What is the point of learning to write in a world with AI?
You aren't sending a message - you're running a marathon
In the last few months, I’ve read and heard so many stories of teachers and university professors getting frustrated with students handing in written assignments that have completed by generative AI.
But plenty of students don’t understand their teachers’ frustration or what the problem is. In their eyes, they were given an assignment and they got it done in the most efficient way.
And it isn’t just the students. Many educators are arguing that if generative AI is going to be used in lots of jobs, why not let students use it in school and university too? Surely it is just Luddite backwardness to insist on AI-free classrooms and assessments?
In this post and the next, I’ll explain why it isn’t.
Filling the unforgiving minute
Technology obviously does transform society and the economy, and it obviously does make certain skills more or less valuable in the workplace.
Think about marathon-running. Ancient Greek ‘day-runners’ like Pheidippides, the original marathon runner, were a vital part of the functioning of Greek city states. Their aim was to get a message from A to B as quickly as possible, and in a world with no combustion engine or telegrams or internet, if you needed to send a message you needed a day-runner.
Nowadays that is obviously not the case. If you need to send a message to a friend 26 miles away, you'd probably send a text message. If they needed a particular physical set of documents, you still wouldn't hire a day-runner - you'd hire a taxi driver instead.
And yet, running marathons is incredibly popular and more people run the London marathon every year than there were ever day-runners. None of them run it to courier a message. The vast majority run it for reasons of personal development. A tiny fraction do run it to gain economic benefit - not from being messengers, but from winning races and providing entertainment which spectators are willing to pay for.
The difference between ancient Greek marathon runners and today's marathon runners is at the heart of so much confusion about AI and its role in society and education.
Students are baffled by teachers who want them to deliver a message via a marathon run when you could just send a text message.
Teachers are baffled by students who want to get a taxi round a marathon course and expect a medal at the end of it.
So partly, this confusion is about what you think the purpose of education is.
If you think the point of education is for students to develop their human potential, then the teacher is by definition right. In order to develop your human potential you can’t get someone or something else to do it for you!
If you think the point of education is to get a job, you will probably be more open to using technology in education.
Education for personal development
However good AI gets at writing, there is always going to be a personal development use case for writing which is analogous to running.
That is, even if the direct economic value of being a good writer gets completely obliterated by technology, the general life value of being able to write will remain.
Being able to run a mile or two without getting out of breath is good for your general health and fitness, and that’s why lots of people choose to go out running even though there is no direct economic benefit in it for them. Likewise, I suspect that people will value being able to read and write complex texts even if they can’t make money out of it.
One reason I think this is because writing has so many underappreciated benefits. Often, we assume the main purpose of writing is communication, as though it works like this: I have a series of well-worked out and complete thoughts in my head; I write them down; other people can read them.
That is certainly one way we use writing, and the one that generative AI is perhaps best placed to help with. If you have a series of thoughts in your head, you can speak them into a voice memo and get AI to polish them up into a perfect paragraph or series of paragraphs. A good example is where you have a relatively simple decision to make - perhaps whether to say yes to an invitation or not - you make the decision, and then you struggle to communicate it. Generative AI can be helpful in these cases.
But there is another important function of writing which generative AI is less well placed to help with: writing helps extend working memory and is a tool for thinking. A lot of the time we don’t have complete and well-worked out thoughts in our head. It is only by writing our thoughts down that we discover the gaps and the flaws in them. “If you’re thinking without writing, you only think you’re thinking.”1
This process is obviously important if you are doing some kind of deep academic work, but it’s still important in everyday life - if you are writing an email to friends with the logistics for a day out, or a memo to colleagues on the decisions that have to be taken at your next meeting. Generative AI might already be a better writer than you in terms of being able to quickly draft readable and error-free texts. But is it better than you at working out what you actually think?
Cars are faster than humans, but we still teach PE at school. Similarly, even if generative AI totally outstrips humans at writing, I think we will still teach it because it is a valuable tool for life.
One of the reasons I love the work of The Writing Revolution is that they explicitly make this link between thinking and writing, and design lessons and curriculum sequences that teach writing as a tool for thinking right from the start. The resources we’ve developed at No More Marking follow this approach too.
But what about education for the job market?
Of course, personal development is not the only purpose of education. Another important purpose is to get a job, and here it would seem the advocates of generative AI are on stronger ground. If AI is going to be used extensively in the workplace, surely it should be used extensively in the classroom too?
I am not so sure. Even if you are solely concerned with the economic function of education, and even if you don’t care very much about the personal development aspect, there are still strong reasons to avoid the excessive use of generative AI in education. I’ll address those in the next post - with another marathon metaphor!
https://www.goodreads.com/quotes/11255502-if-you-re-thinking-without-writing-you-only-think-you-re-thinking
Very timely, Daisy. Wherever I go, it seems that A1 is the subject on everyone's lips - even when I was sitting in the dentist's chair this morning! The two people I was talking to were not so much unenthusiastic as downright terrified about what was coming.
For me, the most challenging question is what happens when we consistently offload cognitive work and outsource it to AI? Does our cognitive ability begin to atrophy? Michael Gerlich from the Center for Strategic Corporate Foresight and Sustainability, SBS Swiss Business School certainly thinks so. He published a study in January, 2025 with 666 participants, which looked at how the use of AI decreases critical thinking through cognitive off-loading.
There's also a new book, The Memory Paradox: Why Our Brains Need Knowledge in an Age of AI by Barbara Oakley, et al. from Oakland University, et al., published in May of 2025. I haven't read it yet but it looks interesting.
Looking forward to your next post. So is Grok, by the looks of things!!
Thank you so much for this - I'm just starting the process of gathering all the 'benefits' of writing without AI to encourage my (online) students to understand why writing (and reading) are vital skills to our humanity. The connection between writing and thinking is one I actively try to build into my teaching - an advantage I have is that I can almost literally sit on a student's shoulder as they choose a word or place a paragraph and ask them to explain their choices to me. I hope that even when I'm not asking them, they may internalise those questions!