In the past fortnight, two teacher friends have told me that they have had to speak to A-level students about trying to hide the fact that they used artificial intelligence to help them write their coursework.

My friends were honest with me about the scale of the challenge. While they found it easier to detect the use of AI in the work of weaker students, for those who could compose relatively well, it was becoming next to impossible to determine if they had relied on AI or not.

This marks a big shift. Even a few months ago the hallmarks of AI were mostly obvious. But that isn’t the case any more.

With a few cursory checks, and perhaps the help of a “humaniser” tool like Phrasly.AI (which rewrites and rephrases text to mask the use of AI) competent students can easily paper over any obvious gaps in their prose, imprinting their personal writing style on to AI-generated analysis in such a way that it becomes very hard to detect.

So, what do these advancements mean for how we assess English?

The AI threat to coursework

In my view, it is time to admit that large language models have effectively killed off independent written coursework. Not only have these tools developed at lightning speed, making it impossible for schools to keep pace, their use is also widespread. In 2024, 77 per cent of 13- to 18-year-olds reported using AI, up from 37 per cent in 2023.

Students are increasingly likely to see coursework and homework as challenges that AI can shrink. This, of course, has implications for the reliability of assessment, but, more importantly, it allows students to bypass having to think for themselves to the degree that English coursework would usually require them to. This shift is perhaps what we should really be worried about, rather than the ability to gain a couple of undeserved extra marks.

What, then, is the answer? How can we reimagine English assessment in a way that allows us to assess students effectively, while also maximising thinking and learning?

One potential solution is to undertake coursework in controlled conditions, where sessions spent working on the assignments are essentially invigilated by a teacher. But those who have delivered controlled assessments before will tell you what a grim experience this can be. Spending hours of precious teaching time undertaking high-stakes assessments tends to strip the joy from the subject.

In a bid to help students get the best results they can, controlled assessment “resits” also become the norm, taking up even more learning time. And that’s not to mention the time wasted on schools investigating allegations of cheating.

In fact, when I suggested the option of controlled assessments to the two teacher friends who had caught their students using AI, they were vehemently against this idea, based on their own memories of delivering these assessments in the past.

So, what other options are there? Perhaps we could look to university-style assessments, such as a written response with a viva (an oral response that accompanies a thesis).

This would not be a quick or easy fix. We know that oral assessments can be unreliable when used to support high-stakes national qualifications. The previous government eliminated them from the final grading of English language GCSEs because of a catalogue of reasons related to this.

We should also ask the question: would a viva approach unfairly benefit students from more affluent backgrounds? We know there is evidence to suggest that pupils from lower socioeconomic backgrounds are more likely to be behind their more advantaged counterparts in early language development, and that this can have an effect later on in school.

Of course, England’s assessment system is mostly based on exams. This offers a natural challenge to cheating via AI. But how many grades issued this summer to A-level students could have been compromised when it comes to coursework?

It’s not yet obvious what the right solution will be. But we have to face the fact that coursework as we know it is very likely dead - and it’s high time we started looking for alternatives.

Alex Quigley is the author of Why Learning Fails (And What To Do About It)

You can now get the UK’s most-trusted source of education news in a mobile app. Get Tes magazine on iOS and on Android