Creating works of art was until today an innate quality to the human being, and that few living beings shared. But the rules have changed. A team of developers designs an artificial intelligence capable of creating their own images.
The evolution of technology in areas such as artificial intelligence it develops in increasingly varied fields. So far, we had seen some software capable of writing a novel, playing chess, or having conversations with some consistency. But the creation of images seemed to be related to the emotional capacity of human beings. It takes feeling, living and experimenting to create visual artworks. However, a team of Microsoft developers wants to prove otherwise.
The Deep Learning Group they’ve been in charge of the project. Nail artificial intelligence which is based on attentional Generative Adversarial Network (AttnGAN). Basically, it comes to be a model centered on the neural system that we have living beings. This neural network i would learn from experience, performing several processes simultaneously that work together.
A logical process to generate an image
A detail to highlight is that the aforementioned artificial intelligence has no imagination of its own. It needs to be provided with a text previously used as the basis for creating the image. According to the work presented by the team, the AI extracts key information by performing several readings to go refining the image generated by pixels. With each scan of the description provided to you, the image gains in quality and definition. Very similar to an artist looking for inspiration in a poem, or an illustrator to a novel.
However, this artificial intelligence approach that is able to design its own content, even through a text, enters a swampy legal field. The theme of authorship, always related to the creating a new building. When we see in the news that an elephant paints and sells his own works, the fact of belonging to the work comes into debate. A complex fact that often gives justice headaches. A case similar to the one we find with this new development: a non-human being creating his own works.
The biggest drawback is that the machine in question it is not able to design objects or invent, as it is based on already existing images. Detail that does not detract from the importance of what has been achieved by this team of developers. The implications can mean a change in the rules of the game. Above all, taking into account the visual need of our contemporary society. Image banks can benefit of this development. The advertising, marketing or design companies they would rely on this software as do architecture studios in 3D software to raise their works. And in the social section everything opens a range of possibilities for those with difficulties. We only have to wait and see how our newborn artist evolves.