development in that the software may be capable of challenging incorrect premises and rejecting inappropriate requests. In other words, the machine can defy requests made by the user. The software is also now available to all.

As a literacy educator and an education researcher, a couple of questions come to mind immediately.

How can educators design assessments so students will not simply channel their homework to bots? Even more fundamentally, with such powerful automated capabilities available, what does it mean to be literate in the digital age?

Need to be multiliterate

Before the turn of the 21st century, a group of academics, the New London Group, published a manifesto in Harvard Educational Review that argued for the need to broaden our understanding of literacy. They pointed to the inadequate practice of confining the education of literacy to reading, writing, speaking and listening, even though larger skills including the critical viewing of multimedia and creative expression involving multimodal formats should be key literacies cultivated as part of the education process. 

They were spot on. Today, we regularly view images and videos on social media and express our ideas and identities through these varied text forms. Students must be multiliterate to navigate the online environment.

Thankfully, this concept of multiliteracies has been included in various curricula around the world. Singapore, like many other countries with forward-looking education systems, has also incorporated multiliteracies, specifically the interpretation and creation of multimodal texts such as posters and digital stories in the English Language syllabus for both primary and secondary schools. 

But oddly enough, while multiliteracies are now part of the Singapore curriculum, assessment practices in the literacy classroom remain language focused. I conducted a study to examine the extent to which the viewing of images is assessed in the English Language papers of the Primary School Leaving Examination and Singapore-Cambridge O-level examinations. I also examined the role of images in international literacy assessments, including in the Programme for International Student Assessment of the Organisation for Economic Cooperation and Development, and the Progress in International Reading Literacy Study.

My study, published in the Australian Journal of Language and Literacy, found that images in literacy tests were mostly decorative, and students’ ability to make meaning of them was rarely assessed. Their ability to represent thoughts and express themselves with images was also not assessed in any examination. But think about it – today, so many of us have to do PowerPoint presentations where complex ideas have to be conveyed through graphics, videos and other forms of communication beyond the written word.

Given the lack of focus on multiliteracies in assessment, it is not surprising that teachers tend not to engage with students or cultivate their competency in this area, choosing to focus instead on preparing them for what exams test them on.

Beyond the traditional

ChatGPT can potentially disrupt teaching and learning – but in a good way. It invites educators to consider what students can create beyond what a machine can produce. Through interacting with ChatGPT, students can be encouraged to ask good questions to deepen their comprehension, uncover new information and learn something new. It encourages educators to recognise students’ abilities beyond memory and regurgitation, to the application of ideas in critical, empathetic and contextually nuanced ways.

Educators can also consider moving beyond the traditional mode of essays as a way for students to express their knowledge and understanding.

In addition to writing, students should be able to deliver presentations, performances and other digital forms, including webpages, videos and animations, to put forward their ideas. For example, in my course at the National Institute of Education (NIE) in Singapore for pre-service teachers, students as part of the final assessment have to create a digital lesson package on a website, complete with learning activities and resources.

Digital technologies can also be harnessed to support assessment, so teachers can focus on qualitative assessment where evaluation requires human judgment. This can include the use of automated essay marking to offer timely feedback on errors in language, and software with feedback on reading fluency. My colleagues at NIE are also exploring the use of learning analytics and adaptive testing to personalise the learning experience and feedback for students.

The assessment that schools do today largely tests and focuses on an individual’s intelligence and competencies. But it is increasingly acknowledged that learning is a social and collaborative endeavour, and having a high emotional quotient can accelerate learning and help an individual reach better outcomes. We learn with others and from others. It is only by tapping the collective smarts of a team that problems can be addressed adequately. Educators should therefore explore new ways of assessing students that reflect the social nature of knowledge building. Whether they can effectively interact with an AI bot like ChatGPT to find answers to open-ended questions could be one way.

Although there are limitations in how much the formal educational curriculum can and should assess, assessments are signals of what educators want students to take away from school. Only by inclusion of these areas in tests and exams will a corresponding shift in what is taught in the classroom happen.

There is little doubt that AI will continue to advance and become even more powerful in the future. ChatGPT must make us ponder the differences in knowledge contribution between man and machine and motivate us to use new tools to evolve our education system. The extraordinary technological marvel presents an opportunity for us to move beyond rote learning to nurture our students to become more creative, thinking individuals as we reflect on what it means to learn and be human in the digital age.

• Victor Lim Fei is an assistant professor in the English Language and Literature Academic Group at the National Institute of Education, Nanyang Technological University. He is principal investigator of a research project on multiliteracies in Singapore’s English Language classrooms and lead author of the book Designing Learning for Multimodal Literacy: Teaching Viewing And Representing (Routledge).