From generating paintings that do not belong to any existing style, to enhancing human creativity with new tools and techniques, to challenging the definition and value of art with questions about originality, authorship and aesthetics, AI is reshaping the art scene in unprecedented ways.

But what does this mean for artists, collectors and consumers of art? How do we evaluate the quality, meaning and worth of AI-generated art? These are some of the questions that we need to grapple with as AI becomes more ubiquitous and sophisticated in producing art.

What are the traditional methods used to evaluate art?

Over the time, the techniques of evaluating conventional art have been well developed. The value of the art is driven by a few factors driven from the wider cultural impact the author had and complemented by technical aspects of the specific work or artists overall portfolio, like quality, rarity, condition and provenance

The quality of the artwork comes from the skills, techniques and style of the artist; rarity refers to how many copies or editions of the artwork exist; condition refers to how well-preserved or damaged the artwork is; and provenance refers to the history and ownership of the artwork. 

For example, a painting by Leonardo da Vinci would be extremely valuable because it is very rare (only about 15 paintings attributed to him survive), very high-quality (he was a master painter and inventor), very well-preserved (he used oil paint that lasts longer than other mediums) and very well-documented (he was famous during his lifetime and his works have been studied extensively).

Digital art is created by human artists using digital tools and software. These artworks are usually unlimited in number, as they can be copied or reproduced easily without losing quality. 

Their value depends largely on their originality, popularity, innovation and authenticity. Originality refers to how novel or unique the artwork is; popularity refers to how many people view or appreciate the artwork; innovation refers to how creative or groundbreaking the artwork is; and authenticity refers to how verifiable or traceable the artwork is.

For example, a digital collage by Beeple would be very valuable because it is very original (he created a new image every day for 13 years), very popular (he has millions of followers on social media), very innovative (he used various techniques and themes to create his images) and very authentic (he used blockchain technology to certify his works).

Beyond the Canvas: The Complexities of Valuing Art Created by Artificial Intelligence

AI art refers to artworks that are created with artificial intelligence using algorithms, data sets and neural networks with the human guidance through instructions called prompts. 

While early explorations of using AI to create art relied more on technological skill and novelty for value creation (For example, already in 2018, a painting created by an AI algorithm sold for over $400,000 at a Christie's auction), the democratization of AI art creation introduced with tools like Dalle, Stable Diffusion, Midjourney and others have turned the valuation tables again. 

Edmond de Belamy, from La Famille de Belamy, sold for more than $400 000 at a Christie's auction

With tools like these the starting point of one art instance is a prompt - a textual guide for the AI engine for image generation. This prompt includes the description of image subject and a styling guides. Mastering the skills of prompt writing results in better quality of generated images. Good prompt writers are well versed in artistic styles as well as the intricacies of each AI model. Also, good prompts deliver fairly consistent results when executed again. So should we take into account the quality of the prompt?

But prompt writing is just half of the equation. The other half is the author’s capability to come up with idea or concept, the story or atmosphere of the image.  

Good AI artists introduce novel ideas, unique styling, catchy stories and atmospheres. 

AI art has found its way to viewers through the social media channels like Instagram or Twitter. These channels allow connecting the author and art piece together and bring in another dimension for AI art valuation - popularity. 

While two persons may generate quite equal depictions of black roses quite easily, the perceived values of two pictures differ significantly between the author who has tens of thousands followers vs the one who has few hundred or so. 

But maybe we are on the wrong track when we try to evaluate an AI art piece from the same perspective as conventional art piece?

While both convey their message visually, there are so deep differences between them in their purpose, production methods and even consumption methods. 

Evaluating conventional art, Jarmo Tuisk, Midjourney, 08.03.2023

In social media the life-span of one artpiece may very well be around 24h or less, rarely being interesting for the users later. Algorithms dictate that in order to stay in the feed of the audience the artist need to post an image per day or even more. 

In my own experience, posting 2-3 art sets or short videos per day result the widest audience reach. Usually I post 4-10 variation images per post, that makes 8-30 art pieces per day. It’s obvious that they cannot very well be compared to the paintings that have taken weeks, months or years to complete. 

Maybe it’s wiser to then evaluate the AI artist’s collection as a whole and take each single image as a part of the larger picture, or stream of ideas. In that case the quality we may want to look into is the consistency of style, the subject or some other uniting aspect that tie the artist’s collection together.

Also we still may want to factor in the artistic impact of the art collection. As in case of conventional art, so in AI art, the value of the AI artist’s stream can perhaps be measured by how many followers will later try to emulate the style or art subject. 

Art in the Age of Algorithms: humans vs machines?

The AI art is still fairly young and still trying to define itself and its place in wider art scene. There is no clear-cut answer or consensus on how AI will transform the value of art, but there are some possible scenarios and implications that we can explore.

One scenario is that AI will create a new category of art that is distinct from human-made art. This would require us to develop new criteria and standards for judging AI-generated art, such as novelty, complexity, coherence and diversity. We would also need to consider the role and contribution of human programmers, designers and curators who shape the algorithms and data sets that enable AI to create art. In this scenario, AI-generated art would have its own niche market and audience that appreciate its uniqueness and innovation.

Another scenario is that AI will compete with human-made art for attention, recognition and revenue. This would create a situation where human artists have to adapt their skills and styles to keep up with or surpass AI's capabilities.

A third scenario is that AI will collaborate with human-made art for mutual enhancement and enrichment. This would create a situation where human artists use AI as a tool, a partner or a source of inspiration for their own creative processes. 

These are not mutually exclusive scenarios, but rather different aspects or dimensions of how AI will transform the value of art.

There may be other scenarios or outcomes that we have not anticipated or imagined yet. The point is that we need to be aware, informed and engaged in this transformation, as it will affect not only our culture, but also our society, our economy and our humanity.

Artificial intelligence is changing the face of art, but it does not have to change its soul. We can still appreciate, celebrate and value human creativity, expression and emotion in all forms of art, whether they are made by humans or machines.

We can also embrace, learn from and collaborate with artificial intelligence, as it can offer us new ways of seeing, thinking and creating.

The value of art is not fixed or static, but dynamic and evolving. And so are we.