Skip to main content

Blog

Exploring AI Applications in Cultural Heritage

Posted on    by CREATe Team
BlogDigital Technologies and HumanismEvents

Exploring AI Applications in Cultural Heritage

By 27 June 2024July 2nd, 2024No Comments

This blog post, by Panagiotis Lampropoulos, Lucy Cunningham and Yi-Ting Lin, contains reflections from a workshop organised by CREATe’s ‘New Technologies and Humanism’ theme and the Digital Cultural Heritage Arts Lab, Glasgow, 31 May 2024.

Panagiotis Lampropoulos and Lucy Cunningham are students enrolled in the LLM Intellectual Property and the Digital Economy run by CREATe, who are deeply interested in copyright law, the regulation of the digital environment and the upcoming challenges caused by the rise of AI technologies.

Yi-Ting Lin is PhD researcher at Information Studies, School of Humanities at the University of Glasgow. Her research is about social sustainability in digital cultural heritage, so she is interested in how technology has been changing the relationship between museums and their communities.

The integration of artificial intelligence (AI) and cultural heritage was the focal point of the ‘AI in Cultural Heritage’ Workshop, organised on 31 May 2024 by the Digital Cultural Heritage Arts Lab and CREATe’s ‘Digital Technologies and Humanism’ theme. This brought together experts from various domains to discuss the opportunities and challenges posed by AI in preserving and enhancing cultural heritage.

For Panagiotis and Lucy, despite having attended for the purpose of informing their specific dissertations, they left the workshop with a greater appreciation of the topic, and with more questions than answers (an obligatory element of any such successful workshop). Yi-Ting had not explored this fast-developing field before, so she welcomed the opportunity that this workshop offered.

The event commenced with a warm welcome from Maria Economou, Professor of Digital Cultural Heritage at the University of Glasgow. As Co-Director of the Digital Cultural Heritage Arts Lab (with Dr Lynn Veschuren, Lecturer in Museums Studies and Museums in the Metaverse Researcher, who co-organised the workshop) and Co-Lead of CREATe’s Digital Technologies & Humanism theme, Maria provided the basis for a stimulating discussion on how AI is revolutionising the cultural heritage sector.

 

Lightning Talks

The workshop included a combination of online and face-to-face lightning talks from experts of various backgrounds, including academics, librarians, and industry representatives. The topics covered were diverse, such as how we should train AI to create art and present history, manage information and rights, serve the audience, and sustain the environment. The recording of all talks is available on CREATe’s YouTube channel here.

Marion Carré, artist, programmer, CEO & co-founder of Ask Mona (a company using artificial intelligence to facilitate knowledge transmission), outlined how AI is transforming visitor experiences in cultural institutions through personalised interactions. This AI Assistant known as ‘Mona’ they have created can be used in the pre-visiting, visiting, and post-visiting gift-shopping stage. This allows visitors to ask questions and receive curated information about the cultural assets from museum teams in a personalised way, making the cultural experience more accessible and inclusive. However, Marion highlighted the challenge of controlling the data that feeds into these AI systems, which is inarguably a sector-wide issue.

Marion Carré (CEO & co-founder of Ask Mona) showed how their “Ask Mona” AI assistant can contribute to the museum experience (Source: Marion Carré).

Amy Adams, Collection Information and Access Manager at the National Museum of the Royal Navy, shared insights from the use of AI for collections management. Amy discussed the Museum’s efforts to manage over 2.5 million collection items using AI. One notable example is the preservation of HMS Victory, where AI has aided in the conservation of the ship through an AI-based algorithm which matches images of HMS Victory from different locations and adds them to the ship’s 3D model. Amy emphasised the importance of the training process and validating AI-generated data, stressing the need to keep an eye on the training results and what is fed to the programme and be cautious of biases in training AI systems. She also invited those interested in AI in CH to keep involved by joining the AI4LAM JISC mail list she co-convenes.

Sarah Cook, Professor of Museum Studies at the University of Glasgow, discussed the use of AI in contemporary art. Sarah shared examples like the AI-driven restoration of Rembrandt’s Night Watch masterpiece, where AI was used to mimic Rembrandt’s style. In addition to restoration, it is clear that AI is becoming a new tool for artists in its own right (e.g. Sara mentioned the artist (and engineer) Tega Brain, who is exploring the creative potentials of AI). She further raised the issue of unsupervised AI machines producing hallucinations and using shoddy code, as well as the question of engaging with AI in an environmentally sustainable manner. Sarah suggested that art museums should consider carefully the engineering and ecological sustainability of AI, using the example of Museum of Modern Art (MoMA) which had trained AI with their digital collections to demonstrated the potentiality of AI art with ‘Unsupervised’ in Refik Anadol’s Machine Hallucinations project.

Cassandra Kist, Chancellor’s Fellow at the Department of Computer and Information Sciences, University of Strathclyde, discussed how AI can create personalised interactions that humanise technology, analysing further some of the issues that Marion raised in her talk. Cassy mentioned a number of interesting examples of innovative and humanised chatbots that are being used in cultural heritage institutions (CHIs) across the world, such as the chatbot project being undertaken by the Akron Art Museum. Cassy identified main trends in design related to humanising conversational agents, of great relevance to CHIs interested in creating user-centred communication-based services with AI. She emphasised the importance of maintaining an empathetic distance and ensuring safety to enhance user engagement without distractions.

Paul Gooding, Professor of Library Studies and Digital Scholarship at the University of Glasgow, addressed the topic of responsible AI in libraries. Paul focused on how AI, particularly machine learning and large language models (LLMs), are being used to create machine-actionable collections which aim to meet the researchers’ need for digitised collections. Highlighted was the importance of responsible development and governance of AI systems. Throughout the talk, Paul linked his discussion to research, such as the work he co-authored titled, ‘The implications of handwritten text recognition for accessing the past at scale’. He also introduced, iREAL, a new project funded by the AHRC BRAID Scoping to Embed Responsible AI in Context programme, and emphasised the need to prepare machine-readable material to support multivocality. The project aims to not only process Indigenous knowledge but also deal with the right to reply and the right to know.

Paul Gooding (Information Studies, UofG) speaking about responsible AI in libraries (source: UofG Photographic Unit)

Sarah Ames, Digital Scholarship Librarian at the National Library of Scotland, presented online on the revolution AI is bringing to national libraries. Sarah discussed the NLS’s current experimentation with AI, among others, to enhance access to and research of cultural heritage. Examples of this include: detection and categorisation of illustration in chapbooks, an AI toolkit to explore datasets at scale, and Handwritten Text Recognition using Transkribus. The National Library of Scotland hopes to use AI to open new ways of interacting with their collections and archives, and Sarah highlighted also their commitment to the responsible adoption of AI. Sarah also referred to a report on the application of AI in cultural heritage that they have carried out. Yi-Ting was impressed by the map Sarah showed and the sophisticated use of geological data and visualisation.

Sarah Ames demonstrated how the National Library of Scotland analysed geological data and created a visualisation with AI (source: Sarah Ames, NLS)

Finally, William Kilbride, Executive Director of the Digital Preservation Coalition, presented online the long-term implications of AI in digital preservation. William outlined four areas where AI and digital preservation intersect: using AI tools for digital preservation, employing digital preservation tools to aid AI, preserving AI outputs, and ensuring the reproducibility of AI systems. Speaking of preparing machine-readable material, William mentioned the helpful guide prepared by DPC on Computational Access: A beginner’s guide for digital preservation practitioners. His presentation highlighted the instalment of AI-based tools in workflow, which acted as a suitable bridge between the talks and the group discussions afterwards.

 

Group discussions

During the second hour of the workshop, attendees of various backgrounds and areas of expertise including librarians, legal experts, and cultural heritage practitioners and academics, were split into groups to engage in discussion related to the presentations but also bring in their own research interests and perspectives.

Workshop participants engaging in group discussions (source: UofG Photographic Unit)

Lucy, who was appointed group 3 convened by Sarah Cook, was part of a discussion centred around the interests of independent artists. Legal solutions offered were anti-AI clauses, where participants highlighted the importance of legal mechanisms in creating a level-playing field and a strong clause to empower artists in negotiations and legal. The group was further challenged to seek legal mechanisms in order to provide the right to authors and artists to object to their work(s) being subject to AI, in a meaningful capacity.

Panagiotis, who was appointed group 4 convened by Paul Gooding, discussed safeguarding artists’ rights and keeping control of work beyond what is already provided by legal mechanisms. Emphasis was placed on keeping CHIs accountable in interacting with work that is out of copyright in order to ensure proper cultural heritage preservation. Legal solutions proposed included a French approach to moral rights as inalienable, unwaivable and imprescriptible to safeguard the integrity of out-of-copyright works.

 

Reflections on #AIinCH

As a current LLM student at the University of Glasgow, Lucy is in the early stages of writing her dissertation which is broadly on cultural heritage, and more specifically on issues regarding digital reproductions of cultural heritage works in the public domain. She has found immersing herself in all aspects of cultural heritage an intellectually enriching experience. For her, the #AIinCH workshop highlighted the humanised and personalised approach being taken towards AI in the cultural heritage sector. She was particularly interested in Paul Gooding’s discussion on the digitisation of collections and archives using AI. His insights explored how digitisation meets the research needs of scholars but also the impact on local communities providing this material. For her, this discussion also underscored the broader societal benefits of making digital cultural heritage accessible online, such as enhancing and preserving cultural heritage, encouraging civic engagement and cultural participation, and enhancing access to knowledge. This experience greatly broadened her knowledge of the cultural heritage sector and gave her a critical understanding of how CHIs are navigating the integration of AI into their sector.

Panagiotis is currently writing his dissertation on reforming the moral right of integrity in light of the AI enhancement of films, balancing the rights of filmmakers in protecting their work with the ability of CHIs to use AI to restore damaged films for the purpose of archiving. Whilst it was initially straightforward to make the distinction between ‘enhancement’ and ‘restoration’, Sarah Cook’s discussion around the AI’s flaws, including the use of shoddy code and hallucinations has triggered more questions around protecting works under restoration from such errors. He also found particularly interesting Amy Adams’ discussion regarding awareness of machine biases. Considering the predominance of colonial history in the west, it is important that CHIs remain aware of the inherent biases which machines cannot use their own judgment to correct. As such, it is significant that at least minimum human intervention is used to guide art resulting from AI technologies, not merely in cases such as preserving the HMS Victory, but also in restoring and interacting with works stemming from other cultures, especially those that have historically been subjected to colonial rule.

Lucy and Panagiotis also gained an appreciation of the importance as legal researchers interacting with experts in other fields, such as those in the cultural heritage sector, in order to better inform their views and broaden their research through understanding the issues that non-legal stakeholders are facing.

For Yi-Ting, #AIinCH was a great success. She thought that the speakers pointed to different directions with rich literature, case studies, and areas to explore. For her future research, she aims to pay more attention to how AI is trained to present history and how museums deal with source communities’ right to reply and the right to know.

The ‘AI in Cultural Heritage’ Workshop by the Digital Cultural Heritage Lab and CREATe promoted the multifaceted role of AI in preserving, enhancing, and engaging with cultural heritage. From improving visitor experiences to aiding in the meticulous preservation of artifacts, AI is proving to be a transformative tool in this area. There is no doubt that its usefulness will only grow. However, as highlighted by the speakers, it is crucial that the integration of AI into the cultural heritage sector is paralleled with caution regarding the challenges of data control bias, and responsible development. It is only if such challenges are raised and addressed, that the sector will be able to fully harness and enjoy AI’s potential. In addition, this event underscored the importance of interdisciplinary collaboration in navigating the complexities of AI in cultural heritage.