Design justice labs: an international and interdisciplinary digital network for community-centered research on the ?generative AI? Modeling of human languages, communication, arts, and cultures [the…

The National Endowment for the Humanities (NEH) awarded a $250,000 Project Grant under the Promotion of the Humanities Research program (CFDA 45.161) to Rutgers, The State University located in New Jersey. The Design Justice Labs project will establish an international and interdisciplinary digital network to conduct community-centered research on the use of "generative AI" for modeling human languages, communication, arts, and cultures. Key activities include:

• Rutgers will focus on probing large language models to assess biases and errors, explore homogeneity and normalization in AI-generated storytelling, and propose new humanities-based benchmarks for machine understanding.
• The Australian National University will examine the visual logics, creativity, and politics enabled or foreclosed by large image models in the context of decolonial and contemporary arts/media practices.
• The University of Pretoria will extend research and data creation capabilities for local African languages while exploring the sociocultural impacts of language models that marginalize these languages.

The 3-year project period runs from October 1, 2023 to September 30, 2026 and will connect researchers and students with community partners in the U.S., Australia, and South Africa.

Design justice labs: an international and interdisciplinary digital network for community-centered research on the ?generative AI? Modeling of human languages, communication, arts, and cultures [the design justice labs are conceived as a networked scholarly digital project that connects researchers and students with community partners in the US, australia, and south africa to develop and share research on the ?AI? Modeling of human languages, communication, arts, and cultures (?generative AI?). Inspired by design justice principles as elaborated by costanza-chock (2020), our labs center groups that are marginalized by design processes. Rutgers will focus on large language models (llms) through ?probing? Experiments that assess bias and errors; explore the tendency for homogeneity and normalization in AI-generated storytelling; and propose new humanities-inflected benchmarks for machine ?understanding.? Australian national university (ANU) will work with large image models (lims) to explore their visual logics, and the modes of creativity and politics they enable or foreclose in the context of decolonial and contemporary arts and media practices. University of pretoria (up) will extend the capacity of research with, and data creation for, local african languages, while exploring the socio-cultural impacts of llms that marginalize these languages. (edited by staff)]