Will Artificial Intelligence ever create Artificial Art?

(This is a post dedicated to my buddy Scott Cooper who routinely provokes these discussions. I’ll take the blame for flying off on preposterous and overkilled tangents….)

The last few posts have explored topics that (at a stretch) can be seen to relate to this question. That’s why I’m here. I’m your fastidious dot connector, your outrageous recycler of mashed up information, your daydreaming spinner of delusion and illusion. What can I say. Its an art….

So…, a few posts ago I raised the question of whether art making was simply a matter of object generation. Chuck Close suggested that the showing up, nose to the grindstone, was more important than anything as ephemeral, inconsistent, and unreliable as inspiration. Machines don’t need inspiration, thunderbolts from out of the blue. There will be no artificial muse for artificial creative intelligence. It will be a matter of routines and subroutines whether a machine will ever be responsible for art in our world. And if all we do as humans to create our art is to churn out product, who’s to say whether we may one day be replaced by machine artists? Industrial production has replaced handcrafting to a large extent, so why not an industrial art? It seems chauvinistic to claim that things are only art due to the touch of a human hand. And didn’t Duchamp dispel that idea when he introduced his ‘ready mades’? And so, if the human hand is not logically necessary for the production of art it must be something else. But what?

That’s the question, isn’t it? Is art really only that object that is left in the world after the producing is done? Is that the art? My next post asked whether there was more to being an artist than producing its objects, more to arting than the ‘art’. The question was whether being an artists was part of living a human life, how we live, whether living itself was a project for art, and whether any particular product or outcome was necessitated in the art process. Is the message in the medium, or is the message in the intention behind the doing?

And obviously if we look in this direction the fit between art and machines starts to unravel. Machines may be a part of an art undertaking, a tool to be used in the process, a means to creating art objects as ends, but they may not square up so well as the inherent source of artness. Its execution, yes, but perhaps not its inspiration. Art seems to need a reason….

But maybe we aren’t looking at the work involved in making things correctly if we invoke some godlike spiritual act in the performance of creation. Conscious intention. A human brain is at least something like an organic computational device. Any inspiration or consciousness we have has to originate somewhere within this mass of jelly. Isn’t the question of AI whether we can simulate a meat mind with a digital one? But setting aside the hardware issue (which as far as consciousness goes seems unknowable) the software seems a bit more approachable. AI is interested in the behavior of machines. Making art seems to be a particular kind of intention. So, if modern day machines are defined by their ability to follow rules, to make this or that decision given the right circumstances or input, does this answer the question of their intentionality?

Following rules does seem to denote an intention. But art isn’t always about providing the right answers. The intention of literally following the evident rules is not the intention of art. Often its as much about what we leave out. Showing rather than telling. The art is sometimes in knowing what not to say. Leaving room for interpretation. Reading between the lines. Knowing how to unlock an audience’s imagination. Can machines do that? Can simple rule following know the difference between enough and too much? If art lies in this sometimes gray area outside the mere rule following can machines intend to make art. Or do they simply do what they are told?

Which seems like an important distinction. The next post I put up was an exploration of the nature of art as something which on the face of it seems uniquely human: The purposeful and intentional breaking of rules. We think of art as not merely reproducing conventions, regurgitating statistical probabilities, but rather that active transgression of categories and the defiance of expectation. We can have intentions that are parallel to the rules, consistent with them, independent of them, or in stark negation of them. To make art a human artist takes a stand against the pressures of conformity. Its not the lowest common denominator but an egalitarian confusion of idiosyncrasy and individuality. Its not a simple case of input A equals output B. New art by definition is something of a revolution against the old. The thread of artistic change passes through the needle’s eye of mystifying, inscrutable, and even ‘bad’ art in order to eventually weave a tapestry of new appreciation. Even the best of art often starts out as something reviled.

So one question would be, if machines were to innovate artistically, would they be producing variations on a programmable theme, or would they be routinely blowing our minds? Would they be trapped in statistical conventions or would they be the mad untethered geniuses of DaVinci, Van Gogh, and Picasso? It seems one thing for a spam bot to carry on a limited conversation with live humans, a computer to play chess, but something wholly different to stand on the cutting edge of human experience, to give new voice to beauty, to move the human race forward. Can machines ever do that? Can they aim for that? Human intelligence can design and program machines to do the stupendous, the gravity defying, the earth shattering, but will an artificial intelligence ever do this of its own volition? That’s the question, isn’t it….

(Here’s an interesting discussion of some of these topics. Well worth a read.)

It seems that the issue also comes down to what role art plays in a human life. And obviously that means one thing for the artist plumbing the depths of their imagination, their creative ‘soul’, and something quite different for an audience accepting the fruits of this labor. In the sense that an artificial intelligence might produce ‘objects an audience classifies as art’ that seems not impossible, but in the sense that art is also an expression, perhaps not so much. This is a question not of how well it can fool the public. This is the question of the motivations for artmaking.

If new art is new expression, then it has to also mean something, and meaning and understanding are not simply reducible to having appropriate answers. Meaning exists only within a web of life. Which isn’t static. Whose growth comes in fits and starts, and often with a retracing of steps and intuitive leaps of non rational proportions. The bots that spam us are trying to give us the right answers, that’s their job, and they play by the rules of certain statistical expectations. But do they ‘understand’ what they are doing? Understanding means having reasons to make a choice. “Yes I will”, or “No I won’t”. Can a machine have an intention to spontaneously NOT play by the rules, to invent its own new path, to skip the rails on a whim? Can an artificial intelligence ever be whimsical? Just how should we make sense of that question? (And keep that in mind as you read further.)

This problem is directly related to the question of art and innovation. If art often breaks new ground then it only does this by upsetting routines. And this is what we call creativity. Its a new logic in contrast to the old. Pseudo logic and irrational reasons. So in some sense all art that sets out in new directions is a kind of ‘mistake’ when viewed in the context of art’s past. Not mistake as in the failure of intentions, but in purposely stepping on the toes of well schooled lines.

And machines can make mistakes. But human mistakes often have the flavor of an advantage. What is the difference between mistakes that go somewhere and those that dead end? Ones that simply abort a logic and those that create something entirely different? New meaning. A new routine. A new reason for doing things a new way. What would the difference be between simply making a mistake in the program/routine and engaging in something wholly unique and self sustaining? At what point does error become poetry?

A human mistake might be adding 2+ 2 and getting 6, or calling something “red” that was green, or calling it something entirely unrelated to our color categories, nonsense like indicating green by calling it “stubborn”, “cooking” or “uymdfy”. And machine errors are often like that. We can sometimes retrain to get the ‘correct’ responses. But there are also error differences that are not always literal rule breaking and which rather than needing to be corrected are a new form of response. And it is in this way that art steps forward. Some transgressions are more like rule bending, e.g. “My love is a flower….” We can learn to make sense of that. We can sing a song in a new key. We can change the words and ad lib. We can improvise. Randomly associate. Use metaphor. Tell jokes. Blow up in anger. The art is in knowing when to set the ‘normal’ rules aside.

In other words, we can be irrational for good reasons. There is simply a whole tonne of nuance to humans not following the literal or conventional rules. Can we say the same of mechanical breakdowns? A human fail is sometimes still a win. If a machine intelligence is based only on following the rules it seems there is only one kind of mistake possible. And that would not lead to art….. Art consists in turning ‘mistakes’ into gold.

But who am I to predict what may or may not be possible in future artificial intelligence? All I can do is point out what it would have to include to better mimic human intelligence. Understanding how a machine might ‘think’ means first understanding what we mean by ‘thinking’. Creative thinking in the sense I have discussed it is a seemingly unique human capacity. And at this point the thinking that machines do is as far from that art as to be unrecognizable….

But what if machine intelligence of that human order were possible? We can imagine many things, so why not that? What if machines could be artists in the sense that humans are? And the question is, what would that art look like? Would it be comprehensible as ‘human art’? Would it be breaking rules only to fit back in the simian cultural paradigm? Would primate chauvinism inform this new intelligence? Would a machine intelligence still be dependent on its human roots? Or would a machine intelligence not also have to have a machine oriented comprehension? Not a human one? And as different as machine life is from human life, would it come to the same conclusions about the world as we do? Why on earth would it? And if art reflects a way of thinking, would a machine art even be something that humans could understand? From the machine’s point of view? Or would it be like humans guessing at the life of insects and gods?

If human art is essentially a breaking of human rules and going off the beaten human path, just what would that endeavor look like to an artificial intelligence? Clearly we would not be expecting the next Vincent van Gogh. He was all too human in his failings and his transcendence. How might a machine fail and transcend? Not facing a human life it would have to be different. Right? Would we even recognize its performances as art? Its an interesting question that has no answer besides the weak human one. It was after all a human question….

This is a 1997 commercial from Steve Jobs and apple inc, who know a thing or two about creative intelligence:

“Here’s to the crazy ones.
The misfits. The rebels. The troublemakers.
The round pegs in the square holes.
The ones who see things differently.
They’re not fond of rules.
And they have no respect for the status quo.

You can praise them, disagree with them, quote them,
disbelieve them, glorify or vilify them.
About the only thing you can’t do is ignore them.
Because they change things.
They invent. They imagine.
They heal. They explore. They create. They inspire.
They push the human race forward.
Maybe they have to be crazy.

How else can you stare at an empty canvas and see a work of art?
Or sit in silence and hear a song that’s never been written?
Or gaze at a red planet and see a laboratory on wheels?
While some may see them as the crazy ones, we see genius.

Because the people who are crazy enough to think they can change the world,
are the ones who do…”

About Carter Gillies

I am an active potter and sometime pottery instructor who is fascinated by the philosophical side of making pots, teaching these skills, and issues of the artistic life in general. I seem to have a lot to say on this blog, but I don't insist that I'm right. I'm always trying to figure stuff out, and part of that involves admitting that I am almost always wrong in important ways. If you are up for it, please help me out by steering my thoughts in new and interesting directions. I always appreciate the challenge of learning what other people think.
This entry was posted in Art, Beauty, Creativity, metacognition, Wittgenstein. Bookmark the permalink.

2 Responses to Will Artificial Intelligence ever create Artificial Art?

  1. If you remember the film Bladerunner you get one take on this question. Its based on the book “Do androids dream of electric sheep?” by Phillip K. Dick. The film suggests a world where machine AIs have been perfected enough to pass almost perfectly as humans. In the film these AIs can perform almost indistinguishably from their biological counterparts. The only difference is that these machines don’t feel empathy. You can spot an AI by trying to provoke its empathy, which either comes out forced, too slow, or wrong.

    This idea revolves around the issue of belonging. If an AI was manufactured to perform as a human in all respects but still thought of itself as something other it might not find the resources for empathy with human beings. And you can see that while much of what we take to indicate human life is tied up in all the normal behavior that goes on with us, the question of empathy is a hard one to get around. I cry out in pain, and how do you respond? In some sense our empathy helps define us. A person without empathy, a sociopath, might even be considered less human than one of these androids in this respect.

    But maybe the question breaks down even here. The book/movie also introduces the character of the girl who doesn’t know that she is an AI. All her memories of a childhood are false but she doesn’t know that. So the question is, would an AI so perfect in all respects be said NOT to understand language or think simply because it was a constructed being? Doesn’t the issue tend to blur once we consider an intelligence that is functionally identical? If androids dream of real sheep, and do all other human things like falling in love, what basis would we have of excommunicating them from the realm of our brothers and sisters? Don’t they exhibit enough humanity to be considered human?

    That they don’t grow old, or start out with full awareness is only held against them with prejudice. An infant is human without resembling an adult in almost all the social categories. Isn’t what we call human more a convention of grammar than anything else? A biological discrimination? We have good reasons to grant an infant its humanity despite its drooling incomprehension of the world. Would there not be different but equally good reasons to grant an AI that status given the criteria laid out in the book?

    Thinking and understanding are not metaphysical operations. They describe what you can do. And if you can do everything required, why on earth would it not be considered those things? There is no special mental state that defines thinking. Take a look at Wittgenstein’s Philosophical Investigations if you have any doubts. And so, a machine intelligence that thought like a human might very well make human art. A machine that thought like a machine would obviously not.

    So the question is “Why do humans make art?” and rather than dismissing this, perhaps our society would be better served if it respected this human impulse more. Making art is what humans do. And its not just what other humans do for our enjoyment and entertainment. Its something we all did as kids while we were busy exploring the unknown of our formative experience of the Universe. A human life is not all obeying the rules, but also an exploring of our limits. Doesn’t that deserve to be honored?

    And while we may not understand the adventures of other creators it does describe a human possibility. These things are possible for human understanding. In fact, other humans are far more alien than the AIs in that book. Alienness seems to depend more on culture than on biology (or the lack there of). And human culture is diverse. There is no one right way of doing it. So how can it not be important to continue exploring? How can it not be important to continue breaking rules? How can it not be important to make art?

    What do YOU think?

  2. “Its reminded me of the late writer and woodworker David Pye, and he said ‘Yeah, people who make things with their hands, its the craftsmanship of risk, and we really don’t know when we make something with our hands what our hands are going to do.’ Because its like life itself where relationships are messy and there’s a degree of unpredictability about it.”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s