AI Definitions in Flux:

Authenticity and Holocaust Testimony in Focus

Abstract

The museum, as an institution, has a long-standing tradition of presenting the public with seemingly authentic and evidence-based narratives. Recent surveys indicate that museums are widely regarded as highly trustworthy across various countries. However, as museum practices increasingly intersect with marginalised and contested politics, this trust can no longer be assumed.
This article offers an in-depth process study of the development of the Swedish Holocaust Museum, with a particular focus on the role of Artificial Intelligence (AI) in its Dimensions in Testimony (DiT) project. DiT features video recordings of Holocaust survivors, each of whom has been filmed responding to approximately 1,000 questions. Visitors engage with these testimonies in real time, through AI driven natural language processing. The analysis draws on a review of documentation and communications, as well as observations and interviews with museum professionals and visitors. It is guided by the conceptual distinction and operationalisation of relational and essential authenticity, and situated within the broader context of the affective turn in museums. A critical dialogue with contemporary scholarship explores the implications of AI technologies for authenticity and trust in the museum sector.
The study identifies a notable shift over time in how the term ‘AI’ is used and understood, accompanied by emerging ethical concerns. Developing a shared institutional language around new technologies has proven challenging. In just a few years, AI has moved from being framed as an exciting innovation in exhibition design to a more cautiously referenced, and sometimes avoided concept. This evolution presents significant challenges to the accurate representation of historical evidence, the authenticity of survivor testimonies, and the ethical integrity of the narratives conveyed. While visitors generally express trust in the DiT narratives, persistent concerns remain regarding the role of AI in shaping evidence-based content and the ethics of digitally mediated representation.