Published on 07 Feb 2023

Commentary: ChatGPT can disrupt education, but it need not be all bad. Here’s how NIE is using it to train teachers

Look at the photo above, which is a screenshot of how we used ChatGPT, an artificial intelligence-powered chatbot from United States firm OpenAI, as a writing aid for this commentary.

Chances are many people would find it hard to tell that the reply was written with the assistance of a chatbot.

Similarly, many users of ChatGPT are likely impressed by its ability to provide coherent, seemingly well-reasoned responses that mimic human writing.

It can also generate outlines, stories, poems, creative writing, summaries, and even computer programs with just a click.

Users can engage in interactive dialogue with ChatGPT to request more information or engage it in a debate that humans are not necessarily able to win.

There has been much discussion in the media and on social platforms about the potential challenges and benefits of using ChatGPT in education.

Educators have expressed concerns that the software's ability to generate essays could lead to increased instances of plagiarism among students.

When asked about its usefulness in education, ChatGPT provided thoughtful responses such as: "One potential use is as a writing assistant or tutor for students, by generating human-like text. It can also assist with language translation and comprehension, making it a valuable resource for multilingual classrooms."

It went on to say: "Additionally, ChatGPT can generate practice problems and quizzes to test students' understanding, and aid in creating educational content like summaries, questions, and answers. Overall, its capabilities make it a valuable tool to assist teachers and students in the learning process."

Yet, many educators have been wondering about the potential negative impact of ChatGPT on school assessments and tests.

Most educators’ anxiety can be linked to the overemphasis on the traditional mindset of privileging summative assessments, and the expectation that students are able to perform tasks from scratch or with no or as little external support as possible.


Here then is the challenge: ChatGPT stimulates us educators to relook the ways we have been assessing our students.

In 2016, a robot test-taker performed better than 80 per cent of students on the University of Tokyo entrance exam, raising concerns about what the tests were measuring, and how both students and machines were being taught.

It should prompt us to question whether we are training students to learn and perform like robots or teaching them skills that computers cannot replicate.

Contemporary educators stress that a crucial skill for students in the 21st century is to "ask the right questions", which is even more important than "answering the right questions”.

In the past, textbooks and teachers served as the main sources of knowledge for students. However, with the advent of the internet and other forms of technology, students now have access to a wide array of sources for information.

The crux is to use the "right" combination of keywords to find relevant, high-quality answers.

For example, suppose a primary school arranges for students to visit a butterfly garden, and students Google before the field trip to learn about personal interests.

Three of the students searched: "Tell me a joke about the butterfly", "How many colours do butterflies have?" and "What is the life cycle of a butterfly?"

If these three questions reflect the learning goals of individual students, who do you say will learn more meaningful and deeper knowledge in preparation before and during the visit?

Also, if students put a great degree of thought into designing a question such as qualifying it with the apt conditions and context, and asking follow-up questions, they are indeed doing some useful learning.


We need to consider how ChatGPT can support teachers to use it in smart ways.

The educational task is not to assess a student as a smart or not a smart individual, but to design smart learning tasks in realistic contexts.

These designs promote engagement and develop learning potential, rather than using formally measured achievement tasks.

The SAMR model — which categorises the use of technology into four levels of sophistication and transformative power: Substitution, Augmentation, Modification, and Redefinition — is a useful framework for thinking about how digital technologies can be used in education.

While ChatGPT can be primarily viewed as a tool for substitution and augmentation of written assignments, it has the potential for significant task redesign and the creation of new learning tasks or assessments.

ChatGPT can be used as a resource for creating smart learning environments where content may already be available or can be generated.

For example, when asked to generate different lesson plans for the science topic of the water cycle based on different theories of learning such as behaviourist, cognitivist, constructivist and situated learning, it generated plausible lesson plans, however, alas only superficially.

In our class at the National Institute of Education, we then create a task for student teachers to critically examine each lesson plan and make improvements.

In another class teaching student teachers the ethical use of AI in education, we get them to listen to an instructor talking about such ethical issues, use a web tool to create a transcript of the talk, and then input into ChatGPT to ask for a summary of the top three key ideas relevant for teachers.

Teachers will also think about their own individual three takeaway messages from the talk earlier, and then compare them with those generated by ChatGPT, thereby stimulating debate and reflections amongst themselves.


The key takeaway is to not resist the integration of ChatGPT, but rather plan for its beneficial and ethical uses.

In education, this presents a chance for students and teachers to learn how to effectively incorporate it into their lives.

With the information generated by ChatGPT, it is important to develop tools to verify its accuracy and detect plagiarism. We may need to utilise more technology to address the issues arising from technology.

The World Wide Web has accustomed us to searching for information online rather than relying on libraries and printed materials.

Similarly, ChatGPT may become an automated learning companion that is smart enough to generate readily available information at our fingertips.

However, it is important for smarter educators and learners to verify the accuracy of this information and know how to treat it critically, use it effectively, and build on it creatively.

Our human capabilities will still play a crucial role in making the most of human-machine collaboration.

In the coming months, we will find out if a new education normal will indeed emerge from the disruption caused by ChatGPT.

Looi Chee Kit is Professor of Education at the National Institute of Education (NIE), Nanyang Technological University (NTU), and co-Director of the Centre for Research and Development in Learning, NTU. Wong Lung Hsiang is Senior Education Research Scientist at the Centre of Research in Practice and Pedagogy, NIE.

Read the original article here.

Source: TODAYOnline © Mediacorp Pte Ltd. All rights reserved