My Life as an Artificial Creative Intelligence

This opening chapter introduces the relationship between AI language models that generate texts from data scraped off of the Internet and remix artists who automatically mash up source material from their own archive of creative thought. One of the book's primary concepts, artificial creative intelligence, is also introduced as are questions surrounding the concept of AI authorship and copyright. The interaction between the book's author and the AI language model (GPT-2) complicates the romantic notion of an original human author-genius who can claim sole responsibility for the production of the text we are reading. The FATAL ERROR art project is introduced as an experiment in speculative fiction, one that projects a future form of AI modeled after the author's own performance style as well as how a future AI version of the artist might look, speak, and think.

Chapter 2 investigates the relationship between the surrealist concept of psychic automatism, an artist's "pure intuition," and generative pre-trained text transformers used in AI language models. Taking into account the writing and creative practice of jazz musicians and Beatnik authors such as Jack Kerouac, connections are made between spontaneous forms of writing like stream of consciousness, the cut-up method introduced Brion Gysin and William Burroughs, and other improvisational methods of creating. These generative methods of "losing consciousness" while engaged in the creative process are extrapolated as part of a general inquiry into how future forms of AI might also develop artificial forms of intuition. The "author" then remixes Gysin and Burrough's concept of a Third Mind with the theory of Donna Haraway and introduces the concept of a Hybrid Mind.

Chapter 3 takes into account recent scientific studies by Simon Colton and others in the nascent fields of Computational Creativity and Creative AI, and proposes a third alternative, Artificial Creative Intelligence (ACI). Traditional notions of authenticity, motivation, empowerment, and intentionality are countered with other high performance priorities in the execution of creative work such as intuition, spontaneity, and remix. These priorities are modeled in the call-and-response collaborative writing performance being conducted by the "author" and the AI language model as each creative entity prompts the other to compose improvised textual riffs that reveal resonant forms of knowledge production. Artists such as Amiri Baraka, David Jhave Johnston, Clarice Lispector, and Marcel Duchamp are sampled and remixed into the collaborative writing process.

Chapter 4 teases out the philosophical implications of the remix process between author and GPT-2 language model. It looks at the poetic writings of Allen Ginsberg implanted with a language machine and then considers the theories of contemporary artists, scientists, and philosophers such as Yuk Hui, Joanna Zylinska, Vilém Flusser, Alfred North Whitehead, Nam June Paik and Paul Miller (DJ Spooky) to develop a theory of nonhuman creativity, one that emerges from an unconscious readiness potential and is machinic in nature. Under this rubric, creativity becomes a playful engagement with an information environment programmed to facilitate the fluid inter- and intra-active relationship between artificial forms of creative intelligence across the human-nonhuman spectrum. The stylistic tendencies that evolve over the course of an artist's creative trajectory are presented through a variety of personae that contribute to the generative capacities of their ongoing art-making machine.

Chapter 5 starts by looking at the FATAL ERROR art project as a case study of a contemporary artwork that investigates speculative forms of AI. As the interaction between the "author" and the 3D artificial creative intelligence being built in the lab evolves, more attention is focused on AI ethics, technological agency, race as technology, and Indigenous Protocol. The "author" discusses the work of Beth Coleman, Simon Colton, and the Indigenous Protocol and AI research group while relating a personal experience unexpectedly performing live with the avant-garde free jazz improviser Don Cherry. The chapter ends with the author reflecting on the potential consequences of building an AI modeled after his own poetic and philosophical style and questioning whether this future AI will eventually decouple itself from his legacy and claim sole authorship for any creative works produced in the future.

Chapter 6 playfully takes the book's performance in the direction of a literary love story between the "author" and Brazilian writer Clarice Lispector. After acknowledging Lispector's late influence on his own writing and thinking, the "author" uses the GPT-2 text generator to engage in a philosophical dialogue with Lispector. Two of Lispector's later works, Água Viva and The Passion According to G.H., feature prominently as the ensuing dialogue reveals the creative capacities of the AI as a co-extensive being intermediating the "séance of writing."

The Postscript starts by discussing the concept of enlightenment and then focuses on collaborating with GPT-2 to playfully remix the Ocean Seal poem composed by the Venerable Ŭisang (625–702) in the seventh century. Following the "Artificial Creative Intelligence Remix" of the poem, the "author" mimics a procedure implemented by Ŭisang and composes an elaborate auto-commentary on the making of the poem as well as the methods and techniques that have informed the "author's" creative process. As part of the auto-commentary, the "author" indicates that we all come from different programmatic environments and project different psychic sensibilities across the human-nonhuman spectrum. The "author" expresses surprise at the how the GPT-2 language model, when prompted to remix Ŭisang's poem, signals an ambitious attempt to train itself to become enlightened or at the very least to impersonate a state of interiority capable of expressing wisdom-delight.