A carregar...
Publicação
Problemáticas da produção de imagens e produtos culturais gerados por sistemas de inteligência artificial: Apropriação e incorporação de representações e elementos surrealistas e do realismo mágico pela arte generativa digital
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 4.89 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Ao longo desta investigação pretende-se compreender que tipo de relação
estética existe entre a arte generativa e a arte realista e surrealista, incluindo os seus
hibridismos e variantes das quais destaquei o realismo mágico. Apesar das nítidas
diferenças entre ambas, o automatismo presente tanto na arte surrealista como na arte
generativa evidencia uma clara necessidade de compreender o funcionamento do
processo criativo, mas também o cruzar de ideias, símbolos e objetos imprevisíveis no
mesmo espaço. A constante procura e fascínio pelo irracional, o estranho e o absurdo
está, no seu cerne, ligada às ruturas com os ideais iluministas e a primazia do
pensamento racional enquanto única forma de atingir o verdadeiro conhecimento.
Este fascínio cresceu com a evolução dos novos aparelhos tecnológicos como a
fotografia e o cinema e, mais tarde, com o surgimento das artes tecnológicas durante os
séculos XX e XXI. O desenvolvimento dos primeiros programas de edição estabeleceu
uma via de contacto fundamental entre a cultura e o digital aprofundado através da
evolução da cultura do software, da sociedade em rede e da Inteligência Artificial.
Inicialmente aplicada como uma ferramenta de catalogação, a inteligência
artificial1 demonstrou desde cedo ser uma grande vantagem na classificação e
categorização de imagens com a introdução dos Convolutional Neural Networks (CNN).
Estes sistemas aceleraram o processo de digitalização e de reconhecimento de imagens
e de acesso e manipulação de produtos culturais disponibilizados pelas bases de dados
na internet. Atualmente, com os sistemas de modelos generativos como os Generative
Adversarial Networks (GAN) e os Artificial Intelligence Creative Adversarial Network ou
AICAN, a IA tornou-se capaz de gerar ou produzir inúmeras variações e falsificações com
base em imagens reais.
A comercialização e integração destes sistemas de IA na área de produção de
conteúdos culturais e artísticos tem visto um crescente aumento do seu número de
utilizadores. Este fenómeno é também refletido pelo crescimento do número de
trabalhos de Arte Generativa nas diversas plataformas, um estilo de arte popularizado
no século XXI. DALL-E 2 e Midjourney, sistemas desenvolvidos pela OpenAI, e ainda
Stable Diffusion desenvolvido pela Stability AI, são alguns destes modelos generativos
capazes de reconhecer e categorizar tendências artísticas e culturais, combiná-las e
reproduzi-las.
Ultimamente, temos observado também um investimento na aplicação de
sistemas de IA generativos em filtros de imagem nas redes sociais, muitos dos quais
recorrem ao uso de elementos do fantástico e do mágico.
Esta dissertação pretende então clarificar o porquê de os sistemas de IA fazerem
uso de elementos do realismo mágico e surrealistas no seu processo criativo. Poderemos
considerar os sistemas de IA como sendo meras ferramentas de criação de produtos
culturais baseados em variações do que já existe? Será que existe uma tendência na
escolha de incorporar elementos surrealistas e do realismo mágico nas imagens criadas
por IA? E como podemos determinar o valor cultural da obra de arte, gerada na nova
etapa de industrialização e automação da cultura, através de sistemas computacionais
e da economia de dados que lhes está associada?
Throughout this dissertation, there will be an attempt to understand what type of aesthetic relation exists between generative art and surrealist as well as realist art, including its hybridisms and variants from which I will highlight magical realism. Although there are clear differences between both art types, automatism is present both in surrealist art and in generative art which indicates a need to understand how the creative process works but also, the need to combine ideas, symbols, and unpredictable objects in the same space. The constant search and fascination for the irrational, the uncanny, and the absurd are mainly related to the rupture of the enlightenment ideals and their precedence over rational thoughts as the only way to achieve true knowledge. This fascination has only grown with the evolution of the new technological apparatus and the emergence of the technological arts field during the 20th and 21st centuries, as was the case of photography and cinema. The development of the first edition programs established the first contact between culture and the digital which would take the next step towards the evolution of the software culture and a network society. Originally used as a cataloging tool, artificial intelligence proved from early on to be a great advantage in the classification and categorization of images with the introduction of Convolutional Neural Networks (CNN). These new systems speeded the digitalization process and the access to images and cultural products through the use of online databases. Nowadays, with generative model systems such as Generative Adversarial Networks (GAN) and Artificial Intelligence Creative Adversarial Networks (AICAN), AI has become able to generate or produce variations based on real images. The commercialization and integration of these AI systems in the artistic and cultural production area has seen a clear rise in its user count. This phenomenon is reflected even more with the apparent increase of generative artworks on the several platforms, a popular art style in the 21st century. DALL-E 2 and Midjourney, systems developed by OpenAI, and Stable Diffusion developed by Stability AI, are amongst the list of generative AI models capable of recognizing and cataloging artistic and cultural tendencies, combining them and reproducing them. Lately, we have observed an investment in the application of generative AI systems in image filters in social media, many of which make use of fantastic, surrealist as well as magical elements. This dissertation’s purpose is to, therefore, attempt to clarify why AI systems make use of these magical realism and surrealist elements in their creative process. Are we able to still see AI systems as mere creation tools of cultural products based on variations of what already exists? Is there an actual tendency in the choice of adopting surrealist and magical realist motifs in AI-generated images? And how can we determine the cultural value of a work of art in this new stage of industrialization and automation of culture through computer systems and the data economy associated with them?
Throughout this dissertation, there will be an attempt to understand what type of aesthetic relation exists between generative art and surrealist as well as realist art, including its hybridisms and variants from which I will highlight magical realism. Although there are clear differences between both art types, automatism is present both in surrealist art and in generative art which indicates a need to understand how the creative process works but also, the need to combine ideas, symbols, and unpredictable objects in the same space. The constant search and fascination for the irrational, the uncanny, and the absurd are mainly related to the rupture of the enlightenment ideals and their precedence over rational thoughts as the only way to achieve true knowledge. This fascination has only grown with the evolution of the new technological apparatus and the emergence of the technological arts field during the 20th and 21st centuries, as was the case of photography and cinema. The development of the first edition programs established the first contact between culture and the digital which would take the next step towards the evolution of the software culture and a network society. Originally used as a cataloging tool, artificial intelligence proved from early on to be a great advantage in the classification and categorization of images with the introduction of Convolutional Neural Networks (CNN). These new systems speeded the digitalization process and the access to images and cultural products through the use of online databases. Nowadays, with generative model systems such as Generative Adversarial Networks (GAN) and Artificial Intelligence Creative Adversarial Networks (AICAN), AI has become able to generate or produce variations based on real images. The commercialization and integration of these AI systems in the artistic and cultural production area has seen a clear rise in its user count. This phenomenon is reflected even more with the apparent increase of generative artworks on the several platforms, a popular art style in the 21st century. DALL-E 2 and Midjourney, systems developed by OpenAI, and Stable Diffusion developed by Stability AI, are amongst the list of generative AI models capable of recognizing and cataloging artistic and cultural tendencies, combining them and reproducing them. Lately, we have observed an investment in the application of generative AI systems in image filters in social media, many of which make use of fantastic, surrealist as well as magical elements. This dissertation’s purpose is to, therefore, attempt to clarify why AI systems make use of these magical realism and surrealist elements in their creative process. Are we able to still see AI systems as mere creation tools of cultural products based on variations of what already exists? Is there an actual tendency in the choice of adopting surrealist and magical realist motifs in AI-generated images? And how can we determine the cultural value of a work of art in this new stage of industrialization and automation of culture through computer systems and the data economy associated with them?
Descrição
Palavras-chave
Surrealismo Realismo Mágico Arte Generativa Inteligência Artificial Fantasia Surrealism Magical Realism Generative Art Artificial Intelligence Fantasy