Image courtesy of Atilla Kefeli

Is beauty subjective or objective?

Image courtesy of Atilla Kefeli

Everyone wants to be beautiful. Yet, we are told, beauty is in the eye of the beholder. If that’s the case, can you crowdsource beauty? If, say, 9/10 people prefer thin people, is it possible to be beautiful and be fat? Only in the eyes of ten per cent of the population. A small number, but still a significant minority.

Perhaps you need to find a set of objective standards that everyone can agree on. Some people like tall people, some people like short people, some people like blonde people, et cetera — but perhaps we come closest to agreement in the following statement: there is beauty in symmetry.

From art and architecture to the human face, people feel more “comfortable” with symmetrical designs. It’s partly instinctive: we’re hard-wired to find people with good genetic material attractive, and symmetry is the most obvious sign of good, strong, healthy DNA. (Consider the reverse stereotype: the slack-jawed yokel of redneck myth is patently depicted as unsymmetrical and considered universally unattractive). But is it a true judge of beauty? I think not.

Anaface is a computer program which claims to do just this – judge your attractiveness by the symmetry of your face. It is of course wrong. You can’t judge beauty on symmetry. I happen to have a very symmetrical face. However, I’m fairly average. I scored almost twice as high on Anaface’s test as a friend of mine who’s a professional model. Symmetry, it seems, is no great judge.

In fact, perfectly symmetrical features are bland and characterless – I’d never cut it as a model because my face doesn’t have “character”. So what do you do? Do you reprogram the computer to understand that some asymmetrical faces are beautiful (because they have character) while others aren’t?

Computers can’t handle variance without record to mathematical equations. Computers are essentially binary – on or off, yes or no, all or nothing. Computers seek veracity — they’re problem solvers, literally. You could plug an equation for “variance based on asymmetry” into a computer but in doing so you have introduced an element of subjectivity rather than objectivity (based on one person’s personal perception, or on crowdsourcing, etc) into the formula, and computers alone will never be able to do this. A computer, in short, cannot make judgements about beauty.

And that is why computers will never be able to make meaningful art.

There is no objective truth in beauty. Charles Bronson, for example, was considered handsome – despite his unconventional, weatherbeaten looks. What do we tell the computer? Well, we can only tell it that some people find weatherbeaten faces attractive some of the time. The computer is confused. You feed it a picture of Charles Bronson and tell it that sort of face is attractive to some people, therefore the computer gives it a score of “above average”. Then you feed the computer a similar picture of someone, say, an old fisherman who’s been exposed to the open seas for two or three decades. Is he attractive? Based on the new information you’ve fed the computer, yes. The computer is incapable of making the same “judgment call” we as humans are.

It’s not simply a matter of updating the formula, either. Adding further exceptions, refining the formula to include fishermen, all other types of weatherbeaten / asymmetrical faces etc — the most the computer will ever be able to reply is that there is variance. ie, some people find this sort of face attractive, others don’t. And as for why some types of weatherbeaten / asymmetrical faces are attractive and others aren’t… how can the computer tell? A scar on one side of the face might be considered attractive by some people because it makes the person look dangerous. But others might be repulsed. Again, the most the computer can say is that there is variance. There is no “objective truth in beauty”.

Mathematics and art don’t match.

If you ask the computer what it thinks, it doesn’t have an answer except recourse to mathematics, ie “I have been shown a photo of a weatherbeaten face a little like Charles Bronson and there is a 50% chance that this face is attractive” — in order for the computer to say yes or no, you would have to program it to either like or dislike asymmetrical / weatherbeaten faces. A computer cannot simply decide if it likes a symmetrical face or a weatherbeaten one until you tell it.

Consider Shakespeare’s famous love sonnet, ‘my mistress eyes are nothing like the sun’ — in which he goes on to explain that the object of his affections is unappealing by conventional standards, but he loves her and finds her attractive. By objective standards he is wrong. But art, like beauty, is not held to objective standards.

This is why computers cannot be good artists. They have no perception of beauty. If you fed a computer every Rembrandt, Picasso, Poussin and Pollock, and told it to extrapolate from that what’s beautiful and create art based on that, you’d come up with a nightmare splattering of nothing.

If you tell an “art producing” computer to imitate one of those styles, it is merely a very clever photocopier. A computer decides on mathematical formula what is or isn’t beautiful. It can either say yes or no based on preprogrammed criteria, or it can tell you what the variance is, what the likelihood that certain types of people will find certain types of art appealing. This is not an adequate.

The creation of new art cannot be based on formulaic analysis of what people have found attractive in the past.

We cannot simply deconstruct what has gone before, evaluate its attractiveness, and create new art based on that.

This method might be a good way of creating, say, Ikea prints, but it’s not art. Art, like beauty, requires a subjective element. Let us say that a vandal throws a tin of black paint over the Mona Lisa. Let us then say that vandal is a currently respected but controversial artist, someone like Banksy or Damien Hirst. The art community is divided. 90% say it’s a travesty and our artist / vandal should be locked up. But 10% say that it’s a powerful statement about the nature of art, that it’s an act of iconoclasm and should be praised. Those 10% may have radical political ideas, for example — that make them differ from the other 90%. Of those 10%, 5% would leave Mona Lisa as she is, covered in black paint, hanging for all to see. The other 5% believe that, point having been made, the painting should be restored.

Feed this information into our computer busily churning out brilliantly made (from a technical perspective) oil paintings to sell at Ikea. It will continue to make 90% of its output as it was. But from now on, it has a difficult decision. Should it make 10% of its new canvases with a big black splodge of paint in the middle? Or should it make 5%? Or should it make none, because 10% of people only consider a black splodge on a painting to be art if that painting happens to be the Mona Lisa?

The answer is probably the latter. It’s very unlikely that anyone will find black splodges attractive except by reference to famous artworks. But some might. It may become a counter-cultural symbol for people with radical political ideas. The black splodge could be imbued with meaning. Having a black splodge on your paintings may become iconic as a Marxist symbol. It’s unlikely, but possible. And a computer cannot adjust to this new fact until it has again been programmed, until the formula for artwork has been updated. Again, a computer cannot lead, cannot make new ideas. It can only copy what people do, what the general consensus becomes, and follow.

Computers are problem solvers, not artists. Computers do not create, they copy. Computers cannot account for subjective beauty.

My example of the Mona Lisa was there for a reason. As was the mention of Banksy. One of Banksy’s oldest works, the “Mild West” mural in Bristol, was vandalised in just this way, with a tin of red paint streaked from side to side. It was done as a political statement, not as an act of vandalism, by a group that disagrees with graffiti as it is an inappropriate use of public space. The council restored the “artwork” — why? If it was an ordinary graffiti artist, the council would be painting over the mural itself. The “is it graffiti / is it art?” debate is entirely subjective. You cannot ask a computer what it thinks, because it cannot tell you until you have first told it. The most it can tell you is that there is variance.

So the computer churns out lots of different types of artwork based on this variance formula, based on what it thinks certain people will like. It doesn’t produce art, it is a photocopier. It is a thousand monkeys on a thousand typewriters, and it may eventually churn out something that some people find beautiful. But it isn’t an artist. It doesn’t have an artistic vision. It doesn’t imbue its artworks with meaning. The same is exactly true of a computer that generates poetry, novels or music. Art is imbued with meaning when an artist makes a subjective decision and applies that to his work. A computer cannot do that because it cannot be subjective, it either accounts for variance or makes a decision based on what is random, based on a flip of a coin.

Computers only follow, they never lead, and they never create anything that is new. For these reasons, computers will never replace the role of the artist / creative in anything but a technical sense. They can create a fine copy, but they can’t create fine art.

One Comment

  1. heythereistea says:

    truth strikes again