AI DOESN’T HAVE TASTE.
It has patterns, preferences and an extraordinary ability to imitate. That isn't the same thing.
By Nick Sharma · Founder & Creative Director, Rude X
There is a sentence I keep hearing whenever people talk about the future of creative AI:
“Eventually, AI will be able to create anything.”
Maybe.
But I think we're asking the wrong question.
The more interesting question is whether AI will ever know what is worth creating.
Because those are very different problems.
We tend to confuse capability with creativity. A system can generate an image, write a script, compose music, build a website, create a product concept or produce a moving sequence. It can do an astonishing number of things.
But being able to produce something isn't the same as knowing whether you should.
That's where taste comes in.
And taste is one of those strange things that is difficult to define, difficult to measure and incredibly easy to recognise when it's missing.
GOOD TASTE ISN'T THE SAME AS GOOD LOOKS
We often talk about taste as though it means making things beautiful.
It doesn't.
A beautifully designed chair can have terrible taste.
A deliberately ugly film can have extraordinary taste.
A minimalist campaign can be brilliant.
It can also be painfully boring.
A maximalist visual can be exciting.
It can also look like someone emptied every visual effect they had onto a single canvas and hoped for the best.
Taste isn't a style.
It's judgment.
It's knowing what belongs and what doesn't.
It's knowing when an idea needs more.
It's knowing when it desperately needs less.
It's understanding context.
It's understanding restraint.
It's knowing when something feels familiar in a good way and when it feels familiar because you've seen it a thousand times before.
And perhaps most importantly, taste involves the ability to reject.
That matters more than people realise.
Creative work isn't only about choosing what to make.
AI IS EXCELLENT AT GIVING US MORE
This is one of the things I love about working with generative systems.
You can explore.
You can iterate.
You can test ideas that would traditionally take hours, days or sometimes weeks to visualise.
You can take a half-formed thought and push it in ten different directions before lunch.
That is extraordinary.
But it creates a new problem.
When possibilities become almost infinite, selection becomes more important.
Imagine walking into a restaurant where every dish you've ever imagined is available.
Every combination.
Every ingredient.
Every technique.
Every cuisine.
Every presentation.
Would that automatically give you a great meal?
Of course not.
Someone still has to decide what belongs on the plate.
And that decision is where the craft lives.
AI gives us an enormous kitchen.
It doesn't automatically give us a chef.
THE MACHINE DOESN'T KNOW WHAT YOU'RE TRYING TO SAY
This is where I think the distinction between generation and creative direction becomes important.
A generative model can respond to an instruction.
It can infer patterns.
It can combine visual concepts.
It can produce something surprisingly unexpected.
But your creative intention still has to come from somewhere.
Suppose I ask for a lonely man standing in an empty train station at night.
The system may give me something visually stunning.
Rain on the glass.
Blue light.
Long reflections.
A dramatic composition.
Beautiful atmosphere.
But why is he lonely?
What happened?
Why is the station empty?
Why is the camera where it is?
What does the audience need to notice?
What am I trying to make them feel?
The machine can help me answer those questions visually.
But the questions themselves are mine.
And sometimes the best creative decision is the one the model would never suggest because it isn't conventionally impressive.
Maybe the station isn't beautiful.
Maybe it's ugly.
Maybe the light is fluorescent and unpleasant.
Maybe the camera is static.
Maybe the character is tiny in the frame.
Maybe nothing happens for eight seconds.
Maybe that's precisely the point.
THE PROBLEM WITH “MORE”
Generative AI has a very particular temptation built into it.
When something isn't working, it is incredibly easy to add.
More detail.
More atmosphere.
More dramatic lighting.
More particles.
More movement.
More characters.
More colour.
More spectacle.
More everything.
And sometimes the image gets better.
But very often, it just gets louder.
I've learned that one of the most powerful questions in creative work is not:
“What can I add?”
It's:
“What can I remove?”
Removing something requires confidence.
You have to understand the idea well enough to know what is essential and what is decoration.
You have to trust the audience enough not to explain everything.
You have to be comfortable with empty space.
That's taste.
AI CAN IMITATE CULTURE. IT DOESN'T EXPERIENCE IT.
This gets even more interesting when we move beyond aesthetics.
Creative work exists inside culture.
It carries history.
References.
Memory.
Politics.
Class.
Fashion.
Architecture.
Music.
Cinema.
Human behaviour.
A filmmaker isn't working in a vacuum. They're borrowing from things they've seen, experienced, loved, hated, misunderstood and obsessed over.
That's part of why human creative work can feel personal.
A machine can recognise patterns across enormous amounts of cultural material.
It can reproduce the appearance of those patterns with remarkable fluency.
But recognising a pattern and having lived inside it are different things.
That's why I think the human perspective remains so important.
The weird references.
The personal obsessions.
The accidental influences.
The things you can't quite explain.
The memory that suddenly changes the way you approach a scene.
Those are difficult to reduce to a prompt.
And honestly, I hope they stay that way.
ORIGINALITY IS NOT THE ABSENCE OF INFLUENCE
There's another misconception I think AI has made worse: the idea that originality means creating something that has never existed before.
Almost nothing works that way.
Artists steal.
Filmmakers reference other filmmakers.
Designers borrow ideas.
Musicians sample.
Architecture evolves through repetition and reinvention.
Originality is rarely about creating from nothing.
It's about what you do with what already exists.
Two people can have access to the same references and produce completely different work because their judgment is different.
Their experiences are different.
Their obsessions are different.
Their point of view is different.
That's why two people can use the same model, the same tools and even similar prompts and still produce work that feels completely different.
The tool is shared.
The taste isn't.
AI CAN SURPRISE YOU. THAT'S PART OF THE MAGIC.
None of this means I want AI to become predictable.
Quite the opposite.
Some of my favourite moments working with generative systems happen when the result isn't what I expected.
The machine introduces something I hadn't considered.
A shape.
A composition.
A detail.
A visual relationship.
A strange accident.
And suddenly the idea moves.
That's exciting.
Because creative collaboration isn't necessarily about controlling every variable.
Sometimes it's about knowing when to listen to the accident.
But even then, someone has to recognise that the accident is worth keeping.
The machine can surprise me.
I still have to decide whether the surprise is good.
MAYBE THE MOST IMPORTANT AI SKILL IS KNOWING WHEN TO SAY NO
This is probably the part nobody puts into an AI workflow diagram.
We talk about prompting.
Generation.
Iteration.
Upscaling.
Editing.
Post.
We rarely talk about rejection.
Yet rejection might become one of the most valuable creative skills in an environment where production is nearly unlimited.
If I can generate fifty variations of an idea in the time it used to take me to produce one, then I don't need more ideas.
I need better judgment.
I need to know which five are interesting.
Which one deserves another round.
Which one is technically impressive but emotionally empty.
Which one is weird in exactly the right way.
And which one should be deleted immediately.
More output doesn't automatically create better work. Better decisions do.
SO, DOES AI HAVE TASTE?
I'm not convinced that's the right question either.
Maybe one day AI systems will develop increasingly sophisticated models of aesthetic preference, context and quality.
Maybe they will become remarkably good at predicting what humans consider beautiful, moving, funny, elegant or provocative.
But predicting taste and possessing taste may not be the same thing.
For me, taste is deeply connected to intent.
It's knowing what you care about.
It's knowing what you believe.
It's understanding why you're making something.
It's taking responsibility for the choice.
And that's why I don't think the future of creative AI belongs to whoever can generate the most.
I think it belongs to whoever can see the difference between something impressive and something meaningful.
That's a much harder skill.
And thankfully, it's still a human one.
At least for now.
Nick Sharma
Founder & Creative Director, Rude X
Nick explores the intersection of filmmaking, generative AI, visual storytelling and creative technology, with a particular obsession for cinematic worlds that don't exist — yet.
Comments