ArchAIvist: Methodological and Technological Support for Working with Archival Audiovisual Material

The rapid development of artificial intelligence technologies has opened new horizons for the audiovisual sector, offering opportunities for creativity and efficiency, while at the same time presenting a number of challenges. One of these areas is work with audiovisual archives, where the presence of artificial intelligence (AI) is associated, on the one hand, with the possibility of automating parts of archival workflows—including the efficient creation of metadata descriptions for large data corpora—and, on the other hand, with issues related to its responsible use. These include the limitations of current sound and image analysis tools, ethical concerns, and copyright issues. Such a situation calls for an interdisciplinary approach combining insights from computer science, law, ethics, (film) history, and archival studies. Moreover, from the perspective of law and media ethics, the use of older archival audiovisual materials also represents a challenge for film producers, directors, and other professionals involved in the production and exploitation of documentary films. The core of the project consists of ethical and legal research into the obstacles and dilemmas faced by creators using archival audiovisual material. The ethical research will focus on addressing ethical issues (work with sources, forms of recontextualization, and respect for depicted individuals; the use of AI models) that arise when working with reused material, while the legal research will focus on issues related to the multitude of rights attached to archival materials; the large number of rights holders; the difficulty of identifying and locating information about these holders; unclear interpretations of exceptions to the protection of archival materials; specific requirements of institutions involved in film production regarding the scope of secured rights to film archives; and legal risks (violation of personality rights, copyright infringement) associated with the use of AI models for describing archival material. The results of the ethical and legal research will be applied in testing the effectiveness of state-of-the-art AI models in archival contexts, with a focus not only on pattern recognition in historical collections but also on their fair and responsible use.