Museums and Artificial Intelligence

On this page we present a summary of the international conference “Innovation and Integrity: Museums paving the way in an AI-driven society" that took place on 20 March 2024 in Brussels, Belgium.

Results and recommendations

The main outcome of the international conference are the three recommendations for policy makers. Together with experts and four of its members, NEMO has developed the recommendations to support museums to tap their full potential and to find their role in the rapid development of artificial intelligence (AI). Museums are excellent places to negotiate technological developments along with the public and offer spaces to learn, experience, and build knowledge. 

  1. A political vision for museums and cultural heritage in an AI-driven society  
    Recognising the unique position of museums and cultural heritage as pillars of trust within society, it is imperative to integrate them into a regulatory framework. Artificial intelligence in museums needs to be addressed and shaped so that technological developments do not simply reshape museums from the outside. Collaborative efforts between governments, regulatory bodies, and museum professionals can ensure that museums play a pivotal role in the development of ethical practices related to emerging technologies.
  2. Financial investments to apply AI successfully in the Public Cultural Domain
    Financial resources must be allocated for infrastructure, equipment and highly qualified human resources, enhancing museums’ professional capacities.AI needs to source high-quality, interoperable data and properly described metadata. Copyright issues must be resolved. Museum professionals need adequate skills to perform these tasks, to keep pace with rapidly evolving AI capabilities and to address sector-specific concerns. Furthermore, standing commitments to support the cultural heritage sector should be expanded to ensure the quality and quantity of digitalisation required by Cultural Heritage Data Spaces and the European Collaborative Cultural Heritage Cloud.
  3. Establishment of a European AI innovation hub for cultural heritage 
    To foster creativity, innovation and collaboration, to centralise expertise and knowledge and to face challenges for the sector associated with AI, there is a need for a dedicated competency centre in Europe. This space would serve as a hub to bring together expertise and practices, knowledge and resources in a network of and for professionals, ensuring digital innovation and development across the diverse European Cultural heritage sector - in alignment with the values of human-centred design, privacy, and open-source practices. 

Innovation and Integrity: Museums paving the way in an AI-driven society

The international conference was co-organised on 20 March 2024 by NEMO, FARO, ICOM Wallonie-Brussels, ICOM Flanders and the House of European History to facilitate dialogue between the museum sector and policy makers including perspectives from local, community, national and European level.

Participants got to investigate museums’ use of AI in terms of art form and cultural practice development, data creation, manipulation, and interpretation. Participants also got to explore how museums and cultural heritage at large can help counter fake reality and disinformation. Additionally, the conference discussed AI in terms of new ways of work, new competences and changing business models for the sector.

Welcome

By Iliana Ivanova European Commissioner for Innovation, Research, Culture, Education and Youth, and Sabine Verheyen, Member of the European Parliament and Chair of the Committee on Culture and Education (CULT)

Iliana Ivanova, European Commissioner for Innovation, Research, Culture, Education and Youth, sent a video message of welcome and support to her colleagues present at the conference along with all participants. She highlighted the opportunities AI offers museums as well as the ways in which the EU seeks to support researchers, institutions, and professionals to innovate with emerging technologies.

Sabine Verheyen, Member of the European Parliament and Chair of the Committee on Culture and Education (CULT) took the stage and underscored the profound impact of technological advancements and the digital revolution on society, emphasising the increasing role of technology in daily life. Verheyen mentioned the EU's recognition of AI's significance, citing the enactment of the world's first AI act. Museums were portrayed as vital in addressing societal challenges such as fake news and fostering understanding among diverse communities. Verheyen stressed the importance of dialogue between museums and policymakers to ensure a smooth integration of AI and digitalisation, envisioning AI as a crucial tool for museums' future development.

Introduction by the partners

With Olga Van Oost, Director (FARO), Julia Pagel, Secretary General (NEMO), Alexandre Chevalier, Chair (ICOM Wallonia-Brussels), Isabelle Vanhoonacker, Chair, (ICOM Flanders). Moderated by Johannes Bernhardt, Lecturer (University of Konstanz).

While opening the partner panel discussion, Johannes Bernhardt highlighted a growing focus on AI since the publication of ChatGPT. He emphasised the need to view AI soberly, as a technology for pattern recognition, amidst distracting dystopian and utopian narratives. The panellists then discussed the profound impact of AI on museum processes and accessibility, noting the necessity of human intervention to address cultural sensitivities and biases. Benefits such as improved research and collections management were acknowledged, but panellists cautioned about the digital divide and the need for comprehensive datasets. They urged for museum training to include AI education and emphasised the challenge of attracting expertise. Furthermore, they emphasised the significance of networking among museum professionals to bridge knowledge gaps and collectively address challenges.

Keynote speech

By Paul Keller, Director of Policy (Open Future Foundation). Moderated by Sergio Servellón, Director (Felix Art and Eco Museum).

In the keynote speech, Paul Keller (download presentation) began by addressing the multifaceted nature of AI technology, emphasising its impact on human perception and the way individuals engage with information. He noted the initial limitations of AI but highlighted its potential to significantly shape how people perceive the world. Keller raised concerns about the profit-driven motives behind AI development by major corporations, cautioning against the idea that AI learns in the same way humans do. Museums were discussed in the context of AI regulation and Keller highlighted the vast difference in scale between museum collections and the data used to train AI models; museum collections being much smaller than the vast data sets fuelling generative AI. He also delved into the complexities of bias in AI datasets, noting that while museums may contribute to mitigating bias through their collection processes, they can also inadvertently perpetuate bias.

Keller cautioned against hastily integrating AI into museum collections without adequate consideration for trust preservation, advocating instead for addressing existing problems with AI rather than using it for experimentation. Furthermore, he underscored the importance of transparency in data management and copyright, pointing out forthcoming regulatory changes that would require AI developers to disclose data sources. He proposed structuring access arrangements and considering taxation mechanisms to ensure equitable access to digital infrastructures.

Lastly, Keller emphasised the broader role of museums and cultural heritage institutions as integral components of the digital public infrastructure, advocating for their inclusion in broader European initiatives. During the Q&A session following the keynote presentation, participants discussed issues surrounding access and copyrights, highlighting the complexities museums face. Concerns were raised about the dominance of private companies in the information landscape. The need for caution in integrating AI into education without a thorough understanding of which data fuels the generative mechanisms was also mentioned. The rapid evolution of discussions around AI legislation was noted, acknowledging the necessity for timely regulatory action.

Policy perspectives: EU, Belgian Federal, French and Flemish Community

With Hans van der Linden, Policy Advisor, Department of Culture, Youth and Media (Flemish Government), Jean-Louis Blanchart, Acting adjunct General Director, Heritage Service (French Community of Wallonia and Brussels), Giuliana De Francesco, Policy Officer, European Commission (Directorate-General Research and Innovation (DG RTD)). Moderated by Lars Ebert, Secretary General (Culture Action Europe).

Alongside an introduction of the Ghent Manifesto, Hans van der Linden (download presentation) referenced policy from a digital transformation position paper, noting the absence of one specifically for AI. He emphasised the importance of supportive ecosystems to ensure that data benefits cultural organisations and museums. Linden cautioned the interference of AI with the sector but advocated the potential for audience engagement and data management. He called for an EU ecosystem to consider public values when collaborating with commercial companies and stressed the importance of cultural institutions learning how to navigate technology and being supported by funding schemes for innovation. Additionally, he cautioned about the potential deepening of the digital divide.

Jean-Louis Blanchart (download presentation) delved into crucial issues that the next government must address regarding the intersection of AI and cultural production. He spoke on the necessity of regulating AI's influence on cultural production, highlighting both negative and positive aspects. Additionally, Blanchart emphasised the importance of ensuring visibility for cultural content from the FWB (Federation Wallonia-Brussels) in recommendation algorithms, necessitating the collection and management of high-quality data. He also questioned whether changes in the legal regime for intellectual property law were necessary, particularly in determining the status of AI products and considering exemptions for Text and Data Mining. Blanchart also addressed the pressing issue of fake news and the important challenges regarding reliability of information disseminated to the public.

In her presentation, Giuliana De Francesco(download presentation) underscored the political significance of AI, highlighting its prioritisation by the European Commission. She emphasised the AI Act as the first legal framework on AI globally, as well as the establishment of an AI office by the commission, which is designed to serve as a hub of expertise on AI. Quoting, "We can build on our areas of strength including excellent research, leadership in some industrial sectors like automotive and robotics, a solid legal and regulatory framework, and very rich cultural diversity at both regional and subregional level," Francesco emphasised the strengths upon which the EU can leverage its AI endeavours. Providing insights into recent EU actions and priorities, she outlined the work programme for 2023-2024, emphasising that digital innovation spans multiple Directorates-General (DGs). She recommended resources such as The Digital Deal Podcast as well as the planned AI 4 Culture access platform. Francesco concluded by advocating for museums to actively engage in discourse on AI, positioning them as crucial in discussions and decisions regarding AI's impact on society, she stressed that this discourse is not distant but rather already changing our lives.

Panel 1: Critical technology discourse in the museum

With Oonagh Murphy, Senior Lecturer in Digital Culture and Society (Institute for Creative and Cultural Entrepreneurship, Goldsmiths, University of London), Thierry Dutoit, Professor (UMONS) and Director (Numédiart) and Nico Verplancke, Director (meemoo, the Flemish Institute for Archives). Moderated by Peter Bary, General Director (Museum M Leuven).

Drawing from her background in museum studies and arts management, Oonagh Murphy(download presentation) emphasised the need for dialogue between these sectors to effectively address the operational impacts of AI on everyday life. Murphy provided recent examples, such as facial recognition technology tests, which highlighted issues of bias and operational shortcomings. She underscored the importance of understanding the practical applications of AI beyond its theoretical power, pointing out that many institutions already engage with AI through email services or visitor mapping. Arguing that museums are inherently data-centric institutions (or even, as she claimed, the first datasets), Murphy asserted that they possess valuable experience in creating and organising data. Drawing parallels between cataloguing practices of the past and current data labelling processes, she positioned museums as crucial to navigating the complexities of AI. Furthermore, Murphy discussed the relevance of the Museums and AI toolkit aimed at guiding institutions in implementing AI projects. She urged museums to lead the conversation on AI by embracing transparency, critical discourse, and digital literacy. When asked about how to advocate for funding for necessary projects, she replied that the boring projects are the ones that make the biggest difference; funding better dataset processes will support AI projects and capacities, and fortunately there is a growing understanding of digital as critical infrastructure. Additionally, Murphy stated that digital security is the more critical resource.

In his presentation, Thierry Dutoit(download presentation) began with an explanation of the variations of AI, stating the distinctions between perceptive, active, and generative AI. He outlined their functions and potential applications for museums. Dutoit included the optimisation of collection management and bolstering of security measures, as well as fostering audience engagement through personalised experiences and facilitating educational initiatives as potential benefits. Dutoit illustrated the real-world impact of AI projects with diverse examples from museums. By weaving together these tangible examples, he illuminated the transformative potential of AI within the museum landscape. In his concluding remarks, Dutoit offered a perspective on the future trajectory of museums in relation to AI technologies. He suggested an increase in the prevalence of avatars in museum spaces, for example, drawing from his belief that museums are fundamentally about human experiences rather than objects. When asked about the strategy at play in the museums which developed the earlier described AI projects, Dutoit shared that many of the projects were initiated in line with the European Year of Cultural Heritage.

In his presentation, Nico Verplancke(download presentation) discussed the application of AI within the realm of museums, particularly focusing on the digitisation, archiving, and management processes by highlighting the collaborative efforts undertaken by meemoo, the Flemish Institute for Archives. Verplancke described the significance of the scale of digital processes and how meemoo leveraged expertise and resources from a diverse set of partners. He delved into the development of meemoo, addressing the challenges associated with metadata management for digitised archival materials. Verplancke detailed the development of a facial recognition system, discussing the decision-making process behind choosing to build an in-house solution rather than relying on external services. He underscored the importance of addressing privacy concerns and ethical considerations throughout the project, involving stakeholders and acting on feedback. Verplancke stressed the importance of the overarching digital strategy and provided valuable insights about practical application and the importance of collaboration, transparency, and strategic planning in such ‘in-house’ endeavours. When the moderator turned to the audience to ask whether building our own metadata sets based on values determined by museums rather than relying on big companies was better, it was criticised that even this approach will hold a bias. Verplancke replied that everything with a human input, including AI, will have biases, therefore the transparency in how decisions are made is key. He also shared examples of tasks that were simply too large for the museums’ human capacity, which could now be completed with the help of technology.

Panel 2: Museums and democracy in an increasingly AI driven world

With Johan Oomen, Head of The Netherlands Institute for Sound and Vision R&D Department and researcher (Web and Media group of VU University Amsterdam), Veerle Vanden Daelen, Curator and director of collections & research (Kazerne Dossin: Memorial, Museum and Research Centre on Holocaust and Human Rights) and Marion Carré, CEO (Ask Mona). Moderated by Guido Gerrichhauzen, Head of Learning and Outreach (House of European History).

In his presentation, Johan Oomen (download presentation) offered insights into their work within the 'AI for media' project. They highlighted the drafting of a white paper and outlined key themes under 'Culture for AI,' which included strategies for museums to support responsible AI use and the need for trustworthy AI requirements to be translated into organisational strategies.

Drawing from earlier discussions on acquiring or developing datasets, Oomen emphasised the importance of understanding collections to tailor datasets effectively. He shared examples of AI applications, such as generative interfaces enhancing archive use. Additionally, Oomen proposed strategies for responsible AI use and advocated for supporting sovereign digital public infrastructure. He stressed the need for cultural organisations to demystify AI and promote diverse representations. Oomen highlighted the educational potential of heritage organisations, citing initiatives like the "AI parade" hosted by libraries. He suggested museums as safe spaces for facilitating discussions around AI and its societal implications, advocating for broader public engagement.

Veerle Vanden Daelen highlighted the potential applications of AI in museums while also emphasising the importance of cautious implementation and acknowledging the varying needs and challenges faced by very different institutions within the museum sector. She acknowledged the benefits AI offers, such as aiding in research and oral history preservation, but cautioned against relying on AI without human curation. She stressed the necessity of verifying sources and maintaining transparency to uphold trustworthiness. Ethical considerations, including GDPR compliance and data privacy, were addressed, particularly concerning sensitive information like oral histories. Ethical questions persist, especially regarding data ownership and usage rights. Challenges for smaller institutions were highlighted, as well as the struggle to introduce new tools while ensuring adherence to FAIR principles (Findability, Accessibility, Interoperability, and Reusability). Moreover, the role of museums in combatting misinformation, particularly concerning sensitive topics like the Holocaust, was discussed; while a museum cannot combat all external misinformation campaigns, they must remain a reliable resource in and of themselves. Daelen concluded by advocating for safe learning spaces to navigate biases and uphold trust in museums amidst increasing use of AI. Transparency and traceability were emphasised as essential to maintaining the integrity and reputation of the profession.

Marion Carré delved into the evolving landscape shaped by AI, highlighting two significant shifts: the recent broadening of AI usage and the now constant exposure to AI-generated content. This accessibility to AI tools without specialised knowledge has led to a proliferation of generated content across various platforms, blurring the distinction between human-created and AI-generated material. Carré underscored the societal impact of this widespread exposure to AI, noting the polarisation of opinions ranging from extreme optimism to deep pessimism. She shared a recent project, aimed at assessing people's ability to distinguish between AI-generated and genuine archive content, highlighting the challenge of combating misinformation. Furthermore, she highlighted the widening gap in AI literacy and access to employment opportunities, suggesting the role of cultural organisations as hubs for lifelong learning and trusted sources of information. Carré shared an ongoing project exploring audience engagement through AI experimentation, illustrating the practical applications of AI within cultural institutions.

 

Wrap-up by organising partners

With Olga Van Oost, Director (FARO), Julia Pagel, Secretary General (NEMO), Alexandre Chevalier, Chair (ICOM Wallonia-Brussels), Isabelle Vanhoonacker, Chair, (ICOM Flanders). Moderated by Johannes Bernhardt, Lecturer (University of Konstanz).

The final panel started with a clip of The Zizi Show by the artist Jake Elwes, a project about submerging biases in drag culture with AI, recently exhibited at the V&A, as a topical artistic intervention related to the discourse of the day.

Bernhardt then highlighted recurring themes from the conference, emphasising trust and transparency as crucial elements to be handled with care in the context of AI. Transparency was identified as essential for museums’ AI strategy, especially considering the lack of transparency in much commercial AI technology. Art was also discussed as a means to consider ethical concerns surrounding AI, suggesting the inclusion of artists in AI-related conferences across all sectors. Paul Keller's opening remarks were revisited, as Bernhardt underscored museums' potential to engage with AI, despite contributing only a fraction of the data used by AI systems. It was suggested that museums should be more conscious of the value of their cultural data and negotiate accordingly (and collectively) with the commercial sector. Additionally, the need to protect museum collections from overly open data approaches was emphasised, with a focus on considering human use and experience. He also advocated for a solutions-based approach to AI utilisation rather than adopting it simply because it's trending and urged careful consideration of trust, transparency, and purposeful application of AI within the museum and cultural sector.

The panel discussion revolved around the three proposed recommendations established to support museums to effectively engage with AI.

  1. A political vision for museums and cultural heritage in an AI-driven society  
    Recognising the unique position of museums and cultural heritage as pillars of trust within society, it is imperative to integrate them into a regulatory framework. Artificial intelligence in museums needs to be addressed and shaped so that technological developments do not simply reshape museums from the outside. Collaborative efforts between governments, regulatory bodies, and museum professionals can ensure that museums play a pivotal role in the development of ethical practices related to emerging technologies.
  2. Financial investments to apply AI successfully in the Public Cultural Domain
    Financial resources must be allocated for infrastructure, equipment and highly qualified human resources, enhancing museums’ professional capacities.AI needs to source high-quality, interoperable data and properly described metadata. Copyright issues must be resolved. Museum professionals need adequate skills to perform these tasks, to keep pace with rapidly evolving AI capabilities and to address sector-specific concerns. Furthermore, standing commitments to support the cultural heritage sector should be expanded to ensure the quality and quantity of digitalisation required by Cultural Heritage Data Spaces and the European Collaborative Cultural Heritage Cloud.
  3. Establishment of a European AI innovation hub for cultural heritage 
    To foster creativity, innovation and collaboration, to centralise expertise and knowledge and to face challenges for the sector associated with AI, there is a need for a dedicated competency centre in Europe. This space would serve as a hub to bring together expertise and practices, knowledge and resources in a network of and for professionals, ensuring digital innovation and development across the diverse European Cultural heritage sector - in alignment with the values of human-centred design, privacy, and open-source practices. 

Throughout the day, participants emphasised the need for a political vision to guide the sector forward, recognising museums' pivotal role. Collaboration at the highest levels, as exemplified by the EU AI Act, was highlighted as crucial for addressing AI's impact on society comprehensively. However, there was a concern about the dual influence of governments and large private companies. The concept of sovereign digital infrastructure raised questions and established the need for ongoing sector involvement in related discussions. Regarding urgent needs for museums, digitisation remained a priority, with subsidies supporting these efforts. Participants stressed the importance of investing in human resources and tools to prevent museums from falling behind. Navigating public-private partnerships was deemed crucial, as they are challenging due to the power dynamics involved. The idea of an innovation hub for museums to interact with AI technology garnered interest: participants envisioned such a hub facilitating collaboration, innovation, responsible data collection and use, and knowledge sharing at a European level. They emphasised the need to integrate AI considerations into broader sector challenges. Additional audience comments highlighted the importance of maintaining museums as human-centred spaces amidst increasing AI integration. Connecting the future results of various European projects and initiatives, ensuring sustainability, and respecting the different speeds of organisations were also noted as crucial considerations for the sector's future. In conclusion, the panel discussion underscored the need for a political vision to guide museums in navigating AI's impact, emphasising collaboration, responsible data use, and support for innovation across the sector.