AI, machinery, drawing, and the problem of letting something else into the studio

First Contact With Paper

Image above: ‘Gedling’. Hand carved mining boot soles. (With additional imagery and text field by AI).

AI, machinery, drawing, and the problem of letting something else into the studio

I did not set out to make an AI drawing project.

That sounds too deliberate. Too clean. Too much like a proposal written after the event to make the accident look respectable.

What actually happened is less tidy.

AI entered the studio because it had become impossible to ignore. At first it seemed like another tool to test. Then it began to affect more than the tool shelf. It interfered with drawing, correction, proportion, printmaking, engraving, judgement, and time.

Especially time.

That is the part people leave out. These systems promise speed, but they also produce endless detours. One test leads to another. One mistake becomes interesting. One interesting failure suggests a new process. Soon the machine has not saved time. It has occupied it.

This may be useful. It may also be a sophisticated way of avoiding the work by making more work.

That is still undecided.

Slogan bastards

The language around AI and art is already full of slogans.

Human-machine collaboration.
Expanded intelligence.
The future of creativity.
Post-digital whatever.

Some of these phrases may contain a useful thought. Many of them are slogan bastards: phrases that sound convincing until asked to explain themselves.

One of mine was:

“The machine gives the hand something new to answer.”

I disliked it almost immediately.

It was not entirely false. It was trying to say that when a drawing is passed through AI, plotting, engraving or print, the artist receives something back: a second condition, a misreading, a correction, a material change. The hand then has to decide what to do with it.

But the phrase was too smooth. It made the process sound resolved, when the actual work is not resolved at all.

The real question is simpler and less comfortable:

What changes when a hand-drawn mark is corrected, translated, machined, engraved, printed and returned?

And after that:

Is the result worth keeping?

The machinery problem

The machinery is part of the attraction.

A laser engraver moving across wood is compelling to watch. So is a pen plotter slowly making a drawing. So is a 3D printer building something over hours, layer by layer, while the room has gone quiet.

There is a strange pleasure in seeing work continue after you have stopped.

But that is also where the danger begins.

A machine can make things while you sleep. That does not mean the things matter. A full surface is not the same as a strong image. A completed process is not the same as a finished thought.

Artists have always used delegated processes. A kiln fires without the potter inside it. A press bites paper through pressure. A photographic exposure continues after the shutter opens. Etching, casting, printing, moulding, cutting and drying all involve waiting, risk and material behaviour outside immediate control.

So the problem is not delegation.

The problem is judgement.

Did I set the conditions carefully?
Do I understand what the machine is doing?
Can I read the result?
Can I reject it when it is weak?
Does it change what I see or only produce more stuff?

Without judgement, the machine is just active.

Activity is not art.

The hand has not disappeared

The hand remains central because drawing is not only image production.

Drawing records looking, hesitation, pressure, error, habit, correction and return. A line is never just a line. It carries a decision, even when the decision is wrong.

The red correction works interest me because they leave the disagreement visible. The first drawing is not erased. The correction does not politely replace it. Both remain present.

That matters.

The correction layer may improve the drawing. It may also expose its weakness too brutally. It may produce a better proportion but a colder image. It may tidy away something that should have remained awkward. It may reveal that accuracy and life are not the same thing.

This is where AI and machine process become interesting to me: not as replacement, but as disturbance.

The machine does not understand the drawing. It does not feel the pressure of the mark. It does not know why an image matters.

But it can still alter the conditions of looking.

It can produce a version, a correction, a structure or a material translation that forces the artist to decide again.

That decision is where the work either survives or fails.

Paper, wood, burn, ink

I keep returning to physical processes because they make the question harder to avoid.

A screen image can look convincing too quickly. It can arrive polished, complete and slightly dead. It can flatter the eye before the hand has had a chance to object.

Paper, wood, burn and ink are less forgiving.

The plotter has to move.
The laser has to burn.
The pen has to touch the surface.
The material has to accept, resist, stain, split, blur or hold.

Once the image enters material process, it becomes accountable in a different way. It is no longer only a possibility on a screen. It has consequences.

Sometimes the machine improves the image. Sometimes it exposes how thin the image was. Sometimes the engraved or printed version has more authority than the drawing that began it. Sometimes the process makes a weak idea look important for about ten minutes.

That last danger is real.

A process can dignify nonsense. A machine can give a bad image better manners.

So the test has to remain severe.

Does the process deepen the work, or only decorate it with effort?

What the machine does not do

I do not want to pretend the machine is more mysterious than it is.

It does not see.
It does not care.
It does not understand my work in any human sense.
It does not know my history, my family, my drawings, my teaching, my printmaking, or why certain images return again and again.

It works through systems, patterns, procedures and outputs.

But that does not make it irrelevant.

A press does not understand grief. A burin does not understand history. A camera does not understand love. Tools and machines do not need human feeling in order to alter what humans make.

The important question is not whether the machine is creative in the way a person is creative. I do not think it is.

The better question is whether the machine can become part of a situation in which human judgement is sharpened, interrupted or changed.

Sometimes the answer appears to be yes.

That is why I have not yet stopped.

The difficult middle

This work is not yet a method. It is not yet a conclusion. It is not a triumphant announcement that AI has transformed my practice for the better.

It has disrupted it.

It has taken time I did not plan to give it. It has opened doors and damaged certainties. It has produced images I would not have made otherwise. Some are probably dead ends. Some may not be.

That is the difficult middle.

If the results were obviously worthless, I could stop. If they were obviously extraordinary, the decision would be easier. Instead they occupy the more troublesome territory where something may be happening, but it is not yet clear what it is.

So for now I am treating the work as a test.

Does it change what I see?
Does it make me draw differently?
Does it survive translation into print, engraving or object?
Does it connect to the older concerns in my work: memory, trace, displacement, material witness, survival?
Does it still matter the next day?

If not, it goes.

If yes, it continues.

For now

I do not know whether this thing can be tamed. I am not sure taming is even the right aim.

Perhaps the better task is to keep it under enough pressure that it has to produce evidence: drawings, prints, engravings, failures, decisions, text, and changed ways of looking.

Not promises. Evidence.

That is where the work is at the moment.

Unsettled, demanding, not entirely welcome, but not yet dismissible.

And yes, still producing bastard slogans.

They will be dealt with individually.