The Instructor Who Keeps Thinking Necessary: Writing Pedagogy and the Survival of Cognitive Formation in the Age of Artificial Intelligence

 

Note to the Reader

If writing is one of the primary conditions under which thinking is formed, then the most urgent question in the age of artificial intelligence is not whether students will continue to produce writing, but who will ensure that writing still requires thought. Artificial intelligence does not abolish language; it detaches language from the labor that once made it cognitively formative, creating a condition in which expression can exist without the sustained effort that gives it intellectual substance. A student can now generate a paragraph that appears complete, structurally coherent, and rhetorically persuasive without ever having undergone the uncertainty, revision, and judgment that would normally produce such writing. The sentence is present, but the thinking is not necessarily so, and that distinction alters the meaning of learning itself. What appears to be understanding may instead be the simulation of understanding’s visible outcomes, a performance of coherence without the internal processes that make coherence meaningful. The danger is not immediate collapse but deferred absence, a gradual erosion of the habits by which students form, test, revise, and ultimately possess ideas. This erosion is difficult to detect because it aligns with visible success: fluency remains, completion remains, confidence remains, and the finished text may even appear stronger than before. What begins to recede is less visible but more consequential—the necessity of thinking as a condition of producing language. What remains, therefore, is a role that has always existed but has rarely been defined with sufficient precision: the writing instructor as the designer of cognitive necessity. This role is not supplemental or remedial; it operates at the level where thinking becomes durable rather than merely visible, shaping the conditions under which ideas are formed rather than simply expressed. It exists in the moments when a student must decide between competing claims, revise a position that no longer holds, or confront the realization that understanding has not yet been earned. When writing no longer automatically demands effort, the instructor becomes the figure who restores that demand through intentional design rather than inherited expectation. This restoration is not a return to past practices, but a recalibration of present ones, ensuring that new tools do not eliminate the need for thinking but instead make its necessity more visible. When effort disappears from writing, one of the most reliable conditions for sustained thinking disappears with it. This essay does not defend that role nostalgically; it defines what disappears if it is weakened.

The Instructor Who Keeps Thinking Necessary: Writing Pedagogy and the Survival of Cognitive Formation in the Age of Artificial Intelligence

Carl Jean

I. The Work That Remains Unseen

The work of the writing instructor has always been difficult to see because writing is evaluated as text but taught as thought, creating a persistent misalignment between what is measured and what is formed. Institutions tend to assess the finished product—the thesis, the paragraph, the evidence, the sentence—while instruction operates within the unstable movement that produces those outcomes. When a student hesitates over a claim, revises a paragraph that no longer holds, or realizes that evidence fails to support the argument it was meant to prove, thinking is being formed in ways that the final text cannot fully capture. These moments rarely appear in grading rubrics, yet they are the conditions under which intellectual development occurs. Linda Flower and John R. Hayes describe writing as a recursive process in which planning, translating, and reviewing continually reshape cognition, revealing that writing is not linear production but iterative construction. Lev Vygotsky shows that language does not merely express thought but organizes it, making writing not a record of thinking but one of its primary engines. To teach writing, then, is not to refine expression alone but to intervene in the formation of cognition itself. This work remains largely invisible because its results are distributed across drafts, revisions, and internal shifts in understanding. Artificial intelligence exposes this invisibility by bypassing it, producing language that appears complete without requiring the processes that would normally generate it. A student can now submit a paragraph that reads as finished but cannot explain how its reasoning holds or why its claims matter. The language remains intact, but the thinking behind it cannot be sustained. The instructor must therefore learn to read not only what is present on the page, but what is absent behind it. The instructor’s work is not in the sentence—it is in the mind that learns how to arrive there. When that work disappears, writing persists, but its role as a site of cognitive formation begins to erode.

II. When Thinking Becomes Optional

The central problem introduced by artificial intelligence is not misuse but optionality, a condition in which students can produce acceptable writing without undergoing the processes that once made writing cognitively formative. When thinking becomes optional, it does not reliably occur, because effort is not self-enforcing and difficulty is easily avoided when alternatives are available. Yet the issue is not the disappearance of difficulty itself, but the disappearance of the right kind of difficulty, the kind that produces intellectual transformation rather than frustration. Not all difficulty is valuable. Confusion without direction leads to disengagement rather than growth. Busywork without consequence produces compliance rather than understanding. Arbitrary constraints generate fatigue rather than insight. These forms of difficulty consume effort without reorganizing cognition. But there exists another kind—formative difficulty—in which effort is structured, purposeful, and cognitively generative. This is the difficulty that arises when a student must choose between competing claims, revise an argument that collapses under scrutiny, or recognize that clarity has not yet been earned. It is the moment when a seemingly coherent paragraph fails under questioning and must be rebuilt from its conceptual foundation. In these moments, thinking is not expressed; it is constructed. Artificial intelligence selectively removes this formative difficulty by providing resolution before struggle has completed its work, offering coherence at precisely the moment when uncertainty should still be productive. The answer appears before the question has fully formed. The student arrives at a conclusion without undergoing the transformation required to produce it. The process is shortened, but the thinking is not strengthened. A mind that arrives too quickly arrives unchanged. Under such conditions, ease is mistaken for progress, when in fact it signals the absence of the processes that make knowledge durable.

III. The Necessary Counterargument

It would be intellectually incomplete to ignore the strongest argument in favor of artificial intelligence within writing pedagogy, because the case for AI is not superficial but structurally significant. Artificial intelligence introduces forms of access that address longstanding inequities in education, lowering the threshold of entry into writing for students who might otherwise remain excluded by fear, linguistic barriers, uneven preparation, or organizational difficulty. A student who once faced the blank page with paralysis may now begin with a generated draft, transforming avoidance into engagement. AI can support multilingual writers navigating unfamiliar academic discourse, assist neurodivergent students in structuring thought, and provide iterative feedback in environments where human attention is limited. Scholars such as Ethan Mollick argue that AI can function as a collaborative tool, enabling experimentation, revision, and increased participation when used transparently. These gains are not incidental; they reshape who is able to write and under what conditions. They matter because access matters, and any pedagogical position that ignores this fact risks reinforcing the very exclusions it seeks to critique. Yet it is precisely this reduction of friction that creates the deeper pedagogical problem. Tools that remove unnecessary difficulty can also remove necessary formation. The same system that helps a student begin can prevent that student from needing to continue the work of thinking independently. If AI initiates thinking, it can support learning; if it completes thinking, it replaces it.

It is also possible that artificial intelligence is not merely weakening existing cognition but reorganizing it into a distributed process. Students increasingly think through interaction with AI—posing questions, refining prompts, testing variations, revising outputs, and developing claims through exchange. A student drafting an argument may ask an AI system to identify counterclaims, then challenge the response, then revise the claim again in light of a new example. In this sense, prompting can become a form of iterative reasoning, and AI can function as externalized working memory, holding fragments of thought long enough for them to be rearranged. Thought appears to unfold across the interaction rather than within a single mind. That possibility should be taken seriously, because it prevents the essay from reducing AI to mere threat. Still, this shift introduces a decisive distinction between extended cognition and displaced cognition. If a student can step away from the system and still explain, adapt, and defend the argument, cognition has been extended. If the reasoning collapses without the system, cognition has been displaced. The difference is not visible in the output; it is visible in what remains when the tool is removed. If artificial intelligence can reliably produce better writing than most students, then perhaps the insistence on struggle is not pedagogy, but nostalgia. This possibility must be faced before it is rejected. Yet it collapses under closer examination because better writing is not the same as better thinking. As Emily M. Bender demonstrates, language can appear meaningful without being grounded in understanding. When tools think for students, students do not become thinkers—they become selectors.

IV. Labor at the Level of Thought

The labor of teaching writing operates where outcomes are unstable, developmental, and resistant to simple measurement. A correct answer can be verified, but a developing idea must be interpreted within a process that is still unfolding, often unevenly and without clear markers of progress. A student sits with a drafted paragraph open on the screen, its sentences smooth, its claims confident, its structure intact. When asked to explain how the argument works, the student pauses. “It just makes sense,” they say, relying on the apparent coherence of the text rather than the underlying reasoning. When pressed—what does the source actually prove, why this claim instead of another, what happens if the counterargument is stronger than expected—the explanation begins to falter. The language remains, but the reasoning fragments. The paragraph, once fluent, begins to feel unfamiliar even to its author, revealing a gap between production and understanding that was not visible at first glance. In this moment, the distinction becomes visible: the writing was produced, but the thinking was not secured. The instructor is not correcting the paragraph; the instructor is revealing the absence beneath it, making visible what the fluency concealed. What appeared as understanding dissolves under pressure, and what remains is not failure, but exposure—the point at which thinking must begin. This is why feedback is not merely correction. It is cognitive intervention, reshaping not only the draft but the student’s capacity to think through complexity. Artificial intelligence intensifies this labor by improving surface fluency while leaving underlying cognition unverified, making the instructor’s role both more difficult and more necessary. The instructor must now read against coherence, distinguishing between language that reflects thought and language that merely simulates it. The more convincing the language becomes, the more carefully thinking must be verified.

V. The Stability of Knowledge

Writing stabilizes knowledge not by preserving it passively, but by forcing it to be built through deliberate acts of articulation, organization, and revision. When students write, they must decide what matters, how ideas connect, what evidence proves, and where reasoning fails, and these decisions transform information into structured understanding. What is merely encountered may feel familiar, but familiarity is not possession, and the distinction becomes evident when knowledge is tested beyond its original context. A student may recognize a concept, repeat a definition, or produce a coherent explanation without being able to use the knowledge under pressure, and this gap reveals the limits of surface understanding. Artificial intelligence intensifies this fragility by allowing students to display knowledge without building it, creating a condition in which comprehension appears present but remains unstable. The result is not always ignorance; it is a more subtle condition in which knowledge cannot endure transfer. The student appears to know until the idea must be adapted, challenged, or defended under altered conditions, and it is at that moment that the weakness becomes visible. Knowledge that is not built cannot be kept. What distinguishes durable knowledge is not that it can be restated, but that it can be transferred—carried into new contexts where the original wording, structure, or scaffolding is no longer available. This transfer requires more than recognition; it requires reconstruction. Writing, when properly structured, forces that reconstruction, making knowledge both visible and durable. The instructor’s role is therefore not simply to evaluate what students know, but to ensure that knowledge has been formed in a way that allows it to persist beyond the moment of submission. Without that formation, writing becomes display, and knowledge becomes performance rather than possession.

VI. Not Parallel, but Foundational

Writing is not the only site of cognitive formation, because thinking also develops through dialogue, problem-solving, practice, experimentation, and embodied experience. Yet writing occupies a distinct position among these forms because it externalizes thought in a durable form that can be examined, revised, and sustained across time. Conversation may disappear after it occurs, leaving only memory, and problem-solving may remain procedural, focused on arriving at answers rather than interrogating reasoning. Writing, by contrast, preserves thought long enough for it to be tested, challenged, and reshaped. It slows thinking down so that it can be seen. It forces ideas to take a form that can be revisited, revised, and evaluated under different conditions. This is why writing is foundational across disciplines. In history, it makes interpretation visible. In science, it makes reasoning accountable. In literature, it makes analysis precise. In civic life, it makes claims answerable. Writing is not foundational because it replaces other forms of thinking, but because it gives thinking a structure that allows it to be examined rather than assumed. When students struggle in lab reports, literary analysis, or research papers, the issue is often not only content knowledge but the absence of structured reasoning. The problem is not that they do not know enough, but that they have not yet learned how to organize and test what they know. When writing weakens, one of the primary structures that sustains extended thinking begins to thin across disciplines. The writing instructor therefore does not merely teach essays; the writing instructor helps sustain the cognitive infrastructure that allows knowledge to be formed, examined, and transferred.

VII. Accountability Under Conditions of Automation

When language can be generated externally, the relationship between writer and idea must be actively restored, because the presence of language no longer guarantees the presence of thought. Coherence, once treated as evidence of understanding, becomes unreliable under conditions where fluency can be produced without cognitive engagement. A student may submit a polished argument that appears complete on the page yet cannot explain how the reasoning works, why the evidence matters, or how the claim would change under pressure. This gap between expression and understanding requires a fundamental shift in how writing is evaluated. The instructor must ask not only whether the writing is effective, but whether the thinking within it is owned. Can the student defend the claim without relying on the phrasing of the text? Can the student revise the argument when a counterexample exposes its weakness? Can the student explain why one source matters more than another? Can the student identify what changed between drafts and why? These questions move assessment from correctness to ownership, from performance to possession. They do not punish students for using tools; they require students to remain responsible for the thinking those tools may assist. They also force the student to encounter the limits of their own reasoning in real time, which is where thinking becomes visible as action rather than artifact. Artificial intelligence does not eliminate the need for such accountability; it makes accountability central. A paragraph can be submitted. Thinking cannot. Writing belongs to a student only when the thinking can survive without the page. Without that condition, authorship becomes attribution without responsibility, and education risks confusing the appearance of reasoning with its presence.

VIII. The Consequence of Absence

If the role of the writing instructor weakens, the system will not immediately collapse, and this is precisely what makes the danger difficult to recognize. Students will continue to submit essays, and those essays may even improve in visible ways. Sentences will become more fluent, structures more coherent, and arguments more polished, creating the impression that learning has advanced. Yet beneath that surface, the processes that once formed thinking begin to recede. Revision becomes cosmetic rather than cognitive, focusing on clarity rather than restructuring ideas. Reflection becomes summary rather than ownership, repeating what has been said instead of interrogating it. Research becomes retrieval rather than judgment, gathering information without evaluating its meaning or relevance. What disappears in this shift is not information, but the act of deciding what information means, a process that requires time, uncertainty, and the willingness to risk being wrong before becoming right. The institution may mistake improved language for improved learning, because the visible indicators of success remain intact. But the deeper question is whether students can still sustain the thinking that the language appears to contain. A system can improve its outputs while losing its mind. This loss is subtle because it aligns with visible success, making it difficult to detect until the absence of thinking becomes unavoidable. The instructor prevents this hollowing by maintaining the conditions under which thinking remains necessary rather than optional. Without that role, education risks becoming fluent in the appearance of thought while less capable of producing thought itself.

IX. Designing Cognitive Necessity

To design cognitive necessity is to structure writing so that thinking cannot be bypassed without being exposed, ensuring that the absence of cognition becomes visible rather than hidden behind fluent language. This does not mean increasing workload or adding difficulty for its own sake. It means aligning assignments with the cognitive processes writing is supposed to develop. Iterative authorship requires students to show how their ideas evolve across drafts, making revision a record of thought rather than a cosmetic correction. Early drafts should reveal uncertainty, false starts, and developing claims, while later drafts must account for change, showing why an argument moved in one direction rather than another. Oral defense requires students to explain and defend their reasoning beyond the written text, revealing whether the argument is owned or merely submitted. Revision under constraint asks students to adapt an argument to a new audience, new evidence, or a counterclaim, testing whether understanding is flexible rather than fixed. Reflective ownership requires students to account for their decisions: why a claim changed, why evidence was selected, why a paragraph was reorganized, and why an objection mattered. In practice, this may involve short defenses in which a student explains a paragraph without reading it, timed revisions that force adaptation, or targeted reflections that expose decision-making. These structures do not eliminate artificial intelligence; they reposition it. AI may assist with drafting, brainstorming, organization, or revision, but it cannot account for why a claim was abandoned, why a line of reasoning failed, or why a different path became necessary. It cannot replace the student’s responsibility to explain, adapt, and defend the work. Pedagogy succeeds when thinking must be demonstrated, not merely submitted. The instructor’s task is not to resist new tools, but to ensure that those tools cannot replace the cognitive work that defines learning.

Conclusion: The Condition That Must Be Enforced

Education now stands at a threshold it cannot avoid. It must decide whether it values the production of language or the formation of mind, because the two can no longer be assumed to coincide. Artificial intelligence has separated them. What once appeared naturally linked—writing and thinking—must now be deliberately reconnected through pedagogy. If that reconnection fails, the system will not collapse. It will continue to function. Writing will persist. Outputs will improve. But beneath that fluency, something essential will have receded. Students may become increasingly capable of producing language they cannot fully explain, defending claims they did not build, and submitting knowledge they do not possess. The danger is not failure. The danger is successful simulation. It is a condition in which nothing appears wrong because the visible signs of success remain intact. Students will write. They will submit. They will succeed. And thinking—slow, effortful, necessary thinking—will occur less often and be required less still. The ethical task of education is to refuse that confusion. The production of language is not the formation of thought. The sentence is not the mind. The danger is not that students will stop writing. It is that writing will no longer require a mind.

Reflection

The easier language becomes to produce, the more necessary it becomes to require the effort that gives language meaning. What has changed is not simply the availability of tools, but the conditions under which thinking is expected to occur. Writing instructors no longer operate in an environment where effort can be assumed; they must now design the conditions that make thinking unavoidable. This shift clarifies rather than diminishes their role. The goal is not to preserve struggle as an artifact of tradition, but to preserve the forms of difficulty that produce transformation. When students revise a claim that fails, defend an idea under pressure, or confront the limits of their own understanding, they are not merely completing tasks; they are becoming thinkers. Artificial intelligence can accelerate language, but it cannot replace this process without consequence. If the labor of thinking is removed, what remains is not learning but its simulation. That simulation may look persuasive, even impressive, but it cannot sustain itself when the student is asked to explain, transfer, or defend the idea beyond the conditions that produced it.

At the same time, artificial intelligence forces a deeper question: where does thinking reside? If cognition increasingly occurs through interaction—with tools, systems, and external forms of reasoning—then the boundary between internal and external thought becomes less stable. A student may develop an idea through prompting, revision, and exchange, and such a process should not be dismissed as meaningless. But the central issue is ownership. If the student can carry the idea beyond the system, adapt it to a new context, and defend it independently, then cognition has been extended. If the student cannot, then cognition has been displaced. Writing instruction must therefore ensure that thought remains transferable, defensible, and owned. Its task is not to reject the future, but to preserve the condition without which the future of education becomes hollow. What must be preserved is not writing itself, but the necessity of thinking through it.

Related Reading

If this essay has clarified what is at stake—the necessity of thinking itself—then the next step is to return to its foundation. Begin with The Work That Cannot Be Outsourced: Teaching Academic Writing in the Age of Artificial Intelligence, where writing is first established not as expression, but as the condition under which thought is formed.

 

From there, continue to The Cognitive Cost of Ease: How Artificial Intelligence Reshapes Attention, Memory, and Intellectual Ownership, which reveals what begins to erode when that condition is weakened—how fluency can persist even as cognition thins.

 

Then read What Still Forms the Mind: Preserving Wri/'.pppppppppppppop[[ting as a Condition of Learning in the Age of Artificial Intelligence, where the argument turns toward preservation, identifying the conditions that must remain if learning is to endure.

 

Taken together, these essays do not repeat this argument—they deepen it, tracing the movement from foundation, to loss, to preservation, and finally to enforcement.

 

***If you're an educator navigating the shift toward AI-resistant pedagogy, subscribe to join our weekly discussion on preserving cognitive formation

Works Cited (MLA)

Bender, Emily M., Timnit Gebru, Angelina McMillan-Major, and Margaret Mitchell. “On the Dangers of Stochastic Parrots: Can Language

Models Be Too Big?” Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, Association for Computing Machinery, 2021, pp. 610–623.

Carr, Nicholas. The Shallows: What the Internet Is Doing to Our Brains. W. W. Norton, 2010.

Flower, Linda, and John R. Hayes. “A Cognitive Process Theory of Writing.” College Composition and Communication, vol. 32, no. 4, 1981, pp. 365–387.

Graham, Steve. “A Revised Writer(s)-Within-Community Model of Writing.” Educational Psychologist, vol. 53, no. 4, 2018, pp. 258–279.

Mollick, Ethan. Co-Intelligence: Living and Working with AI. Portfolio, 2024.

Vygotsky, Lev S. Thought and Language. Translated by Alex Kozulin, MIT Press, 1986.

Wolf, Maryanne. Reader, Come Home: The Reading Brain in a Digital World. HarperCollins, 2018.