If AI Can Write, Why Should We Keep Teaching Writing?

Reading Time: 3 minutes

A question echoed through the minds of writing instructors across the country when ChatGPT was released in late 2022: If students can ask AI to write their papers, why would they bother learning to write?

But maybe we should have had more faith.

Now, research on four years of writing instruction in the context of ever-increasingly sophisticated chatbots has given us a richer understanding of student motivation for using — and not using — AI.

What research reveals after 4 years of students using AI

In a 2026 study of six college writing sections, students were explicitly taught responsible uses of AI and were permitted to use ChatGPT in their writing processes. Only 18.6% of students prompted the chatbot to write for them. Researchers found that just 8.2% of the writing submitted by students was generated by AI. In other words, most student interactions with AI supported or preceded their own writing.

Another recently published study, based on a two-year pilot integrating AI into first-year writing courses, found that students primarily used AI for pre-writing activities like brainstorming and summarizing rather than for generating final text. The instructors deliberately taught students how to use AI while foregrounding the process-oriented practices that support writing and learning.

The findings from these studies raise questions less about whether students should use AI in writing and more about the purpose of writing: Why should students bother learning to write when a machine can do it far more easily?

How does AI change the writing process?

As writing instructors know, teaching writing has never just been about creating a finished product. Writing is also a way of thinking, a process through which students develop ideas, test claims, wrestle with uncertainty, make rhetorical choices, and discover what they think. In this understanding, the struggle in composing isn’t an obstacle to learning — it IS the learning.

AI tools remove much of the struggle associated with writing. They can near-instantaneously brainstorm ideas, summarize readings, create an outline, generate a fully finished draft, and revise sentences. No doubt these capabilities can be useful. However, when students routinely hand over the cognitive work that writing is meant to develop, they risk losing more than the opportunity to learn how to support a thesis or flesh out a literary analysis. They risk losing opportunities to practice thinking, a muscle that they will need for much more than writing.

How should writing instructors approach AI now?

While many colleges and universities are promoting red-light policies against using AI, much of the most sophisticated thinking about the issue suggests the best approach lies somewhere between banning it outright and allowing students to use it without limits. It’s far better, say scholars like Derek Bruff and Annette Vee, to help students see the benefits and pitfalls of AI in writing, and to teach them to make thoughtful decisions about when and how to use AI in their writing processes.

To that end, it helps to think about three kinds of literacy:

• Critical AI literacy means understanding what AI tools can and can’t do. To be critically literate, students must learn to question the outputs of AI. They need to understand that a fluent response isn’t necessarily an accurate one, that AI can reproduce biases embedded in its training data, and that relevance, accuracy, privacy, and other considerations matter when evaluating outputs.
• Functional AI literacy means learning to use these tools without ceding agency to a machine; to maintain control of the intellectual work of writing. There is a meaningful difference between asking AI to generate an argument and using it to help brainstorm possibilities, identify questions, or experiment with alternative approaches to a draft.
• Ethical AI literacy means understanding that choices about using AI have bigger implications than just an individual assignment. Students must consider questions of academic integrity, transparency, intellectual ownership, and what it means to allow a technology to mediate their learning.

These forms of literacy offer a path beyond the false dichotomy that “AI will destroy writing/AI will make writing better.” Technologies don’t determine their own uses: Humans, institutions, and technologies shape one another. The task for writing instructors, then, is not simply to block AI or welcome it uncritically, but to help students become more thoughtful writers and more discerning users of the technologies that increasingly shape how writing happens.

This is the premise behind “A Writer’s Guide to Generative AI.” Rather than prescribing a policy for AI use, the guide offers activities and approaches that instructors can integrate into existing curricula. Its goal is to help students understand the possibilities and limitations of AI while preserving the productive struggle, critical thinking, and rhetorical decision-making at the heart of writing.

Written by Jodie Nicotra, author of “A Writer’s Guide to Generative AI” and previous director of the composition program at the University of Idaho.