︎

A moodboard to illustrate AI 


While generative machine learning models have created a horizon of uncertainty for illustrators, and the companies operating them have unilaterally decided that creative work should be “improved” and made more efficient and less costly, illustrators are systematically framed as passive actors of this fate. We just need to wait and see how this inevitable wave will change our lives. 

I suggest here that illustrators have a specific power to claim as their own : that of creating representations. We’ve seen many blueish robots and countless ethereal digital creatures used as metaphors for machine learning and its impact on social life. Yet, machine learning is first and foremost applied statistic, human labour, and material resources. Illustrators are at the core of how things look like, and the visual language the rest of the population uses to visualize an issue often stems from the visual labour of illustrators. This means that by shifting slightly how we represent AI in the articles we’re hire to illustrate, we might slowly shift its public representations and open the way for more critical engagement with it. What if instead of blueish, futuristic tones, AI illustrations were using the pinks, greens  and yellows from lithium mines used to power computers, or the muddy grey of data centres consuming millions of gallons of water? What if instead of abstract robots AI was personified in the global south workers annotating data, or the obscure spreadsheets of datasets? 

Using another technology of decontextualization and objectification of illustrators’ work, the moodboard, I propose here a growing repertoire of inspirations for illustrators and art directors to draw from when reprensenting machine learning becomes necessary in their work. This project echoes other initiatives directed at a wider audience like Philipp Schmitt’s Blueprint For Intelligence or the Better Images for AI project, but is meant to be a working tools specifically for commercial artists. 


Comunidad de Software Libre Hackem [Research Group], CC BY-SA 3.0 , via Wikimedia Commons ShantoShahriar, CC BY-SA 4.0 , via Wikimedia Commons Doc Searls - The Chemetall Foote Lithium Operation in Clayton Valley, a dry lake bed in Esmeralda County, Nevada پیمان مجیدی معین, CC BY-SA 4.0 , via Wikimedia Commons Sergei Antonov, Alexei Antonov, Public domain, via Wikimedia Commons Swilsonmc, CC BY-SA 3.0 , via Wikimedia Commons Langs, G., Röhrich, S., Hofmanninger, J. et al., CC BY 4.0 , via Wikimedia Commons Joseph Racknitz, Public domain, via Wikimedia Commons Abyssus, CC BY-SA 4.0 , via Wikimedia Commons Ioanna Giorgi, CC BY-SA 4.0 , via Wikimedia Commons Nacho Kamenov and Humans in the Loop, CC BY 4.0 , via Wikimedia Commons Joseph Racknitz, Public domain, via Wikimedia Commons Nicoguaro, CC BY 4.0 , via Wikimedia Commons Exploring Randomly Wired Neural Networks for Image Recognition Saining Xie, Alexander Kirillov, Ross Girshick, Kaiming He; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 1284-1293 Dr assem Badr, CC BY-SA 4.0 , via Wikimedia Commons Kai3952, CC BY-SA 4.0 , via Wikimedia Commons Pro8055, CC BY-SA 4.0 , via Wikimedia Commons Tmy350, CC BY-SA 4.0 , via Wikimedia Commons U.S. Government Accountability Office from Washington, DC, United States, Public domain, via Wikimedia Commons Nacho Kamenov and Humans in the Loop, CC BY 4.0 , via Wikimedia Commons Vicente Oyanedel M., CC BY-SA 4.0 , via Wikimedia Commons Pete Woodhead, CC BY 2.0 , via Wikimedia Commons NASA Earth Observatory, Operational Land Imager (OLI) by Landsat 8, Public domain, via Wikimedia Commons Vicente Oyanedel M., CC BY-SA 4.0 , via Wikimedia Commons Visitor7, CC BY-SA 3.0 , via Wikimedia Commons 123net, CC BY-SA 3.0 , via Wikimedia Commons