Guidelines

Innovative technology can offer interactive and immersive in-situ experiences bridging past and present. This can be achieved by linking spatial history using geo-data, archival photos, audio-visual records, historical research, digitized tangible artifacts, and physical geography. It can also involve re-mediatizing older forms of evocative media such as prisoner and survivor artworks, testimonies, and diaries. Personal documents such as diaries, letters, reports, and testimonies offer insight into multiple perspectives on Nazi persecution that constitute a living heritage. This humanizes the historical account and cultivates a multi-layered digital environment in which users can explore various reconstructed histories and historical viewpoints. This can help in shifting attention from the existing historical canon to biographies, sites, objects, and stories that are less known and often overlooked or ignored. 

However, turning Heritage on Nazi Persecution into the subject of digital enterprises is not without ethical and logistical-technological challenges. 

• How can this be done in thoughtful and responsible ways?
• How can we make heritage on Nazi persecution more accessible and inclusive? 
• How can we foster engagement with such sources and create new resonance in the present?

Computer-based visualization offers new ways of exploring historical sources in interactive environments and linking and interconnecting them with related information. This allows active forms of searching, exploring, and curating digitized heritage objects and materials.

For computer-based visualization of heritage on Nazi persecution, in particular 3D modeling of historical sites, prisoner paintings, and exhibitions, we considered the following principles that should also assist in guiding user engagement with MEMORISE applications and prototypes:

Digitization and computer-based visualization can help preserve and safeguard cultural heritage on Nazi persecution by digitally documenting and archiving historical sites, artifacts, and stories, ensuring their preservation for future generations and mitigating the risk of physical deterioration or loss. In this process, we prioritize respect for the integrity of historical sources. Creating 3D models and other computer-based visualizations based on historical sites and sources demands careful decisions regarding historical accuracy. Considerations include whether or not to ‘fill in’ missing information, to geo-locate models when the specific location is uncertain, to ‘crop’ or alter objects, ‘fix’ ‘mistakes’, and acknowledge the extent of historical reconstruction, all of which must be approached thoughtfully and transparently. 

Ultimately, the model should delicately navigate between reconstruction and stimulating imagination, while maintaining transparency about any alterations made to the original artwork. By openly acknowledging places where the work has been ‘touched,’ the model encourages users to explore the original and engage critically (and comparatively) with the historical context. Users must understand the processes and limits of digital and virtual reconstruction, avoid artificial authenticity effects and emphasize the multi-layered structure of history and the temporality and constant transformation of heritage on Nazi persecution.

Digitization and computer-based visualization can help safeguard and certify historical knowledge, particularly in the face of increasing misinformation and falsification of historical facts. The subject matter must be approached with the utmost sensitivity, recognizing its profound emotional impact on survivors, their families, and the broader community. This includes the careful handling of copyrights and the acquisition of informed consent from relevant stakeholders, ensuring that rights and privacy are respected throughout the development and presentation of computer-based visualizations of heritage on Nazi persecution. 

3D models, virtual environments, digital maps, and digital videography allow users to experience memorial sites remotely and access topographic memory and traces online or through digital applications. They can help make historical sites of Nazi persecution accessible to audiences from a distance.

Content and experiences of computer-based visualizations of heritage on Nazi persecution must be made available to diverse audiences, including people with disabilities, different languages, or limited technological access, ensuring inclusivity and equity in-memory representation.

Digitization and computer-based visualization should emphasize the relevance of engaging with the past by resonating with present concerns while establishing connections between past and present beyond mere analogies. They should offer possibilities for users to connect different sources, documents, and objects with each other, additional information, metadata, etc.

Digitized and visualized heritage can function as an entry point into a broad, multi-layered virtual environment(s) that seamlessly connects various tangible and intangible heritage sources; these may include testimonies, diaries, artworks, historical information, archival sources, and other objects. This holistic approach has the potential to convey more intricate and nuanced narratives while underscoring the interconnectedness of the diverse individual experiences and traumatic fragments that characterize the heritage of Nazi persecution.

Digitization and computer-based visualization can stimulate the imagination of past events, while not authentically reconstructing but offering models for a better understanding of historical experiences and decisions. Instead of fully reconstructing sites or objects, we utilize aesthetics of fragmentation or abstract forms to stimulate the imagination of users, provide orientation, and captivate curiosity. It is important to ensure that users understand and reflect on the difference between a digital or virtual reconstruction and their own present lives to enable resonance effects.

For that reason, heritage sources must not be employed as a mere ‘gimmick’. There should be a clear justification for selecting specific sites or objects as data for computer-based visualization and modeling. It is imperative to ensure that the primary focus remains educational and commemorative, steering clear of any tendency towards entertainment or sensationalism.

Computer-based visualization and narration allow us to connect different perspectives by relating objects, documents, testimonies, and other sources in digital and virtual environments based on 3D modeling. In doing so, users can actively engage with individual experiences and stories derived from personal documents such as diaries, testimonies, or prisoner paintings. This allows for multi-perspective storytelling and cognitive-empathic engagement with biographies from the Nazi past. Telling personal stories and connecting with individual biographies strengthens empathy and allows users to relate to a variety of perspectives on the history of Nazi persecution as well as a better understanding of the interrelation of past and present and significant differences between past and present experiences.

It is essential to recognize that digital technology has the tendency to amplify social biases. 3D models and other visualizations should provide an opportunity to rectify this by giving space to the voices and stories of underrepresented victim groups and mnemonic communities. Furthermore, the digital realm should be leveraged to offer multiple narratives simultaneously, spanning different periods and perspectives, echoing the multifaceted experiences that encompass the heritage on Nazi persecution. We recommend shifting attention from the existing historical canon to biographies, sites, objects, and stories that are less known and continuously overlooked or ignored. They can give voice to marginalized experiences of Nazi persecution.

Digital infrastructures provide new ways of collecting and mediating personal memories and experiences by developing applications that allow generating, documenting, and sharing testimonies in a variety of media formats. They should aim for the empowerment of users while respecting individual boundaries. Their aim should be to generate new experiences of engaging and interacting with cultural heritage. This also includes asking for and taking serious user feedback and creating a responsive environment.

 The participatory nature of digital environments allows users to creatively engage with the heritage and memories of Nazi persecution through acts of searching, interacting, sharing, and creating content. Utilizing digitized heritage on Nazi persecution for user engagement allows one to discover and interact with historical sources such as diaries, letters, photographs, artworks, and films, as well as survivor testimonies and biographies. In doing so, it fosters an active search for information and meaning. This can bridge generational gaps by providing opportunities for intergenerational exchange and dialogue, allowing younger generations to connect with the past and learn from the memories of older generations while promoting intergenerational understanding and dialogue.

However, thoughtful consideration is necessary to determine when it is appropriate to promote interactivity and when a passive walkthrough is more suitable, aligning the computer-based visualization with its intended purpose and target audience.

Artificial Intelligence (AI) does not operate in a social vacuum but rather contributes to shaping the experience, perception, and understanding of human users, whether they belong to the present or to future generations.

AI methods used in MEMORISE should not impose harm on the lives of users, their health, well-being, or security. However, the use of digitally processed heritage on Nazi persecution (HNP) may affect users’ perception and understanding of the past. This might include a confrontation with disturbing historical content selected and distributed through AItrained systems – a risk that can be mitigated through constant monitoring. These guidelines have been closely coordinated with the ethical guidelines developed as part of the MEMORISE project and are based on our further reflections on the possibilities and limits of visualizing HNP.

AI and machine learning can be important tools for analyzing and processing digitized materials related to the heritage on Nazi persecution, especially when dealing with large quantities of documents, photographs, and testimonies. However, digitization of Holocaust-related materials using AI presents both opportunities and challenges, and it is crucial to approach this task with a great sense of responsibility and respect. Concerning the visualization of HNP, we have established conceptual and ethical considerations that form the foundation for the development of applications and prototypes in the MEMORISE project, including AI elements. Here we present some aspects that are of particular importance for dealing with AI in the context of HNP.

● Privacy, Confidentiality, and Dignity: The digitization of HNP-related materials often involves sensitive and personal information about perpetrators, victims, and their families. It is therefore important to make sure the system does not infringe on the right to privacy and that private and sensitive data is protected. We will ensure that the data is protected from misuse or access by unauthorized parties. All data will be handled with care and respect for the privacy and dignity of those affected.

● Diversity, Fairness, Bias, and Discrimination: AI algorithms can sometimes perpetuate or amplify existing biases and discrimination, which could have serious implications when dealing with HNP-related materials. It is crucial to ensure that the AI algorithms used are fair and trained on unbiased and well-curated data collection and that they do not perpetuate any harmful biases or stereotypes.

● Inclusivity in AI Design: It is crucial to ensure inclusivity in digital design to make AI technologies fair, unbiased, and accessible to all individuals, regardless of their backgrounds or characteristics. Designing AI systems with accessibility in mind ensures that individuals with disabilities can fully engage with the technology. Conducting user testing, incorporating user feedback, and ongoing refinement of the design based on user input will help to contribute to the creation of more inclusive and user-friendly AI-reinforced interfaces.

● Accuracy and Authenticity: Digitisation using AI must be accurate and preserve the authenticity of the materials being digitised. This includes preserving the quality and integrity of the data or its representation. This is particularly important for HNP-related materials, as any inaccuracies or alterations could have serious implications for research and historical understanding. Yet, AI methods, like humans, are prone to errors. We rely on manual curation and feedback from user testing to minimize the impacts of such errors as well as to ensure the absence of disruptive flaws.

● Informed Consent: In some cases, the digitization of HNP-related materials may involve personal information about individuals who did not consent to their information being made public. It is important to obtain informed consent from individuals or their families where possible and to ensure that any use of personal information is done in a way that respects the individual’s privacy and dignity.

● Protection from Harm: The AI systems should be designed with a focus on human well-being and empowerment. Human agency, liberty, and dignity should be upheld when developing AI technologies, especially when using technologies that might augment human capabilities.

● Cultural Sensitivity: HNP-related materials are often deeply personal and culturally significant, and it is important to approach their digitization with sensitivity and respect for the cultural context and importance of the materials.

Preventing, and mitigating potential bias, discrimination, and stigmatization in HNP digitization using AI requires a multifaceted approach that involves careful selection of data sources and algorithms, expert review, and regular monitoring and evaluation of the processes. These measures can help to ensure that the digitization process is respectful, transparent, and accountable. It is also important to recognize that AI systems are not neutral or objective and to be transparent about the potential limitations and sources of bias in the digitization process. AI algorithms learn from the data they are trained on, and if the data is incomplete or contains a particular bias, the resulting AI system may perpetuate these biases. For example, as the archival documentation of Nazi persecution is incomplete, certain groups or experiences may be over- or under-represented in the data. Ensuring that the input data used is representative and diverse, and derived from multiple sources can help to prevent bias or discrimination based on a limited or skewed data set.

MEMORISE prototypes and digital infrastructures employ advanced AI technologies to enhance the user experience with digitized historical documents and personal stories. A set of guidelines aims to provide transparency about the use of AI in organizing and processing data, including automatic translation of documents in the development, improvement, and implementation of these prototypes and digital infrastructures, matching users with specific historical biographies, and topics.

MEMORISE prototypes and digital infrastructures utilize:

● AI algorithms to categorize and organize historical documents. This process involves machine learning models that analyze patterns, themes, and metadata to provide a systematic and efficient exploration experience.

● AI-assisted automated tagging to identify key themes, events, and people within documents and other heritage materials. This enhances search functionalities and allows for a more nuanced understanding of the content.

● AI language processing to improve automatic translation features, breaking down language barriers and making content accessible to a global audience.

● Machine translation models that analyze and convert historical documents into the chosen language. Users can enable or disable this feature based on their preferences. Machine-translated texts and documents are labeled accordingly.

● Generative AI, to create descriptions, summaries, and other informative texts. Thorough fact-checking protocols ensure that these texts do not contain any errors or false information.

● Matching and recommendation algorithms to personalize content and propose points of contact that prompt users to engage with specific biographies, stories, and heritage materials.

MEMORISE prototypes and digital infrastructures do not utilize:

● Any AI tools that could harm the health, well-being, and safety of users, or that might have overwhelming emotional effects.

● AI-based machine learning that uses or stores personal user data without the user’s knowledge.

● Generative AI for the artificial production of historical sources. In accordance with our ethical principles, the integrity of sources is given the highest priority.

In order to enhance visibility and transparency when interacting and engaging with MEMORISE prototypes and applications, users:

● Are informed about the use of particular methods, machine learning, and AI-trained systems for development, improvement, and implementation, and their implicit and explicit bias.

● Receive visible indications of the use of AI when they interact with certain tools and information.

● Can access a summary of the content used for training and the nature of algorithms involved in the process.

● Can independently decide on the use of machine learning-based algorithms and deactivate matching and recommendation algorithms

The project is committed to introducing benchmarking standards that better contextualize the data used, to avoid irritation and confusion in the human-computer interaction (HCI), and to a prudent, user-oriented, and ethically informed curation of data.