When someone says "artist", a host of greats from past centuries come to mind. Michelangelo, Renaissance painter of the Sistine Chapel; Leonardo da Vinci, maker of the gently smiling Mona Lisa; Picasso, heralder of the abstract art form Cubism. Perhaps you think of more ‘modern’ forms of art; like clay-ring sculptors, graphic designers, and realistic cake decorators. One thing is for sure: all the names of history and today were human. However, in recent years, a new type of expression has come to the forefront; behold, the rise of AI art!
While AI is far from a brand-new concept, here's a quick rundown of how it works: AI, or Artificial Intelligence, refers to systems and machines that aim to mimic human intelligence to perform tasks, and continuously improve themselves based on the knowledge that they collect. This imitates how the human brain builds on new experiences to shape actions we undertake in the future. In AI art, a research area called computer vision is utilised. Computer vision is dedicated to designing algorithms that process meaningful visual information.
Generational models, a subclass of computer vision algorithms, are AI’s version of human neural networks. By feeding the neural network large datasets containing millions of images, the network is essentially ‘trained’ to analyse and produce new images of its own with components of images it has ‘seen’. Most AI art programs out there require users to input prompts for pieces of artwork; by describing the results users want, the AI network is guided to produce those results from images that contain similar subjects.
Popular art programs like Dream: Wombo and DALL-E-mini rely on prompts as mentioned above, and their results have quickly flooded the internet with memes and trends, its creators vying to use these apps for both creative and comedic purposes.
So, is AI art ‘real’ art? Do computer-generated images live up to their predecessors? A craft like oil painting requires years of practice to hone skill; to become a true master at traditional art, one normally has to pour hours, days and years into perfecting what they do. Unlike AI art, traditional art usually requires expensive supplies and materials (ink, fabric, clay), and one has to be prepared for the cost a career or hobby in art can bring. Additionally, traditional artists have to price their pieces quite high, considering the time and money that go into making each individual product. In contrast, AI art summons brushstrokes with each press of a button and draws from already existing images; how will artists of today be able to compete with this endlessly lucrative behemoth?
Although the barrier to being an ‘artist’ has been lowered by AI technology, we have to consider the skill and talent that goes into the process of curation. It’s not easy; if a dataset contains a small variety of images, a network will be unable to represent features detailed in a user’s prompt and not perform up to its full potential. If a dataset is biased (choosing images according to one’s personal preference), a network can reproduce bias and affect results. Artist Mario Kinglemann carefully curated a dataset of thousands of portraits from the 17th to 18th century; here, Kinglemann’s bias of choosing images for his dataset shows hints of unique personality captured in his final artworks.
If photography is considered an art form, then AI art is not that different. Both mediums capture existing images and tweak them until the end product suits the creator’s fancy. It is hard to get the lighting, shadows and mood of a photo right; similarly, it is hard to input the perfect prompt to get the best image from an AI algorithm.
Although AI and traditional art may seem vastly different on the surface, there is no denying that both mediums require large investments of money and time; instead of shying away from new technologies and berating the arrival of new competitors, we should embrace this new art form with open arms.
Sources:
Millière, R. (2022, July 17). AI Art Is Challenging the Boundaries of Curation. Wired. https://www.wired.com/story/dalle-art-curation-artificial-intelligence/
Comentarios