Augmented Art – 2021-22

// This work was conceived and executed 18 months before the release of “ChatGPT” in November 2022, at a time when LLMs were largely unknown outside of computer science circles. The AI interface available at this point was very different and involved tedious fiddling with parameters and priming to get only occasionally intelligible intended results. With the release and ‘perfection’ of the streamlined “ChatGPT” and similar models, this area ceased to hold any interest to me and I no longer use AI in the creation of my work. //

Blurring the lines between the artist’s and AI’s intentions, Augmented Art, uses a disinterested AI as a collaborator in the process of creating intensely personal videos exploring identity and neurodiversity. GPT-3, an AI released in beta in mid-2020, has capabilities 10x greater than previous deep learning language models. After priming with personal information, the artist used GPT-3 to augment the creative process, co-authoring scripts that are then filmed and edited by the artist.

Commissioned by Chronic Insanity for Puncture the Screen digital art festival, with funding from the Arts Council England and National Lottery.

Motivation
I was interested in different types of ‘brains’: How do neurodivergent people process differently from neurotypical people, and how do neurodivergent people process differently from each other? How does a language-predictive AI model process differently from humans? How do all of these function together creatively, and can they ‘understand’ each other?

GPT-3 is often marketed as a “writing partner” to bounce ideas off. Could you use an AI as a collaborator when you don’t have another human who gets ‘it’? This type of collaboration starts to look attractive to those with severe social anxiety or who are alienated from communities of people, whether in our dystopic megacities or depopulated rural areas.

Collaborators
Film / Music / Coding – Tom Shennan
Writing – GPT-3 via OpenAI