ChatGPT ‘a powerful tool for education if used correctly’

The emergence of smarter, more accessible artificial intelligence tools such as ChatGPT will undoubtedly change higher education – but that isn’t necessarily a bad thing, according to experts.

OpenAI’s new bot has rapidly gained attention since its launch, with much of the commentary focusing on how students may be able to use it to plagiarise answers to essay questions that are undetectable by software such as Turnitin.

While academics have been concerned about AI-generated assignments for some time, ChatGPT “moves the field leaps ahead in generating text from prompts”, according to Kate Devlin, reader in artificial intelligence and society at King’s College London.

THE Campus resource: ChatGPT and the rise of AI writers - how should higher education respond?

She said this poses the “immediate concern” that students will be able to generate plausible answers to essays “in seconds”, but many institutions are already changing the way they conduct assessment anyway, and ChatGPT will only speed up this process.

Dr Devlin said such changes could lead to assignments that prioritise “well-documented and reflective pieces of work tailored specifically to [students’] learning in modules” and urged institutions to resist the urge to return to formal, timed exams to combat the risks of AI.

Thomas Lancaster, an academic integrity expert and senior teaching fellow at Imperial College London, said ChatGPT was “powerful technology if used correctly”.

It had two main advantages over previous systems, according to Dr Lancaster. “The first is ease of use. The second is memory. You can continue a conversation with the bot and it will remember what you talked about. You can also refine text, so if you don’t like something that it generated, you can extend it, refine it or ask for it to be written in a different voice,” Dr Lancaster said.

“Universities have to act swiftly to understand this technology and the implications for education. But rather than simply looking for ways to ban it, they should consider changing policies to make clear to students when it is and is not acceptable to use.”

As well as working out appropriate ways to integrate the technology into education, “perhaps we also have a duty to help students understand the dangers of over-relying on it”, Dr Lancaster cautioned, given ChatGPT has been shown to produce wrong information when asked questions in a particular way.

This, agreed Dr Devlin, made universities’ role in teaching the need to critically evaluate sources ever more important.

Ethan Mollick, an associate professor at the Wharton School of the University of Pennsylvania who studies innovation, said he had introduced ChatGPT to his class on a Tuesday and 70 per cent of them were using it by the Thursday.

“We have to both panic about how it is changing everything but also think about how we can use it,” he said. “As educators, we are trying to train people for the world. Without too much exaggeration, what that world looks like is changing due to these AI companions and…we have to seize the effect that it also extends our capabilities and lets us do things we couldn’t do before.”

Dr Mollick said that with so much of the attention being on combating cheating, there was little focus on several “urgent questions” educators needed to be grappling with.

Students are asking: “Can the AI write a draft that I can then improve?” he said. Or: “Can I use it to make my answers better?” According to Dr Mollick, “what plagiarism is has essentially now become an open question”.

tom.williams@timeshighereducation.com