2. Creativity of, around, and with AI

In our last lesson, we explored the meaning of artificial intelligence, and identified several types of intelligence that we humans have evolved, and that we might wish to impart upon an algorithm or machine. In this lesson, we will narrow in on one of these intelligences: the ability or skill to be creative.

Creativity is the part of humanity that is difficult to quantify, difficult to separate out from the rest of the human experience in a clear and logical way. What is creativity? Can we reasonably extract whatever mysterious thing that it is--the seemingly random firing of synapses and neurons that sparks a new idea, strings together a sentence, changes the brushstroke just so--and condense it to an algorithm, feed it with zeros and ones to a computer? In this lesson section we will be framing the context of what art is, and how it has developed in conjunction with technology, from the very beginning of human history.

We then explore how generative AI is challenging the current mainstream art world, and develop questions to help challenge our perception of art and creativity.

The human experience with creativity is bound tightly with our evolution. Origin myths such as Prometheus stealing fire from Mount Olympus to bring to humanity, or the god Khnum crafting us out of clay, demonstrates how ancient cultures revered the act of creativity and reflects the awe they held for creative thinking and acts, which were necessary for our survival and the building of civilisation.

The oldest known cave painting, in Borneo, is over 40,000--potentially even 52,000--years old. These cave paintings, which might seem simplistic compared to, say, the CGI artistry of Avenger's Endgame, are in fact rich repositories of information and capture the creativity efforts of early humans. Some theories have arisen suggesting that, in some locations, they act as a library of agricultural information, such as the fertility cycles of important prey. Other studies suggest that the paintings were placed and drawn so as to appear animated when in the light of a fire, a cinematic experience for the prehistoric age.

Oral storytelling also thrived from the very beginning of humanity. Much of our knowledge of the ancient world has indeed been lost because of the enduring nature of oral storytelling throughout many of the world's cultures. Examples include the story of the Trojan war, and the soldiers that dealt with the repercussions of it afterwards--The Iliad and The Odyssey, preserved only due to transcription by the mysterious literary figure Homer (who may be one person or many). Music, too, thrived. In Slovenia, a 60,000 year old Neanderthalian flute has been found, and instruments ranging from lyres to harps to lutes to water-organs to flutes have been found around the world.

Across the world, humans: painted, sculpted, sung and played instruments, told and wrote stories, performed, danced, and dreamed of architecture throughout ancient history.

Technology also always had a role to play in our creativity. Those meticulous painters of the caves were limited by the tools at their disposal, just as oral storytellers were limited by their respective culture's ideas and abilities surrounding literacy. As cultures evolved, so did ideas about literacy, with the majority of cultures thinking more about preserving their stories, be that through developing alphabet or permanently altering edifices to record information. The ability to paint evolved too, with an increasing number of dyes being developed to create colours, canvases becoming hardier and more structured, and so on.

In the Western world, a significant technological milestone was reached in the 1440s with the invention of the printing press by Johannes Gutenberg. This radically changed the landscape of the literary world, democratising access to information and allowing for other types of literature to evolve, as no longer were books only available to wealthy patrons who typically were illuminating Bible manuscripts. Another big technological milestone is the invention of the photographic camera (interestingly, and pertinent to our above discussion--cameras have been theorised about, and some rudimentary forms engineered, since ancient history). Whilst these technologies are generally considered, on the whole, in a positive light today, there were backlashes and drawbacks at the time. The printing press threatened the power structures of the time, and the increased literacy it fostered also stoked fires of social revolution that threatened stability in Europe. The camera, similarly, evoked vocal criticism from some artists; for example, Charles Baudelaire, poet and critic, wrote:

...I am convinced that the ill-applied developments of photography, like all other purely material developments of progress, have contrib­uted much to the impoverishment of the French artistic genius, which is already so scarce.

Yet photography did allow for the expansion of new forms of artistry. There was economic upheaval involved in this process, as patronage from aristocrats withered, the wealthy no longer obligated to pay for portraiture (though the development of portrait photography is not the sole reason for the slow dissipation of patronage). Those artists who were able to survive this upheaval (typically due to being wealthy themselves), were able to break free of realism, and explore fantasy and mythological themes. This paved the way for Impressionism, which emphasised the feeling a painting was able to express rather than portraying a strictly realistic landscape or individuals.

We can see even from this example that what gets called art, and what doesn't, is a discussion very familiar to the art world. The Western art community has grappled with questions of patriarchy, sexism, classism, and racism in dealing with this topic, and extensive research and discourse in this community already exists to critically examine the hegemonies and structures that exist within said community. Thus, the computer science field has much to learn from art communities.

We pull now from a case study based on the works of one of digilab's own: artist and software engineer Dr Freddy Wordingham.

Part of the ethical discourse that we are experiencing live, in front of us, as these generative AI tools take over the landscape in front of us, is this discussion of whether it counts as art or not. This can be see in recent decisions, such as the Colorado State Fair awarding an art prize to a Midjourney-generated piece. The artist clearly labelled the use of Midjourney, but nevertheless experienced quite a backlash as other artists questioned the level of work and the nature of the creative process that goes in to creating a piece with Midjourney versus drawing (digitally or otherwise) a painting wholly themselves. Another recent example is more legislative--the United States Copyright Office has repeatedly rejected copyright protection for Stpehen Thaler's AI creation, entitled A Recent Entrance to Paradise. The ruling cited a lack of human authorship as the core reasoning. This implication is wide-ranging, particularly as we experience increasing number of strikes across creative industries. If you cannot copyright protect AI-generated content, will that be the prevailing factor behind entertainment companies protecting the rights of writers and actors, which can be copyrighted?

This echoes decisions made across publications such as the academic research journal Nature and the UK news outlet The Guardian, both of whom have updated their editorial standards to indicate that AI-generated output cannot claim authorship, and instead reserves authorship exclusively for humans.

Of course, these conversations are boundaried by the current limits of technology. There is yet to be algorithm that has indisputably, with peer-review, been able to demonstrate the passing of the tests that we have constructed for self-awareness. Once we do, we must further confront the prospect of defining humanity.

Art has had many definitions over the years, ever since humanity has gathered the first animal hair to create paintbrushes, or invented tales around the hearth fire. We even often describe machines and mathematics beautiful.

Pure mathematics is, in its own way, the poetry of logical ideas. --Albert Einstein

However, we tend to use these tools to replicate or simulate us, and our behaviour. These are separate motivations to art, which is typically an expression of inspiration or self-discovery, exploring the artist's self-awareness and their relation (coloured by their experiences) to the subject of the art. Thus can machines never create until they themselves have self-awareness, and the ability to make artistic decisions (or indeed: choose not to be artists)?

If we can bestow authorship on to an algorithm or a machine--what does that mean about what it means to be human?

How can we protect the livelihood and creative processes of humans if algorithmically produced content is going to replace film, art, writing, and music ventures? What levels of transparency are required in the case of generative AI--is labelling sufficient to protect consumer discretion?

If you received nothing but algorithmically generated emails all day--would you be more or less motivated to go to work? Why or why not?

Finally, if you are interested in detecting whether your work has been included in AI algorithms, there is some ability to capture this with Have I Been Trained? and to combat the inclusion of your art work, if desired, with Glaze. For websites or writing work, Neil Clarke's blog provides some details on how to prevent AI bots from scraping data.

Recommended books and longer reads:

We recommend either utilising a public library to access these books, or, if buying online, using Bookshop.org, which supports independent bookshops. Some books can also be accessed with JSTOR, if you have access to an educational license through a school, university, or research organisation.

• The Creativity Code: How AI is Learning to Write, Paint and Think by Marcus Du Sautoy. This book, written by University of Oxford mathematician and professor, examines how machine learning algorithms advanced from playing simple video games all the way to winning games of the incredibly complex Go against masters of the game. Sautoy engages with complex questions about creativity, and embraces the interdisciplinary nature of book's subject--never do you feel he is deriding the value of artists or assumes pretensions about his ability to contribute to the world of fine art, literature, and music.