The History of AI ART that Wikipedia and Journalists Dare not Admit or Write About

It makes me cringe and laugh to see people who have recently joined the AI ART space talk about its history with zero fact-checking based on hype and speculations made by journalists. This is disrespectful not only for the community of artists and scientists but for all information integrity on Twitter and all social media.

After many requests, and with one last piece no longer missing, I am writing this article with plenty of confidence, references, and links. I am writing this article also to introduce myself and the domain in which I have invested most of my passion and time. I am also writing this to mass-media-consumers and relayers having little to no time for research and fact-checking.

I am finally writing this to share my humble prospects for Generative Adversarial Networks (GANs) that I have been researching and applying to art since 2016 that is way before the buzz hit the fan and the art market started commercializing it. I am happy that some talented artists have used GANs and evolved it since then.

AI-ART History Timeline

Contrary to common belief, one of the first auctions of AI art made was using deep dream in 2016 by Gray Area in San Francisco https://www.businessinsider.com/google-ai-images-raise-100000-at-auction-2016-2?r=US&IR=T and it featured Google’s Deep Dream Artworks by Alexander Mordvintsev, Mike Tyka, Memo Atken, Mario Klingemann … I have been following Alexander’s and Mike’s Twitter accounts and been inspired by their encounters.

That was around the same time I had finished my postdoc and joined an AI company in Paris. Deep dreams did not offer much freedom in terms of creativity. I was reading about Generative Adversarial Networks (GANs), and training them on a portion of 1001 portraits I have captured myself. By the end of 2016, I have generated thousands of people that do not exist using GANs in 2016, that is two years after Ian Goodfellow authored the paper about GANs in 2014 that then got accepted at NeurIPS in 2016 and gained more attention. More precisely, I used DCGANs that was published in 2015 and tested on the MNIST dataset in 2016.

First AI-generated people that do not exist, children of the cloud

The exact date of creation around autumn 2016 is not exactly sure, however, one of the latest editions of my children of the cloud was modified around mid-2017. That was around the deadline for submitting my artwork to exhibit at the Grand Palais the following year.

I hold two records about the creation of these historical people that do not exist. One is the file modification date that was on July 30th, 2017 and another is the publication date that happened to be on Aug 10th of 2017 with several Instagram posts on that same day starting with this post https://www.instagram.com/p/BXnWl7OjzuW/ and since I don’t own Instagram, I have taken a screenshot of the post in case it is ever wiped someday.

First published GAN Art created around the end of 2016, modified on 30 july 2017 and publish on 10 august 2017

These dates are very important as they preceded Robie Barrat’s github repository https://github.com/robbiebarrat/art-DCGAN which gained popularity once artnome and others exposed the use of this library by a paris-based collective for an auction by Christies in 2018, but I will get to that later on.

As you can tell from Robbie’s github, although the initial commit was on Aug 9th, it wasn’t until Aug 13th that the package was ready to use with bugs fixed. Whether we refer to the date of published images or their last modification date, we realize that these children of the cloud came into existence before Robbie’s github. Where did they come from?

There was a DCGAN library on github, which was using the MNIST dataset to classify numerals. I believe it was on Nov 2016 that I came across this github https://github.com/rajathkmp/DCGAN and instead of classifying numerals I wanted to generate art.

I was lucky to find it and adapt part of my 1001 faces (pictures of people I have met and snapped) as features that it can train on as illustrated in my

I was a Sr Data Scientist in an AI company in and one night around spring 2017 I was invited to submit an application to exhibit at the Grand Palais. Since the children of the cloud were so low in resolution (210x210 pixels), my idea was to exhibit a mosaic of thousands of these AI-Generated faces of people that do not exist aka “Children of the Cloud” at the Grand Palais for the following year.

Exhibition of AI Art at the Grand Palais on Feb 13 2018

Prior to blockchains and smart contracts like NFTs, it was hard to prove anything except with official contracts and timestamped letters or links to social media and luckily I posted this picture the day of the opening at the Grand Palais https://www.instagram.com/p/BfIlPzjB0Mr/ that was immediately followed by https://www.instagram.com/p/BfVCteGhwF6/ marking a sale in a video that has not made it to Twitter until later on and has been pinned there ever since.

Salon d’automne prix art digital AI ART

Little did I know that my exhibition will find an art collector, an invitation to show at the Salon D’automne for 6 months later and it would win an award on Aug 25, 2018, at the very Salon d’Automne that once featured Picasso and Rodin starting cubism and fauvism. This might have been enough of a distraction from all the news made about AI ART with zero fact-checking or references. I had written to a few journalists and artists such as Robbie, Artnome, Quansimondo, and sent a few tweets none of which did art history justice.

On that same day of Aug 25 of 2018, in New York, Christie’s had auctioned their AI-Generated Artwork that was by a collective based in Paris known as obvious. What is not so obvious still is whether they have used Robbie’s github and training dataset. However, they seized the opportunity with enough media power on their side. The article featuring their auction sale has made news on the front page of every article in many countries. Following is a perspective unexplored by mass media.

How much brainwashing can a scandalous article and its viral spread inflict on the industry and all the artists that have been innovating in this field, you have no idea. How much disrespect to an entire community of researchers and artists could this marketed press release cause to our society?

I wish it ended here…

As a scientist, I have always been skeptical about Wikipedia and any non-peer-reviewed collection of information. However, it wasn’t until an article about AI ART has emerged, that I had first-hand experience with fake news imposed by guardians of specific topics and their biased perspectives with zero fact-checking.

Surely, I have had my account blocked for editing the article and contradicting the interest of the art authorities. You can check all the edit history of the topic.

I am not alone in this since not a single mention of Deep Dream Artists and their Auction made it to the Wikipedia of biased information. Not a single mention of my sale that preceded Christie's or my Children of the Cloud that preceded Robbie’s github. Only Christie’s sale was hailed and praised in the sale’s history! What else would you expect?

The journalists, documentarists, art festival organizers, and museum curators who have been aware of this will have to update their scripts. Those who have not been aware of it will have to write new ones. The speculations of the art market will eventually come to an end. People can be fooled once but not twice and trice. Bloggers and Micro-bloggers will make a huge difference once they start feeling responsible for the biased information and fake news that is being shared around.

I will print this medium article on a t-shirt that I will wear rather than having to introduce myself in the least pleasant of circumstances. As for alternative opportunities for sharing my passion for art and AI, here is one TEDx talk that comes to mind:

Last but not least and to honor the children of the cloud and make sure that their memory lives on, I have dedicated an NFT collection with 1000 tokens as a form of humanity and information integrity and you can check it out now as it is being patiently pushed into the Polygon blockchain https://opensea.io/collection/choc . This is the first AI-generated series until proven otherwise. My scientific skepticism has kept me from announcing it when others were faking it till they make it! I strongly believe that what goes around comes around!

As for the future of GANs ART, come to my exhibition on Feb 16th till Feb 20th at the Grand Palais FMR to find out!

You can surely help by collecting a token, discussing, and sharing this article, of course after doing your own research!

Thank you!

--

--

--

Traditional & AI Artist, Data Scientist, Filmmaker exploring Creative AI. Started the GAN AI Art Movement in 2016. Former Postdoc @CNRS & PhD @INFORMATICS

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

Puppy Dog Eyes for the Smartphone Generation

The Most Underrated AI Achievement of the Past Decade

Why Making Cars Self-Driving Is So Difficult

Arms Race Dynamics of Lethal Autonomous Weapons Systems

Facial recognition dataset and hats

AI and Neuroscience — Part 3: How can deep learning help neuroscience

AI is the new electricity

Designing the Decision-making Center of an Autonomous Vehicle

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
ALAgrApHY

ALAgrApHY

Traditional & AI Artist, Data Scientist, Filmmaker exploring Creative AI. Started the GAN AI Art Movement in 2016. Former Postdoc @CNRS & PhD @INFORMATICS

More from Medium

The Choice of Umoanjah Useshen

Catching a Fish

The Rotation of the Earth Sucks.

History of Fantasy Football