In 2021, the MIT Press published my first full length book, Techno-Vernacular Creativity & Innovation. Artificial intelligence or AI is briefly mentioned through artists such as Stephanie Dinkins. Dinkins created a training model for an AI agent based on living oral histories and literature including Toni Morison’s Sula (1973) and W.E.B. Du Bois’ The Souls of Black Folk (1903). Known as N’TOO for short, the project exists as an artificially intelligent and socially engaged sculpture. This is reappropriation.
The narrative is experienced as a dynamic conversation between N’TOO and the user, in which the stories are altered according to the user’s questions or the AI’s mood. Over time, N’TOO’s storytelling skills and available vocabulary will grow with each user interaction, folding the conversations into its multiple data sources. — Imani Cooper
In my book I define TVC reappropriation as the counterhegemonic practice of repurposing things in ways that revalue, resignify, and relocalize artifacts from mainstream, or dominant, cultures. Reappropriation can overlap with remixing because both modes involve the manipulation and use of preexisting artifacts.
Appropriation is on the other side of this equation: someone from a socio-politically dominant or privileged group takes from a less dominant one. We’ve seen this time and again from Little Richard vs. Pat Boone to Willie Mae Thornton vs. Elvis. This flipped with sampling, when youth from disenfranchised communities re-purposed audio equipment and rhythm sequences… layering and rearranging sounds to create something new.
Rap music production or sampling, especially in the 1980s and 1990s, was seen as a confrontation to the existing status quo and this (counter-hegemony) can be observed in other spheres of life, such as history, media, music, visual art, etc. Besides the legal factors, sampling was viewed by some as stealing or not creative. However, sampling, when done well, is simply not stealing. If used in the incorrect way, at worst, it’s copyright infringement, which is implicitly different than theft.
We have seen examples of appropriation and reappropriation throughout history. Akira Kurosawa seemed to be okay with George Lucas taking ideas from his films but Kurosawa sued and won a dispute against late Italian filmmaker Sergio Leone who took the entire plot, scenes and characters from “Yojimbo” for his film “A Fist Full of Dollars.” More recently, the Supreme Court has taken up a dispute involving pop artist Andy Warhol and a photographer who claimed Warhol violated her rights in creating images of Prince based on her photo (see above).
But I digress. Back to TVC reappropriation and AI. Reappropriation involves the recontextualization of materials and their redeployment. This includes when a work of art is taken from its original context and given new meaning by both the artist as well as the new context it is made in. In the process of recontextualization, the new work continues to have a connection with the original it references. I assert that the use of text and image prompts in AI tools such as Midjourney is recontextualization, which can blur the lines between creativity, invention, and copyright infringement. AI art can also level the playing field between dominant and less dominant groups that have access to the tools.
The other two TVC modes of production are remixing and improvisation. I will briefly explore these terms as it relates to AI art. In the book, remixing is defined as adding, removing, and changing artifacts. The most obvious connection is the Remix Prompt in Midjourney. AI art remixing is basically the concept of modifying, changing or building from a specific image output. Improvisation is the spontaneous and inventive use of materials.
Improvisations stage the emergence of new forms and practices in which one moves from event to event, decision to decision, and retrospectively creates connections only to dissolve them again.
Improvised activities are less rule-based, more fluid, chaotic or reactive, and are more process-oriented. AI has been making significant strides in this area. NLPs such as Midjourney takes words and sentences apart, allowing software to recognize them by name. Then it approximates images based on these words. For example, if the prompt is “butterflies” Midjourney will find instances of the image and attempt to accommodate the prompt. In other words, it improvises based on user input.