Artificial Intelligence: Development Potential on the Way to Create New Digital art

As neural networks continue to advance, their applications become increasingly diverse. Tesla autopilots are trained with the assistance of their team, and face recognition technology has various applications including photo processing and security systems. Artificial intelligence is utilized for diagnosing diseases. With his assistance, they were able to achieve victory in the elections.

Historically, creativity has been viewed as an area reserved for men. There is growing doubt surrounding this statement. Lee Sedal, the player who was defeated by AlphaGo, expressed that the loss raised doubts about the limits of human creativity. After observing AlphaGo’s gameplay, I experienced uncertainty regarding my own abilities. Today’s post will discuss the ability of robots to enter the realm of art, creativity, emotions, and perception.

Robot Creativity

Self-learning systems have been examined for their ability to generate creative output for a significant period. In 1970, scientists created an algorithm that had the ability to generate prose texts, although the texts were often nonsensical.

Neural networks have acquired the skills to generate images, write music and poetry, and create scripts for movies. The operation principle of algorithms is similar. They analyze a vast collection of artworks, and based on the patterns derived, generate their creation, such as a painting, musical composition, or novel.

The institutionalization of neural networks’ creativity is a gradual process. In 2016, a competition featuring works of art created by robots took place for the first time. The top prize of $40,000 for this year was awarded to PIX18 algorithm of Creative Machines Lab. The algorithm was commended for its ability to create artwork from its stored photographs and produce good brushstrokes.

The victory in the committee was commented on by stating that the composition and brush work was reminiscent of Van Gogh. The color combination is intriguing. The comment appears to be a genuine evaluation of a painting by a hopeful artist.

Perception of Works

The paintings produced by Google’s DeepDream algorithm are considered art due to their creation by artificial intelligence.

Additionally, a significant factor to consider is innovation. According to this criterion, we assess the works of artists. The question remains whether algorithms can generate new works of art if they do not rely on drawing or processing photographs, but instead create abstract paintings.

The Artificial Intelligence and Art Laboratory at Rutgers University used a generative adversarial network (GAN) to explore the answer to this question. In the past, the algorithm was trained using a single discriminator’s feedback: it examined the images, generated its own, and verified the outcome. The images he created were reminiscent of the ones he had previously examined.

The team implemented a second discriminator in the network to enhance its development, which operates in competition with the first. The neural network has analyzed around 81,000 paintings and generated a list of criteria to determine whether a painting can be considered a work of art based on this substantial sample. The second discriminator creates a compilation of styles and conducts a verification operation on the picture to determine similarity. A painting becomes a unique work of art when it is recognized as distinct from any previously existing styles.

The process of drawing a portrait typically involves marking the areas of interest, such as the eyes, mouth, and eyebrows. The program utilizes the neighborhood-growing method for image segmentation and provides justifications for the boundaries of each region.

Here are some additional pieces of information that may be of interest. During a performance, a robot’s musical composition was so impressive that the audience believed it was created by a human. A literary prize was nearly won by a short novel written by a robot from Japan.

Now we come to a significant matter – the issue of how art is received by the viewer. Is there a distinction between the way we perceive a piece of art produced by a human versus that created by a robot? There exists a website called Bot or Not that can determine whether a poem was written by a human or a machine. The solution may not be readily apparent. This territory is unclear or uncertain.

The website Bot or Not features poems that were written by robots, even though they were originally credited to human authors. It can be concluded that these algorithms have successfully passed the Turing test for poetry. The Turing Test requires a computer to demonstrate a level of “humanity” that convinces 30% of its human judges. Oscar Schwartz, the creator of Bot or Not, points out that it is possible to mistake work created by bots for human work and vice versa, highlighting the potential for confusion between human and robot creativity. The interpretation of texts and meanings is evolving, resulting in blurred distinctions between illusion and authenticity.

Creativity is Linked to Emotions

Another issue arises when considering the nature of artwork: what distinguishes it from simply copying or replicating previous experiences.

Colin Martindale, a psychologist from America, presented a unique theory regarding creativity. Based on the research, the creator’s main objective is to elicit emotional stimulation in the consumer. There are multiple ways to achieve this, including novelty, complexity of ideas, intellectual challenge, and ambiguity in interpretations and messages. A society experiencing a decline in excitement levels is undergoing degradation.

Martindale observed and categorized two distinct stages within the cognitive process. The primary process consists of unguided, non-rational thinking similar to the state of dreaming or daydreaming. The secondary process is a conscious, conceptual process that involves problem-solving and the application of logic. The author applied a perspective to the creative process in which conceptual awareness can identify and reason logically, but lacks the ability to generate or derive novel ideas, adhering to the principle of ex nihilo nihil fit – “nothing comes from nothing”. The cognitive process of primordial thinking involves creating connections, making comparisons, and generating unique combinations of mental elements. It provides the necessary material for the processing of conceptual thinking.

The GAN network operates based on the principle of one neural network distinguishing and the other finding associations and making comparisons. The algorithm is based on the theory of creativity and generates new pieces of art that elicit emotional reactions from individuals.

Neural Networks Can Aid both Artists and Musicians

The relationship between art and technology has a long history, with examples such the Renaissance and the experimentation of Leonardo and Michelangelo. Throughout history, artists have utilized new materials, approaches, and inventions in order to create masterpieces and even invent new art forms. Neural networks are now being utilized to aid scientific research in the creative field, alongside individual artistic expressions such as poems, paintings, and music.

The modern music industry is guided by patterns that build a mathematical model of music and aim to create a desired effect from listening to a composition.

An international research team from universities in Japan and Belgium, in collaboration with Crimson Technologies, has developed a machine learning-based device capable of detecting the emotional state of listeners and creating new content based on the collected data.

The Art of Seeing

According to John Berger’s “The Art of Seeing,” vision holds a primary role for individuals in comparison to language. Knowledge has an impact on our assessment. Berger suggests that an image can be interpreted in various ways, and our perception of it is influenced by the method of interpretation we employ.

The topic of algorithmic creativity prompts consideration not only of how programs generate content, but also of how humans perceive creativity. Neural networks are capable of generating poetry, and at times they may be mistaken for human compositions. However, it is our interpretation and comprehension that imbues them with significance. An algorithm perceives words, strokes, colors, and sounds as a collection of signs that can be combined into a cohesive structure. The raw material is the area that the robot cannot perceive, which pertains to the semantic field. It has not been observed yet.

According to Umberto Eco, robots are unable to assign significance to objects or create works of global cultural value. He also notes that despite the abundance of poems being written, there is a lack of true poetry. Artificial intelligence has the ability to create music and poetry that is technically correct, but it is only when a human recognizes it as true art that it can achieve the desired status.