A few weeks ago a student circulated an open letter calling for the abolition of generative AI within the art school. It was carefully written, strongly felt, and widely shared. The language was urgent, at times absolute: there is no place for it here; it harms the environment; it erodes thinking; it steals. The document gathered signatures quickly. It also produced, almost immediately, a different kind of response—quieter, procedural—centred on conversation, on complexity, on how such concerns might be taken up within a shared space rather than imposed upon it.
What was striking was the tone: the sense that something had tipped beyond negotiation into principle. The question shifted from how one might work with a tool to whether one ought to stand against it altogether. This kind of shift—away from adjustment and toward refusal—is not new, though it often feels new when it arrives. It touched, perhaps unexpectedly, on earlier habits and traditions of the art class and the studio. There are small rules we remember without quite knowing when we first absorbed them: not to use a ruler when drawing a straight line; not to trace unless it is clearly declared; to be wary of the grid as anything more than a temporary support. These are not universal prohibitions, and they vary by discipline, but they carry with them an implicit claim: that certain kinds of assistance alter not just the outcome, but the quality of attention that produces it.
A ruler produces a straight line quickly and reliably. It also removes the slight variation that comes from the hand—those small hesitations and accelerations that give a line its character. For a beginner, this distinction can feel abstract. The ruler appears to improve the drawing. Only later does it become possible to see that something has been traded away, and that the point of the exercise was never simply straightness.
Tracing and grids have a somewhat different role because although they’re often treated as shortcuts, they are also quietly built into teaching. They enable a likeness where none might yet be possible. They provide a structure that can be worked through rather than confronted all at once. Used lightly, they support looking; used heavily, they can replace it. The difference isn’t always visible in the result, but it is felt in the process.
In each case, what counts as “cheating” depends less on the tool itself than on what we believe the task to be. If the task is to produce an image that corresponds to another, then these methods are efficient. If the task is to produce an image that corresponds to another, then these methods are efficient. If the task is to judge proportion, weight, and interval, then they may interfere. The tension sits not in the object, but in the relation between tool and intention. This makes it difficult to speak about new tools in categorical terms. Generative AI arrives not as a single technique, but as something that seems to gather many of these earlier forms together. It can produce images without drawing, text without drafting, structure without the slow calibration that usually accompanies it. It can also be used in smaller, more provisional ways—as a prompt, a suggestion, a means of testing a direction. The uncertainty arises because it is not always clear, either to the user or the observer, where along that range a particular use sits.
The student’s letter treats this uncertainty as danger: if a tool can replace thinking, then it will. There is some truth in that. Anyone who has worked with students knows the temptation to bypass the difficult middle of a task—the part that is neither first idea nor finished outcome, but something slower and less defined. Tools that promise immediacy have always exerted that pull. But it is also true that practices adjust around their tools. Photography was received, in its early decades, with a comparable intensity—accused of displacing the hand, reducing art to mechanism, and seducing the public with surfaces. Baudelaire described it as a threat not simply to art, but to thought itself—an intrusion that would reduce imagination to mechanical reflection. The force of that response is familiar but it is now possible to see that photography did not eliminate drawing so much as redistribute its purpose. What had been necessary in one context became optional in another; what had been valued for its fidelity became valued for its inflection.
The arrival of digital photography produced a similar disturbance. The darkroom, with its particular forms of knowledge, was set aside by many. Certain skills faded quickly; others emerged in their place. Again, some losses were real, and were experienced as such. But the practice did not end. It shifted, sometimes awkwardly, sometimes productively.
There is a risk, in moments like this, of stepping back too far in the name of principle. To refuse a tool entirely is to remove oneself from the conditions in which work is now taking shape. That refusal can be a position, but it also has consequences. It can leave decisions about the use and meaning of the tool to others—often those less concerned with the questions that prompted the refusal in the first place. I’m not arguing for uncritical adoption. If anything, I’m suggesting the opposite: that engagement requires a clearer sense of what we are trying to preserve. Not outcomes, perhaps, but forms of attention. The question becomes less “is this tool allowed?” and more “what does this use of the tool enable, and what does it displace?”
There is also the quieter matter of access. Assistance has never been evenly distributed. Some have always had the means to refine their work through others—through editors, technicians, assistants, various forms of unseen labour. When a new tool appears that makes certain kinds of support more widely available, it can unsettle not only practice, but the expectations that surround it. What looks like a levelling or democratisation from one perspective can appear as erosion from another. Neither description is entirely wrong, but neither is sufficient on its own.
Within teaching, these questions tend to surface in small ways. A piece of work arrives that is unusually resolved, or curiously flat. A draft appears too quickly. The difficulty is not identifying the use of a tool, but understanding what has been learned through it, and what has been missed. That understanding cannot be legislated in advance. It has to be worked out, case by case, through looking and discussion.
The open letter, in this sense, can be read not only as a demand, but as a symptom of uncertainty about where those discussions are taking place, and how visible they are. To call for a ban is to simplify the problem into something decidable. It removes the need for more granular judgements—the kind that are slower, and sometimes inconclusive.
It may be that the more useful task is to stay with that difficulty a little longer. To notice where attention shifts under new conditions; to ask what is being eased, and at what cost; to recognise that what once felt like a shortcut may, over time, become ordinary, and that this ordinariness is not neutral but reorganises how we work and what we expect of ourselves and others.
The question, then, may not be whether we permit or refuse such tools, but how we live with the changes they introduce into our habits of attention. What we call cheating often marks a moment where expectations have not yet settled. Given time, some of those judgements harden into convention, while others fall away. What remains constant, perhaps, is the need to notice what our tools are doing on our behalf, and what they quietly ask us to relinquish in return.
