The Creative Mind: Artwork of Artificial Intelligence
creativity
/ˌkriːeɪˈtɪvɪti/
noun
1. the use of imagination or original ideas to create something; inventiveness.
Ex Machina (2015)
The idea of creative inspiration has had a variety of supposed origins throughout history. For the Greeks, the artist or poet was believed to travel into furor poeticus; a divine realm of ectasy beyond their own mind from which their creative instincts flow. In modern times, with the human understanding of the mind moving away from the divine and towards psychological theory, Sigmund Freud theorized the origin of creative inspiration as being the result of repressed unconscious trauma. The current dictionary definition of creativity defines it as being the 'use of imagination', but that creates a whole new question in itself. What does it actually mean to imagine?
Alex Garland's sci fi film Ex Machina is deeply concerned with what it means to be conscious. In the following scene, multibillionaire CEO tech genius Nathan explains the complexities of his new AI's mind using the artwork of Jackson Pollock. Pollock was a painter who 'let his mind go blank' and, using paint splatters, created his art seemingly unconsciously. In this scene, Kaleb, the character to the left, is one of Nathan's employees who has been invited into a research facility to interact with and test out Ava; a new artificially intelligent being that Nathan has been working on. Watch the clip below:
This idea of 'some place in between' is particularly relevant to this discussion. When we paint, dance, or sing, to what extent are we conscious of the choices that we make? We believe we have full control over creative actions, but, as explained by Nathan, it would be an impossible task to explain the motivation behind every single subtle component of any creative endeavour. We are in 'some place in between'. 'Not deliberate, not random'.
If a pianist performs a complex and immaculate musical performance, can they said to be creative? Or is it just a case of well rehearsed muscle memory? Or perhaps then improvisation can be said to be creative? Rather than simply performing a rehearsed, 'programmed' task, improvisation requires a multitude of on- the- spot decisions in order to craft a sound that the audience will recognise as sounding 'good' in accordance to our shared cultural rules of music. But that still begs the question of why? On what basis do we 'decide'?
Over the last few years, AI developers have become increasingly interested in the concept of creativity in relation to the artificial mind. A concept such as creativity, that is believed to be so innately human, provides perhaps the ultimate simulation test of human behaviour. If a machine can 'imagine' of its own accord, can it be said to be conscious? The following clip shows the invention of French roboticist and ex-painter Patrick Tresset. He has created a classroom of 'artistic robots' which not only communicate with each other through morse code, but can draw your portrait completely unassisted. Watch below.
Whilst these robots are eerily precise in the way that they seem to 'watch' their artistic subject whilst putting pen to paper, the extent to which they can be said to be independently creative is fairly limited. Through use of cameras, visual information is transcribed into physical movement via a set of programmed aesthetic rules implemented by Tresset's programming. They are acting automatically, but through a human creativity presets.
The following piece of art takes these ideas a bit further. This image has been generated by an algorithm with no human guidance. There are no cameras to offer direct inspiration, but rather this image was generated purely based on internal algorithms; from the 'mind' if you like.
The most fascinating part of this image is the sense that it 'looks like something'. Almost every aspect of this image appears to be something we recognise, and yet when closer attention is paid, it is impossible to pin down just exactly what we are looking at. So what causes this strange 'familiar yet unfamiliar' spectacle?
To create AI artwork, artists create algorithms that go further than just following a set of rules, but actually “learn” different aesthetics and conventions by analysing thousands of existing paintings. By consulting the aesthetic rules that it has picked up through its analysis, the algorithm generates new and original pieces of artwork. We can therefore perhaps consider the artwork above to be raw aesthetic data; an application of thousands of subtle aesthetic rules and regulations without any sort of conscious guidance as to what they are attempting to achieve through their use. Our brain recognises the creative 'rules', yet cannot go as far as to identify any unified concept depicted on canvas. So in some ways, these algorithms share our definition of creativity. It 'creates something' in accordance to a set of learned aesthetic regulations, but cannot create something coherent as a result.
If a machine independently consults learned aesthetic rules and regulations, and is therefore able to create something that we recognise completely unassisted, can it be said to be creative, then? The following clip shows Mario Klingemann's 'Memories of Passersby I'; a machine that continuously generates portraits of non-existent people. Watch the clip below:
Through a series of neural networks, this algorithm can generate endless portraits by feeding back upon itself. The machine is completely unaided, and can churn out an infinite numbers of 'paintings' completely on its own. This algorithm uses a set of aesthetic regulations to create something coherent and original. Can it therefore be said to have an imagination?
It's interesting to note that both Patrick Tresset and Mario Klingemann still see themselves as the artists, despite the apparent complexity and independence of their machines. When asked if he thinks robots will ever be able to be truly creative, Tresset responds with the following:
'It depends what you mean by 'creative'. For me, being creative would be a robot in a factory that welding cars and suddenly starts to make a sculpture by itself. Then it would be creative by itself, so it would be the author of the sculpture because he would have to make the decision to make a sculpture instead of a car.'
So maybe we can conclude that essence of creativity lies in its effect, rather than its cause. The why rather than the how. AI may be becoming increasingly advanced in the ways in which it can simulate aspects of human behaviour. The ways in which our minds produce creative works are definitely something that can be simulated to some extent. As we create, we consult a vast database of aesthetic information that we have acquired throughout our lifetimes, and we apply them to create something that is intelligible to an audience or to ourselves. However, the difference between us and AI simulations lies in the motivation behind being creative, rather than the process of creativity itself. Although a machine can 'imagine' in the sense that it can produce artwork of its own accord, it can only do so when instructed, and does not have an awareness of what it's doing. So whilst a machine can simulate the mechanics of creativity, we have a long way to go before a machine can possess the emotional complexity required to be 'want' to be creative. So for now, it's probably safe to conclude that conventional artists are safe from automation replacement. However, in terms of being in 'some place in between', automatic artists like Jackson Pollock may have cause to look over their shoulders in the coming years.
