Artificial Intelligence in Heritage Tourism: Innovation, Accessibility, and Sustainability in the Digital Age

Abstract

Artificial intelligence (AI) is profoundly transforming heritage tourism through the incorporation of technological solutions that reconfigure the ways in which cultural heritage is conserved, interpreted, and experienced. This article presents a critical and systematic review of current AI applications in this field, with a special focus on their impact on destination management, the personalization of tourist experiences, universal accessibility, and the preservation of both tangible and intangible assets. Based on an analysis of the scientific literature and international use cases, key technologies such as machine learning, computer vision, generative models, and recommendation systems are identified. These tools enable everything from the virtual reconstruction of historical sites to the development of intelligent cultural assistants and adaptive tours, improving the visitor experience and promoting inclusion. This study also examines the main ethical, technical, and epistemological challenges associated with this transformation, including algorithmic surveillance, data protection, interoperability between platforms, the digital divide, and the reconfiguration of heritage knowledge production processes. In conclusion, this study argues that AI, when implemented in accordance with principles of responsibility, sustainability, and cultural sensitivity, can serve as a strategic instrument for ensuring the accessibility, representativeness, and social relevance of cultural heritage in the digital age. However, its effective integration necessitates the development of sector-specific ethical frameworks, inclusive governance models, and sustainable technological implementation strategies that promote equity, community participation, and long-term viability. Furthermore, this article highlights the need for empirical research to assess the actual impact of these technologies and for the creation of indicators to evaluate their effectiveness, fairness, and contribution to the Sustainable Development Goals.

1. Introduction

The incorporation of artificial intelligence (AI) in the field of heritage tourism is an emerging area of research, characterized by its cross-cutting nature and its transformative potential in the management, conservation, enhancement, and accessibility of cultural heritage. This technological convergence is based on the development of advanced tools such as machine learning, natural language processing, computer vision, generative models, and recommendation systems (Figure 1), which allow for innovative approaches to the challenges inherent in sustainability and the tourist experience in heritage contexts.

The recent literature highlights the strategic role of AI in optimizing the management of heritage destinations by facilitating evidence-based decision-making processes, improving resource allocation, and increasing operational efficiency [1,2]. In particular, the automated analysis of large volumes of data—from social media, sensors, digital platforms, and historical archives—allows for the identification of behavior patterns, visitor preferences, and dynamics of heritage space use, contributing to more adaptive and user-centered planning [3,4,5]. In this sense, the emergence of advanced digital technologies such as artificial intelligence (AI) [6], virtual reality (VR) [7], and the metaverse [8] is profoundly transforming cultural tourism and heritage preservation, opening up new possibilities for interaction, accessibility, and sustainability of cultural assets.

Traditionally, cultural tourism has focused on the in-person experience and direct appreciation of tangible and intangible heritage. However, the integration of AI and immersive digital platforms is revolutionizing the way visitors interact with cultural instruments and practices, enabling personalized, adaptive, and immersive experiences that were previously unattainable. Instead, artificial intelligence (AI) has now established itself as a driver of innovation in the field of cultural heritage, facilitating both its preservation and dissemination [9,10]. Among the most notable applications are opinion mining systems and semantic analysis of tourist comments, which, supported by large-scale language models, make it possible to detect attributes valued by visitors and adjust the heritage offering according to their expectations [11]. This capacity for continuous feedback promotes both the sustainability of sites and the improvement of the visitor experience [2].

In terms of heritage governance, AI facilitates the participation of multiple actors and the identification of diverse cultural values, promoting more inclusive and collaborative approaches. Automated analysis of unstructured texts and data—such as regulations, specialized literature, and social media content—allows for the identification of perceptions, tensions, and priorities of different stakeholders, which is essential for the design of sensitive, data-driven cultural policies [12,13,14].

Its application also enables the planning and conservation of heritage sites through advanced techniques such as image recognition, 3D reconstruction, and predictiveal analysis, contributing to the documentation and restoration of monuments and artifacts at risk of disappearing [15,16,17]. It also highlights the important role it plays in improving accessibility and inclusion for a large part of society, including people with disabilities, in enjoying experiences. AI, therefore, facilitates the creation of uses adapted to people with different abilities, being inclusive through the use of conversational interfaces, machine translation, and user-centered design, expanding access to heritage both online and on-site [18,19]. In addition, AI has proven to be effective in the digitization, restoration, and automated cataloging of cultural assets, as well as in monitoring their state of conservation. In this regard, it should be noted that tools based on computer vision and deep learning have been developed that allow heritage items to be identified, classified, and documented with high precision, facilitating their management and research [19,20]. However, technical challenges remain, such as improving the accuracy of automatic image classification and the availability of quality training data [21].

In the field of tourism, AI enables the creation of immersive virtual experiences through technologies such as virtual reality (VR), augmented reality (AR), and the metaverse, making it possible to explore historical reconstructions and cultural narratives without compromising the physical integrity of heritage sites [22,23]. This capability is complemented by real-time analysis of user preferences and behaviors, which allows for the generation of personalized content, the offering of recommendations, and the facilitation of multilingual accessibility, increasing visitor satisfaction and inclusion [18,22]. Therefore, the personalization of the tourist experience, supported by the processing of large volumes of data, promotes the inclusion and democratization of access to cultural heritage [16,22]. Likewise, predictive analysis algorithms and recommendation systems contribute to the sustainable management of heritage tourism by distributing visitor flows, preventing destination overload, and optimizing ecological carrying capacity [24,25,26]. AI also supports decision-making in tourism planning and the protection of cultural resources through spatial analysis and the correlation of geographic and demographic data [16,25].

In this context, the literature proposes the development of specific ethical frameworks to guide the responsible use of these technologies [27]. Similarly, it enables the creation of personalized digital narratives that promote intercultural empathy and the revaluation of local identities. Along with this, the use of generative models and blockchain technologies has opened up new avenues for the global dissemination of heritage, while ensuring traceability, authenticity, and the protection of cultural rights [28,29].

On the other hand, robotics and automation are also emerging as complementary tools in cultural tourism management, especially in visitor assistance, facility maintenance, and tourist flow control tasks [30]. These technologies, integrated with AI systems, improve operational efficiency and user experience in heritage environments.

It should also be noted that the transformative potential of AI in the economic, social, and cultural spheres has not yet been fully realized. Therefore, it is necessary to create a research and development agenda that addresses both technological advances and ethical and social challenges, including data protection, equity, and respect for cultural diversity [17,19,27,31]. In this regard, the literature warns of risks associated with the use of AI, such as the commodification of culture and the possible loss of authenticity in digital representations. Therefore, the need for a balanced approach is emphasized, where technology complements and enriches the experience without replacing the intrinsic value of heritage [19,22].

The integration of AI with immersive technologies—such as augmented reality (AR) and virtual reality (VR)—has redefined the way visitors interact with heritage, creating educational, inclusive, and sustainable experiences that align with the Sustainable Development Goals (SDGs). These solutions not only enrich cultural interpretation, but also allow for visitor flow management, mitigate environmental impact, and reduce pressure on vulnerable heritage sites [28,32].

Both AR and VR have evolved from simple visualization tools to platforms that enable the digital reconstruction of historical sites, the simulation of past environments, and participation in cultural narratives. These immersive technologies facilitate the creation of virtual environments where users can explore, interact, and learn about heritage without the physical, economic, or logistical limitations that traditionally restrict access [33,34]. Both expand the scope of cultural tourism and also contribute to its conservation by reducing the impact of in-person tourism on original assets, mitigating physical wear and tear, and promoting sustainable practices [35,36]. At the same time, the digitization of heritage experiences allows for the preservation of intangible elements, such as oral traditions, rituals, and artistic expressions, which can be recreated and transmitted through interactive virtual environments [37].

The metaverse, understood as a persistent and collaborative virtual environment, expands these possibilities by offering spaces where users can interact in real time, participate in educational and social activities, and experience heritage in a multisensory and participatory way. The creation of digital twins of historic cities and heritage sites in the metaverse allows for the recreation of historical contexts, the simulation of events, and the co-creation of cultural experiences, facilitating interpretation and meaningful learning [38,39].

Complementarily, gamification and the design of interactive experiences in these digital environments have proven to be particularly effective in capturing the interest of new generations, facilitating the transmission of cultural values and knowledge through participation and playful learning [34,38,40]. However, this digitization process poses significant challenges related to authenticity, cultural representation, the digital divide, and the protection of digital rights and privacy [41,42]. In the face of these challenges, the literature emphasizes the importance of establishing ethical and collaborative frameworks that involve local communities, cultural institutions, and technology developers to ensure responsible and representative preservation.

In line with this, the participation of communities in the digitization and representation of their heritage is essential to ensure the authenticity and cultural appropriation of technological processes [34,35,41]. Likewise, interdisciplinary collaboration between experts in technology, heritage, education, and public policy is fundamental to designing sustainable strategies that balance innovation with the protection of cultural values [16,22,35].

It follows from all of the above that the integration of AI, VR, and the metaverse into heritage management must focus on sustainability, accessibility, and community participation. On the one hand, sustainability implies not only reducing the environmental impact of tourism, but also the long-term viability of the technological solutions implemented, considering their maintenance, updating, and adaptability to changing contexts [35,36]. On the other hand, accessibility must be approached from an inclusive perspective, ensuring that digital experiences are understandable, usable, and culturally relevant to diverse audiences, including people with disabilities or language barriers [22,32,33]. Finally, community participation is key to legitimizing digitization processes and ensuring that technologies respond to the needs and aspirations of the communities where the heritage is located [34,41].

The convergence of AI, VR, and the metaverse offers unprecedented opportunities to enrich cultural tourism and preserve heritage, if they are approached in an ethical, sustainable, and people-centered manner. Despite this, the literature agrees that AI is a powerful tool for innovation in heritage tourism, but its implementation must be responsible, ethical, and oriented toward sustainability and cultural diversity. Interdisciplinary collaboration and strategic vision are required to maximize the benefits and minimize the risks associated with this transformation [16,18,19,22,31].

This analysis, in the authors’ opinion, argues that heritage tourism is facing a profound structural transformation driven by artificial intelligence (AI), conceived not only as a technological tool, but also as a strategic vector for sustainability, accessibility, and cultural innovation. AI enables the optimization of destination management through predictive analytics and recommendation systems, promoting the balanced distribution of tourist flows and the preservation of carrying capacity. It also enhances preventive conservation through computer vision, IoT sensors, and generative models, which facilitate the continuous monitoring of the condition of cultural assets and the digital restoration of damaged elements. On an experiential level, AI enables immersive and adaptive narratives through augmented reality, virtual reality, and the metaverse, democratizing access to heritage and promoting the inclusion of diverse audiences. This technological convergence, if implemented under ethical and participatory frameworks, redefines the relationship between visitors, communities, and heritage, consolidating a more resilient, equitable, and Sustainable Development Goals-oriented model of cultural tourism.

Considering these aspects, this article aims to critically and systematically analyze the current applications of AI in heritage tourism, identifying its benefits, limitations, and emerging challenges. To this end, it is structured around a review of the scientific literature, the analysis of use cases, and the proposal of a conceptual framework that articulates the main axes of technological transformation in this field. This is followed by a workflow focused on analyzing the main technologies applied, showing the main benefits and identifying the challenges detected, to make proposals and draw strategic conclusions (Figure 2).

2. Methodology

This study adopts a qualitative, exploratory, and analytical approach aimed at systematically examining the applications, benefits, limitations, and challenges of artificial intelligence (AI) in heritage tourism. This research is structured in four complementary methodological phases: (1) systematic literature review, (2) thematic analysis, (3) case studies, and (4) theoretical validation through conceptual triangulation.

The first phase consisted of a systematic review of the scientific literature, following the principles of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol. The objective was to identify relevant studies that addressed the application of AI in heritage and tourism contexts from a multidisciplinary perspective. Academic databases indexed in JCR and Scopus were consulted, including Web of Science, Scopus, SpringerLink, ScienceDirect, and IEEE Xplore. The search was conducted between January 2019 and June 2025, using Boolean operators and combinations of keywords such as: “artificial intelligence” AND “cultural heritage,” “smart tourism” AND “machine learning,” “digital heritage” AND “virtual reality,” “AI” AND “accessibility” AND “ethics,” among others.

The inclusion criteria were:

-

Peer-reviewed articles.

-

Contributions to relevant scientific conferences.

-

Empirical, theoretical, or review studies.

-

Publications mainly in English.

-

Direct relevance to the subject of study.

Duplicate documents, non-academic publications, and studies focused exclusively on technologies unrelated to AI were excluded. After initial screening and quality assessment, 78 applied case documents were selected for in-depth analysis (Figure 3) (Appendix A).

The second phase consisted of an inductive thematic analysis aimed at identifying patterns, emerging categories, and lines of discussion in the selected literature. Open and axial coding techniques were applied, following the principles of grounded theory, with the support of qualitative analysis matrices. In open coding, the texts were segmented into meaningful units and initial codes were assigned that reflected emerging concepts, using analysis matrices and manual recording in spreadsheets. These concepts included all applications that had been made with any form of artificial intelligence, the existence of experiences with these technologies, and the benefits and problems that could arise from using various forms of artificial intelligence. To this end, a detailed analysis of the abstracts provided and the keywords assigned by the authors themselves was carried out. The results and discussion were analyzed in a complementary manner. Subsequently, in axial coding, the codes were grouped into categories by identifying conceptual relationships (conditions, actions, consequences), defining properties and dimensions for each category. This process was carried out iteratively, with successive revisions to eliminate redundancies and ensure internal consistency, for which the analysis of the artificial intelligence methodologies used was employed. In this sense, the texts were analyzed iteratively and coded manually, grouping the findings into three major thematic blocks:

-

Applications of AI in heritage tourism

-

Immersive, adaptive, and multisensory heritage experiences

-

Challenges in the application of artificial intelligence to heritage tourism

This categorization allowed us to construct a conceptual framework that articulates the main lines of technological transformation in heritage tourism, as well as the challenges associated with its implementation.

The third phase incorporated the analysis of application cases documented in the scientific and technical literature. Representative examples were selected for their innovation, geographical diversity, scalability, and thematic relevance. Selection criteria included:

-

Explicit use of AI technologies (computer vision, PLN, generative models, etc.).

-

Application in tangible or intangible heritage contexts.

-

Sufficient documentation on results, impacts, and limitations.

The cases were analyzed using a comparative matrix that considered variables such as: type of technology, purpose (interpretation, conservation, accessibility), level of community participation, sustainability, and replicability. This analysis illustrated the potential and limitations of AI in real contexts, complementing the theoretical review with applied evidence.

The fourth phase consisted of a conceptual triangulation between the findings of the literature review, the results of the thematic analysis, and the case studies. This methodological strategy made it possible to contrast perspectives, identify convergences and divergences, and strengthen the internal validity of this study. Theoretical frameworks from studies on digital heritage, technology ethics, universal accessibility, and cultural sustainability were also incorporated in order to contextualize the findings and propose well-founded lines of action. All of this facilitated the discussion.

3. Results

3.1. Applications of AI in Heritage Tourism

3.1.1. Interpretation and Enrichment of the Tourist Experience

AI-based virtual assistants represent one of the most significant innovations in heritage mediation and tourism [43,44]. These systems, integrated into mobile applications, augmented reality (AR) devices, or interactive kiosks, allow visitors to interact with heritage in real time, answering questions, suggesting itineraries, and tailoring information to individual needs. They use natural language processing (NLP) and generative models to offer historical, artistic, and symbolic explanations tailored to the user’s profile, language, and preferences, increasing the accessibility and personalization of the visit. For example, in Tamil Nadu (India), an intelligent platform integrates multilingual voice assistance, GPS navigation, and personalized historical content generation, facilitating interactive tours and contextual recommendations for all types of audiences, including people with disabilities [45]. In addition, there are virtual assistants developed specifically to improve accessibility to information within a museum, emulating a conversational agent [46]. Likewise, there are numerous assistants capable of answering complex questions about works of art or historical events, improving visitor autonomy and learning [47].

Generative AI, in combination with AR and virtual reality (VR), allows for the creation of personalized historical narratives and immersive experiences that transcend the mere transmission of data. These technologies enable visitors to explore virtual reconstructions of spaces, participate in educational games, or receive narratives tailored to their age, interests, or educational level. In metaverse-type virtual environments, AI adapts the narrative, and interaction based on user behavior and preferences, creating a dynamic and participatory experience [22]. For example, at Notre-Dame Cathedral in Paris, following the fire, immersive virtual tours have been developed that allow users to experience the history and architecture of the monument, even when physical access was restricted [48,49,50]. These experiences not only enrich the visit, but also contribute to conservation by reducing pressure on physical spaces.

Computer vision and deep learning have revolutionized the way heritage is documented and presented. These techniques make it possible to create accurate three-dimensional models of monuments and archaeological sites, facilitating the exploration of how they looked at different times and the visualization of missing or damaged elements. In Montenegro, for example, a web application allows users to photograph monuments and receive AI-generated historical information, revitalizing interest in little-known sites and facilitating the documentation and management of heritage [15]. In addition, systems based on convolutional neural networks (CNN) are able to classify and recognize architectural elements with high precision, as demonstrated by the model developed by Folino et al., which achieves 90% accuracy in the classification of heritage elements [20]. These tools are essential for both interpretation and preventive conservation.

Accessibility is a fundamental pillar for the democratization of heritage. Neural translation and speech synthesis systems make it possible to offer heritage content in multiple languages and accessible formats, including sign language and audio for people with visual impairments. This eliminates language and sensory barriers, broadening the scope of the heritage experience. A notable example can be found at the World Heritage Sites of Ayutthaya and Suphan Buri (Thailand), where an intelligent platform integrates generative AI, PLN, and IoT to provide multilingual and accessible information in real time, enhancing the experience of international visitors and people with special needs [51]. These solutions contribute to inclusion and equity in access to knowledge and culture.

3.1.2. Personalization of Offerings and Recommendation Systems

Cultural recommendation systems apply deep learning algorithms, collaborative filtering, semantic analysis, and fuzzy logic to suggest routes, activities, and heritage content tailored to each visitor. These recommendations are based on multiple variables: browsing history, geographic location, stated interests, real-time behavior, age, language, and environmental context (such as weather or crowd size).

For example, a tourist interested in contemporary art may receive suggestions for temporary exhibitions, guided tours, or related immersive experiences, while another with an interest in medieval history may be directed to castles, themed museums, or virtual reconstructions. These recommendations can be integrated into mobile applications or web platforms that dynamically adjust suggestions based on ticket availability, opening hours, or special events.

Recent studies highlight the use of hybrid systems such as TourOptiGuide, which combines fuzzy logic, deep learning, and trajectory data to offer personalized recommendations based on the visitor’s location, estimated age, and preferences [52]. Likewise, the use of probabilistic models and GERT (Graphical Evaluation and Review Technique) networks has been proposed to optimize the temporal performance of recommendation systems in digital heritage collections, allowing for accurate evaluation of response times and service quality [53].

AI enables advanced analysis of tourist profiles using clustering, intelligent segmentation, and data mining techniques. These techniques identify the behavior patterns and preferences in large data sets, facilitating the creation of cultural products tailored to different segments: families with children, school groups, heritage experts, international tourists, seniors, among others. As a result, a museum can offer interactive tours with gamified content for schoolchildren, while for experts it can enable access to heritage databases, detailed 3D reconstructions, or technical visits. This segmentation allows for a more relevant and attractive offering, aligned with the real expectations and needs of each group, and contributes to more efficient management of cultural resources [54].

Adaptive interfaces represent a significant advance in terms of accessibility and inclusion. These interfaces modify their content, structure, and level of complexity based on the user’s cognitive profile, age, language, or abilities. For example, they can simplify language for people with cognitive disabilities, offer audio descriptions for people with visual impairments, or include subtitles and sign language for people with hearing impairments. A notable initiative in this area is AI4Culture, a European project that applies AI to adapt the presentation of heritage content in museums and cultural sites. This tool allows for the creation of personalized tours for audiences with specific needs, such as older adults, children, or visitors with autism spectrum disorders, thus promoting equity in access to heritage [13]. Similarly, the ReInHerit Toolkit project, developed under the H2020 program, has created interactive applications based on computer vision and AI to enrich the museum experience, adapting content to different user profiles and encouraging visitor participation [55].

In summary, it can be said that in the field of heritage tourism, technologies based on artificial intelligence are playing an increasingly important role in optimizing the management, conservation, and interpretation of cultural heritage.

Machine learning is a key tool for developing predictive models and analyzing visitor behavior patterns, which allows strategic decisions to be based on empirical evidence. At the same time, computer vision is applied to automated image recognition and three-dimensional reconstruction of heritage elements, contributing significantly to their documentation, conservation, and dissemination. Likewise, natural language processing provides multilingual support and enables sentiment analysis, facilitating a deeper understanding of the perceptions, assessments, and expectations of the public. Added to this is the potential of generative artificial intelligence, which enables both the virtual restoration of cultural assets and the automated production of interpretive narratives adapted to different user profiles. Finally, recommendation systems allow the tourist experience to be personalized according to the interests, needs, and individual characteristics of visitors. Together, these technologies form an intelligent ecosystem aimed at strengthening the sustainability, accessibility, and appreciation of cultural heritage in contemporary tourism contexts.

3.1.3. Conservation, Documentation, and Digitization of Heritage

AI, combined with computer vision and IoT sensors, enables continuous and accurate monitoring of the state of conservation of monuments and historic sites. Deep learning algorithms analyze images and sensory data to identify cracks, erosion, moisture, vandalism, and other damage, even in the early stages. For example, at Cologne Cathedral and other European sites, AI-assisted visual inspection systems have been implemented to detect structural defects and their evolution over time, improving the efficiency and objectivity of traditional assessments [56]. In India, AI-based predictive models analyze the impact of extreme weather conditions on artifacts and monuments, allowing damage to be anticipated and conservation resources to be optimized [57]. In addition, the integration of digital twins and H-BIM facilitates the simulation of deterioration scenarios and the planning of preventive interventions [13,58].

Generative adversarial networks (GANs), deep learning, and AI image generation have transformed the digital restoration of damaged or incomplete works. These techniques make it possible to reconstruct paintings, sculptures, and manuscripts while respecting the original style and context. A landmark case is the virtual reconstruction of Palmyra (Syria), where AI generated accurate visualizations from textual descriptions and historical images, facilitating the documentation and digital revitalization of sites destroyed by conflict [59]. Projects such as DeepScribe have restored medieval manuscripts, while the 3D reconstruction of archaeological artifacts using AI enables immersive digital exhibitions and the conservation of physically irrecoverable pieces [60,61].

The cataloging and enrichment of heritage archives has been optimized through AI, especially with natural language processing (NLP) and computer vision. These tools automatically tag, describe, and relate objects and documents in digital databases, improving accessibility and knowledge management. For example, the model by Folino et al. [20] achieves 90% accuracy in the classification of architectural elements, facilitating documentation and interactive experience [53,62]. Platforms such as Aïoli use deep learning for the semi-automatic annotation of 2D/3D models, identifying architectural components and degradation patterns [16]. In addition, AI is used to analyze large volumes of chemical and material data, optimizing the conservation of pigments, metals, and ceramics [61].

AI is also essential in the preservation of intangible heritage, such as traditional music, endangered languages, and dances. Through voice recognition, semantic analysis, and generative modeling, cultural expressions that would otherwise be at risk of disappearing are digitized and disseminated. For example, the digitization of Shanxi opera in China has used AI to analyze, catalog, and revitalize musical repertoires, facilitating their transmission to new generations [63,64]. In this regard, various interactive experiences based on AI and virtual reality have enabled the preservation and dissemination of traditional dances and rituals, involving communities in the creation and validation of content [63]. In addition, AI is used to create illustrations and visualizations of intangible heritage, facilitating its access and reinterpretation in contemporary contexts [65].

AI allows heritage to be documented and digitized more efficiently and accurately. For example, the use of computer vision and deep learning facilitates the automated visual inspection of monuments, detecting cracks, erosion, moisture, or vandalism in real time. At Cologne Cathedral (Germany), AI-assisted visual inspection systems have been implemented to identify structural defects and their evolution, improving the speed and objectivity of traditional assessments [56]. In addition, the generation of 3D models from point clouds and their integration into HBIM (Heritage Building Information Modeling) systems allows for the creation of detailed digital replicas of historic buildings, as has been carried out in the ancient city of Side (Turkey), facilitating conservation and remote study [66,67].

It is also used to restore ancient paintings and manuscripts, respecting the original style and facilitating access for researchers and the general public [68].

Through natural language processing and computer vision, it automates the classification and cataloging of large volumes of heritage objects and documents. For example, deep learning algorithms can automatically identify and tag architectural elements in digital databases, improving accessibility and knowledge management [53,69]. In addition, AI systems have been applied to analyze historical photographs, optimizing the production of 3D models and the detection of damage to heritage [21]. It also enhances the creation of immersive and personalized experiences for visitors to museums and heritage sites. Virtual assistants, interactive tours, and machine translation facilitate access to information in multiple languages and accessible formats. For example, augmented reality and virtual reality applications have been developed that allow users to explore historical reconstructions and participate in educational activities tailored to different audiences [37]. These technologies also improve accessibility for people with disabilities, enabling an inclusive experience [18].

Despite the undoubted advantages of using AI in cultural heritage, it poses ethical challenges, such as managing bias, protecting privacy, and the need for community participation. It is therefore essential to establish ethical frameworks that guarantee transparency, inclusion, and sustainability in the application of these technologies [14,70].

3.2. Immersive, Adaptive, and Multisensory Heritage Experiences

AI, combined with augmented reality (AR), virtual reality (VR), mixed reality (MR), and digital twins, allows historical environments to be reconstructed with unprecedented realism and interactivity. Projects such as “Rome Reborn” [71,72] and “Assassin’s Creed: Origins” [73] allow visitors to explore ancient Rome or Pharaonic Egypt, interacting with AI-generated historical characters who answer questions and adapt their stories according to the user’s profile [74]. In the village of Dongguan Nanshe, China, AI and VR were used to create personalized virtual tours, where visitors can choose themed routes (architecture, daily life, festivities) and receive explanations tailored to their age, language, or interests [75]. In addition, the integration of haptic sensors and 3D sound enables multisensory experiences: feeling the texture of a virtual sculpture, hearing the bustle of a medieval market, or perceiving historical aromas, which increases immersion and learning [38]. These experiences extend beyond the museum, allowing for pre-visit preparation and narrative continuity on digital platforms, such as the Metaverse, where users can relive and share their experience [22].

AI is reducing barriers for traditionally excluded audiences. At the Ayutthaya and Suphan Buri sites (Thailand), intelligent systems have been implemented that offer information in audio, text, sign language, and voice navigation, with real-time automatic translation, facilitating universal accessibility [52]. Collaborative platforms allow any user to contribute stories, images, or reinterpretations, enriching heritage knowledge from multiple perspectives and promoting cultural diversity [13]. For example, in European museums, digitization and AI have enabled the creation of tours adapted for people with visual impairments, using automatically generated audio descriptions and interactive tactile maps [38]. However, challenges remain, such as the digital divide and the need for inclusive policies and digital training to ensure that these advances reach all audiences [76].

This type of tool is becoming established as a strategic tool for the smart management of heritage destinations. Predictive models anticipate tourist flows and manage carrying capacity, as is the case in sites of high heritage value where saturation and physical deterioration are prevented through real-time data analysis [77]. In Montenegro, an AI-based application allows users to identify monuments through images and receive personalized historical information, revitalizing tourist interest and the conservation of little-known monuments [15]. In addition, sentiment analysis on social media helps to assess the perception and reputation of destinations, adjusting communication and conservation strategies [13,76]. Decision support systems integrate cultural, economic, and environmental data to guide public policy and improve sustainability. Alongside this, it facilitates heritage co-creation, allowing local communities to actively participate in content generation, the interpretation of their history, and the management of their resources. In the virtual reconstruction of Dongguan Nanshe, AI allowed residents to contribute stories and memories, integrating them into personalized virtual tours [75]. Collaborative platforms with AI assist in oral documentation, the translation of traditional knowledge, and the creation of participatory digital exhibitions, strengthening the social appropriation of heritage and promoting narrative diversity [38].

Consequently, the advancement of AI in heritage contexts requires regulating the authorship and intellectual property of AI-generated content, protecting sensitive cultural data, and ensuring the transparency and explainability of algorithms. For example, the use of NFTs (non-fungible tokens) is being explored to certify the authenticity and ownership of digital heritage works [78,79]. In addition, it is necessary to rethink epistemological frameworks from critical and participatory perspectives, ensuring that the production of heritage knowledge is inclusive and pluralistic [13].

Technological interoperability enables collaboration between museums, universities, and technology companies, as is the case in European heritage digitization projects [38]. Open data and content foster innovation and transparency, while technological and cultural sustainability ensure long-term preservation and respect for diversity. Examples such as the digitization of museums and the creation of digital twins of historical sites demonstrate the potential of these ecosystems to transform the management, access, and conservation of cultural heritage [22,77].

3.3. Challenges in Applying Artificial Intelligence to Heritage Tourism

It is clear that artificial intelligence (AI) is profoundly transforming heritage tourism, redefining the way heritage is preserved, interpreted, and experienced in all its forms. However, its implementation poses a number of ethical and technological challenges that must be addressed to ensure responsible, inclusive, and sustainable integration (Figure 4).

3.3.1. Ethical Challenges

One of the most significant challenges is the potential distortion of cultural narratives when AI mediates their interpretation and dissemination. Algorithms, if trained with unrepresentative data or without the participation of local communities, can reproduce historical biases, render subaltern voices invisible, and reinforce stereotypes [14,22,70]. This compromises the authenticity and plurality of heritage narratives, especially when dominant interpretations are prioritized or cultural elements are trivialized through their algorithmic commodification [80]. The literature recommends the active inclusion of communities and experts in the supervision of AI-generated content, as well as the development of specific ethical frameworks.

On the other hand, the personalization of tourist experiences through AI involves the massive collection of personal data, which poses risks in terms of privacy, informed consent, and secondary use of information [81,82,83]. Cases such as platforms that track visitor behavior without transparency or opt-out options illustrate these concerns. The need to apply “privacy by design” principles, with mechanisms for anonymization, access control, and transparency, is emphasized [83].

Furthermore, the implementation of AI-based solutions can accentuate the digital divide if they are not designed with universal accessibility criteria in mind. Technological, linguistic, or economic barriers can exclude older adults, rural communities, or people with disabilities [76,84]. Interfaces that are not adapted or that require advanced devices limit equitable access to digital heritage. The literature advocates for policies of digital inclusion, universal design, and multilingual adaptation [14,70]. The opacity of many AI systems makes it difficult to understand the criteria that guide their decisions, affecting accountability and public trust [70,81]. For example, systems that favor popular tourist destinations without explicit justification can undermine the diversity and sustainability of the sector. The development of explainable algorithms (XAI), audits, and human oversight is proposed [85].

The use of AI platforms developed by large corporations can create structural dependency, compromising the cultural and technological sovereignty of heritage institutions [14,22,80]. Technological obsolescence and lack of interoperability also threaten the long-term preservation of digital content. The use of open standards, interoperable repositories, and the strengthening of local capacities are recommended [86].

In addition, AI can amplify biases present in data, affecting equity in representation and access to heritage [87]. Algorithms trained with historical data can reproduce gender, race, or class biases. The literature emphasizes the need to audit and correct these biases, as well as to promote diversity in development teams [70,81]. Therefore, the complexity of these challenges requires specific governance frameworks that integrate experts in ethics, technology, heritage, local communities, and policy makers [14,22,70]. In this vein, the development of codes of conduct, training in digital ethics, and continuous evaluation of sociocultural impacts are recommended. Thus, international collaboration and the harmonization of standards are key to responsible management [86].

In summary, the integration of AI into heritage tourism offers significant opportunities, but also poses complex ethical challenges. Addressing these challenges requires robust regulatory frameworks, community participation, bias auditing, and inclusive governance that ensures technology enhances cultural value without compromising its integrity or equity.

3.3.2. Technological Challenges

The development of AI models requires large volumes of high-quality data. In the heritage field, this data is often fragmented, outdated, or protected by copyright, which limits its reuse [2,13,20,76]. The lack of standardization in cultural metadata and the scarcity of high-resolution digitization hinder interoperability between institutions. In addition, data governance is a critical issue: the absence of open data policies and common standards restricts collaboration and the development of scalable solutions [88]. In this regard, the literature recommends collaborative strategies and the adoption of international standards to improve the quality and accessibility of heritage data.

Interoperability is also essential for AI solutions to integrate information from multiple sources. However, the heterogeneity of the systems used by museums, archives, and heritage sites hinders data integration [2,22]. The incompatibility of formats and protocols limits the creation of personalized tourist experiences. Projects such as AI4Culture highlight the need to adopt open standards, APIs, and shared ontologies to ensure technological sustainability [89].

Many AI applications in heritage tourism are developed as pilot projects, but face scalability challenges due to limitations in infrastructure, funding, and technical training [22,76,90]. Similarly, the lack of specialized personnel in small or rural institutions, coupled with high maintenance costs, compromises their sustainability. In addition, rapid technological evolution can render implemented solutions obsolete. In this regard, the literature proposes the development of modular and flexible platforms, as well as collaborative ecosystems between the public, private, and academic sectors [88,91].

AI models, especially those based on deep learning, present problems of explainability and robustness when faced with new or changing data [13,24,76]. For example, classification systems may fail when faced with heritage objects from non-Western cultures or variable lighting conditions [20]. For this reason, the use of explainable algorithms (XAI), periodic audits, and cross-validation techniques is recommended to ensure transparency and reliability [92,93].

All of this means that the use of proprietary platforms developed by large corporations poses risks of technological dependence, rapid obsolescence, and lack of control over models and data [2,20,22]. Problems such as restrictive licenses, high costs, and service discontinuity can compromise the digital sovereignty of cultural institutions. In this regard, the literature recommends the use of open technologies, interoperable repositories, and the strengthening of local capacities [77,88]. In addition, the need to consider the environmental impact of digital infrastructures is highlighted [32].

Despite everything, the digital divide limits the adoption of AI in regions with poor infrastructure, affecting equity in the preservation and dissemination of heritage [76,90]. Differences in connectivity, access to devices, and technical training exclude rural communities and small actors in the sector. It is proposed to implement policies of digital inclusion, continuous training, and support for local innovation to democratize access to advanced technologies [52].

On the other hand, the combination of AI with technologies such as IoT, blockchain, augmented reality, and virtual reality poses new challenges in terms of integration, security, and data management [77,90]. The continuous updating of models and data lifecycle management require planning, resources, and inter-institutional collaboration [2,13].

Together, these technological challenges must be addressed through ethical governance, specialized training, the adoption of open standards, and cross-sector collaboration. Only then will it be possible to harness the transformative potential of AI in heritage tourism in a sustainable, inclusive, and efficient manner. Therefore, the implementation of artificial intelligence technologies in heritage tourism poses a series of critical challenges that must be addressed from an ethical, technical, and sociocultural perspective. First, the issue of cultural authenticity involves the risk of inaccurate or simplified representations of heritage, which can compromise its symbolic integrity. To mitigate this problem, the active involvement of local communities in digitization and cultural mediation processes is proposed. Second, data privacy management is strained by the collection and processing of sensitive visitor information, which requires the implementation of robust anonymization mechanisms and the obtaining of informed consent in accordance with current regulatory frameworks. Likewise, the persistent digital divide can lead to the exclusion of underrepresented groups, requiring inclusive design that guarantees technological and cognitive accessibility. Another relevant challenge is algorithmic bias, which can reproduce or amplify cultural, gender, or ethnic stereotypes; in this regard, systematic auditing of algorithmic models is recommended to identify and correct structural biases. Finally, technological dependence on specific suppliers can lead to vendor lock-in situations, limiting the technological sovereignty of heritage institutions. Therefore, the use of open standards and the strengthening of local capacities are advocated as strategies for technological sustainability. These challenges require an interdisciplinary and critical approach that ensures the equitable, ethical, and contextualized application of artificial intelligence in the field of cultural heritage.

4. Discussion

Artificial intelligence (AI) has established itself as a transformative force that not only optimizes destination management and cultural heritage conservation, but also profoundly redefines the ways in which cultural heritage is understood, interpreted, and legitimized. This transformation, driven by the rapid evolution of digital culture [94], requires the design of a technological-heritage governance framework that coherently articulates ethical principles, technological sustainability, community participation, and digital sovereignty. The implementation of such a framework would not only significantly enrich the tourist experience [95], but also ensure its inclusive and equitable nature. This governance model must be based on contextualized algorithmic ethics, which ensures the explainability of AI systems, the protection of sensitive cultural data, and transparency in the processes of classification, recommendation, and content generation. Likewise, it is crucial to promote the adoption of open standards and interoperable technologies that prevent dependence on proprietary platforms and ensure the long-term preservation of digital assets. The participation of local communities in the processes of digitization, interpretation, and validation of heritage content is essential, as it culturally legitimizes AI-generated representations and prevents the reproduction of colonial or hegemonic biases. This situated heritage intelligence must be complemented by technological and cultural sustainability strategies that take into account the life cycle of data, the environmental impact of digital infrastructures, and the technical training of the actors involved. In this context, the creation of continuous evaluation and auditing mechanisms, such as observatories of digital heritage ethics, is proposed to monitor the sociocultural impact of the technologies implemented, identify algorithmic biases, and promote good practices in the sector.

The incorporation of artificial intelligence in the field of cultural heritage not only introduces new tools for management, conservation, and dissemination, but also profoundly transforms the ways in which heritage is understood, interpreted, and legitimized [2,13]. It significantly enriches the user experience [95]. This transformation implies an epistemological reconfiguration that deserves to be addressed critically. Traditionally, heritage knowledge has been mediated by experts, institutions, and regulatory frameworks that define what is considered heritage, how it should be preserved, and who has the authority to interpret it. By automating processes of classification, analysis, and content generation, AI partially displaces these centers of authority toward algorithmic systems trained with historical, social, and cultural data that are not always neutral or representative [65]. This algorithmic mediation introduces new logics for validating heritage knowledge, based on statistical correlations, behavioral patterns, and predictive models. As a result, there is a risk that heritage narratives will be constructed from biased data, rendering subaltern memories invisible or reinforcing hegemonic views. Furthermore, the automatic generation of cultural interpretations raises questions about the authenticity, authorship, and epistemological legitimacy of heritage discourses produced by AI. On the other hand, AI also opens up possibilities for democratizing the production of heritage knowledge by allowing local communities to participate in data generation, content co-creation, and cultural validation of representations. This epistemic openness, however, requires ethical and methodological frameworks that guarantee inclusion, transparency, and respect for the diversity of knowledge. In this sense, AI should not only be understood as a technical tool, but as an epistemic agent that redefines who produces knowledge about heritage, with what criteria, and for what purposes. This critical dimension is fundamental to avoiding a technocratization of heritage and promoting a truly inclusive, plural, and situated heritage intelligence.

Despite the notable growth in the literature on AI and heritage, significant gaps remain that hinder a comprehensive understanding of the phenomenon. First, there is a shortage of longitudinal studies analyzing the real impact of these technologies on the conservation, accessibility, and sustainability of heritage in the medium and long term [96]. Most research focuses on case studies or theoretical reviews, which limits the possibility of generalizing results and formulating public policies based on robust empirical evidence. Second, scientific output has a considerable geographical bias, with a predominance of experiences in Europe, North America, and East Asia, while contexts in the Global South remain underrepresented. This epistemic asymmetry prevents visibility of the socio-technical, cultural, and economic conditions that shape the adoption of AI in regions such as Latin America, Africa, and South Asia. Third, there is weak articulation between AI applications in heritage and the Sustainable Development Goals (SDGs), especially with regard to inclusive education, reducing inequalities, and urban sustainability. Few studies explicitly link these technologies to social, environmental, or cultural impact indicators. There is also a lack of standardized metrics to evaluate the effectiveness, equity, and sustainability of the solutions implemented. Finally, a more in-depth epistemological reflection is needed on how AI transforms the processes of heritage knowledge production, who defines what heritage is, and what criteria are used to represent it. This critical dimension is essential to avoid the technocratization of heritage and to promote a truly inclusive, ethical, and sustainable heritage intelligence.

Given the rapid advancement of Artificial Intelligence, it is worth asking what role Natural Intelligence should play in guiding the evolution of the former in a field as culturally situated as heritage. In this case, it is clear that natural intelligence—understood as expert judgment, ethical sensitivity, and community participation—must be an irreplaceable element in guiding the application of artificial intelligence. While AI provides unprecedented analytical and operational capabilities, its effectiveness depends on human-defined criteria of authenticity, diversity, and sustainability. Furthermore, it should be noted that algorithms, by themselves, lack contextual understanding and can reproduce historical biases or render subaltern memories invisible. Therefore, human supervision is essential to audit models, validate generated narratives, and ensure that digital representations respect cultural plurality. Likewise, natural intelligence must guide the inclusive design of interfaces and experiences, preventing the digital divide from excluding vulnerable groups. This interaction not only prevents epistemological and ethical risks, but also enhances heritage co-creation, socially legitimizing digitization processes and ensuring that technology complements—rather than replaces—the symbolic value of heritage. In short, the convergence between AI and human judgment allows us to move towards a more resilient and equitable model of heritage tourism that is aligned with the Sustainable Development Goals, where technological innovation is articulated with principles of participatory governance and algorithmic ethics.

Although this study offers a broad and systematic overview, it has limitations that must be considered. First, this review has focused on the academic literature published in English and indexed databases, which may have excluded relevant contributions in other languages or other sources. Second, although multiple applications of AI in heritage tourism have been identified, no direct empirical evaluation of its impact in terms of sustainability, accessibility, or cultural appropriation has been conducted. This lack of empirical validation prevents the establishment of robust causal relationships between technological implementation and observed benefits. Third, this study does not address in depth the economic implications of AI adoption in heritage institutions, nor the emerging business models associated with these technologies. Consequently, the following are proposed as future lines of research: (1) the development of longitudinal studies that analyze the real impact of AI on heritage conservation and enhancement; (2) the development of multidimensional indicators to assess its effectiveness, equity, and sustainability; (3) the exploration of synergies between AI and other emerging technologies such as blockchain, the Internet of Things (IoT), or affective computing; and (4) the formulation of ethical and epistemological frameworks to guide its responsible implementation in diverse heritage contexts.

5. Conclusions

The results of this study confirm that artificial intelligence (AI) is playing a strategic role in the transformation of heritage tourism by introducing new forms of conservation, interpretation, and access to cultural heritage. The application of technologies such as machine learning, computer vision, generative models, and recommendation systems has made it possible to optimize destination management, automate key processes, and generate more personalized, immersive tourist experiences tailored to the needs of diverse audiences.

Furthermore, AI contributes significantly to democratizing access to heritage by eliminating language, sensory, and cognitive barriers through inclusive solutions such as multilingual virtual assistants, adaptive interfaces, and accessible platforms. These innovations not only enrich the visitor experience, but also promote the participation of local communities in the construction of heritage narratives, integrating traditional knowledge and fostering cultural diversity.

However, the implementation of these technologies poses ethical, technical, and social challenges that require priority attention. Among the main challenges identified are the need to ensure the transparency and explainability of algorithms, the protection of personal data, the preservation of cultural authenticity, and the mitigation of algorithmic biases. Added to this are structural limitations such as the lack of interoperability between platforms, technological dependence, the long-term sustainability of the solutions implemented, and the persistent digital divide that affects communities with less access to technological infrastructure.

In this context, it is essential to move towards the development of sectoral ethical frameworks, inclusive governance models, and technological integration strategies that prioritize sustainability, equity, and meaningful participation. It is also recommended to further empirical research that evaluates the real impact of AI on heritage conservation and enhancement, as well as to explore its articulation with emerging technologies such as the Internet of Things (IoT), extended reality, and open digital ecosystems. Only through a critical, interdisciplinary, and people-centered approach will it be possible to ensure that technological innovation effectively contributes to the preservation, accessibility, and relevance of cultural heritage in the digital age.