Exploring emerging technologies in archiving and preservation: Leveraging 3D models,interactive environments,and AI tools

Abstract

This article, derived from a presentation at the NISO Plus 2024 Global Online Conference, explores how cultural heritage practitioners can leverage emerging technologies to enhance their work. New technologies present librarians, archivists, and other cultural heritage practitioners, with tools to streamline repetitive tasks, enrich collections, and build more interactive, accessible resources. This article highlights AI applications and emerging technologies that can generate scripts without needing coding experience, create 3D models that increase accessibility and engagement, and develop virtual exhibits that extend the lifespan and reach of physical exhibits while providing additional interactive elements. Together, these innovations provide valuable strategies for the future of cultural heritage work.

Introduction

The rapid development of AI and other emerging technologies provide significant opportunities for cultural heritage work, bringing increased efficiency, engagement, and productivity. However, these opportunities must be balanced with cultural heritage institutions’ high standards for trust, authenticity, and ethical integrity. The challenge is to preserve the trusted reputation of the cultural heritage field while adapting tools that were not specifically designed with the field in mind. Despite this challenge, recent advances have led to numerous promising applications. Here, I will focus on tools and methods that have proven most valuable in my work as a Digital Curation Librarian in an academic library, particularly in terms of AI, 3D modeling, and creating virtual exhibits.

AI resources for libraries, archives, and museums

Generative AI presents both promise and complexity for the cultural heritage field, especially given the public’s expectation of trust and transparency in our institutions. ¹ While AI’s tendency to create nonexistent or inaccurate responses, or “hallucinations,” remains a very real concern, and a significant barrier for its use in academic settings, it is increasingly possible to find viable applications for generative AI tools ranging from transcription to the automation of repetitive tasks. For instance, generating Python scripts using AI text prompts has been game-changing for handling everything from everyday tasks to complex research projects while saving significant time.

Python scripts are an essential part of our department’s digital workflows, and many are used daily. I, however, have no experience writing code and as a result, I never considered it as a viable solution to tasks that I needed to complete. Gaining the ability to generate scripts using text prompts reshaped how I look at problems and significantly impacts the way I work.

For example, a digital collection clean-up project required consolidating PDFs from numerous folders into a single location and compiling file names into an Excel document. Previously, this would have been a lengthy and repetitive task. However, by generating scripts using AI text prompts, I was able to automate the process in minutes. Simple scripts like these require only a basic understanding of prompt engineering due to their straightforward nature. For more complex scripts, however, crafting well-thought-out prompts becomes crucial. Using AI tools such as ChatGPT significantly reduced the workload, saving hours of effort and alleviating burnout by automating tedious, repetitive busywork.

Given AI’s hallucinations, and my limited ability to read and understand Python, it is absolutely critical to test scripts on dummy data. Without knowing exactly what a script will do always test it. This may add a few extra minutes of work, but it is well worth the effort to avoid data loss. After testing my scripts on a sample dataset to ensure that they worked, I ended up with a resource I could reuse for future projects.

Similarly, generative AI helped streamline other tasks, such as reformatting names in spreadsheets, where potential errors could easily arise if done manually. By changing my mindset, I have been able to find new ways to maximize my efficiency while maintaining a higher degree of accuracy.

These examples highlight how AI tools, when used strategically, can increase efficiency without compromising trust or authenticity.

Generating simple scripts using AI is usually straightforward, but more complex tasks may require more trial and error and the use of more specialized tools. For example, a recent research project of mine involved the extraction of thousands of records from an online database, requiring the creation of a web scraping script. The script needed to navigate past a login screen and extract data from more than fifty fields per record including drop downs, fill-ins, and free-text while also downloading attached PDFs and JPEGs. The tens of thousands of pages of data that I needed made it impractical and unrealistic for me to manually extract the data. After trying to use ChatGPT and encountering limitations resulting in a logic loop, I switched to GitHub CoPilot, which allowed me to successfully complete the project.

For complex, multi-step scripts, prompt engineering becomes a critical part of the process. One limitation of tools like ChatGPT, and to a lesser extent GitHub Copilot, is their tendency to abbreviate data extraction sections or overlook previous errors in long scripts. For example, ChatGPT occasionally “forgot” an error addressed in earlier exchanges, leading to logic loops.

To overcome this, I found it necessary to build complex scripts in smaller, manageable pieces and then combine them manually outside the platform. Crafting these individual components required carefully designed prompts and planning to ensure seamless integration. While the resulting script might appear unconventional or “Frankenstein-like” to a trained coder, it allowed me to effectively achieve the intended goal.

The ability to generate code has been the most impactful application of AI resources for my work, but it is far from the only benefit of the technology. AI’s utility in transcription is another transformative application. AI transcription tools may not be one hundred percent accurate, and still require review, but tools such as Adobe Premiere Pro, YouTube, Whisper AI, and Zoom have significantly shortened the transcription process for audiovisual materials, allowing for increased accessibility at a fraction of traditional transcription costs. Creating a transcript of an hour-long video by hand can take an estimated four to six hours and potentially cost ninety dollars (US) per hour.^2–4 Given the size of cultural heritage collections and the lack of available funding for transcription, these tools provide an opportunity to increase the accessibility of collections not readily available before the emergence of AI.

In a similar fashion to AI transcription, Handwritten Text Recognition (HTR) tools such as Tesseract OCR and Transkribus, though still evolving, are making handwritten materials more accessible by improving readability, searchability, and usability. These tools have advanced significantly in their ability to interpret diverse handwriting styles, offering researchers new ways to engage with previously inaccessible content.

AI tools also have the potential to support research by streamlining workflows, though their role in scholarly writing is more contentious. With many publications restricting or requiring that authors disclose their use of AI tools in the writing process and reviewers potentially viewing AI-assisted work unfavorably, I would avoid using them for scholarly research. ⁵ However, in terms of conducting research for an exhibit, or discovering relevant sources, AI can still serve as a valuable resource.

Tools such as the Web of Science and Research Rabbit can be beneficial when trying to find relevant source materials. Clarivate recently launched an AI-powered Web of Science Research Assistant designed to help researchers find sources quicker and more efficiently. ⁶ These tools are best used to complement rather than replace traditional research efforts. To effectively use Research Rabbit, the researcher needs to already have a list of citations that can then be expanded.

Products such as Assistant by Scite allow users to ask a research question and get a response that includes real citations from academic journals. ⁷ This type of tool may cross the ethical line when it comes to scholarly publications, but it can be useful when conducting exploratory research or background research for an exhibit. However, even if the citations are not hallucinations, the original articles must be read to ensure that the citations accurately represent the text.

3D models

Although AI dominates the emerging tech landscape, other tools like 3D printing have shown substantial potential in the cultural heritage sector. At the University of Alabama at Birmingham (UAB) Libraries, for instance, we have started using 3D printing to reduce handling of fragile objects by creating durable replicas and to enhance exhibitions through interactive models. A recent exhibit titled Navigating Communication: Breaking Invisible Barriers on the history of Otolaryngology provided an ideal case study. The exhibit included a fragile book from 1875, Light for the Blind: A History of the Origin and Success of Moon’s System of Reading for the Blind, that included two pages with raised text illustrating an early reading system for visually impaired readers and a map of the British Isles. Unable to let visitors handle the original, we explored opportunities to use 3D modeling to create an interactive replica that visitors could touch.

The book pages were too undefined and thin for traditional 3D scanning, so we tested a variety of different approaches. We tried using Tinkercad to manually draw raised letters, which did not work because each letter on the page would have needed to be created individually, and the essence of the item would have been lost. We gradually shifted to trying out various 3D conversion software options. Looking Glass seemed promising but there was not enough depth to provide a usable result. ⁸ Eventually we found a tool called 3DPEA and printed the model in gray resin using a Form3+ printer. The result was not an exact match, but it did fulfill our goal of making the object accessible for visitors while ensuring that the original work was not put at risk.

To supplement the book pages, we also printed a few anatomical models related to the exhibit that did not come from the library’s collection. We printed an oversized model of a human ear, including both the external and internal anatomy, with braille labeling on the different features. We also printed an oversized human eye designed for visitors to disassemble and reassemble, making the exhibit more engaging and interactive.

Virtual exhibits

In parallel with 3D modeling, we have begun creating virtual exhibits to extend the reach of our physical displays. Using CenarioVR, we are developing a virtual representation of the Navigating Communication: Breaking Invisible Barriers exhibit, simulating an immersive walk-through experience that includes interactive features such as curator commentary and an animated 3D model of a human eye. This approach preserves the exhibit for future audiences and allows us to make it accessible to remote visitors, ensuring that the considerable investment in temporary exhibits continues to benefit the public long after the physical displays are dismantled.

Conclusion

The integration of AI and other emerging technologies into cultural heritage work offers transformative possibilities. By thoughtfully selecting and adapting tools that align with the field’s ethical standards, cultural heritage professionals can expand access, reduce repetitive work, and create dynamic, inclusive resources. As these technologies continue to evolve, cultural heritage professionals can selectively incorporate them to meet the specific needs of their institutions, enhancing both collection accessibility and the public’s engagement with cultural heritage. This ongoing exploration opens new pathways for innovation, laying a solid foundation for a tech-forward, yet ethically grounded, future in libraries, archives, and museums.

Statements and declarations