Codenrama Logo

Machine Learning Art

After spending years making machines recognize reality, the same tools produced with a simple change of direction can generate a new activity: transforming text into images. A New, albeit not fully recognized Renaissance (a period of cultural and artistic rebirth in Europe, roughly spanning from the 14th to the 17th century) is emerging, thanks to the deadly impact of artificial intelligence, more precisely Machine Learning (abbreviated hereafter as MLA where the A stands for Art) on all human activities.

"MLA" lowers the barriers to entry into the world of art, allowing anyone with access to certain computer tools to express themselves artistically, regardless of traditional technical or artistic training. There is no reason to have only one style when you can have and combine them all.

Image 1

The vast amount of images on which machine learning has been trained, its relationship with the labels associated with each of them should be understood as a kind of visual unconscious, and its artistic use can be interpreted as an experimental psychoanalytic investigation of art itself. Machine learning does not simply "copy" existing art, but rather learns from a vast dataset to generate new compositions. These algorithms identify patterns, styles, and trends within the provided data, but their output is the result of internal processing processes that can produce unique and unexpected results, not directly attributable to pre-existing works. The program relies on managing millions if not billions of images (previously made comparable) associated with a term or a list of any terms and attributes. Then it generates a random image and with refinements and subsequent optimizations (here the mathematically complicated part) it tends towards the text to illustrate.

Image 2

The interesting thing is that the phrases do not necessarily have to make sense since the mechanism in question is essentially an excitatory mechanism that can also be triggered by nonexistent words or compositions of existing and nonexistent words and numbers. The harder an artist struggles to compose a work, the more likely they become uncritical towards it. Conversely, true freedom in art lies precisely in the lack of mercy towards one's own artistic production. The impressive ease with which variants or new instances of the same idea can be produced helps to adopt a much more critical and radical attitude towards one's own work. It's easy to eliminate something that cost very little effort in production. Machine Learning Art generates an extreme freedom, which in the art world also implies an extreme ruthlessness in selection. But it's also the freedom to invent one's own resounding eclecticism, a personal mix, the freedom to traverse non-trivial connections between words and images, the freedom to create one's own new style and schema.

Image 3

So, in the face of those intellectuals who argue that artificial intelligence is nothing but plagiarism and superficial mechanics, we strongly assert once again how these processes increase the individual's freedom in artistic creation. In culture, as in facts, everything is intimately connected. Everything is plagiarism, everything is a reinterpretation of previous experiences. Not even artists who claim to be wild are immune to this powerful mutual influence. It's impossible not to be influenced by all previous culture. Despite the construction of the image having become much easier now, all the semantic and aesthetic challenges inherent in its creation remain unchanged.

Image 4

When freedom in creation is so high, the only limit is the ability to conceive and draw new worlds (in this sense, science fiction is a genre that adapts very well to this type of work), and the possibility of a qualitative leap from all points of view begins to be a reality. In generative art, where algorithms create art without direct human intervention or with limited guidance, the boundary between background and foreground becomes fluid. Algorithms can generate images in which subjects seem to emerge from the background or in which background and subject are intertwined in complex ways, creating a simultaneous effect of presence and absence.

Image 5

The ML's ability to simultaneously treat foreground and background in art is not only a demonstration of computational power, but also an invitation to rethink the modes of artistic creation and perception. MLA can improve itself with a process called GAN (Generative Adversarial Network), essentially creating a mechanism in which one machine creates the image and another judges it and eventually sends it back to the first machine until the result is judged to be good enough.

Image 6

The assertion that MLA in art can be defined as 'The Style of all Styles' effectively captures the transformative and inclusive essence of MLA in the contemporary artistic landscape. This view emphasizes the MLA's unique ability to assimilate, reinterpret, and generate art through a wide range of styles and techniques, offering a new dimension to artistic creation. In general, it's quite limiting to consider MLA as a new artistic style when instead it's precisely its experimental, hybrid, random, and ever-changing nature that defines this new dimension. MLA introduces a new aesthetic, founded on the ability to process and reinterpret the vastness of human visual heritage. Algorithms, trained on millennia of artistic production, generate works that reflect a transcultural and transhistorical synthesis, sometimes bizarrely random and grotesque.

Image 7

MLA challenges and renews the quest of hyperrealism and photography to capture and surpass reality. By generating images that can oscillate between highly realistic and absolutely fantastic, ML-generated art broadens the field of visual and conceptual exploration.

Image 8

With its gaze turned towards infinity, MLA has woven a canvas where the light of intellect dances in harmony with the unpredictable shadows of the machine, creating a rainbow bridge between two worlds, one tangible and palpable, the other ethereal and unfathomable. Its work is a polyphonic chant, a symphony where the notes generated by artificial intelligence merge with the melodies of the human soul, in a psychedelic embrace that transcends the boundaries of known reality.