Generative AI, as an image creation technology, often perpetuates biases and representation harms, yet its use with open-source models and carefully curated datasets presents a tentative opportunity for self-representation and community-driven approaches. My work with Future Archives experiments with these possibilities to address representational gaps and epistemic harms in the SWANA region, with a focus on Levantine histories and Palestinian cultural narratives. For example, I have trained customised LoRA (Low-Rank Adaptation) models on the aesthetics of mid-20th-century Arabic vinyl records. This inquiry remains critical of the structural inequities embedded in AI systems, aiming to cautiously counter futurcide and the digital divide while probing the potential for more ethical, community-centred uses of generative AI.
Future Archives
