This was first written in 2018; the release of Dall.e 2 inspired its revival.
Art exists for two entities — the Creator and the Recipient. For the former, it is the tangible result of various internal mental processes combining with an intent to convey them, coloured by emotion and experience. It is the physical manifestation of the contents of the artist’s cognition. For the latter, it is a medium of connecting with a truth. It is the ultimate cathartic experience, meant to fill the Recipient with the sentiment felt at the time of creation by simply exploring the piece. It is not bound by time, context, or a shared meaning of social structure or understanding of the world — it is simply felt, at some point, by some Observer who has voluntarily or involuntarily opened their mind and become Recipient.
Art need not be scientific or formulaic — it is feeling and emotion driven. Yet it is not without technique — an artist must conform to a certain level of shared meaning with the world, adapting forms, structures and shapes that create representations that can be understood by a wider audience. Yet, there is no constraint, limitation or universal law describing the extent to which the artist must conform to these frameworks. Unlike in science, math, or anything logical, art is individualistic and guided by the Artist’s free will. It is perhaps the freest form of expression humankind possesses — not limited by language, logic or medium.
While art may be without purpose, it is with intent. It is with the artist’s intent of conveying the barest form of truth; the present state of the soul; a portrayal of a version of his existence at the time of creation. Therefore, it is not simply the consequences of an individual’s actions at a time of a given mental state that constitute art — it is her intent to transform those into a vehicle of communication.
Machine learning refers to the idea of feeding computers with huge masses of data, and equipping them with algorithms to be able to sort, filter through and create representations of the information. Advanced image-generation software, then, once fed with enough pictures, paintings, or pieces of art, should be able to churn out its own version of art, using rules, finding patterns, extracting properties and making generalisations. But to what purpose?
Some could argue that the phenomenon of Art has wide-ranging uses and implications in the industrial world. It can be combined with logic, language, consumer-behaviour statistics and an understanding of the human mind to create successful advertisements and commercials. Art can be used with space to create sculptures, civic spaces and public areas. Art could be combined with faith, resulting in purposeful architectural wonders — chapels, cathedrals, mosques and temples.
A consequence of Art is that it can be used to make money, and it is not uncommon for economically and politically motivated parties to capitalize on a fleeting moment of human vulnerability and emotion to sell an experience. Thus was created the science of design.
Perhaps AI systems can churn out effective, captivating designs. Perhaps companies will no longer have to hire designers to create websites, posters and advertisements. Perhaps governments will no longer require creative personnel to craft political campaigns. After being fed with various paintings, styles, descriptions of sentiments and an understanding of space, AIs can use human-supplied tools, frameworks and information to create representations. But as they get smarter, richer, more creative over time, the human intervention and influence required to program, maintain, and monitor the robot will decrease. And one day, the robot will become smart enough to maintain itself. And some days later, it will program another machine — several other machines — to churn out designs as well.
The claim that robots are capable of design and not Art is idealistic — a comforting thought to hold on to as we contemplate the transformation of one of our most authentic forms of communication. Lamus is a computer that composes and produces music. A study conducted by researchers at the University of Malaga asked a group of 250 participants, including professional musicians, to distinguish between human-produced music and music composed by Lamus.
“The computer piece raises the same feelings and emotions as the human one, and participants can’t distinguish them”, researchers said.
AARON, an autonomous art-making program, has already sold paintings to the SF MOMA, the Tate Gallery, and the Brooklyn Museum. It paints not with pixels, but with paint on an actual canvas. Benjamin Grosser’s Interactive Robotic Painting Machine uses background audio, sounds of conversation and its own ‘internal mechanisms’ to create paintings. So AIs appear to be fulfilling the criteria of ‘evoking sentiment and feeling in the Recipients’ to be classified as Art. But do they do so with intent?
The current definition of intention indicates that it involves planning & forethought, which sound suspiciously like elements of cognition. Therefore, this suggests a machine must have the ability to think in order to create Art.
Can a machine think?
Humanity is not equipped to answer that question - yet. We don’t fully understand how the human mind works, so we don’t know exactly what ‘thinking’ entails. We may, through trial and error, and large datasets, be able to create robots that replicate and imitate human behaviour, a result of ‘thinking’. But if we can’t operationalize or define the term ‘thinking’, we can’t devise a test to measure it.
Turing said that the question ‘Can a machine think’ was too broad and “too meaningless to warrant discussion” - his Imitation Game instead measured how well machines could imitate humans. Any machine that could fool a judge into believing that they were communicating with a human would pass this ‘Turing Test’. But, Turing noted, simply demonstrating intelligence need not imply intelligence. Behaviourist measures are not sufficient indicators of cognitive activity.
Sidebar: It is worth noting here that conceptualising a measure for machine cognition is particularly challenging, since we cannot (yet) fathom its bounds, form, nature, or internal function. Technology designed to measure cognitive-activity in carbon-based humans (eg: EEG, MRI, etc.) is likely not transferrable to silicon-based machines. Many deny that intelligence can emerge from non-carbon-based entities at all — but who’s to say that all intelligences must be made of the same ‘stuff’? Species chauvinism seems to be the prevalent sentiment among humans. My personal belief is that the secret to understanding the blueprint of consciousness lies in the evolution of octopi.
What if we somehow invent a test to measure a machine’s intelligence, and find it to be cognitively analogous to that of a human, thus capable of expressing intent?
Then, perhaps, we can with certainty say that these ‘generally intelligent’ entities can produce Art. But it’s worth pondering — if they are creating Art, to whom does it belong — the machine, or the artist that created its inputs? Will these new entities also learn to cultivate emotion-driven intent, or will their art simply reflect the biases of their programmer? Will the emotions that are evoked from machine-generated art be those that their programmer intended to convey?
Or maybe we’re looking at this all wrong, too scientifically, too technically. Maybe the machine is the Artwork — a vehicle of communication, an extra layer of complexity added to convey the creator’s truth.
The earliest Homo sapiens drew on the walls of caves 42,000 years ago. Later, our ancestors discovered paper, canvas, sculpting materials and tools. Years later, graphic design was born. Evolution shows a pattern of discovery, innovation, and an ability of the human race to adapt and to adopt these new paradigms, changing the way we convey meaning and interact with the world at each stage of history.
Maybe embracing this new medium is simply humanity’s way of adapting to make the timeless, fluid concept of Art more relevant in tomorrow’s world.
Such an incredibly deep and nuanced take Avantika . A joy to read
The only fundamental difference between human & AI art would be that the former arises directly from consciousness whereas the latter is drawn from an algorithm . And every truth an artist wishes to express, is essentially a reflection of the stillness from which all inspiration arises
I echo your beautifully articulated views, Avantika...
Art is nothing but non-verbal expressions of Universe in the language of Images, Patterns and Vibration which we feel everywhere in Nature ...from Fractals to Fibonacci...In tunes and tones !
When we are spiralling up on evolution vortex from carbonbased to siliconbased entities, only the conduit has shifted from Subconcious Mind to AI!
Manifestation of Universe will go on evolving thru Art Form ...eternally!