Large language model chatbots like ChatGPT are impressively capable AI tools that can write, edit and even translate text on any number of subjects. That makes them a tempting choice for students, especially those that are writing in a language that isn’t their native tongue or in a field that has a lot of technical jargon. Unfortunately though, while these sorts of AIs can be powerful in the right hands, they can also create as many problems as they solve if used incorrectly – and ultimately, you need to rely on a human set of eyes for all work that you do, whether those are your own or someone else’s!
The biggest issue here is that AIs like ChatGPT, Bing and Bard are often “confidently incorrect” – they sound like experts in any field, but fundamentally what they’re doing is just predicting what the next word they should generate, without any concept of facts or truth. This can lead to glaring errors that make it obvious that a piece of text was run through an AI chat bot – or generated by one entirely. Sometimes this goes as far as inventing terms that simply aren’t used in a given field or generating fake references to scientific papers haven’t been published – so it’s vital to double-check AI-generated work or even not use them in the first place for academic assignments.
Given that many universities consider using AI to complete assignments a type of cheating, it’s therefore best to stick to more traditional methods for proofreading rather than chat bots – such as the spelling and grammar checks built into Microsoft Word or Google Docs, or browser tools like Grammarly. These have all seen AI enhancements in recent years, but they still fundamentally accomplish the same goals – taking text that you’ve written and fixing any mistakes present, rather than generating new text from scratch.
If you do want to use AI chat bots for assignments, it’s best to use it as a ‘fresh set of eyes’ – for example, by summarising a long academic paper while looking at potential sources, brainstorming angles for an essay before you start writing or looking for inconsistencies in your text after you’ve completed a section. Rather than writing the assignment for you, it’s about sparking your imagination or raising points you might not have thought about so that you can write a better paper. Of course, you could ask a friend, family member or classmate to do much the same, and often you’ll find that their feedback or ideas are even more valuable than an AI’s.
If you don’t have a friend or classmate to turn to help you with proofreading – or you don’t want to burden them with what can be quite extensive workload! – then professional services are also available. These firms tend to employ native English speakers with academic backgrounds, making it easier for them to spot errors and suggest improvements that go beyond simple spellchecking – such as identifying logical errors, improving flow, cutting out repetitive content, suggesting proven structures and ensuring adherence to academic standards.
As you might have guessed, I’m a proofreader myself, so I’m biased – but after proofreading hundreds of essays, theses, reports and other academic works over the past decade, it’s clear that ChatGPT has changed how students write – but hasn’t solved how students write. Now, it’s rare to see a piece with many major spelling mistakes, but logical errors and choppy, distinctively-AI-penned prose have become widespread.
No matter what your stance on AI tools like ChatGPT, I implore you: show your assignment to a human being, preferably one that knows writing and your field, before you submit it!
Uni Proofreading has operated in Bristol since 2015, offering academic and careers proofreading across a range of fields, with a particular specialism in proofreading for Chinese international students. Proofreader and co-founder Will Judd received a MEng in Computer Science at the University of Bristol in 2011 and now works as a Managing Editor for video game website Eurogamer and YouTube channel Digital Foundry.