Interactive Heritage: The Role of Artificial Intelligence in Digital Museums

Abstract

Museum use of artificial intelligence (AI) is becoming increasingly common, but its contribution to museum attendance is yet to be confirmed. This paper investigates whether the adoption of AI impacts museum visitation using data from 19 museums. Statistical analyses, including ANOVA and Spearman correlation, were conducted to determine if the use of AI has significant effects on visitors. The findings indicate no statistically significant difference between museums that use AI and those that do not (ANOVA: p = 0.263, F = 1.34), but the Spearman correlation (r = 0.448, p = 0.055) indicates a moderate positive correlation that is not statistically significant. The findings suggest that AI enhances visitor experience rather than increasing attendance. Additionally, this study proposes a conceptual framework for AI prototyping in museums. The study contributes to the ongoing debate on AI in cultural institutions by emphasizing that future research should incorporate longitudinal studies and qualitative visitor feedback in order to capture the overall impact of AI on engagement and sustainability in museums.

1. Introduction

Artificial intelligence is a field within computer science, aimed to develop computing systems that can simulate intelligence based on relevant programming and algorithms [1]. The relationship between AI and art began as early as the 1960s when scientists and artists created abstract art using algorithms. Michael Noll and Harold Cohen were pioneering artists who used this method [2]. Harold Cohen developed AARON, a computer program that generated digital art from rules and patterns coded by him [3]. The initial experiments demonstrated how machine learning methods would later be applied to generate artistic content. The rapid growth of AI has led to a phenomenal increase in its use in the museum setting [4]. In recent years, AI has not only been used to increase the number of visitors but also to improve their experience. Presentation in exhibitions is complemented by AI technologies, facilitating the use of media such as touch screens, virtual reality (VR) devices, and tour programs [5,6]. The integration of AI with digital collections and online exhibition planning offers a new way of educational outreach, enhancing visitors’ interaction with exhibits [7]. The National Museum in Copenhagen possesses a chatbot with a digital guide to help visitors before and while visiting the museum. The chatbot is able to answer questions about anything from opening hours to current exhibitions [8]. Research emphasizes the potential of AI in personalizing exhibitions based on individual interests and increasing engagement through tailored recommendations [9]. Hyun-Kyung Lee proposes an AI-based archival system that converts design sentiments into data and uses metadata to translate images in museums [10]. Jiang et al. [11] created image recognition systems using deep learning algorithms to automatically classify cultural relics and furnish more precise exhibit explanations to visitors. The Sanxingdui Museum in China uses AI and 3D printing to recreate exhibits [12]. At the “Dali Lives” show in the Dali Museum, Florida, USA, visitors were able to interact with Salvador Dali on a digital screen using AI [13]. Google’s “Arts & Culture” platform offers an AI-powered feature called “Art & Selfie”, which matches user selfies with artworks using artificial intelligence [14]. The robot “Berenson” was developed in 2011. In 2016, during an exhibition at the Quai Branly Museum in Paris, it collected visitor reactions to artworks and transmitted them to a computer [15]. In “Unsupervised”, artist Refik Anadol uses artificial intelligence to analyze and reinterpret over 200 years of art at MoMA [16]. The AI Art Gallery platform presents artworks that combine human creation with AI-generated pieces [17]. Such platforms typically display works of art such as Deep Dream [18] and works completed via machine learning algorithms. [18]. Artificial Paintings is operated by a team that creates and sells paintings with the assistance of AI technology [19]. DATALAND is the first museum in the world to showcase art produced through artificial intelligence technology. Opened in Los Angeles in 2025, the museum explores themes of flora, fauna, and fungi, with a pioneering approach to collaboration between art, science, and technology [20]. The Epigraphic Museum, in cooperation with the Institute of Historical Research, hosted the ITHACA project, a deep learning neural network that helps in the completing, dating, and geographic attribution of ancient Greek inscriptions [21].

The research employs a quantitative design to examine the impact of artificial intelligence on museum attendance in Europe by analyzing 19 major museums that have implemented AI technologies. The study uses ANOVA and Spearman correlation to establish whether AI adoption results in higher museum attendance. The research presents a conceptual framework for AI prototyping in museums while demonstrating how these technologies improve both visitor experience and institutional sustainability. The paper is structured as follows: Section 2 describes the methodological approach; Section 3 presents the results of the data analysis; Section 4 discusses the implications of the findings; and Section 5 concludes with a summary and directions for future research.

2. Materials and Methods

2.1. Research Design

The present study employs a quantitative approach, and statistical analysis methods (ANOVA and Spearman correlation) are used to determine whether the adoption of AI technologies is associated with higher visitor numbers.

2.2. Sample Selection

Twenty of Europe’s most visited museums were selected based on attendance data from Statista (2023) [22]. The museums examined, sorted alphabetically, are as follows: Acropolis Museum, British Museum, Centre Pompidou, Cité des Sciences et de l’Industrie, Galleria Degli Uffizi, Humboldt Forum, Louvre, Musée D’Orsay, Museo Nacional del Prado, National Gallery, National Museum of Scotland, Natural History Museum, Rijksmuseum, Science Museum, State Hermitage, State Russian Museum, State Tretyakov Gallery, Tate Modern, Vatican Museums, Victoria and Albert Museum. Data were collected between 3 and 9 January 2025 using the Safari browser to ensure compatibility and consistency across the museum websites.

2.3. Data Collection

Information was obtained from the content publicly available on the official websites, as well as through internal site searches. Based on the design of this research, the walkthrough method was employed, specifically, a systematic examination and analysis of online environments [23]. Data collection involved two primary methods:

1.

Webpage examination

Researchers examined the categories of web pages that had general information regarding the museum, such as the “About”, “History”, and “Museum” pages or their equivalent description pages, to see whether there was any mention of the use of artificial intelligence in the operations or exhibitions of the museum.

2.

Keyword search and content analysis

A keyword search of websites and content analysis [24] was conducted in this research to analyze the presence and usage of artificial intelligence on the websites of the museums selected for this study. To begin with, specific keywords in the English language were selected in lowercase letters: “artificial intelligence” and “ai”. These terms were then searched on the official sites of the museums using the internal search engines of the websites. After data were collected from searching, a content analysis was conducted to examine the way AI is presented on the museums’ sites.

2.4. Categorization of Museums

The data were divided into two groups: (a) museums that actively incorporate AI into their systems—such as object recognition, immersive applications, visitor tools, restoration processes, virtual tours, or automated documentation; and (b) museums that do not use AI. The data thus obtained were then correlated with museum attendance to determine whether the use of AI has any effect on museum popularity. This approach facilitated the systematic and comparative examination of the presence of AI in the museum setting to the evolution of knowledge regarding its use towards enhancing the cultural experience and managing cultural heritage. Quantitative data analysis was applied, examining the relationship between the use of artificial intelligence and museum visits.

2.5. Statistical Analysis Methods

The quantitative analysis applied two main statistical methods to evaluate the connection between AI usage and museum visitation.

• ANOVA: To measure this relationship, the Spearman rank correlation coefficient (p) was applied. The formula used is as follows:

  p = 1 − (6 × Σdᵢ²)/[n(n² − 1)]

  where dᵢ represents the difference in ranks between AI usage (binary variable: 1 for AI, 0 for non-AI) and number of visitors, and n is the number of museums in the sample. The coefficient was calculated based on the ranked values of both variables to assess the monotonic relationship between them.
• Spearman Rank Correlation Coefficient (p): To examine the monotonic relationship between AI adoption and museum attendance.

3. Results

This research examines the use of artificial intelligence in museums, focusing on the different technologies applied, the trends that are emerging, and the differences in the way diverse cultural organizations incorporate AI. The results provide an overview of contemporary AI applications in the museum context (

Table 1. Museum attendance and AI usage: overview.

	Country	Museums	Websites	Most Visited 2023 (Millions)	Uses AI	Main Uses of AI
1	Paris (France)	Louvre	www.louvre.fr	8.860	No	-
2	Vatican (Vatican City)	Vatican Museums	www.museivaticani.va	6.765	Yes	- Artificial intelligence techniques for data processing and analysis [25].
3	London (United Kingdom)	British Museum	www.britishmuseum.org	5.821	Yes	- Natural language processing; - Semantic AI and linked data [26].
4	London (United Kingdom)	Natural History Museum	www.nhm.ac.uk	5.689	Yes	- Machine learning, computer vision, and natural language processing (NLP) [27,28,29,30,31].
5	London (United Kingdom)	Tate Modern	www.tate.org.uk/visit/tate-modern	4.742	No	-
6	Paris (France)	Musée D’Orsay	www.musee-orsay.fr	3.871	Yes	- Chatbot [32]
7	Madrid (Spain)	Museo Nacional del Prado	www.museodelprado.es	3.338	Yes	- Natural language processing; - Semantic analysis; Machine learning; - Computer vision [33,34,35].
8	St Petersburg (Russia)	State Hermitage	www.hermitagemuseum.org	3.274	No	-
9	London (United Kingdom)	Victoria and Albert Museum	www.vam.ac.uk	3.110	No	-
10	London (United Kingdom)	National Gallery	www.nationalgallery.org.uk	3.097	Yes	- Machine learning; - Natural language processing; - Computer vision; - AI-driven interactive systems [36].
11	South Kensington, London (United Kingdom)	Science Museum	www.sciencemuseum.org.uk	2.957	Yes	Machine learning [37]
12	St Petersburg (Russia)	State Russian Museum	en.rusmuseum.ru	2.900	No	-
13	Florence (Italy)	Galleria Degli Uffizi	www.uffizi.it	2.718	No	-
14	Amsterdam (Netherlands)	Rijksmuseum	www.rijksmuseum.nl	2.703	Yes	- Computer vision; - Machine learning; - Deep learning; - Digital restoration AI; - Data science and knowledge graphs [38,39,40,41].
15	Paris (France)	Centre Pompidou	www.centrepompidou.fr	2.622	No	-
16	Edinburgh (United Kingdom)	National Museum of Scotland	www.nms.ac.uk/national-museum-of-scotland	2.187	No	-
17	Moscow (Russia)	State Tretyakov Gallery	www.tretyakovgallery.ru	2.100	No	-
18	Athens (Greece)	Acropolis Museum	www.theacropolismuseum.gr	1.904	No	-
19	Berlin (Germany)	Humboldt Forum	www.humboldtforum.org	1.700	No	-

), highlighting the role of technology in visitor experience, collection management, and research activities. At the same time, limitations of the method are highlighted, such as the availability of information on official museum websites. Another limitation is the unavailability of historical data, as museums do not explicitly report the exact dates when they started employing AI technologies. Therefore, it was not possible to conduct within-institution comparisons of visitor attendance before and after AI implementation. The geographical distribution of the museums examined shows that the UK has the largest number of museums incorporating AI, with a total of five museums: the British Museum, the Natural History Museum, the Tate Modern, the Victoria and Albert Museum, and the National Gallery. In France, four major museums are included in the study: the Louvre, the Musée D’Orsay, the Centre Pompidou, and the Cité des Sciences et de l’Industrie. Russia also has three major museums: the State Hermitage Museum in St Petersburg, the State Russian Museum, and the State Tretyakov Gallery in Moscow. Spain, Italy, the Netherlands, Germany, Greece, and the Vatican each have one museum examined in the study, with institutions such as the Museo Nacional del Prado in Madrid, the Galleria Degli Uffizi in Florence, the Rijksmuseum in Amsterdam, the Humboldt Forum in Berlin, the Acropolis Museum in Athens, and the Vatican Museums. The Cité des Sciences et de l’Industrie museum was removed from the final list of museums as it did not contain a search tool and was not included in the statistical calculations.

The data showed that 8 of the 19 museums (42%) have adopted the use of AI in their operations while 11 museums (58%) have not adopted it in their operations. This shows that although the use of AI in the cultural sector is increasing, the rate of adoption is still low and varies across museums (Figure 1). The use of AI in application is uneven with some countries having more museums that employ AI in their operations. The UK has four of the eight museums using AI (British Museum, Natural History Museum, National Gallery, and Science Museum), demonstrating its significant technological investment in the cultural sector. Spain (Museo Nacional del Prado) and the Netherlands (Rijksmuseum) have also adopted AI, with applications ranging from natural language processing to digital restoration of artworks. In the UK it is more even with AI-using museums receiving 2.96 million to 5.82 million visitors. This means that AI is used by several major UK museums but it has not led to a clear rise in visits. In Spain, the AI-used Museo Nacional del Prado receives 3.338 million visitors; while in the Netherlands the AI-used Rijksmuseum pulls in 2.7 million visitors. AI can be employed to improve the visitor experience but otherwise is not a determining factor in attracting more visitors. On the other hand, in Germany, Greece, and Russia, where none of these museums have AI, the numbers of visitors are fewer. The Humboldt Forum in Germany, for example, has 1.7 million visitors, the Acropolis Museum in Greece has 1.9 million visitors, and the State Tretyakov Gallery in Russia has 2.1 million visitors.

To further explore potential regional differences, the museums were grouped by country and separated according to AI usage. As shown in

Table 2. Average annual visitors per country by AI usage in museums.

Country	Uses AI	Average Visitors
France	No	5.741
	Yes	3.871
Germany	No	1.700
Greece	No	1.904
Italy	No	2.718
Netherlands	Yes	2.703
Russia	No	2.758
Spain	Yes	3.338
United Kingdom	No	3.346
	Yes	4.391
Vatican City	Yes	6.765

, the museums in the UK that use AI have on average 4.39 million visitors annually, higher than those without AI (3.35 million). However, in France, the opposite trend is observed, with non-AI museums showing a higher average (5.74 million) than those with AI (3.87 million). This inconsistency across countries supports the study’s conclusion that AI use alone does not drive visitation, and other factors such as cultural importance or location play a more significant role.

3.1. Relationship Between Artificial Intelligence and Visitor Attendance

The results indicated a Spearman correlation coefficient of 0.448 and a p-value of 0.055 between museum visitation and the use of artificial intelligence in museums. This indicates a moderate positive correlation between the implementation of AI and museum visits. However, given that the p-value is just above 0.05, the correlation is not significant at the 0.05 level. This means that although the museums that incorporate AI appear to have more visitors, the data are not strong enough to confirm this result with statistical certainty. Moreover, the ANOVA revealed no statistically significant difference in visitation between the AI and non-AI museums (F-statistic: 1.34, p-value: 0.263). This suggests that the use of artificial intelligence does not have a major impact on people attending museums. Instead, other elements, like the historical and cultural relevance of the museum, are more likely to affect the attraction of visitors to it. Thus, the museums that do use AI are not statistically more likely to have higher levels of visitation than those that do not. For the analysis by country, only the UK would allow a direct comparison, as it is home to four AI museums and two without AI. For the ANOVA between the two categories for the UK, we see an F-statistic of 0.131 and a p-value of 0.736, indicating that there is no significant difference in attendance among the museums having AI and without AI in the country. Moreover, a correlation of Spearman was performed, and the results are presented below: Spearman correlation coefficient = 0.0, p-value = 1.0. This indicates that there is no correlation between AI use and museum attendance in the UK. In addition, a p-value = 1.0 means that the result is completely non-statistically significant; thus, AI use does not influence visitation to this museum. Clearly, AI can add value to the visitor experience; however, currently, it does not lead to increased visitation. In this scenario, while there is a correlation, the high variability in visitation among museums likely makes it challenging for ANOVA to find statistically significant differences. Other factors, meanwhile, including the reputation of the museum, its location, the size, and the significance of its collection, may have a stronger impact on visitation than AI. As such, AI and visitation are positively related at this point; however, this correlation is too weak to be defined as statistically significant in the analysis. The results underline the challenges of making sense of diverse influences shaping museum visitation and call for further studies to investigate the weight of factors as diverse as the museum-going experience, cultural tourism trends, and the impact of digital technologies on museum attendance.

3.2. Categorization of AI Technologies Used in Museums

The categorization of museums that use AI is presented in

Table 3. Categorization of museums by type of AI technology.

AI Type(s)	Museum	Category
Chatbots	Musée d’Orsay	Visitor interaction
Computer vision	Natural History Museum and Rijksmuseum	Collection management
Image analysis	Prado Museum	Restoration/Research
Knowledge representation	Vatican Museums	Visitor interaction
Machine learning	Prado Museum	Restoration/Research
Metadata automation	Victoria and Albert Museum	Collection management
Natural Language Processing	British Museum, Musée d’Orsay, and Vatican Museums	Visitor interaction
Recommendation systems	British Museum and National Gallery	Visitor interaction
Semantic web	Rijksmuseum	Collection management and visitor interaction

according to the type of AI technology used and its main function. The majority of cases are related to visitor interaction, including chatbots, NLP, and recommendation systems. Other applications include collection management (e.g., computer vision and metadata automation) and research or restoration activities (e.g., machine learning and image analysis). This classification shows the different ways in which AI is integrated into museum operations.

The effects of AI technologies on museums will vary depending on the specific function of each technology. The implementation of NLP and recommendation systems enables personalized visitor experiences which result in greater satisfaction and inclusivity. Computer vision and metadata automation systems are designed to assist internal operations by managing collections and making digital content accessible. Machine learning and image analysis enable museums to innovate through research and restoration activities that support conservation and scholarly interpretation. The strategic implementation of AI requires organizations to match their goals with their available resources.

3.3. AI Grouping of Applications in Museums Based on Frequency of Occurrence

From the analysis of data regarding the use of AI applications in various museums, the following categories emerge, sorted by their frequency of occurrence.

Technologies with a high frequency of occurrence:

• Machine learning (6 occurrences);
• Natural language processing (NLP) (5 occurrences);
• Computer vision (4 occurrences).

These technologies appear more often because they have reached a high level of technological maturity and have extensive support from both the scientific community and the market. The researchers proceeded to Scopus search with the following keywords: “Machine learning and ai and museum”, “natural language processing and ai and museum”, and “computer vision and ai and museum” and yielded 493, 126, and 22 articles on them, highlighting their strong presence in academic research and practical applications in museum contexts.

Technologies with medium frequency of occurrence:

• Semantic analysis and linked data (2 occurrences);
• Chatbots (2 occurrences).

These technologies are used but not widely, probably because they require more customization and expertise. However, they can be relatively easily integrated into the existing infrastructure to improve the visitor experience. The paper proceeded to Scopus search with the following keywords: “semantic analysis and linked data and ai and museum” which yielded 20 articles and “chatbot and ai and museum” which yielded 15 articles.

Technologies with low frequency of occurrence:

• Deep learning (1 occurrence);
• VR and AR (1 occurrence);
• Digital restoration AI (1 occurrence).

These applications are still in the development phase or trial implementation in a limited number of museums. Limited usage might be due to higher costs, complexity of implementation, and the need for specialized staff and equipment. The researchers proceeded to search Scopus with the following keywords: “deep learning and ai and museum”, “virtual reality and ai and museum”, “augmented reality and ai and museum”, and “digital restoration and ai and museum” which returned 25, 66, 42, and 3 articles, respectively.

3.4. Conceptual Framework

The integration of artificial intelligence in museums is a rapidly growing field, affecting both the visitor experience and the management of cultural resources. According to what has been studied, AI can be exploited on multiple levels, from visitors’ understanding of works to automated digitization and conservation of exhibits [32,41]. Based on this analysis, this study proposes the conceptual framework of AI-based tools that could be integrated into museums and cultural organizations:

3.5. Categorization of AI Uses in Museums

• Computer vision and image recognition
  • Artwork Analysis and Restoration [39].
  AI application for digital documentation and collection organization [34].
• Deep learning
  • Object classification and pattern recognition in artworks and historical artifacts, enhancing conservation, documentation, and provenance research [40].
• Machine learning and data processing
  • Analysis and categorization of collections using AI [26,33].
• Natural language processing and chatbot
  • Personalization of visitor experience [32,36].
• Semantic analysis and linked data
  • Structuring and interlinking museum data through semantic web technologies [33,34,35].
• Virtual and augmented reality
  • Creating immersive experiences through VR/AR [36].

4. Discussion

The application of artificial intelligence in the field of art and heritage opens up great possibilities, but it brings critical concerns as well. Among the problems, privacy, the accuracy of historical data, and the absence of bias in the algorithms are some of the important ones [42]. The use of AI in museums is more common in the UK and Spain. However, the analysis does not reveal a clear relationship between AI applications and visitation rates. As French and Villaespesa [43] point out, AI technologies can enhance the visitor experience, but they are not, by themselves, a reason to visit a museum. Both the Spearman correlation coefficient and the analysis of variance (ANOVA) indicate no statistically significant difference in visitation between museums that use AI and those that do not. These results suggest that AI is not a key factor driving museum visitation. The final part of the analysis reveals that while the UK and Spain have embraced AI pilot projects, France, Germany, Greece, Italy, and Russia have shown relatively little interest in integrating the technology. These findings align with the study by Kung and Lin [44], which stated that the use of AI in museums depends on a country’s digitization policies and funding. Thus, AI currently serves as a supporting tool for collection management and digital access, but it is not a key determinant in museum visitorship. Similar results were reported in the study by Rani et al. [45], which found that while digital technology enhances the overall visitor experience, it is not a major driver of actual museum visits.

5. Conclusions

Based on the findings of this study, it is clear that while artificial intelligence has the potential to enhance visitor experiences in museums, there is no direct evidence linking it to increased attendance or making it a primary motivator for visits. AI is, however, emerging in more museum applications, especially in places such as the UK or Spain, but this analysis shows that it is considered to be an enabling performer in relation to collection management and digital interaction and not the leading means for attracting visitors. These findings provide an opportunity for additional research that could expand knowledge of the broader impact of AI on both museums and the cultural sector. Future research could explore visitors’ perceptions of AI through surveys and interviews, employing qualitative methods to gain deeper insights into user experience and engagement. Additionally, expanding the sample to include a larger and more diverse range of museums—potentially from other continents—would enhance the findings. Future work could include a web-based survey to explore how visitors perceive and engage with AI in museums, including user-centered questions related to their expectations, satisfaction, and perceived value. This would provide valuable qualitative insights to complement the current quantitative findings and offer a more comprehensive understanding of the visitor experience. Furthermore, future studies should aim to collect longitudinal visitor data, ideally covering periods before and after the implementation of AI tools, in order to better assess the causal impact of AI integration on museum attendance. Finally, an important avenue for future exploration would be to investigate if AI may be used to attract otherwise unengaged visitor demographics, such as youth or disabled individuals, using inclusive and accessible technologies. Artificial intelligence functions as a useful tool that museums can implement to support their goals and address visitor needs as part of their development.