If you haven’t heard of Google’s DeepDream software, then you’ve almost certainly seen images created by it floating around the web. Resembling something between a magic eye picture and a visual hallucination, images created with DeepDream are memorable for their vivid swirling colours and repeating intricate patterns.
DeepDream started with a piece of software designed to detect faces and other patterns within pictures, much like the technology on your phone camera that can recognise who’s who in your group of friends. The software references a huge bank of stock photos which it compares to patterns within the image, thus identifying faces and objects. In DeepDream, the software focuses on these patterns and teases them out, manipulating them to look more and more like objects, faces or animals. Often, the software will pull images from the tiniest random patterns, leading to a picture populated with faces, buildings and cars that appear to have been generated from thin air.
Is CyberArt the Future?
You’ll typically find that the majority of images generated by DeepDream are not particularly interesting, but there are certainly a few that you could imagine putting in a picture frame and hanging on your wall. The few that are really interesting, meanwhile, have got art experts buzzing, with many people questioning whether they could be the future of art.
Two hundred years ago, an artist was a well-educated person who learnt his or her trade over many years, studying with established artists and learning how to draw, paint and sculpt. Today, “artist” is a far broader term, one that can encompass many different skills and that can overlap with the disciplines of architecture, film, dance, music, and poetry. Walk into any modern gallery and you may be confronted with conceptual art, an offshoot of the modern art movement that prioritises concept over traditional artistic skill.
If the genre of conceptual art has opened up the art world to wider interpretations, then should we not consider DeepDream images to be artworks worthy of the walls of galleries? The answer to that question is largely dependent upon whether or not you think that art can be generated by a machine, or whether it is something that necessitates the direct involvement of a human.
For the experts, the answer is simple; Google DeepDream’s images are art, not because of their aesthetic appeal, but because they do what great art should always do; raise questions about humanity and the state of the modern world. Writer and curator Dylan Kerr was recently quoted in The Guardian arguing for the value of AI-generated images, saying “What really distinguishes DeepDream is the machine learning techniques used to generate the images, rather than the images themselves”.
What Kerr illustrates is that DeepDream technology represents another major step towards the development of artificial intelligence that can rival – and in some areas even outstrip – the human brain. Whether or not you enjoy looking at an AI-generated image, you can’t argue that that’s not a concept worthy of discussion.