I want to reflect on the form and intention of my use of language models when it comes to writing. The concern is simple to formulate and difficult to resolve: I fear I’m suffering the negative consequences of what I criticise in the things I publish. When using these tools, there are times I have the sensation — more fear than anything else — that it’s not me who writes and, worse, that it won’t be me who even has the ideas.

This isn’t a question of copyright but of existential conception. It’s not plagiarism or originality that sit at the front of my concern — it’s the entanglement between authorship and authority over the putative ideas. I question, therefore, the authorship of my own ideas — and the word “my,” in that sentence, already doubts itself.

Recently, preparing a small workshop on the subject made me reread what I’ve written and revisit the way I use these systems. The readings, the conversations, and my incessant and useless reflections left me more attentive to my own practice — and, in particular, to writing.¹

One line of thought leads me to consider that I may be, at times, delegating “only” the task of writing: I orient the systems extensively from my own ideas, give them access to things I’ve already written, and don’t blindly accept the results they produce. I spend time revising and editing. But what if I’m not only delegating what I think I am? What if it’s attention, presence — whose quality has repercussions on associative capacity and memory, not to mention the quality of relationships, when other people are involved — that I’m ceding?²

What worries me is that I’m increasingly becoming an editor and, consequently and progressively, less of a writer. It’s still work — it demands effort, attention, dedication — but it’s not the same kind of work.

Writing is, so often, the encounter with blockages, with the failures and gaps in my reasoning, and with the natural challenge of translating feelings into words that hold depth and breadth but remain simple to read and understand. Better still if they stay in the thoughts of those who read them, and if the memory of those readers has the space and interest to save them from oblivion or, worse, indifference. A writer thinks in order to create something he believes to be original, even if it’s just a point of view; thinks in order to communicate, to share something that seems worthy of being thought by someone else; to understand something complex, to give, share, and disseminate meaning. Of course there are other reasons — narcissistic exercises, exhibitionist impulses, or transactional motives.

An editor doesn’t start from nothing. He starts from what already exists — and his work is to improve it. To do so, he must combine recognition and critique of both the creative act and the creation itself. Is this a form of creativity? I think so. But it’s not a twin, not even a sibling, of writing-creativity. Editing-creativity is a cousin, perhaps.

It’s obvious to me that using these systems represents a way of avoiding the discomfort of the blank page, but also another kind of discomfort: the one that comes from doubt about the quality of what you’ve created. I catch myself looking for justifications — tiredness, circumstances, the lack of time to think — and then censoring myself for listing excuses instead of facing the question.

What occupies me next are the probable consequences. Is my capacity for abstraction being compromised? Is it being amplified? Am I being lazy — in the sense of refusing to spend the energy that will bear fruit in the medium and long term? Am I being efficient in how I spend my attention, or am I confusing efficiency with economy?

I feel these doubts at the moment I write these words — and this sentence, at least, I know is mine.

Will we increasingly become editors? Some believe or wish that to be the near future. If it happens, will machines write for us? And if they do, where will the ideas come from? Before that, how long will we be willing to wait before our natural tendency toward energy conservation overtakes discernment, taste, the pride — and risk — of authorship?

For now, I content myself with the exercise of continuing to write, without the help of language models beyond the one I carry in myself. And with the subsequent task of revising and editing what I’ve just written, without forgetting to wait a little, to allow time and distance between the two kinds of creativity.

The use of non-human language models gives us distance, or at least its illusion. What remains to be seen is what we might end up closer to.

1. For readers unfamiliar with my earlier attempts to think through these questions from the outside: “The (Sad) Story of the Chatbots That Know Us Better Than Our Mothers” was my first sustained engagement with what we outsource when we delegate language to machines. “Four Hundred Truths” examined the uniformity that emerges when four hundred students use AI to write “in their own voice.” This piece turns the lens in the other direction. ↩︎
2. I explored the economics of attention in “On the Beauty of Distraction” and “Slop: On the Viscosity of the Present.” The argument there was structural — about systems designed to capture and monetise focus. Here the concern is more intimate: what happens to my attention when I let a machine carry the weight of articulation? ↩︎