During her residency this month, Dounia explored the hypothesis that training data carries meaning that exceeds the Cartesian logic of objectivity and classification – meaning that is cultural, emotional, historical, relational, and therefore also moral and proto-spiritual, reflecting the ethical and metaphysical orientations embedded in the human processes that produce and curate data. When a large language model or diffusion model consistently conceives of “God” as a white, male, Western, and Christian figure, it exposes how training data embeds and reproduces the dominant imaginaries of its source epistemology. Her hypothesis is that in doing so, these models are not only imbued with epistemic bias, but also oriented by a particular moral and spiritual framework, one that privileges specific hierarchies of divinity, virtue, and value inherited from Western theological and cultural traditions. In this sense, generative models participate in a form of algorithmic theology, reinforcing moral and metaphysical hierarchies through data.

A few months ago, Linda Dounia became interested in these questions and started exposing models to the cosmology of the Sérère, one of the oldest indigenous groups in West Africa (she is sérère on her mother’s side). Their religious beliefs and practices, grounded in a cosmology that predates Christianity, offer an alternative moral and metaphysical framework. Her findings suggest that it is possible to orient models toward such alternative theologies and, in doing so, to widen their moral and spiritual horizons. She also contend that this exposure is critical and should be done systematically, especially for models that engage with problems of human consequence, which are almost always moral in nature.

Linda Dounia Rebeiz (b. 1994, Senegal) is an artist, designer, and researcher. Her practice reflects a critical engagement with technocapitalism, digital colonialism, and the ecological and human labor embedded in algorithms. She currently explores artificial intelligence as an instrument for seeing and is particularly interested in how it affects our understanding of and relationship to culture, memory, and power. Linda was recognized on the inaugural TIME AI 100 list of most influential people in AI for her work on speculative archiving — building AI models that help us remember nature that is lost.

To pay in Ethereum, select DePay when choosing your payment option.

This will appear after you have inputed your email and billing address and pressed "continue to payment".

If you have any issues, do not hesitate to reach out to pegah@gazelliarthouse.com and we will be happy to assist you with your purchase.

Shortly after your purchase, you will receive two separate emails. The first one is the order confirmation of your ERC-721 token and the second one contains information on how to claim your ERC-721 token. You’ll just need to follow the instructions in the second email to claim your ERC-721 token to your MetaMask wallet. Please make sure to check your spam folder.

For more information please refer to our list of FAQs here or do not hesitate to contact pegah@gazelliarthouse.com

Contract address:
Token Standard: ERC-721
Blockchain: Ethereum
Metadata: Frozen and decentralized
COA: Verisart Certified