ChatGPT, AI, and the Future of Education

ChatGPT has caught everyone’s imagination. Here, we catch up with founder of Writer’s Toolbox, Dr Ian Hunter. An award-winning academic turned innovator, Dr Hunter is no stranger to the world of AI computing. He holds a patent in AI technology and now runs one of the world’s fastest-growing education platforms.

What is ChatGPT in layman’s terms?

I think it’s really exciting the work the OpenAI team are doing at their research laboratory in San Francisco. Simply put, ChatGPT is one of a range of natural language applications – think computers working with everyday text – that are taught to recognise and work with information from large databases of written work. Applications can include translating language, or in the case of ChatGPT, answering questions with extended coherent text, where the question itself is posed in everyday language, as opposed to programming code.

Is using computers like this in education a new thing?

Where ChatGPT is taking us is new, but ideas about computers in education were proposed as early as the 1960s, when a range of researchers began building computer models to assess student writing. This work continues today. GMAT, for example, uses the ‘IntelliMetric’ software in conjunction with one human marker to score essays.

Computer Assisted Language Learning (CALL) also proved popular. Here, educators used computer programmes to help people learn foreign languages. And then, as computing advanced, this migrated into students using software applications to learn content and answer closed-loop questions. The advantage: content questions could be targeted to the learner. The disadvantage: knowledge is not the same thing as understanding. And just because you can answer a lot of multi-choice questions, doesn’t mean you can answer compelling questions on a topic in a deep and coherent manner.

So what does ChatGPT mean for education? Have the bots taken over?

That’s a great question. To answer it correctly, you first need to ask: what has ChatGPT been built to do? In short: to generate coherent content in response to human questioning. And it does this better than anything we have seen to date. So, if educators need content generated – and rapidly – then, what an amazing resource. Where we are seeing the debate heading is toward student use of the tool to produce material, and possibly doing so instead of penning their own responses to a question.

“Human beings still need to master fundamental skills.”

Figuring out how to deal with ‘fake’ answers is not a new problem for educators. In the past, it could have been cousin Elizabeth writing the term paper instead of student Mary. Now, it might be ChatGPT. And educators have always had really good tools at their disposal to accurately assess student ability: the depth of the questions we ask, the cognitions we require demonstrated, where we demand those questions to be answered, and the form of those answers. If we stock to first principles – academic honesty and academic integrity – and defend these principles for the sake of the student, the answers will readily come. ChatGPT will not be the end of authentic education.

How is ChatGPT different to the AI you’ve built in Writer’s Toolbox?

Okay, imagine a road. And you come to a fork: one sign reads content generation, and the other sign reads: teaching learners. ChatGPT headed left; we headed right. The AI built inside Writer’s Toolbox doesn’t generate text. Instead, it takes the text a student generates, profiles them as a writer, and begins to teach them – step by step – how to be an even better writer. ChatGPT gives educators content; Writer’s Toolbox teaches you how to write. That is a very different educational outcome.

Does AI really make a difference to student learning?

Since the mid-1990s, we have known that a computer-assisted environment in student learning could make a positive impact – possibly even more so than a traditional classroom. What we have done at Toolbox is to push this even further: to give students feedback at their most teachable moment. And to date, large scale studies are supporting the impact. We’ve seen effect scores double educational norms, and in the largest Australian study to date, writing improvement scores between double and 10 times the state of Queensland in standardized testing. So, yes, I believe this is a goal worth pursuing; we can use AI to make a difference. We can create very different educational outcomes for our students.

Visit the Writer's Toolbox website.