Should AI be allowed to create Art?

Original Source Here

An AI generated landscape — Datascientist55, CC BY-SA 4.0, via Wikimedia Commons

Should AI be allowed to create Art?

Who has the right to make ‘art’?

The questions around AI and art are not new ones. For decades now artists have been using AI in various forms to create art, but it’s only in the last few decades that we’ve begun to see what many of us consider to be ‘AI art’ — namely art made almost entirely by AI, with as little human interaction as possible.

But should AI even be making art, and how will the art world react to what may very soon be the first AI maestro?

Creativity and Art

Photo by Lucas Benjamin on Unsplash

What does it mean to be creative? To ask this question is really to ask what it means to be human in the first place. For millennia people have tried to answer this question and I make no claims to have done so here, but I do think it’s important to distinguish between ‘creativity’ and ‘apparent creativity’.

There is a common adage that if you sat a group of monkeys down in a room with typewriters and had them randomly hit keys, they would eventually type out the entirety of Shakespeare’s Othello completely at random. Given enough time, it’s almost an inevitability. But this is not creation, because it is not intentional.

The monkeys don’t understand what they’re doing, and their ‘creation’ is only possible because of certain parameters we set them at the outset. Many have argued that AI works on the same premise. It can display ‘apparent creativity’, but not actual creativity, because it cannot act beyond the parameters its programmers set it. Put simply, it’s not enough to make a beautiful painting or piece of music, there must be intent behind it, otherwise it is at best mimicry.

This is the core argument against calling what AI’s create ‘art’. If they’re not acting with intent beyond their creators, then are they really acting at all?

A Digital Da Vinci

AI generated painting of a dog — Mark Breadon, CC0, via Wikimedia Commons

Many people who consider AI the future of technology react negatively to arguments like the one above. They appear reductionist and arrogant, an attempt for artists and creators to gatekeep their domain. Some have even identified human ego as the only thing stopping us from accepting AIs as artists. They say that once AI exceeds human intelligence — the singularity — they will far surpass us in all fields, including creativity.

There are many counters to this though. Many see AI as useful tools in creation for artists, but not artists in and of themselves. A common analogy is that of musical instruments. Nobody credits Beethoven’s piano with the creation of his Symphony №5, or Paganini’s skill to his violin. Nor do we ascribe Hemmingway’s prowess with words to his typewriter or Einstein’s theorems to the chalk he used.

AI’s have far more in common with tools than they do with humans. To create a piece of music or to write a novel or devise an elegant mathematical proof is something truly unique to humans. AI has demonstrated that it can competently analyse a database of images and deliver an ‘aesthetic amalgamation’ in return. But is this creativity or the compression of a database of images into a picture? When faced with questions of why it chose to highlight certain features, colours and postures, the AI would shrug its virtual shoulders and respond simply, ‘because that is what was inputted’. Intentionality is lacking and therefore so is creativity.

This logic can even be extended into the world of mathematical proofs, a realm where AI proponents have traditionally speculated AI will excel. Interestingly however, it has been remarked the AI may neither surpass us, nor prove to be that useful here. If the AI can develop proofs that we ourselves can find, it is hardly beyond us, and if the proofs it discovers are so far beyond us as to be incomprehensible, then are they even proofs at all? Mathematics must be interpretable after all and an answer to a question or a theory that can’t be understood by those that ask it isn’t really an answer.

Proponents of AI point to the fact that what an AI makes can be indistinguishable and even better than what a human can make and granted this is sometimes true. It’s also true that when people don’t know a piece is by an AI, they react more favourably that when they do know.

However, therein lies the issue. When viewing art, reading a novel or watching a film, what we are really doing is engaging with the creator in an experience. True, a piece of art by an AI may be beautiful, and AI have even managed to effectively replicate human achievement, with one famously analysing Bach’s music and ‘mastering’ it to the point that humans can no longer distinguish between him and the machine, but this is inherently creative or is it just mimicry? What does this often deceptive imitation do to viewer/creator relationship?

People react negatively to mimicry because the expectations and weight they placed on a piece are tied irreparably to its creator. So, like believing that you have just purchased an original DaVinci only to discover that it is a copy, it should be no surprise that people will often turn hostile. The relationship between viewers and author has been irredeemably damaged.

The Future

Edmond de Belamy, Public Domain, via Wikimedia Commons

All of this points to the fact that the true function of AI may have begun to be overlooked in favour of more wild assertions. AI is fantastic at doing the jobs we don’t want to do, and for the many companies developing the technology, this is exactly the direction they go in. Adobe are particularly prominent in ‘artist AI features’, but many other companies are developing similar features such as finding a specific frame in editing, smart cropping, image tagging, automated audio mixing, and auto colouring frames. This could drastically reduce work times and allow artists to focus more on creating rather than drudgery. AI in artist software may yet become another tool that allows for the next artistic revolution.

It’s been speculated that AI generates between $3.5 and $5.8 billion across all the sectors it’s employed in, and it’s true that some AI ‘art’ has even fetched a high price at some of the best auction houses in the world like Christies in New York, where a piece titled ‘Edmond de Belamy’ sold for $432,500. But is this really a good use of AI? A tool that can drastically cut down on repetitive manual tasks, that has the potential to truly revolutionise huge numbers of industries in as yet unknown ways, is being being used for what appear to be little more than marketing opportunities or thinly veiled ego projects.

AI could prove to be one of the most useful tools ever devised by us, yet we seem insistent on setting it the unenviable task of imitating something as vague and unique as art, something that we ourselves have yet to come up with an adequate definition of. We are, in effect, asking AI to do the impossible — communicate the human experience.


Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: