The New Wave of Digital Art. Uberization?
An inspiring paper in the today's issue of Makery: AI, top ten of prototypes to be keen on (Intelligence artificielle: top 10 des protos à suivre) by Nicolas Barrial. Is it a threat for the "classical" digital art, based on algorithms and random functions?
Let us summarize the first paragraphs. With the deep learning, Ai prides itself on seeing, classifying and even generate images. See it transform a horse into a zebra, a sketch into an Rembrandt and even design imaginary cities... Machines compare and recognise representations (images, texts, videos, sounds) by massively perceiving.
For instance, in 2014, Ian Goodfellow, a leader scientist at Google Brain put into competition generativ e and discriminative models, through Generative Adversial Networks (GAN).*
Then the paper presents 10 projects or pieces:
- Neural City (2017) by Jasper van Loenen,
- Autodraw (2017) by Google,
- Image-to-Image (2017) by Christopher Hesse,
- CycleGAN (2017) by Jun-Yan Zhu and Taesung Parke,
- PredNet (2016) by three scientists in Caltech (California Institute of Technology),
- Invisibles Cities (2016) designed during a workshop at the Milano Open Dot fablab,
- Neural Doodles (2016) by Alex Champandard,
- Deep Dream (2015) by Google,
- GAN (2014)by Ian Goodfellow,
- Ink Poster (2013) by Alex Grave.
A threat to "classical" digital art?
These news confirm ours 2016 intuitions about art and the new AI (see our Roxame's Interim Report). The classical digital art has two main currents:
- The Hollywood-LA mainstreamn with animation cinema (and its Siggrah's big show) and matte painting (Lacda)
- The algorithms*random way, mostly represented in USA and Europe since the 1970's, from Harold Cohen and Schoeffer to the mainly European artists Diccan deals today.
In both cases, progress along the years lies mainly in higher resolution and color palettes, and of course more powerful algorithms due to more powerful machines (even server farms for the mainstreamp.
In both cases also, the market remains active (on quite different scales, of course), as show for instance the sales of digital art in the great exhibits and specialized galleries. And it will probably remain stable, since by its mere nature, well established signatures keep or increase their values with time.
But the new wave (deep learning, big data, cloud) will perhaps bring a quite different kind of not only markets but also practices and sociology. At last for "art" in the traditional sense. Perhaps a kind of art "uberization", letting aside the well greased channels of shows and galleries, selling one-man artistic pieces to fortunate collectors.
Instead, the superplatforms (Gafa, to make it short) brings both the tools and the sales "force", and offers to anybody, if properly gifted and motivated,
- the use sophisticated software and hardware (3D printing, why not virtual reality...)
- the sales place for the pieces through the net.
The years (even perhaps the months) to come will be fascinated to observe, if not to take part in the creation itself. Anyway, journalism itself is an art, and is also undergoing deep changes through social networks but also with robotics.
Pierre Berger. April 26, 2017.