In February Levi Sherman and I co-chaired a CBAA-sponsored session at the 114th College Art Association annual conference in Chicago. We have been co-chairing these sessions for the past three years to bring greater visibility to CBAA and the artist book as both aesthetic object and practice. This year our focus was “Rethinking the Artist’s Book in the Digital Age: AI, Authorship, and Authenticity.” We sought papers that addressed how “digital technologies, AI, and materiality have transformed how we create, consume, and understand art and information. In this digital age, the artist book remains a tactile, immersive medium, emphasizing the importance of haptic engagement. Positioned between tradition and innovation, artist books invite reflection on authenticity, authorship, and material presence.” We were casting a wide net to see what we could receive in terms of the connection between AI and the artist book.
Our session had three excellent presenters who demonstrated the complex frameworks that exist in the creation and function of artist books. This included 1) how artist books critically expose and reinterpret AI systems; 2) how materiality and computation are historically intertwined; 3) how authorship has long been procedural and distributed (or shared); 4) how AI extends older computational and conceptual traditions; 5) how materiality becomes political, epistemological, and resistant; as well as 6) how artist books reveal forms of meaning that AI cannot fully capture.[1]
While reflecting upon the outcome of our session I began to look into other sessions that I either attended or that had some tie to the use of AI and fine art.[2] There were 5 other sessions that addressed issues surrounding AI. Some of the key issues they addressed were:
AI exploits artists' labor, but it also expands creativity and experimentation.
It undermines originality, but increases artistic accessibility.
It threatens artistic livelihood, but supports collaboration.
It reinforces corporate power, and yet enhances education and research.
It encourages homogenization of expression, and yet produces new artistic forms.
It weakens material and embodied art forms, but ironically enables interdisciplinary practices.
It embeds bias and extraction, yet democratizes the tools of creation.
Among all these points and counterpoints, there is no clear trend towards either techno-optimism nor absolute luddism. The continuing trend is toward critical engagement, ethical accountability, protection of creative labor, and reassessing intelligence, authorship and creativity in human-centered ways.
As we train ourselves in the use of AI, we simultaneously train AI. We do so by providing it with metadata that it uses to search and produce synthesized results. These synthesized results are significant, but they are almost always flawed in some way. When junk or incomplete data is entered, that is all that can be accessed or provided in return by searches.
An AI cannot parse and think through texts and implications the way a historian does. Thus, with the information available the AI may answer in an intelligent manner that appears logical and reasonably correct. This makes it easy to “mistake clear reasoning for correct reasoning.” [3] Thus, its answers will be limited and partially correct and false at the same time. Artificial intelligence creates a blended artificial response drawn from a limited library's minor and incomplete collections.
One thing that I have noticed, and which usually comes up at the beginning of most of these sessions or presentations, was a qualification of what version of AI was used to create images or parse data. This leads to the question of how many different types of AI are there that one can possibly use? This is one of those questions that doesn’t really have a precise number available. That is because AI continues to proliferate in all manner of directions, much as Guattari and Deleuze’s concept of rhizomatic networks.
Various types of AI exist and can only really be categorized by their different capabilities, functions and learning approaches.[4]
In terms of capability there are narrow or weak AI’s that are designed for specific tasks such as voice assistants, recommendation systems, chatbots, and self-driving features. Most AI’s around us are of this type. Then there are hypothetical AI systems such as general AI, which are systems that can learn and perform any task that can be done intellectually by a human. This does not fully exist yet. There are potentially also superintelligent AI, a theoretical system which would surpass human intelligence in perhaps every field. This type is still science fiction.
In terms of functionality there are reactive machines that have no memory and only react to whatever is the current input, like IBM’s deep blue chess computer. There is limited memory AI that uses past data, temporarily, to make decisions. This is like self-driving cars and modern chatbots. Then there is also theory of mind AI that could understand emotions, beliefs and intentions. This is still experimental. Finally, there is self-aware AI, that could hypothetically have consciousness or self-awareness. This still does not exist.
AI can also be distinguished by its learning method. There is machine learning, which learns from data patterns. There is also deep learning, which uses neural networks with many layers. This powers image recognition and modern language models. There is also reinforcement learning, which learns through rewards and penalties, used in robotics and game-playing AI. Finally, there is generative AI, which can create new content such as text, images, music or even videos. This includes ChatGPT and image generators.
However, even as AI is parsed into these categories, each one is not distinct because most AI systems combine these categories in different ways. Thus, there are at least dozens to hundreds of AI subtypes. Each type represents a different means with which to enter into a dialogue that can be simultaneously derivative and original, collaborative and threatening, at the same time reinforcing corporate power and enhancing education and research.
In short, I must observer that, as with most things, its complicated. However, what we do with that complication is just as subjective as anything that an AI can produce. It is about as relevant as how we receive any new technology, any new idea, any thesis or dissertation in its relevance or irrelevance.
So, what are we worried about?
What should we be worried about?
What do you think?
Let us know in the comments.
Peter J. Tanner (He, Él, Ele) is Associate Instructor of Spanish at the University of Utah and Editor of Openings: Studies in Book Art, the journal of the College Book Art Association. His research focuses on artist books from Latin America.
[1] These observations were made regarding the materials presented in from the following presentations in our CBAA session: “Material Algorithms: Early Artists’ Books and Computational Practice,” by Regine Ehleiter; “The Artist's Book as a Site for Digital Ethnography”, by Clara Davis; and “Can Artists Books Problematize “Natural Language Processing”? An Exhibition Case Study,” by Xinyue Lulu Yuan.
[2] These observations come from the following sessions with the following presenters:
The Politics of AI as a Medium of Art by the Services to Artists Committee: Brooks Cashbaugh; Mr. Abhishek Narula; Matthew Magill; Morris Fox; Chanhee Choi.
genAI & Copyright: Proactive Protection for Artists: Laura Smith.
What Difference Does AI Make?: Mark J.V. Olson, Julia McHugh and Julianne Miao; Nicola Carboni; Emily A. Pugh; Alison Langmead.
Creativity, Collaboration, and Critique in the Age of AI and Immersive Technologies: Dr. Danilo Ljubomir Bojić; Haiver; Daria Tsoupikova; Hyeyoung Maeng.
Mobilizing Models: How Artistic Intelligence Reprograms AI: Grace Han; Gerui Wang; Miguel Novelo; Barbara Rauch.
[3] Hayt, “Chapter 20,” Frank Herbert, Dune Messiah, 1969.
[4] Primary sources consulted in this section:
Russell, Stuart J., and Peter Norvig. Artificial Intelligence: A Modern Approach. 4th ed. Hoboken, NJ: Pearson, 2021.
Copeland, B. Jack. "Artificial Intelligence." Stanford Encyclopedia of Philosophy. Last modified 2024. https://plato.stanford.edu/entries/artificial-intelligence/
IBM. "Types of Artificial Intelligence." IBM Think. Accessed May 15, 2026. https://www.ibm.com/think/topics/artificial-intelligence-types
