Published via Technological Forecasting and Social Change.
Very pleased to share this timely article that examines the antecedents of the users’ trust in Generative AI’s recommendations, related to travel and tourism planning.
I would like to thank my colleagues (and co-authors), namely, Hari Babu Singu, Debarun Chakraborty, Ciro Troise and Stefano Bresciani, for involving me in this meaningful research collaboration. It’s been a real pleasure working with you on this topic!
https://doi.org/10.1016/j.techfore.2025.124407
Highlights
- •The study focused on the enablers and the inhibitors of generative AI usage
- •It adopted 2 experimental studies with a 2 × 2 between-subjects factorial design
- •The impact of the cognitive load produced mixed results
- •Personalized recommendations explained each responsible AI system construct
- •Perceived controllability was a significant moderator
Abstract
Generative AI models are increasingly adopted in tourism marketing content based on text, image, video, and code, which generates new content as per the needs of users. The potential uses of generative AI are promising; nonetheless, it also raises ethical concerns that affect various stakeholders. Therefore, this research, which comprises two experimental studies, aims to investigate the enablers and the inhibitors of generative AI usage. Studies 1 (n = 403 participants) and 2 (n = 379 participants) applied a 2 × 2 between-subjects factorial design in which cognitive load, personalized recommendations, and perceived controllability were independently manipulated. The initial study examined the probability of reducing the cognitive load (reduction/increase) due to the manual search for tourism information. The second study considers the probability of receiving personalized recommendations using generative AI features on tourism websites. Perceived controllability was treated as a moderator in each study. The impact of the cognitive load produced mixed results (i.e., predicting perceived fairness and environmental well-being), with no responsible AI system constructs explaining trust within Study 1. In study 2, personalized recommendations explained each responsible AI system construct, though only perceived fairness and environmental well-being significantly explained trust in generative AI. Perceived controllability was a significant moderator in all relationships within study 2. Hence, to design and execute generative AI systems in the tourism domain, professionals should incorporate ethical concerns and user-empowerment strategies to build trust, thereby supporting the responsible and ethical use of AI that aligns with users and society. From a practical standpoint, the research provides recommendations on increasing user trust through the incorporation of controllability and transparency features in AI-powered platforms within tourism. From a theoretical perspective, it enriches the Technology Threat Avoidance Theory by incorporating ethical design considerations as fundamental factors influencing threat appraisal and trust.
Introduction
Information and communication technologies have been playing a key role in enhancing the tourism experience (Asif and Fazel, 2024; Salamzadeh et al., 2022). The tourism industry has evolved as a content-centric industry (Chuang, 2023). It means the growth of the tourism sector is attributed to the creation, distribution, and strategic use of information. The shift from the traditional model of demand–driven to the content-centric model represents a transformation in user behaviour (Yamagishi et al., 2023; Hosseini et al., 2024). Modern travellers are increasingly dependent on user-generated content to decide on their choices and travel planning (Yamagishi et al., 2023; Rahaman et al., 2024). The content-focused marketing approach in tourism emphasizes the role of digital tools and storytelling to assist in creating a holistic experience (Xiao et al., 2022; Jiang and Phoong, 2023). From planning a trip to sharing cherished memories, content helps add value to the travellers and tourism businesses (Su et al., 2023). For example, MakeMyTrip (MMT) integrated generative AI trip planning assistant which facilitates conversational bookings assisting the users with destination exploration, in-trip needs, personalized travel recommendations, summaries of hotel reviews based on user content and voice navigation support positioning the MMT’s platform more inclusive to the users. The content marketing landscape is changing due to the introduction of generative AI models that help generate text, images, videos, and interesting code for users (Wach et al., 2023; Salamzadeh et al., 2025). These models assist in expressing the language, creativity, and aesthetics as humans do and enhance user experience in various industries, including travel and tourism (Binh Nguyen et al., 2023; Chan and Choi, 2025; Tussyadiah, 2014).
Gen AI enhances natural flow of interactions by offering personalized experiences that align with consumer profiles and preferences (Blanco-Moreno et al., 2024). Gen AI is gaining significant momentum for its transformative impact within the tourism sector, revolutionizing marketing, operations, design, and destination management (Duong et al., 2024; Rayat et al., 2025). Accordingly, empirical studies suggest that Generative AI has the potential to transform tourists’ decision-making process at every stage of their journey, demonstrating a significant disruption to conventional tourism models (Florido-Benítez, 2024). Nonetheless, concerns have been raised about the potential implications of generative AI models, and their generated content might possess inaccurate or deceptive information that could adversely impact consumer decision-making (Kim et al., 2025a, Kim et al., 2025b). In its report titled “Navigating the future: How Generative Artificial Intelligence (AI) is Transforming the Travel Industry”, Amadeus highlighted key concerns and challenges in implementation Gen AI such as data security concerns (35 %), lack of expertise and training in Gen AI (34 %), data quality and inadequate infrastructure (33 %), ROI concerns and lack of clear use cases (30 %) and difficulty in connecting with partners or vendors (29 %). Therefore, the present study argues that with the intuitive design, the travel agents could tackle the lack of expertise and clear use of Gen AI. The study suggests that for travel and tourism companies to build trust in Gen AI, they must tackle the root causes of user apprehension. This means addressing what makes users fear the unknown, ensuring they understand the system’s purpose, and fixing problems with biased or poor data. Also, previous studies highlighted how the integration of Gen AI and tourism throws certain issues such as misinformation and hallucinations, data privacy and security, human disconnection, and inherent algorithmic biases (Christensen et al., 2025; Luu et al., 2025). Moreover, if Gen AI provides biased recommendations, the implications are adverse. If the users perceive that the recommendations are biased, they avoid using them, leading to high churn and abandoning platforms (Singh et al., 2023). Users’ satisfaction will decline, replaced by frustration and anger as biased output damages the promise of personalized services. This negatively impacts brand reputation and loss of significant market competitive advantage (Wu and Yang, 2023). Such scenarios will likely lead to stricter regulations, mandatory algorithmic audits, and new consumer protection laws forcing the industry to prioritize fairness as well as explainability to avoid serious consequences. Interestingly, research studies draw attention to an interesting paradox, that consumers are heavily relying on AI-generated travel itineraries, even when they are aware of Gen AI’s occasional inaccuracies (Osadchaya et al., 2024). This reliance might stem from a belief that AI’s perceived objectivity and capacity for personalized recommendations indicate a significant transformation of trust between human and non-human agents in the travel decision-making process (Kim et al., 2023a, Kim et al., 2023b). Empirical findings indicate that AI implementation in travel planning contributes to the objectivity of the results, effectively mitigates cognitive load, and supports higher levels of personalization aligned with user preferences (Kim et al., 2023a, Kim et al., 2023b). Despite the growing body of literature explaining the role of trust in Gen AI acceptance and its influence on travellers’ decision making and behavioural intentions, the potential biases in AI-generated content continue to pose challenges to users’ confidence (Kim et al., 2021a, Kim et al., 2021b). Therefore, this research aims to examine the influence of generative AI in tourism on consumers’ trust in AI technologies, particularly their balance between technological progress and ethical responsibility, concerning the future of tourism (Dogru(Dr. True et al., 2025).
Existing research has focused more on the technology of AI as a phenomenon rather than translating those theories into studies on how the ethics involved would affect perceptions and trust (Glikson and Woolley, 2020). In addition, there is still the black box phenomenon, which is the inability of the user to understand what happens in AI. It also emphasizes the need for more integrative studies between morally sound AI development, user trust, and design in tourism (Tuo et al., 2024).
Moreover, scant research has examined the factors that inhibit tourists from embracing Generative AI technologies, resulting in limited understanding of travellers’ reluctance to Generative AI adoption for travel planning (Fakfare et al., 2025). Despite a growing body of literature examining the antecedents and outcomes of Generative AI (GAI) adoption, large body of research has been based on established frameworks such as Information Systems Success (ISS) model (Nguyen and Malik, 2022), Technology Acceptance Mode; (TAM) (Chatterjee et al., 2021), and the Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh, 2022).
However, the extensive reliance on traditional acceptance models might face the risk of ignoring the critical socio-technical aspects, which are paramount in the context of GAI (Yu et al., 2022). While most of the studies explore the overarching effects of user acceptance and use of GenAI using TAM, UTAUT, and Delone and McLean IS success models, there has been a lack of consideration of ethical factors as well as responsible AI systems. Addressing these gaps could significantly broaden our theoretical understanding of how individuals evaluate and adopt generative AI technologies within users’ ethical behaviour and socio-technical perspective.
Therefore, this research aims to fill this gap by investigating factors that facilitate or inhibit trust in generative AI systems, considering responsible AI and Technology Threat Avoidance Theory, and advancing the following research questions:
RQ1
How does the customer experience of using generative AI in tourism reflect the impact of enablers (such as responsible AI systems) and inhibitors (such as ambiguity and anxiety) on trust in generative AI?
RQ2
Does perceived controllability moderate the enablers and inhibitors of trust in generative AI in tourism?
This research includes responsible AI principles and the technology threat avoidance theory to explicate the relationship between generative AI and trust in tourism. Seen from the conceptual lens of Ethical Behaviours, responsible AI principles are crucial for enhancing trust in Gen AI within tourism (Law et al., 2024). When users perceive Gen AI recommendations as fair, transparent, and bias-free, they are more likely to perceive the systems as trustworthy, which in turn mitigates user skepticism and promotes trust (Ali et al., 2023). Also, when Gen AI promotes sustainable and environmentally friendly practices, it demonstrates ethical responsibility and enhances trust in alignment with shared social values (Díaz-Rodríguez et al., 2023). By operationalizing responsible AI principles like transparency, fairness, and sustainability, Gen AI transforms from a black-box tool into a more trustworthy and responsible system for travel decisions (Kirilenko and Stepchenkova, 2025). From the socio-technical perspective, the Technology threat avoidance theory (TTAT) supports the logic of how perceived ambiguity and perceived anxiety act as inhibitors of trust. In tourism, users’ experience holds paramount importance (Torkamaan et al., 2024). When users encounter Gen AI content that is difficult to comprehend, recommendations are unstable or ambiguous, and users’ data is exposed to privacy concerns, these apprehensions will turn into a threat to using Gen AI (Bang-Ning et al., 2025). According to TTAT, when users perceive a greater threat, they are more inclined to engage in avoidance behaviours, which also erodes trust in the system. Hence, TTAT explains why users might hesitate or avoid using Gen AI tools, even if they offer functional benefits such as personalized recommendations and reduced cognitive load (Shang et al., 2023).
The study adopted an experimental research design that would help us to explore the independent phenomenon (use of Gen AI for content generation) and observe and explain its role to establish a cause-and-effect relationship between factors of responsible AI systems and TTAT (Leung et al., 2023). The experimental setting helps us to understand the differences empirically between human and non-human generated content from users’ travel decision-making perspective towards destinations. The study enriched the literature on both the ethical aspects and environmental aspects (perceived fairness and environmental well-being) and the perceived risks (perceived ambiguity and perceived anxiety) perspective in the tourism context. The situation of perceived controllability as a moderator is tested in the literature, offering help to managers on how to develop AI systems responsible for lowering user fear and building trust. The study also facilitated practitioners in understanding how the personalized recommendations & cognitive load facilitated by Gen AI in content generation impact the Gen AI Trust of the tourists.
Access through your organization
Check access to the full text by signing in through your organization.
Section snippets
Responsible AI systems
Responsible AI adequately incorporates ethical aspects of AI system design and implementation and ensures that the systems are transparent, fair, and responsible (Díaz-Rodríguez et al., 2023). Responsible AI includes ethical, transparent, and accountable use of artificial intelligence systems, ensuring they are fair, secure, and aligned with societal values. It is also an approach to design, develop, and deploy AI systems so that they are ethical, safe, and trustworthy. It is a system that
Cognitive load, personalized recommendations, and perceived fairness
Cognitive load is the mental effort to process and choose information (Islam et al., 2020). A cognitive load can also be high when people interact with complex systems such as AI. Thus, high cognitive load may affect the ability of users to judge whether the AI-based decisions can be considered fair, since they may not grasp enough of the workings of the system and its specific decisions (Westphal et al., 2023). On the other hand, whereas perceived fairness refers to the users’ feelings about
Research methods and analysis
The experiments adopted in this study are scenario-based. Participants’ emotions cannot be manipulated easily in an ethical manner (Anand and Gaur, 2018). Also, the scenario-based approach helps test the causal relationship between constructs used for experimentation in a given scenario. This approach also reduces the minimal interference from extraneous variables. In this method, respondents answered questions based on hypothetical scenarios developed in each scenario. Therefore, scenarios
Discussion
Study 1 shows that cognitive load is detrimental to an individual’s notion of justice or environmental wellbeing, indicating that such factors may be difficult for a user to rate properly based on expending greater cognitive effort. However, cognitive load can also limit the extent of open-mindedness and critical evaluation of AI-assisted communication (T. Li et al., 2024), which could leave people resorting to mental shortcuts or simple fairness and environmental fairness issues. Under such
Theoretical implications
Trust is an important element in the design of organizations and systems, and the current study’s theoretical implications extend the understanding of trust in generative AI systems by integrating constructs of responsible AI and Technology Threat Avoidance Theory. This research underscores the significance of moral factors in creating and using AI systems by exploring relationships between perceived justice, environmental concern, and trust. In this context, the study notes that the degree of
Practical implications
To develop and retain users’ confidence, professionals in the field should observe responsible AI principles, particularly perceived equity and ecological sustainability. It is possible for consumers to be amused by and trust that AI recommendations are perceived as fair. This involves developing algorithms that align with users’ interests while promoting green aspects in AI. It also becomes important for management to note that during AI interface design, cognitive load should be considered so
Limitations and future research
This study has certain limitations. First, the use of self-reported measures could pose certain biases, as the participants’ experiences with generative AI or social desirability could affect their judgment. The reliance on self-reported data introduces potential biases from participants’ prior engagements with generative AI, social desirability bias, or limited technological competence. Secondly, focusing on a particular context (i.e., tourism) can be seen as a limitation when it comes to
Conclusion
A thorough examination of advancing artificial intelligence in the tourism industry draws attention to the fact that there is no way of avoiding the issue of encouraging responsible AI use. Extending user satisfaction with rhetoric based on AI suggests that user perceptions are not only shaped by the quality of the recommendations but also by the ethical implications of the system and users’ affective states. A range in the effect of personalized suggestions on some parameters that influenced
Substantia Mea 
You must be logged in to post a comment.