Understanding the Main Functions of Human Writing: The Confusion Around Naming AI-Generated Text

As another school year returns, large language models (LLMs) present difficult questions around learning, thinking, plagiarism and authorship for educators.

New approaches to assignments and assessment are required. Student papers that use LLM technology require additional labour on many fronts. Professors have expressed frustration, worry and anxiety.

As an assistant professor of English whose research has focused on the histories of writing and how it’s taught, I have been involved in many discussions at institutions of higher learning about this topic.

The immediate issue of LLMs in the classroom points to a larger reality. For too long, instructors and universities have been treating students’ writing difficulties as a deficiency in the mind — instead of confronting the truth that writing, as a technology for thinking, is cognitively demanding.

Writing is a technology that helps us understand our own ideas better. Where people are involved in thinking and invention — including at universities — it needs to be taught in that spirit.

Why consider AI output ‘writing?’

Given what LLM technology actually does and does not do, why do we even consider it writing — as in, “they used an LLM to write their assignment?”

LLM technology creates — at best — a facsimile of a textual object, meaning that based on its training on existing texts, it can create output that resembles a “text.”

When generated for the purpose of a university assignment, it resembles the standard academic English that has ruled the academy (historical and present-day institutions of higher learning including colleges, technical schools and universities) and has been endowed with a special relationship to truth.

I’ve seen anger and frustration expressed towards the student who uses LLM output in their writing process or submitted work. The reason seems to be that it undermines that special relationship to truth that academic writing has long held.

The artificial output reveals that academic writing is not an “absolute.” Rather, the LLM shows us how academic writing is a social construction. But academia at large has long resisted acknowledging this.

Historical and social concerns

Using standard academic writing is a choice. It is a style of writing that has been invisible within academia, thought to be the default and “correct.”

Many of the problems and questions that arise with LLM technology in relation to writing are really historical and social concerns that get to the very heart of what we understand writing to be.

As I continue to discover in my research, writing in the context of the English-speaking academy since the 19th century has been taught within two distinct streams: literary and technical communication. Both of these streams flourished within the larger context of the British Empire because both are adept strategies for maintaining the status quo.

The combination of the veneration of canonical British literature and the instruction of a utilitarian language that acts as a “neutral” communication tool (one that is not at all neutral), and a standard one, serves to create an understanding of writing that over-prioritizes the finished text.

Read more:
‘It don’t be like that now’ — the English history of African American English

Writing and context

Still today, what is technically “correct” is seen as “good” writing, while “good” writing has come to be represented as the product of a strong mind. This, of course, implies “bad writing” is the product of a mind not as able, or even suited for post-secondary study. Scholars in writing studies have pointed to academia’s focus on correctness and how ways of teaching academic writing can perpetuate inequities, including those related to colonialism and racism.

“Correctness” in writing is, as writing scholars have long discussed, subjective and contextual. For example, “ill c u l8r” would work as a perfectly fine text message in 2008. In another setting — including texting today with QWERTY-keyboard equipped phones — it would seem “off” or even incomprehensible.

When a student’s ability to write academically is taken to represent their intelligence, we should not be surprised when some use LLMs.

Read more:
Students cheating with generative AI reflects a revenue-driven post-secondary sector

Revisiting the final product

Because LLM technology can mimic standard academic writing, it becomes the perfect context through which to address how we think about the final product of writing. The truth is, if university instructors are only interested in a “correct” piece of writing, it sort of makes sense for a computer to do it.

However, if we are interested in a way of teaching that supports students’ inquiry and thought as they interrogate the often oppressive systems they find themselves in, we need to broaden our understanding of writing beyond the confines of a “correct” standard academic English.

It is important that educators begin to see our students’ writing as part of a social situation which requires clarity on their end, yes, but also our listening.

Without reconsidering how western institutions have positioned writing, instructors risk educating a generation of students who are alienated from their own ability to think and create new knowledge, perspectives and understandings because of an over-reliance on LLMs. Western institutions will create students alienated from themselves and the potentialities of their ideas.

Read more:
How a first-year university writing course for Indigenous students fostered skills and belonging

Re-think what writing is

Part of re-thinking what writing is means allowing students to submit work in process, providing for feedback from instructors on ungraded drafts of assignments and developing scaffolding assignments that reveal the process behind writing.

It also means teaching students a new approach to writing that draws attention to the power structures implicated in the instruction of writing and embraces invention — a rhetorical concept under-utilized within the western academy. “Invention” implies many considerations, including around audience, relationships and circumstances.

Writing experts need to be brought into AI-policy creation and the implications of this field’s research must be considered by a wider audience. If LLMs are implemented in universities without careful consultation with writing experts, I fear we will soon find that all that can be written is the status quo of received ideas.

It’s time for us to reconsider writing in universities or risk our students losing access to their own writing. The study of writing is uniquely situated to help our institutions navigate these urgent questions.