In ancient times, artists sold their paintings on the streets. Ordinary citizens and country folk wandered about, freely pointing and commenting on their work. And the finest art from that era still strikes us today with a beauty that seems to transcend time.

There was a time when the foolish common folk marveled — and then hesitated — because suddenly they could no longer understand avant-garde art. Works that experts praised to the heavens looked, to many people, like outright visual pollution.

Art had entered a new era of "highbrow refinement," and the vulgar masses were no longer worthy of these artists. Or so they said, right?

First, we need to establish a premise: all art — whether classical, modern, postmodern, or contemporary, whether images or sounds — exists for human experience. It is presented before your eyes, it surrounds your ears, all to express one or several specific emotions, or to let you feel a certain experience. If it falls short of expectations, then it is a failed demonstration.

In those ancient times, craftsmen only needed to "paint more realistically" or "carve faster" to sell their work at a good price. So during that period, painters put enormous effort into finding ways to describe three-dimensional spatial relationships on a two-dimensional plane. The stone carvers not only had to carve the shape of, say, a certain appendage in stone, but also spent an extraordinary amount of care trying to accurately depict the forms of tendons and veins on the body's surface.

The trendiest artists of that era, on one hand, were accumulating all kinds of theoretical foundations — space, light and shadow, color, and even mechanics and anatomy — while on the other hand, they still had to produce a complete work to prove these ideas.

Until cameras and 3D modeling — those plain, modern technologies — killed the game. A thick scroll painted with someone else's wife was still too crude; film could capture a moment instantly, and automated ink could record everything. If needed, a modern personal computer could enlarge any required detail and remove every blemish on the skin. The avant-garde among painters were like abacus masters staring at calculators, like chess players staring at AlphaGo. Speechless.

I remember when AlphaGo first began rapidly improving at Go, some people consoled themselves — saying things like it had only beaten those European amateurs, or that it had merely "read" a lot of "game records." But what they failed to grasp was the core point: humanity had found a sound way to define a loss function and ground truth, allowing a computer to propose "the next move" within a reasonable time at every step. Sure enough, very soon, in a devastating sweep, no human could beat AlphaGo. Not long after, along came AlphaGo Zero, which could handle humanity without any game records at all.

Let's dig a bit further into ground truth and loss functions. Ground truth you can understand as "the standard answer." That is, every time we review a game, we might scan the whole board and search for the correct move for that position at that moment. The loss function defines the gap. Training a model is, at its essence, continuously adjusting its parameters so that the gap gradually shrinks to zero. Previously, the barrier in Go was that although an optimal solution existed, the computational cost was potentially too enormous to define the loss.

And this is where art immediately shows its difference:

Faithful recording is indeed beautiful. The sea breeze, the sky, and beautiful people can all be defined as "beauty," so a camera can imitate this kind of "beauty" with great ease — and soon surpass humans at it. But these are only instances of "beauty"; the "standard answer" itself is not nearly so concrete. So many people began working as feature engineers of "aesthetics."

What is a "feature engineer"?

One example I think illustrates this nicely is The Devil Wears Prada, in which Miranda explains to the rookie played by Anne Hathaway: look at these garish clothes on the Paris runways today; tomorrow, elements of them will become mass-produced garments, and eventually they will reach every obscure little factory, reused by street-stall clothing. Although their designs are never used directly, the colors, the pairings, and various structural logics within them become a kind of source, living on broadly in all sorts of places.

That "source" is the "feature." By capturing the features, we can easily transform a moment of inspiration into mass-produced yet effective industrial goods using more modern methods.

Another example: Morandi colors. They come from an Italian painter, Giorgio Morandi, a guy who never left his small town his whole life. I've been to his exhibition — many rooms showing his works from different periods, plus a little attic displaying how his studio was arranged back then, and at the very back a few sheets of draft paper with a few pencil lines on them, not so different from what your deskmate doodled when bored in class as a kid. Along the way, there were also people, left hand holding a little sketchbook, right hand sketching away, copying the works. Without context, it all feels a bit bewildering.

So I tagged along with a class — an art teacher leading her young students, introducing the era of his works, his thinking, the core logic, and so on. Of course, this included the widely known Morandi color palette.

Honestly, some of the works, apart from a few highlighted features, are quite sketchy in the rest. Many modern products built on Morandi colors are actually far more refined than his pile of works — to my crude senses, they elevate things by quite a few notches. But without those theoretical foundations, those modern industrial products would lose much of their luster. Those theories are the "features."

The problem with some self-proclaimed "professionals" in today's society is that they don't know the why behind things, so they always like to describe a material with phrases like "as an artist, Tom's work inherits A, breaks through B, and is the founding work of C-ism." And they forcibly declare these things beautiful. This kind of beauty and a vulgar person like me's understanding of beauty are like a certain doctor's liver versus my liver, or a certain religion's self-proclaimed faith versus my faith — they exist in a state of "similar-yet-different duality," depending on what the "professionals" need.

And at this point, a very serious problem arises: regarding A, B, and C, the "professionals" always love to act mysterious, yet can never define these terms properly. Press them harder, and they'll say the public's aesthetic sense is just too low, "do you like watching cats meow or something," and so on.

But I just like listening to bells, and I like Richard Clayderman, and I love putting some Beatles albums on infinite repeat. And when I'm working, if anyone dares to assault my mood with Yoko Ono's voice, I'll just grab a stick — either the song or their skull is getting shut off. Perfectly reasonable, right?

And if this crowd of "professionals" unfortunately becomes the contemporary "mainstream," then I can only sigh: the profession of the artist is dead. All we can do is wait for a force from another dimension, in the near future, to sweep away these minds — not necessarily old in age, but already rotten with age.