Raymond Klassen – Ideals and Identities

I deliberately used artificial intelligence 4 times today. That sentence would have sounded strange to me only a few years ago. It may still sound suspect to many people, especially in an academic context. To say that one has used AI in writing, teaching, editing, or thinking now carries a shadow of possible dishonesty. Did I really write this? Did I think it through? Did I use the machine to deepen my work, or to escape the labour of thought?

My workplace has been adopting AI with increasing speed. There is a feeling that postsecondary institutions are behind the gun in developing policy, in educating about using AI, and in buttressing the threat to learning and human agency that AI poses.

These are not trivial questions. They go to the heart of academic integrity. But they also go deeper than academic integrity, because they ask what it means to be an author at all.

Authorship is not simply the production of words. It is not merely the ability to generate a fluent text. Nor is authorship reducible to originality in the narrow sense of never having been helped, influenced, edited, challenged, or formed by another. No serious writer writes alone. We write with teachers, books, friends, editors, enemies, memories, communities, traditions, and arguments pressing upon us. The solitary author is partly a myth.

But authorship still matters. It matters because someone must be responsible for what is said. This is where artificial intelligence becomes morally and educationally dangerous. It can give us a text that looks finished before the person who submits it has become answerable for its claims.

That is the academic integrity problem.

The problem is not simply that AI can “write the paper.” The deeper problem is that AI can make it possible to appear as though one has done the work of understanding when one has only managed the appearance of understanding. It can produce the legibility of thought without the discipline of thought.

I have worried about this for some time. In The Human Person and History in the Age of Mass Reproduction, I drew from Walter Benjamin’s concern that mechanical reproduction detaches a work from its unique presence in time and space. Benjamin’s concern about the work of art becomes, in our age, a concern about the human person. The avatar, the profile, the data image, the digital version of the person, all begin to circulate independently of the actual person. The person is reproduced, analyzed, distributed, and made available without necessarily being encountered.

AI intensifies this problem. It does not merely reproduce an image of the person. It can reproduce the style, tone, argument, and intellectual appearance of the person. It can make thought appear where no one has yet taken responsibility for thinking. This is why authorship cannot mean only “who typed the words?” That question is too small. The more important question is: who can answer for them?

To author something is to become answerable for it. It is to say: I have taken responsibility for this claim, this argument, this phrasing, this omission, this source, this judgment. I may have been helped. I may have been edited. I may have argued with others. I may have used a tool. But I am not hiding behind any of them. The words have passed through my judgment. I stand behind them.

In The Aura of Thought: McLuhan and Benjamin on AI, I imagined McLuhan and Benjamin in conversation over AI. The dialogue asks whether the aura of thought itself is at stake. If the work of art once risked losing its aura in the age of mechanical reproduction, then perhaps thought risks losing its aura in the age of artificial intelligence. The issue is not whether AI can produce something impressive. It can. The issue is whether thought can remain connected to presence, struggle, memory, and moral cost.

That is where academic integrity must be rethought. Integrity is not protected by pretending tools do not exist. Nor is it protected by panic, detection software, or the romantic fantasy that true thought only happens when no mediation is involved. Thought is always mediated. Language is a medium. Books are media. Classrooms are media. The university itself is a medium.

But some forms of mediation deepen responsibility, while others dissolve it.

A teacher who challenges my argument may make me more responsible for it. An editor who asks me to clarify a sentence may require me to know more precisely what I mean. A peer who disagrees with me may force me to test the strength of my claim. Even AI, used carefully, can play a role in this kind of dialogical refinement. It can ask questions, propose distinctions, identify repetition, help organize material, suggest counterarguments, and assist with formatting.

But AI becomes destructive when it allows me to avoid ownership. If I submit words I do not understand, sources I have not checked, claims I cannot defend, or arguments I did not judge, then the problem is not merely technological. It is moral. I have allowed the appearance of authorship to replace authorship itself.

Instead of asking whether AI was used, it might be more appropriate to ask: what kind of use was it?

Was AI used to avoid the work, or to deepen it? Was it used to fabricate competence, or to refine real competence? Was it used to impersonate understanding, or to help articulate understanding that the writer is willing to test and defend? Was it used secretly in order to deceive, or transparently as one tool within a larger process of human judgment?

These distinctions matter.

In On the Bridge of Khaza-dum: A.I. and April Fool’s Day, I used the image of Gandalf confronting the Balrog to describe the AI as a demon that our greed has called forth. It was an intentionally dramatic image, but the drama was not misplaced. AI forces us onto a narrow bridge. Behind us is a world in which many of our old assumptions about writing, learning, originality, and academic work no longer hold. Ahead of us is a world in which the human person may be tempted to disappear into simulation, efficiency, and control.

The temptation is not only that students will cheat. That is the surface problem. The deeper temptation is that institutions will quietly accept the logic of mastery that made AI so attractive in the first place. If education becomes only credential completion, output production, task efficiency, and measurable performance, then AI is not a corruption of the system. It is the system’s logical conclusion.

The pressure test that is AI in academia is not only about the integrity of the person. What I have seen the instructors acting blind to is that it pressure-tests the system itself.

This is why the question of academic integrity cannot be separated from the question of education itself. What is education for? If education is merely the production of correct outputs, then AI will often appear superior to students. But if education is the formation of judgment, attention, truthfulness, memory, courage, and responsibility, then AI can never replace the learner. It can only assist, distort, or obscure the learner’s formation.

Authorship belongs to that formation. To become an author is not simply to become expressive. It is to become responsible. It is to learn how to speak in such a way that one’s words remain connected to reality, to sources, to other people, to history, and to one’s own moral judgment. The author is not the person who has never been helped. The author is the person who can answer.

This also means that responsible AI use requires practices of integrity. I should be able to say when AI has helped me. I should not pretend that machine-generated fluency is the same as my own understanding. I should check sources. I should revise. I should remove what is not mine in any meaningful sense. I should refuse claims I cannot defend. I should allow the tool to serve judgment, not replace it.

This may be especially important for writers, teachers, and students. We should not teach young people that integrity means never receiving assistance. That is false, and it misunderstands how learning happens. But neither should we teach them that all assistance is equal. Some assistance forms the person. Other assistance bypasses formation.

A calculator can help with arithmetic, but it cannot make someone mathematically thoughtful. A grammar tool can improve a sentence, but it cannot make someone truthful. AI can help generate, revise, summarize, organize, and clarify, but it cannot make the human being answerable. Only the human being can accept responsibility.

So perhaps the right question for academic integrity in the age of AI is not, “Did you write every word by yourself?” The better question is: “Can you answer for this?”

Can you explain it? Can you defend it? Can you revise it? Can you identify the sources? Can you say what changed in your understanding? Can you tell the difference between what the tool supplied and what you judged to be true? Can you stand behind the final work? If the answer is no, then authorship has been hollowed out. If the answer is yes, then AI has not replaced authorship. It has been placed under authorship.

This is the line we must learn to hold.

It helps us write, but the danger is that it may help us disappear from what we write. It may detach words from presence, claims from judgment, and fluency from truthfulness. But if we use it under the discipline of responsibility, then we may yet preserve the aura of thought.

Don’t refuse every tool; refuse to become one.

Author’s note: This article was developed through AI-assisted dialogue and revision. I remain responsible for the argument, the judgment, the links, the final wording, and any errors that remain.