I used to think art belonged to a different world.
Paintings belonged in galleries.
Music belonged on stages.
Poetry belonged in books.
Engineering belonged somewhere else entirely.
Engineering was mathematics. Physics. Equations. Constraints. Systems that either worked or did not.
Then I began to build.
And I realized that the boundary was never there.
Art Begins Before the Artifact
A painting does not begin with paint.
A song does not begin with a sound.
A building does not begin with concrete.
They begin earlier—with someone seeing something that does not yet exist and deciding that it should.
That decision is invisible from the outside.
No dataset contains it.
No equation requires it.
There is no measurement that tells you what should be created.
The world simply contains possibilities.
A human chooses one.
That is where art begins.
Science Discovers
Science asks what is already true.
What laws govern the universe?
What patterns exist?
What structures are hidden beneath observation?
Science gives us knowledge about reality.
That knowledge is extraordinary because it removes illusion. It tells us what the world permits and what it does not.
But science alone does not tell us what should come next.
Knowing that something is possible does not tell us that it is worth building.
The equation does not choose its application.
The discovery does not choose its purpose.
Something else has to decide.
Engineering Realizes
Engineering takes what is possible and makes it real.
It turns principles into systems.
Ideas into mechanisms.
Theory into objects that survive contact with reality.
But engineering also has a choice hidden inside it.
There are usually many ways to solve a problem.
Many architectures.
Many abstractions.
Many compromises.
The textbook rarely tells you which one is beautiful.
The specification rarely tells you which one will matter.
That requires judgment.
And judgment is where engineering begins to resemble art.
Art Gives Direction
This is the part I have come to value most.
Art is not merely the production of something aesthetically pleasing.
At its deepest level, art is direction.
It asks: what should exist? What is worth making? What deserves to be seen? What should the future feel like?
These questions cannot be answered by optimizing an objective function that has already been defined.
Someone has to define the objective.
Someone has to care about the outcome before the outcome exists.
That is vision.
The Moment of Insight
There is a moment in difficult engineering problems when the work changes character.
At first, everything is technical.
You measure. You calculate. You test. You fail. You try again.
Then, sometimes, something happens.
You see the problem differently.
The solution does not become easier because you calculated faster.
It becomes easier because you finally understood what the problem actually was.
That moment is not reducible to a checklist.
It is insight.
And insight is one of the most artistic experiences I know.
When Mathematics Becomes Intuition
When I first encountered machine learning, I saw linear regression as an equation.
A line through data.
A mathematical object.
But after enough time with the idea, the equation stopped feeling like an equation.
It became intuition.
I could begin to see relationships rather than calculate them.
That transformation changed how I understood technical knowledge.
The highest form of mastery is not remembering more formulas.
It is when the formulas become part of how you see.
A mathematician develops intuition. A musician develops an ear. A painter develops an eye. An engineer develops judgment.
The disciplines look different from the outside.
The internal transformation is remarkably similar.
AI as an Art
This is why I eventually stopped thinking about AI purely as mathematics or statistics.
The mathematics matters. The statistics matter. The optimization matters.
But they are tools.
At a certain depth, building intelligent systems begins to feel less like solving exercises and more like learning a medium.
You learn what the medium can express.
You learn its limitations.
You learn when to push it.
You learn when a seemingly clever solution is actually wrong.
And eventually, you develop intuition for structures that you could not have articulated when you first began.
That is what I mean when I call AI an art.
Not because it is vague.
Because mastery transforms technique into intuition.
Machines Can Reproduce the Artifact
This is where the question becomes more complicated in the age of AI.
A machine can generate an image. It can compose music. It can write poetry. It can imitate an artist's style. It can even create combinations that no human has explicitly produced before.
The artifact is there.
The pixels exist. The waveform exists. The words exist.
But the existence of the artifact does not answer the most interesting question.
Why this?
Why this subject? Why this form? Why this moment? Why was this worth making?
A model can reproduce the visible consequence of a decision.
That does not automatically explain the origin of the decision.
Meaning Comes Before Optimization
An optimizer can find the best solution to an objective.
But who decides what the objective should be?
A system can maximize beauty according to a metric.
Who decides what beauty means?
A model can generate ten thousand possible futures.
Who decides which future is worth wanting?
These questions matter because civilization is not merely an optimization problem.
Before we optimize, we choose what matters.
Before we build, we choose what should exist.
Before we create, we imagine.
That is why art is not a decorative layer added after engineering.
It is upstream of engineering.
The Engineer as Artist
The best engineers I admire are not merely excellent at implementation.
They see.
They notice structure that others overlook.
They reject assumptions that have become invisible.
They find a simpler abstraction.
They know when a system has become unnecessarily complicated.
They can look at a problem and imagine a form that nobody has built yet.
That is artistic judgment.
Not decoration.
Not aesthetics.
Vision expressed through technical constraints.
This Is How I Think About My Own Work
In brain-computer interfaces, the obvious question was: how can we decode the brain?
The more interesting question became: how can we make the brain–machine interface feel like a limb?
The difference is not more computation.
It is a different question.
In neuro-symbolic systems, the obvious question is: how can we make a model more accurate?
The deeper question became: how can perception and reasoning be separated so that the system knows what it has evidence for?
Again, the difference begins with how the problem is seen.
In autonomous scientific systems, the obvious question is: how can we process this dataset?
The deeper question becomes: how can a system learn what the dataset means before deciding how it should be processed?
The pattern repeats.
The breakthrough often begins before the implementation.
It begins with the question.
Art Is Not the Opposite of Rigor
There is a misconception that artistic thinking means intuition without discipline.
I believe the opposite is closer to the truth.
Art at a high level requires enormous discipline.
A great musician practices scales. A great painter studies anatomy. A great architect learns structure. A great engineer learns mathematics.
Rigor is not the enemy of art.
Rigor gives vision a language.
Without technique, vision remains an idea.
Without vision, technique becomes execution without direction.
The two need each other.
Why This Matters More in the Age of AI
As machines become better at execution, execution becomes less scarce.
Generation becomes cheaper.
Optimization becomes faster.
Iteration becomes nearly instantaneous.
That makes the human ability to choose what is worth doing more valuable, not less.
When everyone can generate a thousand images, the difficult question becomes which image matters.
When everyone can generate a thousand products, the difficult question becomes which product deserves to exist.
When machines can explore millions of solutions, the scarce resource becomes the ability to recognize the right problem.
The future may therefore reward people who can combine technical depth with unusual clarity of vision.
Not people who merely know how to use powerful tools.
People who know what to ask them to create.
The Work Is the Art
I don't think art has to hang on a wall.
It can be a piece of mathematics.
A laboratory instrument.
A scientific theory.
A beautifully designed algorithm.
A prosthetic device.
A new interface between a human and a machine.
Anything can become art when it carries a human act of understanding and intention into reality.
The medium changes.
The underlying act does not.
Someone saw a possibility.
Someone believed it mattered.
Someone spent years making it real.
Looking Forward
I don't know whether machines will eventually create things that deserve to be called art.
Perhaps that question will become harder to answer.
Perhaps the distinction between human and machine creation will become philosophically complicated.
I don't think that should make us afraid of machines.
It should make us more serious about what we value.
Because the purpose of technology is not merely to increase what can be produced.
It is to expand what can be possible.
And possibility still requires direction.
Someone has to look at the world as it is and imagine the world as it could be.
Someone has to decide that the new world is worth building.
That is the part of creation I care about most.
The artifact comes later.
First comes the vision.
And perhaps that is what art has always been.