Last Update: August 2024

Critical Topics: AI Images is an undergraduate class I created and delivered for Bradley University in Spring 2023. Now, it’s something of a beginner’s textbook for critical AI. It’s been used by all kinds of people — from career academics to high school students — to help them think critically about AI images. I’m sharing videos and transcripts of the class lectures — a labor of love and deep research, with lots of dynamic visuals and archival footage. An overview of the emerging context of AI art-making tools, the talks connect media studies and history, AI ethics and critical data studies.

Thank you to Harvard’s and FU Berlin’s metaLAB for recommending this course across their AI Pedagogy Project, which is an excellent resource for instructors teaching critically with or about AI!

If you are using any of these materials in a class, paper, or other way, please let me know. I am happy to hear it, but documenting it helps me out.

I continue to write on these topics on my mailing list, Cybernetic Forests, which is free to read. You can sign up at the button below.

A guest talk from Paul Pangaro of Carnegie Mellon University & President of the American Society for Cybernetics. Pangaro has restored one of Gordon Pask’s artworks, Colloquy of Mobiles, shown at the Cybernetic Serendipities show in 1968. In this talk, Pangaro speaks from his own expertise on cybernetics as we walk through Pask’s body of work connecting conversations and machines.

Contemporary AI is defined by the use of data. But what is data? Where does it come from? We look at the most recent developments in AI imagery over the last decade. We begin with a unique concept of a dataset: the evolution of a system for labeling and organizing colors, from hand-painted scientific folios of the 1800s to contemporary digital color palettes. How does data change our definitions and experiences of the world?

CLASS 5

Images and Surveillance: Nobody is Always Watching You

Some of the same AI technologies and training data used to make AI art tools are also used in surveillance systems, both in the United States and abroad. In this class session, we'll explore some of the social issues around AI images, surveillance, race, and predictive policing. Particularly, AI image datasets of faces as they relate to race and gender in surveillance technologies. 

Adapted from my 2021 keynote talk, From Big Brother to Big Data: Nobody is Always Watching You.

Our second artist talk is with Dr. Anuradha Reddy. Dr. Reddy describes a particular way of making art with AI generated images that reflects some of the conversations we’ve had in our class: using AI to connect to the roots and origins of computing back to Ada Lovelace and the Jacquard loom. The video is embedded alongside s other short artist interview videos from other sources, including Amelia Winger-Bearskin, Refik Anadol, and Sougwen Chung.

Anuradha Reddy (website) is an interdisciplinary design researcher based in Sweden. Her research practice includes interaction design, user research, data technologies, creativity, and hacking. She has a practice-based PhD in Interaction Design from Malmö University, Sweden. By day, she works in the software industry as a design researcher. Outside her day job, she continues researching ways to bridge professional ML/AI expertise with informal communities of makers/hackers through craftivist and critical approaches to design. Her recent design work has been covered in conferences, journals, periodicals and zines. She is an open advocate of FOSS/libre design and collaborative technologies. She holds workshops, talks, and writes for magazines.

We look at Generative Adversarial Networks (GANs) and how they might reinforce — or disrupt — the logic of archives. How do the images generated by these systems tell stories about the underlying datasets? How can artists use this process to tell their own stories? We will also look at the history of GANs and their early reliance on datasets of European oil painting. How do the datasets available to us shape the work we make? How do these datasets reflect cultural contexts of museums and digitization?

We look to the work of Hannah Hoch, Lorna Simpson, Helena Sarin, Mario Klingemann, Gene Kogan, Obvious, and Robbie Barrat.

How does a Diffusion model turn pure noise into an image of flowers in bloom? We discuss Diffusion models, the technology at the heart of DALLE, Stable Diffusion and MidJourney. We’ll explore how Diffusion works and how language models steer images into what you write. Then we’ll think about where the artistry lies in this process: is the AI making the art? Is it dreaming or imagining these images? We’ll look at John Searles’ “Chinese Room” thought experiment to think through those questions. Finally, we look at whether AI art is a radical shift in art making, or serves to extend a 60-year-old history of Generative Art from computers.

Artists in today’s lecture include Mezei Leslie, Georg Nees, Robert Mueller and Tim Klein and the 1972 Computer Art exhibition organized by Laxmi Sihara in New Delhi.

A deeper look at the actual data that was used to build open source diffusion models. We’ll examine the logic of image datasets, noting that composite photography, statistical correlations and eugenics were all created by the same man — the British sociologist Francis Galton. Today’s image generators categorize and represent people based on the same principles. How should artists respond to that? What lurks behind seemingly innocuous prompts such as “brave person” or “black girl?” We also offer a simple introduction to the idea of weights and biases in AI systems.

We’ve all heard this expression: Great artists steal. But the whole quote is more complex, and so is the tradition of appropriation in art. We’ll look at appropriation through the lens of power. Several lawsuits around image synthesis suggest that making the work could be seen as derivative; others suggests the models themselves are a copyright violation. We compare image synthesis tools & artistry from the early days of music synthesis to today’s visual synthesis to find insight.

Artists mentioned in today’s lecture include Heather Dewey-Hagborg, René Magritte, Greg Rutkowski, Henri Pousseur, Delia Derbyshire, Robert Rauschenberg, Sherry Levine, Walker Evans, Marcel Duchamp, Patricia Caulfield, Andy Warhol, Elaine Sturtevant, Agnieszka Kurant, and James Bridle.

How do you explore an image dataset? In this class, I’ll do a live dataset dissection. You’ll learn how to find image datasets, how to open them up, and what to look for when you do. We’ll think about how to decide whether datasets are fit for purpose or not.

In the bonus talk, I’ll discuss my own practice of collecting datasets for making art with GANs. I’ll connect the dots to my work in photography and collage, but also discuss what I learned by shifting my photographic eye from traditional photography to building a dataset out in the world with my own camera. That’s the Class 11 Bonus Talk option.

Artist Talk

Caroline Sinders

Button links to our interview with Caroline Sinders on YouTube.

Caroline Sinders (website) is a critical designer and artist. For the past few years, she has been examining the intersections of artificial intelligence, intersectional justice, systems design, harm, and politics in digital conversational spaces and technology platforms. She has worked with the United Nations, Amnesty International, IBM Watson, the Wikimedia Foundation, and others. Sinders has held fellowships with the Harvard Kennedy School, Google’s PAIR (People and Artificial Intelligence Research group), Ars Electronica’s AI Lab, the Weizenbaum Institute, the Mozilla Foundation, Pioneer Works, Eyebeam, Ars Electronica, the Yerba Buena Center for the Arts, the Sci Art Resonances program with the European Commission, and the International Center of Photography. Her work has been featured in the Tate Exchange in Tate Modern, Victoria and Albert Museum, MoMA PS1, LABoral, Wired, Slate, Quartz, the Channels Festival, and others. Sinders holds a Masters from New York University’s Interactive Telecommunications Program.

We discuss datasets and how they're assembled, and how they are "seen" (or not seen). Human labor is behind even the most fundamental technologies we describe as "automated," including the datasets we're looking at. That includes the workers hidden away behind interfaces and content moderation systems (thanks to Sarah T. Roberts for the title and the reading!). For artists specifically, we look to the role of automation and the labor it replaces, to the humans behind the art we treat as data, to the question of where the human fits into AI creativity at the copyright office.

Dr Eleanor Dare is an academic and critical technologist who works with Game Engines and virtual spaces, who now works at Cambridge University, Faculty of Education, as well as UCL, institute of Education. Dr. Dare has a PhD and MSc in Arts and Computational Technologies from the department of Computing, Goldsmiths, and has exhibited work at galleries and festivals around the world. Dr. Dare was formerly Reader in Digital Media and Head of Programme for MA Digital Direction, at the Royal College of Art.

The button will open a link to this conversation directly on YouTube.

At their heart, movies are images — in sequence, fast enough to trick our eye into seeing motion. With synthetic video on the rise, what does AI make possible for this form of storytelling? We look at the history of cinema without cameras, touching on the writers Vilem Flusser and Gene Youngblood. We look to how artists create “motion pictures” beyond the limits of frames on film, using digital tools to shape stories — and suggest ways of seeing the world — that traditional cinema can’t. This is only a brief introduction to a genre with a rich, vibrant history and contemporary practice.

Artists whose works are shown or referenced include Lee Harrison, Nikolai Konstantinov, Deniz Kurt, Refik Anadol, Sofia Crespo, Steina & Woody Vasulka, Vadim Epstein, Ed Emshwiller, Memo Akten, Ed Catmill & Fred Park, and Nam June Paik.

Utilizing cutting edge machine learning technology, Derrick Schultz's work explores multisensory perception, generative abstraction, and the future of ecology. In addition to creating his own work, Derrick also teaches machine learning to artists, designers, and image makers.

On a personal note, Derrick’s online courses on generative AI for artists are a delight for those of you looking for to expand your technical skills.

Coca-Cola was one of the first corporations to make a foray into the world of generative AI for a marketing campaign. We’ll look at the history of media, marketing and advertisements to understand how and why they work, and how we can make informed decisions as consumers about the messages that pay to be there. Are AI systems advertising themselves to us through their interfaces and the myths of their design? How can we see through our own imaginations and treat these systems critically?

We also ask how biased generated content normalizes certain expectations about the world through their content.

Born in Caracas, Venezuela, Moises Sanabria is an artist interested in technology, internet culture and contemporary branding. He is one of the co-founders of the new media collective Art404 (Art Not Found), whose works often deal with legitimacy, value, and perception. Art404 often uses the online world as a medium for creating works, including projects involving Photoshop, the hacker group Anonymous, and the Sims.

Art404 has shown at Transmediale 2k+12 in Berlin, Gucci Vuitton in Miami, Ars Electronica Festival in Austria & Conflux Festival in NYC. Moises is currently attending the Cooper Union for the Advancement of Science and Art, where he is focusing on Interactive 3D graphics and physical computing. He lives and works in New York City.

CLASS 15

The Opposite of Information

What is misinformation, and what is the role that AI plays? In this talk, we explore computer-edited photographs going back to 2006, the controversies around Photoshop, and ask what information even is. We’ll look at recent AI-generated deepfakes, how they circulate, and what their true impact is on the nature of documentary evidence and democratic discourse.

This lecture was written for the Elisava Barcelona Responsible AI Master’s course in October 2023 and was added to this online syllabus in August 2024. There is no video.

Vladimir Alexeev (Merzmensch), born in 1979 in Moscow, living in Germany, speaking Japanese is an international writer, artist, and researcher, who explores areas of tension between Historical Avant-Garde (especially Dadaism) and AI art. In his art and essays, he investigates the horizons and possibilities of human-machine creative collaborations. Harper's Magazine describes him as Data Journalist. In his artworks, he uses mixed media, his own poetry, and photography, but also advanced AI-driven models and approaches.

The pseudonym Merzmensch is coming from MERZ-Art by Kurt Schwitters, who is the big inspiration and spiritual mentor for Vladimir Alexeev. In his works, Merzmensch is trying to follow Schwitters' path to create new realms and realities from mixed media, where every element has its significance and is equal to others.

This course has pulled apart the threads of AI image synthesis. In this final review session, we reconnect them, systematically, through four lenses: 1) Data, 2) Interface, 3) Images, 4) Culture and Society. For each layer of this system, we rethink the ethical implications they pose.

We conclude with a challenge to the idea that machines are inspired, or see the world, or make art, in the same way people do. If they don’t, then what are they doing? And more importantly, what might we do with them?

This is an artist talk held with Soyun Park as part of my role in the AIxDesign Story & Code Program.

Soyun Park is an interdisciplinary artist, designer and educator from South Korea living in The Hague, The Netherlands. She’s also a founder of a community-based studio for bonding technology, RGBdog. She’s fascinated by technology’s historical, emotional, literary, comical, filmic, fictional, humane, societal, political, gamified, capitalised, decentralised, and solidarity aspects. She often examines ‘tools’ from different perspectives and feels the joy of exploring them in connection to a larger context, often with humour. Often in collaborations, her work takes the forms of videos, installations, and audiovisual performances experimenting variety of new media and technology, investigating connection and the gap between the RGB world and reality which is getting thinner every day. She has exhibited and performed at media art festivals, cultural venues and film festivals including for example Rewire Festival (NL), Nederland Fotomuseum (NL), Ars Electronica (AT), CTM Festival (DE), Jeonju International Film Festival (KR) and iii workspace (NL).

A longer essay exploring the role of Flickr photographs in building AI infrastructure, and how archives “haunt” the output of diffusion models. Written as a Flickr Foundation Research Fellow in Spring 2024.

Eryk Salvaggio: Additional Talks & Lectures

Reach out!