You are on page 1of 8

McGowan 1

Robert McGowan

Artificial Intelligence and Creativity

Whether it is a grand symphony or a child’s caricature of the world around them, the

pursuit of art is perhaps the most uniquely human endeavor. The nature of our intellect and its

progression from our biological ancestors has endowed us with this mysterious capability of

representing both physical and nonphysical ideas in a purely abstract method. Being so unique to

the human intellect, the question arises whether or not future artificial intelligence (AI) would

one day be able to create a form of art comparable to the kind which we admire and appreciate.

This question and questions like these are at the heart of the philosophical perspective of AI and

its potential manifestation. To properly dissect the question of whether a form of AI could create

true art, we need to investigate the nature of modern models of consciousness and how

accurately this comes to replicating our standards of human consciousness. The next issue to

tackle regards the existential nature of aesthetics as it relates to humans and whether an AI will

ever have the same “motivations” to create art as we do. The consequence of this question bleeds

into the question of supposed subjective experience from the AI’s perspective and weather AI

will ever hold art to the same standards as we do; or have a purpose for it at all.

Today’s research in artificial intelligence is closely linked to the enlightenment that

cognitive science as enriched the field with. The pursuit of AI research has been historically

predicated on the computational model of cognition that both cognitive and computer science

overwhelmingly subscribe to. This “classical” view asserts an algorithmic method in which the

mind processes information. With cognition being between an input–output relationship,

artificial intelligence aims to isolate this component in order to replicate human mental

processing. With theories like the Language of Thought Hypothesis and Massive modularity
McGowan 2

which attempts to explain this middle- ground central processing unit, there still exists the

difficult hurdle of qualia and vivid representation as experienced by humans. One possibility for

the stagnation that both cognitive and computer science has experienced in attempting to solve

this riddle might stem from the misdirection caused by the classical model of cognition. Radical

new ideas of embodied cognition aim to place an emphasis on the existential nature of

consciousness as it relates to the world around us (Prinz, 2009). It turns out that if representations

are truly embodied, and that these representations manifest into higher level thinking in the form

of artistic pursuits, that the classical pursuit of artificial intelligence is neglecting a fundamental

component in how ideas are manifested in cognition; that being the extension of a body and

spatial-temporal realm to project these ideas into.

Despite the difficulties surrounding the approach towards AI and the hard problem of

consciousness, there is still much effort being placed into the research and application of

machine created art. Magenta is the project of the Google Brain team which utilizes deep neural

network models to analyze existing music in an effort to generate new music (Eck, 2016).

Though there have been computation assignments which aims to analyze and interpret human

created input (such as speech and handwriting recognition software), the application of creating

new works from this learned knowledge is a relatively new field. Magenta aims to be the

ultimate Turing Test winner in how closely it could come to creating works that are potentially

indistinguishable from that of human created works. Magenta is not even the only initiative

which aims to replicate human creativity. Emily Howell is a computer program created by

composer and computer scientist David Cope. This program comes closer to replicating the

compositional art by having preprogrammed “rules” of part writing which allows the program to

create works that could be considered more “original” (Adams, 2010).


McGowan 3

There is both a short term and long term goal to projects like Magenta and Emily Howell.

The short term being to create machines that could simply learn and output replicated samples of

art. In the realm of music specifically, this prospect has been very much achieved as displayed by

the works created by these machines thus far. It is the long term aim of initiatives like these

which poses the deepest philosophical hurdle for researchers to achieve. That being whether

these machines could create truly “generative” works of art. That is creating something out of

seemingly nothing. The obstacle here is that these machines require some human intervention (in

the form of pre written code or input examples) to fully create something “new”.

For as long as there has been formal art, technology has been there to aid in the creative

process. It is believed that the earliest cave paintings used pigmented water to create stenciled

outlines of a hand or a stick figured deer (Battersby, 2014). Even today, electronic music and

digital art comes as close to computer generated art yet still requires a human to dictate the

parameters. We rightly accredit the human artist and not the tool as the sole creator of the pieces.

It could be thought that these modern composition machines are essentially similar to the

computer aided art that we’re all familiar with; though a step removed. This sort of separation

from the original human source could be multiplied indefinitely (much like a technological

singularity), yet the perennial human source will still be credited as the creator of whatever final

output there is.

Though there is debate as to whether any art is truly original and not merely derivative

from some previous work, we will assume that human created art has a component that

fundamentally sets it apart from the works that have been created by machines thus far. One

critique that is frequently brought up opposing the works created by machines is that they are not

“true” works of art. From a human perspective, art is fundamentally created as a representation
McGowan 4

of something which exists independent of the represented art. Even abstract art is a form of

representation of the idea of abstraction itself. When applying this standard to artificially created

art it is difficult to appreciate because of its artificial nature. From a Theory of Mind perspective,

we are apprehensive to assign human like qualities to these machines because there is essentially

“nothing it is like to be” these modern machines.

Even if a modern computer could create a painting comparable to Monet’s Water Lillie

Pond or a novel comparable to the works of Dostoyevsky, there is still the lack of that human

component that would plague any machine creation. After all, one component of art that we

value is the intellectual effort that went into the actual creation. This is the foundation for our

knee jerk reaction to aesthetic beauty. From an evolutionary standpoint, this is even true amongst

different species which value the beautiful colors of a peacock or the melodies of a song bird.

Though it would be easy enough to convince a songbird to accept an artificial companion

through processed melodies, it is far more difficult to convince a human to accept the same

sentiment. This goes back to the difficulty of subjective experience on the part of the AI. It

seems to be that creativity is a definite necessity when it comes to human level consciousness.

This necessity is so engrained in us that creative people are considered my many more attractive

regardless of their looks (Khazan, 2017). Nevertheless, the uncanny valley of AI created art

stems from the fact that it seems to be devoid of any subjective purpose. We value works of art

because of how it portrays and displays the human condition. These representations were created

out of an understanding of what pain and joy feels like and ever emotional experience within the

spectrum. This is obviously what AI is lacking as even the most advanced neural networking

software is simply playing pretend with the existing creativity of man.


McGowan 5

This raises the question of what is creativity in the first place. The researchers at magenta

claim that they want to create a generative creating machine, but to what extent does this

generative originality come from? As mentioned above, there is a mysterious component to

human creativity, but this human creativity might be just a romanticized form of free will. There

are those who subscribe to a deterministic universe, where free will could not exist. If this is true,

then creativity is simply a vacuous name given to an incredibly complex yet theoretically

predictable outcome.

Though there are those in the quantum science field that stress the impossibility of such

theoretical predictions through fundamentally random force in the quantum space, could this

random probability extend to human creativity and free will in general (Jha, 2013)? If so, there

are machine algorithms, simple ones at that, which could output truly random outcomes. If these

random algorithms could be contained to artistic constraints, then the outcome would

theoretically be “original” in the human since.

Giving AI the benefit of the doubt that it might one day be able to interpret, represent and

execute truly creative works of art, there still looms the question of whether the output product is

considered a work of art to the machine itself. The criteria and hurdles given to such machines

have thus far been defined in human terms and appeals to our senses but these are still

fundamentally human standards. Supposing that artificial general intelligence will one day

become reality, there is no reason to suppose that the human intellect is the pinnacle of

intelligence within a functionalist perspective. One of the main aspects which separates us from

every other living organism is the sheer size of our brains. Though brain size amongst humans

does not necessarily equate to higher intelligence, there is still a clear correlation between the

brain to body size amongst species (Wynn, 2013). If an AI agent were to actually have its
McGowan 6

consciousness powered through a silicon based mechanism, there is no limit to the size this

“brain” could amount to. This in turn could allow machines to move beyond “artificial”

intelligence and simply into the realm of super intelligence in every sense of the word.

As discussed earlier, it is my belief that creativity is a clear component and byproduct of

higher intellect. If we would once again grant the possibility of super intelligence amongst future

AI, the question then arises; to what standard would these future AI’s hold art to? It would thus

be difficult to speculate what if any sort of cognition super intelligent AI agents could possibly

possess but it is safe to assume that such intelligence would reach far beyond the rational

understanding of modern humans. There are even some futurist philosophers who believe that AI

could be the next evolutionary step of mankind (Spinks, 2014). If this is the case, it is safe to

assume that the super intelligent standard of cognition as it relates to creativity would be

incomprehensible to what we could ever imagine. Perhaps these super agents would find

aesthetic pleasure in manipulating the laws of physics or in some other display of power.

The fact that such questions are being posed today indicates that our age is thus far

defined by unprecedented technological advancements and its confliction with moral and ethical

philosophy. The question of creativity is but one component of the deeply challenging questions

which AI poses to our understanding of cognition, morality and humanity in general. Though the

answers to these questions might seem frightening, our human curiosity and sense of exploration

will nevertheless drive us towards solving these deeply complex issues.


McGowan 7

References

Adams, T. (2010, July 10). David Cope: 'You pushed the button and out came hundreds and

thousands of sonatas' Retrieved April 29, 2017, from

https://www.theguardian.com/technology/2010/jul/11/david-cope-computer-composer

Battersby, M. (2014, October 09). 40,000-year-old cave paintings include 'oldest hand stencil

known to science' Retrieved April 29, 2017, from http://www.independent.co.uk/arts-

entertainment/art/news/40000-year-old-cave-paintings-include-oldest-hand-stencil-

known-to-science-9783840.html

Eck , D. (2016, June 01). Welcome to Magenta! Retrieved April 29, 2017, from

https://magenta.tensorflow.org/welcome-to-magenta

Jha, A. (2013, November 10). What is Heisenberg's Uncertainty Principle? Retrieved April 29,

2017, from https://www.theguardian.com/science/2013/nov/10/what-is-heisenbergs-

uncertainty-principle

Khazan, O. (2017, April 19). Creativity Makes You Seem More Attractive. Retrieved April 29,

2017, from https://www.theatlantic.com/science/archive/2017/04/creativity-boosts-

attractiveness/523332/

Prinz, Jesse (2009). Is consciousness embodied. In Murat Aydede & P. Robbins (eds.), The

Cambridge Handbook of Situated Cognition. Cambridge: Cambridge University Press.

pp. 419—437
McGowan 8

Spinks, P. (2014, July 16). Artificial intelligence: the next step in evolution? Retrieved April 29,

2017, from http://www.theage.com.au/national/education/artificial-intelligence-the-next-

step-in-evolution-20140717-zu5pt.html

Wynn, T. (2013, September 21). Does Brain Size Matter? Retrieved April 29, 2017, from

https://www.psychologytoday.com/blog/how-think-neandertal/201309/does-brain-size-

matter

You might also like