Next Article in Journal
Cuteness vs. Usefulness: A Dual Perspective on Service Robot Acceptance in the Travel Industry
Previous Article in Journal
Effects of Perceived Price Dispersion on Travel Agency Platforms: Mental Stimulation to Consumer Cognition
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JTAER
Volume 20
Issue 1
10.3390/jtaer20010046
jtaer-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Echoes of Innovation: Exploring the Use of Voice Assistants to Boost Hotel Reputation

by
Fang Yang
1,2,
Tianyu Ying
1,3,4,* and
Xuling Liu
2
1
School of Management, Zhejiang University, Hangzhou 310058, China
2
College of Tourism, Xinjiang University of Finance and Economics, Urumqi 830000, China
3
Center for Balance Architecture, Zhejiang University, Hangzhou 310058, China
4
Zhejiang-Hong Kong Joint Laboratory for Human-Computer Collaboration and Organizational Transformation, Hangzhou 310058, China
*
Author to whom correspondence should be addressed.
J. Theor. Appl. Electron. Commer. Res. 2025, 20(1), 46; https://doi.org/10.3390/jtaer20010046
Submission received: 26 January 2025 / Revised: 3 March 2025 / Accepted: 5 March 2025 / Published: 10 March 2025
Browse Figure
Versions Notes

Abstract

:
Internet platforms and self-media have become vital online communities for promoting positive reputations for hotels. Previous studies have primarily focused on enhancing positive electronic word-of-mouth (eWOM) through improvements in hotel infrastructure and staff services. As hotels deepen their digital transformation, the application of various artificial intelligence technologies in hotel service encounters significantly impacts the service experience. This study explores the effects of voice assistant (VA) attributes on the online reputation of hotels. Specifically, it examines how the attributes of VAs (anytime connectivity, information association, and interactivity) influence positive customer evaluations in hotels. Utilizing a questionnaire survey method, we collected 529 valid questionnaires offline and employed structural equation modeling along with the PROCESS plugin in SPSS to conduct path analysis, as well as mediation and moderation effect analyses. The results indicate that perceived value and the existence of human–AI rapport mediate the impact of VA attributes on positive eWOM, although the direct effect of some attributes (information association) was not supported. Furthermore, anytime connectivity enhances the influence on human–AI rapport through social presence, while privacy concerns negatively affect the relationship between perceived value and intentions to engage in eWOM. These insights are critical for hotels seeking to maximize the benefits of digital transformation.

1. Introduction

Artificial intelligence (AI) technology is a pivotal domain within the digital economy and its applications are progressively catalyzing substantial productivity enhancements and advancements in industry [1]. In particular, elevating the operational efficacy of service enterprises through AI technology is a central concern of contemporary digital practice [2,3]. By seamlessly weaving AI innovations, including robotic automation, voice assistants (VAs), and chatbots, into service interactions, it is envisioned that a fundamental reshaping the dynamics of engagement between clientele and service entities will be achieved [4]. This metamorphosis redefines the roles of the involved parties in both the production and consumption realms.
More than other AI technologies, VAs have garnered favor in the service industry, showing immense market potential for the future. With substantial advances in VAs, the global VA market is projected to grow at an annual rate of 31.8% from 2023 to 2027 and reach a value of $13.66 billion [5]. Seeking not to be left behind, in recent years, the hospitality industry has accelerated its digital transformation through the adoption of intelligent automation, with VA service devices being installed in many guest rooms as AI concierges [6]. Makers of well-known VA devices, such as Amazon Alexa and Google Home, have also partnered with large hotel groups to modify VAs for the hotel space [7]. As an important component of artificial intelligence, the VAs offering service interactions can disrupt conventional service paradigms and exert a profound influence on customer service encounters, perception of encounters, and subsequent behaviors, including eWOM endorsements [3]. Research on hotel customer evaluations have shown customers to offer positive mention of intelligent voice services [8].
Voice interaction is widely acknowledged to be the most innate mode of communication for humans. With the advancement of enabling technologies such as deep learning, speech recognition, and natural language processing technologies, VAs have significantly enhanced their ability to convey information and offer social cues [4], rendering them closely akin to human-to-human exchanges [9]. Maximizing the interactive benefits of VA technology and leveraging eWOM from customers to businesses through VA services is, thus, a crucial concern in the ongoing digital metamorphosis of service enterprises.
Understanding the impact of VAs on eWOM in service industries is critical because of the significant influence of eWOM on users’ perception of and inclination to adopt a product, service, or technology such as VAs [10,11]. WOM plays a pivotal role in the success of products and services through the network effect [12]. Moreover, individuals’ WOM behaviors are influenced by their experiences and the gratification they derive from utilizing target products or services [13]. And in the digital era, the impact of eWOM holds especially great significance [14]. Presently, research on eWOM within service enterprises focuses predominantly on four categories: perceptual aspects, consumption related factors, individual elements, and social factors [15]. In addition, studies exploring the impact of AI technology on eWOM in service interactions have found factors such as chatbot interactivity [16], self–AI connection [17], chatbot service quality [18], and chatbot anthropomorphism [13] to be more pertinent. However, the literature on the specific interaction between VAs and eWOM is scarce, and of the available research, hardly any center on service enterprises in the hospitality industry [19,20].
VAs deployed in hotel rooms stand out from other commercial applications of VAs (e.g., shopping assistant, telephone-based intelligent customer service assistants) due to the complexity of their functional configuration. Hotel service VAs fulfill dual service roles: as utilitarian service providers and as hedonic services companions [21]. Besides offering traditional utilitarian functions, such as information delivery, they also deliver hedonic services to enhance the customer experience. Building affective hedonic services can range from casual chats to complex storytelling [21]. Although previous studies have investigated the cognitive effect of human–computer interaction on word-of-mouth [6], it is not clear whether and how human–computer interaction would simultaneously elicit both cognitive and affective effects in generating WOM.
Previous research has confirmed that the integration of AI in service encounters offers certain advantages [6]. However, the issue of privacy security in personalized human–machine interactions cannot be overlooked [17]. Scholars have focused on how privacy concerns in human–machine interactions affects customers’ continued use of technology, customers’ uneasiness, and brand loyalty [2,7]. Nevertheless, the role of privacy concerns during the AI service on eWOM remains unclear [19,20].
Therefore, in this study, we ask the following questions: Against the backdrop of service enterprises’ digitalization, can VA technology elicit eWOM for hospitality enterprises? If so, what are the origins of this influence, and how do these factors impact eWOM? To address the questions raised, we utilize service dominant logic and affordance theory to scrutinize the impact and mechanisms of VA affordances on eWOM in the context of hospitality enterprises. Some prominent theoretical contributions of this paper are as follows: (1) it applies affordance theory and service-dominant logic to voice assistant services, expanding our understanding of how these affordances influence customers’ eWOM intentions; (2) advances knowledge of eWOM by examining both cognitive and emotional cues influence eWOM intentions after AI interactions; (3) deepens insights into the role of AI attributes, particularly social presence and affordances, in shaping human–AI rapport and eWOM outcomes; and (4) explores the negative impact of privacy security issues on eWOM intentions. Among its various practical implications, this paper offers particularly useful insights for designers and hotel managers of VAs for the hospitality industry.

2. Literature Review and Theoretical Background

2.1. Literature Review

According to Arndt [22], word-of-mouth (WOM) is any informal, personal communication—oral or written—about a brand, product, service, or organization in which the recipient perceives the sender to have a noncommercial intent. Taking a slightly different approach, Barreto and Margarida [23] emphasized WOM as a communication process (oral or written) between a sender and one or more recipients—who may or may not be part of the same social network—for the purpose of information exchange and acquisition. Significantly, the advent of the internet has facilitated consumers to share WOM about products, services, and companies on online platforms, which is termed electronic word-of-mouth (eWOM) [3]. WOM and eWOM are regarded as instrumental marketing communication tools, as they influence consumer purchase decisions and enable companies to better comprehend customers’ needs and desires [14]. In the digital era, eWOM plays an especially vital role in influencing the sustainable development of service-based enterprises [3].
Contemporary explorations of eWOM in service businesses largely investigate four broad categories: perceptual dimensions, consumption-oriented variables, individual constituents, and sociocultural aspects [15]. In service encounters, the perception of service capacity and competency of the personnel are significant determinants of customers’ eWOM [16]. With the infusion of artificial intelligence in service encounters, technology has emerged as a unique type of “employee”, with its service performance becoming a key aspect of customer evaluation [21]. Previous scholarly attention on perceptual exploration of AI’s impact on eWOM in service contexts have primarily focused on the positive eWOM effects induced by chatbots, involving aspects such as chatbot interactivity [16], self–AI connection [17], chatbot service quality [18], and chatbot anthropomorphism [19].
However, advancements in VA technology have seen a surge in voice-enabled AI technology in service contexts, such as smart customer telephone service, shopping assistant, and hotel intelligent butler. Researchers in probing the advantages and shortcomings of VA services in service contexts tend to focus on continued usage intention [24], user satisfaction [25], brand loyalty [26], and privacy concerns [27]. Though the WOM and eWOM effects triggered by VAs have also drawn scholarly attention [10,19], investigations mostly emphasize solely the positive evaluations, acceptance, and recommendations of the technology itself. However, in the hotel service context, VA devices perform as surrogate butlers, offering personalized and emotionally interactive services to guests. Thus, while it has been proven that certain chatbot attributes can enhance customers’ positive eWOM intention toward enterprises, it remains largely unclear whether customer interactions with intelligent voice devices in hotel service scenarios could similarly enhance eWOM attitudes beyond the functionality of the technology itself, and what the underlying modalities would be. Prior studies have established that incorporating AI in service encounters yields distinct benefits [6]. Nonetheless, the significance of privacy security in personalized human–machine interactions warrants attention [17]. Research has explored how privacy concerns affect customers’ sustained engagement with technology, induce unease, and influence brand loyalty [2,7]. However, the impact of these privacy concerns on technology affordances and their subsequent effects on eWOM is still not well understood [19,20].

2.2. Theoretical Background

2.2.1. Affordance Theory

Affordances represent the potential actions suggested by an artifact’s qualities or features [28]. Affordance perception is a cognitive process wherein individuals recognize possible actions and often occurs during their interactions with target objects or technologies. The nature of affordances can vary based on the specific study context. For instance, Treem and Leonardi [29] outlined visibility, editability, persistence, and association as social media affordances, while Rice et al. [30] proposed six organizational media affordances: pervasiveness, editability, self-presentation, searchability, visibility, and awareness. While Chan et al. [31] described four social network site affordances (accessibility, information retrieval, editability, and association), Moreno and D’Angelo [28] suggested identity, social, cognitive, emotional, and functional affordances as relevant to social media.
In regard to the AI-driven customer service context, Li and Chang [32] pinpointed anytime connectivity, association, interaction, and personalization as affordances of AI technology by drawing from in-depth interviews. More relevantly, Li et al. [33] conducted a quantitative evaluation of AI affordances, although they omitted the aspect of personalization, which can be construed as either a corporate strategy or as a customer’s view of tailored service. In particular, Tiihonen and Felfernig [34] defined personalization as furnishing personalized information that caters to individual consumer needs, while Lee and Park [35] characterized it as the extent to which a customer perceives content as customized based on personal data. Hence, personalization may be regarded as a value perceived by consumers rather than a technological affordance.
Thus, we adopt Li et al.’s [33] framework of AI affordances, where anytime connectivity denotes the creation a connection between users and technological tools that enables users to avail themselves of services round the clock [36]; information association refers to the forging connections between individuals and informational content [37]; and interactivity represents the subjective perception of seamless interaction between customers and AI [33].
Within the hotel room context, VA technology is often integrated with local Internet of Things to control lighting and air conditioning temperatures through instant voice-based human–computer interaction, in addition to providing various types of information for customers on request. In the interaction process, VA technology’s ability to promptly respond to customer needs, and customers’ strength of control over VAs [21] are measures of the affordance of VAs.

2.2.2. Service-Dominant Logic

Service-dominant logic (SDL), as expounded by Vargo and Lusch [38], proposes that customers act as operant resources or collaborative partners, interacting with product or service offerings and other resources to jointly create value with the offering firms. In opposition to goods-dominant logic, which stresses delivering products already enriched with value to customers, SDL accentuates the collaborative value co-creation process between customers and firms [39]. SDL conceives of service functions as an input offered by firms, which is then used to actualize envisioned value through a collaborative creation process with customers [40]. In an SDL engagement between enterprises and consumers, customers partake in value co-creation in two ways: firstly, they interact with products or services to fulfill their individual needs; and secondly, as they engage in the service procedures of the enterprises and interact with the company’s assets, they impact the actualization of the firm’s value [38].
From an SDL view, the integration of AI into service interactions is seen as incorporating technology agents into the service ecosystem, which allow customers to combine resources by engaging with technology to achieve jointly created value. In hotels services, VAs can be seen to serve a twofold function: they function as utilitarian service renderers and as agents offering pleasurable hedonic services [21]. Furthermore, in line with the principles of SDL, engagements with VAs have the potential to enable customers to co-create value through behaviors such as heightened customer loyalty and WOM endorsements. Several scholars have adopted the SDL perspective to investigate the sources of brand loyalty [36]. Other researchers have also explored customer interactions with chatbots using the SDL as a theoretical foundation to understand the relationships with eWOM [33]. Empirical evidence from these studies suggests that SDL is an appropriate theoretical framework for this paper.

2.2.3. Integration of Service-Dominant Logic and Affordance Theory

Hotels often introduce VAs to replace human employees in engaging with human customers to carry out various tasks; VAs aid guests with room-specific activities like adjusting temperatures and seeking information [21]. Fundamentally, the VA acts as a service agent that connects with customers. And as in other contexts involving human-to-AI engagement, the conversational dynamics shape both the interaction and customers’ assessments of interaction quality [41]. In other words, VA features that empower customers to execute specific actions greatly influence both the interaction and customers’ assessments of the interaction. Relevantly, affordances delineate the structural features of objects that possess particular functionalities through the medium to elicit distinct behaviors [42]. Hence, the lens of affordance theory is employed in this research to investigate potential actions that hotel customers might undertake from the SDL viewpoint to acquire information or resolve their issues.
The integration of SDL with affordance theory is motivated by two primary factors. Firstly, there has been a clear progression towards enhancing technological capacities to cater to customers using AI-driven services. Notably, AI-powered VAs can functionally address customers’ needs in ways that expand on the functional possibilities of human service staff (i.e., anytime connection). Secondly, by fostering trust-based relationships with customers, VAs can serve as a strategic instrument enabling service enterprises in the hospitality industry to deliver greater experiential value to their clientele. Especially as the emphasis in service provision has transitioned from merely accomplishing the exchange of goods and services to the exchange process itself, and from tangible to intangible elements.
In both scenarios, the affordances provided by VAs present opportunities for customers to undertake specific actions for task fulfillment, with value being co-created through customers’ active involvement in the service provision process. Moreover, this involvement may easily encompass not only the service interaction with affordances but also extend to post-service evaluation, including the propagation of eWOM. Consequently, to investigate the mechanisms of this process, this study adopts a combination of affordance theory and SDL to investigate the influence of VA affordances on customers’ eWOM intention in regard to businesses in the hotel industry.

3. Hypotheses Development

3.1. VA Affordances and Perceived Value

According to the theory of Service-dominant Logic, value is realized in the process of use or consumption, rather than being embedded in the product itself. In other words, the buyer creates or perceives value through their interaction with the product or service [35,38]. Perceived value is defined as consumers’ overall assessment of the usefulness of a product or service, based on the trade-off between the benefits received and the costs incurred to obtain those benefits [43]. The attributes of a product play a pivotal role in shaping customers’ perception of value [33]. In the joint realm of technological products and service components, customer engagements with AI can yield both utilitarian and experiential value [44].
To be specific, information association is a key characteristic of voice assistants’ affordance [33]. In room services, the voice assistant acts as a smart housekeeper, providing customers with relevant information and services upon request. For instance, it can provide hotel-related information like room WiFi passwords or even remind the front desk of deliveries. The second key characteristic of voice assistant affordance is interactivity [33]. Swift responsiveness, punctuality, and ease facilitated by technological interactivity can enhance customer satisfaction [45]. Furthermore, interactive exchanges, which involve customers actively engaging with robots, can result in distinctive experiences and unique service relationships [46], thereby impacting perceived value. Another significant aspect of voice assistant affordance is anytime connectivity [33]. In a hotel room, voice assistant can connect with curtains, lights, air conditioning, and television through the Internet of Things (IoT), allowing customers to call upon the AI for services at any time [21]. Concurrently, based on personalized customer needs, voice assistants can also provide information services at any given moment, creating a customer experience of being served at all times. This approach can help to enhance the customer’s perceived value. Specifically, it is hypothesized that each of the three characteristics mentioned plays a significant role in influencing this perceived value. To clarify, the following hypotheses are proposed:
H1. 
Information association has a positive effect on customers’ perceived value.
H2. 
Interactivity has a positive effect on customers’ perceived value.
H3. 
Anytime connectivity has a positive effect on customers’ perceived value.

3.2. VA Affordances and Human–AI Rapport

Human staff members proactively partake in rapport-building actions, furnishing customers with corresponding “script” prompts that trigger akin responses, thereby fostering a rapport-building process [47]. Moreover, seamless interaction forms the bedrock of establishing rapport [47]. Similarly, through vocal interactions, customers can conduct information searches, access voice-automated amenities, and offer tailored services to customers in real time [21,48]. Drawing from the Media Equation Theory, it is posited that individuals intuitively interact with a technological interface exhibiting specific cues, aligning their engagement in accordance with societal norms [49]. Moreover, according to Kim et al. [50], favorable interactive performance can bolster the rapport between humans and machines. Based on the above, it is proposed that:
H4. 
Information association has a positive effect on human–AI rapport.
H5. 
Interactivity has a positive effect on human–AI rapport.
H6. 
Anytime connectivity has a positive effect on human–AI rapport.

3.3. Perceived Value and eWOM Intention

According to the theory of service-dominant logic, customers act as operant resources or collaborative partners, interacting with product or service offerings and other resources to jointly create value with the offering firms [38]. eWOM depends on customers’ emotional reactions, perception of value, and stance towards the enterprise [11]. Customers’ perception of value emerges from their appraisal of their interactions with products and services, serving as a pivotal gauge of customer experience. Perceived value can serve as a substantial forecaster of customers’ affirmative intentions and actions post consumption [51]. When customers perceiving utilitarian values and experience hedonic pleasure from a service [10], they are inclined to formulate positive assessments and sentiments, often leading to greater likelihood or intention to engage in eWOM [33].
H7. 
Customers’ perceived value has a positive effect on their eWOM intention.

3.4. Human–AI Rapport and eWOM Intention

In interactions between customers and human employees, the establishment of a harmonious relationship can enhance the customer’s hospitality experience, making it more likely to induce positive word-of-mouth and loyalty from customers [8,47]. According to the Media Equivalence Theory, when enough social cues are provided in human–machine interactions, they can generate social effects similar to those formed in human-to-human interactions [47]. Moreso than traditional technologies, service robots act as quasi-employees serving customers [52]. In particular, VAs provide more immediate and personalized feedback, enhancing customers engagement and interactive experiences [19]. This fosters a harmonious human–machine rapport, strengthening users’ emotional connections and further motivating them to share these positive interactive experiences, generating positive WOM behaviors [13]. Thus, the following is proposed:
H8. 
Human–AI rapport has a positive effect on customers’ eWOM intention.

3.5. The Moderating Effect of Social Presence

Typically, VAs more effectively emulate human traits than traditional machines, which offers advantages in conveying information and emotions to customers [53]. For example, VAs’ ability to closely replicate human speech and expressions renders human–service robot interactions more akin to authentic interpersonal exchanges [53]. Araujo [54] posited that physical embodiment is not a prerequisite for evoking social responses in human–machine interactions and argued it is the interactive linguistic style that triggers social reactions. Notably, in comparison to text chatbots, VAs necessitate more sophisticated machine learning capabilities, resulting in enhanced fidelity and interactivity in human–machine conversations [55]. Furthermore, referring to the social influence threshold models, authentic interactive behavior by AI agents such as can prompt social responses from technology users, and the degree of social presence AI agents achieve plays a crucial role in human–AI interactions [56]. Hence, the following are proposed:
H9. 
Social presence significantly moderates the effect of VA affordances on human–AI rapport.
H9a. 
Social presence significantly moderates the effect of information association on human–AI rapport.
H9b. 
Social presence significantly moderates the effect of interactivity on human–AI rapport.
H9c. 
Social presence significantly moderates the effect of anytime connectivity on human–AI rapport.

3.6. The Moderating Effect of Privacy Concerns

When consumers harbor concerns about privacy security, they begin to question the overall security of the consumption process. Such doubts prompt consumers to reassess the value of the services [19]. Even if the service is perceived as high in functionality and quality, privacy risks are considered a negative factor, diminishing the overall evaluation of the service by the consumer [7].
From the perspective of consumer decision making, when considering whether to engage in eWOM, consumers weigh the benefits of recommending (based on perceived value) against potential risks (privacy concerns) [19]. If privacy concerns are significant, consumers may believe that recommending AI service could expose others to privacy risks, which conflicts with their sense of moral responsibility or they may fear being blamed for any privacy issues that arise after their recommendation [3,20]. Hence, the following is proposed:
H10. 
Privacy concerns significantly moderates the effect of perceived value on customers’ eWOM intention.
When privacy concerns are heightened, customers shift their focus towards the protection of personal information. Even if customers have established a harmonious relationship with AI, privacy concerns lead them to question the entire interaction process [50]. For example, they may worry that personal information disclosed during communication with AI could be improperly collected or leaked. Such concerns trigger a risk-averse mentality in customers.
In considering whether to engage in eWOM, customers weigh the potential benefits of recommending (based on a rapport relationship with AI, such as the satisfaction derived from sharing quality service experiences) against potential risks (risk of privacy breaches) [50]. If privacy concerns are substantial, customers may believe that recommending could expose others to privacy risks, and they themselves could suffer negative evaluations or consequences due to privacy issues arising from their recommendations [2,19]. Hence, the following is proposed:
H11. 
Privacy concerns significantly moderates the effect of human–AI rapport on customers’ eWOM intention.
The proposed model guiding this study is depicted in Figure 1.

4. Methodology

4.1. Sampling

In January 2024, this study received ethical approval from the ethics committee, based on which survey work commenced. Various ethical guidelines were also observed in the survey process including obtaining informed consent before administration of questionnaires, notifying respondents of their right to terminate the process at any time, and anonymizing all data collected to protect respondents’ privacy.
To ensure the reliability and validity of the questionnaire, a pilot study was conducted before main survey distribution. The pretest was administered offline, and 100 questionnaires were administered. In total, for the pretest, 72 valid questionnaires were collected and analyzed, and the questionnaire was found to demonstrate good reliability and validity. During the distribution of the questionnaires, researchers explicitly informed participants about the primary purpose of the questionnaire and assured the respondents’ privacy during the completion process. Cases where respondents provided only one answer to all questions, offered multiple responses, or submitted incomplete forms were excluded. For the formal data collection phase, a total of 580 questionnaires were similarly distributed offline in 4 smart hotels in Nanjing, Hangzhou, and Urumqi, China. The hotels utilized either Xiao Ai Classmates or Xiao Du devices for VA services. After invalid questionnaires, characterized by incomplete responses or uniform answer selections throughout the entire questionnaire, were eliminated, a total of 529 valid questionnaires were retained.

4.2. Measured Items

Affordance adopted and modified by Li et al. [33], wherein affordance is regarded as a second-order construct comprising three primary variables: interactivity, anytime connectivity, and information association. The concept of perceived value was adopted from work by Maroufkhani et al. [26], while human–AI rapport was adapted from work by Kim, So, and Wirtz [50]. eWOM intention was defined based on the work of Lai, Liu, and Lu [57]. Social presence as a construct was adapted from the work of McLean and Osei-Frimpong [58]. Privacy concerns was adopted from Maduku et al. [19]. The specific measurement items can be found in Table 1. Given the potential sensitivity attached to the nature of self-reported inquiries under consideration, we elected to incorporate a version of the Loo and Thorpe (2000) [59] (with a reliability coefficient of 0.68), as a control variable. Items embodied in this scale encompass statements such as “I never hesitate to go out of my way to assist someone in distress” and “I perpetually maintain courtesy, even when dealing with individuals who may be deemed disagreeable”.
Following Brislin’s [60] “translation-back-translation” method, the English scales were initially translated into Chinese and then back into English to ensure measurement impartiality. Subsequently, two service management experts, both associate professors, were enlisted to scrutinize the scales for authenticity, precision, and clarity of translation and expression. In this study, a five-point Likert scale was employed for measurement purposes, with a rating of ‘1’ indicating complete disagreement and ‘5’ indicating complete agreement.

5. Results

5.1. Descriptive Analysis

The 529 valid questionnaires collected were relatively balanced in distribution between male (n = 277) and female (n = 252) participants, with males representing 52.36% and females 47.64% of the sample. In terms of education level, individuals with high school education (junior college) or below comprised 44.98% of the respondents, while those with college-level education (junior college, undergraduate) or higher accounted for 55.02%. In terms of income, the majority fell within the 6001–9000 RMB range, followed closely by the 9001–12,000 RMB range (see Table 2). The demographic of our research sample, encompassing aspects such as age, occupation, and income, aligns substantially with those employed by previous scholars [49,52].

5.2. Common Method Bias

Data are best gathered from diverse sources to mitigate common method bias. However, to ensure the consistency of the dataset and better safeguard the confidentiality of participants, our research exclusively utilized a singular quantitative methodology (anonymized questionnaire). Fornell and Larcker [61] highlighted that, even in Exploratory Factor Analysis (EFA), the bias associated with opting for a single technique is negligible if the explained variance of the primary factor remains below 50%. We, therefore, scrutinized for common method bias using Harman’s one-factor test. Results indicated that the initial component explained 34.89% of the variance, signifying that common method bias does not pose a significant concern in this inquiry and does not unduly influence the outcomes.

5.3. Measurement Model

This study used confirmatory factor analysis (CFA) to assess the measurement model. The model’s fit was examined using AMOS 24.0, yielding the following results: χ2 = 504.084, χ2/df = 2.879, CFI = 0.888, TLI = 0.866, IFI = 0.890, RMSEA = 0.06. These findings indicated that the model’s goodness of fit was within an acceptable range [62]. Furthermore, all latent variables demonstrated Cronbach’s α values exceeding 0.7, indicating strong internal consistency. The items measuring these variables exhibited factor loadings above 0.6. The Average Variance Extracted (AVE) values exceeded 0.5, confirming the convergent validity of the variables under study, see Table 3. Additionally, we verified the multicollinearity issue within the model. The Variance Inflation Factor (VIF) values of the independent variables ranged from 1.03 to 1.296, with tolerance values exceeding 0.772 and condition indices below 10, indicating the absence of multicollinearity concerns.
Discriminant validity was assessed by examining the diversity in characteristics represented by the latent variables. Favorable discriminant validity is indicated when the square root of the AVE for each latent variable surpasses the correlation coefficient between the variables. In this study, the square root of the AVE values for the latent variables exceeded the correlation coefficients, underscoring robust discriminant validity within the measurement model, see Table 4.

5.4. Structural Equation Model (SEM)

Based on the collected valid samples, a structural equation model (SEM) of the research framework (depicted in Figure 1) was established. During data analysis, we incorporated demographic characteristics, including age, gender, and educational level, as control variables. AMOS 24.0 was used to test the parameter estimation and model fit. The results (χ2 = 571.32, χ2/df = 3.01, CFI = 0.868, TLI = 0.844, GFI = 0.918, and RMSEA = 0.055) demonstrated that the model fit the data relatively well [63]. Furthermore, the R2 values of the endogenous variables—human–AI rapport, eWOM intention, and perceived value—are 0.582, 0.484, and 0.503, respectively, indicating that the model explains a substantial proportion of the variance in these variables.
Path analysis revealed that interactivity significantly and positively influenced perceived value (β = 0.609, p < 0.001), while anytime connectivity also exhibited a significant positive effect on perceived value (β = 0.419, p < 0.001). However, information association did not demonstrate a significant effect on perceived value (β = 0.110, p = 0.106 > 0.05). Additionally, anytime connectivity had a significant positive impact on customer human–AI rapport (β = 0.191, p < 0.001), the effect of information association on customer human–AI rapport was not statistically significant (β = 0.051, p = 0.456 > 0.05). Furthermore, both perceived value and customer human–AI rapport significantly contributed to willingness to engage in eWOM (β = 0.551, p < 0.001; β = 0.531, p < 0.001). Overall, hypotheses H2, H3, H5, H6, H7, and H8 were validated, see Table 5.
To ensure the robustness of our model, we conducted group effect analysis, encompassing tests for measurement invariance and multi-group analyses. We employed the AMOS multigroup structural equation model for a significance test of variations and similarities among individuals with differing characteristics. This further improves the empirical validity of our study based on the effect of the availability of voice technology on eWOM mechanisms. By analyzing the differences within model fit indices, namely the default (no parameter constraints adjusted model), covariance equality, variance equality, path coefficient equality, and model invariance, we found the x2/df in the default model and the multi-group model ranging from 2.104 to 2.921. The NFI ranged from 0.892 to 0.927, the CFI from 0.917 to 0.938, the IFI from 0.902 to 0.938, TLI from 0.903 to 0.933, and the RMSEA from 0.034 to 0.055. These values generally met the model fit requirements. This indicates that the multi-group structural model fits the sample data well and allows for subsequent path difference analysis based on factors such as education, occupation, and gender grouping. By examining multiple groups divided based on different occupations, educational backgrounds, and genders, it was observed that the path coefficients of the model do not exhibit significant differences.

5.5. Moderating Effect Test

We used the PROCESS macro to test the moderating effect of H9, using social presence as a moderator. Social presence did not significantly moderate the impact of information association on human–AI rapport (β = −0.042, p = 0.3701 > 0.05, 95% confidence interval: [−0.1349, 0.0504]); H9a was, hence, not supported. It also did not significantly moderate the impact of interactivity on human–AI rapport (β = 0.066, p = 0.8361 > 0.05, 95% confidence interval: [−0.1163, 0.1436]); therefore, H9b was also not supported. However, social presence significantly moderated the impact of anytime connectivity on human–AI rapport (β = 0.096, p = 0.011 < 0.05, 95% confidence interval: [0.0223, 0.1711]), supporting H9c.
We used the same method to test the moderating effect of H10 and H11. We found privacy concerns significantly moderated the impact of perceived value on customers’ eWOM intention (β = −0.046, p = 0.047 < 0.05, 95% confidence interval: [−0.0975, −0.005]). Privacy concerns did not significantly moderate the effect of human–AI rapport on customers’ eWOM intention (β = 0.089, p = 0.958 > 0.05, 95% confidence interval: [−0.0600, 0.0633]).

6. Discussion and Conclusions

6.1. Discussion of Findings

While voice assistant (VA) technology has found application in diverse service contexts including serving as intelligent customer service agents, hotel concierge aides, and shopping guides, the impact of customer–VA interactions on fostering eWOM behavior towards service enterprises in the hospitality industry remains unclear [40]. Hence, we examined how eWOM intentions are influenced by human–AI VA interaction experiences. Drawing on service-dominant logic and affordance theory, we built an eWOM intention framework with three dimensions of VA affordances as antecedent variables and perceived value alongside human–AI rapport as pivotal mediators.
Among the three dimensions of affordances, interactivity and anytime connectivity had validated positive impacts on perceived value. These findings align with those of Fang [36], suggesting that leveraging interactivity and anytime connectivity and empowering individuals to actively engage in human–AI interactions can enhance user experiences and perceived value [31]. This may include offering users the ability to proactively customize their information preferences (active control), receive instant responses from VAs (synchronicity), and engage in two-way conversations with VAs.
Drawing on service-dominance logic, we also validated the theoretical pathway of technical VA attributes contributing value and, thus, eWOM (VA affordances —> perceived value —> eWOM intention). Like Gronroos and Voima [63], we find co-creation of value hinges on interactivity, with value is generated or realized through active customer participation in the service delivery process. Our findings also support the viewpoint that smart technologies, specifically VAs, by facilitating customer interactions, can foster value creation [33] and positively impact eWOM intentions. Moreover, we validated an additional pathway, where technical VA attributes foster VA rapport which in turn positively impact eWOM intentions (VA affordances —> human–AI rapport —> eWOM intention), a discovery that aligns with the research of Fatima et al. [13].
The two hypotheses in this article—H1 and H4—were not supported. The positive impacts of information association on perceived value and human–AI rapport were not significant. In past research, Li et al. [33] found that information association could positively enhance the customer’s experience and perception of personalized service during interactions with a chatbot. Compared to other service scenarios, hotel service involves not only completing the cognitive tasks advanced by the customer but also offering a certain degree of emotional service, such as companionship [21]. To date, even though voice-assisted technology is rapidly advancing, within the context of hotel services, voice-assisted applications can make responses, but these replies might not always accurately satisfy customers’ personalized emotional needs. This could potentially result in reduced usefulness of information, thus, affecting the perceived value of AI and the human–AI relationship.
We further found that VA social presence enhances the impact of anytime connective ity on human–AI rapport of customers. This suggests that a strong VA social presence can reduce feelings of detachment and foster a sense of companionship and support for users interacting with VAs [50]. Moreover, strong social presence can bolster users’ recourse to AI’s capabilities, thereby aiding in rapport building and positively influencing the impact of anytime connectivity. Interestingly, we observed that social presence did not have a statistically significant moderating effect on the influence of interactivity and information association on human–AI rapport. This could be attributed to the fact that the ability of AI agents’ social presence to enhances the perception of warmth in human–AI interactions is mostly beneficial for rapport building [50]. However, it is also possible greater VA social presence may interfere with the sense of control in interactions [48]. From the result of our study, that means, the positive effect of social presence may be offset.
In the digital era, privacy security represents a significant ethical issue within human–machine interactions. Previous studies have identified that privacy concerns can influence customers’ emotional responses to using VAs, thereby exerting negative impacts, such as reducing the willingness to use technology and the intention to repurchase hotel products [7]. In this study, privacy concerns modulate the effect of the availability of VAs on perceived value and rapport in human–machine interactions. Notably, compared to other artificial technologies, the negative effects of privacy concerns are more pronounced due to the unique advantages offered by VA services, such as anytime connectivity [7]. Therefore, in promoting VA devices, a transparent strategy should be adopted. This includes clearly informing customers about the scope of data collection, usage, and protection measures of the device. Additionally, providing customers with more options for privacy control, such as allowing them to choose whether to activate certain privacy-sensitive functions, can mitigate their privacy concerns, thereby enhancing the usage rate of VA devices and improving the reputation of hotels.

6.2. Theoretical Implications

This study offers several valuable contributions to the field of human–AI interaction and eWOM/WOM behavior. First, being situated within the realm of VAs services, this research leverages affordance theory and service-dominant logic to investigate and confirm the influence of affordances (specifically information association, interactivity, and anytime connectivity) on customers’ eWOM intention towards businesses. This is particularly valuable as prior studies on WOM and eWOM stemming from consumer interactions with AI have centered on text chatbots predominantly [16], and only limited research has delved into eWOM effects elicited by VA technologies [10]. Moreover, in the limited examples of the latter, the focus has mainly on assessing the technology’s reputation rather than its impact on perceived value and eWOM. The outcomes of this research especially enrich the existing literature by introducing and validating a service-dominant logic model that incorporates AI affordance constructs to elucidate customers’ perceived value, human–AI rapport, and eWOM intentions towards enterprises.
Second, this research contributes significantly to the comprehension of how VA fea tures prompt cognitive and emotional cues to affect customers’ intention to engage in eWOM following interactions with AI services. Previous studies on WOM stemming from consumer interactions with AI have predominantly concentrated on cognitive pathways [33] or emotional pathways [10] exclusive of the other. In contrast, this study jointly models both cognitive cue (perceived value)- and emotional cue (human–AI rapport)-based pathways for various dimensions of affordances to impact eWOM.
Third, this study identifies the moderating effects of social presence on affordances on human–AI rapport, extending the understanding of the impact of affordances on human–AI rapport as it pertains to eWOM. Moreover, while social presence has been viewed as an attribute of technology in previous studies [10], this study also explores the interaction effect between different attributes (social presence and affordance) of VAs, contributing to a comprehensive understanding of how the AI that influence human–AI rapport.
Furthermore, this research investigated the moderating influence of privacy concerns on the relationship between affordances and both human–AI rapport and perceived value and examined how privacy concerns moderate the impact of perceived value and human–AI rapport on eWOM intentions. This analysis contributes to a more nuanced comprehension of the adverse implications of privacy intrusions within VA services, enhancing the theoretical framework surrounding consumer interactions with AI technologies.

6.3. Practical Implications

Enhancing eWOM intention is undeniably valuable; eWOM has been acknowledged to significantly aid in expanding an enterprise’s customer base and market [2]. Presently, the installation of intelligent voice devices in hotels has gained favor among certain hotel groups. For example, Xiaomi Group and Huazhu Hotel Group have collaborated to roll out a hotel version of ‘Xiao Ai’. In the context of hotel services undergoing digital transformation, the research findings of this study have practical implications for enterprises seeking to use VAs service systems to enhance consumer eWOM intention, to bolster interactivity, anytime connectivity, or to foster heightened perceived value and human–AI rapport. A few recommendations are offered below.
Firstly, improving the interactivity of VA means enhancing their technical performance in crucial ways. Designers should prioritize the stability of VA service systems to minimize crashes and delays [64,65]. This enhancement will significantly augment the immediacy of the interactive process and improve the overall interactive experience. Additionally, it is essential to improve on the natural language recognition and understanding capabilities of intelligent voice devices. In particular, mimicking human speech patterns more closely, being highly expressive, and utilizing linguistic diversity will enable VAs to convey richness more effectively [65]. This approach will not only make VAs’ language expression and understanding abilities akin to those of humans but also optimize two-way interactions between humans and machines [66]. Moreover, as voice-based interactive communication may exhibit relatively weaker controllability, strengthening VA controllability is crucial [48,67]. This will empower customers and lead to greater satisfaction and more seamless and engaging user experiences.
Secondly, designers should focus on anytime connectivity abilities in VAs. VAs should be on standby, ready to respond to customers’ calls promptly. Interaction efficacy should be measured not only by the speed and responsiveness but also by the ability to be activated at any given moment. Moreover, enhancing the connectivity between VAs and service providers both within and outside the enterprise is essential. Maximizing the integration of internal service operating systems can streamline various functions such as check-out services and in-room dining orders, elevating the overall customer experience [21]. In addition, establishing robust connections with diverse third-party service platforms, such as ride-hailing and food delivery services, enhances the ability of VAs to fulfill customer requests more effectively and seamlessly, creating a more tailored and comprehensive service experience.
Thirdly, VAs’ social presence is of significance in amplifying the impact of anytime connectivity on human–AI rapport offers a valuable way to enhance customers’ eWOM intention. To bolster social presence, it is imperative to focus on VAs’ ability to deliver emotional value and enhance their empathic abilities. VAs can offer heightened emotional value by enhancing their natural flexibility of language and incorporating elements of cuteness, playfulness, and humor in their linguistic expression [66]. Meanwhile, enhancing the empathetic capabilities of VA systems can mean incorporating interjections like “ah”, “yeah”, and “hmm”, and appropriate variations in intonation to match contextual emotional expressions [67]. These strategies collectively contribute to creating more empathetic and emotionally resonant interactions, fostering a stronger human–AI rapport and ultimately enhancing eWOM intention.
In addition, critical factors such as cost, benefit, implementation timelines, and risk mitigation strategies warrant careful consideration. The estimated renovation and installation cost per guest room is approximately RMB 5500. Direct costs encompass equipment purchases (including smart VAs and associated hardware), installation, staff training, and ongoing maintenance, while indirect costs may arise from business disruptions during system failures or maintenance.
The integration of smart voice technology significantly enhances guest satisfaction and increases occupancy rates, fostering positive feedback. Advances in technology have improved the personalization of these devices, with guests interacting with VAs 11 to 15 times per night [68]. By 2023, 40% of newly opened hotel rooms in China were equipped with VA terminals, underscoring their importance in digital transformation. According to the 2024 “Digital Transformation and ROI” industry report, 71% of hotel managers acknowledge that technology boosts guest experiences and service performance. Additionally, 62% of guests feel their service experience has improved due to technological enhancements. Customer reviews for hotel smart products have surged, rising from over 9000 in January 2020 to approximately 30,000 monthly by 2023 [69]. This increase in online reputation has significantly boosted hotel customer attraction, with hotels reporting a 4.88% rise in occupancy rates within three months of upgrading to smart rooms equipped with Xiaodu VAs.
The implementation timeline involves selecting vendors and products, purchasing and installing equipment, training staff, conducting tests, and system operationalization, typically within a three-month period. However, the deployment of VA devices introduces risks such as technical failures, low user acceptance, and data security concerns. To mitigate these risks, we propose the following: (1) Technical Failures: regular maintenance and system updates, along with emergency plans for unexpected interruptions, are essential. Continuously improve technology based on customer needs and maintain close contact with VA suppliers to prepare for future upgrades; (2) Low User Acceptance: provide comprehensive training for staff and education for guests to ensure they can understand and use the VAs effectively; (3) Data Security Issues: implementing stringent data encryption and adhering to data protection regulations are imperative to safeguard customer privacy. These measures will not only enhance security but also build trust among users, thereby fostering a more secure and engaging technological environment in the hospitality industry.

6.4. Limitations and Future Research

As with any research, this study has a number of limitations. Firstly, in terms of research method, this study solely employs a mono instrument survey approach, obtaining data via hotel customers’ self-report, which is limited in validity. Future studies could consider the experimental method to verify the identified causal relationships between variables. Additionally, our data exhibited a good fit for a linear relationship and, indeed, we cannot completely rule out potential non-linearity, future research may explore alternative models or non-linear relationships.
Secondly, individual factors may potentially influence the effect of VA affordance on eWOM; however, this study does not delve into the internal heterogeneity of customer elements such as usage motivation [70], interaction frequency [64], and so on. Subsequent research could investigate the differential effects arising from these elements to offer valuable insights for market development and segmentation. Third, it is important to acknowledge that the survey data gathered for this study predominantly comes from China; therefore, it mirrors the particular cultural influences that shape views about technology in that context. In light of this, future research could yield interesting insights by introducing a comparative study of Eastern and Western sentiments in relation to intelligent voice technology. Fourthly, scholars have noted that customers’ brand perceptions of the hotel itself can indirectly influence their assessment of VA service performance [20,26]. However, the study does not specifically examine the impact of different brand and service types due to time and scope constraints. Future research could employ experimental methods to compare various service types and enterprise brands. Lastly, the study does not differentiate among various customer segments. VA services offer a natural mode of interaction which is particularly beneficial for digitally disadvantaged groups such as the blind and illiterate [71]. However, this study focuses on the impact of eWOM generated by VA services on enterprises and does not address disadvantaged groups. Future research avenues could explore the influence of affordances on digitally vulnerable groups within the realm of VA services.

Author Contributions

Conceptualization, F.Y., X.L. and T.Y.; formal analysis, F.Y.; funding acquisition, T.Y.; investigation, F.Y. and X.L.; methodology, F.Y., X.L. and T.Y.; supervision, T.Y.; writing—original draft, F.Y. and X.L.; writing—review and editing, F.Y. and T.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Social Science Foundation of China (22BJY148).

Institutional Review Board Statement

The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of Xinjiang University of Finance and Economics (Date of approval: 11 January 2024).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

Data supporting reported results are available from the authors on request.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Lee, W.-L.; Liu, C.-H.; Tseng, T.-W. The Multiple Effects of Service Innovation and Quality on Transitional and Electronic Word-of-Mouth in Predicting Customer Behaviour. J. Retail. Consum. Serv. 2022, 64, 102791. [Google Scholar] [CrossRef]
  2. Borghi, M.; Mariani, M.M. Service Robots in Online Reviews: Online Robotic Discourse. Ann. Tour. Res. 2021, 87, 103036. [Google Scholar] [CrossRef]
  3. Belhadi, A.; Kamble, S.; Benkhati, I.; Gupta, S.; Mangla, S.K. Does Strategic Management of Digital Technologies Influence Electronic Word-of-Mouth (eWOM) and Customer Loyalty? Empirical Insights from B2B Platform Economy. J. Bus. Res. 2023, 156, 113548. [Google Scholar] [CrossRef]
  4. Schepers, J.; Belanche, D.; Casaló, L.V.; Flavián, C. How Smart Should a Service Robot Be? J. Serv. Res. 2022, 25, 565–582. [Google Scholar] [CrossRef]
  5. Kang, W.; Shao, B.; Du, S.; Chen, H.; Zhang, Y. How to Improve Voice Assistant Evaluations: Understanding the Role of Attachment with a Socio-Technical Systems Perspective. Technol. Forecast. Soc. Chang. 2024, 200, 123171. [Google Scholar] [CrossRef]
  6. Liu, S.Q.; Vakeel, K.A.; Smith, N.A.; Alavipour, R.S.; Wei, C.; Wirtz, J. AI Concierge in the Customer Journey: What is It and How Can it Add Value to the Customer? J. Serv. Manag. 2024, 35, 136–158. [Google Scholar] [CrossRef]
  7. Kim, J.; Erdem, M.; Kim, B. Hi Alexa, do hotel guests have privacy concerns with you? A cross-cultural study. J. Hosp. Mark. Manag. 2024, 33, 360–383. [Google Scholar]
  8. Yin, D.; Li, M.; Qiu, H.; Bai, B.; Zhou, L. When the Servicescape Becomes Intelligent: Conceptualization, Assessment, and Implications for Hospitableness. J. Hosp. Tour. Manag. 2023, 54, 290–299. [Google Scholar] [CrossRef]
  9. Poushneh, A. Impact of Auditory Sense on Trust and Brand Affect through Auditory Social Interaction and Control. J. Retail. Consum. Serv. 2021, 58, 102281. [Google Scholar] [CrossRef]
  10. Mishra, A.; Shukla, A.; Sharma, S.K. Psychological Determinants of Users’ Adoption and Word-of-Mouth Recommendations of Smart Voice Assistants. Int. J. Inf. Manag. 2022, 67, 102413. [Google Scholar] [CrossRef]
  11. Donthu, N.; Kumar, S.; Pandey, N.; Pandey, N.; Mishra, A. Mapping the Electronic Word-of-Mouth (eWOM) Research: A Systematic Review and Bibliometric Analysis. J. Bus. Res. 2021, 135, 758–773. [Google Scholar] [CrossRef]
  12. Akbari, M.; Foroudi, P.; Zaman Fashami, R.; Mahavarpour, N.; Khodayari, M. Let Us Talk about Something: The Evolution of e-WOM from the Past to the Future. J. Bus. Res. 2022, 149, 663–689. [Google Scholar] [CrossRef]
  13. Kayeser Fatima, J.; Khan, M.I.; Bahmannia, S.; Chatrath, S.K.; Dale, N.F.; Johns, R. Rapport with a Chatbot? The Underlying Role of Anthropomorphism in Socio-Cognitive Perceptions of Rapport and e-Word of Mouth. J. Retail. Consum. Serv. 2024, 77, 103666. [Google Scholar] [CrossRef]
  14. Verma, S.; Yadav, N. Past, Present, and Future of Electronic Word of Mouth (EWOM). J. Interact. Mark. 2021, 53, 111–128. [Google Scholar] [CrossRef]
  15. Ismagilova, E.; Rana, N.P.; Slade, E.L.; Dwivedi, Y.K. A Meta-Analysis of the Factors Affecting eWOM Providing Behaviour. EJM 2021, 55, 1067–1102. [Google Scholar] [CrossRef]
  16. Yim, M.C. Effect of AI Chatbot’s Interactivity on Consumers’ Negative Word-of-Mouth Intention: Mediating Role of Perceived Empathy and Anger. Int. J. Hum.–Comput. Interact. 2023, 40, 5415–5430. [Google Scholar] [CrossRef]
  17. Huang, B.; Philp, M. When AI-Based Services Fail: Examining the Effect of the Self-AI Connection on Willingness to Share Negative Word-of-Mouth after Service Failures. Serv. Ind. J. 2021, 41, 877–899. [Google Scholar] [CrossRef]
  18. Shahzad, M.F.; Xu, S.; An, X.; Javed, I. Assessing the Impact of AI-Chatbot Service Quality on User e-Brand Loyalty through Chatbot User Trust, Experience and Electronic Word of Mouth. J. Retail. Consum. Serv. 2024, 79, 103867. [Google Scholar] [CrossRef]
  19. Maduku, D.K.; Mpinganjira, M.; Rana, N.P.; Thusi, P.; Ledikwe, A.; Mkhize, N.H. Assessing Customer Passion, Commitment, and Word-of-Mouth Intentions in Digital Assistant Usage: The Moderating Role of Technology Anxiety. J. Retail. Consum. Serv. 2023, 71, 103208. [Google Scholar] [CrossRef]
  20. Cai, X.; Cebollada, J.; Cortiñas, M. Impact of Seller- and Buyer-Created Content on Product Sales in the Electronic Commerce Platform: The Role of Informativeness, Readability, Multimedia Richness, and Extreme Valence. J. Retail. Consum. Serv. 2023, 70, 103141. [Google Scholar] [CrossRef]
  21. Buhalis, D.; Moldavska, I. Voice Assistants in Hospitality: Using Artificial Intelligence for Customer Service. JHTT 2022, 13, 386–403. [Google Scholar] [CrossRef]
  22. Arndt, J.E. Role of Product-Related Conversations in the Diffusion of a New Product. J. Mark. Res. 1967, 4, 291–295. [Google Scholar] [CrossRef]
  23. Barreto, A.M. The Word-of-Mouth Phenomenon in the Social Media Era. Int. J. Mark. Res. 2015, 56, 631–654. [Google Scholar] [CrossRef]
  24. Cao, D.; Sun, Y.; Goh, E.; Wang, R.; Kuiavska, K. Adoption of Smart Voice Assistants Technology Among Airbnb Guests: A Revised Self-Efficacy-Based Value Adoption Model (SVAM). Int. J. Hosp. Manag. 2022, 101, 103124. [Google Scholar] [CrossRef]
  25. Kumar, A.; Bala, P.K.; Chakraborty, S.; Behera, R.K. Exploring Antecedents Impacting User Satisfaction with Voice Assistant App: A Text Mining-Based Analysis on Alexa Services. J. Retail. Consum. Serv. 2024, 76, 103586. [Google Scholar] [CrossRef]
  26. Maroufkhani, P.; Asadi, S.; Ghobakhloo, M.; Jannesari, M.T.; Ismail, W.K.W. How Do Interactive Voice Assistants Build Brands’ Loyalty? Technol. Forecast. Soc. Chang. 2022, 183, 121870. [Google Scholar] [CrossRef]
  27. Hu, Y.; Min, H. The Dark Side of Artificial Intelligence in Service: The “Watching-Eye” Effect and Privacy Concerns. Int. J. Hosp. Manag. 2023, 110, 103437. [Google Scholar] [CrossRef]
  28. Moreno, M.A.; D’Angelo, J. Social Media Intervention Design: Applying an Affordances Framework. J. Med. Internet Res. 2019, 21, e11014. [Google Scholar] [CrossRef]
  29. Treem, J.W.; Leonardi, P.M. Social Media Use in Organizations: Exploring the Affordances of Visibility, Editability, Persistence, and Association. Ann. Int. Commun. Assoc. 2013, 36, 143–189. [Google Scholar] [CrossRef]
  30. Rice, R.E.; Evans, S.K.; Pearce, K.E.; Sivunen, A.; Vitak, J.; Treem, J.W. Organizational Media Affordances: Operationalization and Associations with Media Use: Organizational Media Affordances. J. Commun. 2017, 67, 106–130. [Google Scholar] [CrossRef]
  31. Chan, T.K.H.; Cheung, C.M.K.; Wong, R.Y.M. Cyberbullying on Social Networking Sites: The Crime Opportunity and Affordance Perspectives. J. Manag. Inf. Syst. 2019, 36, 574–609. [Google Scholar] [CrossRef]
  32. Lin, J.C.; Chang, H. The Role of Technology Readiness in Self-service Technology Acceptance. Manag. Serv. Qual. Int. J. 2011, 21, 424–444. [Google Scholar] [CrossRef]
  33. Li, C.-Y.; Fang, Y.-H.; Chiang, Y.-H. Can AI Chatbots Help Retain Customers? An Integrative Perspective Using Affordance Theory and Service-Domain Logic. Technol. Forecast. Soc. Chang. 2023, 197, 122921. [Google Scholar] [CrossRef]
  34. Tiihonen, J.; Felfernig, A. An Introduction to Personalization and Mass Customization. J. Intell. Inf. Syst. 2017, 49, 1–7. [Google Scholar] [CrossRef]
  35. Lee, E.-J.; Park, J.K. Online Service Personalization for Apparel Shopping. J. Retail. Consum. Serv. 2009, 16, 83–91. [Google Scholar] [CrossRef]
  36. Fang, Y.-H. An App a Day Keeps a Customer Connected: Explicating Loyalty to Brands and Branded Applications through the Lens of Affordance and Service-Dominant Logic. Inf. Manag. 2019, 56, 377–391. [Google Scholar] [CrossRef]
  37. Wagner, D.; Vollmar, G.; Wagner, H.-T. The Impact of Information Technology on Knowledge Creation: An Affordance Approach to Social Media. J. Enterp. Inf. Manag. 2014, 27, 31–44. [Google Scholar] [CrossRef]
  38. Vargo, S.L.; Lusch, R.F. The Four Service Marketing Myths: Remnants of a Goods-Based, Manufacturing Model. J. Serv. Res. 2004, 6, 324–335. [Google Scholar] [CrossRef]
  39. Yazdanparast, A.; Manuj, I.; Swartz, S.M. Co-creating Logistics Value: A Service-dominant Logic Perspective. Int. J. Logist. Manag. 2010, 21, 375–403. [Google Scholar] [CrossRef]
  40. Vargo, S.L.; Akaka, M.A. Service-Dominant Logic as a Foundation for Service Science: Clarifications. Serv. Sci. 2009, 1, 32–41. [Google Scholar] [CrossRef]
  41. Go, E.; Sundar, S.S. Humanizing Chatbots: The Effects of Visual, Identity and Conversational Cues on Humanness Perceptions. Comput. Hum. Behav. 2019, 97, 304–316. [Google Scholar] [CrossRef]
  42. Sundar, S.S. The MAIN Model: A Heuristic Approach to Understanding Technology Effects on Credibility; Digital Media: Cambridge, MA, USA, 2018. [Google Scholar]
  43. Han, H.; Hwang, J. Multi-Dimensions of the Perceived Benefits in a Medical Hotel and Their Roles in International Travelers’ Decision-Making Process. Int. J. Hosp. Manag. 2013, 35, 100–108. [Google Scholar] [CrossRef]
  44. Longoni, C.; Cian, L. Artificial Intelligence in Utilitarian vs. Hedonic Contexts: The “Word-of-Machine” Effect. J. Mark. 2022, 86, 91–108. [Google Scholar] [CrossRef]
  45. Shao, Z.; Zhang, J.; Zhang, L.; Benitez, J. Uncovering Post-Adoption Usage of AI-Based Voice Assistants: A Technology Affordance Lens Using a Mixed-Methods Approach. Eur. J. Inf. Syst. 2024, 1–27. [Google Scholar] [CrossRef]
  46. Tung, V.W.S.; Au, N. Exploring Customer Experiences with Robotics in Hospitality. IJCHM 2018, 30, 2680–2697. [Google Scholar] [CrossRef]
  47. Qiu, H.; Li, M.; Shu, B.; Bai, B. Enhancing Hospitality Experience with Service Robots: The Mediating Role of Rapport Building. J. Hosp. Mark. Manag. 2020, 29, 247–268. [Google Scholar] [CrossRef]
  48. Fan, A.; Lu, Z.; Mao, Z. To Talk or to Touch: Unraveling Consumer Responses to Two Types of Hotel in-Room Technology. Int. J. Hosp. Manag. 2022, 101, 103112. [Google Scholar] [CrossRef]
  49. Kothgassner, O.D.; Griesinger, M.; Kettner, K.; Wayan, K.; Völkl-Kernstock, S.; Hlavacs, H.; Beutl, L.; Felnhofer, A. Real-Life Prosocial Behavior Decreases after Being Socially Excluded by Avatars, Not Agents. Comput. Hum. Behav. 2017, 70, 261–269. [Google Scholar] [CrossRef]
  50. Kim, H.; So, K.K.F.; Wirtz, J. Service Robots: Applying Social Exchange Theory to Better Understand Human–Robot Interactions. Tour. Manag. 2022, 92, 104537. [Google Scholar] [CrossRef]
  51. Liu, C.-H.; Horng, J.-S.; Chou, S.-F.; Yu, T.-Y.; Ng, Y.-L. Antecedents and Consequences of New Technology Application Behavior on Word of Mouth: The Moderating Roles of Perceived Interactivity. J. Hosp. Mark. Manag. 2022, 31, 872–898. [Google Scholar] [CrossRef]
  52. van Doorn, J.; Mende, M.; Noble, S.M.; Hulland, J.; Ostrom, A.L.; Grewal, D.; Petersen, J.A. Domo Arigato Mr. Roboto: Emergence of Automated Social Presence in Organizational Frontlines and Customers’ Service Experiences. J. Serv. Res. 2017, 20, 43–58. [Google Scholar] [CrossRef]
  53. Hu, P.; Lu, Y.; Wang, B. Experiencing Power over AI: The Fit Effect of Perceived Power and Desire for Power on Consumers’ Choice for Voice Shopping. Comput. Hum. Behav. 2022, 128, 107091. [Google Scholar] [CrossRef]
  54. Araujo, T. Living up to the Chatbot Hype: The Influence of Anthropomorphic Design Cues and Communicative Agency Framing on Conversational Agent and Company Perceptions. Comput. Hum. Behav. 2018, 85, 183–189. [Google Scholar] [CrossRef]
  55. Huang, M.-H.; Rust, R.T. Engaged to a Robot? The Role of AI in Service. J. Serv. Res. 2021, 24, 30–41. [Google Scholar] [CrossRef]
  56. Blascovich, J. A Theoretical Model of Social Influence for Increasing the Utility of Collaborative Virtual Environments. In Proceedings of the 4th International Conference on Collaborative Virtual Environments, Bonn, Germany, 30 September–2 October 2002; ACM: New York, NY, USA, 2002; pp. 25–30. [Google Scholar]
  57. Lai, I.K.W.; Liu, Y.; Lu, D. The Effects of Tourists’ Destination Culinary Experience on Electronic Word-of-Mouth Generation Intention: The Experience Economy Theory. Asia Pac. J. Tour. Res. 2021, 26, 231–244. [Google Scholar] [CrossRef]
  58. McLean, G.; Osei-Frimpong, K. Hey Alexa … Examine the Variables Influencing the Use of Artificial Intelligent In-Home Voice Assistants. Comput. Hum. Behav. 2019, 99, 28–37. [Google Scholar] [CrossRef]
  59. Loo, R.; Thorpe, K. Confirmatory factor analyses of the full and short versions of the Marlowe–Crowne Social Desirability Scale. J. Soc. Psychol. 2000, 140, 628–635. [Google Scholar] [CrossRef] [PubMed]
  60. Brislin, R.W. A Culture General Assimilator: Preparation for Various Types of Sojourns. Int. J. Intercult. Relat. 1986, 10, 215–234. [Google Scholar] [CrossRef]
  61. Fornell, C.; Larcker, D.F. Evaluating Structural Equation Models with Unobservable Variables and Measurement Error. J. Mark. Res. 1981, 18, 39–50. [Google Scholar] [CrossRef]
  62. Hair, J.F.; Black, W.C.; Babin, B.J.; Anderson, R.E.; Tatham, R.L. Multivariate Data Analysis: A Global Perspective, 7th ed.; Pearson Prentice Hall: Upper Saddle River, NJ, USA, 2010. [Google Scholar]
  63. Grönroos, C.; Voima, P. Critical Service Logic: Making Sense of Value Creation and Co-Creation. J. Acad. Mark. Sci. 2013, 41, 133–150. [Google Scholar] [CrossRef]
  64. Bronner, F.; De Hoog, R. Vacationers and eWOM: Who Posts, and Why, Where, and What? J. Travel Res. 2011, 50, 15–26. [Google Scholar] [CrossRef]
  65. Aeschlimann, S.; Bleiker, M.; Wechner, M.; Gampe, A. Communicative and Social Consequences of Interactions with Voice Assistants. Comput. Hum. Behav. 2020, 112, 106466. [Google Scholar] [CrossRef]
  66. Buteau, E.; Lee, J. Hey Alexa, Why Do We Use Voice Assistants? The Driving Factors of Voice Assistant Technology Use. Commun. Res. Rep. 2021, 38, 336–345. [Google Scholar] [CrossRef]
  67. Bălan, C. Chatbots and Voice Assistants: Digital Transformers of the Company–Customer Interface—A Systematic Review of the Business Research Literature. JTAER 2023, 18, 995–1019. [Google Scholar] [CrossRef]
  68. Sohu News (2024, January). Artificial Intelligence in Hospitality: A Paradigm Shift in Contemporary Hotel Design. Available online: https://www.sohu.com/a/753546578_121864338 (accessed on 20 July 2024).
  69. Ips Consulting (2023, June). China Online Hotel Accommodation Market status and consumption Insights report. Available online: https://www.ipscg.com/detail-538.html (accessed on 7 March 2025).
  70. Choi, T.R.; Drumwright, M.E. “OK, Google, Why Do I Use You?” Motivations, Post-Consumption Evaluations, and Perceptions of Voice AI Assistants. Telemat. Inform. 2021, 62, 101628. [Google Scholar] [CrossRef]
  71. Adam, M.; Wessel, M.; Benlian, A. AI-Based Chatbots in Customer Service and Their Effects on User Compliance. Electron. Mark. 2021, 31, 427–445. [Google Scholar] [CrossRef]
Jtaer 20 00046 g001
Figure 1. The research framework.
Figure 1. The research framework.
Jtaer 20 00046 g001
Table 1. Measurements of constructs.
Table 1. Measurements of constructs.
Construct and SourceItems
Interactivity
Li et al. [33]		I have much control over my interaction with the voice assistant (VA).
When I talk to the VA, I can freely choose the topic.
The VA allows two-way communication between me and the VA.
The VA gives me the opportunity to talk back.
The VA responds to my questions quickly.
I can get information from the VA rapidly.
Anytime connectivity
Li et al. [33]		I have access to the voice assistant (VA) service and information whenever I need it.
I have access to the VA service and information at all times.
I have access to the VA service and information anywhere in the hotel room.
Information association
Li et al. [33]		The voice assistant (VA) enables me to find additional information I did not know.
The VA enables me to discover new products I am unaware of.
The VA tells me information I need.
Human–AI rapport
Kim, So, and Wirtz [50]		I look forward to seeing the voice assistant (VA) when next I visit a hotel.
The VA takes a personal interest in me.
I have a close rapport with the VA.
Perceived value
Maroufkhani et al. [26]		I find it easy to get the voice assistant (VA) to do what I want it to do.
In my experience with the VA, it has satisfied my needs and wants.
Overall, the value of my experience with the VA is high.
eWOM intention
Lai, Liu and Lu [57] 		I am willing to share my experience of the hotel on social media.
I am willing to share my experience of the hotel on the Internet if someone asks.
I am willing to post a positive comment about the hotel on review platforms after the service is over.
Social presence
McLean and Osei-Frimpong [58]		When I interact with the voice assistant (VA) it feels like someone is present in the room.
My interactions with the VA are similar to those with a human.
When I communicate with the VA, I feel like I am dealing with a real person.
I communicate with the VA similarly to how I communicate with humans.
Personal information could be inappropriately used by the manufacturers of this VA.
Privacy concerns
Maduku et al. [19] 		In general, it would be risky to give my personal information to this VA.
There would be a high potential for privacy loss associated with giving personal information to this VA.
Providing my personal information to this VA would involve unexpected problems.
Table 2. Profile of respondents.
Table 2. Profile of respondents.
VariableCategoryFrequencyPercentage (%)
GenderMale27752.36%
Female25247.64%
Educational LevelJunior high school and below6612.47%
Senior high school (including vocational and technical school)17232.51%
College degree (junior college, undergraduate)22843.10%
Postgraduate degree6311.92%
Income Level3000 RMB and below8115.31%
3001–6000 RMB9117.21%
6001–9000 RMB16130.43%
9001–12,000 RMB9818.51%
12,001–15,000 RMB6311.92%
Above 15,001 RMB356.62%
Ages18–2512423.44%
26–3514827.98%
36–4513124.76%
46–559918.72%
Above 56275.10%
Note: n = 529.
Table 3. Reliability and validity indices.
Table 3. Reliability and validity indices.
Constructs/ItemsFactor LoadingsCronbach’s αAVECR
Information association 0.7230.6350.839
IS10.778
IS20.823.
IS30.788
Interactivity 0.8830.5560.882
IN10.762
IN20.721
IN30.687
IN40.794
IN50.733
IN60.772
Anytime connectivity 0.8620.5560.790
AC10.774
AC 20.763
AC 30.698
Human-AI rapport 0.7960.5540.789
HAR10.711
HAR20.735
HAR30.787
eWOM intention 0.7880.5700.799
EI10.789
EI20.745
EI30.729
Perceived value 0.7190.5670.797
PV10.783
PV20.701
PV30.773
Privacy concerns 0.8500.5780.742
PC10.763
PC20.773
PC30.789
PC40.738
Table 4. Discriminant validity.
Table 4. Discriminant validity.
ISINACPVHAREIPC
IS(0.796)
IN0.426(0.746)
AC0.0970.108(0.745)
PV0.1070.6260.420(0.753)
HAR0.0480.3740.1900.619(0.744)
EI0.3340.4350.2590.5560.530(0.755)
PC0.2280.4170.1290.2160.4930.403(0.761)
Note: HAR = human–AI rapport; IS = information association; IN = interactivity; AC = anytime connectivity; PV = perceived value; EI = eWOM intention; PC = privacy concerns.
Table 5. Hypothesis testing.
Table 5. Hypothesis testing.
HypothesisPathβS.E.T-ValueDecision
H1Information association → Perceived value0.110 NS0.0861.583Unsupported
H2Interactivity → Perceived value0.609 ***0.0677.520Supported
H3Anytime connectivity → Perceived value0.419 ***0.0436.872Supported
H4Information association → Human-AI rapport0.051 NS0.1110.708Unsupported
H5Interactivity → Human-AI rapport0.389 ***0.0725.292Supported
H6Anytime connectivity → Human-AI rapport0.191 **0.0493.446Supported
H7Perceived value → eWOM behavior intention0.551 ***0.0817.605Supported
H8Human-AI rapport → eWOM behavior intention0.531 ***0.0578.155Supported
Note: ** = p < 0.01; *** p < 0.001; NS = not significant.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Yang, F.; Ying, T.; Liu, X. Echoes of Innovation: Exploring the Use of Voice Assistants to Boost Hotel Reputation. J. Theor. Appl. Electron. Commer. Res. 2025, 20, 46. https://doi.org/10.3390/jtaer20010046

AMA Style

Yang F, Ying T, Liu X. Echoes of Innovation: Exploring the Use of Voice Assistants to Boost Hotel Reputation. Journal of Theoretical and Applied Electronic Commerce Research. 2025; 20(1):46. https://doi.org/10.3390/jtaer20010046

Chicago/Turabian Style

Yang, Fang, Tianyu Ying, and Xuling Liu. 2025. "Echoes of Innovation: Exploring the Use of Voice Assistants to Boost Hotel Reputation" Journal of Theoretical and Applied Electronic Commerce Research 20, no. 1: 46. https://doi.org/10.3390/jtaer20010046

APA Style

Yang, F., Ying, T., & Liu, X. (2025). Echoes of Innovation: Exploring the Use of Voice Assistants to Boost Hotel Reputation. Journal of Theoretical and Applied Electronic Commerce Research, 20(1), 46. https://doi.org/10.3390/jtaer20010046

Article Metrics

Back to TopTop