Creative AI newsletter #8 — Auctions, continued hype and AI art overload
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 😀
Art 🎨
2019 is now in full swing and AI art is as trendy as ever, though there are signs of normalisation. Last week saw the Sotheby’s auction of Mario Klingemann’sMemory of Passersby I for £40,000, which met its estimate but not the price-tag of the controversial Obvious piece, sold at Christie’s last October for $432,500. Lawyers are still deliberating the legal challenges of AI-generated works in the UK and beyond, philosophers are arguing against considering AI as an artist, researchers are exploring autonomy, authenticity, authorship and intention and journalists are attempting their own AI art. There’s even a gallery owner who created the world’s first humanoid AI artist called Ai-Da.
A selection of last year’s AI artworks can be found in the NeurIPS Gallery aiartonline.com. New artworks include the Barrot / Barrat skulls collaboration; Hannah Davis’s Laughing Room which found these lines the funniest; oil paintings based on AI-generated landscapes by Chris Peters; Trevor Paglen’s classifier ImageNet Roulette; a simulation of an automated world in Stark Choices from Superflux; Madeline Gannon’s Manus with autonomous robots behaving as a pack; beautiful abstractions in Origami from Alex Mordvintsev; Marta Revuelta’s AI Facial Profiling, Levels of Paranoia which judges who is allowed to handle firearms based on a biometric analysis of the face; Ultimate Emoji by Albert Barqué-Duran; Daniel Ambrosi with even more detailed Grand Scale Dreamscapes; cloud-based AI Auria that draws and writes poems; Gillian Wearing’s ad campaign Wearing,Gillian.
Some in-depth interviews with Daniel Ambrosi, Ian Cheng, Patrick Tresset, Madeline Gannon, Mario Klingemann and write-ups on AICAN (here in a more academic setting), Helena Sarin’s Neural Bricolage, Egor Kraft’s Content Aware Studies and Caroline Sinders’ Feminist Data Set.
The past few months have also seen the launch of a few AI arts related initiatives, including Goldsmiths’ Museums and AI network, British Library’s Living with Machines and Quartz AI’s portal for journalists. Dancewise, Qosmo collaborated with flamenco dancer Israel Galván and Google with British choreographer Wayne McGregor.
Text 📖
Following the release of OpenAI’s GPT-2 , text generation has swanned back into fashion thanks to the much improved results, with everyone from journalists and charity workers experimenting with the new tool, including our usual suspects Janelle Shane, Kyle McDonald and Mario Klingemann. The quality of GPT-2 generated text has inspired GLTR, a tool to detect automatically generated text.
Here’s a list of games that use natural language generation techniques and a list of NaNoGenMo projects. If you enjoyed the latter, Golan Levin’s students made a book of computer-generated chapters. You can also choose-your-own-adventure in Botnik’s Goosebumps or read some Japanese AI poetry. James Ryan finally published his PhD thesis on emergent narrative and you can read it here.
Experimentation with TED talks strikes again, via Kean Patti and Alex Reben. For refererence, here is Samim’s original experiment. And here’s another ad made with AI, this one attempts to differentiate itself as the “first commercial scripted by artificial intelligence”. See for yourself (or maybe not).
Music🎵
Music with GANs! Jesse Engel from Magenta published GANsynth (online supplement with examples here). Max Frenzel made a track out of samples generated by neural networks, Clara uses an LSTM to compose piano and chamber music and Aleksey Tikhonov made an exploration tool of drum patterns. There’s even an AI now that autocorrects your karaoke singing and isolates vocals from stereo music.
Last October while curating Impakt Festival I organised a Robot Folk Jamsession at Museum Speelklok in Utrecht, where Shimon the Robot played alongside a folk band. Watch it! In other news, Holly Herndon and Matt Dryhurst have been working with their AI called Spawn for their new album Proto, due to launch on 10th May. ‘Godmother’ with machine learning. Huawei has finished Schubert’s Unfinished Symphony.
Hannah Fry wrote about valuing art made with algorithms, with particular reference to music and the musician Nick Cave responded to a fan’s question about whether AI will ever be able to write a good song. OpenMind has a lengthy piece about computational creativity, with a decent overview of music. MusicAlly provides a snapshot of the startups and commercial projects in the space, while Technosphere magazine has a whole issue dedicated to Machine Listening.
Design 👗
In the world of design, we have seen projects designing floorplans and chairs from leftovers as well as those which combine Nordic design, Chinese manufacturing and machine learning. Chenoe Hart came up with new configurations of portable electronic devices based on those from the 90s/00s and Sarah Friend focussed on digital interfaces through the eye of the machine. Both Google Arts and Culture and Shi Weili found inspiration in satellite imagery — one resulting in NASA’s Visual Universe and the other in a generated Terra Mars.
Robbie Barrat spoke to SSENSE about his Balenciaga AI project, Kate Rose designed a dress to introduce noise in hostile datasets, Victor Dibia generated African masks and string&&loop has been selling QuickDraw! socks. IBM is still working on the AI fragrance. And if knitting’s your thing, there’s a paper on Neural Inverse Knitting, which translates images into manufacturing instructions.
Actors are digitally preserving themselves, clearly to make the most of scenarios like this one.
Resources 📓
Datasets galore! 70,000 images of faces from FFHQ, 50,000 collection images from the The Art Institute of Chicago, 50,000 detailed 3D buildings from Zurich, 50,000 hotels, 50,000 graphic design images from the Letterform Archive, 30,000 additional digitsed works from the Cleveland Museum of Art ,11,000 books from 1923 released into the public domain and there’s 4 years of keylogging data. Kyle McDonald made a face labelling dataset list and some face-related releases include Disguised Faces in the Wild, Google’s Open Images extended and IBM Diversity in Faces dataset, which has already come under fire. There’s also an article about Lena, one of the famous test images.
GROW Paris have published their talk recordings including mine, Code Ecologies recorded their live-stream here and ProcJam has theirs up here. You can also find their Seeds Issue 3 with many contributions on procedural generation and then a Connection Science issue on computational creativity,
Like a good newsletter writer, I should occasionally remind you about my work. The videos from the Impakt Festival I curated are now online, including for the two panels I moderated on AI art and Financial Speculation. There’s also podcasts from my Retune Festival Creative AI panel, Serpentine Gallerypodcast on work and the odd time The Guardian interviewed me for Science Weekly. If you missed my NeurIPS Workshop on Machine Learning for Creativity and Design, you can find the accepted papers and the AI art gallery. Keunwoo Choi did a write-up too so I don’t have to!
Liam Young published a book Machine Landscapes: Architectures of the Post Anthropocene, Lev Manovich on AI Aesthetics, Julian Togelius on Playing Smart and Nora Khan on Seeing, Naming, Knowing. Here’s an analysis of imagery used to represent AI and IBM’s attempt to depict AI.
Google Magenta have released Magenta Studio, a collection of music creativity tools built on Magenta’s open source models, Arts and Culture shared the code for Art Palette and Creatibility is full of accessible creative tools.
Research 📋
StyleGAN came out and the internet is rife with fake people, rental appartments, waifus, cats and even resumes. To test your skills, there’s a game where you check which person is real and Kyle McDonald even has a guide on how to recognise fake images. We need it as the GAN progress has been astounding. Before StyleGAN, BigGAN dominated the twitter feed and Joel Simon’s Ganbreeder is still there if you want to create weird images. Gray Crawford build a system that lets you maneouvre BigGAN in real time, while Alex Reben connected Ganbreeder to body sensors to make Amalgan.
CariGAN makes caricatures out of photos, Comixify turns videos into comics and Photo Wake-Up 3D animates subjects from a single photo. GANwise, you can now edit faces based on user input and colour with SC-FEGAN and paint objects with GANpaint. Spatially-consistent Feature Learning has been applied to discover visual patterns in art collections and neural networks helped spectral reproduction of paintings. Here’s also a simple guide to using CPPNs to create art.
Opportunities 🚀
Residencies: Google Arts and Culture have teamed up with Jacquard for a residency exploring synergies between art, technology and fashion, apply until 1st April. Waag’s AI Lab residency is on understanding living materiality in outer space with use of AI and Djerassi’s Scientific Delirium Residency explores how the creativity of scientists and artists is connected (apply for both until 15th March!). NEW INC is looking for people at the interesction of art, design and technology until 20th April and Eyebeam Centre for the Future of Journalism awards grants for journalistic projects including using AI. Retreats for women working on open-source projects, research and art.
Prizes: The Lumen Prize for digital art is open for submissions until 3rd May and this year I’m on the selection committee for its BCS AI Award. NTAA plans to award a prize of €5,000 to an art project using technology in a new way. Apply until 31st March.
Open Calls: Musical MetaCreation (MUME) held at ICCC is seeking submissions around music and computational creativity until 24th March.
Academia: Monash University is looking for a Lecturer in Creative AI and also has a PhD Scholarship in the Emerging Technologies Lab. EPRSC Edinburgh has a studentship on intelligibility through design in AI and data-driven media and DMU Institute of Creative Technologies invites applications for a PhD scholarship until 26th March.
The Meetup🎨
The meetup group is no more, but you can sign up to be notified of future events here. Slides from last year’s talks ‘Three Images of the New’ by Richard Hames (October), ‘Neural Psychedelic Aesthetics’ by Eyal Gruss (October) and ‘Seen by Machine: Computational Spectatorship and the BBC Archive’ by Daniel Chávez Heras (November). Periscope recordings of the October (Richard Hames / Eyal Gruss) and November (Libby Heaney / Daniel Chavez Heras). Next meetup will be on 28th March at Somerset House with Scott Eaton and Andrew Brock, registrations should open on Thursday 14th so keep an eye on your emails!
Things to do 😃
London: The next Zabludowicz Collection Invites is dedicated to Jake Elwes, who will be showing a new work which explores the intersection between nature and machine, opening on the 21st March until 28th April. Watermans is showing Nye Thompson’s self-surveillance themed SKRBT from 23rd March until 26th May.
Oxford: AI-Da Robot will showcase her drawings and performance art in her first exhibition Unsecured Futures at Lady Margaret Hall and St John’s College in Oxford. 9th May-31st May.
Switzerland: I’ve curated an exhibition of AI art by women for the Women in Data Science Conference in Zurich, running between 4th April — 30th April at Swiss Re in Rüschlikon. HeK Basel opens a group exhibition of AI art titled Entangled Realities on the 9th May until 11th August.
Germany: Berlin hosts Egor Kraft’s Content Aware Studies at Alexander Levy(until 13th April) and Compressed Cinema by Casey Reas at DAM Gallery(23rd March — 4th May). There’s a workshop around AI and creative processes on 22/23 March at KreativBund’s Gestaltungsmachine in Berlin and the EvoMusart conference takes place in Leipzig between 24–26 April. Hannover Kunstverein is putting on Artistic Intelligence from 27th April until 30th June.
US: The Algorithmic Art Assembly, featuring artists using algorithms for aesthetics, takes place at the Gray Area on the 22/23 March. In New York between 8–12 May, you can see a performance of Discrete Figures by Rhizomatiks Research x ELEVENPLAY x Kyle McDonald. Refik Anadol’s Infinity Room is on view at Wood Street Galleries in Pitsburgh until 4th July.
Elsewhere: Rome sees the MAXXI Low Form exhibition on imaginaries and visions in the age of AI extended until 24th March. UCCA Beijing hosts an exhibition of Qiu Zhijie’s work, which includes an AI-based piece that generates maps in the artist’s style based on certain words, while Ujazdowski Castle Centre for Contemporary Art in Poland presents Janek Simon’s Synthetic Folklore until 5th May, featuring mosaics with universal folk patters generated by AI.
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.
