Generative AI in hospitality and tourism: a dual-stakeholder perspective on tourist and workforce experience dynamics Open Access

Artificial intelligence (AI) has evolved from a niche field of computer science in the mid-20th century into one of the most transformative forces shaping the global economy. Often cited as a key driver of the Fourth Industrial Revolution (Krafft et al., 2020), AI’s applications now extend far beyond automation to include advanced decision-making, creativity and human-like reasoning (Hermann and Puntoni, 2024). Among its most disruptive advances is Generative AI (GenAI), exemplified by tools such as OpenAI’s ChatGPT, which are redefining how content, insights and experiences are produced at scale.

The extent of this transformation is remarkable. The global AI market is projected to exceed $4.8tn by 2033, growing at an annual rate of 37.9% (United Nations, 2025). Adoption is accelerating across industries, with more than 78% of large enterprises expected to integrate AI by 2025, driving measurable improvements in productivity, decision-making and customer engagement (McKinsey & Company, 2025).

Within this broader technological revolution, the hospitality and tourism (HT) sector, which contributes nearly 10% of global gross domestic product and employs one in ten people worldwide (WTTC, 2025), is also taking part in this technological shift. As new technologies are integrated, many foundational processes are being redefined, with AI gradually deployed to personalize services, streamline operations and enhance guest experiences (Dogru et al., 2025).

Yet, AI’s impact in HT is uneven, with effects on performance, productivity and service quality varying by context and application (Mohapatra and Mishra, 2024). Despite heavy investments, many organizations still struggle to realize AI’s full value because of fragmented data, integration issues, ethical and regulatory concerns and workforce resistance (Fouad et al., 2024; Husain, 2025).

Crucially, AI’s influence often manifests asymmetrically across stakeholders, enhancing guests’ experiences while inadvertently eroding employee autonomy or improving operational efficiency while diminishing the authenticity of human interactions (Prentice and Nguyen, 2020). These tensions are particularly pronounced in HT, where AI is not hidden in the background but directly embedded in the service encounter itself. This centrality makes the sector uniquely sensitive to the dual impacts of GenAI and highlights the need for approaches that consider both customer and workforce perspectives as interdependent rather than isolated dimensions of technological change.

Existing studies on GenAI in HT frequently adopt narrow perspectives, focusing either on customer-facing applications or workforce impacts in isolation. For example, Khan and Khan (2024) examine marketing-related opportunities and risks without linking them to customer outcomes, while Dogru et al. (2025) adopt a stakeholder lens without exploring interactions between stakeholder domains.

Because the field is still quite recent, many studies remain largely descriptive in nature, seeking to map the phenomenon rather than developing integrative theoretical frameworks that link concepts (Fouad et al., 2024). As a result, a critical blind spot persists: there is currently no integrative framework explaining how customer experience and workforce outcomes interact and co-evolve in the context of GenAI adoption. This gap limits our understanding of the mechanisms through which AI-driven transformation shapes service quality, employee well-being and consequently organizational performance. Without such a framework, organizations risk implementing technologies that optimize one domain while inadvertently harming the other, ultimately constraining value creation and long-term competitiveness. Therefore, we propose the following research question:

This study addresses this gap through a systematic literature review of GenAI’s evolving role in HT, with a distinctive focus on its dual impact on customers and employees. The significance of this review lies in its ability to synthesize fragmented research and provide a unified analytical lens for understanding these dynamics. Our contribution is twofold. First, we offer a structured synthesis of where, how and why GenAI is being implemented across the sector and how it is being researched. Second, we introduce the concept of GenAI-shaped Tourist Experience–Workforce Dynamics, which explains how customer and workforce outcomes align, diverge or interact to shape value co-creation or co-destruction. By framing these dynamics as an ideally constitutive system, this study advances theoretical understanding of GenAI’s dual impact and informs the development of more balanced, human-centered strategies for AI adoption.

Recognizing the multidisciplinary and context-sensitive nature of the HT industry, this study adopts an exploratory and integrative approach to capture the blooming discussion on GenAI and its applications within this industry’s context. To achieve this, we conducted a systematic literature review, which offers the necessary transparency and rigor to synthesize research evidence in fields that are both emerging and fragmented across disciplines (Tranfield et al., 2003).

The review process followed the Preferred Reported Items for Systematic Reviews and Meta-Analysis (PRISMA) protocol, which is structured around four steps: identification, screening, eligibility and inclusion (Figure 1) (Moher et al., 2009). To ensure academic rigor and credibility, we restricted our search to peer-reviewed journal articles indexed in Scopus and Web of Science (WoS), two databases widely recognized for their broad coverage of high-quality research, particularly in the social sciences fields (Mongeon and Paul-Hus, 2016). Building on this foundation, a targeted search string was designed based on a preliminary review of GenAI–HT literature. This string included terms related to the technology of interest (GenAI) and the sectoral context. To maximize relevance and specificity the search was limited to the “Article Title, Abstract, Keywords” setting, ensuring that only studies directly connected to GenAI in HT were retrieved. The initial search yielded 553 articles across the two databases (Table 1).

After the identification step, the research team then applied clearly defined inclusion and exclusion criteria, refined through discussion, to ensure that only the most relevant and methodologically robust studies were retained (Table 2). Any studies that did not meet these criteria were excluded, leaving us with 279 of papers eligible to be downloaded and reviewed.

After removing duplicates, 165 unique articles remained (WoS = 19; Scopus = 146). Titles, abstracts and keywords were then reviewed to select only empirical studies that directly address GenAI and related tools in the HT industry. At this stage, we intentionally excluded theoretical or conceptual works, as well as studies discussing broader AI applications or offering conceptual reviews. This process resulted in a final pool of 89 articles eligible for analysis.

Throughout the screening process, inclusion decisions were double-checked for consistency, and any ambiguous cases were revisited and discussed to ensure alignment with the study’s objectives. To enhance the transparency of the screening process, two researchers independently reviewed the titles, abstracts and keywords of all retrieved records. Disagreements were resolved through discussion until consensus was reached. Once the final set of articles was established, the analysis proceeded in several stages to capture both the breadth and depth of GenAI research in HT. First, we conducted a descriptive overview to map publication trends and disciplinary distributions. This was followed by a qualitative content analysis of all 89 articles, with each paper read in full and manually coded using an inductive approach to identify recurring patterns, themes and conceptual linkages (Krippendorff, 2013). Coding categories were developed and refined iteratively, focusing on the types and contexts of GenAI applications, their value for both customers and employees and the main opportunities and challenges highlighted in the literature. To ensure the effectiveness and reliability of the coding scheme, two researchers first jointly coded a subset of ten articles to develop an initial category framework. This framework was then iteratively refined through multiple rounds of testing and discussion.

The results of this review are, therefore, presented in three main sections: a descriptive analysis of the field, an identification of key gaps and limitations revealed by this analysis and the introduction of the Tourist Experience and Workforce Dynamics Framework, which addresses these gaps by synthesizing opportunities and challenges across customer and workforce domains.

This section provides a descriptive analysis of the findings presented individually in  Appendix. The table includes information on the author, year of publication, journal, theoretical approach, method, sample, context, GenAI tool and whether the paper focused on employees or customer and identified mostly advantages or challenges in the application of GenAI in HT.

Literature on GenAI in HT is still nascent but rapidly expanding. The earliest studies appeared in 2023 (5.61%), grew modestly in 2024 (23.59%) and peaked in 2025 with 59 publications (66.29%). With four early-access papers from 2026, GenAI research shows strong momentum and is set to remain a pulsing topic. Regarding publication sources, research is widely dispersed across journals (42 in total), though International Journal of Hospitality Management (11 papers) and Current Issues in Tourism (10 papers) stand out as leading sources.

The context analysis hints a significant imbalance in the literature: GenAI research is overwhelmingly tourist-focused, with 69.66% of studies examining customer perspectives. Employee-focused research remains limited (19.10%), while only 4.49% address both groups. Six studies focus solely on GenAI tools without considering human stakeholders. This uneven distribution shapes the field’s insights. Tourist-oriented studies often emphasize convenience, personalization and enjoyment (Arora et al., 2024; Pham et al., 2024), whereas those involving employees highlight risks such as deskilling, ethical challenges and cultural misalignment (Demir and Demir, 2023; Dogru et al., 2025). Collectively, these patterns underscore a fragmented evidence base. Opportunities and challenges vary across stakeholders, yet direct comparisons remain rare, obscuring the interdependencies that shape GenAI’s adoption and impact.

The landscape of tools studied adds another layer of complexity. In all, 17 GenAI platforms are mentioned, with ChatGPT dominating more than half the sample (51.68%). Yet 38.20% of studies fail to specify the tool used, creating transparency and replicability concerns and making cross-study comparison difficult.

The field is relatively well anchored theoretically (62.92% apply at least one theory), but conceptual diversity reveals deeper tensions. The Technology Acceptance Model (TAM) is dominant (approached by five studies), reflecting a rationalist view of GenAI adoption based on perceived usefulness and ease of use. However, this perspective is increasingly challenged by alternative lenses. Innovation Resistance Theory and perceived risk frameworks foreground privacy, accuracy and ethical concerns as barriers to adoption, suggesting that rational cost–benefit evaluations are insufficient on their own. Emotional and experiential models, such as PAD theory (Xu et al., 2025) and parasocial interaction theory (Duong et al., 2025), further complicate the picture by demonstrating that affective responses like enjoyment, trust and companionship can override utilitarian considerations. These competing explanations reveal a field grappling with foundational doubts: is GenAI adoption primarily a calculated decision, a risk-avoidance behavior or an emotionally driven process? Few studies attempt theoretical integration, and even fewer explore how these mechanisms might interact or conflict across stakeholder groups and contexts.

In line with the tourist-focus approach, research remains dominated by quantitative methods (64.05%), particularly survey-based studies (38%), which privileges generalizability but often oversimplifies a rather complex topic. This has resulted in a strong bias toward measuring acceptance rather than exploring its antecedents, dynamics or consequences in depth. By contrast, qualitative research (22.47%), although less frequent, successfully provides richer insights into employee anxieties, ethical dilemmas and cultural frictions (Wang, 2025; Limna and Kraiwanit, 2023). The relative scarcity of longitudinal, comparative and mixed-methods designs limits our understanding of how attitudes evolve over time or diverge across cultural and organisational contexts. Moreover, a small number of studies rely on simulated data (e.g. ChatGPT-generated responses), raising questions about validity and the robustness of empirical claims.

Overall, the descriptive results reveal a fragmented field: most studies isolate tourists from employees, emphasize either opportunities or challenges and privilege single-theory or single-method approaches. This siloed perspective reflects the early stage of research in this field and underscores the need for more integrative models. Without them, current studies fail to capture how these dynamics interact to co-create or co-destroy value. These limitations point to the necessity of a holistic framework, which the following section introduces.

Among the most cited opportunities, personalization stands out (Table 3). GenAI systems can generate highly tailored travel recommendations, itineraries and accommodation suggestions based on behavioral patterns, preferences and contextual data (Shin and Kang, 2023; Gao and Wang, 2024; Han et al., 2024; Casaló et al., 2025). This capacity for real-time customization allows travelers to co-design their journeys interactively, turning GenAI into an indispensable planning companion (Arora et al., 2024; Ghorbanzadeh et al., 2025; Xu et al., 2024a, 2024b). Beyond convenience, personalization reflects a shift toward algorithmic co-creation, where value emerges from iterative human–AI collaboration rather than one-way service delivery.

Importantly, personalization also extends to inclusivity. GenAI enhances accessibility for travelers with disabilities, provides real-time translation and supports cross-cultural communication, thereby broadening participation and reducing barriers (Dogru et al., 2025; Suanpang and Pothipassa, 2024). These features position GenAI as a democratizing force in tourism, capable of promoting equity and engagement across diverse traveler groups.

Yet this empowerment contains inherent tensions. Personalized recommendations may reduce serendipity and constrain cultural immersion, reinforcing algorithmic path-dependency and experiential homogeneity (Zhang et al., 2026; Tosyali et al., 2025; Pham et al., 2024). Moreover, employees increasingly serve as mediators between algorithmic suggestions and guest expectations, stepping in when personalization falters or becomes overly prescriptive. From a labor perspective, personalization shifts cognitive and emotional demands rather than eliminating them, illustrating a symbiosis–strain paradox, whereby technology augments guest value while intensifying employee interpretive labor.

On the operational side, efficiency and automation represent another major advantage. GenAI automates routine functions such as bookings, inquiries and feedback management, through intelligent chatbots and virtual assistants available 24/7 (Wayne Litvin and Pei-Sze Tan, 2024; Ng et al., 2024; Wan, 2024). These systems continuously learn, improving contextual awareness and freeing staff for creative or high-touch tasks (Dogru et al., 2025; Koc et al., 2023). Consequently, GenAI not only reduces response time but also enhances service consistency and scalability.

However, efficiency gains also reshape labor dynamics. As routine tasks diminish, remaining work often becomes more complex, emotionally intense and cognitively demanding. This shift is consistent with skill-bifurcation effects observed in digital transformation research. Thus, automation may alleviate workload yet simultaneously create new pressures around monitoring, exception-handling and emotional authenticity.

GenAI is also transforming marketing, creativity and strategic decision-making. By analyzing vast data sets, it enables hyper-personalized campaigns and rapid content generation (Dogru et al., 2025), supports immersive digital experiences such as virtual tours and AR guides (Chakraborty, 2024) and fuels innovation in experience design (Luo et al., 2025; Huang et al., 2025a, 2025b). Strategically, these capabilities support experience orchestration, allowing firms to prototype and scale novel service concepts more rapidly than before.

Financially, GenAI contributes to cost optimization and agile resource deployment (Saleh, 2025; Zhang and Prebensen, 2024). Yet such efficiencies raise normative questions about equitable value capture: while firms benefit from scalability, employees may experience role compression and identity disruption, reinforcing the need for governance frameworks that ensure technology complements, rather than substitutes, human capability.

Taken together, GenAI’s opportunities are multilayered and interdependent. Personalization can deepen engagement yet reshape labor meaning; automation enhances efficiency while transforming skill structures; creative augmentation fuels innovation while challenging traditional roles. These dynamics emphasize that GenAI does not simply add value but reconfigures value creation systems. To harness its potential requires managing paradoxical tensions between efficiency and empathy, automation and authenticity and co-creation and control, ultimately advancing a human-centered, relational approach to digital hospitality.

Despite the transformational promise of GenAI in HT, its integration also brings complex challenges affecting both customers and employees (Table 4).

One of the most widely recognized risks concerns accuracy and misinformation. GenAI-generated outputs often include factual errors or “hallucinations,” leading to misleading travel advice or incorrect information about destinations and services (Xu et al., 2024a; Zhang and Prebensen, 2024; Ayyildiz et al., 2025). Such inaccuracies can erode consumer trust and brand credibility, particularly where algorithmic output is treated as authoritative. Critically, these failures do not remain confined to digital channels. Employees must frequently resolve inconsistencies and guest frustrations, effectively absorbing the “last-mile burden” of AI errors. This dynamic reinforces the human fallback paradox: technology promises labor reduction but often redistributes complexity to frontline staff through emotional and cognitive labor demands.

This challenge links to a deeper tension around authenticity and trust, foundational to hospitality experiences. As AI-generated language and emotional expressions increasingly mimic human interaction (Arora et al., 2024; Wan, 2024; Thakur et al., 2025), guests may experience ambiguity in relational cues and interpersonal meaning. The replacement or dilution of human warmth risks an erosion of affective value, consistent with debates on algorithmic authenticity in service-dominant logic. For employees, the perceived commodification of emotional labor and service warmth raises concerns about identity, dignity and the shifting meaning of “care” in technologically mediated encounters (Koc et al., 2023; Mladenović et al., 2024).

Cultural and ethical concerns further complicate AI integration. GenAI models trained on broad, often Western-centric corpora may fail to reflect local nuance, emotional norms or hospitality traditions (Christensen et al., 2025; Zhao et al., 2024). This may homogenize tourism experiences and weaken “sense-of-place,” a critical dimension of value creation in experiential services (Koc et al., 2023). Paradoxically, while GenAI promotes personalization, it can drive cultural standardization and emotional flattening, illustrating a personalization–homogenization paradox. For many travelers, this undermines the very experiential richness and cultural distinctiveness they seek.

Privacy and ethical governance concerns are also growing as GenAI relies on large-scale data capture and behavioral inference, which raises anxieties among both guests and employees (Limna and Kraiwanit, 2023; Altinay et al., 2025; Chakraborty, 2024). Ambiguous regulation and uneven global standards compound fears around surveillance and data misuse (Dwivedi et al., 2024). These dynamics introduce a “trust gap,” where even high-performance AI systems may be rejected because of perceived ethical opacity, reaffirming that legitimacy, not only functionality, shapes technology acceptance (Yaşar and Yayla, 2025).

Operationally, GenAI integration remains resource-intensive and uneven. Successful implementation demands capital, infrastructure, technical literacy and continuous training (Zhu et al., 2024). Smaller firms, which is a large share of tourism operators, face disproportionate barriers, risking a widening digital divide. Meanwhile, employees confront job insecurity, role redefinition and intensified monitoring (Dogru et al., 2025; Limna and Kraiwanit, 2023; Wang et al., 2024).

Finally, the dynamic nature of GenAI introduces strategic and psychological uncertainty. Rapid innovation cycles and shifting ethical norms challenge organizational planning and employee confidence (Zhang et al., 2025). Guests, too, struggle to form stable expectations about service quality and data handling. This volatility illustrates a moving-target risk in digital transformation: technology evolves faster than organizational capacity, governance structures and social norms.

In summary, the challenges of GenAI in HT are not isolated constraints but systemic tensions: between automation and authenticity, personalization and privacy, efficiency and empathy and global scalability and cultural specificity. Addressing them requires not only technical solutions but also sociotechnical alignment, ethical stewardship and workforce-centered change management. In this sense, sustainable GenAI adoption depends on safeguarding the very relational and cultural foundations that distinguish hospitality from transactional service delivery.

While the preceding sections delineated the advantages and challenges of GenAI adoption in HT, the listing of these elements revealed that these dynamics are not discrete but interwoven. The same mechanisms that enable personalization, efficiency and creativity are frequently pointed to causing tensions around authenticity, emotional labor and trust. This suggests that GenAI’s effects cannot be meaningfully understood when customer and employee outcomes are examined in isolation. Yet, as our review shows, existing studies overwhelmingly adopt siloed perspectives, focusing either on guests or on employees. In contrast, we argue that GenAI introduces a system of interdependent dynamics in which technological shifts in one domain reverberate through the other. We term this configuration GenAI-shaped Tourist Experience–Workforce Dynamics, which captures the reciprocal evolution of guest experiences and employee practices under AI augmentation.

Across studies, we were able to identify three relational modes characterize how GenAI reconfigures this system: alignment, divergence and negotiation (Figure 2).

Alignment occurs when GenAI’s affordances simultaneously enhance customer experience and employee performance. Automation of routine inquiries, for instance, frees staff to engage in more meaningful, creative or emotionally resonant tasks, while guests benefit from faster and more consistent service delivery (Ng et al., 2024; Dogru et al., 2025). Similarly, data-driven personalization enables employees to anticipate needs and deliver contextualized experiences that deepen perceived authenticity and satisfaction. In such cases, GenAI operates as a collaborative complement, supporting human capabilities rather than substituting them. This alignment illustrates a technologically mediated form of co-creation in which value emerges from distributed human–machine collaboration across the service ecosystem.

However, divergence arises when GenAI optimizes one domain at the expense of the other. Personalized recommendations may improve guest convenience while generating cognitive strain for staff tasked with correcting algorithmic errors or managing unrealistic expectations, a manifestation of the human fallback paradox. Similarly, standardization for operational efficiency can erode local distinctiveness and relational authenticity, undermining both employees’ sense of identity and guests’ search for meaningful connection (Koc et al., 2023; Zhang and Prebensen, 2024). This misalignment reflects a redistribution rather than elimination of complexity, as the pursuit of technological optimization introduces new emotional and ethical burdens across the service encounter.

Finally, negotiation captures the ongoing recalibration through which both tourists and employees adapt to AI-augmented realities. Tourists increasingly interpret AI outputs through a lens of human mediation, valuing hybrid service forms that blend technological precision with human warmth. Employees, in turn, reframe their professional roles, from information providers to curators of authenticity and interpreters of algorithmic intelligence. This dynamic process of mutual adjustment underscores that GenAI integration is not a linear adoption but an evolving relationship that continually reshapes expectations, competencies and affective boundaries.

Taken together, these dynamics reveal that the effects of GenAI in HT are systemic rather than additive. Opportunities and challenges co-exist within a relational matrix where each advantage contains its shadow tension: personalization can breed homogenization; automation can deepen emotional labor; data-driven insight can erode privacy or trust.

As we conceptualize these relationships as GenAI-shaped Tourist Experience–Workforce Dynamics, we aim to call attention to these relations that allow for a more nuanced understanding of how value is simultaneously co-created and co-destroyed across interconnected human and technological systems.

To better illustrate these dynamics, we present a visual representation in Figure 2. Conceptually depicted as a Venn diagram, there are two interlocking shapes: experience dynamics and workforce dynamics. At their intersection, the relational core where the two domains converge and mutually shape one another, reinforcing the theoretical message of “one cannot be understood without the other.” Surrounding the dynamics are the emergent modes of the system, which stand in for the possible states that arise from the interplay between customer and workforce outcomes.

Regarding the practical implications, the results call for the attention of managers, who must recognize GenAI as a relational intervention rather than a purely technical one. Because customer experiences and employee workflows shape each other, decisions about automation, personalization or AI-based support should be evaluated for their dual impact. For instance, deploying AI without adjusting job design may improve efficiency but erode service authenticity and employee motivation. Second, organizations should strategically pair technological implementation with human capability development. For example, integrating AI itinerary tools alongside staff training enables employees to contextualize, personalize and enrich AI suggestions, transforming automation into a co-creative process rather than a transactional one. Third, firms should develop clear communication strategies around human–AI collaboration and demonstrate how technology supports rather than replaces human service. Doing so can foster trust, reduce employee resistance and increase guest acceptance. Fourth, policymakers should incorporate this relational understanding into regulatory design, creating guidelines that safeguard both workforce well-being and consumer interests. Policies addressing transparency, accountability and skill transitions can mitigate negative externalities and encourage ethical AI use.

Theoretical contributions lie primarily in bridging fragmented perspectives. While prior studies typically examine guest-facing applications or employee outcomes in isolation, our framework conceptualizes them as mutually constitutive and co-evolving. This shift offers a more accurate explanation of how technological change unfolds in service contexts: guest expectations and satisfaction are shaped by workforce responses, while employee practices evolve in response to changing service demands. In doing so, this study reframes GenAI’s impact not as a linear input–output process but as a relational system of co-evolving interactions. Moreover, the proposed framework demonstrates that value creation and destruction emerge from interaction effects rather than isolated drivers. By highlighting how different combinations of personalization, automation, emotional engagement and skill dynamics shape service outcomes, it enriches service-dominant logic and offers a nuanced understanding of GenAI’s non-linear impacts. It also provides a basis for linking micro-level processes (e.g. trust and deskilling) to macro-level consequences (e.g. loyalty and performance). Finally, the study advances digital transformation theory by framing GenAI as a socio-technical phenomenon. Beyond technical capabilities, its effectiveness depends immensely on relational, cultural and emotional factors within service encounters. The framework identifies conditions under which GenAI enhances authenticity, well-being and trust.

Moving to future research directions, we highly invite new studies for the need of empirical work to adopt a comparative stakeholder lens. Current studies examine guests and employees in isolation. Future studies should adopt a comparative stakeholder lens to investigate how their responses interact and shape one another over time.

The framework shows that outcomes emerge from interaction effects, not single variables. Future research should, therefore, investigate how specific combinations, such as personalization plus employee training or automation plus cultural sensitivity, influence value co-creation or co-destruction. Under what circumstances do these synergies enhance authenticity and satisfaction, and when do they lead to disengagement or mistrust? Experimental or quasi-experimental designs could test these causal relationships.

Future work should interrogate the tensions and trade-offs inherent in GenAI adoption. For example, when personalization boosts satisfaction but compromises privacy, how do managers and policymakers prioritize these competing outcomes? Should regulatory frameworks privilege transparency over convenience? These questions could be tackled using comparative case studies or policy experiments to shed light on how different actors navigate these dilemmas and what governance mechanisms best support balanced outcomes.

Another limitation of current literature stands against the fact the grand majority of studies over rely on the study of ChatGPT. Future research should test whether similar patterns emerge with alternative systems, for instance, domain-specific chatbots, image-based generative tools or multimodal assistants.

As GenAI is an everchanging tool, it means that its impacts are dynamic rather than static. Future research should examine how trust, service authenticity, workforce skills and guest expectations evolve over time as technologies mature. For instance, do employees initially resist but later embrace AI as they acquire new skills? Does guest enthusiasm wane as novelty fades or deepen as AI becomes more personalized? These questions could be approached using longitudinal panel studies to capture these adaptive processes.

In the age of algorithmic imagination, hospitality is no longer merely about service; it is about cohabiting with intelligence. GenAI systems do not simply automate tasks; they converse, recommend and imagine alongside humans, quietly rewriting what it means to host and to be hosted. From itinerary planning to emotional support, AI is becoming an invisible co-designer of the tourist journey and the workplace alike.

This study has traced that transformation through a systematic review of 89 publications on GenAI in HT. We found a literature rich in optimism yet divided by disciplinary silos: one celebrating personalized, frictionless experiences, the other warning of dehumanization and algorithmic opacity. To bridge this divide, we introduced the Tourist Experience–Workforce Dynamics framework, which positions GenAI not as a technological artifact but as a constitutive system where guest and employee experiences co-evolve through mutual adaptation, negotiation and resistance.

Our synthesis reveals three defining characteristics of this new era: Augmentation represents GenAI’s dual capacity to amplify human potential while intensifying cognitive and emotional demands. Negotiation reflects the continuous recalibration of roles, authority and authenticity as employees and guests learn to coexist with algorithmic decision-making. Divergence is defined by how GenAI simultaneously unifies and fragments experiences, streamlining operations yet diversifying expectations, empowering users while deepening digital divides. These dynamics suggest that value in AI-mediated hospitality is co-created through an ongoing balancing act between efficiency and empathy, automation and autonomy.

For practitioners, these insights demand a rethinking of leadership, training and design: organizations must define a human–AI choreography rather than manage discrete technologies. For researchers, the agenda ahead lies in decoding this choreography and examining how GenAI reshapes emotional labor, cultural authenticity and power asymmetries across the service ecosystem.

Ultimately, understanding GenAI’s role in hospitality is not about forecasting technological futures but about reimagining the human condition within them. As hospitality moves from service to symbiosis, the challenge and opportunity is to ensure that intelligence, whether human or artificial, continues to serve what is most distinctly human: connection, meaning and care.

This work was supported by the Research Unit on Governance, Competitiveness and Public Policies (UIDB/04058/2020)+(UIDP/04058/2020), funded by national funds through FCT-Fundação para a Ciência e a Tecnologia.

The table describes the authors' contributions to the paper. Maria Leonor Ferreira contributed to the conception, data collection, analysis, and drafting of the article. Elisabeth Kastenholz contributed to the conception, data analysis, critical revision, and final approval of the version to be published.