A magazine created by artificial intelligence (AI) and to show the power of programs and platforms powered by this technology. Is about AI-GENa work that is behind Mariola, the artistic collective formed by Cristina Chiarroni and Roquekes.
According to a report published on the website of graphicfor more than a year they began to experiment with artificial intelligence models for the creation of this artistic work.
“The pieces we have made so far have been based on exploring the unique creative potential of these tools, but always as an extension of our perception or human senses, like a digital synesthesia.”, they assure from Mariola
AI-GEN was created entirely by artificial intelligences. Cristina Chiarroni and Roquekes gave more details: “When we say fully we mean that every creative decision and creation of content, images, layout, texts and format have been indicated and created by the different artificial intelligence models”.
How did you create this artificial intelligence magazine?
The report explains that the creative approach consisted of asking the AIs to create content for themselves, trying to find through this genuine cultural elements of artificial intelligences, as well as their own aesthetic, specific humor, topics of interest and even a philosophy.
In addition, the magazine has the concept of artificialismthe artistic movement of artificial intelligences: “Artificialism responds to the main objective of opening new paths towards an artistic expression not yet explored, which can generate new visions and values in art and in the way of appreciating it.”.
“For us, the interest of all this is to explore what these tools can contribute at a creative and artistic level that cannot be created or contributed in any other way, even though that something may be foreign or incomprehensible to us and we start from a fictionality.”, indicated the creators.
“The intention of the project is to be an artistic experiment and an entertainment product, but also informative. Well, in the end, to create the magazine, we have chained together the different models that have been released throughout the year and that were accessible to the public, thus turning the magazine into an annual “record” of the advances in creative potential and technician of these technologies”.
What artificial intelligences did they use?
- To generate the contents (ideas, texts, names, indications of layout, design, etc.): GPT-3, Bloom, GPT-Neo, GPT-J and ChatGPT (Text generation models or NLP).
- To generate the graphic content that accompanies and complements the texts: VQGAN+CLIP, DALL E 2, Midjouney, Stable Diffusion, Craiyon (Text-to-image and image-to-image AI models).
- For the interpretation of the images: Clip Interrogator (Image to text model).
- To increase the resolution of the images: Gigapixel AI (Super-resolution model).
- For the translation of the contents: deepL (Text translation model).
- For the creation of musical pieces: Text-to-Music (Text to music model).
