Can we keep ChatGPT out of the classroom?

As schools and universities prepare once again for the upcoming academic year, the threat and promise of AI tools like ChatGPT looms large. Some schools will open the classroom doors to AI; others will seek to ban it. But everybody agrees we can’t go on with business as usual.

After more than a year and a half since the launch of ChatGPT, some of the initial hysteria surrounding generative AI seems to be subsiding. It hasn’t taken over the world, resulted in era-defining job loss, or overthrown traditional education. Even as they get better and can be used for an increasing range of sophisticated tasks, generative AI tools may remain this generation’s calculator or automatic translator. Powerful yet imperfect tools for getting things done.

But in terms of school, one thing ChatGPT and its competitors are good at getting done is homework. Specifically, the kind of written essays and summaries that are staple of homework assignments in schools around the world. Write a 500-word essay about the importance of the Magna Carta. About the circulatory system. About the advantages and disadvantages of remote work. About the causes of the Russian Revolution.

Which naturally makes generative AI perfect for cheating. Once upon a time you might have paid – or bullied – a better student in your class to write your homework for you. With the internet, you could pay an online essay mill for someone on the other side of the world to do it. Now, with ChatGPT, you don’t have to pay anybody. In many ways, generative AI is just the newest face of an age-old problem.

Can schools stop it? Fighting technology with technology doesn’t seem to be the answer. Plagiarism detection tools don’t work on AI generated essays. AI-detection tools don’t work very well either. Last year OpenAI, maker of ChatGPT, shut down its own AI detection tool, citing a “low rate of accuracy” – a real danger when a false accusation of AI cheating could ruin a student’s academic career.

But there may be another way. Some schools are beginning to take a new look at an old solution: handwriting. According to one survey published last year, in order to combat the use of generative AI, 50% of surveyed educators were returning to handwritten assignments. These may include some who formerly allowed or encouraged the use of artificial intelligence in drafting assignments, like James Stacey Taylor, a professor philosophy at The College of New Jersey.

As he wrote recently in the Times Higher Education, he has a ban on AI that he achieves by doing the work in class. He devotes class time to reading, discussing, understanding, and drafting summaries, and then formulating arguments to support an essay in class. All this work is handwritten, with no recourse to electronic devices. While his students may write – and presumably even type – the final version outside class, they submit it to the professor with their previous handwritten drafts as well.

The point isn’t to try to catch students out, but to give students all the tools they need to successfully engage with the texts they read, formulate their own opinions, and craft their essay. It’s to slow students down and help them challenge their own understanding. And by helpfully shifting the focus away from blame, cheating, and mistrust, to actual teaching and learning, Taylor’s analogue approach may be a model for other teachers to follow. As teachers on all sides of the ChatGPT divide will say, gen AI is here to stay. Students and adults will find a thousand ways to use it (and abuse it). But as part of a growing trend pushing back against the tech takeover of the classroom, it may be that in order to learn to really think, we have to pick up a writing instrument and write.