In just a few years, the number of artworks produced by self-proclaimed AI artists has increased dramatically. Some of these works have been sold by major auction houses at staggering prices and have ended up in prestigious collections. Initially led by a few tech-savvy artists who embraced computer programming as part of their creative process, AI art has recently been embraced by the masses as image-generating technology has become both more efficient and easier to use without coding skills.
The AI art movement draws on technical advancements in computer vision, a field of research dedicated to designing algorithms capable of processing meaningful visual information. A subclass of computer vision algorithms, called generative models, take center stage in this story. Generative models are artificial neural networks that can be “trained” on large datasets containing millions of images and learned to encode their statistically salient features. After training, they can produce entirely new images that are not contained in the original dataset, often guided by text prompts that explicitly describe the desired results. Until recently, the images produced by this approach somewhat lacked consistency or detail, although they possessed an undeniable surreal charm that captured the attention of many serious artists. However, earlier this year, tech company Open AI unveiled a new model, dubbed DALL·E 2, capable of generating remarkably consistent and relevant images from virtually any text prompt. DALL·E 2 can even output images in specific styles and imitate famous artists quite convincingly, as long as the desired effect is correctly specified in the prompt. A similar tool was made freely available to the public under the name Craiyon (formerly “DALL E mini”).
The coming of age of AI art raises a number of interesting questions, some of which – like whether AI art is really art, and if so, to what extent it is really created by AI – are not particularly original. These questions echo similar concerns once raised by the invention of photography. With just the push of a button on a camera, someone with no painting skills could suddenly capture a realistic representation of a scene. Today, a person can press a virtual button to run a generative model and produce images of virtually any scene in any style. But cameras and algorithms don’t make art. People do. AI art is art, created by human artists who use algorithms as another tool in their creative arsenal. Although both technologies have lowered the barrier to entry for artistic creation – which calls for celebration rather than worry – the amount of skill, talent and intentionality involved should not be underestimated. in creating interesting works of art.
Like any new tool, generative models introduce significant changes to the artistic creation process. In particular, AI art expands the multifaceted notion of curation and continues to blur the line between curation and creation.
There are at least three ways in which making art with AI can involve acts of curation. The first, and the least original, concerns the curation of releases. Any generative algorithm can produce an indefinite number of images, but not all of them will generally be given artistic status. The output curation process is very familiar to photographers, some of whom regularly capture hundreds or thousands of shots from which a few, if any, can be carefully selected for display. Unlike painters and sculptors, photographers and AI artists have to deal with an abundance of (digital) objects, the preservation of which is an integral part of the artistic process. In AI research generally, “picking cherries” on particularly good results is considered bad scientific practice, a way to deceptively inflate a model’s perceived performance. When it comes to AI art, however, cherry picking can be the name of the game. The intentions and artistic sensibility of the artist can be expressed in the very act of promoting productions specific to the status of works of art.
Second, curation can also take place before the images are generated. In fact, while “curation” as applied to art generally refers to the process of selecting existing works for display, curation in AI research colloquially refers to the work required to create a whole of data on which to train an artificial neural network. This work is crucial because if a dataset is poorly designed, the network will often fail to learn to represent the desired features and function properly. Moreover, if a dataset is biased, the network will tend to reproduce, or even amplify, that bias, including, for example, harmful stereotypes. As the saying goes, “garbage in, garbage out”. The adage also applies to AI art, except that “trash” takes on an aesthetic (and subjective) dimension.
For his work Memories of passers-by I (2018), German artist Mario Kinglemann, one of the pioneers of AI art, carefully curated a dataset of thousands of portraits from the 17th to 19th centuries. He then used this dataset to train generative algorithms that could produce an endless stream of new portraits that shared similar aesthetic characteristics, displayed in real time on two screens (one for female portraits, one for male portraits). This is an illustration example of AI that does not involve output curation. Yet the meticulous curation of training data played a fundamental role in its design. Here, “bias” is a blessing: the dataset was heavily skewed according to the artist’s personal aesthetic preferences and tastes, and this aesthetic bias is reflected in the final work, albeit through the distorting lens of the computer-driven generation process.
Another novelty spurred by recent advances in generative algorithms is the ability to produce images by describing the desired outcome in natural language. This is called “prompting” or guiding the algorithm with text prompts, as opposed to sampling random outputs. Consider the illustration accompanying this article: the collage features several images generated by inviting DALL E 2 with the phrases “an AI image generation algorithm, concept art”, “collage with images created by a generative AI model, illustration from Wired magazine,” and “an artist who curates artwork produced with an AI algorithm, concept art.”