About the GenAI Studio
The GenAI Studio aims to support Milieux and AI2 students, staff and faculty develop collaborative and personal research projects using generative AI. The studio’s technical and social governance is part of the ongoing research-creation of the studio itself and is subject to change. As a research-creation project, the studio privileges free software approaches to AI first, open models second, commercial models last in its adoption and design.
The studio serves as a central hub for students from any discipline to engage in peer supported research and research-creation with generative AI tools. Studio members are expected to share expertise and resources, and this can be expanded with basic and advanced workshop, residency, and internship programs as interest and participation.
Seed Funding Pilot
The Milieux Institute and the Applied AI Institute partnered to provide seed funding to an initial cohort of eligible students for generative AI research and research-creation projects between May 15-August 31, 2024. The purpose of this pilot is to encourage student-driven cultural research-creation while also building a peer network around the Machine Agencies GenAI Studio. Funded students are not meant to work in silos but rather are encouraged and required to assist each other, share knowledge and experience, and help build the studio infrastructure and research culture.
Projects
Aurélie Petit – B̷l̷a̷c̷k̷ ̷P̷u̷d̷d̷i̷n̷g̷
B̷l̷a̷c̷k̷ ̷P̷u̷d̷d̷i̷n̷g̷ is an animation project based on Black Pudding (1969), a lost adult-oriented experimental animation film from Canadian-American visual artist Nancy Edell. By using Black Pudding’s existing film reviews, film criticisms, synopsis, and director interviews as guiding prompts to generate related AI images, B̷l̷a̷c̷k̷ ̷P̷u̷d̷d̷i̷n̷g̷ will interrogate the ethical and creative limits of GenAI to (re)produce feminist sexual representations.
François Lespinasse – Mechanical Meanderings
Mechanical Meanderings is a computationally-augmented audiovisual experience that explores the intersection of creative writing and embodiment through two installations: Mechanical Brainstorm and Mechanical Silhouettes. Mechanical Brainstorm invites participants to engage in a reflective writing process with an AI system, while Mechanical Silhouettes generates visual landscapes based on the participants’ writing and movements, creating a collage of human and non-human narratives.
Kamyar Karimi – Deconstructed Selfies: Regenerated
Deconstructed Selfies: Regenerated is an audiovisual experience in which the audience is invited to participate in taking a selfie at the installation, then a trained AI model deconstructs the selfies in the style of the works’ first edition by dissecting facial characteristics and providing a digital identity of each individual in the digital realm. This experience reflects on how “selfies” were thought of previously, presently, and how generative AI will shape a future for them.
Luciano Frizzera & Maurice Jones – The Consultation Machine
The consultation machine is a satirical, all-in-one AI-driven public consultation tool for Canadian citizens that takes on the AI hype. Public consultation improves democratic governance by helping governments consult their citizenry on the key public policy issues they face.
Maurice Jones – feral.ai
feral.ai reimagines our more-than-human relations inspired by the critical openings of Black Ecologies and Black Technoscience. Beyond histories of extractivism and settler-colonialism, feral.ai draws upon the mutli-faceted relations black communities have formed with their natural and built environments – relations that center more-than-human recuperation, relationality, reciprocity, and kinship across space and time. To do so, the generative sound work draws upon the learnings from the Maroon communities of the Great Dismal Swamp, black environmental justice movements, and black techné to explore liberatory relations with the more-than-human world.
Rowena C & Hamidreza Nassiri – Machine Dreams of a Cosmopolitan Past
Machine Dreams of a Cosmopolitan Past is a custom-trained, bilingual (English-Farsi) image generation model which focuses on reproducing Ilkhanate and Safavid Iranian art. Through opening up this window to Iran’s cosmopolitan past for remixing and remaking, it asks a question of our global cosmopolitan present: how will our cultures continue to evolve in an era of digital cultural diffusion and international mobility?