FIAT/IFTA World Conference 2025

The growing reliance on artificial intelligence technologies in film archiving has raised increasing concerns regarding the neutrality of these systems: particularly in representing non-Western productions. Global tools such as Google Cloud Video Intelligence and Amazon Transcribe are commonly used in audiovisual archiving; however, their linguistic and cultural training raises questions about their ability to accurately represent Arab films.

Although advancements in Arabic language processing have improved AI’s capacity to interpret Arabic content, challenges remain. These systems may still reproduce algorithmic biases stemming from unbalanced training data or design assumptions. Furthermore, the complexity of the Arabic language, especially its dialectal and cultural diversity, continues to pose unresolved technical and archival obstacles. This gap has yet to be systematically studied through comparative analysis with specialized Arabic-language AI models. If unaddressed, the integration of such global technologies into Arab film archives risks producing digital repositories that misclassify or marginalize local narratives, thereby distorting cinematic memory.

This research proposes an applied analytical study to examine the potential bias of global smart archiving systems toward Arab films. It evaluates the performance of Google Cloud Video Intelligence and Amazon Transcribe, comparing their results to those of specialized Arabic AI models. The study also explores the potential for integrating Arabic models via a dedicated API to enhance the performance of global systems and proposes practical solutions to foster a more inclusive Arab digital film archive.