Navigating the AI horizon in hospitality: a novel classification and future research agenda

Abstract

Purpose

This study aims to evaluate Artificial Intelligence (AI) research in the hospitality industry based on the service AI framework (mechanical-thinking-feeling) and highlight prospective avenues for future inquiry in this growing domain.

Design/methodology/approach

This paper conceptualizes timely concepts supported by research spanning multiple domains.

Findings

This research introduces a novel classification for the domain of AI hospitality research. This classification encompasses prediction and pattern recognition, computer vision, NLP, behavioral research, and synthetic data generation. Based on this classification, this study identifies and elaborates upon five emerging research topics, each linked to a corresponding set of research questions. These focal points encompass the realms of interpretable AI, controllable AI, AI ethics, collaborative AI, and synthetic data generation.

Originality/value

This viewpoint provides a foundational framework and a directional compass for future research in AI within the hospitality industry. It pushes the industry forward with a balanced approach to leveraging AI to augment human potential and enrich customer experiences. Both the classification and the research agenda would contribute to the body of knowledge that will guide the industry toward a future where technology and human service coalesce to create unparalleled value for all stakeholders.

Citation

Pan, T. and Fu, R.J.C. (2024), "Navigating the AI horizon in hospitality: a novel classification and future research agenda", International Hospitality Review, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/IHR-01-2024-0003

I think that because of artificial intelligence, people will have more time to enjoy being human beings. ----- Jack Ma, co-founder and chair of Alibaba Group (Hotel of the Future owner)¹

1. Introduction

In recent years, interest in artificial intelligence (AI) in hospitality research has grown significantly. Many investigations have focused on consumer behavior in the context of AI adoption, reflecting its importance in the hospitality industry (Chi, Jia, Li, & Gursoy, 2021; Gursoy, Chi, Lu, & Nunkoo, 2019; Ribeiro, Gursoy, & Chi, 2022; Shi, Gong, & Gursoy, 2021). To facilitate a deeper comprehension of AI's role in the hospitality domain, Bock, Wolter, and Ferrell (2020) coined the term “service AI,” signifying the strategic employment of technology to deliver value in both internal and external service environments through the integration of sensing, learning, decision-making, and actionable capabilities. Against this backdrop, the present review aims to offer an inclusive perspective on the trajectory of the hospitality industry and AI research.

Machine learning (ML) and deep learning (DL) often come to mind when discussing AI in the industry. However, it is important to avoid limiting the field of AI solely to ML and DL, as AI encompasses a broader spectrum of capabilities. Thus, a clear and comprehensive definition of AI that accommodates its associations with ML and DL is provided in this paper. According to Bishop (2006), AI refers to the capacity of machines to execute tasks commonly associated with human intelligence—service delivery robots, hotel check-in robots, and advanced computer systems for capturing consumers’ biological information for security and safety.

ML represents a specific subset of AI wherein computers acquire the ability to perform specific tasks based on distinct sets of experiences. DL, conversely, constitutes a subfield of ML that employs artificial neural networks (ANNs) to simulate human thought processes (Bishop, 2006; Izenman, 2008). By utilizing ANNs, DL aims to enable machines to learn and adapt autonomously through intricate layers of abstraction, facilitating sophisticated decision-making capabilities. Figure 1 illustrates the interconnectedness among AI, ML, and DL, showcasing the hierarchical relationship and depicting how each concept fits into the broader framework of AI.

AI has profoundly influenced marketing literature, prompting numerous scholars to conduct reviews and explore research agendas from diverse perspectives. De Bruyn, Viswanathan, Beh, Brock, and Von Wangenheim (2020) addressed two crucial challenges associated with AI adoption in marketing: first, the potential for managers to be lured by the allure of AI-driven solutions, leading them to underestimate associated risks, limitations, and pitfalls; and second, the risk of misjudging areas where AI is most likely to encounter setbacks. Their study emphasized the importance of marketing organizations facilitating and systematizing interactions between AI and marketing stakeholders, fostering an ecosystem that cultivates a form of “intimacy” between AI and consumers through two-way observation, imitation, and practice.

Vlačić, Corbo, e Silva, and Dabić (2021) conducted an extensive review of 164 published articles on AI and marketing that revealed four major research themes: marketing channels, marketing strategy, performance, and segmentation, targeting, and positioning (STP). Furthermore, the authors identified six potential research trends in this field, namely the acceptance of AI technology, adoption and use of AI technology and applications, the transformation of the labor market and marketers' competencies, the role of institutional support, the significance of data protection and ethics, and the impact of the COVID-19 pandemic on AI in marketing.

Despite these contributions, few scholars to date have explored existing research or investigated future potential research topics in this domain. Among recent research, Bock et al. (2020) and Huang and Rust (2021) have made notable advancements. Bock et al. (2020) conducted an examination of AI's current and potential impact on prominent service theories focusing on the service encounter, exposing significant deficiencies in nine leading service theories and concepts concerning their ability to predict AI-related phenomena and research. Consequently, the authors emphasized the need for new theory development and highlighted the relevance of technology-centric theories. On the other hand, Huang and Rust (2021) developed a strategic framework for utilizing AI to engage customers and gain various service benefits. They categorized AI into three dimensions: mechanical (standardization), thinking (personalization), and feeling (relationalization). This framework recommends using mechanical AI for routine and repetitive service tasks, thinking AI for data-based, analytical, and predictive tasks, and feeling AI for experience-based or emotional tasks that require interaction and communication.

Despite AI's increasing prominence in the hospitality industry, scholars and practitioners must remain cognizant of the challenges and drawbacks associated with its implementation. Law, Lin, Ye, and Fong (2023) adopted the theory-context-methods framework to analyze current AI research in the hospitality industry. They mainly classified the AI hospitality research into two main streams: AI application research and AI methods research, followed Doborjeh, Hemmington, Doborjeh, and Kasabov (2022). Their findings indicated that AI methods research was identified as mostly data-driven and lacked theoretical engagement. However, AI methods research could be more powerful since consumers’ insights are significantly revealed by data. This study aims to assess current AI methods research in the industry using the service AI framework proposed by Huang and Rust (2021), conduct a comprehensive trade-off analysis for utilizing ML models, and delineate potential avenues for future inquiry.

The structure of this paper is as follows: the next section develops a new hospitality AI theoretical framework by classifying existing research and integrating it with Huang and Rust’s (2021) service AI framework. Section 4 provides a research agenda for AI in the hospitality industry. Finally, Section 5 concludes the paper by summarizing its key contributions while also discussing its limitations and proposing opportunities for further study.

2. AI in hospitality research

The three dimensions of service AI – mechanical, thinking, and feeling – each offers unique contributions to the industry (Huang & Rust, 2021). Mechanical AI involves automated tasks and processes, improving efficiency and reducing human intervention. Thinking AI refers to the cognitive capabilities of AI systems such as problem-solving, decision-making, and learning. Feeling AI encompasses emotional intelligence, enabling machines to understand and respond to human emotions, thereby enhancing user experiences. For both scholars and managers, harnessing AI as a strategic tool holds great significance in advancing research and enhancing hospitality operations. Understanding the applicability of AI across different dimensions is critical to leveraging its potential effectively. To facilitate a comprehensive understanding of AI utilization in the hospitality industry, we have developed a classification framework for the relevant literature, which can be divided into five categories: prediction and pattern recognition, computer vision, natural language processing (NLP), behavioral research, and synthetic data generation. Different scholarly contributions have been made to distinct categories. Table 1 presents a comprehensive literature review in this field.

Descriptions and examples of each category are provided in the following:

Prediction and pattern recognition encompasses a range of AI applications primarily concerned with predictive modeling, data analysis, and identification of patterns within extensive datasets. This AI domain is of utmost importance in forecasting customer behavior, demand trends, and hospitality requirements. Specifically, it finds significant applications in analyzing supply-and-demand data in the hospitality industry. Developing an exceptionally accurate and explicable prediction model holds immense potential to optimize operational efficiency and reduce variability (Jordan & Mitchell, 2015). In the realm of hospitality research, scholars frequently employ ML models such as random forest and Long Short-Term Memory (LSTM) compared to traditional benchmark approaches (canonical statistics) and enhance prediction accuracy (Bi, Li, & Fan, 2021; Höpken, Eberle, Fuchs, & Lexhagen, 2021; Sun, Wei, Tsui, & Wang, 2019). Notably, various scholarly articles have introduced novel frameworks for identifying data patterns and conducting forecasting such as the STL-DADLM proposed by Zhang, Li, Muskat, and Law (2021) and the ML-based search query selection framework presented by Li, Li, Pan, and Law (2021). These studies have made significant contributions to the literature on time-series forecasting and offer notable implications for enhancing service practices.

Computer vision pertains to visual data analysis, enabling service providers to automate various processes, including object detection, image recognition, and facial expression analysis. This area of research offers valuable insights into consumer behavior by analyzing photos and images in the hospitality industry, playing a crucial role in enhancing customer service experiences, product promotion, and advertising. For instance, Barnes (2022) conducted a study exploring the factors of visual color complexity in listing photographs on Airbnb and their impact on consumer behavior, revealing that accommodation providers should be attentive to color characteristics in listing images to enhance occupancy rates. Additionally, Deng and Li (2018) and He, Deng, Li, and Gu (2022) investigated user-generated content (UGC) photos using machine learning (ML) models to identify suitable advertising images for Destination Marketing Organizations (DMOs). Additionally, Li, Ji, Liu, Cai, and Gao (2022) employed deep learning (DL) algorithms to explore consumer sentiment derived from review photos, investigating the effects of photo sentiment on consumers’ perceived review helpfulness and review enjoyment.

NLP entails research combining computer linguistics with statistics, ML, and DL to examine consumer sentiment and behavior and develop effective marketing and management tactics. Despite the ubiquity of images and videos, text remains an indispensable tool in all academic fields (Gandomi & Haider, 2015). Within this domain, various applications such as chatbots, sentiment analysis, and language translation significantly impact customer engagement and communication. Since e-commerce is still rising, online reviews have become an invaluable element for businesses. Using NLP, Martinez-Torres and Toral (2019) identified the distinctive characteristics and subjects of misleading and honest evaluations in the hospitality industry. Chang et al.’s (2020) study was among the first to detect management reactions objectively using hotel reviews and responses using visual analytics and NLP. Leveraging TripAdvisor reviews, Hassan, Zerva, and Aulet (2023) devised a novel method for building context-driven brand personality categories employing psychometrics and NLP.

Behavioral research in the field of AI includes studies that use data obtained from surveys and experiments, focusing on two primary sub-categories: AI adoption in behavioral marketing and AI applications in understanding decision-making processes. In these studies, consumer perception data are typically collected through surveys and experiments to reveal insights into consumer psychology concerning the incorporation of AI in hospitality and decision-making processes. For instance, Chi, Gursoy, and Chi (2022) conducted a study to explore the attitudes of tourists toward the use of AI devices in tourism service delivery by gathering perception data through the mental simulation method. Employing the Multidimensional Emotion Questionnaire, Pantano and Scarpi (2022) analyzed the impacts of different types of AI on the emotions of service consumers. As chatbots gain increasing popularity, scholars have started examining the dynamics of engagement between consumers and service providers. Flavián, Pérez-Rueda, Belanche, and Casaló (2022) investigated how customers' technology readiness and service awareness influence their intention to use analytical AI investment services such as robo-advisors. Their findings revealed that feelings of technological discomfort positively influenced robo-advisor adoption, challenging conventional insights into technology adoption. Similarly, Xie, Pentina, and Hancock (2023) explored the relationship between chatbot engagement and psychological dependence, highlighting that developing an attachment to a social chatbot enhances positive consumer engagement. Due to the inherent limitations of applying algorithms to survey and experiment data, the majority of research within this domain has focused on the implementation of AI in behavioral marketing. These studies have effectively addressed theoretical gaps in consumer understanding of AI adoption in marketing research and have provided valuable insights to businesses seeking a deeper comprehension of their customer's behaviors and preferences.

Synthetic data generation techniques have been prevalent since the 1980s in the domain of computer vision, primarily aimed at the validation and testing of novel computer vision algorithms (Nikolenko, 2021). However, these profound developments in deep learning and artificial intelligence have instigated a notable resurgence in interest in these synthetic data generation methods, along with their application in entirely novel domains. It is noteworthy that the exploration of this subject within the hospitality industry remains relatively underexplored. In this context, the recent research conducted by Huang and Rust (2022) introduced the concept of “AI customers,” denoting those customers who are either partially or entirely AI-driven entities. These entities can encompass AI augmenting human customers, either replacing them or even representing the customers themselves. Huang and Rust (2022) integrated the “AI customers” concept into the framework proposed in a previous study (2021) to examine AI's role in catering to human customers through the three dimensions of AI intelligence: mechanical AI as customers, thinking AI as customers, and feeling AI as customers. Additionally, the research advances a forward-looking perspective by outlining six potential avenues for future inquiry. These avenues encompass discerning the primary advantages stemming from the incorporation of AI customers, identifying the most compelling new applications of AI customers, gauging the manageability and control aspects of AI customer interactions, evaluating the benefits arising from the collaborative engagement of human customers with AI customers, pinpointing the specific demographic of human customers that gain the most from AI-driven interactions, and addressing issues pertaining to the ownership and agency of AI customers (Xu, Shieh, van Esch, & Ling, 2020; Huang & Rust, 2022). Given the rapid evolution of AI technologies, a thorough and comprehensive investigation into this subject holds substantial promise for researchers.

This paper introduces a comprehensive categorization of AI in the domain of hospitality research, which is subsequently integrated into the framework proposed by Huang and Rust (2021). The delineation of AI categories is as follows: Prediction and pattern recognition exhibit dual characteristics encompassing both mechanical AI and thinking AI. Entities falling under this research category undertake not only standardized data forecasting but also engage in the identification of consumer behavioral patterns. These insights are then directly furnished to managers for the purpose of tailoring services to individual preferences. Synthetic data generation is situated within the realm of feeling AI, as they can acquire knowledge and adjust based on experiential relationalization.

Furthermore, the remaining three research clusters exhibit an overlap between thinking AI and feeling AI. Thinking AI, in particular, exhibits the capacity to learn and adapt with a heightened comprehension of contextual intricacies, while feeling AI proves advantageous in cultivating and managing customer relationships, thereby catering to interpersonal dimensions (Huang & Rust, 2021). Within the domain of hospitality AI, namely computer vision, NLP, and behavioral research, a synthesis of insights emanating from both thinking and feeling AI is attainable. Hence, Figure 2 elaborates on the relationship between this classification and Huang and Rust's framework.

3. Research agenda

Incorporating the literature review (Table 1) and the research classification, this paper asserts that the scope of potential topics for AI hospitality research is as expansive as the breadth of subjects within canonical hospitality research. Firstly, the classification reveals the complexity of abstraction in AI applications within hospitality. Interpretable AI becomes a critical topic to ensure transparency and trust, especially as thinking AI and feeling AI become more prevalent in customer-facing roles, where decisions need to be understandable to both managers and consumers. With developing technology, increasingly sophisticated AI models boasting heightened precision and diminished computational expenses have emerged. In the realm of hospitality research, the prominence of AI interpretability transcends considerations of mere accuracy and computational expenditure. Mechanical AI could be investigated for its straightforward interpretability while feeling AI could pose challenges due to its complex and experience-based nature. As such, exploring this topic warrants deeper scholarly inquiry and scrutiny. Below are the research questions that could be explored in the future:

RQ1.

To what extent are AI models more interpretable than traditional statistics?

RQ2.

How can we develop interpretable AI models that effectively convey the rationale behind personalized hospitality recommendations to customers while maintaining accuracy and relevance?

RQ3.

What methods can quantify the trade-off between interpretability and performance, allowing hospitality providers to make informed decisions about model complexity and explainability?

RQ4.

To what extent are AI decisions understandable, particularly in finance or hotel services where transparency is crucial?

RQ5.

How can the concepts of “local” and “global” interpretability be applied in the context of hospitality AI, enabling both fine-grained explanations of individual predictions and high-level insights into model behavior?

By improving the transparency of how AI systems make decisions, companies can build greater customer trust, leading to increased customer loyalty and repeat business. Research could explore methods to make AI decision processes more understandable for non-technical stakeholders, which in turn could enhance customer satisfaction and adherence to regulatory requirements.

Secondly, AI customers emphasize emotional and relational interactions within the dimension of feeling AI, necessitating a balance between autonomy and control. The topic of controllable AI is justified by the need to manage the output of AI systems to ensure they remain within acceptable parameters of service delivery while maintaining the personal touch that is important in the hospitality industry. This research area assumes significance in its pursuit of constructing AI systems capable of producing outcomes within defined parameters (De Bruyn et al., 2020). Within the hospitality industry, controllable AI assumes heightened importance because it offers the potential to customize service interactions to align with diverse prerequisites and preferences. While AI's integration has gained pervasive recognition across academia and the industry, there remains a compelling need for deeper exploration of the dynamics governing the interplay between human employees and AI in work environments. Specifically, the empowerment of human employees to oversee and manage AI functionalities effectively warrants comprehensive investigation. Several research questions are presented below that might inspire future research in this area:

RQ6.

What methods can enable real-time adjustment of AI responses during customer interactions, allowing human agents to guide AI behavior as needed?

RQ7.

How can AI-driven recommendation systems be enhanced to provide users with a range of controllable options, allowing customers to customize their suggestions based on specific preferences or constraints?

RQ8.

What approaches can make AI-generated content adaptable to different cultural contexts, ensuring that the content is respectful and culturally appropriate for diverse customer bases?

RQ9.

What trade-offs exist between controllability and creativity in AI- generated content, and how can these trade-offs provide innovative yet controlled service interactions?

RQ10.

To what extent can AI contribute to enhance collaboration between human employees and AI systems in the service industry, allowing for seamless joint decision-making?

RQ11.

Whether AI is a service provider or customer, how can human employees efficiently manage it?

Research into finding the optimal balance between AI autonomy and human control could lead to more efficient operations, reducing labor costs and minimizing errors. This could involve developing frameworks that dictate when and how AI should act autonomously and when it should defer to human judgment, increasing both efficiency and customer satisfaction.

Thirdly, the advancement and widespread implementation of AI have foregrounded a multitude of ethical concerns—for instance, offensive responses generated by ChatGPT (Deng & Lin, 2022). AI ethics is critical and evolving, especially as AI becomes more integrated into various aspects of our lives such as chatbots, ChatGPT, service robot, and ML algorithms employed in business contexts. This can be linked to each dimension by assessing how ethical concerns such as privacy, bias, and misuse vary with the use of mechanical, thinking, or feeling AI. Ensuring that AI is developed and deployed in ways that are fair, transparent, accountable, and beneficial to the industry is necessary. The following research questions are provided for future research:

RQ12.

How can AI in the hospitality industry be designed to protect customer privacy and sensitive information while still providing personalized and efficient service?

RQ13.

What is the best method for measuring and addressing biases in AI algorithms that impact customer experiences?

RQ14.

How can AI technologies in the hospitality industry be safeguarded against potential misuse that leads to negative social, economic, or political consequences?

RQ15.

How can diverse stakeholder groups, including marginalized communities and affected individuals, be included in the design, deployment, and evaluation of AI systems in the hospitality sector to prevent concentration of power and bias?

RQ16.

What are the potential long-term social, economic, and psychological impacts of AI-driven services? How can we address these challenges?

Ethically deployed AI can prevent biases and ensure equity in customer interactions, thereby enhancing brand reputation and customer loyalty. By investigating the above questions, studies could identify ethical dilemmas specific to hospitality and tourism and propose guidelines for ethical AI use that enhances corporate social responsibility (CSR).

Fourthly, mechanical, thinking, and feeling AI are introduced and incorporated into many hospitality businesses in nowadays, which might affect job roles, employee morale, and the potential displacement of human workers. The classification in this paper provides a basis for examining the interplay between AI and human labor within the industry – collaborative AI. With AI making significant strides in replicating human capabilities, concerns have arisen regarding the potential displacement of human workers across the broader economic spectrum (Wilson & Daugherty, 2018; He, Teng, & Song, 2023). Engaging in research at the juncture of labor and AI within the hospitality industry offers a valuable avenue for gaining nuanced insights that not only encompass strategies for effectively harnessing AI's benefits within organizational frameworks but also emphasize the importance of nurturing employee engagement, empowerment, and adaptability to accommodate the ever-changing nature of work. Below are the research questions that might help scholars who want to explore this area in the future:

RQ17.

To what extent can AI fully replace human employees in the hospitality industry?

RQ18.

How can AI systems seamlessly integrate into hotel workflows to enhance human productivity and decision-making while maintaining a positive work environment?

RQ19.

What training and reskilling programs are most effective in preparing hospitality workers for roles that involve working alongside AI systems, and how can these programs be scaled and implemented?

RQ20.

How can AI integration impact employee job satisfaction, morale, and overall well-being in the hospitality industry, and how can these impacts be measured and addressed?

RQ21.

What is the broader impact of AI integration on the labor market within the hospitality industry, including potential shifts in job roles, new opportunities, and challenges related to job displacement?

Lastly, the classification in this paper identifies data scarcity as a limitation in current AI applications within hospitality. Synthetic data generation could address this issue by allowing researchers to create robust datasets where the limitation appears. Interest in AI data generation has greatly increased due to advances in image and text generation with the advent of models like Stable Diffusion and ChatGPT. These advances are already having an impact across many disciplines and have also raised interesting copyright concerns (e.g. AI art generation) due to model performance (Stability AI, 2023; OpenAI, 2023). Synthetic data generation enables the creation of larger and more diverse datasets for training AI models, and it is poised to meet the demand for open-ended interactive multimodal simulations that enable the analysis of complex systems dynamics (de Melo, Torralba, Guibas, DiCarlo, Chellappa, & Hodgins, 2021). Research in this topic can contribute to overcoming challenges related to data scarcity, bias, and privacy, ultimately leading to improved AI model performance and better customer experiences. The following research questions could be beneficial to future research:

RQ22.

How can AI models be trained effectively using a combination of real and synthetic data to improve generalization and enhance performance in service tasks?

RQ23.

Can synthetic data generation techniques help mitigate biases in real-world data, ensuring fair and unbiased outcomes in service interactions?

RQ24.

How can synthetic data be enriched to include realistic variations and outliers observed in actual service interactions, enhancing the robustness and adaptability of AI models?

RQ25.

How can human experts contribute to creating and refining synthetic data to ensure that the generated data accurately represents complex service scenarios?

RQ26.

What ethical considerations arise when using synthetic data in hospitality research, and how can these considerations be addressed to ensure the responsible and ethical use of synthesized datasets?

Overcoming data limitations through synthetic data can enable smaller entities to compete more effectively, democratizing the use of AI in hospitality and tourism. By investigating the above questions, the research could focus on creating robust synthetic datasets that mimic real customer interactions, thereby allowing companies to refine AI applications without compromising data privacy.

4. Conclusion

This study introduces a novel conceptual classification for AI hospitality research building upon the work of Huang and Rust (2021). The proposed classification encompasses diverse domains, including prediction and pattern recognition, computer vision, NLP, behavioral research, and synthetic data generation. This innovative research taxonomy furnishes scholars with valuable insights into the trajectory of AI hospitality research. Notably, the exploration of synthetic data generation and feeling AI represents pioneering endeavors within the industry, and this paper serves as a catalyst for advancing such research.

A research agenda with five trended topics is identified in line with the classification. Each topic not only resonates with the current industry challenges but also aligns with the theoretical and practical implications of the AI dimensions this paper has delineated. The research agenda covers topics such as the transparency of AI decisions, the balance between AI autonomy and control, the ethical deployment of AI, and the use of synthetic data generation to overcome data limitations.

In conclusion, this viewpoint provides a foundational framework and a directional compass for future research in AI within the hospitality industry. It pushes the industry to move forward with a balanced approach to leverage AI to augment human potential and enrich customer experiences. Both the classification and the research agenda would contribute to the body of knowledge that will guide the industry toward a future where technology and human service coalesce to create unparalleled value for all stakeholders.

Figures

Figure 1

The relationship among AI, ML, and DL

Figure 2

Research classification and service AI framework

Table 1

Comprehensive literature review about AI in hospitality research

Category	Article	Journal (Year)	Contribution	Limitation and future direction
Prediction and pattern recognition	Demand forecasting model using hotel clustering findings for hospitality industry	Information Processing and Management (2022)	(1) Innovative demand forecasting model: The study introduces a forecasting model that uniquely combines time series demand data with additional features derived from hotel clustering, such as the top 10 hotel features and hotel embeddings obtained through Neural Networks (NN). (2) Utilization of hotel clustering: By clustering hotels based on certain criteria and utilizing the findings, the model accounts for the nuanced preferences of travelers and the similarity in demand patterns among hotels that cater to similar customer segments	(1) Data and geographic specificity: The study primarily utilizes data from hotels in Turkey, which may limit the generalizability of the findings to other geographic locations or markets with different characteristics. Future research could explore the applicability of the proposed model across diverse hospitality settings. (2) Model complexity and interpretability: While the model incorporates advanced techniques like NN embeddings and Attention-LSTM, these methods can be complex and less interpretable than simpler models. Further research could simplify the model without significantly compromising accuracy, thus making it more accessible for practitioners. (3) Dynamic market conditions: The hospitality industry is influenced by various dynamic factors such as economic conditions, political events, and technological advancements. Future studies could enhance the model's adaptability to rapidly changing market conditions. (4) Integration with other data sources: The study could be extended by integrating additional data sources such as social media trends, customer reviews, and macroeconomic indicators to further improve forecasting accuracy
	Hierarchical pattern recognition for tourism demand forecasting	Tourism Management (2021)	(1) Innovative hierarchical forecasting approach: The study proposes a novel hierarchical pattern recognition (HPR) method that systematically incorporates the complexities of calendar patterns, including floating holidays and varying workday and weekend patterns. This approach enhances demand forecasting models' adaptability to intricacies related to real-world tourism flows. (2) Floating holidays integration: Unlike traditional models that may struggle with the irregularity of floating holidays, the HPR method effectively accounts for these variations. By identifying calendar patterns and adjusting forecasts accordingly, the model offers a more precise prediction mechanism that can handle the unique challenges posed by these holidays. (3) Empirical validation and superior forecasting performance: Utilizing daily visitor data from three attractions in China, the paper demonstrates the effectiveness of the HPR method. It outperforms several benchmarks forecasting models, indicating its potential utility for practitioners in the hospitality and tourism industry seeking accurate, short-term demand predictions	(1) Geographic and contextual specificity: The model's current validation is based on data from specific attractions in China. Future studies could explore its applicability and effectiveness in other geographic locations and contexts, including different types of tourism destinations or broader hospitality settings. (2) Model complexity and user accessibility: While the HPR method provides improved forecasting accuracy, its complexity might pose challenges for implementation by practitioners without advanced statistical or computational backgrounds. Simplifying the model or developing user-friendly tools could enhance its practical applicability. (3) Integration of additional variables: The current model focuses primarily on calendar patterns and historical data. Integrating additional variables such as economic indicators, weather conditions, and social media trends could further improve the accuracy and robustness of forecasts. (4) Long-term forecasting capability: The study focuses on short-term forecasting. Assessing the model's performance in long-term forecasts and its ability to adapt to changing tourism trends over longer horizons represents a valuable direction for future research
	Market segmentation and travel choice prediction in Spa hotels through TripAdvisor's online reviews	International Journal of Hospitality Management (2019)	(1) Hybrid machine learning methods: The study introduces a hybrid machine learning approach to analyze and interpret social data from online reviews effectively. This approach allows for a nuanced segmentation of spa hotel markets and predicts travel choices with a higher degree of accuracy. (2) Incremental recommendation agent: The proposed methods serve as an incremental recommendation agent by utilizing big data from online social media platforms like TripAdvisor. This system facilitates the segmentation of spa hotels and resorts, enabling targeted marketing strategies and personalized customer engagement	(1) Data and geographic specificity: The study focuses on data collected from TripAdvisor, which may only partially represent part of the spectrum of spa hotel customers. Future research could expand the dataset to include other sources of online reviews and cover a broader geographic area to test the model's generalizability. (2) Model complexity and accessibility: The complexity of hybrid machine learning algorithms might pose challenges for practitioners in the hospitality industry who lack a technical background. Simplifying the model or developing more user-friendly tools could increase accessibility and practical application. (3) Dynamic nature of online reviews: The sentiment and content of online reviews can rapidly change due to various factors, including service improvements or declines, new trends, and external events. Future studies should consider these dynamics and update segmentation models accordingly. (4) Integration with other data sources: Combining online reviews with other data sources, such as customer demographic information, booking patterns, and economic indicators, could enhance the model's predictive power and segmentation accuracy
	Predicting hospitality financial distress with ensemble models: the case of US hotels, restaurants, and amusement and recreation	Service Business (2018)	(1) Introduction of ensemble models: This study innovatively applies ensemble models combining Support Vector Machines (SVM), Neural Networks (NN), and Decision Trees (DT) for predicting financial distress. This method leverages the strengths and compensates for the weaknesses of individual models, offering a robust approach to financial distress prediction in the hospitality industry. (2) Segment-specific analysis: By focusing on different segments within the hospitality industry, the research acknowledges and addresses each segment's varying characteristics and risks. This tailored approach enhances the relevance and applicability of the predictive models for practitioners within these segments	(1) Data and geographic scope: The study is based on U.S. firms within specific Standard Industrial Classification (SIC) codes, which may limit its generalizability to other countries or hospitality segments not covered by the SIC codes used. (2) Model complexity: The ensemble models, while powerful, are complex and may require significant expertise to implement effectively. Future research could explore simpler models that still offer strong predictive accuracy. (3) Temporal dynamics: The study covers data up to 2010, and the hospitality industry has undergone significant changes since then, including the impact of digital platforms and the COVID-19 pandemic. Future studies should update the models with more recent data and consider these industry changes. (4) Broader predictor variables: While financial ratios and certain market-based variables were included, integrating additional non-financial indicators such as customer satisfaction scores, online review analytics, or macroeconomic variables could potentially improve the models' predictive power
	A new forecasting approach for the hospitality industry	International Journal of Contemporary Hospitality Management (2015)	(1) Innovative forecasting framework: This paper introduces a multiple-input-multiple-output approach based on ANN models designed to improve forecasting by considering the cross-correlations among different tourist markets. (2) New forecasting accuracy measure: Beyond traditional accuracy metrics, the study proposes a new measure based on the proportion of periods in which the model achieves lower absolute forecasting errors compared to a benchmark model. (3) Empirical validation using tourism data: The findings demonstrate that multivariate architectures, particularly radial basis function (RBF) networks, can significantly outperform single-market models in predictive performance	(1) Geographic specificity: The findings demonstrate that multivariate architectures, particularly radial basis function (RBF) networks, can significantly outperform single-market models in predictive performance. (2) Comparison with other forecasting models: While the study highlights the superiority of RBF networks over other ANN architectures, it leaves room for comparison with a broader range of forecasting models, including non-AI-based approaches. Further research could explore hybrid models or compare the proposed ANN models against state-of-the-art statistical and machine learning techniques. (3) Dynamic tourism market conditions: While the study highlights the superiority of RBF networks over other ANN architectures, it leaves room for comparison with a broader range of forecasting models, including non-AI-based approaches. Further research could explore hybrid models or compare the proposed ANN models against state-of-the-art statistical and machine learning techniques. (4) Data granularity and quality: The accuracy and granularity of the input data are critical to the performance of any forecasting model. The paper utilizes monthly tourist arrival data, which might not capture short-term fluctuations or the impact of specific events. Investigating the effects of using more granular or diverse data sources could be a fruitful area for future research
Computer vision	Progress on image analytics: Implications for tourism and hospitality research	Tourism Management (2024)	(1) Methodological framework for image analysis: The paper introduces a structured approach to conducting image analytics in tourism and hospitality studies. This framework includes steps from data collection and transformation to applying computer vision techniques and advanced statistical analysis, addressing methodological challenges and opportunities. (2) Theoretical insights and research directions: The paper introduces a structured approach to conducting image analytics in tourism and hospitality studies. This framework includes steps from data collection and transformation to applying computer vision techniques and advanced statistical analysis, addressing methodological challenges and opportunities. (3) Ethical considerations in image analysis: The study discusses the ethical implications of using images as data sources, emphasizing the importance of anonymity, consent, and mitigating biases, especially when employing automated tools and algorithms for image analysis	(1) Scope of literature review: The paper focuses on journal publications within tourism and hospitality, predominantly in English, suggesting a broader examination of work in other languages and disciplines could provide additional insights. (2) Emphasis on academic publications: The review may not capture the full extent of discussions and methodological innovations presented in grey literature or industry reports by concentrating on journal articles. (3) Advancement of method: While offering a comprehensive framework for image analytics, there's a continuous need to develop and validate new analytical tools and algorithms that can handle the complexities and nuances of visual data in tourism research. (4) Deepening theoretical engagement: The paper calls for a more profound theoretical engagement with the visual content, urging future studies to delve into how images influence consumer behavior and decision-making processes more nuancedly
	Research on user-generated photos in tourism and hospitality: A systematic review and way forward	Tourism Management (2023)	(1) Advancements in method: The paper highlights the evolution from manual qualitative content analysis to sophisticated AI-based quantitative methods, enabling the processing of large visual data sets. This methodological progression enhances the ability to derive meaningful insights from user-generated photos (UGP). (2) Insightful themes identification: The review offers a structured understanding of how UGPs influence and reflect tourism and hospitality dynamics by categorizing the existing literature into four major themes. These themes cover a wide range of research interests, from why people share photos to how these photos impact tourism destinations and businesses. (3) Theoretical development: Despite the dominance of data-driven studies, the review notes a gradual increase in theoretical engagement within this research area, especially in studies focusing on destination image perception and the impact of UGPs. The paper suggests that future research should aim for a balance between empirical data analysis and theoretical contributions	(1) Geographical and cultural generalizability: Most studies concentrate on specific geographic locations or platforms, which may limit the applicability of findings across different cultural and tourism contexts. Future research could address this by incorporating diverse geographical locations and cultural perspectives. (2) Integration with other data sources: The paper observes that while UGPs provide valuable insights, integrating these photos with other types of data (e.g. textual reviews and economic indicators) could offer a more comprehensive understanding of tourism and hospitality phenomena. (3) Dynamic nature of social media: The review underscores the rapid evolution of social media platforms and how user behaviors change over time. It calls for continuous monitoring and analysis to keep pace with these trends
	Applying image recognition techniques to visual information mining in hospitality and tourism	International Journal of Contemporary Hospitality Management (2023)	(1) Innovative image recognition application: It demonstrates how advanced image recognition techniques, especially CNNs, can efficiently process and analyze large volumes of visual content from platforms like TripAdvisor and Airbnb. This allows for a more nuanced understanding of customer preferences and behaviors based on the visual information they share online. (2) Enhanced market insights: By extracting detailed features from images, such as object recognition and facial analysis, businesses can gain deeper insights into the visual appeal of their services and how consumers perceive them. This could lead to more targeted marketing strategies and improvements in service delivery based on consumer feedback reflected through images. (3) Reduction of information asymmetry: The paper addresses how visual information can mitigate information asymmetry between consumers and service providers. Visual content provides a tangible way for customers to evaluate the quality of hospitality services before experiencing them, potentially increasing trust and reducing uncertainty in booking decisions	(1) Technical complexity and accessibility: Applying CNNs and deep learning models requires significant technical expertise and computational resources, which may not be readily available to all hospitality businesses, especially smaller operators. (2) Generalizability of findings: While the study showcases the potential of image recognition techniques, further research is needed to understand how these technologies can be adapted and applied across different hospitality industry segments and in diverse geographic locations. (3) Evolution of visual content: The dynamic nature of online content and consumer preferences necessitates continuous updating and refinement of image recognition models to stay relevant and effective. (4) Ethical considerations: The use of facial recognition and analysis of personal attributes raises privacy and ethical concerns that need to be addressed, including consent and data protection issues
	Color and engagement in touristic Instagram pictures: A machine learning approach	Annals of Tourism Research (2021)	(1) Integration of color psychology and data science: The study provides a novel integration of color psychology with machine learning and data science methodologies, offering a fresh perspective on analyzing visual content in tourism marketing on social media platforms like Instagram. (2) Insights on color influence: It reveals that certain colors enhance user engagement with touristic Instagram photos. For instance, photos featuring natural scenery, high-end gastronomy, and sacral architecture that include the color blue tend to receive more engagement. Red/orange hues boost engagement for pictures related to local delicacies and ambiance, while a combination of violet and warm colors benefits cityscapes and interior design photos. (3) Practical guidelines for marketers: The findings offer actionable guidelines for hospitality and tourism marketers on leveraging color in their visual content strategy to enhance engagement on social media platforms	(1) Geographical and cultural generalizability: The study’s findings are based on a global dataset from Instagram, which may not account for cultural or geographical differences in color perception and social media usage patterns. Future research could explore these aspects more deeply. (2) Focus on color attributes: While the study provides valuable insights into the role of color, it acknowledges that other factors (e.g. image composition, hashtags, and captions) also influence user engagement. Subsequent studies might examine these elements in conjunction with color. (3) Dynamic nature of social media trends: The rapidly changing trends on social media platforms and user preferences highlight the need for ongoing research to keep the findings relevant and actionable for marketers
	Effects of user-provided photos on hotel review helpfulness: An analytical approach with deep leaning	International Journal of Hospitality Management (2018)	(1) Introduction of deep learning techniques: The study introduces deep learning models for analyzing large quantities of visual content in online reviews, highlighting their superiority in predicting review helpfulness over traditional machine learning methods. (2) Insight into user-provided photos' impact: It is demonstrated that user-provided photos, when combined with review texts, significantly enhance the prediction accuracy of a review's helpfulness. This suggests that visual content is crucial in online consumers' perception and decision-making processes	(1) “Black Box” nature of deep learning models: One limitation highlighted is the opaque nature of deep learning models, which makes it difficult to understand how specific features of user-provided photos contribute to the helpfulness of reviews. Future research could focus on developing more interpretable models. (2) Focus on a single tourist destination: The study's dataset is primarily from Orlando, FL, which may limit the generalizability of the findings. Future studies should include a more diverse set of locations to validate the model's effectiveness across different contexts. (3) Interaction between text and images: The study does not explicitly examine how the interaction between text and images influences review helpfulness. Investigating this aspect could offer deeper insights into how consumers process and evaluate online reviews. (4) Technological and methodological advancements: The paper calls for future research to employ advanced machine learning techniques, such as attention mechanisms or influence functions, to unlock the potential of deep learning in hospitality and tourism research further
Natural language processing	Text classification in tourism and hospitality – a deep learning perspective	International Journal of Contemporary Hospitality Management (2023)	(1) Comprehensive review: This study offers a systematic review of deep learning applications in text classification within tourism and hospitality, marking it as one of the first of its kind to do so comprehensively. (2) Focus on deep learning methods: It highlights using various deep learning models for classifying text features, sentiment, and ratings in tourism and hospitality research. Notably, it highlights the predominant use of older deep learning methods like feed-forward neural networks. It suggests that newer models proposed in computer science with better performance have yet to be widely adopted in tourism and hospitality. (3) Practical implications: The findings provide practical implications for managers leveraging deep learning algorithms to analyze user-generated content efficiently. The study emphasizes the potential of these algorithms in extracting valuable insights from customer reviews, which can aid in decision-making and service improvement	(1) Adoption of newer deep learning methods: The study acknowledges a gap in adopting the latest deep learning methods within the field. It suggests that newer models with demonstrated success in computer science, like BERT and OpenGPT, should be explored for their applicability in tourism and hospitality. (2) Manual data annotation and use: This paper calls for more effort in manual data annotation for specific tourism contexts to enhance the quality and applicability of text classification research. This could improve the accuracy of models in capturing the nuances of tourist experiences and sentiments. (3) Expanding research scope: Future research should expand beyond publicly available datasets like Yelp reviews and explore a wider range of data sources and languages. This would help understand the global and cultural nuances in tourist behavior and satisfaction
	Differences in Chinese and Western tourists faced with Japanese hospitality: a natural language processing approach	Information Technology and Tourism (2021)	(1) Cultural insights into tourist satisfaction: The study provides significant insights into how cultural backgrounds influence tourists' satisfaction, especially focusing on renowned Japanese hospitality. It demonstrates that while Chinese tourists prioritize room quality, Western tourists are more inclined toward staff behavior and overall service quality. (2) Innovated method: By employing machine learning and natural language processing techniques to analyze a vast amount of hotel review data, the study introduces a novel, scalable approach for assessing and comparing satisfaction factors across different tourist demographics	(1) Cultural and language specificity: The study primarily focuses on Chinese and Western tourists, leaving room for further research on tourists from other cultural backgrounds and languages to provide a more comprehensive understanding of global tourist satisfaction in the Japanese hospitality context. (2) Dynamic nature of tourist expectations: As the tourism industry evolves, so do tourists' expectations. Future studies could explore changes in satisfaction factors over time, considering emerging trends and the impact of global events on tourist behavior. (3) Broader application of method: While the study showcases the effectiveness of machine learning and natural language processing for analyzing tourist satisfaction, further research could expand this methodology to other aspects of the tourism industry or apply it in different hospitality contexts globally
	Understanding homophily of service failure within the hotel guest cycle: Applying NLP-aspect-based sentiment analysis to the hospitality industry	International Journal of Hospitality Management (2020)	(1) Cultural insights on service failure perceptions: The research provides valuable insights into how service failure perceptions vary between Asian and non-Asian guests, emphasizing the influence of cultural backgrounds on guests' expectations and experiences. It reveals specific service aspects where these differences are most pronounced, such as engineering problems and housekeeping services. (2) Method advancement: Utilizing NLP for aspect-based sentiment analysis represents a methodological advancement, allowing for a more detailed and scalable analysis of textual data from online reviews. This approach enhances the understanding of guest feedback by systematically categorizing service failures. (3) Practical implications for service improvement: By identifying the stages of the guest cycle where service failures are most frequently reported and understanding the cultural nuances in guest feedback, hoteliers can tailor their service improvement strategies more effectively. This includes focusing on specific operational areas and adopting a more culturally sensitive approach to service design and delivery	(1) Cultural and geographic specificity: The study's focus on Asian and non-Asian tourists primarily reviewing UK hotels may limit the generalizability of the findings to other cultures and geographic regions. Future research could broaden the scope to include more diverse cultural backgrounds and hotel contexts. (2) Dynamic nature of guest expectations: The study captures a snapshot of guest expectations and perceptions that could evolve over time. Ongoing research is needed to track these changes, especially in response to emerging trends in hospitality and shifts in cultural attitudes towards travel and accommodation. (3) Integration with other data sources: While the study effectively utilizes textual analysis of online reviews, integrating these findings with other data sources, such as guest surveys or operational performance metrics, could provide a more comprehensive understanding of service failures and guest satisfaction
	Automatic analysis of textual hotel reviews	Information Technology and Tourism (2016)	(1) Innovative NLP platform: Introduction of the OpeNER platform, providing a comprehensive suite of NLP tools for automatically analyzing hotel reviews and facilitating the extraction of valuable information from customer-generated content. (2) Enhanced text analysis capabilities: The platform offers functionalities for language identification, tokenization, part-of-speech tagging, named entity recognition and classification, sentiment analysis, and opinion mining, among others, supporting six languages. (3) Demonstrated applicability of the platform in analyzing hotel reviews to extract insights regarding customer perceptions and experiences, aiding in understanding and managing online reputation. (4) Open source and accessibility: The Open Source nature of the platform and its modular design encourage adaptation and extension by other researchers or practitioners, potentially expanding its use beyond the hospitality domain	(1) Domain-specific challenges: The Open Source nature of the platform and its modular design encourage adaptation and extension by other researchers or practitioners, potentially expanding its use beyond the hospitality domain. (2) Language coverage: Currently supporting six languages, extending the platform to include additional languages would broaden its applicability and utility across more geographic regions. (3) Integration with additional data sources: Future enhancements could include integrating the platform with other data sources, such as social media platforms, to enrich the analysis and insights generated. (4) Advancements in NLP techniques: Ongoing developments in NLP and machine learning could further enhance the platform's analytical capabilities, necessitating continuous updates and improvements
Behavioral research	Customers’ acceptance of artificially intelligent service robots: The influence of trust and culture	International Journal of Information Management (2023)	(1) Integration of trust and cultural dimensions: The study introduces trust as a pivotal factor in AI robot acceptance, arguing that trust directly influences the intention to use AI robots. It further explores the moderating effects of national culture (U.S. vs China) and individual cultural values on this relationship. (2) Empirical evidence on cultural influence: By comparing responses from the U.S. and China, the research provides empirical evidence on how national and individual cultural dimensions, such as uncertainty avoidance and individualism/collectivism, affect consumer acceptance of AI service robots. (3) Practical guidelines for cross-cultural deployment of AI robots: The findings offer actionable insights for deploying AI service robots in the hospitality industry across different cultures, emphasizing the need to consider cultural sensitivity in service design and delivery	(1) Dynamic nature of technology and culture: The study acknowledges the rapidly evolving landscape of AI technology and cultural dynamics. Continued research is necessary to keep pace with technological advancements and shifting cultural attitudes toward AI. (2) Broader spectrum of cultural dimensions: The research focuses on certain cultural dimensions but does not encompass all aspects of culture that might influence AI acceptance. Future studies could explore additional cultural factors and their impact on consumer behavior toward AI robots
	Artificial intelligence service recovery: The role of empathic response in hospitality customers’ continuous usage intention	Computers in Human Behavior (2022)	(1) Emphasize emotional intelligence in AI: The study highlights the significance of incorporating emotional intelligence, specifically empathy, into AI services for effective service recovery. This approach shifts the focus from improving AI's “intelligence quotient” to enhancing its emotional quotient. (2) Psychological mechanism exploration: It uncovers the psychological mechanisms—namely, the sequential mediation of psychological distance and trust—through which empathic responses by AI facilitate service recovery. (3) Impact of interaction modality: The research also examines how the interaction modality (e.g. text-only vs text and voice) moderates the effect of AI's empathic response on service recovery outcomes. The findings suggest that multisensory interactions enhance the recovery effect of empathic responses, highlighting the importance of how AI communicates empathy	(1) Cultural and contextual specificity: The study's context, primarily based on hospitality settings in specific geographical locations, may limit the generalizability of findings. Future research could explore the impact of AI's empathic responses across different cultures and service settings. (2) Dynamic AI capabilities: As AI technology evolves, its capability to exhibit genuine empathy and understand complex human emotions will improve. Investigating how advancements in AI affect service recovery strategies will be crucial. (3) Broader application of empathy: Exploring other forms of emotional intelligence beyond empathy, such as compassion or humor, and their role in AI service recovery could provide a more comprehensive understanding of AI's potential in hospitality
	Developing a formative scale to measure consumers’ trust toward interaction with artificially intelligent (AI) social robots in service delivery	Computers in Human Behavior (2021)	(1) Innovative trust scale: The study introduces the Social Service Robot Interaction Trust (SSRIT) scale, a novel tool for measuring trust in AI social robots. It highlights factors such as familiarity, robot use self-efficacy, social influence, technology attachment, and trust stance in technology as key components influencing consumers' propensity to trust these robots. (2) Multidimensional approach: It proposes a third-order reflective-formative scale to capture the complexity of trust in human-AI robot interactions, integrating dimensions related to the robot's function and design, service task and context, and individual propensity to trust	(1) AI and consumer expectations: As AI technologies and consumer expectations evolve, ongoing updates and revisions of the SSRIT scale may be required to maintain its relevance and accuracy. (2) Broadening the scope of trust indicators: Future research could explore additional trust indicators in AI social robots, considering rapid technological advancements and changing consumer behaviors
	Adoption of AI-based chatbots for hospitality and tourism	International Journal of Contemporary Hospitality Management (2020)	(1) Integration of context-specific variables: The study enriches the Technology Adoption Model (TAM) by integrating additional variables like anthropomorphism and perceived intelligence, offering insights into how these factors influence the adoption intention of chatbots. (2) Insight into behavioral intention and actual usage: It provides a comprehensive understanding of the factors driving customers' intentions to use chatbots for tourism and their engagement with the technology	(1) Geographic and cultural specificity: The study is focused on India, suggesting that findings need to be tested in other cultural and geographic contexts to ensure broader applicability. (2) Emerging technology dynamics: As AI and chatbot technologies evolve rapidly, ongoing research is needed to keep pace with technological advancements and their impact on consumer behavior in hospitality and tourism
	Consumers acceptance of artificially intelligent (AI) device use in service delivery	International Journal of Information Management (2019)	(1) Theoretical model development: The paper introduces a comprehensive model that incorporates various factors influencing consumer acceptance of AI in service delivery, highlighting customers' three-stage process in determining their acceptance. (2) Understanding of consumer behavior: It provides insights into how various factors, such as social influence, hedonic motivation, and anthropomorphism, impact the perception of AI devices' usefulness and ease of use, ultimately affecting their willingness to engage with AI in service settings	(1) Dynamic nature of AI technology: As AI technology evolves, consumer perceptions and acceptance criteria may also change. Future studies should consider longitudinal approaches to capture these dynamics. (2) Expansion of antecedents and variables: The model could be further refined by including additional factors influencing consumer acceptance of AI in service delivery, such as privacy concerns or trust in technology. (3) Integration with other technologies: Considering the interplay between AI and other emerging technologies (e.g. virtual reality, blockchain) in service delivery could provide a more holistic view of future service landscapes
Synthetic data generation	Generative Artificial Intelligence in the Hospitality and Tourism Industry: Developing a Framework for Future Research	Journal of Hospitality and Tourism Research (2023)	(1) Innovative framework for Generative Artificial Intelligence (GAI) integration: The study introduces a comprehensive framework to understand the impact of GAI applications on a wide range of stakeholders within the hospitality and tourism industry. This framework aims to advance academic research by integrating practical insights with academic perspectives. (2) Stakeholder theory application: By employing the stakeholder theory, the research provides a nuanced analysis of how GAI applications can benefit or challenge different groups within the hospitality and tourism industry, from customers and employees to suppliers and society at large. (3) Ethical and legal considerations: It emphasizes the need for hospitality and tourism firms to carefully consider the ethical and legal implications of deploying GAI technologies, including privacy concerns, data protection, and the potential for biases in AI-generated content	(1) Rapid technological evolution: Given the fast-paced development of GAI technologies, ongoing research is necessary to keep up with new advancements, applications, and implications for the hospitality and tourism industry. (2) Integration with existing system: Future studies could explore how GAI technologies can seamlessly integrate with existing operational and management systems within HT firms to maximize efficiency and effectiveness without disrupting established workflows
	Autonomous travel decision-making: An early glimpse into ChatGPT and generative AI	Journal of Hospitality and Tourism Management (2023)	(1) Efficiency in trip planning: ChatGPT streamlines the trip planning process, offering tourists comprehensive and cost-effective travel solutions by leveraging vast amounts of online data. (2) Customized recommendations: It provides personalized travel recommendations based on user interactions, improving the travel experience and satisfaction. (3) 24/7 personal assistant: ChatGPT serves as a round-the-clock personal assistant, capable of offering real-time guidance and assistance in multiple languages. (4) Enhanced sharing experience: The technology facilitates enriched sharing of travel experiences, enabling tourists to create and share high-quality, creative content easily	(1) Rapid evolution of AI technology: As AI technologies like ChatGPT continue to evolve, ongoing research will be needed to keep pace with their capabilities and implications for the tourism industry. (2) Ethical considerations and privacy concerns: The use of ChatGPT raises important questions about data privacy, security, and the potential for biases in AI-generated content. Addressing these concerns will be crucial for its successful integration into tourism services

Source(s): Table by authors

Names are blinded for peer review.