AI Meets Art: Can Creativity Be Replicated?

January 16, 2021

Art is subjective, reactionary, and culturally dependent. AI is objective, logical, and universal. What happens when these two worlds collide? What does this collaboration of man and machine mean for the future of the creative artistic process?

Art makes us aware of our human condition.

In the past, all sorts of art ever created came from the human mind.

This is not the only case anymore.

The advances of Artificial Intelligence (AI) are making machines able to mimic humans day by day, be it the ability to learn, see, or speak. it has come the time to ask ourselves if this is true to creativity.

In this article, we explore how AI has been contributing to multiple manifestations of art and what we should expect about the integration of creativity and technology for the future of society.

Past and Present of Art & Creativity

Artists are often perceived as storytellers, dreamers, or out of the box persons. They cultivate their artwork in many forms, expanding ideas and appealing to the senses of audiences through visuals, sounds, and many other forms.

History has made artistic works and technological tools in the former eras to work along with each other to delve into the new ways to express ourselves with increasingly complex partners — paint, write, sculpt, and music.

The desire to create uniqueness and beauty has gifted mankind with remarkable pieces of art that remain highly valued and have inspired generations after centuries through the several stages of schools or art.

Advances in AI techniques have been walking hand in hand with the AI and art timeline in the past decades. Machine Learning “state of the art” models and algorithms make our lives easier by doing tasks at a human level accuracy. Deep Learning and Natural Language Processing are at their finest growing curves now, and applications of Computer Vision and text analysis are becoming the most common ones when it comes to AI and art.

We can now ask ourselves: What happens if AI becomes a creator? What happens if some sophisticated intelligent entity starts building creative artwork alongside humans?

In the early 21st century, the artistic environment dramatically changed. AI-driven artworks invaded museums and galleries, such as the “Portrait of Edmond Bellamy” which had an auction pre-sale-estimate of $433,500 USD.

Although AI-generated artwork is recognized as an esteemed technical piece of art, it can also be associated with a lack of emotional value and passion by the absence of human inspiration and social context.

Being able to replicate creative processes via digital devices is a complex task, as the understanding of the human mind is still significantly unknown even to human researchers and scientists.

From a psychology perspective, creativity may be defined as a new idea that has value in solving a problem, such as a new useful object or a revolutionary idea. From a mathematical or statistical point of view, creativity may come from a random stochastic process occurring in one’s intellect, inspired by learning from previous experiences and outcomes.

In the past, technology was an auxiliary tool for artists. However, with the advances of tools like AI, the entire concept of art is exposed to a new interpretation. The art of the 21st century is no longer inspired purely by feelings and thoughts — characteristics that are intrinsically human — but also by patterns and randomness not necessarily understandable by humans.

Is creativity triggered by random event in our heads that give life to an empty canvas? Is there a set of statutes to art? Or can it be implemented as a set of predefined algorithmic instructions?

AI meets Art

AI is built to have the capacity to think like the human brain. The advances in AI for social impact showed the potential of such techniques for innovation, sustainability and humanitarian factors. Surprisingly to some, AI can be related to social impact through art too.

You may now be asking yourself whether AI can replicate complex tasks such as creating art: the catch lies in how these models are trained.

It’s like teaching a child how to read and how to perceive senses. One has an intuition on these principles of nature. But this intuition breaks as soon as we try to define artistry and creativity.

While some may be anxious that the evolution of AI can be threatening for artists, it is also a fact that it produces a creative roadmap for thinking in various original directions.

Augmenting creativity through AI can act as a guiding force for artists and a driving force for the young generation to get involved with art.

As AI and art integration, there are multiple applications that can be brought to light, from bringing back life of old portraits to the endless creativity of innovative Art in the 21st century.

AI for Art and History Conservation

It is not quite rare that our parents show us the pictures of our ancestors and we are worried that they are too deteriorated for us to still perceive details.

In a similar fashion, as years and centuries go by, the conservation of art pieces is crucial to keep these suitable to be studied by future generations of artists and keep inspiring art enthusiasts.

Generative design is another side of augmented creativity which will have various designs generated with a blend of ‘intuition’ and ‘emotions’. The neural networks can be trained to produce a variety of artistic features that can be incorporated in various ways to produce desirable effects.

An example of an AI application for historical architecture is Deep Image Prior, a neural network created by an Oxford University research team and Nvidia’s image reconstruction, which demonstrated the process of digital restoration.

Moreover, moving backward in history, a number of portraits and paintings have lost their significance, particularly because we are unable to interpret their purpose.

AI helps us to decode the artist’s minds by catching details that would not be visible to human eyes through the identification of patterns. This way, historic art can be re-written and ultimately become better understood by historians and art researchers.

AI also plays a significant role in understanding history. From understanding the lives of dinosaurs to mapping historical locations by 3D model recreations. AI is expected to help in remembering the history of places where war and destruction events took place.

Bringing Artists Back to Life

One of the most exciting aspects of applying AI to art is the idea that creative artwork from remarkable artists does not have to end when the artists die.

“When you are a genius, you do not have the right to die, because we are necessary for the progress of humanity.” — Salvador Dali’s last public statement, 1989.

One beautiful application of AI in art is the awarded Dali Lives project, in the Dali Museum in Florida. This project, developed to celebrate what would have been Dalí’s 115th birthday, allows museum visitors to interact with an equivalent to a deepfake Salvador Dalí through screens around the Museum.

The artificial intelligence technology used to bring Dalí back to life was trained on hundreds of interviews, quotes, pictures, and original writings from Dali. In this way, it was possible to capture the most realistic possible version of the artist, based on all his recorded essence left in the world through voice, image, and language.

Dali Lives Project. Source: Youtube
.

The technique behind the scenes is called Generative Adversarial Networks (GANs), which is a powerful tool based on game theory, in which two neural networks contest with each other.

The most famous use of GANs is deepfakes. Apart from the negative connotation associated with fake news, GANs can be used in positive art applications, such as bring movie stars back to life in iconic roles and also help actors who suffer from performance anxiety to have a backup.

…

Recurrent Neural Networks (RNNs), a class or artificial neural networks associated with sequential models and temporal series, are other powerful tools for bringing the work from writers back to life.

RNNs can be trained to learn the patterns of the entire life work of famous writers or stories, and later extrapolate it to new content. That was the case of the new chapter of Harry Potter written by an AI, which contains parts that could fool even the most devoted fans of the series.

See below an example of an RNN applied to recreate Shakespeare plays (originally posted here):

ROMEO: how I, away too put That you shall have thieffort, are but love.

JULIET: Go, fight, sir: we say ‘Ay,’ and alack to stand and not to go to; And washt us him to-domm. Ay, my ows young; a man hear from his monsher to thee.

KING RICHARD III: Come, cease. O broteld the costime’s deforment! Thou wilt was quite.

PAULINA: I would you say the hour! Ah, hole for your company: But, good my lord; we have a king, of peace?

BALTHASAR: Cadul and washee could he ha! To curit her I may wench.

GLOUCESTER: Had you here shall such a pierce to temper; Or might his noble offery owe and speed Which seemest thy trims in a weaky amidude By this to the dother, dods citizens.

Third Citizen:
Madam sweet give reward, rebeire them With news gone! Pluck yielding: ’tis sign out things Within risess in strifes all ten times, To dish his finmers for briefily.

JULIET:
Gentlemen, God eveI come approbouting his wife as it, — triumphrous night change you gods, thou goest:
To which will dispersed and France.

Similar to actors, painters, and writers, AI can also be inspired by deceased musicians, architects, and many other remarkable artists.

Creating AI-based Art

When AI meets art, it democratizes the artistic process by making it more accessible to those who do not have artistic skills.

The partnership between AI and creators helps to build AI-inspired work.

This means that inspirations from several artists can be designed to be merged by a human designer, but implemented by an AI, generating a unique piece of art through the collaboration of human and machine.

In this sense, humans still remain as the core tool in the form of “Augmented Intelligence” to propel into the future of art — whereas AI acts as a tool to enhance our skills and craftsmanship with collaborative-driven effort, rather than invoking the idea of autonomous machines surpassing human beings.

An example of collaboration between humans and AI to build art is through the Drawing Operations Unit: Generation 2 (DOUG) — which is an AI worker who collaborates with a human artist by learning her drawing styles and patterns and late helping her to create more art.

…

Neural Style Transfer (NST) is another amazing application of AI in a creative context. It is a process made possible due to the virtue of deep learning, which migrates the semantic content of one image to different styles.

The below image is an amalgam of the art styles of “Mona Lisa” by Da Vinci and “The Starry Night” by Vincent van Gogh.

One might argue that this is just copying an art style and using it and that no element of uniqueness is present. However, the process is not about replicating an art form but understanding what art itself is.

Comprehend such an abstract concept and apply it to different domains is art itself and part of the creative process of many artists who are inspired by the giants from the past.

The combination of several works of art to inspire AI to create has applications in diverse domains of art, such as:

Music: check out the Top 10 AI composers of 2020;
Themes: the MIT Shelley which transforms pictures into horror images;
Paintings: GANs allow for creating realistic paintings that are authored by algorithms, such as in the website “thisartworkdoesnotexist.com”.
Combinations of art domains: another form of AI-driven art is by the combination of art. The voice-driven sound sculpture, allows anyone to create sculptures based on the human voice.

This artistic image seems to be created by an artist or designer with his bold brush strokes on the canvas. But what’s interesting about it, is that it wasn’t created by a human being! Yes, this piece of art was generated using a GAN.

You can go try generating your own unique art using the GAN provided and see creativity literally being created on your screen.

After all the examples seen so far, you might argue that this definition of creativity is still not unique and dependent on random scientific discoveries and mathematical discoveries.

It’s indeed true that these artworks are fueled by certain principles but what matters is that our digital devices are beginning to think virtually. GANs models are able to infer these principles and start amending them and creating a set of data of their own.

AI in the Art Industry

The domain of AI is not limited to fine arts only.

When AI meets art, it is also used to create digital art solutions, rendering environments, architecture, and generating game scenes for visual designers.

After reading about how AI is revolutionizing art and creating its own pieces, you may be wondering how the companies in the Art Industry are taking full advantage of this new aspect.

To answer this question let’s explore some of these sectors.

Game industry: One such pioneer in this field is Promethean AI. It is the world’s first AI-based approach that works together with artists, assists them in the process of building virtual worlds, and helps creative problem solving by suggesting ideas. EA recently built an AI agent that taught itself to play Battlefield 1 and developed a 3D game environment specifically for deep learning networks to navigate.

Movie industry: one of the most promising applications of AI in the art industry goes for sure to the introduction of deepfakes into movies. Deepfakes can play the role of original actors, as well as create persons who do not exist to be a part of a movie. Moreover, AI has also created short film scripts, such as Zone Out and Sunspring based on science fiction scripts.

Design and architecture: the possibilities of AI for the design of environments and products are promising for the new decade.

The 2021 year started with exciting advances from OpenAI on the release of two deep learning models, CLIP (Contrastive Language-Image Pre-training) and DALL·E (named in combination of Salvador Dalí and Pixar’s WALL·E). The models combine text and image data and released what can be the future AI for design, through what became known as the avocado armchair. As the name suggests, the AI generates an image of an armchair inspired by an avocado.

Moreover, 3-D spaces can also be rendered using AI-based models that take care of reflections, shadows, and ray tracing. The model can distinguish between objects at human-like accuracy and can potentially ease up the process of visual designers by giving them a base to work on.

Police Investigation: one positively surprising application of AI and art is by use of AI for crime investigation. It is common that artists work in investigations making sketches of faces, in a somewhat rudimentary attempt to find criminals or missing people. If AI is combined with the artists’ drawings, it is possible to create faces based on the drawings, which can ultimately help police investigations.

Music industry: as mentioned in the previous sections, AI-based music is already a hit and the possibilities of combinations of diverse music styles to form new ones is an exciting idea in the sector.

Culinary arts: another nice application of AI is in the art of cooking. IBM’s Chef Watson uses AI to help develop recipes and advice on food pairings and unique flavors.

Behind the avant-garde uses of AI in Art industry, lies a disturbing question: Who Owns the Art Created by Artificial Intelligence?

As AI becomes more ubiquitous, legal scholars and engineers alike ask themselves who actually gets the copyright from what AI robots produce — man or machine? Is the art form idiosyncratic to a particular entity?

As pointed out by researcher Robert Hart, the ownership laws need to adapt to the reality of the modern world to not delay the advances of the industry.

The Future of Art with AI

In the past, artists had the ability to touch the human mind and convey intense emotions through the use of several elements, such as music, paintings or movies.

Given the advances of technology, it is natural to ask ourselves if this will be an extra element in the equation of the artistic process and the entire art sector.

AI is changing the relationship between art and the until now art enthusiasts, by democratizing the artistic process.

In the near future, the perception of what art will mean may change drastically.

At the moment it is hard to say, but it is possible that an entirely new genre of art associated with AI creation takes place, and things such as AI movies and AI music start to be consumed in the media.

In the same way, the price of art pieces can have a significant impact when AI is added to the creative process. On one hand, there may be that the price of human artwork increases by the public choosing to value human creativity as genuine art. On the other hand, AI art can make prices cheaper, as they make the creation and design processes automated, becoming appealing to industry sectors that need original and cheaper pieces of art.

Deep learning models are becoming experts at understanding art styles to the extent that they can accurately predict if a particular art style has been forged. This shows the potential of AI for the future of art authentication.

In terms of industry, it is expected that companies will further improve AI algorithms to generate better quality content in gaming, movies, or any other creative sectors. AI may also help to enhance human creativity, through AI algorithms that are told what to create.

Additionally, it is believed that generative design, where the designer inputs their design goals and specifications, will disrupt many industries, including architecture, consumer product design, and other engineering fields.

At the moment, it is still hard to say what the outcome of the AI and art encounter in many sectors is going to be.

But for the technology enthusiasts, this is for sure a topic to keep an eye on in the next few years.

This article has been written by Rosana de Oliveira Gomes, Ajaykumar Palaniswamy, Sara El-Ateif, and Harini Suresh through the Omdena Writers’ Academy and under the mentoring of Mohit Sharma