Creative AI newsletter #7 — AI art updates, NIPS workshop, Impakt Festival
Issue #7 September 1st — October 15th
This is an occasional newsletter with updates on creative applications of artificial intelligence in art, music, design and beyond. Below is issue #7 and you can subscribe for future editions here 😀
Since I last wrote this newsletter, the interest in AI art has exploded with a variety of dedicated festivals, gallery and museum shows. Now Christie’s is set to auction work made by the French collective Obvious, which has sparked debate from established artists in the field and prompted Jason Bailey to investigate further. There has also been recognition — Mario Klingemann scooped the gold Lumen Prize and Forbes interviewed artists from my Machine Dreams exhibition. Artsy wrote on ownership in AI art, Helena Sarin talked about training GANs on her own drawings and Kyle McDonald shared his work on dance and machine learning.
A multitude of new AI-related artworks have appeared: Refik Anadol and Parag Mital projected transformed archival materials onto the Walt Disney Concert Hall, Es Devlin and Ross Goodwin invited you to feed words to a lion sculpture for AI poetry, Anna Ridler generated tulips and Pierre Huyghe reconstructed images from human brain activity to be enjoyed in the company of live flies. Shinseungback Kimyonghun created portraits that confuse facial recognition, Jenna Sutela and Memo Akten attempted interspecies communication, Kate Crawford and Vladan Joler mapped the anatomy of an AI system and Mike Tyka made AI-generated political tweets by imaginary, generated people. Finally, Casey Reas has a GAN sketchbook and Paolo Cirio reveals technology enabling social manipulation.
In literature news, Ross Goodwin published his book 1 the Road, a road trip narrated by a neural network, while the Using Electricity series of computer generated books added three new books to its collection by Li Zilles, Ranjit Bhatnagar and Milton Läufer. Meanwhile, Janelle Shane has written up her tips on how to tell whether a bot is really a human.
Botnik Studios have been busy producing an album with AI generated lyrics. Their efforts to date can be found on their Songularity project website and include a combination of Morrissey with Amazon Reviews and a song in the style of the Strokes. They just successfully closed their Kickstarter campaign, so expect more soon. If you’re looking for AI involvement in the tunes, Bob Sturm has an experimental album of Irish traditional music and computer-generated tunes that you can listen to here. Then there’s also the “first AI-human collaborated album”, Hello World, by music collaborative Skygge.
Google Magenta released a Magenta.js implementation of their piano transcription model, while Tero Parviainen experiments with posing AI-generated drum patterns with your hands. He also has a talk on “Musical Neural Networks in the Browser”.
For some reading, check out Rebecca Fiebrink’s interview on machine learning for audio, Bob Sturm’s advice on the Nottingham Music Database and Amir Saffari’s list of ICLR 2019 Submissions on Generative Models for Music and Audio.
The ECCV Computer Vision Art Gallery, which I co-organised, presents 30+ accepted art submissions representing recent practice in computer vision and AI art. It was part of the first ECCV Workshop on Computer Vision for Fashion, Art and Design and you can find the accepted papers here.
Gene Kogan is back at ITP and this time is recording the lectures from his course on The Neural Aesthetic. Slides and videos from the first four sessions are up already. For your inspiration, Eyeo Festival have uploaded the talks from this year’s event with speakers including David Ha, Janelle Shane, Hannah Davis and more.
New datasets include the Savoias visual complexity dataset and InteriorNetdataset of indoor scenes with 20 million images and 1 million CAD models from furniture manufacturers. If you’re looking for others, Google now has a dataset search.
The ICLR submission that has been making waves in the community is BigGAN, a model that produces high resolution and photorealistic images using lots and lots of TPUs. Browse the images yourself here.
For the world of manga and anime, lllyasviel developed colorization tools: MangaCraft for black-and-white manga and PaintsTransfer for line drawings. Alternatively, you can generate full-body and high resolution anime characters with GANs.
Matthew Guzdial has been working on automated game design via conceptual expansion, resulting in such generated games as “Killer Bounce”, where both level design and rules are learned from scratch. Julian Togelius published Playing Smart, an overview of research on AI and games for non-technical audiences.
I’m organising the second NIPS workshop on Machine Learning for Creativity and Design on 8th December in Montreal. Submit your paper and/or art to our open call until 28th October, and check out last year’s accepted papers and AI art gallery for inspiration. And, if you want to play music or VJ at this year’s NIPS banquet, there’s an open call for that too.
Artists can apply on the European Media Art Platform for residencies at FACT, Ars Electronica, Impakt and other top art and technology instititutions in Europe (deadline 3rd December) and the Re: Humanism contest is looking for art projects on the theme of AI (deadline 20th November). ACM Creativity & Cognition 2019 invites submissions on transformational creativity (deadlines Feb/March) and EvoMUSART on computational intelligence for artistic tasks (deadline 1st November).
Oh and Kaggle has a competition for building a classifier on the Quick,Draw! dataset.
The Meetup 🎨
I’m still planning future events, so if you have speakers or themes to recommend, please let me know. Venue leads would also be helpful, ideally to host around 70–100 people.
In the meantime, enjoy the slides, write-ups and/or periscope recordings from the recent events. Michael Castelle wrote up his July talk on “Social Theory for Generative Networks” as part 1 and part 2. In September, we had Sander Dieleman’s talk on “Generating music in the raw audio domain” (slides) and Natasha Jacques’ “Learning via Social Awareness” (slides / recordings) . In October, look at Eyal Gruss’s talk on ‘Neural Psychedelic Aesthetics’ (slides / recording) and Richard Hames’s “Three Images of the New” (recording).
Cool things to do 😃
Netherlands: I’ve spent the past year curating Impakt Festival 2018: Algorithmic Superstructures, a 5-day event with discussions, workshops and an exhibition themed around AI and post-truth taking place between 24–28 October. While you are there, check out the exhibition Robots Love Music at Museum Speelklok, which traces the uses of robotics to make musical instruments and the Robot Love exhibition in Eindhoven.
Russia: The Moscow Museum of Modern Art has an exhibition Daemons in the Machine, “an artistic re-thinking of artificial intelligence, myths and ghosts of the epoch of autonomous machines” featuring Memo Akten, Egor Kraft, Dmitry Kawarga and more until 11th November.
South Korea: Seoul Mediacity Biennale includes a section on AI art featuring Gene Kogan, Mike Tyka, Mario Klingemann and others as well as Sey Min’s visualisation presenting views on AI, diversity, creativity and the future from all participating artists (including me!). On until 18th November.
UK: Pierre Huyghe’s UUmwelt (AI art with flies!) is on at the Serpentine Gallery until 10th February. Memo Akten will Jennifer Walsh present ULTRACHUNK, an AI and human vocal improvisational performance, on Sat 17th November at Somerset House Studios. There’s also the Artificially Intelligent display on at the V&A until 31st December.
Germany / France: I’m involved in a couple of events in Germany including Art and AI: Shamans of the Digital Renaissance on 18th October in Berlin and Post Binary on 10th November in Frankfurt. If you’re around, come say hi or find me at GROW conference in Paris on 12th November.
Thanks for getting this far. It has been a long one. Anything I missed? Drop me a line if you have any updates I should include or if you need any creative AI-related help.